Digital Telephony

DigitalTelephony

WILEY INTELECOMMUNICATIONS SERIES ANDSIGNAL PROCESSING JohnG. hoakis. Editor Northeastem University Introduction to Digital Mobil Communications Yoshihiko Akaiwa Digital Telephony, Sril Etlition John Bellamy E lements of I rfonuttion Theory Thomas M. Cover and Joy A, Thomas Fundame ntals of Telecommunicat ions Roger L. Freeman P ractic al Data Communicat ions Roger L. Freeman Radio SystemDesign for Telecommunications, Znd Edition Roger L. Freeman Telecommunication SystemEngineering, 3rd Edition Roger L. Frceman Telecommunications Transmission Handb ook, 4th Etlition Roger L. Frceman Introduction to Communications Engineering, 2nd Edition Robert M. Gagliardi Optital Communications, Znd Edition Robert M. Gagliardi and Sheman Ksxp Active Noise Control Systemt: Algorithm"s and DSP Implementations Sen M. Kuo and Dennis R. Morgan Mobile Communications Design Fundamentals, 2nd Edition William C, Y, Lee Expen SystemApplications for Telecommunications Jay Liebowitz Digital Signal Esilrndtion Robert J. Mammone, Editor Digital Communication Receivers: Synchronization, Channel Estimation, and Sigtnl Processing Heinrich Meyr, Marc Moeneclaey, afld Stefan A, Fechtel Synchronization in Digital Comntunications, Volume I Heinrich Meyr and Gerd Ascheid Business Earth Stationsfor Telecommunications Walter L. Morgan and Denis Rouffet Wirele ss I nfo rmat ion N etwo tk Kaveh Pahlavan and Allen H. lcvesque Satellite Communicationt: The First Quarter Century of Senice David W. E. Rees Fundamentals of TeIecommunicat fun N etw orks Tarek N. Saadawi, Mos'tafa Ammar, with Ahmed El Hakeem Meteor Burst Communicalions: Theory and Practice Donald L, Schilling, Editor Vector Space Projections: A Numerical Approar:h to Signal and Image Processing, Neural Nets, and, Optict: Henry Stark and Yongyi Yang Signaling in Telecommunitation Networl
(( L

DigitalTelephony Third Edition JohnC. Bellamy TX CoPPell,

glgf ef$*.

r:rll6lnt jt

f

ru

adrar*hr

ITnn"f

r.r.R.c.

Publication A Wiley-lnterscience JOHNWILEY& $ONS,lNC. r TOronto . Brisbaner Singapore NewYork. Chichesterr Weinheim

-l* n 5 \oE .1 r t

'B

b\.ts

2 ooo

This book is printed on acid-freepaper,€ Copyright@2000by JohnWiley & Sons,Inc, All rights reserved.Publishedsimultaneouslyin Canada. No part of this publicationmay be reproduced,storedin a retrieval systemor transmittedin any form or by any means,electronic,mechanical,photocopying,recording,scurningor otherwise,exceptaspermit. ted by Sections107or 108of the 1976United StatesCopyrightAct, without eitherthe prior written permissionof the Publisher,or authorizationthmughpaymentof the appropriateper-copylee to the copyrightclearancecenter,222RosewoodDrive,Donvers,MA 0t9?3, (s0g)7j0-g400,fax (50g)750_ 4?44.Requeststo the Publisherfor permissionshouldbe addressedto the PermissionsDepaxtrnent, John wiley & sons,ftrc.,605Third Avenue,New york, Irry 10158-0012, (212)8s0-601l,fax (zl?) 8506008,E-Mail: PERMREQ@ WILEY.COM. For orderingandcustomerservice,call I-800-CALL-WILEY. lihrary of CongressCataloging-in-PuhlfuationData: Bellamy,John,l94l* Digital telephony/ JohnBellamy.*3rd ed. p.cm,- (Wiley seriesin t'elecornmunications andsignalprocessing) "A Wiley-Lrtersciencepublication." Includesindex, ISBN0-471-34571-7 l. Digital telephone systems. I. Title. IL Series, TK5103.7.844 2000 6?1.385-dc2l 99-34015 Printedin the United Ststesof America 10987654321

To myfather for passingon the enioymentof being an engineer

CONTENTS xvll

Preface Acknowledgment

xtx

Acronyms

xxl

/ chapter 1 Backgroundand Terminology StandardOrganizations3 1.1 Telecommunications 1.? TheAnalogNetworkHierarchY 5 HierarchY 6 1.2.I Bell SYstem U.S.Network 10 I.2.2 Postdivestiture L23 SwitchingSYstems 12 SYstems l8 I-2.4 Transmission "1.2.5 Pair-GainSYstems24 1.2.6 FDM MultiplexingandModulation 26 Media 28 1.2.7 WidebandTransmission Impairments 33 1.2.8 Transmission I'2.9 Powerkvels 4l 1.2.10 Signaling 42 1.2.11 AnalogInterfaces46 1'Z.lZ TheIntelligentNetwork 49 Routing 51 1.2.13 DynamicNonhierarchical I.2.14 CellularRadioTelephoneSystem 52 1.2.15 VoicebandDataTransmission 54 1.3 TheInfioductionof Digits 56 1.3.1 VoiceDigitization 56 I.3.2 Time DivisionMultiplexing 58 VII

CONTENTS

1.3.3 DataunderVoice 63 1.3.4 DigiralMicrowaveRadio ffi 1.3.5 FiberOpticTransmission 65 1.3.6 DigitalSwitching 65 1,.3.7 Digital NerworkEvolution 67 References69 Problems 7l Chapter 2 Why Dlgital? .

73

2.1 Advanrages of Digital VoiceNetworks 7j 2.1.1 Easeof Multiplexing 7j 2.1.2 Easeof Signaling i4 2.1.3 Useof ModernTechnology 75 2.1.4 Inregrationof Transmission andSwitching 77 2.1.5 SignalRegenerarion78 2.1.6 PerformanceMonitorability79 2.1.7 Accommodation of OtherServices g0 2.1.8 Operabilityat Low Signal-to-Noise/Interference Ratios 80 2.1.9 Easeof Encryption 8l 2.2 Digital SignalProcessing 8l 2.2.1 DSPApplications Bz 2.3 Disadvantages of Digital VoiceNetworks g4 2.3.1 Increased Bandwidth 84 2.3.2 Needfor Time Synchronization85 2.3.3 TopologicallyResrricred Multiplexing g5 2.3.4 Needfor Conference/Extension Bridges g6 2.3.5 ftrcompatibilities with AnalogFacilities g7 References88

Chapter 3 Volce Digitizatlon 3.I

PulseAmplitudeModulation 93 3.1.1 NyquistSamplingRate 94 3.1.2 FoldoverDistortion 95 3.2 PulseCodeModulation 98 3.2.1 Quantization Noise 99 3.2.2 Idle ChannelNoise l0Z 3.2.3 Uniformly EncodedpCM 103

91

CONTENTS iX

3.?.4 Companding 106 3.2.5 EasilyDigitally LinearizableCoding 108 3.2.6 SyllabicCompanding 116 3.2.7 AdaptiveGainEncoding 119 3.3 SpeechRedundancie$121 3.3.1 NonuniformAmplitudeDistributions 122 I22 3.3.2 Sample-to-SampleCorrelation 122 3.3.3 Cycleto-CycleCorrelations Correlations 123 3.3.4 Pitch-Interval-to-Pitch-Interval 3.3.5 InactivityFactors 124 3.3.6 NonuniformLong-TermSpecnalDensities IZ4 3.3.7 Short-TermSpectralDensities 127 3.4 Differential PulseCodeModulation 127 3.4.1 DPCM Implementations 129 3.4.2 HigherOrderPrediction l3l 3.4.3 AdaptiveDifferentialPCM 131 3.5 DeltaModulation 133 3.5.1 SlopeOverload 134 3.6 AdaptivePredictiveCoding 136 Coding 138 3.7 Subband 3.8 Vocoders 141 3.8.1 ChannelVocoder 142 3.8.2 FormantVocoder lM 3.8.3 LinearPredictiveCoding 144 LinearPredictiveCoding 147 3.8.4 Enhanced-Excitation 151 3.9 Encoder/DecoderSelectionConsiderations 3.9.1 VoiceQuality 151 for NonvoiceSignals 152 3.9.2 Transparency Enors 153 3.9.3 Toleranceof Transmission 3.9.4 Delay 154 3.10 ITU-T CodingStandards 154 References 155 Problems 158 Ghapter 4

Dlgital Tranemission and Multlplexing 4.1 PulseTransmission 162 4.1.1 IntersymbolIntederence 164 4.1.2 Timins Inaccuracies 164

161

X

CONTENTS

4.1.3 InsufficientBandwidthl& 4.1.4 AmplitudeDistortion 165 4.1.5 PhaseDistortion 165 4.2 Asynchronous versusSynchronous Transmission 165 4.2,.1 AsynchronousTransmission 166 4.2.2 SynchronousTransmissionI6i 4.3 Line Coding L7l 4.3.1 LevelEncoding L7I 4.3.2 BipolarCoding 173 4.3.3 Binary N-ZeroSubstirution 176 4.3.4 PairSelectedTernary L19 4.3.5 TernaryCoding 180 4.3.6 DigitalBiphase 181 4.3.7 DifferentialEncoding 183 4.3.8 CodedMark Inversion 183 4.3.9 Multilevel Signaling 184 4.3.10 Partial-Response Signaling 185 4.4 Eror Performance 189 4.4.1 SignalDetection 190 4.4.2 NoisePower 190 4.4.3 Enor Probabilities191 4.5 PerformanceMonitoring198 4.5.1 Redundancy Checks 198 4.5.2 SignalQualityMeasurements201 4.5.3 FramingChannelErrors 203 4.5.4 Performance Objectives 2O3_ 4.5.5 ForwardErrorCorrection 2O4 4.6 Time DivisionMultiplexing 207 4.6.I Bit Interleavingver$usWord Interleaving 208 4.6.2 Framing 209 4.6.3 DSI ExtendedSuperframe Zl5 4.7 Time DivisionMultiplex LoopsandRings 216 References 219 Problems 221 Chapter 5

Digltal$witching 5.1 SwitchingFunctions ZZE 5.2 SpaceDivisionSwitching 227 5.2.1 Multiple-StageSwitchingZ3O

ZZs

CONTENTS

5.3

5.4

5.5

5.6

5.2.2 BlockingProbabilities:L.eeGraphs 234 Jacobaeus238 5.2.3 BlockingProbabilities: 5.2.4 FoldedFour-WireSwitches 242 5.2.5 Pathf,rnding243 5.2.6 SwitchMatrix Control 24 Time Division Switching 246 5.3.1 AnalogTime DivisionSwitching 246 5.3.2 Digital Time Division Switching 247 Switching 251 Two-Dimensional 5.4.1 STSSwitching 255 5.4.2 TST Switching 257 5.4.3 No. 4 ESSToll Switch 262 5.4.4 System75 Digital PBX 264 Systems 265 Digital Cross-Connect 5.5.1 ConsolidationandSegregation 267 5.5.2 DCSHierarchy 268 Equipment 269 5.5.3 IntegratedCross-Connect Digital Switchingin an AnalogEnvironment 27O

Switching 270 5.6.1 Zero-Loss 5.6.2 BORSCHT 272 5.6.3 Conferencing 272 References273 Problems 274

Chapter6 Dlgltal Modulatlon and Radlo Sy$tsms 6.1 Digital Modulation 279 6.1.1 AmplitudeModulation 280 6.1.2 FrequencyShift KeYing 284 6.1.3 PhaseShiftKeying 288 AmplitudeModulation 30I 6.I.4 Quadrature AmplitudeandPhaseModulation 309 . 6.1.5 Carrierless 6.1.6 Partial-Response QAM 310 311 6.1.7 Trellis-CodedModulation 315 6.1.8 MulticarrierModulation 6.2 Filter Partitioning 317 3I7 6.2.1 Adjacent-Channellnterference 318 6.2.2 OptimumPartitioning 6.3 EmissionSpecifications 320

277

xll

coNTENTS 6.4 RadioSysternDesign 322 6.4.1 FadeMargins jZZ 6.4.2 SystemGain 323 6.4.3 FrequencyDiversity 3.26 6.4.4 SpaceDiversity 327 6.4.5 Angle Diversity 327 6.4.6 AdaptiveEqualization 3ZB 6.4.7 RouteDesign 3ZB References 329 Problems 332

,,/ Chapter 7

Network Synchronization Controland Management 7.1 Timing 336 7.1.1 TimingRecovery: Phase-Locked Loop 336 7.1.2 ClockInsrabitity 337 7.I.3 ElasticStores 339 7.1.4 JitterMeasurementsj4Z 7.2

7.3

7.4

7.5 7.6

7.1.5 Systematic Jitter 345 Timing Inaccuracies 346 7.2.1, Slips 346 7.2.2 AsynchronousMultiplexing 351 7.2.3 WaitingTimeJitter 359 NetworkSynchronization361 7.3.I Plesiochronous362 7.3.2 NetworkwidePulseStuffing 363 7.3.3 MutualSynchronization3M 7.3.4 NetworkMaster 364 7.3.5 Master-SlaveSynchronization 365 7.3.6 Packetization366 7.3.7 NetworkTiming Performance Measurements366 U.S.NetworkSynchronization 370 7.4.1 Synchronization Regions 370 7.4.2 PrimaryReference Sources 372 7.4.3 1996AT&T Synchronization Architecrure j73 NetworkConhol 373 1.5.1 HierarchicalSynchronization Processes374 NetworkManagement 376

335

coNrENrs xlll 7.6.1 RoutingControl 376 '7.6.2 Flow Control 377 References380 Problems 382

-€6apter I

FiberOptlcTransmisslonSystems Ll

8.2

8.3 8.4

8.5

8.6

383

SystemElements 386 FiberOpticTransmission 8.1.I OpticalFiberFundamentals387 Transducers390 8.1.? Electrical-to-Optical 393 8.I.3 Optical-to-ElectricalTransducers Line Codesfor FiberOpticTransmission 395 8.2.I mBnBLine Codes 396 8.2.2 Bit InsertionCodes 399 WavelengthDivisionMultiplexing 401 FiberSystemDesign 403 andSplices 404 8.4.1 FiberConnectors 404 Switching Protection 8.4.2 8.4.3 SystemGain 405 SONET/SDH 406 8.5.1 SONETMultiplexingOverview 408 8.5.2 SONETFrameFormats409 Administration,and 8.5.3 SONETOperations, Maintenance 4l I Justification 4I4 8.5.4 PayloadFramingandFrequency 417 Tributaries Virtual 8.5.5 MaPPing 422 8.5.6 DS3PaYload MaPPing 423 8.5.7 E4 PaYIoad 8.5.8 SONETOPticalStandards 425 8.5.9 SONETNetworks 426 SONETRings 429 Ring 4Zg 8.6.1 UnidirectionalPath-Switched 43I Ring Line-Switched 8.6.2 Bidirectional

References 433 Problems 434 Chapter I

Digital Moblle Telephony 9.1 NorthAmericanDigital Cellular 437 Format 438 9,1.1 D-AMPSTransmission

437

xlv

coNTENTS 9.1.2 D-AMPSSpeech Coding 43l.9 9.1.3 D-AMPSControlChannel 439 9.1.4 D-AMPSError Conrrol 440 9.2 GlobalSystemfor Mobile Communications 44I 9.2.1 GSM ChannelStructure Ml 9.2.? GSM Speech Coding M3 ' 9.2.3 GSM ChannelCodingandModulation 443 9.2.4 GSM Mobile Station 443 9.2.5 GSM FrequencyHopping 444 9.2.6 GSM ShortMessageService 444 9.3 CodeDivisionMultiple-AccessCellular 444 9.3.1 CDMA ChannelEstablishment 445 9.3.2 CDMA MultipathTolerance MB 9.3.3 CDMA powerConhol M9 9.3.4 CDMA SoftHandoff 449 9.4 PersonalCommunication System 450 9.5 VolcePrivacyandAuthenticarion 450 9.6 Iridium 45I 9.7 TrunkedRadio 45? 9.8 CellularDigiralPacketDara 453 References453 Problems 454

Chapter 10 Data and Asynchronous Transfer Mode Networks 10.I Message Switching 456 10.2 PackerSwitching 458 10.2.1 PacketFormats 460 1O.2.2 StatisticalMultiplexing 461 IO-2.3 RoutingControl 46j 10.2.4 Flow Control 466 10.2.5 X..25 46e 10.2.6 FrameRelay 471 10.2.7 TCP/rP 473 10.3 Asynchronous TransferModeNetworks 474 10.3.1 ATM Cells 474 10.3.2 ATM ServiceCategories474 10.3.3 ATM Connections4i7

455

CONTENTS XV

10.3.4 ATM Switching 477 10.3.5 ATM Applications 484 10.4 InternetProtocolTransport 490 References 492 Problems 494 '/Chapter 11 Dlgital $ubscriber Accese I 1.I

Il.2

495

IntegratedServicesDigital Network 496 11.1.1 ISDN BasicRateAccessArchitecture 497 11.1.2 S/Tlnterface 499 I Ll.3 ISDN U Interface 501 11.1.4 ISDN D ChannelProtocol 503 loops 503 Digital Subscriber High-Data*Rate Line 503 11.2.1 AsymmetricDigital Subscriber

rr.2.2 VDSL 507 11.3 Digital Loop CarrierSystems 507 I L3.1 UniversalDigital Inop CarrierSystems 507 11.3.2 lntegratedDigital Loop Carier Systems 508 Digital Loop Carrier 1l .3.3 Next-Generation Systems 509 11.4 Fiberin theLooP 510 I L5 HYbridFiberCoaxSYstems 5l I I 1.6 VoicebandModems 512 11.6.1 PCMModems513 II.7 Local MicrowaveDistributionService 515 I1.8 DigitalSatelliteServices 516 References516 Problems 5I7

/Cnapter tZ Traffic Analysis 12.1 Traffic Characterization520 12.1.1 ArrivalDiskibutions 5M 12.1.2 HoldingTime Distributions 527 12.2 LossSystems 530 12.2.1 LostCallsCleared 531 12.2.2 Lost CallsReturning 536 12.2.3 LostCallsHeld 539 12.2.4 Lost CallsCleared-Finite Sources 54I

519

xv|

CoNTENTS 12.2.5 Lost CallsHeld-Finite Sources 544 12.3 NetworkBlockinghobabilities 547 12.3.I End-to-EndBlockingprobabilities 549 I2.3.? Over{towTraffic 551 12.4 DelaySystems 552 12.4.1 ExponentialServiceTimes 555 12,.4.2 ConstantServiceTimes 559 . l?,.4.3 FiniteQueues 561 12.4.4 Tandemeueues 566 References 567 Problems 568

AppendlxA Derlvatlzationof Equations

573

AppendixB Encodlng/Decodlng Algorlthmsfor segmgntedPcM

579

Appendk G AnatyticFundamentalsof DigitalTransmisslon

587

AppendixD TrafficTabtes

607

Gloseary

613

Answersto SelectedProblem*

631

lndex

635

PREFACE As mentionedin the prefaceof the first two editions,the termdigffaI telephonytefers netpath of voice communications to the useof digital technologyin the message works.In thiscasethetetmdlgital refersto a methodof encodingthesignal-that is, andswitcha form of modulation.Hencedigitaltelephonyimpliesvoicetransmission of Althoughthe primaryfocus this book not datacommunications. ing applications, treatmentof datacomthis editioncontainsanexpanded is not datacommunicafions, municationsnetworks,particularlyastheyrelateto providingvoicecommunications servicesin additionto datacommunications. technologyandnetThis bookcoversall aspectsof digitalvoicecommunications works.It is not a technicalbookin thetraditional,analyticalsenseof communications theoryarealreadyavailable, theory.Sincenumerousbookscoveringcommunications system the applicationandoperationalaspectsof communications this book stresses design.Somebasictheoryis presentedin both qualitativeand, when appropriate, terminology, terms.Themainputpose,however,is to introduceconcepts, quantitative [n mostcasestheconceptsaresupinfluenceimplementations. andhow applications network,althoughexin theU.S.telephone portedby citingexampleimplementations amplesfrom other(ITU) publictelephonenetworksarealsoprovided' Theprimaryaudiencefor this book aregraduateelectricalengineers.The electrical to corffnureferences engineeringstudentis mostcapableofappreciatingoccasional nicationstheoryandits influenceon the practice.However,becauseanalyticalrigor lessanalyticallyorientedreadersshould is waivedin favorofoperationaldescriptions, theprinciples.Chapter6 (coveringdigitalradioand haveno difficulty understanding modulation)is the mostanalyticalbut is easilyskippedwithoutlosingcontinuityfor Similarly,Chapter12(coveringtrafficanalysis)containsnumerous theotherchapters. for understandingthe materialin otherchapters. equationsthat areunneces$ary Whenthe first editionof Digital Telephonywaswritten (1980)'public telephone with analogtechnology,but networksaroundtheworld wereprimmily implemented it wasclearthatdigitaltechnologywasrapidlytakingover'Whenthe secondedition waswritten (1990),the inlemalportionsof the networkhad,for the mostpart,been convertedto an all-digitalnetwork.Thenandtoday(1999)the mainremnantsof the xvll

xviii

PREFAcE

original analogtelephonenetworks are analog subscriberloops and analog telephones connectedto them. Although Integrated services Digital Nerwork (ISDN) technology was developed as a means of replacing analog loops to complete the transformation of the network to suppofi end-to-end digital connections,ISDN deployment is below expectationsfor severalreasons.One of thesereasonsis a growing needformore bandwidth than what is available from a basic rate IsDN subscriberloop (128 kbps). There is currenrly much activity within the industry to develop new technologiesfor medium- and highbandwidth digital subscriberaccess.A new chapter (chapter l1) has been added to this edition to specifically addressalternative technologiesfor digital subscriberacCESS.

Anotherrelativelyrecentapplicationof digital technologyaddedto this edition involvesdigital cellulartelephones,which first appearedin the marketplacein the mid-1990s.Digital mobileradio is enabledby the emergence of low-cost,highperformance digitalsignalprocessing (DSp)technologyfor compressing speechsignals to low bit rates and for providing sophisticatedcoding, modulation,and equalization requiredfor digitalradiosin a bandwidth,constrained mobileapplication. A completelist of chaptertopicsis; chapterl: overview of analogtelephonetechnologyfollowedby an inhoduction of how digitaltechnologyis usedro fulfrll the samefunctions chapter2; Discussionof advantages anddisadvantages of digital technologyfor voicecommunications chapter3: Descriptions of themostcomrnonvoicedigitizationalgorithms chapter4: Fundamenral$ of digiralwire-linerransmission andmultiplexing chapter5: Basicconceptsandoperationsof digital switchingmachines Chapter6: Digital modulationandradiofundamentals chapter 7; Network synchronization,control, andmanagement requirements Chapter8: Fiberoptictransmission systemsandSONET Chapter9: Digital cellulartelephonesystems Chapterl0: Datanetworks Chapter1l: Digital subscriber accesstechnology Chapter12:Fundamentals of traffic analysisfor designingnetworks Theappendices coverthederivationofequations,pcM voicecodingrelationships, fundamentals of digitalcommunications theory,andtraffic tables. JoruqC. BeLLnr4y

Coppell,Tems October1999 j [email protected]

ACKNOWLEDGMENT

Once again I am indebted to Wanda Fox and Alcatel USA for allowing me accessto the corporate library for researchmaterials for this edition. I also owe a great deal of gratitude to Gerald Mitchell of the University of Colorado for thoroughly reviewing and enhancing the last chapter on traffic theory'

J.B,

xlx

ACRONYMS AAL ABR ACD ACK ADM ADPCM AIN AMPS ANI APC APON ARPANET ARQ ATM ATPC BSZS BER BISDN BLSR CAC CAP CAS CBR

ccc

CCIS

ccn"T CCS CDMA CDO CDPD

layer ATM adaptation availablebit rate automaticcall distributor (Positive) acknowledgment adaptivedeltamodulation;add-dropmultiplexer adaptivedifferentialPCM advancedintelligentnetwork advancedmobilephonesYstem automaticnumberidentification adaptivepredictivecoding ATM basedpassiveopticalnetwork advancedresearchprojectsagencynetwork automaticrepeatrequest transfermode asynchronous adaptivetransmitpowercontrol binary8 zerosubstitution bit-enor rate broadbandintegratedservicesdigital network bidirectionalline switchedring connectionadmissioncontrol modulation amplitude/phase competitiveaccessprovider;carrierless signaling channelassociated constantbit rate(AfUl clearchannelcapability coilrmoncharurelinteroffice signaling ConsultativeCommitteefor IntemationalTelephonyand Telegraphy (now tTU) cornmonchannelsignaling;hundredcall seconds codedivisionmultipleaccess communitydial office cellular digital packetdata(for AMPS networks)

xxll

AcHoNyMS

CELP codeexcitedlinearprediction CES circuitemulationservice(ATM) CGSA CellularGeographic ServiceArea CLASS cu$tomlocal areasignalingservices CLEC competitivelocal exchangecarrier CLP cell losspriority (ATM) CMI codedmarkinversion CODEC CODer/DECoder CPFSK continuousphasefrequencyshift keying CRC cyclicredundancycheck CSMA/CD carriersensemultipleaccess/collision detection CSU channelserviceunit CTD cell kansferdelay CTI computertelephonyintegration D-AMPS digitaladvanced mobilephoneservice DAVIC digitalaudiovideocouncil DBS directbroadcast satellite DCM digitalcircuitmultiplication DCME digitalcircuitmultiplicationequipment DECT digitalenhanced cordlesstelephony DFE decisionfeedbackequalization DID direct inward dialing DLC digital loop carrier DM deltamodulation;degraded minute DMT discretemultitone DNIS dialednumberidentificationservice DPCM differentialpulsecodemodulation DQDB distributedqueuedualbus DSI digital signallevel I at 1.544Mbps DS3 digitalsignallevel3 at,14.736 Mbps DSI digital speechinterpolation DSS digital satellitesystem DTE dataterminalequipment DTMF dualtonemulrifrequency (signalingtones) DVB digital video broadcastinggroup DVD digitalvideodisc El Europeandigital signallevel I at ?.048Mbps E3 Europeandigital signallevel 3 at 34.368Mbps ECMA EuropeanComputerManufacturersAssociation EMI elechomagneticinterference ERMES enhanced radiomessage system ESF extendedsuperframe ETSI EuropeanTelecommunications StandardsInstitute FDDI fiber dishibuteddatainrerchange

ACRONYMS XXIii

FDM FEC FEXT FIFO FRAD FSK FTTC FTTH GPS GSM HDB3 HDLC HIPPI HTTP IDLC IEC IETF ILEC IMT IP ISDN ISI ISO ITU M JPEG LAN LATA LD-CELP LEC LMDS MAN MCM MLCM MMDS MPEG MPLS MSK MTIE MTSO MULDEM NAK NCP NEXT

frequencydivisionmultiplexing forward error correction far end crosstalk first in-first out framerelayaccessdevice frequencyshiftkeying fiber to the curb fiber to the home globalpositioningsYstem globalsystemfor mobilecommunications high densitybipolarof order3 high-leveldatalink control high performanceparallelinterface hypertexttransportprotocol integrateddigital loop carrier Commission InternationalElectrotechnical task force engineering internet incumbentlocal exchangecarrier internationalmobiletelecommunications internetProtocol integratedservicesdigitalnetwork intersymbolinterference Organization IntemationalStandards Union InternationalTelecommunications interactivevoiceresPonse ExpertsGroup JointPhotographic network local area local accesstransPoflarea low-delayCELP carrier local exchange local microwavedistributionservice metropolitanareanetwork multicarriermodulation multilevelcodedmodulation multichannelmultipointdistributionservice Motion PicturesExpertsGroup multiprotocollabelswitching minimumshift keYing maximumtime interval enor mobiletelephoneswitchingoffice multiplexer-demultiplexer (negative) acknowledgment networkcontrolpoint;networkcontrolprotocol(ARPANET) nearendcrosstalk

XXIV

ACRONYMS

NMT NNI NRZ OFDM OSI PABX PAM PBX PCM PCME PCR PCS PDC PDH PHS PLL PON POTS PRC PRK PRS PSK PSTN PVC Qos QPRS QPSK RADSL RCC SDH SDLC SES SF SIM SLIC SMDS SMR SMS SNMP SOHO SONET SRTS S57 STM

Nordicmobiletelephonesystem network*to-networki nterface nonreturnto zero orthogonalfrequencydivisionmultiplexing opensystemsinterconnection privateautomaticbranchexchange(alsopBX) pulseamplitudemodulation privatebranchexchange pulsecodemodulation packetcircuitmultiplicationequipment peakcell rate (ATM) personalcommunication system(or service) personaldigitalcellular(Japan) plesiochronous digitalhierarchy personalhandyphone sy$tem(Japan) phaselockedloop passiveopticalnetwork plainold telephoneservice primary referenceclock phasereversalkeying partialresponse signaling;primaryreferencesource phaseshift keying publicswitchedtelephonenetwork permanent virtual circuit qualityof service quadrature partialresponse signaling quaternary phaseshift keying(4-PSK) rateadaptivedigital subscriber loop radio commoncarrier synchronous digitalhierarchy synchronous datalink control severelyerroredseconds superframe Subscriber IdentificationModule(GSM) subscriber loopinterfacecircuit switchedmultimegabitdataservice specialized mobileradio shortmessage service simplenetworkmanagement protocol small office/homeoffice synchronous opticalnetwork synchronous residualtime stamp signalingsystemversion7 synchronoustransfermode

ACRONYMS

STS STS-n SVC TI T3 TACS TASI TCM TCPAP TDM TDMA TETRA TMN TST UBR UDP UI UMTS UM UPSR URL UTP VBR VCC VCI VPC VPCI WAN WATS

digital switchingstructure space-time-space transportsignal-n synchronous switchedvirnralcircuit systemat 1.544Mbps TDM transmission TDM hansmissionsystemat 44'736Mbps system total accesscommunications speecbinterpolation time assignment trelli$codedmodulation protocol controlprotocol/internet transmission time divisionmultiPlexing time divisionmultipleaccess trunkedradio Trans-European Network Management Telecommunications digital switching$tructure time-space-time unspecifiedbit rate UserdatagramProtocol unit interval universalmobiletelephoneservice user-to-networkinterface unidirectionalpathswitchedring universalresourcelocator unshieldedtwistedpair variablebit rate virtual channelconnection virtual channelidentifier virtualpathconnection virtual pathconnectionidentifier wide areanetwork services wide areatelecommunications

XXV

1 ANDTERMINOLOGY BACKGROUND in theUnitedStatesandaroundtheworld Beginningin the1960s,telecommunications differentareas'First, the conventional in several changes Uefanundergoingradical upon to provide many new and different being called was *ulog telephonenetwork industry'Second,themardataprocessing from the emanated mostof which services. stimulatedcompetitionin States in the United agencies ketplaceandthe regulatory Third, digitaltechnolservices. monopolistic both old andnew areasof traditionally andswitchingfunctransmission fundamental many of the ogyemergedto implement the world' The around networks and other network tions wittrin ttreU.S. telephone operationalasand application, the design, is to describe main purposeof this book analogteleof the the technology As background, pectsof tfis newdigitalequipment. of digital introduction for the a framework provide phonenetworkis reviewedto equipment. thattheintroductionof digitaltechnologyinto thetelephone limust beemphasized networkwasmotivatedby desiresto improvethe quality,addnew features,andreDigitizationofthe networkdid notarise ducethecostsofconventionalvoiceservices. for betterdatahansmissionservices' industry processing from the needsof the data into thenetworkwasinitially inacintroduced Indeed,mostof thedigitaltechnology channels' Of course,a digitalnetwork analog cessibleto datafiaffic, exceptthrough As moreof the network services. is a naturalenvironmentfor datacommunications availablefor data became of the facilities use becamedigitized,morertupportfor direct high-endbusifor relatively only exi$ted Initially, directdigital access applications. Network Digital Services the Integrated facilitiesrof It wasnot until nessapplications. be usedby could digital channels (ISDtiibecameavailablethatend-to-end{switched apother 1990s numerous ny tfre late for bothvoiceand'datp,l individualsubscribers for primarily available, became facilities to digital proaches roprovidingdigitalaccess 1l' Chapter in are described technologies Thesevariousdigitalaccess Intemetaccess. idea that the to show has been included Figurel. I .A$a pointof historicalreference, of integratedvoiceanddatais not new.This figuredepictsa conceptof a Germaninto the prevailingmeans ventornamedPhillip Reis [1] to addvoicecommunications developedthe equipReis at the time-the telegraph' of electricalcommunications GraharnBell reyears Alexander before ment in tfie 1860sanddied in 1874-two

2

BAcKcRoUNDANDTERMtNoLocy

ceived his patentfor the telephone.As indicated,the figure implies altemateuseof the wires for voice or data cornmunications(i.e., integratedtransmission).Reis actually used the telegraphattachmentto signal infbrmation pertaining to voice tests,an indication of inadequatevoice quality. To implement simultaneousvoice and telegraphcommunications,the telephonein Figure 1.1 would have to have been digital. Becauseof technology limitations at the time, such an implementation was impossible and telephone systems necessarily evolved with analog technology. one hundred years later the situation chrurgedsignificantly. Telephoneequipment developersand service providers had an abundance of new technology, and they were challengedwith how to make effective use of it. This book describesdigital telephonetechnology from two perspectives.The first perspectivedescribesindividual equipmentsor subsystemsand technical reasonsfor transitions from conventional analog equipment to seemingly less natural digital counterparts.Thus, one purposeofthis book is to describehow digital technology improves and expandsthe capabilitiesof various subsystemswithin voice telephonenetworks' Another purposeof the book is to describethe ultimate benefits derived when an entire network is implementedwith digital techniques.A greatdegreeof synergism exists when individual systemsare designedinto one cohesivenetwork utilizing aigital implementations throughout. The synergistic effect benefits conventional voice servicesand newer $ervicessuch as ttre in-G:iiiei.

Flgure 1.1 Back to the future: the first integratedvoice/data communication $vstem.

1.1 TELECOMMUNICATIONSSTANDAHDORGANIZATIONS3

Most of the equipment descriptions urd design examples presentedin this book come from material authoredby engineersat AT&T Laboratories(now Lucent Technologies) and other suppliersfor the public telephonenetwork. The basic principles, however, are by no meansunique to the public telephonenetwork. The conceptsand implementation examples are applicable to any communications network: public or private, voice or data.An inherent attribute of a digital network is that it can,to a large extent, be designedindependentlyof its application. Terminals, Transmiesion, and Swttchlng The three basic elements of a communications network are terminals, tran$mission systems,and switches.The first part of this chapterprovides an overview of theseelements as implementedin analogtelephonenetworks. Then, the last part of this chapter provides a,lrief overview of digital implementationswithin the analog network. Following a detailed discussionof the motivation for digital implementationsin Chapter 2, the next four chaptersdescribethe operation and design of the basic elementsof a digital voice network. Chapter 3 discussesdigital voice tetminals and the most common algorithms used to convert analog voice signals into digital bit streams.Chapter 4 presentsthe basicsof digital transmissionsy$tems.Fundamentalsof digital switching follow in Chapter 5. Basic digital modulation techniquesand their application to point-to-point digital microwave and digital cellular systemsare describedin Chapter 6. A discussionof various synchronizationand control considerationsfor digital networks is provided in Chapter 7. Chapter I describesfiber optic transmissionsystems and the synchronousmultiplexing stiurdard(SONET)' Chapter 9 discussesthe basic architectureand operation of prevailing digital cellular $ystemsin use in the United Statesand around the world. The main emphasisof the first nine chaptersinvolves circuit switching astraditionally implementedfor voice telephonenetworks.A circuit-switched network is onp_that assignsa completeend-to-endconnectionin responseto eachrequestfor service\Each connection, with its asrrociatednetwork facilities, is held for the duration of the'call. ) Chapter l0 describesa different type ofnetwork, generically referred to as a packetswitched network, that is particularly suitedto servicing datatraffic. Included in Chapter 10 is a discussion of Asynchronous Transfer Mode (ATM), a form of a packet-switchednetwork. Chapter 1l discussesvarious technologiesand systemsfor achieving direct digital accessto a digital network (voice or data).The last chapterpresentsthe basicsof Faffic theory: the fundamentalmathematicsfor analyzing and predicting telecommunicationsnetwork performance.

ORGANIZATIONS STANDARD 1.1 TELECOMMUNICATIONS standl, 1984,telecommunications onJanuary of theBellSystem Priortothebreakup ards in North America were essentially establishedby the dominant equipment designer and supplier to the Bell System: Bell Telephone Laboratories and Westem Electric. Independenttelephonecompaniesthat provided local service to the 207oof the country not coveredby the Bell Systemrelied on the U'S. IndependentTelephone

l

4

encxcRouNDANDTEHMINoLOGY

Association[usITA; larerreferredto asrheu.s. TelephoneAssociation(usTA)] to formulateand disseminate particularlyfor interconnecting standards, with the Bell System. In anticipationof the divestitureof the RegionalBell operating companies (RBocs) from AT&T, the Exchangecarriers standardsAssociation(ECSA) was formedin 1983as a nonprofittradeassociation to representthe interestsof all exchangecarriers(RBocs and independents). In Februarylgg4 the ECSA sponsored the establishment of the Tl standards committeeto formulatenew interconnection standards for theu.s. nationalnetwork.TheTl commifteeis accredited by theAmerican NationalStandards Institute(ANSD to ensurethat standardsapprovalsfollow principlesof openness. ThusTl committeestandards aredesignated asANSI Tl.nnndate(T I standsfor Telecommunications standards entitynumberI ). TableI . I liststhe majorsubcommittees within Tl andtherespective responsibilities. Otherorganizations in North Americathat establishstandards relatedto telecommunicationsaretheElecffonicIndustriesAssociation(EIA), theInstituteof Electrical and ElectronicEngineers(IEEE), and Bell communicarion$Research(Bellcore). Bellcorewasan organizationcharteredto establishstandards andqualify equipment for the RBOCs.Bellcorehassincebeenreorganized asTelcordiaTechnologies. The IEEE is mostknownfor its datacommunications standards listedin Table1.2but has alsoestablished numerousstandards for measuringandcharacterizing telecommunicationseguipment. Most of the world outsideof North Americarelieson internationaltelecommunications$tandards committees e$tablished undertheauspices of theInternational Telecommunicationunion (ITU). In the past,two major entitieswithin the ITu were established:the InternationalTelegraphand TelephoneConsultativecommittee (ccITT) andtheInternarional Radioconsultativecommittee(ccIR). ccITT establishedrecommendations for telephone,telegraph,anddatatransmission circuitsand equipment.ccIR was concernedwith coordinatingthe useof the radio specrrum. CCITT andCCIR activitiesareno longeridentifiedasbeingdistinctfrom the ITU. ccITT hasbecomeITU-T andccIR is now ITU-R. In the united states.useof the radiospectrumis controlledby theFederalcommunicationscomrnission(FCC). North AmericanstandardsandITU standardshaveoften beenincompatiblein the past.Notth Americanstandards established by theBell SystemwerethereforeincorTABLE1.1 T1 StandardsSubcommltteee Committee T1A1 T1E1 T1M1 T1P1 T1S1 T1X1

Responsibility Performance andsignalprocessing powerandprotectionfor networks Interfaces, Internetwork operations, administration, maintenance, (IOAM&p) andprovisioning Wireless/mobile services andsy$lems Services, architeclures, andsignaling Digitalhierarchy andsynchronization

HIERARCHY NETWORK 1.2 THEANALOG Area Nstwork(LAN/MAN)Data TABLE1.2 IEEELocal Arsa NetworldMetropolltan 'CbmmunlcstionsStandards 802"1 802.2 802.3 802.4 802.5 802.6 802.7 802.8 802.9 802.10 80e.11 802:12 802.14

VirtualbridgedLAN(VLAN) Bridging, andArchitecture, Overview (LLC) LogicalLinkControl (cD) (Ethernet) Access(csMA)withcollisionDetection carriersenseMultiple TokenBus(Arcnet) TokenRing(lBMBing) (QPSX) Exchange QueuedPacketSynchronous Broadband OpticalFiberTechnologies Seryices Integrated Security Wireless DemandPriority CableTV

asa subset.Due to the needfor moreinternaporatedinto CCITT recommendations now work closely andITU-T committees iional compatibility,theT1 subcommittees joint effort is the the of first major example joint A standards. to establish together fiber transmissionsystemsrefenedto as SONET in the standardfor synchronous united statesandsynchronousDigital Hierarchy(sDH) in ITU-T standards. Organization(ISO) is an organizationwith standards The InternationalStandards activitiesin a wide rangeof subjectmatters,someof which involvetelecommunicawithin ISo work closelywith ITU studygroupsin tions.Technicalsubcommittees particularlythoserelatedto ISDN protocolsthat formulatingITU recommendations, standaldfor OpenSysadhere,asmuchaspossible,to theISO datacommunications Model. (OSI)Reference temsInterconnection

1.2 THE ANALOG NETWORKHIERARCHY of theworld evolvedovera periodof almost Becausethe analogtelephonenetworlLs alsodeveloped' 100years,a greatarnountof diversityin equipmentimplementations thatvastnetworks,like theU.S' network,couldaccomachievement It is a remarkable modatethemyriadof equipmenttypesandfunctionproperly.In 1980-in theunited [2], almostall of whichcoulddirectly Statesalone.therewere181million telephones good qualityconnection'This achievea and have number public telephone dial any andwell-definedfunctionalhierinterfaces mentwa$madepossibleby standardized it necessarilyadheredto the installed, archies.As newer digital equipmentwas *Although

1980 is there is no specific date at which digital technology took over from analog.technology, analog displacing emerged to begin fiber optics in which time frame repfesents the significant in that it portions of the rirlios for intercity transmission, the last stronghold of analog technology in the intemal public network.

6

BecxcRouNDANDTERMtNoLocy

standardizedpracticesof the analog network. The fact that the equipment was implemented with digital technology was transparentto the rest of the network.

1.2.1 BellSystemHierarchy Alexander Graham Bellinvented thefirstpractical telephone in 1876.It soonbecame

apparent,however, ttrat the telephonewas of little use without some meansof changing connections on an "a$-needed"basis. Thus the flrst switching office was established in New Haven, connecticut, only two years later. This switching office, and othersfollowing, was locatedat a central point in a serviceareaand provided switched connectionsfor all subscribersin the area.Becauseof their locations in the serviceareas,the$eswitching offices are often referred to as cenffal offices. As telephoneusage grew and subscribersdesired longer distance connections,it becamenece$saryto interconnectthe individual serviceareaswith trunks betweenthe central offices. Again, switches were neededto interconnecttheseoffices, and a second level of switching evolved. Continued demandfor even longer distanceconnections, along with improved long-distancetransmissionfacilities, stimulatedevenmore levels of switching. In this manner the analog public telephonenetwork in the United states evolved to a total of five levels. Theselevels are listed in Table I .3. At the lowest level of the network are class5 switching offices, also called central offices (cos) or end offices (Eos). The next level of the network was composedof class4 toll offices. The toll network of the Bell Systemcontainedthree more levels of switching: primary centers,sectionalcenters,and regional centers. To illushate the structure and motivation for hierarchical networks, a symbolic, three*levelexample is shown in Figure 1.2. In contrast,Figure 1.3 depicts a different network structurefor interconnectingall of the firstlevel switches;a fully connected mesh structure. obviously, the hierarchical network requires more switching nodes but achievessignificant savings in the number of trunks; the transmission links between switching offices. Detetmination of the total number of trunk circuits in either network is necessarilya function of the amount of traffic betweeneachpair of switching nodes.(Chapter 12 provides the mathematicsfor determining the number of trunk circuits.) As a first approximation, the trunk costsof a mesh can be determinedas the total number of connections(trunk groups) N" between switching off,rces: TABLE 1.3 PubticNetworkHierarchyot the Bell $ystem (1gSA)tgl SwitchClass I

2 Q

4

5

Functional Designation Regional center Sectional center Primary center Tollcenter Endoffice

No. in Bell System

No. in lndependents

Total

10 52 148 508 9803

0 0 20 425 9000

10 67 168 933 18,803

HIERAHCHY 7 NETWORK 1.2 THEANALOG

l) N"=+N(N-

(l'1)

whereN is thenumberof nodes. aScomparedto 12conThusthe meshnetworkof Figure1.3has36 connections' thecostcomparisonof nectionsin Figure1.2.In the caseof fiber optictransmission

Figure 1.2 Three-levelswitchinghierarchy.

network. Flgure 1.3 Mesh-connected

BACKGROUND ANDTERMINOLOGY

the trunks is almost exactly 3 ; I becausea single fiber systemcan provide more voice capacity than is neededbetweenany two switches. A lessobvious difference betweenthe networks of Figures 1.2 and 1.3 involves the method of establishingconnectionsbetweentwo offices. In the hierarchical network thereis one and only one route betweenany two switching nodes.[n the meshnetwork most connectionswould be establishedon the direct route between the two offices. However, if the direct route is unavailable (becauseof a traffic overload or an equipment failure) and the first-level switches can provide trunkto-funk connections (called tandemswitching functions), the mesh network provides many altemativesfor establishingconnectionsbetween any two nodes.Hence the reliability of a network architecturemust be consideredin addition to just the costs.In general,neither a pure mesh nor a purely hierarchical network is desirable. Taking these factors into account, Figure 1.4 depicts alternate routing as implemented in the former Bell System.As indicated, the basic backbonehierarchical network was augmentedwith high-usagetrunks. High-usage trunks are used for direct connectionsbetween switching offices with high volumes of interoffice traffic. Normally, traffic betweentwo suchoffices is routed through the direct trunks. If the direct trunks are busy (which may happenfrequently if they are highly utilized), the backbone hierarchical network is still available for alternaterouting. Traffic was always routed through the lowest available level of the network. This procedurenot only usedfewer network facilities but also implied better circuit quality becauseof shorterpaths and fewer switching points. Figure 1.4 showsthe basic order of selection for alternateroutes. The direct interoffice trunks are depicted as dashed lines, while the backbone,hierarchical network is shown with solid lines.

):*;;; -t-t-

,F,

\\\\\\\

l;:;----

I

I

I

f

Figure 1.4 Altemateroutingin North Americannetwork.

1,2 THE ANALOGNETWORKHIERARCHY

In addition to the high-usagetrunks, the backbone network was also augmented with additional switching facilities called tandem switches.These switcheswere employed at the lowest levels of the network and provided switching betweenend offices. Tandem switcheswere not pafr of the toll network, as indicatedin Figure I .5, but were (and are) pafr of what is referred to as an exchangearea.Generally speaking,an exchangeareais an area within which all calls are consideredto be local calls (i.e., toll free). Il generalterms, any switching machine in a path betweentwo end offices provides a tandem switching function. Thus toll switchesalso provide tandem switching functions. Within the public telephonenetwork, however, the term tandem refers specifically to intermediateswitching within the exchangearea' The basic function of a tandem office is to interconnectthose central offices within an exchange area having insufficient interoffice fiaffic volumes to justify direct trunks. Tandem offices also provide alternateroutes fbr exchangeareacalls that get blocked on direct routesbetweenend offices. Although Figure 1.5 depictstandem offices as being physically distinct from end offices and toll offices, tandem switches were often colocatedwith either or both types.Operationally, exchangeareaswitching and toll network switching in the Bell systemwere always separated.The primary reason for the separationwa$ to simplify tandem switching by avoiding billing and network routing. A toll switch had to measurecall duration for billing purposesbut a tandem switch did not. Prior to the introduction of computer-controlled switching, billing functions were a significant consideration.The operationalseparationalso implied that toll-connecting trunk groups were separatefrom tandem trunk groups. The flexibility of computer-controlled switching has eliminated the need for the separation. The separation of exchange facilities from toll facilities had an important effect on the transmissionand switching equipment utilized in the respectiveapplications.Exchangeareaconnectionswere usually short and only involved a few switching offices. Toll connections,on the other hand, coulcl involve numerous switching offices with Toll netrrvork

Tandom office Di.€ct trunk

Figure 1.5 Exchangeareanetwork.

10

BAcKGRoUNDANDTEHMtNoLocy

relatively long ffansmission links between them. Thus, for comparable end-to-end quality, individual analog exchangeareaequipment did not have to provide as much quality as did toll network counterparts.

1.2.2 PostdivestitureU.$. Network In thedecade of the1980s thestructure of thepublictelephone networkin theUnited stateschangedsignificantlyasaresultofchangesin thetechnologyandtheregulatory environment. Themaintechnological changeswere(1) extensivedeploymentof very largedigital switchingmachines,(2) theadaptationof computer-controlled switches to providemultipleswitchingfunctionsin onemachine(e.g.,the integrationof endoffice, tandem,andtoll switchingfunctions),and (3) the deploymentof fiber optic hansmissionsystemsthatcouldcarryvery largecrosssectionsof traffic. Noticethat all threeof thesetechnological developments suggesta networkwith fewerandlarger switchingoffices.Thesetechnologicalinfluenceson the networktopologyare discussedmorefully in Chapters8-10. The most dramaticand immediateeffect on the network occurredon Januaryl, 1984,whenthebreakupof AT&T officially tookeffect.Because thebreakupinvolved divestitureof Bell operatingcompanies(BoCs) from AT&T, the networkirselfbecamepartitionedat a new level.The new partitioningis shownin Figure I.6, which depictsAT&T asoneof severalcompetinglong-distance carriersreferredto a$interexchangecarriers(IXCs) andlocal accessandtranspoftareas(LATAs), which were originally the exclusivedomainof local exchangecarriers(LECs).In additionto AT&T, the othertwo main IXCs areMCI andu.s. sprint. The LECs originallyincluded23 Bocs (organized into 7 RBocs), formerindependenr relephone companies like GTE, contel, and united relecommunications, and some1500mostly smalltown telephonecompanies. Mergerswithin the industryhavesubsequently reduced thenumberof LECsandRBOCs. Thenumberof LATAs in theunited stateswasinitially 164,but the numberhas changedasadjustments in serviceboundaries aresometimes made.Becausea LATA entailsan areathatincludesmanyexchangeareas,LECscompletetoll callsthatkaversedifferentexchangeareaswithin oneLATA. TheIXCs werenot allowedto carry intra-LATA traffic. similarly, an LEC wasnot allowedto carrytraffic betweentwo LATAs evenwhenbothLATAs mightbe serviceareasof a singleBoc. only anIxc wasallowedto carryinter-LATA traffic.To ensurethattheseservicepartitionswere adheredto, eachIXC interfacedwith a LATA at a singlepointin theLATA, referred to asa point of presence (PoP).IXC equipmentat a Pop couldbe a switchingoffice or merelyajunctionfor collectingtraffic carriedelsewhere to be switched. A majoraspectof themodifiedfinaljudgment(MFJ)thatspecifiedthedivestiture wastheconditionof equalaccess, whichmeantthatanLEC (specificallya Boc) was to treatall IXCs equallyin regardsto exchange access. Theconditionsofequalaccess meantthat accessto all endofficesin a LATA would be equalin type,quality,and pricefor all IXCs. The LATA nerworkritructure[4] established to accomplishequal access is shownin Fieure1.7.

NETWORK HIERARCHY 1 1 1.2 THEANALOG

Figure 1.6 U.S, network partitioning.

POP: AT: TO: EO: TIC: OIG: TCTC;

Poinl ol Ptesence Accets tandent Tarrdcnt office Entl ollice Tsndetn inter"LATA connecting Direct inter-LATA connecling Tdnderfi connecling

Figure 1.7 LATA hierarchyhndaccessarchitecture'

12

BACKGROUND ANDTEHMINOLOGY

The design of the LATA network for intra-LATA traffic was left to the discretion of the LECs. Thus intra-LATA connectionscan involve multiple switching offices between end offices. However, connectionsbetweenan Eo and a PoP could involve at most one intermediateswitching office referredto as an accesstandem (AT). with respect to the previous Bell system hierarchy, an AT takes the place of a class 4 toll switch. However, long-distancebilling functions, which were formerly performed in class 4 switches, are now performed within the IXC network. Although Figure l.7 shows accesstandem and basic tandem switching functions as being distinct, access tandem functions can be integratedinto regular tandem switchesif the tandem switch provides AT features.Foremost among thesefeahrresare the ability to forward automatic number identification (ANI) information to an IXC for billing and the ability to route calls to different IXC POPsdependingqn presubscriptionor per-call three-digit carier designations. In 1997 the FCC issued some rulings with the intent of stimulating competition in both the local exchange and long-distance networks. under this ruling, LECs that want to enter the long-distancemarket can do so if they open their local exchangefacilities to long-distancecarriers or other competitive accessproviders. A key aspect of making the local facilities available to competition is the establishmentof unbundled pricing for local seryices;the separationof the cost of the local loop, the local switching equipment, maintenance,and ancillary services such as 9ll emergency calling. Another key requirementis number portability, which allows a subscriberto changelocal service providers without having to changetelephonenumbers.The introduction of competition for local distribution instigated the use of two tennrr:competitive local exchange carrier (CLEC) for the competition and incumbent local exchangecarrier (ILEC) for the establi$hedcarrier.

1.2.3 SwitchingSystems Manual Swltchboards The first telephone switching equipment utilized operatorsat manual swirchboards.The operatorsaskeda caller for the number they wanted to call and then establishedthe connection by plugging in a cord between terminal jacks. Although switchboards are no longer used, a legacy of their existencelives on: the use of the terms "tip and ring." As shown in Figure 1.8, one wire of a wire pair was connectedto the tip of a plug comector and the other wire was connectedto the ring. Ever since,one wire of a wire pair is com-

Switdrboard jacf

Switdrboard plug

Flgure 1.8 switchboardplug with corresponding jack (R, s, andr arering, sleeve,andtip, (FromFreeman,Fundamentals respectively). of releconzmunications, wiley, New york.)

1.2 THEANALOGNETWORKHIEHARCHY

13

monly referredto asthetip andthe otheris referredto asthering, evenon digital wire On someof theoriginalswitchbomds pairs,whichhaveneverusedplugsin a swirchboard. provided the sleeve conductor shownin Figure1.8. would be by a thirclconnection Automated Switching with anyparticularswirchingmachinecan In generaltermstheequipmentassociated asprovidingoneof thefollowing functions: be categorized 1. Signaling 2. Control 3. Switching Thebasicfunctionof the signalingequipmentis to monitortheactivityof theincoming linesandforwardappropriatestatusor controlinformationto thecontrolelement of the switch.Signalingequipmentis alsousedto placecontrolsignalsontooutgoing linesunderdirectionof theswitchcontrolelement. incomingsignalinginformationandsetsup connecThecontrolelementprocesses tionsaccordingly.The switchingfunctionitself is providedby a switchingmatrix:an usedto completecomectionsbetweeninput linesand arrayof selectable crosspoints of a switchingmachineareshownin Figure1.9. outputlines.Thesebasicconstituents Electrcmechanicalswitching. Prior to the introduction of digital electronic switchingmachinesin thelate 1970s,switchingofficesin NorthAmericaandaround switches; the world wereequippedwith oneof two basictypesof electromechanical step-by-step 1.10, crosspoints of a As shown in Figure step-by-step* and crossbar. to dial pulses.As thepulsesof switcharewipercontactsthatmovein directresponse "step"theverticalwiperto a horizonthey immediately thefirst digit entertheswitch, to the first digit. After the properrow is selected,the wiper is tal row corresponding rotatedacrossanothersetof contactsuntil an idle line to the next stageof switching theseconddigit,thenstepsthesecis located.Thenextsetofdial pulses,representing throughhowevermany stagesare process The continues ond stagein like mannet. particular size. switch neededfbr a switchusesdirectprogressivecontrol:SuccesAs thenameimplies,a step-by-step aseachdigit is dialed.Wittr of a paththroughtheswitchareestabli$hed sivesegmenrs are progressive control,thecontrolelementsof the switch integratedinto the switching matrix.This featureis very usefulfbr implementinga varietyof switchsizesand allowing relativelyeasyexpansion.A progre$sivecontrol switch,however,has a numberof significantlimitations; paththroughtheswitchexists l. A call maybeblockedeventhoughanappropriate in anearlystage. anunfortunatepathgetsselected but is not attempted because possible. That is, the outgoingline trunks is not 2. Alternateroutingfor outgoing pulses be substituted' incoming dial and cannot is directlyselected by *A

step-by-stepswirch is also referred to as a Strowger switciz,in honor of its inventor Almon B. Strowger.

14

BACKGHOUND ANDTEBMINOLOGY

Figure 1.9 Switchingsystemcomponents.

3. Signaling schemesother than dial pulses (e.g., tone signaling) are not directly usable. 4 . Number translation is impossible. In conffast to a step-by-stepswitch, a crossbarswitch is one that used centralized, coillmon control for switch path selection.As digits were dialed, the conhol element of the switch received the entire addressbefore processingit. when an appropriate path through the switch was determined(which may have involved numbertranslation or alternaterouting), the control element transferredthe necessaryinformation in the form of control signals to the switching matrix to establishthe connection. The fundamentalfeature,and advantage,of a common control switch is that control function implementation is separatefrom the switch implementation. Common control cross-

SLE E BANI( SLE EVE Ii,IPER

LINE WIPER

V ER T I C A L W I P EH

V EF T I C A L COMMUTATOB I U S E OI N L I N E F I N D ER S I

W I P ER CORDS

Figure l.l0 Step-by-stepswitching elemenl (Copyright 1977 by Bell Telephone Laboratories. Reprintedby permission.)

HIERARCHY 15 1,2 THEANALOG NETWOHK

(telephone numbers)inbarsystemsintroducedtheability to assignlogicaladdresses of physicalline numbers. dependently The crosspointsof a crossbarswitch(Figure Ll l) aremechanicalcontactswith magnet$to setup and hold a connection.The term crostbdrarisesfrom the useof crossinghorizontalandverlicalbarsto initially selectthecontacts.Onceestablished, energizedwith directcunentpassthe switchingcontactsareheldby electromagnets circuit. When the circuit is opened,the loss of current ing throughthe established to be releasedautomatically. causesthecrosspoints switches limitationsof progressivs control,step-by-step Because of theoperational Crossbar switches, on the 5 switching offices. primarily in smaller class were used netand within the toll predominantly in metropolitan areas hand, were used other by with coillmon control were augmented In cases step-by-step switches work. $ome the processing the request, equipment. After the digits into special control receiving pulsesthat setup a connectionasif the switchwasrecontrolequipmentgenerated pulses directly. ceivingdial

l,lt( '-

.c'*I.

cffiEl

xfrss

luc,'F-5

ErilI

mlt'f,

Figure LlLll Crossbar switching element. (Copyright 1977 by Bell Telephone Laboratories. Reprinted by permission.)

16


Stored Program Control. Step-by-step andcrossbarswitchingsysrems usedelectromechanical components for boththe switchingmatrixandthecontrolelements. In somecasettthe electromechanical controlelementsin theseswitchesrepresented rudimentaryforms of special-purpose digital computers.The hardwiredelectromdchanicallogic, however,had limited capabilitiesand was virtually impossibleto modify. A majormilestonefor telephonywasestablished in 1965whentheBell Systeminstalledits first computer-controlled switchingsystem:theNo. I ElectronicSwitching system(Ess).- This switchingsystemusesa stored-program digitalcomputerfor irs controlfunctions.Thestored-program control(SPC)featureof theNo. I ESSallowed the introductionof new featuressuchas abbreviateddialing,call forwarding,call waiting,andthree-waycalling. Theintroductionof SPCnot only providedsignificantadvantages to endusersbut alsosimplifiedmanyadministrative andmaintenance tasksfor the operatingcompanies.A largeparl of line administration thatformerlyrequiredmanymanualmodifrcations(mainframecross-connects) could insteadbe accomplished with changesin computerdatatablesof an SPCswitch.Furthermore, physicalline numberswereindependent of thelogical(directory)line numbers,thusmakingnumberchanges easy. other benefitsenabledby sPc areautomated recordkeeping,lower blockingprobabilities,generationof traf-ficstatistics,automated call tracing,andmessage unit accounting(per-callchargesasopposedto flat-ratebilling for unlimitedlocalcalling). The switchingmahix of the No. I ESS(andalsorheNo. 2 ESS,No. 3 ESS,and No. IA ESS)is implementedwith electromechanical reedrelays.ThusthetermESS refersin generalto computer-controlled switchingandnot to thenatureof theswitching makix itself.However,AT&T's No. 4 ESS,which wasfirst installedin 1976.is a high-capacitytoll switch using computercontrol and digital electronicsfor its switchingmatrix.Thusthe No. 4 ESSis "electronic"in its controlandits switching matrix.Fufthermore,thedigital multiplexsystem(DMS) switchesof NorthernTelecom,theNo. 5 eleckonicautomaticexchange(EAX) of GTE, andthe No. 5 ESSof AT&T alsoutilize.digital logic circuitsfor thecrosspointmatrix. ("''f't

\

Private Branch Exchanges. In the united statesthe termprivate branchexchange(PBX) refersgenericallyto any switchingsystemownedor leasedby a businessor organizationto provideboth internalswitchingfunctionsand accessto the publicnetwork.Thusa PBX in theunited statesmayuseeithermanualor automatic control.The term PABX is alsousedin the united States,andparticularlyin other countries,to referspecificallyto automaticallycontrolledPBXs. Thehistoricaldevelopment of PBX systemshasfollowedcloselythatof switches in thepublicnetwork.PBXswith computerized controlbecameavailablein 1963(beforetheNo. I ESS)whenAT&T's No. l0l ESSwasfirst installed.sincethattime a *Computer-controlled

PBXs were available befbre 1965. The No. I E$S represents the first instance of computer control in the public network hierarchy,

*rra ? 1.2 THEANALOGNETWORKHIERARCHY

17

have developedcomputer-controlled large numberof independentmanufacturers PBXs.In fact,thePBX markethaslong beenoneof themostcompetitivebusinesses in telecommunications. newfeaturesfor users. Theuseof computercontrolfor PBXsintroducednumerous dialing)provided,butnuNot only werecustomized callingfeatures(e.g.,abbreviated alsobecameavailable.Someof themoreusemerousfacilitiesfor costmanagement featuresin a PBX arethefollowing: ful, commonplace by individualemployeeor department 1. Accountingsummaries 2. Multiple classesof servicewith prioritiesandaccessreshictionsto areacodes, WATS lines.andsoon 3. Least-costroutingto automaticallyselecttie lines,foreignexchangecircuits, WATS. DDD, andsoforth 4. Automaticcallbackwhencircuitsareavailable 5. Traffic monitoringandanalysisto determinetheutilizationof existingcircuits blockingprobabilitiesandnetworkcosteffectiveness or to a$certain Centr6x. Many of thefeaturesinitially providedby PBXsarealsoofferedby operAs indicatedin Figure1.12,Centrexis abusiness asCentrexfeatures. atingcompanies customerserviceoffering supportedby switchingequipmentin the centraloffice. Every telephoneor datadeviceat the customerpremiseshasa dedicatedchannelto the switchin the centraloffice. Originally,eachchannelimplieda dedicatedpair of wires.It is now morecommonto usemultiplexingtechniques(describedin Section from thecentraloffice point of costs.Nevertheless, I.2.5) to reducethehansmission appearance at theCO with a uniquepubview eachCentrexextensionhasa dedicated number.A softwarepartitionin thecentralofficetreatstheCenlic networktelephone trex linesasa closedusergroupto providethefollowingbasicfeatures:

Locll Common Chantel Slgnaling

Sit6 I

Sit6 2

Site 3

Figure 1.12 Centrexserviceto multiplesites.

18

BACKGRoUNDANDTERMINoLoGY

l. Directdialingto Centrexextensions from thepublicnetwork. 2. Station-to-station callingusingextensionnumbersasopposed to full, 7- (or l0-) digit publicnumbers. 3. Commonvoiceuserfeaturessuchascall forwarding,call transfer,call waiting, call pick up, andthree-waycalling. 4. Multiple siteswith transparent numberingplansand features.With citywide Centrexthe sitescan be supportedfrom multiple offices interconnected by common-channel signaling(CCS)describedin Section1.2.10. 5. Centralizedattendant/message desk with call origination information for informedprocessing of forwardedcalls. 6. High availabilitybecause CO equipmentanddirectlyconnected analogphones arepoweredat theCO with backuppowersources. 7. Virtually unlimitedgrowth. 1.2.4 Transmission Systems Functionally,thecommunications channelsbetweenswitchingsystemsarereferredto astrunks.In thepast,thesechannelswereimplemented with a varietyof facilities,includingpairsof wires,coaxialcable,andpoint-to-pointmicrowaveradiolinks.Except for specialsituations,tunk facilitiesnow utilizeopticalfibers. Open Wire A classicalpictureofthe telephonenetworkin the pastconsistedoftelephonepoles with crossarms andglassinsulatorsusedto $upportuninsulated open-wirepairs.Exceptin rura]environments, theopenwire hasbeenreplacedwith multipaircablesystemsor fiber. Themain advantageof anopen-wirepair is its relativelylow attenuatiofl (a few hundredths of a decibelpermile at voicefrequencies). Hence,openwire is particularlyusefulfor long,rural customerloops.Themaindisadvantages arehavingto separatethe wires with crossarmsto preventshortingandthe needfor largeamounts of copper.(A singleopen-wirestrandhasa diameterthatis five timesthediameterof a typical strandin a multipaircable.Thusopenwire usesroughly25 timesasmuch copperasdoescable.)As a resultof copperco$tsandtheemergence of low elecffonics costs,openwire in rural environments haribeenmostlyreplacedwith cablesystems using(digital)amplifiersto offsetattenuation on long loops. Paired Cable In responseto overcrowdedcrossarmsandhigh maintenartce costs,multipaircable sy$temswereintroducedasfar backas 1883.Todaya singlecablemay containanywherefrom 6 to 2700wire pairs.Figure 1.13showsthe structureof a rypicalcable. Whentelephonepolesareused,a singlecablecanprovideall thecircuitsrequiredon the route,therebyeliminatingthe needfor crossarms. More recentlythe preferred meansof cabledistributionis to bury it directlyin the ground(buriedcable)or use underground conduit(underground cable).

NETWoRK HIERARGHY19 1.2 THEANALOG Table 1.4 lists the most conrmon wire sizes to be fbund within paired-cablesystems. The lower gauge(higher diameter) systemsare urtedfor longer distanceswhere signal attenuationand direct-cunent (dc) resisturce can becomelimiting factors. Figure l . I 4 rrhowsattenuationcurves [5] for the common gaugesof pairedcable as a function of frequency. An important point to notice in Figure l.14 is that the cable pairs are capableof canying much higher fiequencies than required by a telephonequality voice signal(approximately3.4 kHz). In the past, the exchangeareasof the telephonenetwork used paired cable almost exclusively fbr short-haul interoffice transmission.Up until the introduction of mul-

Laboratorics. Reprintedby Figure 1.13 Multipair cable.(Copyright1977by Bell Telephone permission,)

20

BACKGROUNDAND TERMINOLOGY

TABLE1.4 WireGaugeand Resletance of Common PalredCable Gauge 30 28 26 24 22 20 19

(in.) Diameter

Direct-Current (fy1000ft)a Resistance

0.010 0.013 0.016 0.020 0.0?5 0.032 0.036

104 oo 41

26 16 10

eNote that the loop resistanceof a pair is twice the resistanc€of a single wire given in th€ table.

#

# g g

t

E

E

c o tl

18 17 18 15 14 13 12 1l

r0 I I 7 6 5 4 3 2 I 0

1,000 l.lfr Frequercy (Hz)

1,1d

1.100

1.107

Figure 1.14 Attenuation versus frequency of common gaugesof paired cable, (From W. D. Reeve, ,SuDscriDerLoop Signaling Transmission Handbook, IEEE Press, New york, Fig. 7-16a.)

1.? THEANALocNETwoHKH|FRARCHY 21

with groundretum, Figure 1.15 Single-wiretransmission

Figure 1.16 Two-wiretransmission. tiplexing techniques,describedlater in this chapter,eachvoice circuit (trunk) was carried on a separatepair of wires. Beginning in the early 1960selectronicsbegan to be used for short-haul interoffice transmissionsystemsusing multiplexing techniquesto carry multiple channelson a single pair of wires.

Tw*Wire Versue Four-Wire through All wire-linetransmission in thetelephonenetworkis basedon transmission pairsof wires.As shownin Figure L15, transmission througha singlewire (with a groundreturn)is possibleandhasbeenusedin thepast.However,theresultingcircuit pairsof wiresasshownin FigInstead,balanced is too noisyfor customeracceptance. ure 1.16areusedwith signalspropagatingas a voltagedifferencebetweenthe two wires.The electricalcurrentproducedby the differencesignalflowing throughthe wiresin oppositedirectionsis calleda "metalliccurrent.'rIn contrast,currentpropagatingin thesamedirectionin bothwiresis referredto ascommon-mode or longitudinal current.Longitudinalcurrentsarenot coupledinto a circuit outputunlessthere is animbalancein thewiresthatconvertssomeof thelongitudinalsignal(noiseor ininto a differencesignal.Thustheuseof a pairof wiresfor eachcircuitproterference) vides much better circuit quality than doessingle-wiretransmission.Someolder transmission switchingsystemrr to minimizethenumusedsingle-wire(unbalanced) ber of contacts.Unbalanced circuitswereonly feasiblein smallswitcheswherenoise andcrosstalkcouldbe controlled. networkareimplemented with a sinVirtually all subscriber loopsin thetelephone If users glepairof wires.-Thesinglepairprovidesfor bothdirectionsof ffansmission. their conversations are superimon both endsof a connectiontalk simultaneously, posedon thewire pairandcanbeheardattheoppositeends.In contrast,wireJine(and fiber) transmission overlongerdistances, asbetweenswitchingoffices,is bestimpleare separated wire pairs. mentedif the two directionsof transmission onto $eparate *It

is no* commonplace to use fiber for the tbeder portion of a subscdber loop, but the drop to a te sidence is a single pair per telephone.

22


Longer distance transmissionrequires amplification and most often involves multiplexing. These operationsare implemented most easily if the two directions of ffansmission are isolated from each other. Thus interoffice trunks typically use two pairs of wires or two fibers and are referred to as four-wire $ystems.The use of two pairs of wires did not necessarilyimply the use of twice as much copper as a two-wire circuit. After 1960, four-wire systemsincreasingly used some form of multiplexing to provide multiple channelsin one direction on one wire pair. Thus, a net savingsin copper could result. Sometimesthe bandwidth of a single pair of wires was separatedinto two subbands that were used for the two directions of travel. These systemswere referred to as derivedfour-wire systems.Hence, the term/aur-wire has evolved to imply separate channelsfor eachdirection of transmission,even when wires may not be involved. For example, fiber optic and radio system$that use separatechannelsfor each direction are also referred to asfour-wire systems. The use of four-wire transmissionhad a direct impact on the switching systemsof the toll network. Since toll network circuits were four-wire, the switches were designed to separately connect both directions of hansmission. Hence, two paths through the switch were neededfor each connection. A two*wire switch, as used in older analog end offices, required only one path through the switch for eachconnection.

Two-Wira-to-Four-Wlre Conversion At somepointin a long-distance connection it is necessary to convertfromtwo-wire transmissionof local loops to four-wire transmissionon long-distancetrunks. In the past,the conversionusually occurredat the trunk interface of the (two-wire) end office switch. Newer digital end office switchesare inherently "four-wire," which meansthe two-wire-to-four-wire conversion point is on the subscriber(line) side of the switch as opposedto the trunk side.A generalizedinterconnectionof two-wire and four-wire facilities for a connection is shown in Figure I.17. The basic conversion function is provided by hybrid circuits that couple the two directions of transmissionas shown. Hybrid circuits have been traditionally implemented with specially interconnected transformers.More recently, however, electronic hybrids have been developed.Ideally a hybrid should couple all energy on the incoming branch of the four-wire circuit into the two-wire circuit, and none of the incoming four-wire signal should be transferred to the outgoing four-wire branch.

Hybrid

Hybrid

Figure 1.17 Interconnection of two-wireandfour-wirecircuits.

1.2 THEANALocNETWoRKHTERARoHy eg When the impedance matching network Z exactly matches the impedance of the two-wire circuit, near-perfectisolation of the two four-wire branchescan be realized. Impedancematching usedto be a time-consuming,manual proce$$and was therefore not commonly used. Furthermore, the two-wire circuits were usually switched connections so ttre impedance that had to be matched would change with each connection. For thesereasonsthe impedancesof two-wire lines connectedto hybrids were rarely matched.The effect of an impedancemismatch is to causean echo, the power level of which is related to the degreeof mismatch. The effect of echoeson voice quality and the meansof controlling them is discussedlater in this chapter.

Loadlng Coils of the curvesshownin FigureI .14indicatethatthehigherfrequencies Theattenuation thanthe lower frequenmoreattenuation voicespectrum(up to 3.4kHz) experience distortsthevoicesignalandis referredto attenuation cies.This frequency-dependent asamplitudedistor"tion.Amplitude distortionbecomesmost$ignificanton long cable pairs,wherethe attenuationdifferenceis greatest. (3The usualmethodof combatingamplitudedistonionon intermediateJength inductance into the lines. The extra lS-mile)wire pairsis to insertartificialinductance comesfrom loadingcoilsthatariinsertedat 30fi)-,4500-,or 6000-ftintervals.Figure 1.18showstheeffectof loadingcoils on a 24-gaugeloop.Noticethatthe voiceband is devresponse up to 3 kHz is greatlyimproved,but theeffecton higherfrequencies astating. Prior to the introductionof wire-lineandfiber carriersystems,loadingcoils were usedextensivelyon exchangeareainterofficetrunks.Loadingcoils arealsousedon havedisplaced loops.Here,too,carriersystem$ thelonger,typicallyrural,subscriber routes. mostof the singlepairsof wiresbeingusedon long

gto E E, I

Szo e E

EI E t 0

Frequency (kllzl

tr'lgure1.18 Effectof loadingon 24-gaugecablepair.

24


1.2.5 Pair-GainSystems Providingserviceto rural subscribers hasalwaysbeenan expensivepropositionbecauseof thelengthof theroutesinvolvedandthesmallnumberof households to suppott thecostof boththeinitial installationandthemaintenance. In thepast,a cornmon meansof reducingthe co$t$wasto usepartylines,whichinvolvedsharingof a wire pair amongmultiplehouseholds. A partyline is particularlyusefulfor satisfyingnew servicerequests on routeswith no sparepairsbut is obviouslyobjectionable dueto the lackof privacyandthelack of availabilityof theline. A pair-gainsy$temis an alternateapproachto sharingpairs of wires that is much moreacceptable to theusers.This sectiondescribes two basictypesof pair-gainsys--;_ * (remoteswirches)andmultiplexers(carriersystems). tems;concentrators Goncentratlon Thefirst form of a pair-gainsystemin FigureL 19depictsa basicline concentration system.Whenviewedftom the stationsetendof the system,a pair-gainsystemprovidesconcentration by switchingsomenumberof activestationsto a smallernumber of sharedoutputlines.At theotherendof thesystem,deconcentration (expansion) occursby switchingfromthesharedlinesto individualinputsof theswitchingofficecorrespondingto theactivestations.Expandingthetraffic backto theoriginalnumberof stationsensure$ thatthesystemis operationallytransparent to boththeswitchandthe user.Noticethata definitionof whichendprovidesconcentration andwhichendprovidesexpansionis dependent on thepoint of view. N surrce$

2 3 /l{ chadnels

N $rbchandels

N tource$

Figure 1.19. Pair-gainsystems:concentration andmultiplexing:(a) concentration (z > M); (b) multiplexing.

1.2 THE ANALOGNETWOHKHIERARCHY

25

connectingall stationsit servis incapableof simultaneously Sincea concentrator When introducedby concentration. ices,a certainamountof blockingis necessarily theactivityof individualstationsis low enough,significantamountsof concenffation blockingprobabilities.For example,40stationsthat canbe achievedwith acceptable 10lineswithablocking areeachactiveonlyT.SVoofthetimecanbeconcentratedonto degradation in servicesinceanequallyacprobabilityof 0.001.* Thisis anacceptable tive calledstationis busy75 timesasoften. systemrequirestheffansferof controlinformationbeNoticethata concentration switchterminals.Whenoneendof thesystemestabtweentheconcentrator/expander to oneof the sharedlines,theotherendmustbe informedto lishesa newconnectron reversecomection. setup theappropriate Multlplexing As shownin Figure1.14,theinherentbandwidthof a typicalwire pairis considerably greaterthanthat neededfor a singlevoicesignal.Thus,multiplexingcanbe usedto imcarrymultiplevoicechannelson a singlepairof wires.Theincreasein attenuation is offsetby amplifiersin themultiplexequipmentand pliedby thehigherfrequencies at periodicpoints in the ffansmissionlines.The particularmultiplexingtechnique shownin Figure1.I9b is a frequencydivisionmultiplexsystem.Anotherform of mul* tiplexing,time divisionmultiplexingof digital voicesignals,is the preferredmultilater' plexingapproachfor digitalpair-gainsystemsdiscussed relationship betweenthecustomer As shownin FigureI . l9b, thereis a one-to-one system, Thus, unlike the concentration of the multiplexer. linesandthe subchannels pair-gain system. Also, type of thereis no possibilityof blockingin a multiplexing one-to-one relationsince the same thereis no needto transferswitchinginformation betweencustomerlinesat oneendandswitchingofshipdefinestheconespondence f,rcelinesat theotherend.A majordrawbackof multiplexingpair-gainsystemsis that arehighlyunderutilizedif thesourcesarerelativelyinactive.In these thesubchannels andmultiplexingis normallyjustified. situationsa combinationof concentfation Time Asslgnment Speech lnterpolation Time assignment speechinterpolation(TASI) is a pair-gainsystemthat dynamically assignsa channelto a circuit only whenthereis actualvoiceactivity.Thus,a TASI to one N; assignsactivesources voiceactivityfrom a numberof sources sy$temsenses of M channels,whereM is typicallyabouthalf aslargeasN; andsignalsthe far end is activefor only Normally,eachparticipantin a conversation abouttheconnections. thatif M = j N,thereissomeamountofsparecapac407oof thetime,whichindicates excessactivityin onedirection.If a sourcebeginsto talk whenall ity to accommodate channelsareutilized,thebeginningofthat speechsegment Setsclippeduntil a channel formulationsto determinethe becomesavailable.Chapter12providesmathematical probabilityof clippingasa functionof N, M, andthe voiceactivityfactor' Initial TASI applicationsinvolved improving the utilization of underseacable theuseof arathercomplicated of thesechannelswarranted pairs.Theobviousexpense *A

discussionof traffic analysis is provided in Chapter 12, from which this result can be obtained.

(') 26

BACKGHOUND ANDTERMINOLOGY

multiplexingtechniquefor thetime.Thesamebasictechniquehassincebeenusedin numerousapplicationswith digital speechfor satelliteand landline applications. Thesesystemsaregenerallycalleddigital speechinrerpolation(DSI) systems[6]. 1.2.6 FDM Multlplexingand Modutation The introductionof cablesystemsinto the transmission plant to increasethe circuit packingdensityof openwire is oneinstanceof multiplexingin thetelephone network. Thisform of multiplexing,referredto asspacedivisionmultiplexing,involvesnothing morethanbundlingmorethanonepairof wiresinto a singlecable.Thetelephone network usestwo otherformsof multiplexing,bothof whichuseelectronics to packmore thanonevoicecircuitinto thebandwidthof a singletransmission medium.Analogfrequencydivisionmultiplexing(FDM) hasbeenusedextensivelyin point-to-pointmicrowaveradiosand to a much lesserdegreeon $omeobsoletecoaxial cableand wire-line$ystems. FDM is alsoutilizedin fiber optic transmission systems,whereit is referredto aswavelength divisionmultiplexing(wDM).- Digitaltimedivisionmultiplexing(TDM) is thedominantform of multiplexingusedin thetelephone nerworks worldwide.(Eventhe fiber optic system$that utilize wDM commonlyusedigital TDM within the signalsof a particularwavelength.) Frequency Division Multlplexing As indicatedin FigureI.l9b, an FDM systemdividesthe availablebandwidthof the transmissionmediuminto a numberof narower bandsor subchannels. Individual voicesignalsareinsertedinto thesubchannels by amplitudemodulatingappropriately selectedcarrierfrequencies. As a compromisebetweenrealizingthe largestnumber of voicechannelsin a multiplexsystemandmaintainingacceptable voicefidelity,the telephone companies established 4 kHz asthe standard bandwidthof a voicecircuit.t If both sidebands producedby amplitudemodulationareused(asin obsoleteNI or N2 carriersystemson pairedcable),thesubchannel bandwidthis 8 kHz, andthecorrespondingcarrierfrequencieslie in the middle of eachsubchannel. Sincedoublesidebandmodulationis wastefulof bandwidth,single-sideband (ssB) modulation wasusedwhenevertheextraterminalcostswerejustified.Thecarrierfrequencies for single-sideband systemslie at eithertheupperor loweredgeof thecorresponding subchannel,depending on whethertheloweror uppersidebandis selected. TheA5 channel bankmultiplexerof AT&T usedlower sidebandmodulation. r...

FDM Htetrarcny In order to standardizethe equipment in the various broadbandtransmissionsystems of the original analog network, the Bell system establishedan FDM hierarchy as provided in Table 1.5. ccITT recommendationsspecify the samehierarchy at the lower *Optical

technology is customarily defined in terms of the wavelength of the optical signal as opposed to the corresponding frequency. 'Actually' the usable bandwidth ofan FDM voice channel was closer to 3 kFIz due to suard bandsneeded by the FDM separation filters.

27 1.2 THEANALOGNETWORKHIERARCHY TABLE1.5 FDMHierarchyof the Bell Nstwork Numberof Voice Circuits

Level Multiplex Voicechannel Group Supergroup Mastergroup Mastergroup Mux Jumbogroup Mux Jumbogroup

Formation

1

12 60 600 1,200-3,600 3,600 10.800

12voicecircuits 5 groups 10 supergroups Various 6 mastergroups 3 jumbogroups

Frequency Band(kHz) H 60*10B 312-552 56rt--3,084 7,548 31?,564-1 56,t-17,548 3,00160,000

FDM modlevels.Eachlevel of thehierarchyis implementedusinga setof standard particular broadband transmission is independent of The multiplex equipment ules. media. All multiplexequipmentin theFDM hierarchyusedSSBmodulation'Thus,every 4 kHz of bandwidth.The lowestlevel building voicecircuit requiredapproximately block in thehierarchyis a channelgroupconsistingof 12voicechannels.A channel groupmultiplexusesa totalbandwidthof 48 kHz. FigureI.20 showsa blockdiagram of an A5 channelgroupmultiplexer,themostcommonA-typechannelbankusedfor carriersgeneratel2 first'level multiplexing.Twelve modulatorsusing 12 $eparate filteredto select signalsasindicated.Eachchannelis thenbandpass double-sideband signal.Thecompositemultiplexsigof eachdouble-sideband only thelowersideband thefilter outputs.Demultiplexequipinentin a receivnal is producedby superposing in reverseorder. ing terminalusesthe samebasicprocessing filter not only removesthe uppersidebandbut Noticethat a sidebandseparation alsorestrictsthe bandwidthof the retainedsigrpl: the lower sideband.Thesefilters thereforerepresenteda basicpoint in the analogtelephonenetworkthat definedthe bandwidthof a voicecircuit. SinceFDM was u$edon all long-haulanalogcircuits, lndividualvoice channelinputs I

rt-r1 1 2

EL

0 4

FI

104 108

l(x

l2 Chennel multiplex output

Fl-r68

60

60 64 BdndpEEE fitter$

Figure 1.20 A5 channelbank multiplexer.

28


long-distance connectionsprovided somewhat less than 4 kHz of bandwidth. (The loading coils discussedpreviously also produce similar bandwidth limitations into a voice circuit.) As indicated in Table 1. 5, the secondlevel of the FDM hierarchy is a 60-channel multiplex refened to as a supergroup.Figure I.2l shows the basic implementation of an LMX group bank that multiplexes five flust-levelchannelgroups.The resulting 60channelmultiplex output is identical to that obtained when the channelsare individually translatedinto 4-kHz bands from 312 to 552 kHz. Direct translation requires 60 separateSSB systemswith 60 distinct cariers. The LMX group bank, however, uses only five SSB systemsplus five lower level modules. Thus two-stage multiplexing, as implied by the LMX group bank, requiresmore total equipmentbut achieveseconomy through the use of common building blocks. (Engineersat AT&T Laboratories also developeda multiplexer that generateda supergroupdirectly [7].) Because a second-level multiplexer packs individual first-level signals together without guard bands,the carier frequenciesand bandpassfilters in the LMX group bank must be maintained with high accurqpy.Higher level multiplexers do not pack the lower level signalsas close together.Notice that a ma$tergroup, for example,does not provide one voice channelfor every 4 kHz of bandwidth. It is not practical to maintain the tight spacingbetweenthe wider bandwidth signalsat higher frequencies.Furthermore, higher level multiplex signals include pilot tone$ to monitor transmission link quality and aid in carder recovery.

',.2.7 WidebandTranemission Medla Wire pairs within multipair cableshave usablebandwidths that rangefrom a little under I MHz up to about 4 MHz dependingon the length, the wire gauge,and the type of insulation u$edon the pairs. Multiplexed transmissionson thesewire pairs consequently have capacitiesthat rangefrom ?4 channels(on analogN3 or digital Tl carrier systems)up to 96 channels(on obsoletedigital T2 carrier systems).In contrast,an ana12 Channel group inputB

t-]

60

n l r

380 408

s28

r-1

t-]

50* s52

504 552 Bandpeac fltterE

Figure 1.21 LMX groupbankmultiplexer.

NETWORK HIERARCHY 29 1.2 THEANALOG

log L5E coaxialcable$ystemwasdeveloped to carry 13,200voicechannels.Optical fiber systemshavebeendevelopedthatcarryin excessof 100,000voicechannelson a singlewavelength.Becauseopticaltransmittersflight-emittingdiodes(LEDs) and laserslfunctionbestin a pulsedmodeof operation,fiber systemsareprimarilydigital in Chapter8. Thissecsystemsaredescribed in nature.Digital fiber optictransmission coaxialcableandpoint-to-pointmicrowaveradio systems. tion discusses Coaxial Cable of Coaxialcablesystemswereusedpredominantlyto $atisfylong-haulrequirements the toll network.The first commercial$y$temwasinstalledin l94l for hansmission of 480 voicecircuitsover a 200-milestretchbetweenMinneapolis,Minnesota,and repeateramplifierswere inStevensPoints,Wisconsin[7]. To combatattenuation, stalledat 5.5-mileintervals.Consideringthe maximumcapacityof l2 voicecircuits on openwire or cableat thetime,theintroductionof "coax" wasa significantdevelby (l) opment.After thefirst installationcoaxialcablecapacitywassteadilyincreased (2) decreasing the disattenuation, usinglargerdiametercables(0.375in.) to r"educe and(3) improvingthenoisefigure,linearity,andbandwidth tancebetweenrepeaters, of the repeateramplifiers. A sumnraryof theanalogcoaxialcablesystemsusedin theBell Systemis provided in Table1.6.Noticethateachsystemreservedonepair of tubesassparesin theevent sinceeachtubecanieda high volume of failure,a particularlyimportantconsideration lower attenuation,lower have wider bandwidths, Because optical fibers of traffic. systems are obsolete. and lower cost, coaxial maintenance, Mlcrowave Radlo Much of the impetu.sfor terreshialmicrowaveradio systemscarnefrom the needto traffic indistributetelevisionsignalsnationwide.As the volumeof long-distance radiosystemsalsobecamethemosteconomicalmeansof diskibutingvoice creased, network.Beginningin 1948,whenthe first systemwas circuitsin the long-distance installedbetweenNew York and Boston,the numberof microwaveradio sy$tems grewto supply607oof the voicecircuitmilesin theU.S.toll networkin 1980[71.It wasjust a few yearsafterthat that opticalfibersbeganto takeoverfor high-density interofficeroutesandeventuallyfor thenetworkasa whole. TABLE1.6 CoaxlalCableSysteme ln th€BellNetwork Pairsper System Designation Systeme SignalDesignation L1 L3 L4 L5 the

3/6 5/6 9/10 10/11

Ma$iergroup Mux Mastergroup Jumbogroup Mux Jumbogroup

numberof pairs are shown a$ workingôtal.

FleDeater Spacing Capacityper (miles) Pair I 4 2 1

600 1,860 3,600 10,800

Total Capacity 1,800 9,300 32,400 108,000

30


Microwaveradiosystemsrequireline-of-sightnansmission with repeaterspacings typically 26 miles apart.The majoradvantage of radio systemsis that a continuous right-of-wayis not required-only smallplotsof land spaced20-30 milesapartfor towersandequipmentshelters. A majorcostof guidedtransmission, for example,wire pairs,coax,or fiber,is theright-of-waycosts.In manymetropolitanareas,microwave routesbecameheavilycongested andcouldnot be expanded with allocatedcommoncarrierfrequencybands.In thesesituations,findinga right-of-wayfor a coaxor eventually an optical fiber systemwas sometime$the only choice for high-capacity transmission.The frequencybandsallocatedby the FCC for common-carrier usein the united statesarelistedin TableL7. of thesebands,4 and6 GHzhavebeenthemostpopular. The2-GHzbandhasnotbeenusedextensivelybecause therelativelynarrowallocated channelbandwidthsdo not permitimplementation of economicalnumbersof voice circuits.The basicdrawbackof the lI-GHz bandis its vulnerabilityto rain attenuation.However,ll-GHz radioshavebeenusedin some$fiort-haulapplications,, The microwaveradio systemsof the analogBelt netwoik arelistedin tabte L8. Noticethateachradiosystemis designed to carryoneof themultiplexhierarchies describedpreviously.All of theseradiosexceprrheAR-6A usedlow-indexfrequency modulation(FM) of the signalgenerated by the FDM multiplexerequipment.Thus, theFM radioshansmitthessB FDM signalasa baseband signalwith a bandwidthas indicatedin Table 1.5. FM modulationwas chosento permit the use of nonlinear poweramplifiersin thetransmitters andto takeadvantage of FM signal-to-noise ratio perforrnance. Examinationof Tables1.7and1.8indicatesthat 13.3kHz of bandwidthis utilized pervoicecircuitin TD-2 radiosand 14.3kHz in TH-3 radios.Thus.theuseof FM introduceda significantincrease in the4 kHz bandwidthof theindividualSSBvoicecircuit. In conrrast,the ssB AR-6A radio introducedin lggl provided6000 voice circuitsin the 30-MHzchannelsat 6 GHz. Sincea largenumberof voicecircuitsarecarriedby eachradiochannel,microwave system$usuallyincludeextraequipmentandextrachannelsto maintainservicedespiteoutagesthatmayresultfrom anyof thefollowing: 1. Atmospheric-induced multipathfading 2. Equipmentfailures 3. Maintenance On someroutes,the most frequentsourceof outagein a microwave radio systemarises from multipath fading. Figure 1.22depicts a simple model of a multipath environment arising as a result of atmospheric refraction. As indicated, the model involves two ray$: a primary ray and a delayed secondaryray. Ifthe secondaryray anives out of phasewith-respectto the primary ray, the primary signal is effectively canceled.The *Micruwave

bands with wide-bandwidth allocations at high camer frequencies are available for some I ocal distribution services and are discussedin Chapter I l.

1.2 THE ANALOGNETWORKHIERARCHY

31

TAELE1.7 MlcrowaveFrequencieeAllocatedfor Common-CarrlerUss in the UnltedStates TotalBandwidth ChannelBandwidths (MHz) (MHz)

Band(MHz) 2110-2130 2160-2180 3700+200 5925-64?5 1,700 10,700-1

20 20 500 500 1000

3.5 3.5 20 30 40,20

amount of cancelation is dependent on both the magnitude and the phase of the secondary ray. Quite often only nominal amountsof fading occur and can be accommodated by excess signal power in the transmitter, called a fade margin. In some instances,however, the received signal is effectively reduced to zero, which implies that the channel is temporarily out of service.

Frequency Diverelty Fortunately,exceptionallydeepfadesnormally affect only one channel(carrier frequency)at a time. Thus,a backupchannelincludinga sparetransmitteranda spare receivercanbe usedto carrythe traffic of a fadedprimarychannel.Selectionof and switchingto the sparechannelis performedautomaticallywithout a lossof service. This remedyfor multipathfading is referredto asfrequencydiversity.Notice thatfrequencydiversityalsoprovideshardwarebackupfor equipmentfailures' A fully loadedTD-3 radio systemused 12 channels:10 main channelsand 2 or to variouslyas2-for-10,1.Q--by12' backupchannelsfor protection.This is Lefe-rred protection switchl-for-l 10X 2 protectionswitching.Someshort-haulsystemsused ingbecauseit i$ $impleft-dimplement.However,sinceonly half of therequiredbandwidth is actuallycarryingtraffic, systemswith l-for-l protectionwereonly allowed environments. in uncongested TABLE1.8 Bell SystemAnalogMicrowaveFsdios System TD.2 TD-3 TH.1 TH.3 TM.1 TJ TL.1 TL-2 AR-6A

Band(GHz) 4 4 o

6 6 11 11 11

6

VoiceCircuits Application 600-1500 1200 1800 2100 600-900 600 240 600-900 6000

Longhaul Longhaul haul ShorVlong haul ShorUlong haul ShorUlong Shorthaul Shorthaul Shorthaul Longhaul(SSB)

32

BAcKcHoUNDANDTERMtNoLoGy

Figure1.22 Two-raymodelof multipathpropagation. Exceptin maintenance situations,protectionswitchingmustbe automaticin order to maintainservicecontinuity.A typical objectivewasto restoreservicewithin 30 msecto minimizenoticeableeffectsin the message haffic. A morecritical requirement is to re$tore service before the loss of signal is interpretedby somesignaling ( schemesasa circuit disconnect.Inadvertentdisconnects occurif an outagelastsfor morethan 1-2 sec. (

Space Diversity since deepfadesonly occurwhena secondary ray arrivesexactlyout of phasewith respectto a primaryray, it is unlikelythat two pathsof differentlengthsexperience fadingsimultaneously. Figure 1.23depictsa technique,calledspacediversity,using differentpathlengthsto provideprorecrionagainstmultipathfading.As indicated,a singletransmitterirradiatestwo receiveantennas separated by somedistanceon the tower. Althoughthepathlengthdifferencemay be lessthana meter,this differenceis adequate at microwavefrequencies, whichhavewavelengths on theorderof tenthsof meters. Rain is anotheratmospherically basedsourceof microwavefading.As already mentioned,rain attenuationis a concemmosflyin higherfrequencyradios(l I GHz and above).unfortunatelyneitherfrequencydiversity(at the high frequencies) nor spacediversityprovidesanyprotectionagainstrain fades. Satellitee FollowingtheApril 1965launchof the sovietunion's Molniya domesticcommuni* cation$satelliteandthefirst international communications satellite,INTELSATI, the useof satellitesfor internationaltelephone traffic grewphenomenally. The 1g70sand early 1980salsoproducedsignificantuseof satellitesin theUnitedStatesfor televisionprogramdistributionandfor corporatevoiceanddatanetworks.Thefirst domes-

== =:: ;: ::: :::::::::::--;h H(-.-= Figure 1.23 Spacediversity.

1,2 THEANALOG NETWOHK HIEHAFCHY

33

tic systemin North AmericawastheCanadianArik A in 1972followedby Westem Union'sWestarsystemfor U.S.servicein 1974[8]. In onesensea satellitesystemis a microwaveradiosystemwith only onerepeater: thetransponder in outerspace.In fact,somesatellitesystemsusethe same4- and6however, GHzfrequencybandsusedby tenestrialmicrowaveradios.In anothersense, for new services natureof thedownlink offersadditionalopportunities thebroadcast not availablefrom point-to-pointterrestrialsystems.Distributionof networktelevinatureof satsionprogrammingis oneapplicationparticularlysuitedto thebroadcast to receive-onlyhomereceivers etlites.Directbroadcastsatellite(DBS) transmission of DBS systemsaretheEuropeanDVB-T is a primeexample.*Two primaryexamples systemt9l andthe North Americandigital satellite$ystem(DSS)systemdeveloped by HughesElechonicsSystems(HES). with mobile Anotherapplicationthatis well suitedto satellitesis communications (INMARSAT) besystem, for example, stations.The internationalmaritimesatellite with digital 1982 and was augmented gan $upporringthe maritime industry in telephoneservicein 1989. Skyphone[10] for aeronautical delayof the is theinherentpropagation Onedrawbackto satellitecommunications (not includingground long transmissionpath.For a stationary$atellite,this delay links) is 250msecup anddown.A completecircuitwith satellitelinks for bothdirecround-trippropagation tionsof travelthereforeimpliesgreaterthana one-half-second but not prohibitime.Delaysof this magnitudearenoticeablein a voiceconversation by pairingeach delayscanbealleviatedsomewhat tive.Theeffectsof thepropagation circuitin theoppositedirection.Thus,theroundsatellitecircuitwith a ground-based trip delayinvolvesonly onesatellitelink. As is the casewith point-to-pointmicrowaveradio,fiber optic transmissionhas displacedthe useof satellitesfor high-density,domesticandinternationaltelephone Hencethe useof satellitesis primarily directedat thin-routetelecommunications. andbroadcastapplications. somemobilesy$tems, phoneanddataapplication$, 1.2.8 Tranemissionlmpairments Oneof the mostdiff,rcultaspectsof designingan analogtelephonenetworkis deterwithin the impairmentsto individualsubsystems mininghow to allocatetransmission for end-to-end by listeners,certainobjectives network.Using subjectiveevaluations manner[11].Afin a relativelystraightforward qualitywereestablished fransmission ter temperingthe goalswith economicfeasibility,theend-to-endobjectiveswereestablished.However,consideringthe myriad of equipmenttypes and connection in theolderanalognetwork,designingtheindividualnetworkelements combinations to meettheseobjectivesin all caseswa$a complexproblem.A greatdealof creditis a nationwideanalognetworkwith thelevel duetheBell Systemfor havingcleveloped it had. performance of consistent *A

DBS system is designed to use rcceiver antenflas that are about I rn in diameter. The older home satellite receiver systems that wefe common in the United States used f- to 5-m anteflnasto rcceive (intercept) commercial program distribution.

34


The major factors to be consideredin establishing transmissionobjectives for an analog network are signal attenuation,noise, interference,crosstalk, distortion, echoes, singing, and various modulation- and carrier-relatedimperfections.

Signal Attenuatlon Subjectivelistening testshave shown that the prefened acoustic-to-acousticloss [12] in a telephoneconnectionshouldbe in the neighborhoodof I dB. A study oflocal telephone connections[l3] demonsftatesthat the typical local call had only 0.6 dB more loss than ideal. Surveys of the toll network [4] indicated that the averageanalog toll connection had an additional 6.7 dB of loss. This same survey also showed that the standarddeviation of loss in toll connectionswas 4 dB (most of which was attributable to the local loops). Since trunks within the toll network used amplifiers to offset transmissionlosses, it would have been straightfbrward to design thesetrunks with zero-decibelnominal insertion loss. However, as discusrtedlater, echo and singing considerationsdictated a need for ceftain minimum levels of net loss in most analog trunk circuits.

hterterence Noise and interferenceare both characterizedas unwanted electrical energy fluctuating in an unpredictable manner. Interference is usually more structured than noise since it arisesas unwanted coupling from just a few signals in the network. If the interf'erenceis intelligible, or nearly so, it is referred to as crosstalk.*some of the major sourcesofcrosstalk are coupling betweenwire pairs in a cable,inadequatefiltering or carrier offsets in older FDM equipment, and the effects of nonlinear componentson FDM signals. Crosstalk,particularly if intelligible, is one of the most disturbing and undesirable imperfections that can occur in a telephonenetwork. Crosstalk in analog systemsis particularly difficult to control sincevoice signal power levels vary considerably(i.e., acro$sa dynamic range of40 dB). The absolutelevel ofcrosstalk energy from a highlevel signalmustbe small comparedto a desiredlowJevel signal.In fact,crosstalkis most noticeableduring speechpauses,when the power level ofthe desiredsignalis zero. Two basic forms of crosstalkof concernto telecommunicationsengineersare nearend crosstalk(NEXT) andlitr-end crosstalfr(FEXT). Near-endcrosstalkrefersto coupling from a transmitter into a receiver at a common location. often this form of crosstalk is most troublesomebecauseof a large difference in power levels between the ffansmitted and received signals. Far-end crosstalk refers to unwanted coupling into a received signal from a transmitter at a dirrtantlocation. Both forms of crosstalk are illustrated in Figure 1.24.

Nolse The most common form of noise iuralyzedin communicationssystemsis white noise with a caussian (normal) di$tribution of amplitude values. This type of noise is both -Crosstalk

is also used to characterizesignallike interferencesin nonvoice networks. For cxample, crosstalk in a data circuit would refcr to an interfering signal being coupled in from another similar data circuit.

NETWOFK HIEHAHCHY I.2 THEANALOG

35

Flgure I.24 Near-endandfar-endcrosstalk, easy to analyze and easy to find since it arisesas thermal noise in all electrical components.Battery systemsused to power customerloops are also a sourceof this type of noise. White noise is truly random in the sensethat a sample at any in$tflntin time is completely unconelated to a sample taken at any other instant in time. The other most cornmon forms of noise in the telephonenetwork are impulse noise and quantization noise in digital voice terminals (Chapter 3). Impulse noise can occur from switching transientsin older electromechanicalswitching offices or ftom rotary dial telephones.Step-by-stepswitcheswere the most frequent culprits. More modern electromechanical switches that use glass-encapsulatedreed relays for crosspointsproduce much less noise. Whereaswhite noise is usually quantified in terms of average power, impulse noise is usually measuredin tems of so many impulses per second. Impulse noise is usually of less concernto voice quality than backgroundwhite noise. However, impulse noise tendsto be the greatestconcernin a datacommunicationscircuit. The power level of any disturbing signal, noise or interf'erence,is easily measured with a root-mean-square(rms) voltmeter. However, disturbancesat some frequencies within the passbandof a voice signal are subjectively more annoying than others. Thus. more useful measurementsof noise or interferencepower in a speechnetwork take into accountthe subjectiveeffects of the noise as well as the power level. The two morttcornmon such measurementsin telephony use a C-messageweighting curve and a psophometric weighting curve, as shown in Figure 1.25. These curves essentially representfilters that weight the frequency spectrum of noise according to its annoyanceeff'ectto a listener. C-messageweighting represent$the responseof the 500-type telephoneset. As f'ar as perceived voice quality is concerned,only the noise that gets passedby the telephonesetis imporlant. Notice that disturbancesbetween I and 2 kHz are most. perceptible. C-message weighting is used in North America while psophometricweighting is the European (ITU-T) standard. A standardnoise referenceusedby telephoneengineersis I pW, which is l0-12 W, or -90 dBm (dBm is power in decibelsrelative to a milliwatt). Noise measuredrelative to this reference is expressedas so many decibels above the reference (dBrn). Thus, a noise level of 30 dBrn correspondsto -60 dBm, or 10=eW of power' If the readingsare made using C-messageweighting, the power level is expressedby the abbreviation dBrnC. Similarly, psophometrically weighted picowatts are expressedby

36

BACKGROUND ANDTEBMINOLOGY

0

c !

-10

tl fl d

CE

-20

250

500

1000

zilro

40(n

(Hz) Frequency Figure 1.25 C-messageand psophometic weighting.

the abbreviation pWp. The relationshipsbetween various noise power measurements are given in Table L9. The quality of an analog voice circuit is usually not specified in terms of the classical signal-to-noiseratio. The reasonis that relatively low levels of noise or interference are noticeable during pau$e$in speech,when there is no signal. on the other hand, high levels of noise can occur during speechand be unnoticeable.Thus, absolute levels of noise are more relevant than signal-to-noiseratios for speciffing voice quality.- The objectives for maximum noise levels in the analog AT&T network were 2g dBrnC for connectionsup to 60 miles in length and 34 dBrnC for 1000-mile circuits.t

Die|p.rtlon In a previoussectionsignalattenuations wereconsidered with thetacitassumption that a received waveform was identical in shape to a source waveform but merely scaleddown in amplitude. Actually, a received waveform generally contains certain distortions not attributable to external disturbancessuch as noise and interferencebut that can be attributed to internal characteristics of the channel itself. In contrast to noise and interference,distortion is deterministic; it is repeatedevery time the same signal is sentthrough the samepath in the network. Thus distortions can be controlled or compensatedfor once the nature of the distortion is understood. There are many different type$ and sources of distortion within the telephone network. The telephonecompaniesminimized thosetypesof distortion that most affected the subjectivequality of speech.Later on they also becameconcemedwith distortion effects on data transmission. Some distortions arise from nonlinearities in the network, such as carbon microphones, saturating voice-frequency amplifiers, and un*It

is a "orn rron pracrice in the industry to specify the quality of a voice circuit in terms of a test-tone-to-noise ratio. However, the test tone must be at a specific power level so the ratio, in fact, absolutenoise power. *specifies 'These noise power values iue related to a particular point in a circuit, called ^ zero-transmission-Ievel paizt, discussedlater,

1.2 THE ANALOGNETWOHKHIERARCHY

37

betweenVariousNoise TABLE1.9 Relatlonships Measurement8 To Convert From dBm dBm dBm dBrn dBc pW

3 kHzllat 3 kHzflat 3 kHz flat 3 kHzflat

To dBrn dBrnC dBp dBrnC dBp pwp

Add90 dB AddBBdB Add87.5dB Subtract2 dB Subtract 0.5dB Multiply by 0.562

matchedcompandors(Chapter3). Otherdistottionsarelinearin natureandareusually characterizedin the frequencydomainaseither amplitudedistortionor phasedistortion. in thevoicespectrum somefrequencies Amplitudedistortionrefersto attenuating onemeansof eliminatearlierrepresent morethanothers.TheIoadingcoilsdiscussed wire pairs.Amplitudedistortion ing amplitudedistoltionon long voice-frequency Ideallythese filtersin FDM equipment. couldalsobeintroducedby spectrum-limiting up to 4 kHz andrejectall othfilters shoulduniformlypassall voicebandfrequencies "roll-offs" beers.Practicaldesigns,however,imply theneedfor gradualattenuation respon$e ginningat about3 kHz. Figure1.26showstheattenuation-versus-frequency of a typicalanalogtoll connectionof thepast. medium' of thetransmission Phasedistortionis relatedto thedelaycharacteristics in a signalunisy$temshoulddelayall frequencycomponents Ideallya transmission

c E

a 3 c

o

*E

FreguencylkHzf

Figure 1.26 Insertionlossversusfrequencyoftypical toll connection.

38

BAcKcRoUNDANDTERMtNoLoGy

formly so the proper phaserelationshipsexist at the receiving terminal. If individuat frequency component$experiencediffering delays,the time-domain representationat the output becomesdistorted becausesuperpositionof the frequency terms is altered at the output. For reasonsnot discussedherethe delay of an individual frequencycomponent is usually referredto as its envelopedelay. For a good explanationofenvelope delay seereference[5]. Uniform envelopedelay relatesto a phasere$ponsettrat is directly proportional to frequency. Thus systemswith uniform envelope delay are also referred to as linear phasesystems.Any deviation from a linear phasecharacteristicis referred to as phase distortion. The perceptualeffects ofphase distortion to a voice signal are small. Thus only minimal attention need be given to the phaserespon$eof a voice network. The phaserespon$eand correspondingenvelope delay providecl by a typical analog toll connection is shown in Figure 1.27. In addition to the distortionsjust mentioned, analog carrier systemssometimesintroduced other frequency-relateddistortions such as frequency offsets, jitter, phase hits, and signal dropouts.The effects ofthese imperfections and phasedistortion were adequately controlled for voice traffic but presented difficulties for high-rate voiceband data traffic.

Echoee and Slnging Echoesandsingingbothoccurasa resultof transmittedsignalsbeingcoupledinto a returnpathandfedbackto therespective sources. Themostcommoncauseof thecoupling is an impedance mismatchat a four-wire-to-two-wire hybrid.As shownin Figure 1.28,mismatches causesignalsin theincomingbranchof a four-wirecircuitto get coupledinto the outgoingbranchand retum to the source.In the older networkwith two-wireanalogendoffice switches,it wasimpracticalto providegoodimpedance matchesat thispointfor all possibleconnections sincethetwo-wiresideof thehvbrid

(kHrl Fr€quency Figure 1.27 Envelope delay and phaseresponseof typical toll connection.

HIERARCHY 39 NETWOFK 1.2 THEANALOG

interface. of echosat two-wire-to-four-wire Figure 1.28 Generation could be connectedto many different local loops, eachwith its own characteristicimpedance. "talker echo." [f a second If only one reflection occurs,the situation is refened to as "listener echo" results.When the returning signal is repeatedlycoureflection occurs. pled back into the tbrward path to produce oscillations, singing occurs' Basically, singing resultsifthe loop gain at somefrequency is greaterthan unity' Ifthe loop gain is only slightly less than unity, a near-singing condition causesdamped oscillations. Singing and near-singingconditions have a disturbing effect on both the talker and the listener. Talker echo is usually the most noticeable and ffoublesome. The degree of echo annoyiunceexperiencedby a talker is dependenton both the magnitudeof the returning signal and the amount of delay involved U 6' 171' On short connections the delay is small enough that the echo merely appears to the talker as natural coupling into his ear. In fact, a telephoneis purposely designedto couple some speech energy (called sidetone) into the earpiece. Otherwise, the telephone seems dead to a talker. Near-instantaneousechoesmerely add to the sidetoneand go unnoticed. As the roundtrip delay increases,however, it becomesnecessaryto increasingly attenuatethe echoesto eliminate the annoyanceto a talker' Hence, long-distancecircuits require significant attenuationto minimize echo annoyance'Fortunately, an echo experiencestwice as much attenuationas doesthe forward-propagatingsignal sinceit traversestwice the distance.Intermediate-lengthconnections are typically designed with 2-6 dB of path attenuationdependingon the delay. All transmissionlinks within the Bell System were designedwith specific amounts of net loss called via net loss (VNL) that dependedon the length of the link and the position in the hierarchy [17]. In general,the VNL network design establishedend-to-endattenuationin proportion to the length of the circuit Connectionsthat producemore than 45 msecof roundtrip delay (representing1800 miles of wire) require more attenuation for echo control than can be tolerated in the forward path. In thesecasesone of two types of deviceswas used to control the echo: an echo suppressoror an echo canceller. As shown in Figure 1.29, an echo suppressoroperateson four-wire circuits by measuringthe speechpower in each leg and inserting a large amount of loss (35 dB typically) in the oppositeleg when the power level exceedsa threshold.Thus, a returning echo is essentially blocked by the high level of attenuation.Notice that an echo riuppressorconvertsa full-duplex circuit into a half-duplex circuit with energy sensing being the meansof turning the Iine around. for voice circuits,was that they might clip beOne drawback of echo suPPressors If a party at one end of a connectionbegins talksegments. portions ginning of speech

40

BACKGROUND ANDTEBMINOLOGY

Figure 1.29 Echosuppressor. ing at the tail end of the other parfy's speech,the echo suppres$ordoes not have time to reversedirections. Echo suppressorswith good performanceare able to reversedirections in 2-5 msec [16]. For the fastest possible releaseof backward attenuation, split-echo suppressorsare necessary.A split-echo suppressoris one that separatesthe echo control of each direction so the loss insertion of each direction is closest to the point of echo occurrence. A second,and much preferred,form of echo contror is echo cancellation [lg, I ga]. As shown in Figure 1.30, an echo canceller operatesby simulating the echo path to subtract a properly delayed and attenuated copy of a transmined signal from the receive signal to remove (cancel) echo components.Thus echo cancellation requires training to determine how much delay and attenuation are neededto simulate the echo characteristicsof the circuit. Notice in Figure 1.30 that echoesare canceledclose to the source so that delays in the echo canceller are minimized. The important feature of an echo cancelleris that it maintains a full-duplex circuit so clipping doesnot occur. Satellite circuits with greaterthan 500 msec of roundtrip delay required echo cancellers for acceptableperformance.Becausethe cost of digital signal processing(DSp) technology has dropped so dramatically, echo cancellersare now usedin any situation requiring echo control. Full-duplex voiceband modems (v.32 and rater) incorporate echo cancellers directly in their receive circuitry. Thus, network-basedecho cancellers are unnecessary*-and sometimes undesirablebecausetandem echo cancelling may not work properly if two echo cancellers do not cooperatein the haining process. Networkbasedecho cancellerscan therefore be disabled by a modulated 2lfi)-Hz tone (echo suppressor$were also disabled with a 2100-Hz tone) transmitted at the start of a connection[19]. Another method of echo control involves impedancematching of the hybrids to reduce the magnitude of the echo. some electronic hybrids provide dynamic balancing to automatically eliminate or reduceechoes.In fact, a coflrmon way of implementing the impedancematching is to build an echo cancellerwith near-zerodelay. Thesecircuits eliminate, or greatly reduce,echoesoccurring at the associatedhybrid but do not eliminate echoesthat may occur elsewherein the network. For a detailed discussion "If

ttre terminal (e.g., modem) echo canceller has insufhcient delay buffering for very long echoes, the network echo canceller at the far end of a corurection may be necessary.

-Tl I L F 1

-

1.2 THE ANALOGNETWORKHIEFARCHY

41

Figure 1.30 Echocanceller' of all types of echo control, seereference[20]. In general,the proceduresusedto control echoesalso control singing. On some fairly short connections,however, no echo conffol is necessary,and singing may becomea problem'

1.2.9 PowerLevels As indicated in previous paragraphs,voice signal power in a long-distanceconnection needsto be rigidly controlled. The tlelivereclsignal power must be high enough to be clearly perceivedbut, at the sametime, not be so strong that circuit instabilities such as echo and singing result' To maintain rigid control on the end-to-endpower level of a circuit involving a variety of transmissionsy$tems,telephonecompaniesnecessarilycontrolled the net attenuation and amplification of each analog transmissionsystem.These systemswere designedfor a certain amount of net loss (VFil-) as describedpreviously. To administer the net loss of transmissionlinks, the fransmissionlevels of various points in a transmissionsyrttemare specifiedin terms of a referencepoint- ITU-T recommendations call this point the zero-relative-level point and the North American term is a zero-transmission-levelpoint (0-TLP). The referencepoint may not exist as an accessiblepoint but has long been consideredto be at the sendingend terminal of a two-wire switch. In North America the sendingend of a four-wire switch is defined to be a *z-dB TLP. Hence, a 0-dB TLP is only a hypothetical point on a four*wire circuit. Nevertheless,it is useful in relating the signal level at one point in the circuit to the signal level at anotherpoint in the circuit. If a O-dBm (l-mW) test tone is applied at a O-TLP, the power level at any other point in the circuit is
42

BACKGRoUNDANDTERMINOLoGY

-zdB TLP

_l3dB TLP

_4d8 TLP

Figure1.31 TLp valuesfor Example l.l. noise is commonly expressedin units of picowattspsophometricallyweighted (dBm0por pWpO). Example1.1. using Figure1.31for reference, determineeachof thefollowing:(a) the signalpowerto be appliedat point B ro determineif pointsA and c are at the properlevels;(b) the amountof gain (loss)a signalexperiences whenpropagating from A to c; and(c) theamountof noisethatwouldbe measured at c if zj dbrnc of absolutenoiseis measured at B andno additionatnoiseoccurson theB-to-c link. solution. (a) Because pointB is a -l3dB TLp, thepropertesttonelevelis -13 dBm (0.05mw). (b) BecausetheTLP valuesdropby 2 dBm,thereis 2 dB netlossfrom A to c. (c) An absolutemeasuremenr of 27 dBmc at B is 40 dBmc0. This is also40 dBmc0 at C. Theabsolute noisepowermeasured atC wouldbe40 - 4 =36dBmC. To put signalandnoisepowersin perspective, a surveyof voicesignalsin theBell system[21] indicatedrheaveragespeechsignalhas-16 dBm0of power.hr comparison,the noisepowerobjectivefor a 1000-mileanalogcircuit was 34 dBrnc0 (-56 dBmCO).Thus,thenoiseis 40 dB belowthe signalpower. 1.2.10 $ignallng The signalingfunctionsof a telephonenetworkrefer to the meansfor transferring network-related controlinformationbetweenthevariousterminals,switchingnoaes, andusersof the network.Therearetwo basicaspectsof any signalingsystem:specially encodedelectricalwaveforms(signals)andhow thesewaveformsshouldbeinterpreted. Themostcommoncontrolsignalsto a useraredial tone,ringback,andbusy tone.Thesesignalsarerefered to ascall progres$tonesandmay neverchange.The signalingprocedures usedinternallyto the networkarenot constrained by userconventionandwereoftenchangedto suit particularcharacteristics of transmission and switchingsystems.As a result,the analognetworkuseda wide varietyof signaling schemes to transfercontrolinformationbetweenswitchingoffices. Signallng Functlone signalingfunctionscanbe broadlycategorized asbelongingto oneof two rype$:supervisoryor informationbearing.Supervisorysignalsconveystatu$or conffolofnetwork elements.Themostobviousexamplesarereque$tfor service(off-hook),ready

43 NETWORKHIERARCHY 1.2 THEANALOG to receive address(dial tone), call alerting (ringing), call termination (on-hook), request for an operator or feature invocation (hook flash), called party ringing (ringback), and network or called party busy tones. Information bearing signals include called pafiy address,calling party address,and toll charges.In addition to call-related signaling functions, switching nodescommunicatebetween themselvesand network control centersto provide certain functions relatedto network management'Networkrelated signals may convey statussuch as maintenancetest signals, all trunks, busy, or equipmentfailures or they may contain information relatedto routing and flow control. Chapter7 discussessomeof the basic considerationsof network managementfor routing and flow control.

In-GhannelSlgnaling

in-channelsignalingor Signalsaretran;mittedwith oneof two basictechniques: "per-trunk referredto as signaling.In-channelsignaling(sometimes common-channel facilitiesor channelfor signalingasfor voice' signaling")usesthe sametransmission in thenextsection,usesonechannelfor all signaling,asdiscussed Common-channel In the pastmostsignalingsystems voice channels. group of of a functions signaling variety. in-channel were the network in thetelephone In-channelsignalingsystemscan be further subdividedinto in-bandand out-ofIn-bandsystemstransmitthesignalinginformationin thesameband bandtechniques. of in-bandsignalingis usedby the voice signal.The main advantage of frequencies arisesfrom a disadvantage The main medium. thatit canbe usedon anytransmission anda user's waveforms the signaling between needto eliminatemutualinterference (SF) single-frequency was speech.The mostprevalentexampleof in-bandsignaling Altrunks' for interoffice signaling,whichuseda 2600-Hztoneasanon-hooksignal disconnects thoughnormalspeechrarelyproducesa pure26ff)-Hzsignal,inadvertent signals.Two othercommonexamplesof haveoccurredasa resultof user-generated by dual-tonemultifrequency(DTMF) signalsfrom in-bandsignalingareaddressing or multifrequency(MF) signalingbetweenswitchingoffices' push-button telephones signalingusesthe samefacilitiesasthe voicechannel but out-of-band ln-channel a band'Thusout-of-bandsignalingrepresents frequency portion ofthe but a different out-of-band of instance coilmon The most voice circuit. a single form of FDM within signalingis dc signalingasusedon mostcustomerloops.With thisform of signaling, theoff-hookconditionby theflow ofdirect currentin the thecentralofficerecognizes by a rotarydial at a line. Othercommonlyusedloop signalsaredial pulsesgenerated centraloffice' All from the voltage rateof 10 pulsesper secondanda 20-Hzringing Thusthereis no generated in speech' thanthose of thesesignalsuselowerfrequencies of out-of-band possibilityofonebeingmistakenfor theother.Themajordisadvantage SSB carriersyssystem'For example, on thetransmission signalingis its dependence with eachvoicechannel.Thus,the associated temsfiher out the verylow frequencies on-hooVoff-hooksignalmustbe convertedto somethinglike SF signalingfor FDM abovethe Out-of-bandsignalingis alsoimplementedwith frequencies transmission. channel' of a limit the 4-kHz but below filters separation of voice cut-off frequency purpose. for this 3825 Hz the use of CCITT recommends

44

BACKGEoUNDANDTERMINoLoGY

Common-ChannelSignaltng Instead of sending signaling information overthesamefacilitiesthatcarrythemessagehaffic(voicesignals), common-channel (ccs) usesa dedicated signaling data

link between the stored-programcontrol elements(computers)of switching systems. Figure 1.32depicts sucha datalink betweentwo switching offices. Notice that the pertrunk signaling equipment associatedwith the trunks has been eliminated. The data link sendsmessage$that identify specific trunks and eventsrelated to the trunks. Thus the type of ccs shown in Figure I.32 is refened to as "channel-associated"commonchannel signaling. The main advantagesof CCS arel 1. only one set of signaling facilities is neededfor each associatedtrunk group insteadof separatefacilities for each individual circuit. 2. A single dedicated confrol channel allows transfer of information such as addressdigits directly between the control elements (computers) of switching offices. In-channel system$, on the other hand, must have the control information switched from the common control equipment of the originating office onto the outgoing channel, and then the receiving office must switch the incoming control information from the voice channel into its common control equipment' The simpler procedurefor transferringinformation directly between switch processorsis one of the main motivations for CCS. 3. since separatechannels are used for voice and control, there is no chance of mutual interference. 4. Since the control channelof a common-channelsystemis inaccessibleto users, a major meansfor fraudulent use of the network is eliminated. 5. connections involving multiple switching offices can be set up more rapidly since forwarding of control information from one office can overlap a circuit set up through the node. With in-channel systemsthe associatedcircuit must first be establishedbefore the control information can be transferredacrossit. The ccs data link can al$o operateat a much higher data rate than common analog network signaling schemes,implying even faster connection setup. 6. The channel used for ccs does not have to be associatedwith any particular trunk group. In fact, the control information can be routed to a centralieed conhol facility where requestsare processedand from which switching offices PerChanrrel Signeling

Flgure 1.32 Trunk-group-associated common-channel signaling.

HIERAHCHY45 NETWORK 1.2 THEANALOG receive their connection control information. Figure 1.33 depicts a CCS network that is disassociated from the mes$age network structure. One advantageofcentralized control is its ability to processrequestswith knowledge of networkwide traffic conditions. Centralized control is also atffactive for . managing multiple switching offices that are too small to warrant call processingfacilities of their own. The transition from in-channel signaling to disassociatedCCS at the network level is analogousto the lower level transition from direct progressive control switches ($tep-by-step) to common control switches. The major disadvantagesof CCS are: 1. Control information pertaining to an establishedcircuit, such as a disconnect, must be relayed from one node to the next in a store-and-forwardfashion' An in-channel disconnect signal, on the other hand, automatically propagates through the network, enabling all nodes involved in the connection to simultaneouslyprocessthe disconnectand releasethe associatedfacilities. 2. If one node in a common-channel system fails to relay the disconnect information properly, facilities downstream from the disconnect will not be released. Thus, a high degree of reliability is required for the common channel-both in term$ of physical facilities (duplication) and in terms of error control for the data link. 3. Since the control information ftaver$esa separatepath from the voice signal, there is no automatic test of the voice circuit as when the voice channel is used to transfer control information. Initial usageof CCS sy$temsincluded special provisions for testing a voice circuit when it was set up' 4. In some instancesall trunks in a group do not logically terminate at the same switch. Figure 1.34 depicts a multiplexed transmissionlink from a PBX to the

_-t

gNetwork

control c6nter

/'

r/ Control circult

Figure 1.33 Dissociatedcommon-channel signaling network'

46

BAcKcRoUNDANDTERMtNoLocy

ccs

(D Channel)

DigitEl Cros*Connect Systom

*""S9# Local Serving Office

Tie Line FBX PBX

Figure 1.34 TDM Iink with multiple destinationsfor the channels.

public network.As indicated,someof the channelscomposea trunk groupto the local CO but otherchannelsmay representtie lines to otherPBXs or foreign exchangecircuitsro othercos. The digital cross-connect system(DCs) is a specializedswitchingsystem(describedin Chapter5) that routesindividual circuitswithin a trunk groupto individuallocations.If CCSis implemented on thetrunkgroup(asin anISDN primaryrateinterface,describedin chapterl1), the signalinginformationfor the laffertwo trunk groupsterminatesat the co. The CO must then forward the informationto the respectivedestinations (assuming thosedestinations areequippedwith CCS).All-in-all rheselanertwo casesareeasierto handleif thesignalinginformationaccompanies themessage channelsasit doesautomaticallywith in-channelsignaling. As a final note,it shouldbe pointedout thatsomesignalingfunctionsoriginating or terminatingwith an end userinherentlyrequirein-channelimplementations. For example,dataterminalswishingto disableechosuppressors or echocancellers in dialup connections needto sendspecialtonesthatgetrecognized by echocontrolelements in the circuit. In similarfashionautomaticoriginatingfacsimileequipmentgenerate I 100-Hztonesto allowautomaticrecognitionof theoriginatingequipmenttypeat the destination.

1.2.11 Analog Intedaces The design,implementation, andmaintenance of any largeandcomplexsystemrequirepaditioningof the systeminto subsystems. Associatedwith eachsubsystem is an interfacethat definesthe inputs and ouSuts independentof internal implementations'Well-established interfacesarea fundamental requirement to maintaincompatibility between old and new equipment.within the u.s. telephonenetwork

HIERAR0HY47 1.2 rHEANALocNETWoRK standardized interfaces are particularly necertsaryto support competition among equipment suppliers and service providers in almost all facets of the network' The ptiniiput analog interfaces used in the periphery of the network are subscriber loops, ioop-start trunks, ground-starttrunks, direct-inward-dial trunks, and E&M trunks.

$ubscriher LooP lntartace of indiin thenetworkinvolvesthetwo-wireconnection interface Themostcommon Because loopintedace. thesubscriber linesto endofficeswitches: vidualtelephone (2500stationsets)andthe electromeof the natureof industrystandardtelephones thisinterfacehasa number chanicalswitchesto whichtheywereoriginallyconnected, circuittechnology. thataredifficult to satisfywith modemintegrated of characteristics The fundamentalcharacteristicsof this interfaceare; l. Battery:Applicationof dc power to the loop (48 V normally)to enabledc signalingandprovidebiascurrentfor carbonmicrophones' Protectionof equipmentandpersonnelfrom lightning 2. OvervoltageProtection: strikesandpowerline inductionor shorts. 3. Ringing:Applicationofa20-HzsignalatS6Vrmsforringerexcitation.Typical cadenceis 2 secon and4 secoff. by flow/no-flowof dc current' Detectionof off-hoolq/on-hook 4. Supervision: or back to thelineto testin eitherdirection-toward thesubscriber 5. Test:Access into the switch. required;twoIn the caseof a digital endoffice,two morefunctionsarenecessarily (and digital-towire-to-four-wireconversion(hybrtil and analog-to-digitalcoding The as BORSCHT' analogdecoding).Takentogether,thesefunctionsarereferredto of digital context of a BORSCHTinterfaceis describedin the basicimplementation switchingin Chapter5. LooP-Start Trunks A loop-start(LS) h'unkis a two-wireconnectionbetweenswitches(usuallybetween pointof view,anLS trunkis identical a centralofficeanda PBX).Fromanoperational loop.Thusan LS interfacein a PBX emulatesa telephoneby closing to a subscriber the loop to drawcurrentfor call originationandby detectingringingvoltagesfor ininformation,thePBX interfacegenerallywaitsfor a few comingcatls.To sendaddress thata dial toneis presentbeforesendingDTMF tonesor genersecondsandassumes PBXsprovidedial tonedetection atingdial pulsesby interruptingloop cunent)tsome canbe aremordAasilyrecognizedandaddressing sofaulty equipmentor connections sentassoonasthe otherendis ready. | 1 Onesignificantdifficulty with two-#ayLS trunksariseswhenbothendsof theline bothendsof theline seizetheline at thesametime (or nearlythesametime)'Because think they areoriginatinga call, the line becomeshung.This situationis commonly referredio asglare.Ifthe PBX detectsa dial tonebeforesendingdigits,it will recog* nize the glareconditionby timing out on the wait for dial tone andcanthengenerate

48

BAcKGRoUNDANDTERMtNoLocy

a disconnect to releasethe glare condition but drop the incoming call. More commonly, the PBX blindly sendsthe addressdigits and connectsthe originating pBX station to the line. Generally, this meansthe incoming call gets connectedto the wrong station. For this reason,LS trunks are normally used only as one-way trunks: either one-way incoming or one-way outgoing.

Ground-StartTrunks The aforementionedproblem with glare on two-way LS trunks can be largely resolved by augmenting the call origination proces$to use ground-start(GS) procedures [22]. When originating a call, the end office applies a ground potential to the tip lead of the tip and ring pair and waits for the pBX to acknowledgethe seizure by drawing loop current. when the PBX originates a call, it first applies ground to the ring teao ana closesthe loop waiting for loop current. (The co doesnot apply battery during an idle stateas it does in ar Ls interface.) The co acknowledgesthe connect r.quest by upplying battery to the tip/ring pair and momentarily applying ground to the tip. (Mai;taining low-noise circuits requiresremoval of all pathsto ground during the connected $tate.) A GS protocol prevents simultaneousseizuresunless the originations occur within a few hundred milliseconds of each other. In contrast, an LS protocol allows multiple seizuresto occur with windows up to 4 sec (the silent interval betweenring bursts).Moreover, a glare condition on GS trunks can be recognizedby the interface equipment so it can be resolved by redirecting the calls to different trunk circuits. Another advantageof GS trunks is the ability of the CO to signal network disconnectsto the PBX (the co remove$battery). witir I-s trunks the network doesnor generally provide disconnectsignaling so the pBX must rely on the end user to hang up. (A situation that often produces hung trunks with data connections.) Furthermore. when the co placesan incoming call that eventually gets abandoned,becauseno one answers,a co immediately signals the abandonmentby removing ground from the tip lead. with LS trunks, abandoneclcalls can be recognized only by the absenceof ring voltage, which can take 6 sec.

Di rect4nward-Dial Trunke Direct-inward-dial (DID)trunksareparticularly simpletwo-wiretrunkinterfaces because theyarealwaysone-way trunts:incoming onlywithrespect to a pBX.As impliedby thename,theyallowa servingco to forwardtheextension numberof

incomingcallssoa PBX canimmediatelyroutethecall to a destination withoutgoing throughan attendant.In contrastto LS andCS trunks,the pBX endof a DID trunk providesbatteryvoltagesothe co cansignalanincomingcall by merelyclosingthe loop to draw current.After the PBX reverse$ batterymomentarily(winks)to signify it is readyto receivedigits,theco eithergenerates dial pulsesor DTMF tonesto send theextensionnumber(two,three,or four digits).Ai'terthedesignated stationanswer$, thePBX reversesbatteryagainto signifythe connectedstateandhotdsthat statefor thedurationofthe call.DID trunksarealsoreferredto as..loopreverse-battery $upervision" trunkswith variationsin the signalingprotocoldependingon the typeof co

122t.

1.2 THEANALocNETwoRKHIERARGHY 49

E&MTrunks

As indicatedin Figure1.35,anE&M trunkis defrnedasanintetfaceto a transmission sy$temandnot a kansmissionsystemitself.The interfacein Figure1.35hasa fourrenrmlead(SG),andanM leadwith an wire voicepath,anE leadwith an associated in this interface(referredto asa type wires (SB). are eight Thus there associaledreturn are defined with as few as four interfaces of E&M types II E&M interface).Other M leadwith earthgroundreturns). and an path, an E lead, (a voice two-wire wires[23] is alwaysconveyedon theE signaling supervision interface, of E&M In anytype signifiesoff-hookby closing pairs). (or The PBX pair voice not on the andM leadsand indicatesoff-hook equipment the transmission while to draw current the M-SB loop conveys equipment How the transmission current. to draw E-SG loop by closingthe protocols are timing A variety of link. transmission of the is a function thesupervision or dial DTMF tones in-channel can be which signaling, of addless definedfor the$tart E andM leads' by momentaryopenson therespective -pulsesgenerated just as an interface,they are often defined is formally signaling Althbugh E&M of betweenPBXs.Because connections pairs wires) as direct (withup of to four used are PBXs the occur when usually pairs, such applications for multiple therequirement locatedwithin a singlebuildingor campuscomplex.Theavailabilityof extemalconsuchaspagingsysfor specialapplications trol leadsallowstheuseof E&M interfaces loudspeaker' on the used to tum lead can be tems.wheretheM 1.2.12 The IntelligentNetwork Thefirst commonchannelsignalingfacilitiesof AT&T wereinstalledin thetoll netand a No. 4 ESs work betweena No. 4A crossbarswitchin Madison,wiSCOnsin, betweensPc data links ccs switchin chicago,Illinois, in 1976[24].The2400-bps costsand to reduce facilities switchingofficesreplacedin-channelSFA4Fsignaling of thetoll more more and times. As connect significantlyspeedup long-distance-call not that network evolved a CCS switcheswereimplementedwith CCScapabilities, platform a also established network but only improvedtheperformanceof theexisting for theintroductionof newfeatures.As indicatedin Figure1.36,theCCSnetworkbe-

Figure 1.35 TyPelI E&M interface.

50

BAcKcRouNoANDTERMtNoLocy

Figure 1.36 CCSnetworMntelligentnetworkplatform. came a disassociatedpacket-switching network that separatedcontrol of the network from the swirching machinesthemselves. The packet-switching nodes of the CCS network are referred to as signal transfer points (srPs). The network control points (NCps) of Figure 1.36 initially represented network databaseserversthat defined how to route calls, verify credit cards,or process specialservicessuch as 800 numbers.The samebasic structureis also installed within the LATAs to extend ccs featuresall the way to the end offices. The original communicationsprotocol used betweenccs entities was ccITT signaling system No. 6 (ccs6). In the early 1980sthis protocol was replaced by ccITT signaling system No. 7 (SS7). Seereference[25] for a good overview of SS7. As the capabilitiesof the NCps are expandedfrom being databaseserversro processingservicerequests,the concept of an intelligent network (IN) 126l begins to take shape.In its ultimate form advancedservicesof the network are executedin the NCps. or service control points (sCPs), as they are called in the IN, and not in the switching machinesthemselves.The switching machinesprovide only the switching fabric; connection commandscome from an SCp. The fundamentalmotivations behind developing the IN are: I ' To enable the deployment of network-basedfeatures such as citywide centrex mentionedpreviously 2. To allow the introduction of new features without having to change the hardware or software of switching vendors

To speedup thedesign/deployment cycleofnew featuresbecause onlv thescp softwareneedsupgrading 4. To allow customizingof servicesby theoperatingcompaniesaccordingto the needsof theircustomers The obvious disadvantageof such centralized control is the vulnerability of the network to a failure in the SS7 network or the SCP. For this reasonbasic serviceis likely to remain with the switching machines.In this case, softwarc in the switching machines recognizes special service situations as software triggers to involve an scp.

51 1.2 THEANALOGNETWOHKHIERARCHY Leaving basic service processingin the switching machinesalso reducesthe load on the SS7 links and the processingspeedin the SCPs.

1.2.13 Dynamic NonhierarchlcalRoutlng The alternaterouting proceduresshownin Figure 1.4for the hierarchicalnetworkof the Bell Systemwerenecessarilysimpleand straightforwardbecauseof the original relianceon mechanicalswitchingmachines.Suchsimplicity,however,leadsto the following limitations: l. Eventhoughthetopologyof thenetworkwouldallowa largenumberof routing altematives,consfiaintson how the routeswere selectedlimited the actual choicesto a relatively small number. 2. The routing patternscould not be changedaccordingto the time of day or networktraffi c Patterns. 3. The proce$sof progressivesetupprecludedretracinga path to a previous switchingnode'andtrying a new,untriedroutewhena previouschoiceled to a nodethatwasblocked' 4. A completedconnectionmay haveinvolveda large numberof intermediate for oneconnection' facilitiesusinga lot ofnetwork resources All of thesedeficienciesresultedfrom usingrelativelysimpledecisionlogic in each nodewith no knowledgeof the networkstatusas a whole.The introductionof SPC switchingmachinesandCCSchangedthesituationsothatmoreefficientroutineprocedureswerepossible.Along theselines,AT&T completedthe deploymentof dyrouting(DNHR) in 1987127,281.Specificfeaturesof DNHR namicnonhierarchical are: Routingtablesin theNCPslist all possibletwolink routesin the orderof cost areminimized. so useof networkresources ) Routingis dynamicto take advantageof traffic noncoincidence' "crankback,"whichallowsselection nodeproduces Blockingat anintermediate of untriedroutesfrom the originatingnode. 4. Routesthat producetoo muchdelayfor the echocancellersin the circuit are l.

excluded. Reference [28] reportsthatDNHR reducedblockingon ThanksgivingDay fuom34flo in 1986toSVoin 1987.After installationof DNHR Mother'sDay traffic sawthreeto four fimesasmanyflir$t-attemptcompletionsasbefore.Theimplementationof DNHR that a network architectureis asmucha function of the softwarein the demonsffates thatcentralizedcontrolof a networkasof thephysicaltopology.It alsodemonstrates networkis morevulnerableto failures.On January15, 1990,an obscurebug in the

52

BA0KGRoUNDANDTERMINoLoGY

CCS software of the No. 4 ESS switchesin AT&T's long-distancenetwork prevented completion of many calls during a 9-hour period [291.Ironically, the bug occurred in software intended to isolate the network from a node failure. The combination of large digital switches,large-bandwidth (fiber optic) rran$mission links, and developmentof the SS7 network has stimulateda transition from hierarchical networks to "flat" networks wherein switching nodesare interconnectedin a functional mesh-switches are either connected directly to each other or through cross-connectfacilities (DCSs and SONET rings describedlater).

1.2.14 CellularRadioTelephoneSystem Prior to 1983 mobile telephoneusersin the United Stateswere restricted to using the servicesof radio common carriers (RCCs) that had limited bandwidth and limited resourcesavailablefor services.The serviceenvironment was necessarilyone of limited availability and extremecongestion.October 1983 marks a significant datefor mobile telephony as the time when commercial cellular mobile telephone service started in chicago, Illinois. As indicated in Figure 1.37, acellular system consistsof a number of radio basestationsconnectedto a (digital) switching office refened to as the mobile telephone switching office (MTSO). As a mobile subscribermove$ from one cell to another, the MTSO automatically switches connectionswith the respectivebase stations to maintain a continuousconnectionwith the public switchednetwork. The basic advantagesof the cellular architectureare: l. The ability to reuse allocatedchannel frequencieswithin a serving area.Using a combination of antennadirectivity in the basestationsand signal attenuation

Figure 1.37 Cellularmobiletelephonetopology.

53 1.2 THEANALOGNETWORKHIEHARCHY from distant cells, the same channel can be reu$edin one of every sevencells

t301.

2. Reducedtransmitpower requirementsfor the mobiles.The power savings advantagefor automobilesis primarily one of reducing the cost of the transmitter.For hand-heldunitsthepowersavingsis importantfor batterysize acceptance anduseofpersonal In fact,widespread andtimebetweenrecharges. power levels and consequently lower transmit radio systemsrequiremuch smallercells[31]. 3. Reducedocclrrencesof multipathpropagation.Shorterdistancesimply less chanceof signalreflectionscausingmultipathsignaldegradation. 4. Expandability.A systemcan be installedwith comparativelylarge cells to grow andrevenueis being minimizestart-upcosts.After servicerequirements received,the capacityof the systemcan be expandedby subdividingthe cells. congested if thecellsactuallyhavesignificantoverlappingcoverage, 5. Reliability.Because are made. until repairs provide service cells can onecell fails, neighboring The FCC in the United Stateshas defined 728 mobile service areasreferred to as cellular geographicservice areas(CGSAs). Each CGSA is allocated 832 radio channels, which are equally divided between two competing service providers; one a wireline carrier and the other a nonwireline carier. The wireline carrier provides local telephone servicein the areabut competeswith the nonwireline canier (RCC) for cellular service. To ensure effective competition, the wireline carrier must not use facilities that are sharedwith local telephoneservice.Specifically, the MTSOs must be separate from the local switching offices. Thus, both types of carriers must backhaul all traffic to their respectiveMTSOs. The service providers can have more than one MTSO in an areabut cannot interconnectthem by the switched public network. They are typically interconnectedby leasedprivate lines (fiber) or digital microwave. Cellular networks have grown to cover large areas interconnectedby dedicated long-distance facilities that allow $omecafriers to offer free long-distancecalling when using a cellular phone. When a mobile unit first activatesitself, it scansthe channelsto determine which idle channelofapredefined setofcontrol channelshasthe strongestsignal. Using that channel the unit registers with the system to identify itself and place calls- After the initialization proce$$,the network continually monitors signals from the mobile and conffols it to switch channelswhen necessary.To complete calls to a mobile, the cellular network pagesfor the clesignatedsubscriberbeginning with its home cell (unless the cellular control centeralreadyknows where the subscriberis located).When a subscriber crossesan MTSO boundary, in addition to a cell boundary, a common-channel signaling network is used to transfer the call to the new cell and the new MTSO. The connection to the public network is unchanged.Thus, the original MTSO then becomes a tandem node to the public network' Examples of analog cellular system$are Advanced Mobile Phone service (AMPS) Total Access [30, 32], developedin the United Statesand deployedin North America;

54

BACKGRoUNDANDTERMINoLoGY

communications system (TAcs), developedin the united Kingdom and deployed in several European countries; and the Nordic Mobile Telephone (NMT) sy$tem, deployed in Scandinavia.As discussedin Chapter 9, digital cellular systemshave been developedto either replace the analog sy$temsor provide service altematives. 1.2.15

Voiceband DataTransmisslon

The primary concern of telephonecompanies,voice service,is sometimesreferred to as Pors: plain old telephoneservice.In addition to pors, the telephonecompanies also provide a number of special servicessuch as program distribution of radio and television, telephoto, teletype, facsimile, and data transmission.In most cases,these servicesare supportedby adaptingthe sourcesto use the ubiquitous voice-gradetelephone channel. An exception is television distribution, which was an initial stimulus for the nationwide microwave radio network. To a certain extent, wideband FDM voice signals were designedto conform to the network TV distribution facilities. Nationwide TV distriburion is now primarily by sarellitesand fiber. one significant aspectofa voicebandtelephonechannelin regardsto nonvoice applications is the absenceof dc and low-frequency transmission.Although the metallic circuit provided in customerloops passesfrequenciesdown to and including dc, most of the rest of the network does not. The equipmentsthat are primarily responsiblefor blocking the low frequenciesare ffansformers in two-wire to four-wire hybrids, old FDM separationfilters, and digital voice encodersthat purposely block low frequencies to avoid 60-Hz hum (that arisesfrom power line interference).Becausebaseband signalsfrom facsimile and most data applicationshave frequency content down to dc, thesesignals must be modulated for voicebandfransmission. A second consideration when using the telephone network for data is the bandwidth restriction of approximarely 3 kHz originally imposed by FDM separarionfilters, loading coils, and more recently band{imiting filters in digital voice terminals (Chapter3). The main implication of a resfficted bandwidth is a limitation on the signaling rate, or baud rate, which in turn directly relatesto the datarate. A common signaling rate is 2400 symbols per second using carriers between 1700 and 1g00 Hz. symbol ratesof 4800 symbols per secondhave arsobern usedin lower sidebandmodems with a carrier at 2850 Hz. As mentioned, user acceptanceof voice quality does not require stringent conhol of the phaseresponse(envelopedelay) of the channel.High-speed data transmission, however, requires comparatively tight toleranceon the phaseresponseto prevent intersymbol interference.Thus analog voice channelsused for high-speeddata transmission sometimesrequired special treatrnent.The special treatunentwas originally provided at additional monthly rates from the telephonecompaniesin the form of ctype conditioning for leased lines. C-type conditioning was available in several different gradesthat provided various amounts of control of both phase and amplitude distortion. As digital signal processingtechnology becamecommercially viable, the needfor signal conditioning becamelessnecessarybecausemodem equalizersaccomplished the samefunction.

HIEHARCHY55 NETWORK 1.2 THEANALOG Most medium-rate (synchronous)modems include fixed equalizationcircuitry designed to compensatefor the phasedistortion in a typical connection.For higher data rates,automatically adjustableequalizationis needed.An automaticequalizerfust undergoesa training sequencein which the characteristicsof the transmissionchannel are determinedby measuringthe responseto known test signals.Then, equalizing circuitry in the receiver of each modem is adjusted (and continually readjustedduring hansmission) to provide compensationfor the amplitude and phasedistottions in the channel. High-speed transmissionover dial-up lines requires automatic equalization in the modems since the channel characterjsticschangewith each connection. Another form of conditioning, referredto as a D-type conditioning, provided lower levels of noise and harmonic distortion than normally provided in leasedlines, even with C-type conditioning. D-type conditioning usually did not involve special treatment of any particular line. Instead, the telephone company tested a number of different circuits until one with suitablequality was found. Sometimesthis meant avoiding a cable that included pairs from a noisy switching office, Unlike voice, which is reasonably tolerant of impulse noise, a data circuit is more susceptibleto impulse noise than to the normal background (white) noise. The elimination of older equipment (e.g., step-by-stepswitches)has eliminated much of the impulse noi$e in the network. As more and more of the public network becamedigitized, fewer and fewer facilities were analog and, hence,fewer analog impairments were likely to occur in a connecrion.By the early 1990svirtually all of the internal portions of the public telephone network had been digitized. The only remaining $egmentsof a connection that were analog were the subscriberloops and the interfacesin the associatedend offices. The principal source of impairment for voiceband modems then became the analog-todigiral (A/D) conversionequipmentin the line interfaces.Recognition of this situation completely eliminated the need for any type of conditioning on leasedlines (which were often digital and not analog) and allowed for the developmentof V.90 modems describedin Chapter I l. An important considerationfor data transmissionover long-distancecircuits in the analog network was the effect of echo suppressors.As mentioned in Section 1.2'8' an echo suppressorblocked the signal in a return path when the correspondingforward path of a four-wire circuit was active. Thus, operative echo suppressorseffectively precluded a full-duplex operation. Even in a half-duplex mode of operation the echo suppressor$might require 100 msec of deactivation time to reverse the direction of propagation.For theserea$onsthe common carriersprovided a meansof disabling the echo suppressorsby using an in-channel control signal from the data terminals. Echo suppressorswere disabledby tran$mitting a pure tone between 2010 and 2240Hzfor 400 msec. The echo suppressorsremained disabled as long as there was continuous energy in the channel. Thus, the modems could switch to signaling frequenciesand begin full-duplex datatransmissionafter the suppres$orswere first disabled.If the energy in the channel was removed for 100 msec,the echo suppressor$were reactivated. Hence, rapid line turnaroundwas required for half-duplex modemsutilizing the entire bandwidth for each direction of transmission. Contemporary modems have echo can*

56

BACKGRoUNDANDTEHMINoLoGY

cellers for full-duplex transmissionand thereforeautomatically disable echo suppressors (which are probably nonexistent).

1.3 THEINTRODUCTION OF DIGITS Voicedigitization andtransmission firstbecame feasible in thelate1950s whensolidstateelectronics became available. In 1962BellSystem personnel established thefirst

commercialuseof digitaltransmission whentheybeganoperatinga T1 carriersystem for useasa trunkgroupin a chicagoareaexchanget331.After theTl systema family of r-carrier systems(Tl, Tlc, TlD, Tz, T3, T4) weredeveloped,all of which involvedtime divisionmultiplexingof digitizedvoicesignals. The world's first commerciallydesigneddigital microwaveradio systemwasestablishedin Japanby Nippon ElecrricCompany(NEC) in 196g t341.In the earty 1970sdigitalmicrowavesystemsbeganto appearin theUnitedStatesfor specialized datatransmission services.The first digital microwavelink in the U.S. public telephonenetworkwas suppliedby NEC of Japanfor a New york Telephonelink betweenBrooklyn and North statenIslandin lgTz 1341.Digit€I microwavesystems weresubsequently developedandinstalledby severalU.S.manufacturers for usein intermediate-length toll andexchangeareacircuits. Bell System'sf,rrstcommercialuseof digital fiber optic transmission occurredin septemberof 1980on a short-haulroutebetweenAtlantaandsmyrna,Georgia[35]. Threeyearslaterthefirst long-haulsystembetweenNew york andwashington,D.c., wasput into service. In additionto ffansmissionsy$tems,digital technologyhasprovento be equally usefulfor implementingswirchingfunctions.Thefirst countryto usedigitalswitching in thepublictelephone networkwasFrancein 1970t361.Thefirsr applicationof digital switchingin thepublicnetworkof theUnitedStatesoccurredin early lg76 when Bell systembeganoperaringits No. 4ESSt37l in a class3 toll officein chicago.Two monthslaterContinentalTelephone companybeganoperationin Ridgecrest, California,of a digitaltoll switch[38].Thefirst digitalendoffice switchin theUnitedStates becameoperationalinJuly of rgii inthesmalrtownofRichmondHill,Georgia[39]. 1.3.1 VolceDigitizatlon The basicvoice-codingalgorithmusedin T-carriersystemsand most otherdigital voiceequipmentin telephone networksaroundtheworldis shownin Figure1.3g.The first stepin the digitizationprocessis to periodicallysamplethe waveform.As discussedat lengthin Chapter3, all of theinformationneededto reconstruct theoriginal waveformis containedin the samplesif thesamplesoccurat an g-kHzrate.The second stepin the digitizationprocessinvolvesquantization;identifyingwhich amplitudeintervalof a groupof adjacentintervalsa samplevaluefalls into. In essence the quantization process replaceseachcontinuouslyvariableamplitudesamplewith a dis-

OF DIGITS 57 I.3 THEINTRoDUCTION

tl -l -t -t - l - 5 - 6 - l

Outntiz8tion

ttn

ul|l

Coding

Figure 1.38 Voice digitizationProcess'

crete value located at the middle of the appropriate quantization interval. Since the quantized sampleshave discretelevels, they representa multipleJevel digital signal. For transmission purposes the discrete amplitude samples are converted to a binary codeword. (For illustrative purposesonly, Figure 1.38 shows 4-bit codewords.) The binary codes ale then transmitted as binary pulses. At the receiving end of a digital transmissionline the binary data stream is recovered,and the discrete sample values "intelpolate" between Samplevalare reconstructed.Then a low-pass filter is used to ues and re-createthe original waveform. If no transmissionerrors have occurred,the output waveform is identical to the input waveform except for quantization distortion: the difference between a sample value and its discrete representation.By having a large number of quantizationintervals (andhenceenoughbits in a codeword to encode them), the quantization intervals can be small enough to effectively eliminate perceptible quantization effects. It is worth noting that the bandwidth requirementsof the digital signal increaseas a result of the binary encodingprocess.If ttre discrete,multiple-amplitude samplesare transmitted directly, the bandwidth reguirements are theoretically identical to the

58

BACKGRoUNDANDTEHMINoLoGY

bandwidth of the original signal. when eachdiscretesampleis representedby a number of individual binary pulses,the signal bandwidth increasesaccordingly. The twolevel pulses,however, are much less vulnerable to transmissionimpairments than are the multiple-amplitude pulses (or the underlying analog signal).

1.8.2 TlmeDivisionMultlplexing Basically, time division multiplexing (TDM) involves nothing more than sharing a transmissionmedium by establishinga sequenceof time slots during which individual sourcescan transmit signals.Thus the entire bandwidth of the facility is periodically available to eachsourcefor a restrictedtime interval. In contrast,FDM systemsassign a restrictedbandwidth to each sourcefor all time. Normally, all time slot$ of a TDM sy$temare of equal length. Also, each subchannelis usually assigneda time slot with a common repetition period called a frame interval. This form of TDM (as shown in Figure 1.39) is sometimesreferuedto as synchronoustime division multiplexing to specifically imply that each subchannelis assigneda certain amount of transmission capacity determinedby the time slot duration and the repetition rate. In contra$t,another form of TDM (refened to as sfatlstical, or asynchronousrtime division multiplexing) is described in chapter 10. with this second form of multiplexing, subchannelrates are allowed to vary according to the individual needsof the sources. The backbonedigital links of the public telephonenetwork (T-carrier, digital microwave, and fiber optics) use a synchronousvariety of TDM. Time division multiplexing is normally associatedonly with digital transmission links. Although analog TDM transmissioncan be implemented by interleaving samples from each signal, the individual samplesare usually too sensitiveto all varieties of transmissionimpairments. In contrast,time division switching of analog signals is more feasible than analog TDM transmissionbecausenoise and distortion within the switching equipment are more controllable. As discussedin chapter 5, aralog TDM techniqueshave been used in some PBXs befbre rtigital electronics becameso inexpensive that the digitization penalty disappeared.

T-CarrierSystems The volumeof interofficetelephone trafficin the UnitedStateshastraditionally grownmorerapidlythanlocaltraffic.Thisrapidgrowthputsevere strainontheolder

Figure 1.39 Time divisionmultiplexing.

oF Dlclrs 1,s THEtNTRoDUcTroN

59

interoffice transmission facilities that are designed for lower traffic volumes. Telephone companieswere often faced with the necessarytask of expanding the number of interoffice circuits. T-carrier systemswere initially developed as a cost-effective means for interoffice transmission: both for initial inslallations and for relief of crowded interoffice cable pairs. Despite the need to convert the voice signalsto a digital format at one end of a Tl line and back to analog at the other, the combined conversion and multiplexing cost of a digital TDM terminal was lower than the cost of a comparableanalog FDM terminal. The first T-carrier systemswere designedspecif,rcallyfor exchangeareaffunks at distancesbetween l0 and 50 miles. A T-carder systemconsistsof terminal equipment at eachend of a line and a number of regenerativerepeatersat intermediatepoints in the line. The function of each regenerative repeater is to restore the digital bit stream to its original form before transmission impairmentsobliteratethe identity of the digital pulses.The line itself, including the regenerative repeaters,is referred to as a span line. The original terminal equipmentwas referred to as D-type (digital) channelbanks,which camein numerous versions.The transmissionlines were wire pairs using 16- to 26-gaugecable' A block diagram of a T-carrier systemis shown in Figure 1.40. The first T1 systemsusedDlA channelbanksfor interfacing, converting, and multiplexing ?4 analog circuits. A channelbank at eachend of a spanline provided interfacing for both directions of transmission.Incoming analogsignalswere time division multiplexed and digitized for transmission.When received at the other end of the line, the incoming bit streamwas decodedinto analog samples,demultiplexed, and filtered to reconsfuct the original signals.Each individual TDM channel was assignedI bits per time slot. Thus, there were (24X8) = 192 bits of information in a frame' One additional bit was addedto each frame to identify the frame boundaries,therebyproducing a total of 193 bits in a frame. Since the frame interval is 125 psec, the basic Tl line rate became L544 Mbps. This line rate has been establishedas the fundamental standardfor digital transmissionin North America and Japan.The standardis referred to as a DSI signal(for digital signal 1). A similar standardof 2.048 Mbps has been establishedby mJ-T for most of the rest of the world. This standardevolved from a Tl-like $y$temthat provides 32 channels at the samerate as the North American channels.Only 30 of the channelsin the El standard,however, are used for voice. The other two Areused for frame synchronization and signaling. Signaling and control information for Tl systemsare inserted lnto into each voice channel (or transmitted separatelyby CCIS facilities). Digital signaling and conffol techniquesfor both systemsare discussedin Chapter 7'

Figure 1.40 Tl-carrier system,

60

BACKGHOUND ANDTERMINOLOGY

The greatlyincreased attenuation of a wire pair at thefrequencies of a DSI signal (772kHz cbnterfrequency)mandates theuseof amplificationat intermediate points of a Tl spanline.In contrastto ananalogsignal,however,a digitalsignalcannotonly be amplifiedbut al$obe detectedandregenerated. That is, aslong asa pulsecanbe detected,it canbe restoredto its originalform andrelayedto the next line segment. For this reasonTl repeaters arereferredto asregenerative repeaters. Thebasicfunctionsof theserepeaters are: l. 2. 3. 4.

Equalization Clockrecovery Pulsedetection Transmission

Equalizationis requiredbecausethewire pairsintroducecertainamountsof bothphase andamplitudedistoltion that causeintersymbolintederenceif uncompensated. Clock recoveryis requiredfor two basicpur?o$es; first, to establisha timing signalto sample theincomingpulses;second, to transmitoutgoingpulsesatthesamerateasattheinput to theline. Regenerativerepeatersare normally spacedevery 6000 ft in a T1 spanline. This distancewas chosenas a matterof convenience for convertingexistingvoice frequencycablesto T-carrierlines. Interofficevoice frequencycablestypically used loadingcoils that werespacedat 6fi)0-ft intervals.Sincethesecoils werelocatedat convenientaccess points(manholes) andhadto be removedfor high-frequency transmission,it wasonly nafuralthatthe6000-ftintervalbechosen.Onegeneralexception is that the first regenerativerepeateris typically spaced3000ft from a centraloffice. The shorterspacingof this line segmentwasneededto maintaina relativelystrong signalin the presence of impulsenoisegenerated by olderswitchingmachines. Theoperatingexperience of Tl systemswasso favorablethattheywerecontinually upgradedandexpanded. one of theinitial improvements producedTlc systems that providehighertransmission ratesover Z?-galgecable.A Tlc line operatesat 3.152Mbpsfor 48 voicechannels, twice asmanyasa Tl system. Anotherlevelof digitaltransmission becameavailablein 1972whentheT2 system wasintroduced.This systemwasdesignedfor toll networkconnections. In contrast. T I systemswereoriginallydesigned only for exchange areatransmission.TheTZ system providedfor 96 voice channelsat distancesup to 500 miles.The line rate was 6.312Mbps,whichis referredto asa DSZstandard. Thetransmission mediawasspecial low-capacitance 2?-gaugecable.By usingseparate cablesfor eachdirectionof transmission andthespeciallydevelopedcables,T? systemscourduserepeaterspacings up to 14,800ft in low-noiseenvironments. The emergence of opticalfiber systemsmadecopper-based T2 transmission systemsobsolete. TDM Hierarchy In a manneranalogous to theFDM hierarchy,AT&T established a digitalTDM hierarchyttrathasbecomethe standardfor North America.Stafiingwith a DSI signalas

OF DIGITS 61 I.3 THEINTBODUCTION TAELE 1.10 Dlgltal TDM Signale ol North America and Japan DigitalSignal Numberof VoiceCircuits Number

DS1 DSlC

48

DS2

96

DS3

672

D54

Bit Rate MultiplexerDesignation (Mbps)

D channelbank (24analoginputs) M1C (2 DSI inputs) M12 (4 DS1inputs) M13 (28DS1inputs) M34 (6 DSginputs)

Media Transmission

1.544

T1 pairedcable

3.152

T1Cpairedcable

6,312

T2 pairedcable

44.736

Radio,Fiber

274.176

T4Mcoax,WT4 waveguide, radio

a fundamentalbuilding block, all otherlevelsareimplementedas a combinationof of the higherlevel digital mulsomenumberof lower level signals.Thedesignation levels. For example,an M12 multirespective input and output tiplexersreflectsthe DSZ plexercombines form a single signal..TableI.10 liststhe four DSl signalsto mediausedfor each.Novariou$multiplexlevels,theirbit rates,andthetransmission tice that thebit rateof a high-levelmultiplexsignalis slightly higherthanthe combinedratesof the lower level inputs.Theexcessbits areincludedfor certaincontrol in Chapter7. A similardigitalhierarchyhas functionsdiscussed andsynchronization As shownin Tablel. I I' by ITU-T asan internationalstandard. alsobeenestablished this hierarchyis similarto the North Americanstandardbut involvesdifferentnumbersof voicecircuitsat all levels. Dig itaI Pal nGal n Syefems introductionof Tl systemsfor interofficetrunks,mostmajor Followingthesuccessful for local disof telephoneequipmentdevelopeddigitalTDM system$ manufacturers tribution.Thesesystemsaremostapplicableto long ruralloopswherethecostof the electronicsis offsetby the savingsin wire pairs.No matterwhatthe distanceis, uninby addingelectronics, expectedgrowthcanbe mosteconomicallyaccommodated steadof wire,to producea pair-gainsystem.Thepossibilityof trailerparks,apartment nightmares houses,or Internetserviceprovidersspringingup almostovernightcauses provide altera networking plant Pair-gain system$ forecasters. in themindsof cable nativeto dispelthosenightmares. Digital pair-gain$ystemsare also usefulas alternativesto switchingoffices in areoftenservicedbysmallautomaticswitchSmallcommunities smallcommunities. ing systemsnormallyunattended andremotelycontrolledfrom a largerswitchingoffice nearby.Thesesmallcommunityswitchesarerefenedto ascommunitydialoffices *Because T2 transmissionsystemshave becomeobsolete,the Ml2 function exists only in a functional sense within Ml3 multiplexers, which multiplex 28 DSI signals into I DS3 signal.

62

BACKGHoUNDANDTERMtNoLocy

TABLE 1.11 ITU Digital Hierarchy

LevelNumber El E2 E3 E4 E5

Numberof Voice Circuits

30 120 480 1920 7680

Multiplexer Designation M1? M?3 M34 M45

Bit Rate(Mbps) 2.048 8,448 34,368 139.264 565.1 48

(CDOs).A CDO typicallyprovidesonly limitedservicefeaturesto thecustomers and often requiresconsiderable maintenance. Becausedigital pair-gainsystemslower transmission groupsof subscribers, co$t$for moderate-sized theyarea viablealternative to a CDO: Stationsin the smallcommunityareservicedfrom the centraloffice by way of pair-gainsystems. A fundamental consideration in choosingbetweenpairgain systemsandremoteswitchinginvolvesthe haffic volumesandcallingpatterns within thesmallcommunity.Thebasictechniques of analyzingtrafficpaftemsanddeterminingtrunk groupsizesareprovidedin Chapter12. Thefirst two digitalpair-gainsy$tems usedin theBell Systemwerethesubscriber loop multiplex(sLM) system[40, 4l] and,its successor, the subscriberloop carrier (slc-40) system140,421.Althoughthesesysremsuseda form of voicedigitization (deltamodulation)differentfrom thatusedin T-carriersystem$(pulsecodemodulation), they both usedstandardrl repeatersfor digital transmissionat 1.544Mbps. Both systemsalsoconvertedthedigitizedvoicesignalsbackinto individualanalogintetfacesat theendofficeswirchto achievesystemtran$parency. FiguresL4l and1.42 showblock diagramsof thesesysrem$. Noticethat the sLM systemprovidedboth concentration and multiplexing(80 subscribers for 24 channels)while the SLC-40 wasstrictlya multiplexer(40 subscribers assigned in a one-to-one mannerto 40 channels). The sLM andsLC-,{Osystemsuseddeltamodulationvoicecodingbecause it was simplerthan pulsecodemodulationas usedin Tl systemsand was thereforeless costlyto implementon a per-channel basis-a desirablefeaturefor modularsystem implementations. The originalTl systems,on the otherhand,minimizedelectronics costsby usingcommonencodersanddecoders, which precludedimplementation of

Crofibdr witch

Figure l.4l

Subscriber loop multiplexer.

oF DIGITS 63 1.3 THEINTHODUCTIoN

'.*Jill"i,**' loopcarrier(SLC-40), Figure1.42 Subscriber lessthan24 channels(anunnecessary featurein aninterofficeapplication).By thelate pulsecodemodulation 1970slow-cost,integratedcircuitimplementations of standard becameavailablethatled theway to thefirst (1979)installationof theSLC-96,a subusingvoicecodingthatwascompatiblewith T1 systemsandthe scribercarrierrty$tem emergingdigitalendoffice switchingmachines[43]. The SLC-96system(whichis functionallyequivalentto four Tl lines)caninterinto 24 distinctanalog facedirectlywith a digitalendoffice andnot bedemultiplexed interfaces. Thusthis capability,which is referredto asintegrateddigitalloop carrier (IDLC), greaflyreducestheprove-indistancewherethe digital carrierbecomesless of the SLC-96 enhancements expensivethan separatesubscriberpairs.Subsequent supportfor local switchingfunctionsin the $y$temincludeuseof fiber transmission, remotemodule,andconfigurabilityof analoginterfacesfrom the centraloffice [44, digital loop carriersystemshavebeende45.1.Many newercopper-andfiber-based morethoroughlyin Chapter11in thecontextof veloped.Thesesystemsarediscussed digital subscriber access. 1.3.3 Data under Voice AT&T beganoffering After thetechnologyof T-carriersystemshadbeenestablished, This service,knownas servicesfor datacommunications. leaseddigitaltransmission links with specialterminals DataphoneDigital Service(DDS),usesTl transmission (channelbanks)thatprovidedirectaccessto the digital line. An initial drawbackof areaand T-carriersystemswereoriginallyusedonly for exchange DDS arosebecause digital transmission, shorrtoll networktrunks.Without someform of long-distance AT&T's exchangeareascould not be interconnected. the digital circuitsin separate of a spewasthedevelopment digitaltransmission to long-distance originalresponse cial radioterminalcalledthe 1A radiodigitalterminal(IA-RDT) [46].This terminal encodedoneDSI signal(1.544Mbps)into lessthan500kHz of bandwidth.As shown of in Figure1.43,a signalof thisbandwidthwasinsertedbelowthelowest'frequency a mastergroupmultiplex(Table1.5).Sincethis frequencybandis normallyunused in TD or TH analogradio systems,theDSI signalcouldbe addedto existinganalog belowthoseused Theuseoffrequencies routeswithoutdisplacinganyvoicechannels. "dataundervoice"(DUV). for voicesignalsleadsto the designation specifically a specialdevelopment It is importantto pointoutthatDUV represented In fact,DUV wasusedonly andnot for voiceservices. intendedfor datatransmission of facilitiesfor DDS. The emergence digital transmission to providelong-distance

64

BAcKeRouNDANDrERMrNolocy E

E 8.9

*E

Ee I 6

cr

3084 (kHzl Frequency Figure 1.43 Data under voice.

long-distance fiber sy$tem'r obviouslyeliminatedthe needfor DUV equipment(and eventhe analogradiosthemselves). 1.3.4 Digital Mlcrowave Radio In contrastto DUV systems,which piggyback1.544Mbpsonto an analogradiofor dataservices,a common-carrier digitalmicrowavesystemusesdigitalmodulationexclusivelyto transmitandreceivehigherleveldigitalmultiplexsignalsfor voicetraffic. Digital radiosusethesamefrequencybandsallocatedfor analogradiosaslistedin Table 1.7.Thusa primarydesignrequirementof a digitalradiowasthatit mustconfine its radiatedpowerto preventexcessiveinterferenceinto adjacent,possiblyanalog, chamels.Moreover,theFCCstipulatedthatthedigitalradioshadto becapableof providingroughlythesamenumberof voicecircuitsasexistinganalogradios.TableI .12 liststheminimumnumberof voicecircuitsandtheresultingbit ratethatmustbe providedby digitalradiosin eachof thecommon-canier microwavebands[47]. Despitethedesignconstraints imposedby compatibilityrequirements with analog radios,digital radiosprovedto be moreeconomicalthananalogradiosin severalapplications.Because of lowerterminalcosrs(multiplexinganddemultiplexing), digiral radio$ystemsweregenerallylessexpensivethananalogradio systemsfor distances TABLE1.12 Minimum VolceClrcuitRequirements ot DigitatRadlosln the United States

Equivalent Frequency Band Minimum Number Numberof DS1 (MHz) of Circuits Signals 2110-2130 2160-2180 370H200 5925-6425 10,700-11,700

96 96 1152 1152 1152

4 4 48 48 48

Resultant Bit Channel Rate(Mbps)a Bandwidth (MHz) 6,144 6.144 73.7 73.7 73.7

3.5 3,5 20 30 40

alhe actualbit rate is usuallyslightlygreater owingto €xtraframingand synchronizationbits. Furthermore,mosl radio systemBprovidesignificantlymor6 voice circuitsthan the minimum.

r.g rHErNTRoDUcrroN oFDrclrs 65 up to about300milesandon longerroutesrequiringchannelaccess(drop-and-insen) at intermediatepointsin the route [a8]. The major impetusfor digital radio in the of digitaltoll switcheslike theNo.4ESS.TheinterUnitedStateswastheemergence connectionof digitalradiosignalsto digitalswitchesavoidedcostlylowerlevel multiplex equipment. 1.3.5 Fiber Optic Transmiesion Of all the new technologyinhoducedinto the telephonenetwork,fiber hascertainly transmissionengineerswould had the mo$tprofoundeffect.Prior to its emergence, thecombinationof exffemelywidebandwidth,extremelylow attenuhaveconsidered assomethingakinto perpetualmotion.Low atation,andimmunityfrom interference costs-the to low maintenance whichequates tenuationallowslongrepeaterspacings, On December5, reasonfor wholesalereplacement of long-haulanalogradiosystems. fiber system[49].By the 1986,AT&T completedthelastsectionof a transcontinental endof thatdecadevirtuallyall of thehigh-densityrouteswereconvertedto fiber.Radio systemsareand will continueto be usedfor carryinglow-densitytraffic where However,becausetheseroutesare right-of-waycostsarea dominantconsideration. relatively$hortspursoff of digital fiber arteries,intetfacecostsimply digital implethuscompletingtheeconomicfoundationfor theall-digitalnetwork. mentations, As discussed in Chapter8, anopticalfiber is not aninherentlydigitaltransmission system.However,theinterfaceelectronics(driversandreceivers)andtheapplication of connectingto digital switcheshavestimulatedtheir useof digital kansmission.* consideringthe amountof bandwidthavailablethereis little incentive Furthermore, to conservebandwidth,aswasthe originalsituationwith to useanalogtransmission microwaveradios. 1.3.6 DigitalSwltching wasrepoftedby Earle into digitalswitchingat Bell Laboratories Theoriginalresearch theconcept to demonstrate Vaughanin 1959[501.Laboratorymodelsweredeveloped systemswith time division of integratingdigitaltime divisionmultiplextransmission solid-stateelechonicshad not maswitchingsystems.Unfortunatelythe necessary of digitalswitchingwa$not turedsufficientlyat thattime socommercialdevelopment pursued,anddevelopment of theNo. I ESScontinuedalongthelinesof spacedivision technology.Almost l0 yearslater,however,Bell Labsbegandeelectromechanical velopmentof a digitaltoll switch,theNo. 4 ESS. Whenplacedin servicein January1976,the No. 4 ESSprovidedseveralnew capabilitiesfor the toll network.First, it was the first toll switch to be designedfor controlat the outset.tSecond,its capacitywasthreetimesthatof the stored-program prevailingelectromechanical switchat the time: theNo. 44 crossbar.The largerca*Analog of opticalfiber systemshavebeenwidely usedto carrytelevisionsignalsin feedetssegments .CATV systems. lstored-programconfrol was first implementedin the toll networkbeginningin 1969by retrofitting crossbarswitches[51].

66


pacityof theNo.4 ESSmeantthatmanymetropolitanareascouldconsolidate toll haff,rcinto oneswitchinsteadof several.Third,thedigitaltime divisiondesignof theNo. 4 ESSalloweddirectconnectionto digitalT-carrierlines.This lastfeatureillu$trate$ theoriginalattractionof digital switchingfor toll andtandemofficesof thenetwork. By 1976,whenthefirst No. 4 ESSwasinstalled,it wasclearthatdigitaltransmission wasdominatingtheexchange areaandshorterinteftolltrunks.Thussignificanteconomiesandimprovements qualityresultedby eliminatingchannelbanks in transmission at theinterfacebetweendigitalkunks anda switchingsystem. The earlydevelopment of digitalendoffice switchesin theUnitedStareswasundertakenby independent equipmentmanufacturers with theflustsystembeingplaced in servicein 1977.Thesesystemswereprimarilydesignedfor the smallerswitching officesof theindependent telephone companies. Digital switcheswereparticularlyattractiveto rural telephonecompaniesbecausethey couldprovidesignificantcopper savingswhendirectlyconnectedto digital pair-gaintransmission systems.The flust large digital end office switching sy$temto be introducedinto the North American networkwasthe DMs-100 providedby NorthernTelecom.Table Ll3 lists digital switchingmachinesdevelopedfor the North Americanpublic telephonenetwork. The functionalessence of a digital time divisionswitchingmatrixis illustratedin Figure1.44.As indicated,all inputsaretimedivisionmultiplexlinks.Theselinksmay represent digitalpair-gainsystem$, T-carrierinterofficetrunks,or theoutputsof colocatedchannelbanksusedto interfaceanaloglines to the digital switch. In any event the switchingmahix itself is designedro serviceonly TDM input links. Basically,the switchingmatrix is requiredto transferinformationarriving in a specifiedtime slot (channel)on an incomingTDM link to a specifiedtime sloton an

TABLE 1.13 Dlgltal Central Oflice Swltchlng Systems of North America

Manufacturer

Designation

AT&T 4 ESS AT&T 5 ESS C|T-Alcatel E 10-five GTE 3 EAX GTE 5 EAX LM Ericsson AXE10 NEC NEAX.61 ITT System1210 NorthernTelecom DMS-10 NorthernTelecom DMS-100 NorthernTelecom DMS-200 Siemens EWSD Stromberg Carlson DCO Vidar lTS4 Vidar lTS4/5

Dateof lntroduction 1S76 1S83 1982 1S7B 1982 1978 1979 1978 1977 197S 1978 1981 1977 1977 1978

Application Toll Local Local Toll/tandem Local Local/toll Local/toll Local/toll Local Local Toll Local Local Toll/tandem Local/toll

LineSize 107,000 100,000 100,000 60,b00 145,000 200,000 80,000 26,000 7,000 100,000 60,000 200,000 32,000 7,000 12,768

1,3 THEINTRODUCTION OF DIGIT$

67

I FRAME

Figure 1.44 Digital time divisionswitchingoperation. outgoing TDM link. Since an arbitrary connection involves two different physical links and two different time slots, the switching processrequires spatial translation (space switching) and time translation (time switching). Thus the basic operation is sometimes referred to as two-dimensional switching. Space switching is achieved with conventional digital logic selector circuits, and time switching is achieved by temporarily storing information in a digital memory or register circuit. For convenience,the inputs to the switching matrix of Figure 7.44 arc all shown on the left, and the outputs are on the right. Since the links are inherently four-wire, each input time slot is paired with the sametime slot on a correspondingoutput link. Hence, a full-duplex circuit also requires a "connection" in the reversedirection that is achievedby transferring information from time slot 17 of link N to time slot 3 of link I in the example shown. Operational and implementationdetails of a large range of digital time division switches are provided in Chapter 5.

1.3.7 DigltalNetworkEvolutlon The evolution of the analog telephone network into one that is all digital except for the accesslines is summarizedin Figure 1.45.The processbeganin the 1960s(a) with T1 systemsbeing installed on relatively short haul interoffice trunks within the exchangeareas.Next, in the early 1970s(b), digital transmissionwas introducedinto the toll network with T2 systemsfor relatively short routesbetweentoll offices. It was in the late 1970s (c) that digitization really began to take over. Tl coverageexpanded greatly, digital loop carier (DLC) systemscame into use,* and digital switches be* came available at all levels of the network: PBXs (DPBXs), end offices (DEOs), tandem offices, and toll offices (DTOs). Moreover, microwave digital radios (MDRs) proved to be advantageousto use in both the exchangeareasand the shorter toll network routes due to low interface costs to digital switches. Thus, the late 1970sproduced a number of integrated islands where digital switches within a region were interconnectedby digital transmissionlinks but therewas little digital connectivity between the islands. (Data under voice was installed as an overbuild to analogroutes for *Digital

subscribet caffier systemswere actually introduccd in the early 1970s,but these systems utilized a voice digitization technology (delta modulation) that was incompatible with the rest ofthe network and therefore did not figure into the integrated network. These early digital loop carrier systems atE now obsolete.

68

BACKGHOUND ANDTEHMINOLOGY

Exchangn arta Toll mtwuk

/ -\

tO r z--------.'t}..r-*{",7'g o Q o

I

1\n/'1" ,i \

I

T_1 i

t-/

'

,\r

o )\ (- - c

Excharqe arot

.Tr\

--"----*l-/

\----

(dl

trz j I

I

I

I

(1

(Dl

l,

tt*--

\ord

DTO MDR

oTo

DPEX

o

\ - J

(d)

.-**:/\.

,) .'i

I

Fibrt

::-*-]./

'\-----

(d) Fibrr

DLC /

7

'r) ,'I

I r '

I ISDil P B I

oiix

\--'

\*----t6t

Figure 1.45 Digital networkevolution.

I

REFERENoES 6S

digitalnetwork A fully integratedandinterconnected dataservices.) limited-capacity emergedasthe becamea realityin the early 1980s(d) whenfiber optictransmission technologyof choicefor high-densitylong-haulroutes.Digital connectivityto businesscustomerpremisesequipmentalsooccurredin this time frameasTl becamethe preferredvoice trunk interfacefor largePBXs. End-to-enddigital connectivityfor voiceor dataservicesbecamea reality in the late 1980s(e) wittr theintroductionof ISDN basicrate(ISDN BR, 28 + D) andISDN to thecustomer.In addition,fiprimaryrate(ISDN PR,238 + D) digitalconnection$ bertechnologybecamemuchmoreubiquitousasDS3ratesystemseliminatedT2 syssystemsbecamethepreferredtechnologyfor temsin thetoll networkandfiber-based loop carrierandfeedersystems-evenat relativelyshortdistances.

REFERENCES I

S. P, Thompson, Phillip Reis: Inventor of the Telephone, Sylvanus P. Thompson, London. New York. 1883.

"Record'81TelcoConstructionSetat $23.5Billion, Up R. SmithandW. Sonnerryille, Management, Jan.15, 1981,p. 55. Engineer and 5.6Vo,*Telephone Engineering, andOperationsin theBelI System,Znd TechnicalStaff,Bell Laboratories, 4 5 6

ed., Bell Telephone Laboratories, Indianapolis, Indiana, 1983. "Plan for Access*Tandem and Trunking," Telephone Engineer & D. M. Mardon, Nov. 15, 1983,pp. l4O-142. Management, W, D. Reeve, Subscriber Inop Signaling and Transmission Handbook, IEEE Press, New York. 1995, "Digital Speech Interpolation Systems," in Advanced DiSital S, J. Campanella, Communication,r; Systerzs and Signal Processing Tethniques, K. Fehrer, Ed., Prentice-Hall, Englewood Cliffs, NJ, 1985. Technical Staff, Bell Laboratori es,Engineering and Operations in the Bell System,Bell Telephone Laboratories, Indianapolis, Indiana, 1977. "The History and Future of Commercial Satellite Communications," W. L. Pritchard, IEEE Communitations M agaarne,May 1984, pp. 22-37 .

ll

t2 l3

Digital Video Broadcasting (DVB), Framing structure,channel coding and modulation for digital tenestrial television, EN 300 744, (ETSI TM 1545), 1999, "Spech Codec for the I. Boyd, C. B. Southcott, D. P. Crowe, and P. J. Bolingbroke, Skyphone Aeronautical Telephone Service,"BritishTelecommunications Engineering, July 1986,pp. 83-91. "Evaluation of MessageCircuit Noise," Bell System A. J. Aikens and D. A. Lewinski, Technical Jountal, July 1970, pp. 879-909. J. L. Sullivan, t'A Laboratory System for Measuring Loudness Loss of Telephone Connections," Bell SystemTechniml Joumal, Oct. 1971, pp. 2663-2739. "A Basis for Transmission Performance Objective in a Telephone W. K. Macadam, Communications System," AIEE Transactions on Comntunications and Electronics, May 1978,pp. 205-209.

70


14 F. P. Duffy and T. W. Thatcher,Jr,, "Analog Transmissionperformanceon the SwitchedTelecommunications Network,"BelI SystemTechnical Journal,Apr. 1971, p p .l 3 l t - 1 3 4 7 . l5 W. R. BennettandJ. R. Davey,Data Transmission, McGraw-Hill,New york, 1965, 16 Membersof Technicalstaff, Bell TelephoneLaboratories,Transmissionsystems for Communitatiorrs, 4th ed,,WestemElectricCompany,Winston-Salem, NorthCarolina, 1971. 1l H. R. Huntley, "TransmissionDesignof Intertoll TelephoneTrunks,',BeII System TethnicalJoumal,Sept.1953,pp. l0l9-1036, l8 K. Murano,S. Unagami,andF. Amano,"EchoCancellationandApplications,,'IEEE Communictttions Magazine,Jan.1990,pp. 49-55. 18a E, A, Lee and D. G. Messerschmin, Digital communication,Kluwer Academic Publisher,Norwell,MA, 1993. 19 "A TechnicalReporton Echo Cancelling,"ReportNo, ?7, ANSI T14,l.6 Working Groupon Specialized SignalProcessing, Nov. 1993. 20 H. G. Suyderhoud,M. Onufry, and S. J. Campanella,"Echo Connol in Telephone Communications,"National Telecomtnunications ConferenceRecord, 1976, pp. 8 . 1 - l - 8 I. - 5 . 21 F. T. Andrews,Jr. and R. w. Hatch, "Nationa]TelephoneNetwork rransmission

22 23 M 25

26 27

28 29 30 3l 32

Planningin the AmericanTelephoneandTelegraphcompany," IEEE Transactionson Communitations Technology, June1971,pp. 302-314. cornpatibility Information for Telephone Exchange service, Bellcore Technical AdvisoryTA-NPL-000:9/2, Bell Communications Research. Issuel. Feb.1989. Noteson THE NETWORK,AmericanTelephoneand TelegraphCompany,Network PlanningDivision,AT&T, New York, 1980,pp. 44-58. "specialIssue:storedhogram conftol Network,"Bell systemTechnical Joumal,sept. 1982. P. J. Kuhn,C, D. Pack,andR. A. Skoog,"CommonChannelSignalingNetworks:past, Present,Ftfitre," IEEEEJoumalon selededAreasin communications, Apr.1994,pp. 383-393, J. S. Doyle and C. S. McMahon, '-The Intelligent Network Concept," IEEE Transactions on Communicafdons, Dec. I988,pp. 1296-130L J. M. Alcinpelu, "The overload Performance of Engineered Networks with Nonhierarchical andHierarchical Routing,"Bell systemTechnical rountal, sept.1984, p p .l 2 6 l - 1 2 8 1 . G. R. AshandE. oberer,"DynamicRoutingin theAT&T Network-tmproved service Proceedings,lgSg, pp.g.l.l-9.1.6, Qualityat LowerCost,*IEEEGlobecom C. Mason,"SoftwareProblemCripplesAT&T Long-Distance Network,',Telephony, Ian.22, 1990,pp. l0-l l. "specialIssueon Advanced Mobile Phoneservice(AMPS),"Beil systemTechnical Joumal,Jan.1979. D. c. cox, '?ortable Digital Radio communications-An Approachto Tetherless Access,"IEEE Communications Magaaine,July 1989,pp. 30-40. "Mobile stationLand stationcompatibility specificarion,"ANsI/EIA./TIA standard 553,ElectronicIndustriesAssociation,I 989,

PHOBLEMS71 "TheTl CarrierSystem,"BelI Systern TethnicalJoumal, 33 K. E. FultzandD. B, Penick, Sept.1965,pp. 1405-1451. 34 ..SupplyRecordon NEC PCM MicrowaveEquipment,"Nippon ElectricCompany, Tokyo,May 1974. "Optical Fiber Transmission-FromConceptionto Prominencein 20 35 M. Schwartz, Magazine,May1984'pp. 38-48' Years,"IEEE Communications "The New Approaches and solutionsoffered by the Digital ESS.El0 to 36 J. B. Jacob, Deal with Switching, Operation, and Maintenance'Functions," International e Proceedings, I 979' pp. 46.6-I -46.6-5. ationsConferenc Communic "special Issue:No, 4 ESS,"Bell SystemTechnicalJournal,JulylAng'l98l ' 3'l "A 38 J. C. McDonaldand J. R. Baichtal, New IntegratedDigital switching system," NationalTelecommunitationsConference Record, 1976' pp. 3.2-| -3'2-5. Aug. I ' 39 "FirstClass5 Digital CentralOfficeGoesinto Servicein Georgia,"Telephony, 1977,pp.M-25. "CustomerLinesGo Electronic,"Bell LabsRecord,Feb.1972,pp. 40 F. T. Andrews,Jr,, 59*64. "Digital Loopcarriersystems,"Bell 4l M, T. Manfred,G. A. Nelson,andc. H. sharpless, TechnicalJoumal,Apr. 1978,pp. I 129-1142. System "ThesubscriberLoopcarier system 42 S.J. Brolin,G. E. Harrington,andG. R. t eopold, GetsEvenBetter,"BeIl labs Record,May1977, pp. ll9-122. "Insidethe New Digital 43 S. Brolin, Y. s. cho, w. P. Michaud,andD. H. williamson, Loop System,"Belt l-absRecord,Apr' 1980'pp. I 10- I I 6. Subscriber "A 'smart' slc carrier Meetsthe changing 44 G. E. HarringtonandR. w, wyndrum, Natureof servicesin the EvolvingDigital Loop," Bell l-absRecord,sept. 1985,pp. l3-17. "Digital InterfaceBetweentheSLC-96DigitalLoopCarrierSystemanda LocalDigital 45 Bellcore,Aug' 1987' TR-TSY-000008, Switch,"BellcoreTeohnicalRequirement$, 'DataUnderVoice,"' "lA Put Terminals RadioDigital 46 R. R. GradyandJ. w. tuupT, Bell LabsRecord,May1974,pp.16l-166. 4'7 FCC MemorandumOpinionand Order,releasedJanuary29' 1975,modifyingFCC 27,1974. September ReportandOrderofDocketl93l l, released "Digital Economics,"Digital MicrowaveTransmission Transmission 48 W. S. Chaskin, Sept.I 978. sium,RockwellInternational, Engineering Sympo "AT&T's Digital NetworkEvolution,"IEEE InternationalConference 49 R. C. Hansen, 1987,pp. 198'l'l -1. 1.5. on Communications, "ResearchModel for Time-Separation IntegratedCommunication," 50 H. E. Vaughan, Joumal,July 1959'pp.909-932' BelISystemTechnical 'ElectronicTranslatorSystemfor Toll Switching,"IEEE 5l B. T. FoughtandC. J. Funk, June1970'p. 168. on Communications, Transactions

PROBLEMS l.l

andmilliwattsof absolute 33 dBmCof noisein termsof picowatts Express power.

72


1.2 Twenty-seven picowattsof noisewith a flat spectrumbetween0 and3 kHz is how manydBmC? 1.3 A valueof 30dBrnc0 is howmanypicowartsof absolutenoisepowerata -3-dB TLP? t.4 An idle-channelnoisepowermeasurement of 2l dBrnc occursat a -7-dB TLp. Expressthe noisepowerof this measurement in dBrnc0 and determinewhat powermeasurement thisnoisewouldproduceat anotherpointin thecircuitthat is designated asa -2-dB TLP. 1.5 A transmission link has14dBrn of absolutenoisepowerat a -13-dB TLp input testpointand27 dBmof absolutenoi$epowerat a -3-dB TLp outputte$tpoint. How muchabsolutenoiseis addedon thetransmission link? r.6 A transmissionlink with a -Z-dB TLp at the rransmitendanda -4-dB-TLp at thereceiveendis partof a voicecircuitthatproducesthefollowingidle-channel noisemeasurements: 18 dBrnc0 at the transmitend and20 dBrnc0 at the receiveend.what is the conrribution(in pwp0) of the transmission link toward thetotal absolutepowerat thereceiveend? 1.7 why is echocontrolunnece$sary on short-delay voiceconnections? what about singing?

WHYDIGITAL? The first chapterprovides an overview of an analog telephonenetwork and a brief introduction to digital ffansmission and switching technology introduced to replace older analog equipment.This chapterdiscussesthe basic technical advantagesof digi' tal implementationsthat stimulated the replacementof the analog systems.

OF DIGITALVOICENETWORKS 2.1 ADVANTAGES A list of technical featuresof digital communicationsnetworks is provided in Table 2. L Thesefeaturesare listed in the order that the author considersto be their relative importancefor generaltelephony.In particular applications,however, certain considerationsmay be more or lesssignificant. For instance,the last item, easeof encryption, is a dominant feature favoring digital networks for the military. Most of the featuresof digital voice networks listed in Table 2.1 and discussedin the following paragraphspertain to advantagesof digital hansmission or switching relative to analog counteryarts.In some instances,however, the featurespertain only to all-digital networks. Encryption, for example,is practical and generallyuseful only if the secureform of the messageis establishedat the sourceand translatedback into the clear form only at the destination.Thus an end-to-enddigital systemthat operates with no knowledge of the nature of the traffic (i.e., provides transparenttransmission) is a requirement for digital encryption applications. For similar reasonsend-to-end digital transmissionis neededfor direct transmissionof data (no modem)-When a network consistsof a mixture of analog and digital equipment, universal use of the network for servicessuch as datatransmissiondictatesconformanceto the leastcommon denominator of the network: the analog channel.

2.1.1 Ease of Multiplexing werefirst appliedto generaltelephony As mentionedin Chapterl, digitaltechniques thesesystems in interofficeT-carrier(time divisionmultiplex)systems.In essence, pathfor thecostof multiplepairs tradedelectionicscostsat theendsof a transmission 73

74

WHYDIGITAL? TABLE 2.1 Technical Advantages ot Dlgltar communlcations Networks

1. Ease0l multiplexing 2. Easeof signaling 3. U$eof moderntechnology 4. lntegration of kansmission andswitching 5. Signalregeneration 6. Performance monitorability 7. Accommodation of otherservices 8. Operability at lowsignal-to-noi$e/intederence ratios L Easeof encryption'

of wires betweenthem.(A hade that is morecost effectiveeveryyear.)Although FDM of analogsignalshadalsobeenusedto reducecablecosts,FDM equipmentis muchmoreexpensivethanTDM equipment, evenwhenthecostof digitizationis included.After voicesignalshavebeendigitized,TDM equipmentcostsarequitesmall by comparison. Sincedigitizationoccursonly at thefirst levelof theTDM hierarchy, high-leveldigitalTDM is evenmoreeconomicalthanhigh-levelFDM counrerpans. It shouldbe pointedout thatTDM of analogsignalsis alsovery simpleanddoes not requiredigitizationof rhesamplevalues.Thedrawbackof analogTDM lies in the vulnerabilityof nanow analogpulsesto noise,distortion,crosstalk,andintersymbol interference. Thesedegradations cannotbe removedby regeneration as in a digital system.Hence,analogTDM is not feasibleexceptfor noiseless, distortion-free environments.* In essence theability to regenerate a signal,evenat theexpenseofgreater bandwidth,is almosta requirement for TDM fransmission. 2.1.2 Eaeeof Slgnating control information(e.g.,on-hook/off-hook, addressdigits,coin deposits)is inherentlydigital and,hence,readilyincorporated into a digitalFansmission system.one meansof incorporatingcontrolinformationinto a digital transmission link involves time division multiplexingthe control as a separatebut easilyidentifiablecontrol channel'Anotherapproachinvolvesinsertingspecialcontrolcodesinto themessage channelandhavingdigital logic in thereceivingterminalsdecodethat controlinformation'In eithercase,as far asthe transmission sy$temis concerned, controlinformationis indistinguishable from message haffic. In contrast,analogtransmission sy$tems requiredspecialattentionfor controlsignaling.Many analogtransmission systemspresented uniqueandsometimes difficult environments for insertingcontrolinformation.An unfortunateresultwasthatmany varietiesof controlsignalformatsandprocedures evolvedfor theanalognetwork.The controlformatsdependon thenatureof boththetransmission systemundit, terminal .Aoulog

TDM has been used in a fbw telephone applications. Farinon's SubscriberRadio System [l ], for example, used'pulse-width-modulatedTDM. As discussedin Chapter 5 some older PBXs also used analog TDM.

VOICENETWORKS 75 OF DIGITAL 2.1 ADVANTAGES

controlinformationhad equipment.In someinterfacesbetweennetworksubsystems to be convertedfrom one format to another.Signalingon analoglinks thereforerepandfinancialburdento the operatingtelephone resenteda significantadministrative companies. signalingremovedmostof the signalingcostsa$The moveto common-channel sociatedwith interofficetrunksbut did not changethe situationfor individualsubchannel. scriberlines,which mustcarrysignalingon the samefacility asthemessage relative to analog (DSLs) costs reduces thesignaling lines Theuseof digitalsubscriber telephone' a digital lines,which helpsoffsetthehighercostof a DSL and subscriber DSLsarea fundamentalaspectof ISDN, asdescribedin ChapterI l. In summary,digital systemsallow controlinformationto be insertedinto andexmedium of thenatureof thetransmission streamindependently ffactedfrom a message (e.g., cable,fiber, microwave,satellite).Thus the signalingequipmentcan (and system.It thenfollowsthatconfrom thetransmission should)bedesignedseparately subsysof thetransmission trol functionsandformatscanbemodifiedindependently impacting without be upgraded can systems digital transmission Conversely, tem. controlfunctionsat eitherendofthe link. 2.1.3

Use of Modern TechnologY

A multiplexer or switching matrix for time division digital signals is implemented with the samebasic circuits used to build digital computers:logic gatesand memory. The basic crosspoint of a digital switch is nothing more than an AND gate with one logic input assignedto the mes$agesignal and other inputs usedfor control (crosspoint selection).Thus the dramatic developmentsof digital integratedcircuit technology for computer logic circuits and memory are applicable directly to digital transmissionand switching fiystems.In fact, many standardcircuits developedfor use in computersare directly usable in a switching matrix or multiplexer. Figure 2.1 shows the basic implementation of a l6-channel, bit-interleaved, digital time division multiplexer using common digital logic circuits. As indicated, the multiplexing function involves nothing more than cyclically sampling the 16 input data streams.Such an operation assumesall of the data streamsare synchronizedto eachother. As discussedin Chapter 7, the processof synchronizing the data sffeamsrequires logic circuitry that is much more complicatedthan that shown. Nevertheless,the implementationof TDM is much less expensivethan analog FDM. Even greateradvantagesof modern technology have been achievedby using largescaleintegrated(LSD circuits designedspecifically for telecommunicationsfunctions such as voice encoding/decoding,multiplexing/demultiplexing, switching matrices, and special-purposeand general-purposedigital signal processing(DSP). Digital signal processingfunctions are describedin Section 2'2. The relative low cost and high performanceof digital circuits allows digital implementationsto be used in some applicationsthat are prohibitively expensivewhen implemented with comparableanalog components.Completely nonblocking switches, for example, are not practical with conventional analog implementations, except in

76

WHYDIGITAL?

f;TATA*l .rr BA F

a t t

BA

4 Bit Gounter

Figure 2.1 Sixteen-to-one TDM multiplexer. small sizes.In a modern digital switch the cost of the switching matrix itself is relatively insignificant. Thus, for medium-size applications,the size of the switch matrix can be increasedto provide nonblocking operations,if desired.The automaticcall distributor developedby collins-Rockwell [2] is an early exampleof a digital switch operating in an analog environment. A digital implementation was chosen largely becauseit could economically provide a nonblocking operation. The benefits ofmodern device technology arenot confined to digital circuits alone. Analog integrated circuits have also progressed significantly, allowing traditional analog implementationsto improve appreciably.one of the primary requirementsof an analog component, however, is that it be linear. It appears,if only becauseof researchand developmentemphasis,that fast digital componentsare ea$ierto manufacture than linear analog counterparts.In addition, digital implementations appear to have an inherent functional advantageover analog implementations.This advantage is derived from the relative easewith which digital signals can be multiplexed. A major limitation with the full use of LSI componentsresults from limited availability of external connectionsto the device. with time division multiplex techniques,a single physical pin can be used for multiple-channel accessinto the device. Thus the same technique usedto reducecostsin transmissionsystemscan also be usedwithin a local module to minimize the interconnectionsand maximize the utilization of very large scaleintegration. In the end, a "switch on a chip" is possibleonly if a great number of channelscan be multiplexed onto a relatively small number of external connections. The technological development to have the most significant impact on the telephonenetwork is certainly fiber optic transmission.Although fibers themselvesdo not favor digital transmissionover analogtransmission,the interface electronicsto a fiber

VOICENETWORKS 77 OFDIGITAL 2.1 ADVANTAGES

systemfunction primarily in an on-off (nonlinear)modeof operation.Thus digital althoughanalogopticaltechnologyis comfiberapplications, dominates transmission monlyusedin analogvideodistribution. 2.1.4 Integratlon of Transmis$lon and $wltching and switchingsystemsof telephonenetworks Traditionallythe analogtransmission independentorganizations.In theopby functionally weredesignedandadministered thesetwo equipmentclassesarereferredto as outside eratingtelephonecompanies, plant and inside plant, respectively.Theseequipmentsnecessarilyprovide standequipmentwasfunctionallyindebut,otherthanthat,transmission ardizedintdrfaces, pendentof switchingequipment. When TDM of digital voice signalswas introducedinto the exchangeareaand engineersbeganconsideringdigital switching,it becameapparent communications thatTDM operationswerevery similarto time divisionswitchingfunctions.In fact, as describedin laterchapters,the first stagesof digital switchesgeneratefirst-level links.Thusthe TDM signalsby nature,evenwheninterfacedto analogtransmission into the integrated can be easily multiplexingoperationsof a transmissionsystem switchingequipment. of integratingthetwo systemsis shownin Figure2.2.ThedeThebasicadvantage and multiplexingequipment(channelbanks)at the swirchingofficesis unnecessary, first-stageswitchingequipmentis eliminated.If bothendsof thedigitalTDM trunks are integratedinto a digital switch, the channelbanksat both endsof the trunk are eliminated.In a totally integratednetworkvoice signalsaredigitizedat or nearthe sourceandremaindigitizeduntil deliveredto their destination.Furthermore,all interoffice trunksandintemallinks of a switchingsystemcarryTDM signalsexclusively' Thusfirst-level multiplexinganddemultiplexingarenonexistentexceptat theperiphery of the nerwork.Althoughintegrationof DSI signalsinto $witchingmachinesis

Figure 2.2 Integration of transmission and switching: (a) nonintegrated nansmission and swirching, (b) integrated time time division switching and transmission.

78

WHYDIGITAL?

commonplace, integrationof higherlevel signalsis complicatedby higherlevelmultiplexingformats(pulsestuffing)describedin chapter7. A newermultiplexingformat (soNET) describedin chapter 8 has someoperationalmodesthar are more amenable to directinterconnection into a switchingsy$tem. Integrationof transmissionand switchingfunctionsnot only eliminatesmuch equipmentbut alsogreatlyimprovesend-to-endvoicequality.By eliminatingmultiple analog-to-digital and digital-to-analog conversionsand by using low-error-rate transmissionlinks, voicequaliryis determined only by theencoding/decoding pnxesses. In summary,theimplementation benefitsof a fully integrateddigitalnetworkare: I' Long-distance voicequalityis identicalto local voicequality in all aspectsof noise,signallevel,anddistortion. 2. since digital circuits are inherentlyfour-wire,network-generated echoesare eliminated,andtruefull-duplex,four-wiredigitalcircuitsareavailable. 3. Cableentrance requirement$ andmainframedistributionof wire pairsis greafly reducedbecause all trunksareimplemented assubchannels of a TDM signal. 2.1.5 Signal Regeneration As described morefully in thenextchapter,therepresentation of voice(or anyanalog signal)in a digitalformatinvolvesconveftingthecontinuousanalogwaveforminto a sequence of discretesamplevalues.Eachdiscretesamplevalueis represented by some numberof binarydigitsof information.when transmitted,eachbinarydigit is representedby only oneof two possiblesignalvalues(e.g.,apulseversusnopulseor apositive pulseversusa negativepulse).The receiver'sjob is to decidewhich discrete valueswereffansmittedandrepresentthe message asa sequence of binary-encoded discretemessage samples.If only smallamountsof noise,interference. or distortion areimpressed uponthesignalduringtansmission,thebinarydatain thereceiverare identicalto thebinarysequence generated duringthedigitizationor encodingprocess. As shownin Figure2.3,thetransmission process,despitetheexistence of certainimperfections, doesnot altertheessential natureof theinformation.Of course,if theimperfections causesufficientchanges in thesignal,detectionerrorsoccurandthebinary datain thereceiverdoesnot represent theoriginaldataexactly. A fundamentalattributeof a digital systemis that theprobabilityof transmission errorscanbemadearbitrarilysmallby insertingregenerative repeaters at intermediate pointsin the transmission link. If spacedcloseenoughtogether,theseintermediate nodesdetectandregenerate the digital signalsbeforechannel-induced degradations

H:gmretiw rGg€dt€r

Figure 2.3

Ftfttratiw rlPcrtcr

Signal regenerationin a digital repeaterline.

VOICE NETWORKS79 OFDIGITAL 2.1 ADVANTAGES becomelarge enoughto causedecision effors. As demonshatedin Chapter4, the endto-end error rate can be made arbitrarily small by inserting a sufficient number of regenerationnodesin the transmissionlink. The most direct benefit of the regenerationprocessis the ability to localize the effects of signal degradations.As long as the degradationson any particular regenerated segmentof a transmissionlink do not causeerrors, their effects areeliminated. In contrast, signal impairments in analog transmissionaccumulatefrom one segmentto the next. Individual subsystemsof a large analog network must be designed with tight controls on the transmissionperformance to provide acceptableend-to-end quality. An individual subsystemof a digital network, on the other hand, need only be designed to ensurea certain minimum error rate-usually a readily realizable goal. When an all-digital network is designedwith enough regenerationpoints to effectively eliminate channel errors, the overall ffansmissionquality of the network is determined by the digitization process and not by the transmission systems' The analog-to-digital conversion process inherently introduces a loss of signal fidelity sincethe continuousanalogsourcewaveform can only be representedby discretesample values.By establishingenoughdiscretelevels, however, the analog waveform can be representedwith as little conversion error as desired.The increasedresolution require$ more bits and consequentlymore bandwidth for transmission.Hence, a digital transmission sy$tem readily provides a trade-off between transmission quality and bandwidth. (A similar trade-off exists for frequency-modulatedanalog signals.)

2.1.6 PerformanceMonitorability An additional benefit of the source-independentsignal structurein a digital transmission systemis that the quality of the received signal can be ascertainedwith no knowledge of the nature of the traffic. The transmission link is designed to produce well-defined pulseswith discretelevels. Any deviation in the receive signal, other than nominal amount$planned for in the design,representsa degradationin transmission quality. In general,analog systemscannot be monitored or testedfor quality while in service since the transmitted signal sfucture is unknown. FDM signals typically include pilot signals to measurechannel continuity and power levels. The power level of a pilot is an effective meansof estimatingthe signal-to-noiseratio-only in a fixednoise environment.Hence, noise and distortion are sometimesdeterminedby measuring the energy level in an unusedmessageslot or at the edge of the signal passband' In neither case,however, is the quality of an in-service channel being measureddirectly. One common method of measuring the quality of a digital transmissionlink is to add parity, or cyclic redundancycheck (CRC), bits to the messagestream.The redundancy introduced to the data streamenablesdigital logic circuits in a receiver to readily ascertainchannel error rates. If the error rate exceedssome nominal value, the transmissionlink is degraded. Another technique for measuring in-service transmission qualiry is used in T-cartier lines. This techniqueinvolves monitoring certainredundanciesin the signal waveform

80

WHYDIGITAL?

itself' Whentheredundancy patternat thereceiverdeviatesfrom normal.decisionerrorshaveoccurred.A completedescriptionof theline codingformatusedin T-carrier $ystems is providedin Chapter4. Othermethodsof measuringtransmission qualityin digital systemsarediscussed in Chapters4 and6. 2.1.7 Accommodatlon of Other Services It waspreviouslypointedoutthata digitaltransmission systemreadilyaccommodates conffol(signaling)information.This factis representative of a fundamental aspectof digital transmission:any digitally encodedmessage(whetherinherentlydigital or conveftedfrom analog)presentsa coillmonsignalformatto thetransmission system. Thusthetransmission systemneedprovideno specialattentionto individualservices andcan,in fact, be totally indifferent to the natureof the traffic it carries. In an analognetworkthetransmission standard is the4-kHzvoicecircuit.All special servicessuchasdataor facsimilemustbetransformed to ,,looklike voice."In particular,datasignalsmustbeconvertedto ananalogformatthroughtheuseof modems. Thestandard analogchannelwa$necessarily optimizedfor voicequality.In sodoing, certaintransmissioncharacteristics (suchas the phaseresponseand impulse noise)receivedlessattentionthanmorenoticeablevoicequalityimpairments.Some lessemphasized considerations, phasedistortionin particular,arecriticalfor high-rate dataservices.Useof an analognetworkfor nonvoiceservicesoftenrequiresspecial compensation for variousanalogtransmission impairments.If the analogchannelis toopoor,it maybe unusable for a particularapplication.In contrast,themainparameter ofquality in a digital systemis theerrorrate.Low-error-ratechannelsarereadily obtainable.Whendesired,theeffectsof channelerrorscanbe effectivelyeliminated with errorcontrolprocedures implemented by theuser. An additionalbenefitof the commontransmission formatis ttrattraffic from differenttypesof sources canbeintermixedin a singletransmission mediumwithoutmutual interference. The useof a commonffansmissionmediumfor analogsignalsis sometimes complicated because individualservicesrequiredifferinglevelsof quality. For example,televisionsignals,whichrequiregreatertransmission qualitythanvoice signals,werenot usuallycombinedwith FDM voicechannelsin a widebandanalos transmission system[3]. 2.1.8 Operabillty at Low Slgnal-to-Noiee/tnteileronceRailos Noise and inted'erence in an analogvoice networkbecomemost apparentduring pauseswhenthesignalamplitudeis low. Relativelysmallamountsof noiseocspeech curing duringa speechpausecanbe quiteannoyingto a listener.The samelevelsof noiseor interference arevirtuallyunnoticeable whenspeechis present.Henceit is the absolutenoiselevelof anidle channelthatdetermines analogspeechquality.Subjective evaluationsof voice quality t4, 5l led to maximumnoiselevel standards of 2g dBmcO(-62 dBm0)for short-haulsystemsand34 dBrnc0 (-56 dBm0)for long-haul systems.For comparison, the powerlevel of an activetalkeris typically -16 dBm0.

2.2 DIGITALSIGNALPBOCESSING

81

ratiosin analognetworksare46 and end-to-endsignal-to-noise Thusrepresentative ratioson indiSignal-to-noise respectively. 40 dB for short-andlong-haulsystems, highervidualhansmissionsystemsarenecessarily In a digital systemspeechpause$are encodedwith a particulardatapatternand virBecausesignalregeneration transmittedat thesamepowerlevelasactivespeech. is noise demedium,idle channel tuallyeliminatesall noisearisingin thetransmission pauses link. Thusspeech terminedby the encodingprocessandnot thetransmission do not determinemaximumnoiselevelsastheydo in an analogsystem.As discussed at signalin Chapter4, digitalfansmissionlinks providevirtuallyerror-freeperformance to-noiseratiosof 15-25 dB, dependingon thetypeof line codingor modulationused' more systemto rejectcrosstalkis sometimes The ability of a digital transmission One of noise. levels ofrandom high in relatively to operate rhan its ability significant netof the analog maintenance design and in the considerations themosttroublesome The problemwas work was the needto eliminatecrosstalkbetweenconversations. was at maxiinterfering channel while an one channel mostacuteduringpauseson The noticeable. would be crosstalk mum power.At thesetimesrelativelylow level violated therefore and crosstalkwas particularlyundesirableif it was intelligible someone'sprivacy.Again, speechpau$esdo not producelow-amplitudesignalson digilinks maintaina constant-amplitude links.Thetransmission digitaltransmission processin tal signal.Thus,low levelsof crosstalkareeliminatedby theregeneration a digital repeateror receiver.Evenif thecrosstalkis of sufficientamplitudeto cau$e detectionerrors,theeffectsappearasrandomnoiseand,as$uch,areunintelligibleConsideringthe fact that a digital systemtypically needsa greaterbandwidththan a comparableanalogsystemandttratwider bandwidthsimply greatercrosstalkand noiselevels,the ability to operateat lower SNRsis pafily a requirementof a digital systemandpartlyan advantage. 2.1.9 Ease of Encryptlon Althoughmost telephoneusershavelittle needfor voice encryption,the easewith [6] meansthata digital which a digitalbit $treamcanbe scrambledandunscrambled for userswith sensitive an extra bonus provides network(or a digitalcellularsystem) encryptandis gendifficult to In contrast,analogvoiceis muchmore conversations. of common discussion erally not nearlyassecurea$digitally encryptedvoice.For a preAs mentioned seereferences [7], [8], and[9]. analogvoiceencryptiontechniques, viously,easeof encryptionstimulatedearlyuseof digital voicesystemsby themilitary.

2.2 DIGITALSIGNAL PROCESSING The precedingparagraphsemphasizethe advantagesof digital technologyin impleof a network.Anothersignificantapmentingthetansmissionandswitching$ystems plication of digital technologyis the areaof signalprocessing.Basically,signal

82

WHY DIGITAL?

processingrefers to an operationon a signal to enhanceor transform its characteristics. Signal processingcan be applied to either analogor digital waveforms. Amplif,rcation, equalization, modulation, and filtering are cofilmon examples of signal processing functions. Digital signal processing(DSP) refers to the use of digital logic and arithmetic circuits to implement signal processingfunctions on digitized signal waveforms. sometimes analog signals are converted to digital representationsfor the expresspurpose ofprocessing them digitally. Then the digital representationsofthe processedsignals are converted back to analog. These operationsare illustrated in Figure 2.4, where a sine wave comrpted by noise is digitally filtered to remove the noise. The main advantagesof digitally processingsignal waveforms are listed in Table 2.2. It is important to point out that DSP in this context refers to the technology usedto condition, manipulate, or otherwise transform a signal waveform (a digitizert representationthereof). In another context signal processingrefers to the interpretation of conffol signalsin a network by the control processorsof switching $ystems.In the latter casethe logical interpretationofa control code is processedand not an underlying signal waveform

2.2.1 DSPApplications The following four sectionsidentify applications of DSP that either representlower cost solutions to functions that have beentraditionally implementedwith analogtechnology or are functions that have no practical implementationcounterpartwith analog technology.

Echo Cancellers The cost and pedormanceof DSP echo cancellershave improved to the point that they can be justified for any long-distancecircuit, thereby providing full-duplex circuits (no echo suppression)and no artificial attenuation (no via net loss). A particularly critical needfor echo cancellationoccur$in high-speed,full-duplex data modemsthat incorporate near-end echo cancellation-an unnece$saryrequirement for voice circuits. Furthermore, low-cost echo canceling enablespacket-switchedvoice applications that inffoduce artificial delays that are not accommodatedin normal analoe

Anrlog input

Anelog to digitrl

Figure 2.4

Dieitd dFrl procarEof

Digit l to fido0

Digital signal processingof an analog signal.

Anrlog ouFut

srcNALPHocE$slNG 83 a.a DrcrrAL TABLE2.2 DigltalSlgnalProceeslngFeatures and parasiticelements The immunityof digitalcircuit$to smallimperfections Reproducibilitltr withoutline characteri$tics operational withconsistent imiliesthatcircuitscanbe pioducdd or agingtolerances. adjustments A single basicstructurecanbe usedfor a varietyof signaltype.sand Programmability -changing in a digitalmemory. specification or parameilic an algorithmic by'merely applications signalsby circuitcanbe usedfor multiple A singledigitalsignalprocessing Timesharing'. each memoryandprocessing storingtemforaryretuftsbfeachprocessin random-access lashion. signalin a cyclic(time-divided) circuitaredigital. Automatic fesfiSincethe inputsandoutputsof a digitalsignalprocessing storedin to datapatterns testresponses routinely by comparing data,testscanbe perlormed memory. of digitallogic,digital.signal capahilities Because of thedecision-making versatititr. withanalog or impractical thatareimpossible processing canporformmanyfunctions implementations.

intedaces. The aclaptationlogic and delay requirementsof a switched network echo canceller virfually preclude any type of analog implementation.

Tone Receivers Detectionof DTMF, MF, SF,or otheranalogtonesis easilyandeconomicallyrealized for the explicitpurposeof by convertingthe analogsignalsto digitalrepresentations is evenmoreeconomicalwhen detectingthe tone.Of course,a DSPimplementation thetonesarealreadydigitized,whichis thecasewithin a digitalswitch.Theprogrammability featureof a DSPcircuit is particularlyusefulfor tonereceiversbecauseone canbeusedfor multiplefunctionsby selectingdifferentfilhardwareimplementation ter options(programs)dependingon the application[10]. Hlgh-Spead Modems (e.g.,?8.8-kbps)voicebandmoReliableoperation(low bit errorrates)of high-speed modulation dems[tI] over the switchedtelephonenetworkrequiressophisticated signalconditioningreferredto techniques(describedin Chapter6) anttsophisticated Theonly practicalway to implementthesefunctionsis with asadaptiveequalization. DSPcircuitry.Referenceil21 describesan earlyapplicationof DSPto a 14,400-bps the useof DSPfor adaptiveequalizationof a 400modem.Reference[13] describes Mbps digital radio.Previousdigital radiosusedanalogadaptiveequalizersbecause equalizthey werecheaper.Very-high-ratedigital radiosrequiremoresophisticated possible) DSP. with ers,whichareeasierto implement(perhapsonly Low-Bit-Rate Voice E ncodi n g The realizationof low-bit-ratevoiceencodingalgorithmsdescribedin Chapter3 involve$extensivenumericalprocessingto removeredundancyin the digitizedvoice samples.DSPtechnologyis the only economicalmeansof implementingthesealgorithmson a real-timebasis.References [14], [15], and[16] describeDSPimplemenReference respectively. coders, voice and4.8-kbps [17]describes tationsfor32-,16-, to voice compression' of DSP application general theory and moreof the

84

WHYDIGITAL?

2.3 DISADVANTAGES OF DIGITALVOICENETWORKS The first pafr of this chapterdiscussed the basictechnicaladvantages of digital networks.To balancethe discussion, this sectionreviewsthe basictechnicaldisadvantagesof digitalimplemenrarions aslistedin Table2.3. 2.3.1 IncreaeedBandwidth In thebrief introductionto voicedigitizationpresented in Chapter1,mentionis made thattransmission of samplesof ananalogwaveformrequiresno morebandwidththan theunderlyingwaveform(at leastin theory).Thebandwidthexpansioncomeswhen thesamplesareencodedintobinarycodesandffansmittedwith anindividualpulsefor eachbit in the code.Thus a Tl systemrequiresapproximatelyeight timesas much bandwidthasdo 24 analogvoicechannelssinceeachsampleis represented by an8-bit codewordandeachbit is transmittedasa separate discretepulse.Althoughmoresophisticateddigitizationalgorithmscanbe usedto encodevoiceat a lowerbit ratethan thatusedon Tl systems(64 kbps),eventhemostsophisticated algorithms(described in chapter3) cannotprovidecomparable voicequalitywithoutat leasta rwo-to-one bandwidthpenalty. In someportionsof theanalognetwork,suchasthelocalloops,thebandwidthincreasedid not represent muchof a penaltysincethe inherentbandwidthwas(andis) underutilized. In long-haulradiosystems, however,bandwidthwasat apremium,and digital systemswererelativelyinefficientin termsof the numberof voicechannels provided.one mitigatingaspectof a digital radio systemis its ability to overcome higherlevelsof noiseandinterference, which sometimes providescompensation for the bandwidthrequirements, particularlyin congestedtransmissionenvironments wheremutualinterference canbecomea limiting consideration [3]. The inherentrobustnessof a digital systemwith respectto interference is oneimportantattributeof digitalcellularsy$tems describedin Chapter9. Thebandwidthpenaltyimposedby voicedigitizationis directlydependenr on the form of transmission codingor modulationused.With greatersophistication in the modulation/demodulation equipment,greaterefficiencyin termsof the bit ratein a givenbandwidthis achievable. Basically,greatertransmission efficiencyis achieved by increasingthenumberof levelsin theline code.With limitedtransmitpower.however,thedistances betweendiscretesignallevelsin thereceiverarereduceddramatiTABLE2.3 Dlsadvantagesof Digitallmplementatlona 1. Increased bandwidth 2. Needfortimesynchronization 3. Topologically restricted muttiplexing 4. Needfor conference/extension bridges 5. Incompatibilities withanaloglacitities

VOICE NETWORKS85 OFDIGITAL 2.3 DISADVANTAGES cally. Thus, the transmitted signal is no longer as immune to noise And other imperfections as it is with lower information densities Using a combination of advanceddigital modulation, lower rate digitization, and error-correcting codes,point-to-point digital radios could provide voice channel efficiencies comparableto or even better than analog microwave systems.Full development along these Iines never occuffed, however, becausethe emergenceof optical fiber transmission eliminated the incentive to do so.

2,3.2 Needfor TimeSynchronlzation Wheneverdigitalinformationis transmittedfrom oneplaceto another,a timing reference,or "clock," is neededto controlthehansfer.Theclock specifieswhento sample the incomingsignalto decidewhich datavaluewas transmitted.The optimum to themiddleof thetransmittedpulses.Thus,for opiampletimesusuallycorrespond to thepulsearrivaltimes'In clock mustbe synchronized timumdetection,the sample thedigitalsignalis not for detecting reference timing ofa local general,thegeneration neededto establish considerations the design some of aimcutt. Chapter4 discusses linft' transmission a digital receiver of propersampleclockingin the links More subtleproblemsanse,however,whena numberof digitaltransmission elethe individual must Not only to form a network. andswitchesareinterconnected networkwide certain but also mentsof thenetworkmaintaininternalsynchronization, can beforetheindividualsubsystems procedures mustbe established synchronization resynchronization basic network these properly.Chapter7 discusses interoperate quirements andimplementations. is not uniqueto digital nefivorks'Single' Theneedfor someform of synchronization for carriersynchrosystemspresentsimilarrequirements FDM transmission sideband synchronization the nization in analog networks.In analog sy$tems,however, ' I 8] arelesscriticalby abouttwo ordersof magnitudeI requirements 2.3.3 Topologically Restricted Multlplexing To the generalpublic,themostapparentuseof multiplexingis broadcastservicesfor radioandtelevision.In thesesystemstheairspaceis sharedby usingFDM of individual broadcastchannels.With this systemthereare no operationalrestrictionsto the confine andreceivers.As long asthetransmitters geographic locationof transmitters sea sufficiently theii emissionsto their assignedbandwidthandeachreceiveruses mutual inwithout lective filter to passonly the desiredchannel,the networkoperates to applications terference.On the other hand, TDM is not nearly as amenable Sincethe time of ardvalof datain a involvingdistributedsource$anddestinations. on the distanceof havel, distributedTDM systemsrequirea time slot is dependent betweenthe guardtime betweentime slots.FDM systemsalsorequireguardbands Thewidth ofthe FDM guardbands, channelseparation. channelsto achieveadequate In a TDM on the geographiclocationof the transmitters. however,is not dependent betweentransseparation asthegeographic $ystemtheguardtimesmustbe increased

86

WHY DIGITAL?

mitters increases.Furthermore,eachtime division sourcemust duplicate the synchronization and time slot recognition logic neededto operatea TDM $ystem.For these reason$,TDM has been usedprimarily in applications(e.g., interoffice trunks) where all of the information source$are centrally located and a single multiplexer controls the occurrenceand assignmentof time slots. Time division multiple access(TDMA) satellites and cellular systemsare examples of applicationsof TDM for distributed sources.These systemsuse sophisticated synchronizationtechniquesso that each ground station or mobile unit times its hansmission to arrive at the satelliteor basestation at precisely defined times, allowing the use of small guard times between time slots. Notice that these applications involve only one destination; a satellite or a base station. If an application involves multiple, distributed solurcesand destinatians (with transmissionin more than one direction), larger guard times are unavoidable. Figure 2.5 shows such an application but uses FDM insteadof TDM. The main engineeringconsiderationfor this systemis to ensure that the FDM channelshave sufficient isolation to allow a high-powered sourceto be adjacentto a receiverwith the worst-casereceivelevel. obviously, adequateFDM isolation require$a certain amount of bandwidth overhead,but it is usually fairly easyto design filters with adequateisolation for a large rangeofsignal levels so distanceconsiderationsare minimized.

2.3.4 Needfor Conference/Extension Brldges The processof combining multiple analog signals to form a conferencecaII or function as multiple extensionson a single telephoneline can be accomplishedby merely bridging the wire pairs togetherto superimposeall signals.Nowhere is this more convenient than when multiple extensions share a single two-wire line, as indicated in Figure 2.6. When digitized voice signalsare combined to form a conference.either the signals must be converted to analog so they can be combined on two-wire analog bridges or the digital signals must be routed to a digital conferencebridge, as shown in Figure 2.l.The digital bridge selectively adds the (four-wire) signals together (using digital signal processing)and routes separatesumsback to the confereesas shown. FDM srbchrnnels

Multipoint trmcni$ion line

Figure 2.5 Frequencydivisionmultiplexingon distributedmultipointline,

VOICENETWORKS 87 OF DIGITAL E.3 DISADVANTAGES

to two-wireline' connected Figure 2.6 Analogtelephones When conferencing is implemented in associationwith a switching system, the needfor a digital conferencebridge is not much of a disadvantageand in fact can significantly improve the quality of a conferenceby eliminating echoesand signal loss causedby power division. However, when digital extensionsneedto have their signals combined so multiple extensionscan be active in a conversation,the need for a centralized bridge can be an onerousproblem. Residentialtelephonewiring typically follows a daisy-chain pattern, as indicated in Figure 2.6. Thus the need to rewire all outlets and install a centralizedconferencebox is a significant impediment to the deployment of digital station equipment in residential applications'

2.3.5 Incompatibllltieswith Analog Faclllties Whendigital equipmentwas first usedin privateandpublic telephonenetworks,it analoginterfacesto therestof thenetwork.Sometimes providedstandard necessarily The foremostexa majorcostof the digitat subsystem. theseinterfacesrepresented loop analogsubscriber ampleof thissituationarosein digitalendoffices.Thestandard in Chapter1 is particularlyincompatiblewith electronicswitching interfacedescribed its (analog or digital).Anotheraspectof digital switchingthatcomplicates machines mahix' digital a typical inserted by delay aftificial is the usein analogenvironments in Chapter5' Both of theseaspectsof digital switchingarediscussed is to usedigital subinterface the analog problems with Oneway to eliminatethe investmentin the overwhelming Unfortunately, scriberloopsanddigital telephones. of digital deployment a widespread complicates the loop plantfor analogtelephones that complicate Practices equipment.Most notableof thelong-established subscriLer

Digital Conference Bridge bridgefor digitaltelephones. Figure 2.7 Useof conference

88

WHYDIGITAL?

a transitionto digital loopsaresinglewire pairs,loadingcoils,bridgedtaps,-highresistance or intermittentsplices,and wiring gaugechanges.The digital subscriber loop systemsdescribedin ChapterI I accommodate mostof the aboveimpediments but do sowith very sophisticated DSp circuip.

REFERENCES I sft-15-76subscriberRadio,TechnicalDescription, FarinonsR systems,euebec, 2

3 4 5

6 7

I 9 l0 II

12

13

14

Canada,1977. R. J. Hirvela,"The Applicationof computerconkolled pcM swirchingto Automatic call Disfibution," IEEE communicafions ,sysfernsand Technology conference, Dallas,TX, May 1974. M. R. Aaron, "Digital communicarions-The silent (R)evorution?"IEEE Communications Magafine,Jan.197g,pp. 16_26. I. Nasell,"The lg62 survey of Noise and Loss on Toll connections."Bel/,svsrern TechnicalJoumal,Mar. 1964,pp. 697-718. Technical staff, Bell relephone L,aboratories, Transmission systems for communications, westem Electriccompany,winston-salem,North carolina. Feb. 1970. H. J. Hindin, "LSI-BasedDataEncryptionDiscourages the Data Thief,,,Electronics. June21, 1979,pp. 107-120. N. S.Jayant,B. J. McDermott,s. w. chrisrensen, andA. M. Quinn,.-AComparison of Four Methods for Analog speech Encryption," Intemational communication, Conference Record,1980,pp. 16.6,l-16.6.5. A. GershoandR. steele,Ed., "specialIssueon Encryptionof Analogsignals,"rEEE fournal on Selected Areasin Communicariorrs, May 19g4. H. J. Bekerand F. Piper,"speechscrambring,"speechTechnology,Mar,/Apr. r9g7, pp.40-45. A. Fukui andY. Fujihashi,"A single-chip,4-channelMFATFC/PBReceiver,"IEEE GlobecomConference Remrd, 1987,pp. I 2.6.I - I 2.6.4. "A Modem operatingat DatasignalingRatesUp to 2g,gfi)bpsfor use on rheGeneral switchedTelephoneNetwork and on Leasedpoint-to-point2-wire Telephone_Type Circuits,"ITU-T Rec.V.34, Geneva,Switzerland,Sept.1994. T. Kamitake,K. uehara,M. Abe, and s. Kawamura,"rrellis coding 14.4kb/s Dara ModemImplemented with a single-chip High-speedDigital signatilocessor,"rEEE GlobecomConferenceRecord,I 987,pp. I 2.9.I - I 2.9.6. H. Matsue,T. shirato, and K. watanabe,"256 eAM 400 Mb/s MicrowaveRadio system with DSP Fading countermeasures," IEEE International conferenceon Communications, I 988,pp.41.5.l-41.5.6. J' L. so' "Implementarionon an NIC (Nearly Insrantaneous companding) 32 kbps Transcoderusing the TMS320cz5 Digital signal hocessor," IEEE Globecom ConferenceRecord,1988,pp. 43.4.1-43.4.5.

A bridg+'d tap is an unused pair of wires connected at some point to an in-use pair as alother extension or for possible future reassignment of a cable pair.

REFEHENCES89 "softwareConsiderations in theDesignof a 16 l 5 J. L. Dixon. V. Varma,andD. W. Lin, ConferenceRemrd, Globecom IEEE Phone," Portable for a TDMA Coder kbps Speed '5. 1988,pp. 26.7.1-26.7 ,.4.8kbit/s speechcodecusing AdvancedDigital signalhocessors l 6 K. Irie ands. Iai, Record,1987,pp' 20'4't-20'4'5' (DSSP),'IEEE GlobecomConJbrente van Nostrand in communitationssysterns, Processing Digital signal t7 M. E. Frerking, Reinhold,New York, 1994' C' D' Near' l 8 J.E. Abate,E. W. Butterline,R. A. Carley,P' Greendyk,A' M. Montenegro, '?T&T's New Approachto thesynchronization of s. H. Richman,andG. P. Zampetti, pp. 1989' Apr. Magazine, Networks,"IEEE Comruunications Telecommunication 35-45.

VOICEDIGITIZATION in a varietyof apBecauseof the interestingnatureof the subjectandits usefulness be anareaof intense to and continues plications,thefield of voicedigitizationhasbeen algodigitization of voice types hasproducedmanydifferent iesearch.This research implementation on the primarily dependent rithms.Thechoiceof a particularUpe is impliedby theapplication.Thealgorithmchorequirements costandtheperformance excellentqualityfor all typesof input provides PcM) senfor T1 ,yst*rn, (companded (64 kbps)at whatwasoriginallya rate signals(e.g.,voiceor data)at a moderatedata digital PBXs usedlower *ia"tutt iost. The algorithmsusedin the first-generation at thetime,a because, modulation) (higherratePCM or delta costcodingtechniques sensitive less cost and switchingapplicationwasmoresensitiveto digitalconversion introStates, United in the to qualityorhatu rate.For example,the first digitalPBX of rate 144 at a data PCM auceOUy Rolm Corporationin 1975,useduncompanded adYanSubsequent the time PCM at [1]. it wasiheaperthancompanded kbpsbecause drop in the tagesderivedfrom integratingtransmissionand switchinganda dramatic"switchingof the use obsolete "o*t of Tl-.ompatibledigitalvoicecodershavemade Tl-compatible of high-volumeproduction, only" voicedigitizationalgorithms.Because applications in switching used (ICs) (codecs)canbe integratedcircuits coder/decoder a more today, designed to be withouta costpenalty.In fact,if thedigitalnetworkwere 64 below rate significanfly complicatedbut economicallyviablecodecwith a dala kbpswouldprobablybe utilized. applicationswith strict bandwidthlimits suchas high frequency Transmissions voicedigitizationalgo(HF) or digitalcellularradiorequiremuchmoresophisticated in reducingthedata a help rithmsto achievedatarateson theorderof 8-16 kbps'As relaxedasmuch as are also rate,the performancerequirementsof theseapplications the applicationallows. enotttet applicationfor digitizedvoice is voice storagesystems-eitherfor reDigital storageis particularlyapProor for voicemessaging' cordedannouncements becausethe playbackquality doesnot deteriorate priatefor recorded€Innouncements storedin memoryor on a compactdisc(CD) *ittt ti*" andindividualannouncements with limited memoryis an exampleof an SpeechStorage canbe randomlyaccessed. applicationthatcanu$every low ratedigitizationalgorithmswith significantquality

sl

92

votcEDtctlzATtoN

reductions. The original speak-and-spell learning machine developed by Texas In_ struments,for example, stored words with a few hundred bits per word, representing a datarate of about 850 bps [z]. In a similar example requiring better quality rhe same encoding algorithm (LPC) was used in the voice Aleft sysrem of chrysler automobiles to store 20 secof speech(40 words) in 32,000 bits of read*only memory-a data rate of 1600bps [3]. The main reasonvoice messagingsystemsusedigital storageis to have random ac_ cessto the individual messages.To minimize $toragespace,these systemstypically use 8*32-kbps data rates. At the upper end of digital storageapplicationsarehigh-fidelity recordingsof voice and music. Many of the sameadvantagesof digital transmission,as opposedto analog transmission,also apply to digital recordings.Foremostamong theseadvantagesis the abiliry of defining the fideliry at the rime of recording and maintaining the quality in_ definitely by periodically copying (regenerating)the digitally stored Informarion before inevitable deterioration produces bit errors. Thus a high-quality (high-bit_rate) digital recording of Bing crosby, Ervis presley, or Luciano pavarotti (depending on your taste in music) can be savedfor posterity. This feat could not be accomplished with analog recordings no matter how well cared for or preserved.As an example of high-fidelity audio recording, compact disc players [4] record two channelsof audio at 705 kbps each. Speechanalysisand synthesismake up anotherareaofwidespread researchclosely relatedto voice digitization. [n fact, some of the lowest bit rate voice encoders and decoders use certain amounts of analysis and synthesisto digitatly representspeech. In its ultimate form, however, analysisand synthesishave unique goals ana applications fundamentally different from those of general voice digitization. Basically, goals of analysis and synthesis are to recognize words [5] or produce machine-generated speech(e.g.,text-to-speech)[6]. one approachto analyzing speechis to processwaveforms with the intent of recognizing speechphonemes-the basic units of speechfrom which spokenwords are constructed.once the phonemeshave been identified, they are assignedindividual codewordsfor storageor transmission.A synthesizercan then genera:tespeechby recreating the combinations of phonemes.Analysis of this techniqueindicates that the information contentof speechcan be transmittedwith a datarate of 50 bps It must [7]. be emphasized,however, flrat what is hansmiftedis the information content a$sociated with the words themselves,not the more subjective qualities of speechsuch as naturalness,voice inflections, accents,and speakerrecognizability. Thus suchtechniques, by themselves,are not applicable to general telephony, which customarily includes qualities other than the messagecontent of spokenwords. Efficient encoding of facsimile images presentssimilar opportunities and limitations. Facsimile machinestypically scanat 200 dots per inch, which implies there are 3.T4millionbitsofrawinformationonaB.5x ll-in.pieceof paper.If thepapercontains only recognizabletext charactersat l0 charactersand 6 lines per inch, the same information can be encodedas 5610 ASCII characters,ot 3g,27obits, a savings of almost 100 to I. Besidesbeing restricted to text-oriented messages,character-oriented

93 MODULATIOru AMPLITUDE 3.1 PULSE encoding and decoding produces the same output character font independent of the $ource (which could conceivably be hand written). Facsimile machines achieve one significant level of coding efficiency without sacrificing tfansparencyby encoding rhit" ,pu*" into run length codes. Although this does not reduce the number of bits in a worst-case(random-dot) image, it greatly reducesthe number of bits in the average image of interesr. similar processing is possible in voice applications by efficiently encoding silence. However, the voice problem is more complicated because reat-time voice requires reconstructingthe temporal aspectsof the source,restricting silence encoding to relatively large intervals' Another level of speechanalysis involves the actual recognition of spokenwords. High levels of successhave been achieved,with the two main restrictionsthat the system is trained on the speakers and the speakers are trained to speak with isolated goal words. As an example of one implementation [8] that tries to achievethe ultimate recogof speakerindependence,continuous speech,and large vocabularies,7l-96$o nition accuracyis possibledependingon the level of the grammar specified.(A grammar defines allowed $equencesof words.) Voice digitization techniquescan be broadly categorizedinto two classeslthose digitally *n*ding analog waveforms as faithfully as possible and those processing waveforms to encode only the perceptually signihcant aspectsof speech and hearing processes.The first category is representativeof the general problem of analog-toiigitut *O digital-to-analog conversionsand is not restricted to speechdigitization. The three most common techniquesused to encodea voice waveform are pulse code modulation (PCM), differential PcM (DPCM), and delta modulation (DM)' Except in special cases,telephoneequipment designedto transparenflyreproducean analog waveform used one of thesetechniques.Thus, when studying these common waveform encoding techniques,we are, in fact, sfudying the more generalrealm of analogto-digital conversion [9]. The secondcategory of speechdigitization is concernedprimarily with producing very low datarate speechencodersand decodersfor narrowbandtransmissionsystems o. digitul storagedivices with limited capacity. A device from this special class of is commonly referred to as a'*vocoder" (voice coder)' Very low data rate techrr:iques vocoder techniquesgenerally produce unnatural or synthetic sound1e.g.,t-ZOO-Ups) gening speech.As such, low-data-ratevocodersdo not provide adequatequality for eral telephony. A great deal of effort has been expendedto develop medium-rate (e.g.' 8-kbps) voice coders with natural speechqualities, primarily for digital cellular applications' These coders are implemented as a combination or hybrid of the low-bit-rate techniques and the waveform coders. Thus, these techniquesrepresenta third class of voice digitization algorithm.

MODULATION 3.1 PULSEAMPLITUDE timesat a setof discrete is to establish waveform Thefirst$tepin digitizingananalog which the input signal waveform is sampled. hevalent digitization techniques are

94

votcEDtetTtzATtoN

basedon theuseof periodic,regularlyspacedsampletimes.If thesamples occuroften enough,the original waveformcan be completelyrecoveredfrom the samplesequenceusinga low-passfilter to interpolate, or "smoothout,"betweenthesamplevalues. Thesebasic conceptsare illustratedin Figure 3.1. A representative analog waveformis sampledat a constantsamplingfrequency/,= IIT andreconstructed us_ ing a low-passfilter. Notice that the samplingprocessis equivalentto amplitude modulationof a constant-amplitude pulsetrain.Hencethe techniqu"represented in Figure3.1 is usuallyreferredto asa pulseamplitudemodulationqr,lvlr. 3.1.1 Nyqulst Sampting Rate A classicalresulrin samplingsysrems wasestablished in 1933by HarryNyquistwhen he derivedthe minimumsamplingfrequencyrequiredto extractall informationin a continuous,time-varyingwaveform.This result-the Nyquistcriterion-is defined by therelation

l, > (2xBw) wheref = samplingfrequency BW = bandwidthof inpursignal Thederivationof this resultis indicatedin Figure3.2,whichportraysthe$pectrumof theinputsignalandtheresultingspectrumof thepAM pulsetrain.ThepAM spectrum canbe derivedby observingthata continuoustrainofimpulseshasa frequencyspecffum consistingof discreteterm$at multiplesof the samplingfrequency.The input signalamplitudemodulatesthesetermsindividually.Thusa double-sideband spectrum is producedabouteachof the discretefrequencytermsin the spectrumof the pulsetrain. The originalsignalwaveformis recoveredby a low-passfilter designed to removeall but theoriginalsignalspectrum. As shownin Figure3.2,thereconstruc_ tive low-passfilter musthavea cutoff frequencythat lies betweenBW andf, - Bw. Hence,separation is only possibleiff, - Bw is grearerthanBW (i.e.,iffr > zBw).

PAM samplet

Irtllt,,,,, Lour-Fa$ filter

Figure 3.1 Pulseamplitudemodulation.

MODULATIOI'I 95 3.1 PULSEAMPLITUDE

/

|

Inpultpectrum

\

Output filter

B w \ t , /.-sw Figure3.2 Spectrumof PAM signal.

3.1.2 FoldoverDietortion (f, < zBw), the original If the input wavefonnof a PAM systemis undersampled withoutdistortion.As indicatedin Figure3.3,thisoutwaveformcannotberecovered put distortionarisesbecausethe frequencyspectrumcenteredaboutthe samplingfreqo"o"y overlapsthe original spectrumand cannot be separatedfrom the original "folded" backon spectrumby filtering. Sinceit is a duplicateof the input specffum top of thedesiredspecrumthatcausesthedistortion,this typeof $amplingimpairment "foldoverdistortion'" is oftenreferredto as in the desiredfrefoldoverdistortionproducesfrequencycomponent$ In essence, quencybandthatdid notexistin theoriginalwaveform.Thusanothertermfor thisimprrrn""t is "aliasing."Aliasing problemsare not confinedto speechdigitization pr*rrr"r. Thepotentialfor aliasingis presentin anysampledatasystem.Motionpiciuretaking,for example,is anothersamplingsystemthatcanproducealiasing.A comin old Westems-Oftenthe mon exampleoccurswhenfilming movingstagecoaches and wheelmovements, samplingprocessis too slow to keepup with the stagecoach

Distortionenergy

-BW

BW

-3f"

-2f

Input tpeclrum

Figure 3.3 Foldoverspectrumproducedby undersamplingan input,

96

VOICEDIGITIZATION

2 . 5k H z

Figure 3.4 Aliasingof 5.5-kHzsignalinto a 2.5-kHzsignal. spuriousrotation ratesareproduced.If the wheel rotates355" betweenframes,it looks to the eye as if it has moved backward 5". Figure 3.4 demonstratesan aliasing processoccurring in speechif a 5.5-kHz signal is sampled at an 8-kHe rate. Notice that the sample values are identical to those obtained from a 2.5-kHz input signal. Thus after the sampled signal passesthrough the 4-kHz output filter, a 2.5-wlz signal arisesthat did not come from the source.This exampleillustratesthat the input must be bandlimited, hefore sampling, to remove fre_ quency terms greater than j,[, even if thesefrequency terms are ignored (i.e., are inaudible) at the destination.Thus, a complete pAM system,shown in Figure 3.5, must include a bandlimiting filter before sampling to ensurethat no spuriousor source-related signals get folded back into the desired signal bandwidth. The input filter of a voice codec may also be designedto cut off very low frequenciesto remove 60-cycle hum from power lines. Figure 3.5 shows the signal being recoveredby a sample-and-holdcircuit that produces a staircaseapproximation to the sampled waveform. With use of the staircase approximation, the power level of the signal coming out of the reconstructivefilter is nearly the same as the level of the sampled input signal. The respon$eof the reconstructive filter, in this case,must be modified somewhatto account for the spectrum of the wider "staircase" samples.(The modification amountsto dividing the..flat" filter spectrumby the spectrumof the finite width pulse. SeeAppendix C.) The bandlimiting and reconstructivefilters shown in Figure 3.5 are implied to have ideal characteristics.*Since ideal filters are physically unrealizable,a practical implementation must consider the effects of nonideal implementations.Filters with realizable attenuation slopes at the band edge can be used if the input signal is slightty oversampled. As indicated in Figure 3.2, when the sampring frequencyf is somewhargreater than twice the bandwidth, the spectralbandsare sufficiently separatedfrom eachother -An

id-ul filter is one with a frequency-independent time delay (linear phase), no attenuation in the passband(except as might be desired for pulse shaping),an arbitrarily steepcutoff, and infinrte attenuation everywhere in the stopband.

g.t puLSEAMPLITUDE MoDULATtotrt 97

, r l l r , ,,,\

,++

A=

SEmFle cl(sk

PAMsYstem. Figure3.5 End-to-end canbe used.As an example,sampled thatfilters with gradualroll-off characteristics voicesystemstypicallyusebandlimitingfilterswith a 3-dBcutoffaround3.4kHz and attheovera samplingrateof 8 kHz.Thusthesampledsignalis sufficientlyattenuated lap frequencyof 4 kFIzto adequatelyreducetheenergylevel of thefoldoverspectrum. for outFigure3.6 showsa filter templatedesignedto meetITU-T recommendations is proof-bandsignalrejectionin PCM voicecoders.Noticethat 14dB of attenuation videdat 4kllz.

'/L

6'

-r0

.= -tl t'

-t0

''ff.t::.:!:i -tl -lo

.10 r1.00

{000 rnqurrrcy (Hz)

for PCM Figure 3.6 Bandlimiting filter templatedesignedto meetITU-T recommendations coders. voice

98

VOICE DIGITIZATION

As mentioned in chapter l, the perceived quality of a voice signal is not greatly dependentupon the phaseresponseof the channel(the relative delay of individual frequency components).For this reason the phaseresponsesof the bandlimiting filters in the encodersand the smoothing filters in the decodersare not critical. Nonlinear phaseresponsesin thesefilters, however, do impact high-rate voiceband data signals when digitized. Hence a somewhatparadoxical situation ariseswhen voicebanddata are transmittedover a T-carrier line: the processof converting the voicebanddata signal (28,8ffi bps typical maximum) to a virtually error free 64-kbps digital signal causesdistortion in the lower rate data signal. However, becauseofregeneration the transmissionprocessitself does not add to the signal degradation. By interleaving the samplesfrom multiple sources,pAM systemscan be used to sharea transmissionfacility in a time division multiplex manner.As previously mentioned, PAM systemsare not generally useful over long distancesowing to the vulnerability of the individual pulses to noise, distortion, intersymbol interference,and crosstalk.* Instead, for long-distance transmission the pAM samples are converted into a digital format, thereby allowing the use of regenerative repeaters to remove transmissionimperfections before errors result.

3.2 PULSECODEMODULATION The preceding section describespulse amplitude modulation, which uses discrete sample times with analog sample amptitudesto extract the information in a continuously varying analog signal. Pulse code modulation (pcM) is an extension of pAM wherein eachanalog samplevalue is quantizedinto a discretevalue for representation as a digital codeword. Thus, as shown in Figure 3.7, a pAM systemcan be convefied into a PCM system by adding an analog-to-digital (A/D) converter at the source and a digital-to-analog (D/A) converter at the destination. Figure 3.9 depicts a typical quantizationprocessin which a set ofquantization intervals is associatedin a one-toone fashion with a binary codeword. All sample values falling in a particular quantization interval are representedby a single discrete value located at the center of the quantization interval. In this manner the quantization process introduces a certain amount of error or distortion into the signal samples.This error, known as quantizaPAM rampler Digitally encoded

Sdmple clock

to digital

Figure 3.7 Pulsecodemodulation. 'As

discussed in Chapter I 1 the emergence of sophisticated DSP equalization algorithms in V,90 modems enablesPAM ransmission on analog subscriberloops.

3.2 PULSECODEMODULATION 99

of analogsamples' Figure 3.8 Quantization tion noise,isminimized by establishinga large number of small quantizationintervals. Of course, as the number of quantization intervals increases,so mu$t the number of bits increaseto uniquely identify the quantization intervals.

3.2.1 QuantizationNoise A fundamental aspectof the design and developmentof an engineeringproject is the need for analytical measuresof systemsperformance.Only then can a systembe objectively measuredand its cost effectivenesscomparedto alternatedesigns.One of the measuresneededby a voice communicationengineeris the quality of speechdelivered to the listener. Measurementsof speechquality are complicated by subjective aftributes of speechas perceivedby a typical listener. One subjectiveaspectofnoise or distortion on a speech signal involves the frequency content, or spectrum, of the disturbancein conjunction with the power level. Theseeffects of noise as a function of frequency are discussedin Chapter I with the introduction of C-messageand psophometricweighting. Successivequantizationerrors of a PCM encoderare generally assumedto be distributed randomly and uncolrelated to each other. Thus the commutative effect of quantizationerrors in a PCM systemcan be treatedas additive noise with a subjective effect that is similar to bandlimited white noise. Figure 3.9 shows the quantization noi$e as a function of signal amplitude for a coder with uniform quantizationintervals' Notice that if the signal has enough time to change in amplitude by several quantization intervals, the quantization errors are independent.If the signal is oversampled (i.e., sampledmuch higher than the Nyquist rate), successivesamplesare likely to fall in the sarneinterval, causing a loss of independencein the quantizationerrors' The quantization elror or distor"tioncreatedby digitizing an analog signal is customarily expresseda$ an averagenoise power relative to the averagesignal power. Thus the signal-to-quantizing-noiseratio (SQR, also called a signal-to-distortionratio or a signal-to-noiseratio) can be determinedas

100

votcEDtctlzAloN

J

o

o

lnput amplitude

Ouantization error

Figure 3.9 Quantizationelror as a function of amplitude over a range of quantization intervals.

Etfu)l

sQR= -r(r)12} E{Ly(r)

(3.1)

whereE{.} = expectation or averaging x(t) = srulo* input signal y(t) = decodedoutputsignal In determiningtheexpectedvalueofthe quantization noise,threeobservations are necessary: l. The errory(r) - x(r) is limited in amplitudeto 4/?, whereq is theheightof the quantizationinterval.(Decodedoutput samplesare ideally positionedat the middleof a quantizationinterval.) 2' A samplevalueis equallylikely to fall anywherewithin a quantization interva1, implyinga uniformprobabilitydensityof amplitudel/q. 3. signal amplitudesare assumedto be confinedto the maximumrangeof the coder.If a samplevalueexceedstherangeof the highestquantizationinterval, overloaddistortion(alsocalledpeaklimiting) occurs. If we assume(for convenience) a resistance level of I o, the averagequantization noisepoweris determinedin AppendixA as

noir"po*er= d Quantization fr

(3.2)

3.2 PULSECODEMODULATIOT'|101

thequantiIf all quantizationintervalshaveequallengths(uniformquantization), as* the SQR is determined values and the sample of noise is independent zation

lorogrot-+) sQR(db)= =10.8 +201"s,. FqJ

(3.3)

wherev is therms amplitudeof theinput.In particular,for a sinewaveinputtheSQR is producedby uniformquantization

=rorogl,1w#) (dB) sQR

=7.78 +ro"*,. [+J

(3.4)

whereA is the peakamplitudeof the sinewave' is to be digitizedwith Exampte3.1. A sinewavewith a l-V manimumamplitr'rde quantization intervalsare spaced many uniformly a minimumSQR of 30 dB. How each sample? to encode needed,andhow manybits areneeded Solution. Using Equation3.4, the maximum size of a quantizationinterval is determinedas q = (l)10{3F7'78)/20 = 0,078V Thus 13 quantizationintervalsareneededfor eachpolarity for a iotal of 26 intervals in atl. Thenumberof bitsrequiredto encodeeachsampleis determinedas trl = lo9r(26)= 4.7 = 5 bits per sample Whenmeasuringquantizationnoisepower,the spechalcontentis oftenweighted in the samemannerasnoisein an analogcircuit.Unforhrnately,spectrallyweighted do not alwaysreflect the h.ueperceptualquality of a voice ennoise measurements .The

SeRs commonly compaxe unfiltered decoder outputs to unfiltered quantization errors. In actual practice, the decoder output filterreduces the power level ofboth the signal and the noise. The noise power ixperiences a greater feduction than the signal power, since the uncorrelated noise samples have a wider spectrum than the corelated voice samples. Thus filtered signal-to-noise ratios are usually higher thal the values calculated here by l-2 dB'

102

votcEDlctlzATtoN

coder/decoder.If the spectraldistribution of the quantization noise more or less follows the $pectralcontent of the speechwaveform, the noise is masked by the speech and is much less noticeable than noise uncorrelatedto the speech[l0l. on the other hand, ifthe quantizationprocessproducesenergy at voicebandfrequenciesother than those contained in particular sounds,they are more noticeable. High-quality PCM encodersproduce quantization noise that is evenly distributed acrossvoice frequenciesand independentof the encodedwaveforms. Thus quantization noise ratios defined in Equation 3.4 are good measuresof pcM performance.In some of the encoders discussed later (vocoders in particular), quantization noise power is not very useful. Referencestgl, I I I l, and I I z] describeother measuresof encoder speechquality providing bettercorrelationsto quality asperceivedby a listener.

3-2.2 ldle ChannelNoise Examination of Equations 3.3 and 3.4 reveals that the SeR is small for small sample values.In fact, as shown in Figure 3.10, the noise may actually be greaterthrurthe signal when samplevaluesare in the first quantizationinterval. This effect is particularly bothersomeduring speechpausesand is known as idle channelnoise. Figure 3.I t depicts one method of minimizing idle channel noise in pCM systemsby establishinga quantizationinterval that straddlesthe origin. In this caseall samplevaluesin the central quantization interval are decodedas a constantzero output. pCM systemsofthis type usean odd number of quantizationintervals sincethe encodingrangesof positive and negative signals are usually equal. The quantization characteristicsrequired to produce the output waveforms shown in Figures3.10 and 3.ll are shown in Figures3.12 and 3.13, respectively.The first characteristic(midriser) cannot produce a zero output level. The secondcharacteristic (midtread) decodesvery low signalsinto constant,zero-level outputs. However, if the signal amplitude is comparableto the size of the quantization interval or if a dc bias exists in the encoder,midtread quantization will produce about as much idle channel noise as midriser quantization. As mentioned in chapter l, noise occurring during speechpausesis more objectionable than noise with equivalent power levels during speech.Thus idle channel

2.0

r.0

-1.0 -2.O

Figure 3.10 Idle channelnoiseproducedby midriserquantization.

3.2 PULSECODEMODULATION 103

Figure 3.ll

Elimination of idle channel noise by midtread quantization.

from quantizationnoise,whichis specinoiseis specifiedin absolutetermsseparate fied relativeto the signallevel.For example,Bell systemD3 channelbankspecificationslist themaximumidle channelnoiseas23 dBmCO[13]. 3.2.3 Uniformly Encoded PCM quantizationintervalsfor all samplesproducescodeAn encoderusingequal-length wordslinearlyrelatedto the analogsamplevalues.Thatis, the numericalequivalent In this of eachcodewordis proportionalto the quantizedsamplevalueit represents. converterto genanalog-to-digital mamera uniformPCM sy$temusesa conventional eratethe binarysamplecodes.The numberof bits requiredfor eachsampleis deternoisepower.Minimum digitizedvoice quality minedby the maximumacceptable ratioin excessof 26 dB [ 14].For a uniformPCM systemto requiresa signal-to-noise achievea sQR of 26 dB, Equation3'4 indicatesthat4*o = 0' l23A' For equalpositive andnegativesignalexcursions(encodingfrom -A to A), this resultindicatesthatjust over l6 quantizationintervals,or 4 bits per sample,arerequired.*

Flgure 3.12 Midriserquantizercharacteristic. .This

SeR objective is for minimum acceptable perform$nce and assumesall degadations occur in a sin gle encoder, If additional sigrral impairments occur (such as multiple A/D conversions), the encoder must use more bits to ptovide noise margin for other elements in the network.

104

votcEDtctTtzATtoN

Figure3.13 Midtread quantizer characteristic. In additionto providingadequate qualityfor smallsignals,a telephone systemmusr be capableof transmittinga largerangeof signalampritudes,referredto asdynamic range.Dynamicrange(DR) is usuallyexpressed in decibelsastheratio of themaximum amplitudesignalto the minimum amplitudesignal:

DR= to**,.fts] ffirn \_

/

=zorog,oti"_il

(3.s)

A typicalminimumdynamicrangeis 30 dB tt4l. Thussignalvaluesas largeas 3l timesA mustbe encodedwithoutexceedingthe rangeof quantizationintervals.Assumingequallyspacedquantizationintervalsfor uniformcoding,thetotalnumberof intervalsis determinedas496,whichrequires9-bit codewords.* Theperformance of ann-bit uniformPCM systemis determined by observingthat M*^*

q =-T -

(3.6)

whereA.u* is the manimum(nonoverloaded)amplitude. substitutingEquation3.6into E4uation3.4producesthepcM performance equa. tion for uniformcoding:

seR=116+6.ozn+ 2olog,o t^*l

(3.7)

This rcsult is derived with the assumption of minimum performance requirements. Higher performance objectives (less quantization noise and grcater dynamic range) require as many as I 3 bits per sample for uniform PCM systems. This coding performance was established when it was likely that multiple conversions would occur in an end-to-end connection. Now that the possibility of multiple ArD and D/A conversionshas been eliminated, end-to-endvoice quality is much better than it was in the analog network,

105 MODULATIoN 3.2 PULSE CODE Thefirst two termsof Equation3.7providetheSQRwhenencodinga full-rangesine wave.Thelasttermindicatesa lossin SQRwhenencodinga lowerlevelsignal.These in Figure3.14,which showsthe SQRof a uniformPCM relationships arepresented sy$temasa functionof thenumberof bits per sampleandthe magnitudeof an input sinewave. Example3.2. What is the minimumbit rate that a uniform PCM encodermust provideto encodea high-fidelityaudiosignalwith a dynamicrangeof 40 dB?Assume dictatepassageof a 20-kHz bandwidthwith a minimum the fidelity requirements sinusoidalinput signals' ratio of 50 dB. For simplicity,as$ume signal-to-noise Solutinn. To preventfoldoverdistortion,the samplingratemustbeat least40 kHz. to thatusedin D-typechannelbanks Assuminganexcesssamplingfactorcomparable (4000/3400),we choosea samplingrateof 48 kHz as a compromisefor a practical bandlimitingfilter. By observingthata full-amplitudesignalis encodedwith an SQR of 40 + 50 = 90 dB, we canuseEquation3.7to determinethenumberof bitsn required to encodeeachsample: '"

n - G

90 - 1.76 6.02

= 15 bits Thustherequiredbit rateis = 720kbps (15 bits/sampleX48,O00 samples/sec)

e [ *

so t EI

.E 40 E F

o

Teo o c

a - 40

-30 A/A6.\

-20

- l0

(dB)

Figure 3.14 SQRof uniformPCM coding.

106

vorcEDlcrrzATroN

3.2.4 Companding In a uniform PCM systemthe size of every quantization interval is determinedby the SQR requirement of the lowest signal level to be encoded.Larger signals are also encoded with the same quantization interval. As indicated in Equation 3.7 and Figure 3.14, the SQR increaseswith the signal amplitudeA. For example, a Z6-dB SeR for small signals and a 30-dB dynamic range producesa 56-dB SQR for a maximumamplitude signal. In this manner a uniform PCM system provides unneededquality for large signals.Moreover, the large signalsarethe leastlikely to occur. For thesereasonsthe code spacein a uniform PCM systemis very inefficiently utilized. A more efficient coding procedureis achievedif the quantization intervals are not uniform but allowed to increasewith the sample value. When quantization intervals are directly proportional to the sample value, the SQR is constantfor all signal levels. with this technique fewer bits per sample provide a specified seR for small signals and an adequatedynamic rangefor large signals.When the quantizationintervals are not uniform, a nonlinear relationship exists between the codewords and the samples they represent. Historically, the nonlinear function was first implemented on analog signalsusing nonlinear devices such as specially designeddiodes [15]. The basic processis shown in Figure 3.15, where the analog input sample is f,rrstcompressedand then quantized with uniform quantizationintervals.The effect of the compressionoperationis shown in Figure 3.16. Notice that successivelylarger input signal intervals are compressed into constant-lengthquantizationintervals. Thus the larger the samplevalue, the more it is compressedbefore encoding. As shown in Figure 3.15, a nonuniform PCM decoder expandsthe compressedvalue using an inversecompressioncharacteristicto recover the original samplevalue. The processof first compressingand then expanding a signal is referred to as compandlng. when digitizing, companding amounts to as* signing small quantization intervals to small samplesand large quantization intervals to large samples. various compression-expansion characteristics can be chosen to implement a compandor.By increasingthe amount of compression,we increasethe dynamic range at the expenseof the signal-to-noiseratio for large-amplitude signals. one family of compression characteristicsused in North America and Japan is the p-law characteristic, defined as

Compressed digital codewords Compre$ion

Lin6ar PCM encoder

Linear PCM decoder

Expansion

Figure 3.15 Companded PCM with analogcompression andexpansion.

g,E PULSE coDEMoouLATtoN 107

rot t E

E t*rt 9 E

E",,'8 oro$

Input sfirple vrlus

Figure 3.16 Typicalcompression characteristic. ..ln(l+ULtl) Fr(r)=sen(x) ffi[y

(3.8)

wherex = input signalamplitude( -1 {x { 1) sgn(x) = polarity ofx F = pllrameter used to defirneamount of compression Becauseof the mathematicalnatureof the compressioncurve, compandedPCM is sometimes referred to as log-PCM. A logarithm compression curve is ideal in the sensethat quantizationintervals and, hence,quantizationnoise are proportional to the sample amplitude. The inverse or expansioncharacteristicfor a p-law compandor is defined as

1, F'r(v):'s"0) +p;rrr[iJr(1

(3.e)

where ) = thecompressed value,=Fp(x)(-l < y < l) $gn(y) polarityofy parameter F =' companding Thefirst T-carrjersystemsin theUnitedStatesusedDl channelbanks[16], which approximated and expansionfunctions Bquation3.8 for F = 100.The compression wereimplemented with thespeciallybiaseddiodesmentionedpreviously.Figure3.17 depictsa blockdiagramof a Dl channelbank.Noticethatthetimedivisionmultiplexon analogPAM samples. Thusthe ing anddemultiplexingfunctionsareimplemented compandingandencoding/decoding functionsweresharedby all 24 voicechannels.

108

vorcEDtetT|zATtoN

-+E Tl transmirsion line

Analog Inpuli

I a t

Bandlimiting filterB

Figure 3.17 Functionalblock diagramof Dl channelbank. The ability to sharethis relatively expensive equipment was one of the reasonsthat PCM was originally chosenas the meansof digitally encoding speech.subsequentdevelopment of integratedcircuit PCM codecsdiminished the need to sharethis equipment, Thus later generation$ystemscould use per-channelcodecsand provide more flexibility in implementing various $ystemsizes.when most of the cost of a channel bank is in the common equipment,as in the original channelbanks,lessthan fully provisioned sy$temsare overly expensive. Each sample produced by a Dl channel bank was encodedinto 7 bits; I polarity bit and 6 compressedmagnitude bits. In addirion, I signaling bit was added to each channel to produce an 8-bit codeword for eachtime slot. since the sampling rate was I kHz, a 64-kbps channelresulted. Even though the Dl channel banks have been su* percededby newer channel banks utilizing a different coding format, the 64-kbps channel rate has persistedas a standard.

3.2.5 Eaeily Dlgltally LinearlzabteGoding The success of the flrst Tl sy$temsfor interofficeexchangeareatransndssion paved thewayfor fuftheruseof digitaltimedivisionmultiplexsystems. As it becameapparent that digital transmission wasusefulwithin the toll network,it alsobecameclear thattheform of PCM encodingusedin theDl channelbankswasinadequate. In contrastto the exchangearea,an end-to-endconnectionthroughthe toll networkcould haveconceivablyinvolvedasmmy asninetandemconnections. Sincedigitalswitching wasnot in existenceat the time that T-carier systemsfor the toll networkwere beingdeveloped,eachof thesetandemconnection$ implied an extraDiA and A/D

3,2 PULSE CODE MODULATION109 conversion. Thus the quality of each conversion had to be improved to maintain the desired end-to-endquality" The D2 channel bank [ 7] was therefore developedwith improved digitized voice quality. When the D2 channelbank was being developed,digital switching was recognized as a coming technology, implying that channel banks would be paired on a dynamic basis, as opposedto the one-to-onebasis in T-carrier systems.Thus a greater degree of uniformity in companding characteristicswould be required to allow pairing of channelbankson a nationwide basis.The main featuresincotporatedinto the D2 channel bank (and ensuingchannelbanks suchas D3, D4, and D5) to achievethe improved quality and standardization are: L Eight bits per PCM codeword 2. Incorporation of the companding functions directly into the encoder and decoder 3. A new companding characteristic(p255)

TheDl channelbanksuseI bit pertimeslotfor signalingand7 bitsfor voice.Thus for basic which wasmorethannecessary a signalingrateof 8 kbpswasestablished, voiceservice.To providea higherdataratefor voice,signalingbetweenD2 andall subsequent channelbanksis insertedinto the leastsignificantbit positionof 8-bit in everysixthframe.Thuseverysixthp255PCM codewordcontainsonly code-words 7 bits of voiceinformation,implying that the effectivenumberof bits per sampleis actually7f bits insteadof 8. Theuseof 1 bit in everysixthframefor signalingis often signalingis utilized,the referredto as"robbedbit signaling."Whencommon-channel T-carriersystemsno longerneedto carrysignalinginformationon a perassociated channelbasisanda full I bits of voicecanbe transmittedin everytime slot of every foame. of theDl channelbankswereimandexpansioncharacteristics The compression plementedseparatelyfrom theencodersanddecoders.The D2 charurelbankincorpoIn these themselves. operationsinto theencodersanddecoders ratesthe companding ofnonuniformlyspaced channelbanksa resistorarrayis usedto establisha sequence comparingtheinput A samplevalueis encodedby successively decisionthresholds. quantization interval ofdecisionthresholds until theappropriate valueto thesequence theparticular whatevercodeis usedto represent is located.Thedigitaloutputbecomes quantizationinterval.(SeeAppendixB for a detaileddescriptionofthe directencodfuncusedin theD? channelbanks.)By incotporatingthecompanding ing procedure tions directly into the encodersand decoders,D2 channelbanks avoid certain sensitivityof comproblemsassociated with parametervariabilityand temperature Dl pandingdiodesin channelbanks. The D2 channelbanksalsointroducedimprovedperformancein termsof theeffect of channelerrorson thedecodedoutputs.Of paramountconcernin a PCM systemis theeffectof a channelenor in themostsignificantbit positionof a codeword.Bit errorsin otherpositionsof a codewordaremuchlessnoticeableto a listener.A channel errorin themostsignificantbit of a codewordproducedby a Dl channelbankcauses anoutputerrorthatis alwaysequalto one-halfof theentirecodingrangeof thecoder'

11O

Vo|cEDIGITIZATIoN

The D2 channel bank, on the other hand, usesa sign-magnitudecoding format. With this format, a channel error in the polarity bit causesan output error that is equal to twice the samplemagnitude (i.e., the polarity is inverted). In the worst casethis enor may correspondto the entire encoding range.Maximum-amplitude samplesare relatively rare, however, so most channel error$ in a D2 coding format produce outputs with error magnitudeslessthan one-half the coding range.Thus, on average,the signmagnitude coding format of the D2 and ensuingchannelbanks provide superior performance in the presenceof channel enors I I 7]. In addition to a need for improved voice quality, it also became apparentthat as more of the network beganusing digital techniques,it would be necessary,or at least desirable,to implement many signal processingfunctions directly on digital signals and not convert them to an analogformat for processing.Most signal processingfunctions (such as attenuating a signal or adding signals together) involve linear operations. Thus before processinga log-PCM voice signal, it i$ necessaryto convert the compressedtransmissionformat into a linear (uniform) format. To simplify the conversion process,the particular companding characteristicwith =255 was chosen.This compandingcharacteristichas the property of being closely F approximatedby a set of eight straightJine segmentsalso refened to as chords. Furthermore,the slopeof eachsucces$ivesegmentis exactly one-half the slopeof the previous segment.The flust four segmentsof a p255 approximation are shown in Figure 3.I8. The overall result is that the larger quantization intervals have lengths that are binary multiples of all smaller quantizationintervals.Becauseof this properly, a compressedcodeword is easily expandedinto a uniform representation.Similarly, the uniform representationis easily converted into a compressedrepresentation.*In fact, commercially availablePCM codecsdigitally compressuniform codewordsinsteadof using direct compressedencoding, as in the D2 channel banks. These techniquesuse a uniform encoderwith a relatively large number of bits to cover the entire dynamic rangeof the signal. As describedin Appendix B, the leastsignifrcantbits of large sample values are discardedwhen compressingthe code. The number of insignificant bits deletedis encodedinto a specialfield included in the compressed-codeformat. In this mannerdigital compandingis analogousto expressinga number in scientific notation. As shown in Figure 3.18, each major segmentof the piecewise linear approximation is divided into equally sizedquantizationintervals. For 8-bit codewordsthe number of quantization intervals per segment is 16. Thus an 8-bit p255 codeword is composedof I polarity bit, 3 bits to identify a major segment,and 4 bits for identifying a quantizing interval within a segment.Table 3.1 lists the major segmentendpoints, the quantization intervals, and tlre corresponding segment and quantization interval codes. The quantization intervals and decodedsample values in Table 3. I have been expressedin terms of a maximum-amplitude signal of 8159 so that all segmentendpoints and decoderoutput$are integers.Notice that the quantizing stepis doubled in eachof 'The

inexorable advance of semiconductor technology has obviated much of the ingenuity that weflt into selecting EDL coding fbrmats. Brute-force table look-up conversion between codes using read-only memories (ROMs) is now the most cost-effective approach,

3.2 PULSE coDEMoDULAToN 111

o

.9 ll

$ E

ET E

I

31

95

223 Lineartignal

478

Flgure 3.18 First four segmentsof straight-line approximation to p255 compressioncurve.

linearsegments. It is this propertythat facilitatesthe conversionto eight successive and from a uniform format.A completeencodingtableis providedin AppendixB processto andfrom uniformcodes. alongwith a detaileddescriptionofthe conversion reThe straight-lineapproximationof the p255 compandingcurvei$ sometimes arisebecause,although ferredto as a lS-segmentapproximation. The 15 segments nearesttheorithe2 segments and8 negativesegments, thereare8 positive$egments gin arecolinearandthereforecanbe consideredas l. Whenviewedin this manner, the middlesegmentcontains3l quantizationintervalswith I segmentstraddlingthe origin (from -1 to +1 in Table3.1).Codewordsfor this middlequantizationinterval ariseasa positivevaluelessthan+1 or a negativevaluegreaterthan-1. Thereare,in in Table3.1,thesevalues effect,a positivezeroanda negativeeero.As represented areencodedas00000000and 1ffi00000,respectively.However,p255 PCM codecs The smalleramplitudesignals,with mostly0 invert all codewordsfor transmission. bits in the segmentcode,aremostprobableandwouldthereforecauseles$than50% pulseson thetransmission line. Thedensityof pulsesis increased by inversionof the transmitteddata,which improvesthe timing and clock recoveryperformanceof the receiving circuitry in the regenerativerepeaters.Thus the actual transmittedcode111111I1 wordscorrespondingtoapositivezeroandanegativezeroarerespectively an idle channel. for the strong timing line signal of and01111II 1, indicating content p255 in the Tl repeaters, PCM ln the interestof ensuringclock synchronization codecsaltertlrc ffansmitteddatain oneotherway.As indicatedin Table3.1,a maximum' amplitudenegativesignalis all I s, whichwouldnormallybe convertedto all 0's for

I 12

vorcEDrcrrzATtoN

TABLE 3.1 Encodlng/Decoding Table for p2SSPCMd

Input Amplitude Range

Step Size

0-1 1-3 3-5 ; 29-31 31-35 ; 91-95 95-103

Segment Code S

000

Quantization Code Q Code Varue

0000 0001 0010

Decoder Amplitude 2 4

1111 0000

30 33

1111 0000 : 1111 0000

9; 99

001

010 21F.223 2?3-239 16

011

463479 479-511

219 231

rrir

471 495

0000 32

100

oEo-ool 991-1 055

64

1111 000

975 1023

1111 0000

1983 2079

101

r ss1-ZOrS 201ts2143 110 3935-4063 4063-431I

1111 0000

111 112

3999 4191

1111

127

8031

:

256 7903-8159

this tabledisplaysmagnitud€ Bncoding only.Polarity bitser6a8signed a "0"forpositiveenda ,'1"fornegatlve. h transmission ell bitsareinverted.

transmission. Instead, for the all-0 codeword only, the second least significant bit is

setto I sothat00000010is transmitted. In effect,anencodingerroris producedto precludeanall-Ocodeword.Fortunately, maximum-amplitude samples areexffemelyunlikely sothatno significantdegradation occurs.(If theleastsignificantbit wereforced to a 1,a smallerdecodingerrorwouldresult.However,in everysixthframethisbit is "stolen"for signalingpu{poses andthereforeis occasionally setto 0 independently of thecodeword.To ensurethat "all-zero"codewordsarenevertransmitted. thesecond leastsignificantbit is forcedto a I whennecessary.) Example3.3. Determinethe sequenceof codewordsfor a p255 PCM encoded channelbankformatrepresenting a half-maximum-power l-kHz digital signal.

3,2 PULSE CODE MODULATION113 Solution. Since the samplingrate of the standardp255 PCM channelbankis 8 kHz, a sequenceof eight samplescan be repeatedin cyclic fashion to generatethe l-kHz waveform. For convenience,the phasesof the samplesare chosento begin at 22.5o. Thus the eight samplescorrespond to 22.5", 67 .5", II2.5o , l5'l .5", 202.5", 247.5", 292,.5",and 337.5o.With thesephases,only two different magnitudescoresponding to 22.5oand 67.5' are required. The maximum amplitude of a half-maximum-power sine wave is 0.707 x 8159 = 5768. Thus the two amplinrdescontained in the sample sequencesare

= 5329 (5768)sin(67.5o)

(5768)sin(22 .5")= 2207

Using the encodingtablein AppendixB, we determinethe codesfor thesesample Thesequence of eightsamples respectively. magnitudes to be 1100001 and1110100, asfollows: cannow be established SamplePhase (deS)

Polarity

2 2 . 5 0 6 7 . 5 0 1 112.5 0 1 157.5 0 1 2 0 2 .1 51 247.5 1 1 2 9 2 .1 51 1 3 3 7 .1 51 1

Quantization

Segment

1 1 1 1 0 1 1 0

1 1 1 0

0 1

0 0 0 1

0 0

0 0

1

U

0

0

1

0 0 0 0 0

0 1 0 0

1

0

1

0 0 1 1

0 0 1

Note: This sequence defines a 1-kHz test signal at a power l€v6l of 1 mW at the transmissionlevel point (0 dBmo). However,the actualtransmltteddata patternis the complementof lhat provid6dabove. because only two amplitudesamples are requiredto produce the test ton6, this tono does not test all encoding/docodingcircuitry. In general, a 1004-Hztone is a better t6st tone since it is not harmonically relatedto an 8000-Hzsamplingrate and will thereforeexerciseall encoder and decoder levels.

Pertormanceof ap255 PGM Encoder As mentioned, the main motivation for changing the encoding algorithm of the Dl channelbank was to provide better speechquality for digital toll network transmission links. The SQR for a maximum-amplitude $ine wave in the first segmentof a p255 codec is determinedeasily from Equation 3.4 as

SQR(A: 31)= 7.78+ 20t"*r. [*l \.- / = 31.6dB The SQRsfor largeramplitudesinusoidsare not as easyto calculatesincethe lengthsof the quantizationintervalsvary with the samplesize.Thusa generalcalculationof thequantizingnoisepowerinvolvesfindingtheexpectedvalueof thepower of the quantizationerrors;

114

VotcEDtctlzATtoN 7

. l _ power=;lW? noise

(3.10)

r'4

wherep; = probabilityof a samplein ith segment p255coding 4; - quantizationsizefor segmenti, =!i+t for segmented UsingEquation3.10,we determinethe SQRpowerfor a full-rangesinusoidas SQR(A= 8159)= 39.3dB For comparison, if all quantizationintervalshadthemaximumlengthof 256asin theuppersegment,trquation3.4providesan SQRof 37.8dB. Thedifferenceof only 1.5dB reflectsthefactthata full-scalesinewavespends67Voof thetime in theupper segmentwherethequantizationintervalsaremaximum(i-e.,pt = 0.67).A voicesignal,on theotherhand,hasa higherpeak-to-average ratiothana sinewave.Theaveragequantizationerroris smaller,but sois the averagesignalpower.Hencethe SQR is approximately the same. Thedynamicrangeof a segmented PCM encoderis determined asthesignalpower differencebetweena low-levelsignaloccupyingtheentirerangeof the first segment anda high-levelsignalextendingto thelimits of thecode.Thusthedynamicrangeof a segmented 255coderis determinedas = 48.4dB DR(A= 3I toA = 8159)= 20log,o(8159/31) In summary,an 8-bitp255PCM codecprovidesatheoreticalSQRgreaterthan30 dB acrossa dynamicrangeof 48 dB. For comparison, Equation3.4or Figure3.14reveals that a uniform PCM encoder/decoder requires13 bits for equivalentperformance. (Theextraqualityof uniformencodingat high signallevelsis unneeded.) Thetheoreticalperformance of an 8-bit segmented F255coderis shownin Figure 3.19asa functionof theamplitudeof a sinewaveinput.Also shownis thetheoretical performance p255 coderand a 7-bit plfi) coderusedin the Dl of an unsegmented channelbank.Noticethatthe 8-bit codersprovideabout5 dB improvementoverthe 7-bit coderfor high-levelsignalsandevenmoreimprovementfor low-levelsignals. Theperformance shownfor the 8-bit codersdoesnot includetheeffectofusing only 7 bits for voice codingin everysixth frame.Whenthis effectis included,the B-bit coderslose I.76 dB in performance. Thescallopedeffectofthe segmented coderoccursbecause thequantizationintervals changeabruptlyat the segmentendpointsinsteadof continuouslyasin analog companding.AIso as shownin Figure3.19,notethe requiredperformance of a D3 channelbankandcodecsdesignedto meetthis specification[13]. This specification assumesall noisemeasurements are madeusing C-message weighting.C-message weightingreducestheeffectivenoiselevelby 2 dB andthereforeimprovestheSQR. Thusan ideal 8-bit p255 coderactuallyexceedsthe specif,rcation by morethanthat

3.2 PULSEcoDE MoDULATIoN 115 Full load rignal = +3d8m0

6

:.E

g o ,t E o

'n E 3

6 -!

Unfiltered pieccYviBe linEdrI-258 Unfiltered

B-bir#-?55

il c .gl th

-50

-40

-30

-20

(dBm0l Signalporrnrof rinewave Figure 3.19 SQR of pJaw coding with sine wave inputs.

shownin Figure3.19.rilhen the leastsignificantbit of everysixth frameis usedfor signaling,however,theSQRis reducedby a comparable1.76dB. A-Law Compandlng recommended by ITU-T for Europeandmost of the The compandingcharacteristic hasthe This characteristic restof the world is referredto asanA-law characteristic. asdoesthe p-law characteristic. advantages samebasicfeaturesandimplementation by shaight-line In particular,theA-law characteristic canalsobe well approximated segmentsto facilitatedirect or digital compandingandcanbe easilyconveltedto and characteristicis definedas from a uniform format.ThenormalizedA-law compression

a u I| flSsn(Jl . . |f-------

0
l" [1+ln(A)l Fa(x)= i | . It+tntt.ll i s-s n'( x- )[l -1- + l n ( A )| _ J f*r*r
( 3 . 1l )

116

votcEDtctlzAlor\

*entrluT@ o
(3.12)

'*olT

pFl(1+|fl(A)l-l

*fustvt(l

where y = F.A($ and A = 87.6. Notice that the first portion of the A-law characteristicis linear by def,rnition.The remaining portion of the characteristic(l/A < lxl < I) can be closely approximatedby linear segmentsin a fashion similar to the trr-lawapproximation. In all, there are eight positive and eight negative segments.The first two segment$of eachpolarity (four in all) are colinear and thereforeare sometimesconsideredas one straight-line segment. Thus the segmentedapproximation of the A-law characteristicis sometimesrefened to as a "13-segment approximation." For easein describing the coding algorithms of the segmentedcompanding characteristic,however, a ld-segment representationis used,just as in the caseofthe segmentedp-law characteristic. The segmentendpoints,quantization intervals, and correspondingcodesfor an 8bit segmentedAlaw characteristicare shown in Table 3.2. The values are scaled/o a maximum value of 4096 for integral representations.Figure 3.?0 displays the theoretical performance of the Alaw approximation and compares it to the performance of a p-law approximation presentedin Figure 3.19. Notice that the A-law characteristic provides a slightly larger dynamic range.However, theAJaw characteristicis inferior to the p-law characteristicin terms of small-signal quality (idle channel noise). The difference in small-signal performanceoccurs becausethe minimum step size of the A-law standardis 2/4096 whereasthe minimum step size of the p-law is 2/8159. Furthermore, notice that the A-law approximation does not define a zero-level output for the first quantizationinterval (i.e., usesa midriser quantizer).However, the difference between midriser and midtread performanceat 64 kbps is imperceptible tl8l.

3.2.6 SyllabicCompandlng A significant attribute of the companding techniquesdescribed for pcM systems is that they instantaneou$lyencompassthe entire dynamic range of the coder on a sampleto-sample basis. Thus p-law and A-law companding is sometimesreferred to as instantaneouscompanding. However, because the power level of a speech signal remains fairly constantfor 100 or more 8-kHz samples,it is wasteful of code spaceto be able to encode very low level signals adjacentto very high level signals.In other words, it is unnecessaryto allow for instantaneouschangesin the dynamic rangeof a signal. One techniqueof reducing the amount of transmissionbandwidth allocatedto dynamic range is to reduce the dynamic range of the signal at the source before encoding and then restorethe original dynamic range of the signal at the receiver while decoding. when the adjustmentsto the dynamic range occur on a periodic basisthat more or lesscorrespondsto the rate of syllable generation,the techniqueis referred to as syllabic companding.Becausethe dynamic range adjustmentsoccur only every 30 msec or so, there is very little bandwidth neededto communicatethe adjustments.

3.2 PULSECODEMODULATION 117 TABLE 3.2 $egmented A-Law Encodlng/Decoding Tablea

Input Amplitude Range

StepSize

Segment CodeS

r2 000

24 3G-32 32-34 : 62-64 64-€8 i 124-128 12B-136 : 248-256 25tsl272 : 496-512 512-544 i 992*1024 1024-1088 : 1984-2048 ?048-2176 i 3968*4096

2

Ouantization CodeQ CodeValue 0000 0001

D€coder Amplitude 1

3

1111 0000

3; 33

1111 0000

63 66

1111 0000

126 132

1111 0000

252 264

1111 0000 : 1111 0000

504 528

001

4

I

16

3e

64

128

010

0 11

100

101

1008 1056

1111 0000

95 96 ; 111 112

2016 2112

1111

127

4032

110

111

alntransmission, everyoth6rbitis invened.

Syllabic companding was first developed for use on noisy analog circuits to improve the i
1 18

VOICEDIGITIZATIoN

6 E

:.E

a q 30 fl g E o

.E .E c A I

€ .F 6

-70

-60

-50

-40

-30

-20

_lo

Siqnelpowr of rinswrve (dBm0l

Ftgure 3.20 SQR of A-law PCM coding with sine wave inputs.

Input porv€r level(dBm)

Output povt/€t level(dBm)

*to[-'"

*7.5-.r0 TrEnsrfiitt€d l6vBl {dBm}

* ' J * f f i u u,rr\s-a, (//w^c.*oi*'

--sH

Variable

Ponnr ftreasurom€nt Figure 3.21

Expanrion

\\ Varieble\ --59 attenuetion

Power meesuteftent Syllabic companding of an analog signal.

3.2 PULSECODEMODULATION 1 19

coders are consideredas part of the transmissionlink, the processof amplifying lowlevel signalsbefore encoding and attenuatingthem after decoding effectively reduces the quantization noise with no net changein signal level. In practice, syllabic companding as implementedin digitized voice terminals doesnot amplify the signal at the sourceand attenuateit at the destination.Instead,an equivalentprocessof controlling the step sizesin the encoderand decoderis used. As far as the transmitted bit stream is concerned,it makes no difference if the signal is amplified and encodedwith fixed quantization or if the signal is unmodified but encodedwith smaller quantization intervals.Thus syllabic compandorsin digital voice terminalstypically reducethe quantization intervals when encoding and decoding low-power syllables but increasethe quantization intervals for high-power syllables' Although syllabic companding can be used in conjunction with any type of voice coding, the techniquehasbeen applied most often to differential sy$temsdescribedin the following sections.In many of the applications,the adaptationtime has been reduced to 5 or I0 msec,which is somewhatshorterthan the duration of a typical syllable (approximately 30 msec). The technique is still generally referred to as syllabic companding, however, to distinguish it from the instantaneousvariety. To adjust the step sizes in the decoder in synchronism with adjustmentsmade in the encoder.some mean$must be establishedto communicatethe step size information from the sourceto the destination.One method explicitly transmits the step size information as auxiliary information. A generally more usefirl approach is to derive the step sizeinformation from the transmittedbit stream.Thus, in the ab$enceof channel erTors,the decoderand the encoderoperateon the sameinformation. This procedure is analogousto syllabic compandedanalog systemsin which the receiver determines its attenuation requirementsfrom the short-term power level of the received signal. In a digital system the bit streamis monitored for certain datapattems that indicate the power level of the signal being encoded.Indications of high power level initiate an increasein the step size, whereasindications oflow levels causea decrea$e. Determining the step size information from the transmittedbit streamis generally better than explicitly transmitting the step sizes for the following reasons.Because there is no explicit transmissionof step size, the transmissionof sampled speechinformation is never intemrpted, and the speechsamplerate is equal to the transmission rate. Also. the bit sheam does not have to be framed to identify step size infbrmation separatelyfrom the waveform coding. Furthermore, if the step size adjustmentsare made on a more gradual basis,the individual incrementsare small enough that occasional incorrect adjustmentsin the receiver causedby channel errors are not critical' However, on transmissionlinks with very high error rates (one error in a hundred bits or so), better decodedvoice quality can be obtained ifthe step size is transmitted explicitly and redundantly encodedfor error correction [19].

3.2.7 AdaptiveGain Encoding example of Figure3.21thedynamicrangeof a signalis In thesyllabiccompanding signalhas36dB of dyreduced by a factorof 2 (in dBm).Thus,if anuncompanded

120

votcEDtctIzATtoN

namic range, the encoder seesonly 18 dB. The 18-dB reduction implies three fewer bits are neededfor dynamic rangeencoding.In the limit, if the power level of alt input signals is adjustedto a single value, no bits ofthe encoderneed to be allocatedto dynamic rangeencoding.A processthat adjustsall signalsto a standardvalue is referred to as automatic gain control (AGC). AGC is traditionally usedon carrier transmission systemsto adjust all received signalsto a standardvalue, thereby removing variations in propagation attentuation.AGC cannot be applied to a sourcevoice signal without allowancesfor speechpauseswhen there is no signal present.Otherwise, idle channel noise would be amplified to the averagelevel of active voice. Notice that with AGC there is no residual information in the power level of the encodedsignal as there is in syllabic companding.To ascertainthe original power level, AGC must be augmented with adaptive gain encoding (AGE), as indicated in Figure 3.22. There are two basic modes of operation for gain encoding dependingon how gain factors are measuredand to which speechsegmentsthe factors are applied. one mode of operation,as implied in Fi gureS.ZL,involvesmeasuringthe power level of one segment of speechand using that information to establisha gain factor for ensuingspeech segments.obviously, this mode of operationrelies on gradually changing power lev* els. This mode of operation is sometimesreferred to a$ "backward estimation." Another mode of operation involves measuring the power lever of a speechsegment and using the gain factor thus derived to adapt the encoderto the samesegment. This approach,referredto as "forward estimation," hasthe obvious advantage,Ihatthe encoderand decoderuse gain factors specifically related to the speechsegmentsfrom which they are derived.The disadvantageis that eachspeechsegmentmustbe delayed while the gain factor is being determined.Although the availability of digital memory has made the cost of implementing the delay insignificant, the impact of the delay on echoesand singing in a partially analog network must be considered.(As long as the subscriberloops are analog, the network is partially analog.) Adaptive gain control with explicit kansmission of gain factors is not without shortcomings.First, when the periodic gain information is inserted into the transmitted bit stream,some meansof framing the bit streaminto blocks is neededso gain in* formation can be distinguished from waveform coding. second, periodic insertion of gain information disrupts information flow, causinghigher transmitterclock ratesthat might be inconveniently related to the waveform sample clock. Third, correct recep-

adiustment

Flgure 3.22 Adaptive gain encoding.

12I 3.3 SPEECH HEDUNDANcIES tion of gain factors is usually critical to voice quality, indicating a needto redundantly encodegain information. Reference[20] describesa modified form of PCM using forward estimationof gain factors that is referred !o as nearly instantaneously companded PCM' The need for transmitting speechsegmentsin blocks is not a disadvantagein the application mentioned (mobile telephone)becauserepetitive bursts with ertor checking are used as a meansof overcoming shortJived multipath fading. This systemprovides a bit rate reduction of 30Vowith respect to conventional PCM. Another example of the use of AGE is the subscriberloop multiplexer (SLM) system developedby Bell Labs [21]. The SLM systembecameobsoletewhen low-cost PCM codecsbecameavailable and the subscribercarrier systemscould be integrated into digital end offices (with SLC 96 and later DLC systems).All of the encoding algorithms describedin the following sectionsuse syllabic companding or AGE in some form to reduce the bit rate'

3.3 SPEECHREDUNDANCIES Conventional PCM systemsencodeeach sampleof an input waveform independently from all other samples.Thus a PCM system is inherently capableof encoding an arbitrarily random waveform whose maximum-frequency component does not exceed one-half the sampling rate. Analyses of speechwaveforms, however, indicate there is considerableredundancyfrom one sampleto the next. In fact, as repofted in reference t101,the correlation coefficient (a measureof predictability) between adjacent$-kHz samples is generally 0.85 or higher. Hence the redundancy in conventional PCM codes suggestssignificant savings in transmissionbandwidths are possible through more efficient coding techniques.All of the digitization techniquesdescribedin the rest of this chapter are tailored, in one degree or another, to the characteristics of speechsignals with the intent of reducing the bit rate. In addition to the correlation existing between adjacent samples of a speechwaveform, severalother levels ofredundancy can be exploited to reduceencodedbit rates. Table 3.3 lists theseredundancies.Not included are higher level redundanciesrelated to context-dependentinterpretationsof speechsounds (phonemes),words, and sen-

TABLE3.3 SpeechHedundancles redundancies Time-domain amplitudedi$tributions 1, Nonuniform 2, Sample-to-samplecorrelations (periodicity) correlatians 3. Cycle-to-cycle conelations 4. Pitch-interval to pitch-interval 5. lnactivityfactors(speechpauses) Frequency-domain redundancies spectral densities long-term 6. Nonuniform spectral densities 7. Sound-specific shorl-term

122

vorcEDrcrrzATroN

tence$.Thesetopicsarenot coveredbecausetechniquesthat analyzespeechwaveforms to extractonly informationcontenteliminatesubjectivequalitiesessentialto generaltelephony. 3.3.1 Nonuniform Amplltude Dietrlbutlons As mentionedin theintroductionto companding, lower amplitudesamplevaluesare morecommonthanhigheramplitudesamplevalues.Most low-levelsamples occuras a resultof speechpau$e$ in a conversation. Beyondthis,however,thepowerlevelsof activespeechsignalsalsotendto occurat the lower endof the encodingrange.The compandingproceduresdescribedin the previoussectionprovideslightly inferior quality(i.e.,lower signal-to-noise ratios)for smallsignalscomparedto largesignals. Thusthe averagequalif of PCM speechcould be improvedby furthershortening lower level quantizationintervalsand increasingupperlevel quantizationintervals. The amountof improvementrealizedby sucha techniqueis minimal andprobably would notjustify the additionalcomplexities.Themosrbeneficialapproachto processingsignalamplitudesin orderto reduceencoderbit rate$involvessomeform of adaptivegaincontrol,asdiscussed earlier. 3.3.2 Sample-to-SampleCorrelatlon Thehighcorrelationfactorof 0.85mentionedin Section3.3indicatesthatanysignificantattemptto reducetransmission ratesmustexploitthe correlationbetweenadjacentsamples.In fact, at 8-kHz samplingrates,significantcorrelationsalsoexist for samplestwo to threesamplesapafr.Naturally,samplesbecomeevenmorecorrelated if the samplingrateis increased. Thesimplestway to exploitsample-to-sample redundancies in speechis to encode only thedifferences betweenadjacentsamples. Thedifferencemeasurements arethen accumulated in a decoderto recoverthe signal.In essence thesesystemsencodethe slopeor derivativeofa signalat thesourceandrecoverthesignalby integratingat the destination.Digitizationalgorithmsof this type arediscussed at lengthin later sections. 3.3.3 Cycle-to-Cycle Correlatlons Althougha speechsignalrequiresthe enrire300-3400-Hzbandwidthprovidedby a telephonechannel,at anyparticularinstantin time certainsoundsmay be composed of only a few frequencies within the band.when only a few underlyingfrequencies exist in a sound,the waveformexhibitsstrongcorrelationsover numeroussamples corresponding to severalcyclesofan oscillation.Thecyclic natureofa voicedsound is evidentin thetime waveformshownin Figure3.23.Encodersexploitingthecycleto-cycleredundancies in speecharemarkedlymorecomplicated thanthoseconcerned only with removingtheredundancy in adjacentsamples.In fact,theseencoders more

REDUNDANCIES123 3.3 SPEECH

Flgure 3.?3 Time wavefbrmof voicedsound. or lessrepresenta transition from the relatively high rate, natural-soundingwaveform encodersto the relatively low rate, synthetic-soundingvocoders.

Gorrelations 3.3.4 Pitch-lnterval-to-Pitch.lnterval Human speechsounds are often categorized as being generatedin one of two basic "voiced" soundsthat arise as a result ways. The first category of soundsencompasses of vibrations in the vocal cords. Each vibration allows a puff of air to flow from the lungs into the vocal hact. The interval between puffs of air exciting the vocal tract is referred to as the pitch interval or, more simply, the rate of excitation is the pitch. Generally speaking,voiced soundsarise in the generationof vowels and the latter portions of some consonants.An example of a time waveform for a voiced sound is shown in Figure 3.23. "unvoiced" sounds.FricaThe secondcategoryof soundsincludes the fricatives, or tives occur as a result of continuous air flowing fiom the lungs and passingthrough a vocal tract constricted at some point to generateair turbulence (friction). Unvoiced soundscorrespondto certain consonant$such asf j, s, and x' An example of a time waveform of an unvoiced sound is shown in Figure 3.?4. Notice that an unvoiced sound has a much more random waveform than a voiced sound. As indicated in Figure 3.23, not only doesa voiced soundexhibit the cycle-to-cycle redundanciesmentioned in Section 3.3.3, but also the waveform displays a longer qtre of lgtn rEpetitive pattern poffesponding to the duration of a pitch interval. $11p the glo$t Afficient ways of encoding the voiced portions 9f speechis to encode one that encoding,asa templatefor each successivepitch p,te4+Jp*41--1vpv9{pyr+ an"d.use i[GwU"in th.eqamesound,-fi!9,!r in1-elvalstyp,i[a]ly-last gom 5 to 20 mpes f,or men and flrqp 2.5 to 10 msegfor.yomen*"Sincea.typica] voiced.sogndf4ptsfo,rapprq4i5nately 100msec, there may be as many as 20-40 pitch intervals in a single sound. Although iitdh inierval encoding Can provide significant reductions in bit rates, the pitch is sometimesvery difficult to detect. (Not all voiced soundsproduce a readily identifi-

Figure 3.24 Time waveform of unvoiced sound.

124

VOICEDIGITIZATION

able pitch interval as in Figure 3.23.) rf the pitch gets encodeclerroneously, straxge soundsresult. An interestingaspectof pitch interval encodingis that it provides a meansof speeding up speechwhile maintaining intelligibility. By deleting some percentageof pitch intervals from each sound (phoneme),the rate of sound generationis effectively increasedin a manner analogousto more rapid word formation. The pitch of the sounds remains unchanged.In conffast, if the rate of reconstruction is merely increased,all frequenciesincluding the pitch increaseproportionately. Moderate speedupsproduce obvious distortion while greaterspeedupsbecomeunintelligible. Devices designedto simulate faster word formation have demonstratedthat we are capableof assimilating spokeninformation much faster than we can generateit.

3.3.5 InactivityFactor$ Analyses of telephone conversationshave indicated that a pafiy is typically active about4OVaof a call duration. Most inactivity occurs as a result of one personlistening while the other is talking. Thus a conventional (circuit-switched) full-duplex connection is significantly underutilized. Time assignmentspeechinterpolation (TASD describedin Chapter I is a techniqueto improve channelutilization on expensiveanalog links. Digital speechinterpolation (DSI) is a term usedto refer to a digital circuit counterpart of rASI. DSI involves sensingspeechactivity, seizing a channel,digitally encoding and transmitting the utterances,and releasingthe channel at the completion of each speechsegment. Digital speechinterpolation is obviously applicable ro digital speechsrorage$ystems where the duration of a pausecan be encodedmore efficiently than the pauseitself. In recorded me$sage$,however, the pauses are normally short since a "half-duplex" conversationis not taking place. DSI techniqueshave been used to expand the voice channel capacity of digital TDM links. The inputs are standardPCM signalsthat are digitally processedto detectspeechactivity. The DSI operationis often combined with speechcompressionalgorithms to implement digital circuit multiplication (DCM) equipment.when a 2 : I voice compressionalgorithm is combined with a2.5 : I DSI concentrationfactor, an overall 5 : I circuit expansionis achieved.Depending on the quality of speechdesired,even greater concentrationfactors are possible. The use of such equipment in a network must be carefully managedto ensure that voiceband data and digital data channelsbypassthe DCM operations.

3.3.6 NonunlformLong-TermSpectralDenslties The time-domain redundanciesdescribed in the preceding sections exhibit characteristics in the frequency domain that can be judiciously processedto reduce the encoded bit rate. Frequency-domain redundancies are not independent of the redundanciesin the time domain. Frequency-domaintechniquesmerely offer an alternate approachto analyzing and processingthe redundancies.

REDUNDANCIES 125 3.3 SPEECH

a I

-ro E t

T o EL

.H -20 o -6

E

-30 Ftequeilcy (Hrl

Flgure 3.25 Long-termpowerspectraldensityof speech. A totally random or unpredictablesignal in the time domain producesa frequency speckum that is flat acrossthe bandwidth of interest.Thus a signal that producesuncorrelatedtime-domain samplesmakes maximum use of its bandwidth' On the other hand, a nonuniform spectraldensity repre$ent$inefficient use of the bandwidth and is indicative of redundancyin the waveform' Figure 3.25 showsthe long-term spectraldensity of speechsignalsaveragedacross two populations: men and women [22]. Notice that the upper portions of the 3-kHz bandwidth passedby the telephonenetwork have significantly reducedpower levels. The lower power levels at higher frequencies are a direct consequenceof the timedomain sample-to-samplecorrelationsdiscussedpreviously. Large-amplitude signals cannot change rapidly because,on average,they are predominantly made up of lower frequency components. A frequency-domain approach to more efficient coding involves flattening the spectrumbefore encoding the signal. The flattening processcan be accomplishedby passingthe signal through a high-passfilter to emphasizethe higher frequenciesbefore sampling.The original waveform is recovered by passing the decoded signal through a filter with a complementary, low-pass characteristic.An important aspect of this processis that a high-passfilter exhibits time-domain characteristicsof a differentiator and a low-pass filter has time-domain characteristicsanalogousto an integrator. Thus the spectrum-flatteningprocessessentiallymeansthe slope of the signal is encodedat the $ource,and the signal is recoveredby integrating at the destinationthe basic procedure describedpreviously for sample-to-sampleredundancyremoval in the time domain. In studying Figure 3.25 it is natural to think that the remarkably low levels of signal energy at the higher frequencies(2-3.4 kHz) meansthat more bandwidth is being allocated to a voice signal than is really necessary.The error in such a conclusion,however. lies in the distinction between energy content and information content of the voice frequency speckum. As any beginning computer prograflrmer soon leams,- the *I

am assuming that beginning prograflmers seven-character name limitations.

still encounter older languages or file systems with

I

'

H

(

\

a

lrHtl rPUq lrllsds

126

PULSECODEMODULATION 127 3.4 DIFFERENTIAL

meaningof a programvariablecan often be retainedeventhoughit is shortenedby deletingall of the vowels.In speechthe vowelsrequiremostof the energyandprion theother marily occupythe lowerportionof the frequencyband.Theconsonants, hand,containmostof theinformationbut usemuchlesspowerandgenerallyhigher of the originalspeechenHencemerelyreproducinga high percentage frequencies. or storagesystem. goalfor a digital speechtransmission ergyis an inadequate 3.3.7 Short-Term Spectral Densitiee of thespeclong-termaverages shownin Figure3.25represent Thespeechspectrums and vary considerably periods the spectral densities of time Over shorter hal densities. (resonances) peaks some frequencies at with energy structureri exhibitsound-specific occurarecalled at whichtheresonances andenergyvalleysat others.Thefrequencies contain speech sounds typically formants. Voiced formantfrequencies,or simply are density the short-term spectral threeto four identifiableformant$.Thesefeaturesof is a displayofspeechspecofFigure3.26.A spectogram illustratedin thespectogram tral energyas a function of time and frequency.The horizontalaxis representstime, energylevels.Thus frequency,andthe shadingsrepresent the verticalaxisrepresents thedarkerportionsin Figure3.26indicaterelativelyhigh energylevels(formants)at particularinstantsin time. voicecodersprovideimprovedcodingefficienciesby encoding Frequency-domain of the spectrumon a dynamicbasis.As the sounds the mostimportantcomponents change,differentportions(formants)ofthe frequencybandareencoded.The period betweenformantupdatesis typically 10-20 msec.Insteadof usingperiodicspectrum somehigherquality vocoderscontinuouslytrackgradualchangesin measufements, vocodersoftenprovidelower thespectraldensityat a higherrate.Frequency-domain produce lessnaturalsounding bit ratesthan the time-domaincodersbut typically speech.

3.4 DIFFERENTIALPULSECODE MODULATION specificallyto takeadvantage Differentialpulsecodemodulation(DPCM)is designed in a typicalspeechwaveform.Sincetherange redundancies of the sample-to-sample of sampledffirences is lessthan the rangeof individual samples,fewer bits are Thesamplingrateis oftenthesameasfor a comneededto encodedifferencesamples. parablePCM system.Thusthe bandlimitingfilter in the encoderandthe smoothing FCM system$. filter in thedecoderarebasicallyidenticalto thoseusedin conventional A conceptualmeansof generatingthedifferencesamplesfor a DPCM coderis to circuitanduseananalog storethepreviousinputsampledirectlyin a sample-and-hold to measurethechange.The changein the signalis thenquantizedandensubtracter The DPCM structureshownin Figure3.27is morecomplicodedfor transmission. by a feedbackloop because thepreviousinputvalueis reconstructed cated,however, signalis an the feedback In essence, sample differences. that integratesthe encoded

128

vorcEDtctrtzAnoN gandlimiting filter

Prsvioiti input

CBtimate Aftumulrtor Flgure 3.27

F\rnctional block diagram of differential pCM.

estimateof theinputsignalasobtainedby integratingtheencodedsampledifferences. Thusthefeedbacksignalis obtainedin thesamemannerusedto reconstruct thewaveform in thedecoder. Theadvantageof thefeedbackimplementationi$ thatquantizationerrorsdo not accumulateindefinitely.If thefeedbacksignaldrifts from theinputsignal,asa resultof an accumulationof quantizationerrors,the next encodingof the differencesignal automatically compensate$ for thedrift. In a systemwithoutfeedbacktheoutputproducedby a decoderat theotherendof theconnectionmightaccumulate quantization errorswithoutbound. As in PCM sy$tems,the analog-to-digital conversionprocesscanbe uniform or companded. someDPCM systemsalsouseadaptivetechniques(syllabiccompanding) to adjustthequantizationstepsizein accordance with theaveragepowerlevelof the signal.(Seereference[9] for an overviewof varioustechniques.) Example3.4. speechdigitizationtechniquesaresometimesmeasuredfor quality by useof an 800-Hzsinewavea$a representative test signal.Assuminga uniform PCM systemis availableto encodethe sine wave acrossa given dynamicrange, determinehow manybitsper samplecanbe savedby usinga uniformDpcM system. solution. A basicsolutioncanbe obtainedby determininghow muchsmallerthe dynamicrangeof the differencesignal is in comparisonto the dynamicrangeof the signalamplitude.Assumethemaximumamplitudeof the sinewaveis A, sothat "r(f)=A sin(2n'800t) Themaximumamplitudeof thedifferencesignalcanbeobtainedby differentiating andmultiplyingby thetime intervalbetweensamples:

#

= e?n)(A0o) '8oor) cos(2n

PULSECODEMODULATIOru 129 3.4 DIFFERENTIAL

A(2nX800) lAx(t)l*u*= [,#l \

= 0'628A /

The savingsin bits per samplecanbe determinedas

t r \ loc,| 0628l=0.62tits \

/

that a DPCM sy$temcanuseJ bit per samplelessthan Example3.4 demonstratetl a PCM systemwith the samequality.Typically DPCM systemsprovidea full l-bit reductionin codewordsize.The larger savingsis achievedbecause,on average' speechwaveformshavea lower slopethanan 800-Hztone(seeFigure3.25).

3.4.1 DPCM lmplementatlons in a varietyof waysdecanbeimplemented anddecoders DifferentialPCM enCoders pendingon how the signalprocessingfunctionsarepartitionedbetweenanalogand digitalcircuitry.At oneextremethedifferencingandintegrationf'unctionscanbe implementedwith analogcircuitry,while at the otherextremeall signalprocessingcan be implementeddigitally using conventionalPCM samplesas input. Figure 3.28 with differing amountsof showsblock diagramsof tfueedifferentimplementations digital signalprocessing. Figure3.28adepictsa sy$temusinganalogdifferencingandintegration.Analogfot the is performedon thedifferencesignal,andD/A conversion to-digitalconversion feedbackloopis immediatelyperformedon thelimited-rangedifferencecode.Analog (S/H)circuitis usedto provideintegration. $ummationandstoragein a sample-anrl-hold Figure3.28bshowsa systemthat perfbrmsthe integrationfunctiondigitally. Insteadof immediatelyconvertingthedifferencecodebackto analogfor feedback,the differencecodeis summedand storedin a dataregisterto generatea digital representationof thepreviousinputsample.A full-scaleD/A converteris thenusedto producethe analogfeedbacksignalfor differencing.Noticethat the D/A convertersin in FigtheD/A converters Figure3.28bmustconvertthefull amplituderangewhereas signal. limited difference the more ure 3.28aconvert is performedby digital Figure3.28cshowsa sy$temwhereall signalprocessing samplecodes,which full-amplitude-range produces logic circuits.TheA/D converter previous amplitudecode. of the generated approximations are comparedto digitally rangeof the dynamic must encode the entire in this case Noticethat theA./Dconvefter on only the differversions operate in the other two input whereasthe A/D convefters encesignals. components, someof whichconDueto theavailabilityof digitalsignalprocessing (as is generallythe in Figure 3.28c) processing digital tain intemalA./Dconverters, most DPCM apIn fact, mosteffectivemean$of implementinga DPCM algorithm. into been digitized plicationsinvolve processingspeechsignalsthat have already requires no usually standard64-kbpsPCM formats.ThustheDPCM implementation

130

votcEDtGtlzATtoN

analogprocessing. As an aid in processing log-PCMsignals,someDSp components provideinternalp-law andA-law conversionfunctions. Thedecoders in all threeimplementations shownin Figure3.28areexactlylike the feedbackimplementations in theconesponding encoder.This reinforcesthefact that thefeedbackloopgenerates anapproximation of theinputsignal(delayedby onesample).Ifno channelelrorsoccur,thedecoderoutput(beforefiltering)is identicalto the feedbacksignal.Thusthecloserthefeedbacksignalmatchestheinput,thecloserthe decoderoutputmatchesthe encoderinput.

Decoder

Encoder

Oecoder

Encoder

(t)

Figure 3.2t DPCM implementations:(a) analog integrarion; (b) digital integration; (c) digital differencing.

pulsEcoDEMoDULATIoN 131 s.4 DTFFERENTTAL

3.4.2 HlgherOrder Prediction it asa specialcaseof a linear A moregeneralviewpointof a DPCM encoderconsiders of thepredictionerror.Thefeedbacksignal predictorwith encodingandtransmission first-orderpredictionof thenext samplevalue,andthe of a DPCM sy$temrepresents sampledifferenceis a predictionelror.Underthis viewpointtheDPCM conceptcan be extendedto incolporatemorethanonepastsamplevalueinto the predictioncircuitry. Thus the additionalredundancyavailablefrom all previoussamplescan be weightedandsummedto producea betterestimateof the next input sample.With a to allow encodingwith betterestimate,the rangeof the predictionerror decreases fewerbits. For systemswith constantpredictorcoefficients,resultshaveshownthat mostof therealizableimprovementoccurswhenusingonly thelastthreesamplevalof linearpredictionusingthe lastthreesamplevalues ues.Thebasicimplementation showsanalog is shownin Figure3.29.For conceptualpurposesthis implementation differencingandintegrationasin Figure3.28a.The mosteffectiveimplementations usedigital memory,multiplication,andadditionin a DSP componentin lieu of the involvealreadydigmostapplications shown,particularlybecause analogprocessing itized (PCM) signals. As mentionedin Section3.4,analysisof differentialPCM systemswith first-order reductionin codelengthrelativeto predicationtypicallyprovidesa 1-bit-per-sample utilizingthirdExtendedDPCM sy$tems PCM systemswith equivalentperformance. orderpreclictioncanprovidereductionsof f -2 bitsper sample[23]. Thusa standard DPCM systemcanprovide64-kbpsPCM quality at 56 kbps,andthird-orderlinear predictioncan provide comparablequality at 48 kbps. However,somesubjective higherbit ratesareneededto match64-kbps haveindicatedthatsomewhat evaluations PCM quality. 3.4.3 Adaptive Differential PGM of DFCM canprovidesavingsof I -2 bits implementations Relativelystraightforward per samplewith respectto standardPCM encoding.Even greatersavingscan be

Decoder

Encoder

Figure 3.29 Extensionof DPCM to third-orderprediction.

132

votcEDtctlzATtoN

achievedby adding adaptationlogic to the basic DPCM algorithm to create what is refened to as adaptive differential PcM (ADpcM). Many forms of ADpcM have been investigatedand used in various applications.Two of the most prevalent applications are voice messagingand DCM equipment for increasingthe number of voice channelson a Tl line. with respectto the latter application, ITU-T has establisheda 32-kbps ADPCM standard(RecommendationG.721) [24]. This algorithm has been extensively testedand characterizedto not significantly degradetoll quality voice circuits when insertedinto the internal portions of the network. Design considerationsof the standard are: 1. Multiple tandem encodings and decodings berween both pcM and analog interfaces 2. End-to-end signal quality for voice, voicebanddata, and facsimile 3. Effects ofrandom and bursty channel errors 4. Performanceon analog signals degradedby loss, noise, amplitude distortion, phasedistortion, and harmonic di$tortion 5. Easy transcodingwith p-law and Alaw pCM The 32-kbps rate implies a 2 : I savingsin channelbandwidth with respectto standard PCM. A significant impairment inffoduced by implemenrarionsof the ADpcM standard is the comrption of modem signalscarrying data ratesgreaterthan 49fi) bps [24]. Voiceband data at rates of 4800 bps and below are adequatelysupported. The G.721 ADPCM algorithm is conceptually similar to rhat shown in Figure 3.29 but more sophisticatedin that it usesan eighth-orderpredictor, adaptivequantization, and adaptive prediction. Furthermore, the algorithm is designedto recognize the difference between voice or data signals and use a fast quantizer adaptationmode for voice and a slow adaptationmode for data.

4

*q

4

-*o*****__=-

v,

3 3

$r

2

# ---*

2 PCM ADPCM

1

2

3 4 5 I 7 I Numbsrof Encodlner (el

t 2

3 4 E g 7 I Numbsrof Encodl4r

tbl

Figure 3.30 AverageMOS versusnumberof encodingsfor PCM and ADPCM: (a) linear microphone;(b) carbonmicrophone.

g.s DELrA 133 MoDULATtoN subjectiveevaluation[25] of the G.721algorithmusingthe meanopinionscore (MOS) methodof evaluatingspeechquality is shownin Figure 3.30. The MOS methodusestrainedlistenersto evaluatethe speechqualityon a scaleof I : 5. Figure 3.30showstheaveragescoresof 32-kbpsADPCM and64-kbpsPCM asa functionof with multiple The speechqualityassociated thenumberof analogtandemencodings. first transcoding beyond the (to not degrade from PCM) does digitaltranscodings and errors or slips such as channel impairments transmission aslong asthereareno digital be an important (or used to any other coder) of ADPCM (Chapter7). Theperformance of mixturesof analoganddigiwhenthepublicnetworkwascomposed consideration network areall digital, performance porxions of the tal equipment.Becausetheinternal (Multiple PCM-tolonger a consideration. with multiple analogconversionsis no are encountered DCM systems multiple ADPCM conversionscan still occurwhen within a network.) BecauseADFCM at 32 kbpsprovidesgoodqualityat a moderatecostandpower it is usedin severalcordlesstelephoneor low-tier-digitalcellularsysconsumption, tems: System(PACS)(NorthAmerica) PersonalAccessCommunications (CT2) (Europe) CordlessTelephones SecondGeneration (DECT)(Europe) Digital EuropeanCordlessTelephones System(PHS)(Japan) PersonalHandyphone

3.5 DELTA MODULATION Deltamodulation(DM) is anotherdigitizationtechniquethat specificallyexploitsthe in a speechwaveform.In fact,DM canbe considered redundancy sample-to-sample asa specialcaseof DPCM usingonly I bit per sampleof the differencesignal.The singlebit specifiesmerelythe polarity of the differencesampleandtherebyindicates sincethelastsample.An approximation or decreased whither thesignalhasincreased to the input waveformis constructedin the feedbackpathby steppingup onequantizationlevel whenthedifferenceis positive("one")andsteppingdownwhenthedifof'trps" ferenceis negative("zero").In this way the input is encodeda$a sequence Figure3.31showsaDM approximaa staircase. and"downs"in amannerresembling tion of a typical wavefonn.Noticethat the feedbacksignalcontinuesto stepin one directionuntil it crossesthe input,at whichtime thefeedbackstepreversesdirection until the input is crossedagain.Thus,whentrackingtheinput signal,theDM output "bounces"backandforth acrossthe input waveform,allowingthe input to be accuby a smoothingfilter' ratelyreconstructed samplecontainsa relativelysmallamountof information(1 each encoded Since a highersamplingratethanPCMor multibitDPCM systems. require bit), DM $ystems muchhigherthantheminimum(Nyquist)samis necessarily rate In fact,thesampling *A

low-tier cellular system utilizes simple, Iow-power mobile units, and small cells and only suppofts pedesfrian speeds,

134

VoIcE DIGITIZATIoN

Figure 3.31 Waveformencodingby deltamodularion, pling rate of twice the bandwidth. From anotherviewpoint, "oversampling" is needecl to achieve better prediction from one sample to the next. The main attraction of DM is its simplicity. Figure 3.32 shows a basic implementation of a DM encoderand decoder.Notice that the A/D conversion function is provided by a simple comparator. A positive-difference voltage produces a l, and a negative-differencevoltage produces a 0. Correspondingly, the D/A function in the feedback path, and in the decoder,is provided by a two-polarity pulse generator.In the simplest form the integrator can consist of nothing more than a capacitorto accumulate the charge from the pulse generator. In addition to theseobvious implementation simplicities, a delta modulator also allows the use of relatively simple filters for bandlimiting the input and smoothing the output [26]. As discussedin section 3.1, rhe speckum producedby a samplingprocess consist$of replicasof the sampledspectrumcenteredat multiples of the sampling frequency. The relatively high sampling rate of a delta modulator producesa wider separation of these spectrums, and, hence, foldover distortion is prevented with less stringent roll-off requirementsfor the input filter.

3.5.1 SlopeOverload The conceptualoperationof a delta modulator shown in Figure 3.3 I indicatesthat the encoded waveform is never much more than a step size away from the input signal. sometimes a delta modulator, or any differential system such as DpcM, may not be able to keep up with rapid changes in the input signal and thus fall more than a step

Pul* gf,ndtbr

Eruodr Figure

3.32

Delta modulation

encoder and decoder.

135

3.5 DELTAMODULATION

"slope size behind. When this happens,the delta modulator is said to be experiencing overload." A slope overload condition is shown in Figure 3'33. Basically, slope overload occurs when the rate of changeof the input exceedsthe maximum rate of changethat can be generatedby the feedbackloop' Since the maximum rate of change in the feedback loop is merely the step size times the sampling rate, a slope overload condition occurs if

l#l'*

(3.13)

wherex(t) = input signal q = stePsize f; = samplingfrequencY involvesa trade-offbetweentwo types Thedesignof a DM (or DPCM)necessarily of distortion:the moreor lessrandomquantizationnoise,sometimesreferredto as granularnoise,and the slopeoverloadnoise.As indicatedin Figure3-33,granular for slowlychangingsignals,whereasslopeoverconsideration noiseis a predominant loadis dominantduringrapidlychangingsignals.Obviously,granularnoiseis small if stepsizesaresmall,but smallstepsizesincreasethe likelihoodof slopeoverload' TheoptimumDM stepsizein termsof minimizingthetotalof granularandslopeoverUyAbate[271. loadnoisehasbeenconsidered The perceptualeffectsof slopeoverloadon the quality of a speechsignalare significanitydifierentfrom the perceptualeffectsproducedby granularnoise'As inditheencoded its peaksjustbefore catedin iigure 3.33,theslopeoverloadnoisereaches identical components signalreachesits peaks.Hence,slopeoverloadnoisehassfiong Disthe input' of irifrequencyandapproximatelyin phasewith a majorcomponent "masked'by is effectively tortionthatis correlatedin thismannerto thespeechsignal "uncorrelated"distortion.In the speechenergyandthereforeis lessnoticeablethan to a listenerthanrandomor granular fact, overloadnoiseis muchlessobjectionable noiseat an equivalentpowerlevel [28]. Hence,from the point of view of perceived speechquality,the optimummix of granularandslopeoverloadnoiseis difficult to determine.

Figure 3.33 Slope overload and granular noise of delta modulation system.

136

vorcEDtcllzAloN

slope overload is not a limitation of just a DM system, but an inherent problem with any system, such as DPCM in general, that encodesthe difference in a signal fiom one sampleto the next. A difference systemencodesthe slope of the input with a finite number ofbits and hencea finite range. Ifthe slope exceedsthat range, slope overload occurs. In contrast, a conventional pcM systemis not limited by the rate of changeof the input, only by the maximum encodableamplitude. Notice that a differential systemcan encodesignalswith arbitrarjly large amplitudes,as long as the large amplitudes are attainedgradually. Many versionsof DM for voice encoding were investigatedin the 1960sand 1970s [29, 30]. All of theseefforts focused on ways of implementing adaptive delta modulation (ADM) to improve the performance at a given bit rate. The intense interest at that time was relatedto the simplicity, good toleranceof channelerrors,and relatively low cost implementation.The cost factor is no longer relevantbecauseeven relatively complicatedcoding algorithms now have insignificant costscomparedto mosr system costs.ADM is still used in some old pBXs, in some military securevoice radio systems, and as a mealls of encoding the residual error signal of some predictive coders describedin the following sections.

3.6 ADAPTIVE PREDICTIVE CODING Thedifferential sysrems described in theprevious (DpcM,ADpcM,ADM) sections

operatewith lowerdataratesthanpcM systemsbecause theyencodea differencesignal thathasloweraverage powerthantheraw inputsignal.Theratioof theinputsignal powerto thepowerof thedifferencesignalis referredto asthepredictiongain.Simple DPCM system$(firsr-orderpredictors)provide about 5 dB of prediction ga1n. ADPCM providesgreaterlevelsof predictiongaindependingon thesophistication of the adaptationlogic andthe numberof pastsamplesusedto predictthe nextsample. Thepredictiongainof ADpcM is ultimatelylimited by thefact thatonly a few fast samplesareusedto predictthe input andthe adaptationlogic only adaptsthe quantizer,not thepredictionweightingcoefficients(thec's in Figure3.zg). Adaptivepredictivecoding(Apc) t3l, 3zl providesgreaterlevelsof prediction gain by adaptingthe predictioncoefficientsto individual speechsegmentsand, in mostcases,usinghigherordersof prediction(e.g.,up to 12).If thecoefficientsarede_ terminedfrom pasthistoryandusedto predictsubsequent speechsegments (backward estimation),l3 dB of predictiongainis possible[10]. If speectrsegments aredelayed so predictorcoefficientscanbe usedon the samespeechsegments from which they werederived(forwardesrimation), 20 dB of predictiongainis possible[33]. A blockdiagramof a basicApc encoder/decoder is shownin Figure3.34.Theinput to the encoderandthe outputfrom the decoderareassumed to be uniformpcM, most likely representing conver$ionsfrom and to log pcM. The transmitteddata $treamis necessarily composedof blockscontainingthreetypesof information:(l) the encodeddifferencesignal(residual),(z) a gainfactor,and(3) rhepredicrorcoefficients.Themostsignificantdifferencebetweenthis coderanda'DpcM or ADpcM coderinvolvesthe periodicdeterminationand transmissionof the predictorcoeffi-

= (J G

E o .E f

tr o

:E

- E s t i&l 'GE o

t g

c F o

E L

t-'

at

EI o

F E o iJ 4) ts

9J q) F

E

4 +

ca?

.E {) ,9

H

o! FFT

E o .g

E

G

T

E (J r

E

=oF 3

137

138

votcEDtctlzATtoN

cients' Notice that the integratedresidual signal at point A is identical to the input signal delayed by one sample (except for residual quantization error). Thus the corresponding point in the decoderis the reconstructedpCM output. Explicit transmissionof a gain factor, as opposedto deriving a gain factor from the transmitted residual, is useful in this application becausea block-structuredtransmission format is required for transmitting the predictor coefficients anyway. Residual encoding may use any of the waveform algorithms describedpreviously. Arbiharily accuratereconstructionof the input can be achievedif enoughbits are usedin encoding the residual. of course,the purpose of the adaptiveprediction is to achievea low data rate. single-bit PcM thar conveys only the polarity ofthe residual provides adequatepedormancefbr low-data-rateapplications [33]. Notice that becauseof the feedback path, single-bit encoding ofthe residual essentiallyproducesa delta modulator with very sophisticatedfeedbacklogic that is periodically changedto match the shape and energy level of corresponding(forward-estimated)speechsegments. A large variety of algorithms have been investigatedfor calculating the predictor coefficients, most of which involve extensive computation. If the apflication is for voice responsesystems,the computations do not have to occur in rial time and are therefore sometimes done on a large mainframe computer. Real-time encoding restricts the algorithm to one that can be realized with a DSp or special-purposeIC de_ signed to implement a specific coefficient determining algorithm. A linear predictive coding (LPC) algorithm as describedin Section 3.8.3 is a common algorithm because it provides good performance and is available in custom parts. The u.s. defense deparHnentadoptedan APC algorithm using a fourth-order LPC coefficient analysis as a government standardfor secure voice transmission at 9.6 kbps [34]. This system transmitsblocks of 240 bits containing I 80 one-bit samplesof the residual, 54 bits for parametersand gain factors,5 bits for error correctionsof critical most significant bits, and I framing bit.

3.7 SUBBANDCODING A subbandcoder is one form of coder using a frequency-domainanalysisof the input signal insteadof a time-domain analysisas in previously described.od"r*. As shown in Figure 3.35, the coder first divides the input spectrum into separatebandsusing a bank ofbandpassfilters. The signal passingthrough eachofthe rilatively nanow subbands is individually encoded wirh separateadaprive ApcM, pcu, or enpcM encoders.After each subbandis encoded,the individual bit streamsare multiplexed for transmissionto the decoder,where they are demultiplexed, decoded,and combined to recon$tructthe input. separatelyencodingeachsubbandis advantageousfor severalreasons.First, by using separateadaptationfor each band, the quantization step sizescan be adjustedaccording to the energy level in each band. Those bands with relatively high energy levels can be encodedwith relatively coarsequantization.In this mannei the spectrum of the quantization noise is matched to the short-term $pectrum of the signal. This propefly is very desirablein a perceptualsensebecauseit allows the speechsignal to

CODING 3.7 SUBBAND BandpErB filters

139

AdaPtive PCM

Figure 3.35 Subbandcoder.

mask the quantization noise. (The human ear perceives speechby measuring the shortterm energy level of individual frequency bands.Hence,relatively low noise in a band with no speechenergy is perceptually more significant than greater noise in a band with significant speechenergY.) A secondadvantageof subbandcoding is that the bit rate (quality) assignedto each individual band can be optimized according to the Perceptualimportance of each for low be used per can sample band. In particular, a relatively large number of bits of frequencies where it is important to preserve the pitch and formant Structure voiced ,ound*. At higher frequencies,however, fewer bits per sample can be used because noiselike fricatives do not require comparablequality in reproduction' As reported in reference [10], subbandcoders provide significant bit rate reductions compared to the more coilrmon and simpler coding algorithms: adaptive delta modulation and adaptive differential PCM. Specifically, subbandcoders at 16 kbps are reported to be perceptually equivalent to ADFCM coders at ?2 kbps. A subband coder at 9.6 kbps is reportedto be eguivalentto an ADM coder at 19.5kpbs. Extensive in reference description and performance analysis of subband coding are available

t351.

A particularlysignificantexampleof subbandcodingis the ITU-T recommendaprovidesfor encoding nonl.7}Zfbr widebandspeechcodingt36, 371.This standard 7-kHz speechbandwidthwith 64 kbps.Thusit providesa meansof significantlyimproving voice fidelity whenend-to-enddigital channelsare available.Applications andspeakerphones' thatcanbenefitmostfrom thehigherfidelity areteleconferencing it doesnot have the network, of equipment internal Becauseit is not intendedfor Tranconversions. analog tandem support to proces$voicebanddata signalsor bridge conference of in support scodingsto and from uniform PCM are required affangements. As shownin Figure3.36[36], theG.722algorithmdividestheinput speechbandsubbands'Both subequal-sized width from 50 to 7000Hz into two approximately at48 kbpsandtheuppersubband bandsareencodedwith ADPCM:theIowersubband

140

votcEDlctTlzATtoN

l4 blt| l6kHr I

I

04 kbps oulput

Lowr ilbbEnd ADPCM dftoder (3 vrrirntl

Mode indic8tion

Figure3.36 G.7227-kIIzaudiocodec, (Fromreference t361,p. 10.) at 16kbps.Theupperbanddoesnot requireashigh a dataratebecause it is not asimportantperceptuallyand has significantlyless energy.The algorithmpermits the lowerbandto be encodedat 40 or 32 kbps,whichallowsg or 16kbpsfor anauxiliary datachannelin teleconferencing applications or for theuseof56-kbpschannelswhen 64 kbpsis not available.Figure3.37showstheMos performance olthe G.222atgorithm for speechandmusicat threebasicrates[3g].

S i g n sB l 8ndwidth=7kHz

Excellcnt 5

Good 4 Sprech t 6 !

o

,s E {

Muric Fair 3 Hsiearch Gorl for 7-kHu Sp€ech Poor 2

o 24& l$ps PCM r G722:64-.kbps$B-ADPCM Bed I 4$ Bit RErB(kbps)

Figure3.37 AverageMosversusbitrateotG.lzzk-kHzaudiocodec.(Fromreference[3g], p. 15.)

3.8 vocoDEHS 141

3.8 VOCODERS For the most part, the encocling/decodingalgorithms describedpreviously have been possible' concernedprimarily with reproducing the input waveform as accuratelyas and process they the signal nature of of the Thus they u**u-* little or no knowledge occur Exceptions channel. in a voice are basically applicable to any signal occurring when subbandcoding and adaptivepredictive coding are designedfor particularly low to bit rates (20 kbps or iess). At thesebit ratesthe encodershave been closely tailored other sigquality for comparable provide the statisiics of a speechsignal and cannot nals. Differential systems,such as DPCM and DM, also exhibit a certain amount of deficiencies speech-specificproperties by virtue of their high-frequency encoding (slope overload). The digitization procedures described in this section very specifically encode speechsignatsand speechsignals only. For this reasonthese techniquesare referred "vocOders,"an acronym for vOiCecoderS.Since theseteChniquesare to collectively as public designedspecificatly for voice signals,they are not applicableto pottions ofthe be acmust (such signals) as modem telephonenetwork in which other analog signals commodated.The basic goal of a vocoder is to encodeonly the perceptuallyimportant aspectsof can be speechwith fewer bits than the more general waveform encoders.Thus they cannot. used in limited-bandwidth applicationswhere the other techniques"wrong number") Some of the main applications for vocoders are recorded (e.g., tecellular digital message$,encryptedvoice transmissionover niuT0wbandHF radio, over telephony Iephony, digital circuit multiplication, computer output, games,and provided multiple tnl fnternet. A particularly interesting,early use of an LPC vocoder voice channels over a single voice frequency leasedline' Using a well-conditioned time division leasedline to obtain a 9600--bpscircuit, four 2400-bpsvoice signalswere (early 1980) where multiplexed into a single line [39]. This is one of the first cases digitization was usedto actually decreasethe bandwidth of a voice signal' This system pr*ovidedintelligible voice, but the overall quality was below telephone standards' This particular systembecameobsoletewhen long-distanceleased-linecosts dropped to the point that the sacrifice in voice quality was unjustified' voThis sectiondescribesthree of the most basic vocoding techniques:the channel variand coder, the formant vocoder, and the aforementioned LPC. Many other forms some of the ations of vocoders have been proposed and studied. For a discussionof other techniques and an extensive bibliography on the subject, see reference [10]' the Most commercial applications for vocoders have concentratedon adaptationsof LPC algorithm, particularly for digital cellular and voice over data networks'

Phtaselnsensitlvltq

preserving A fundamental requirementfor maintaining good speechquality involves individbetween phase relationship the shot"t-termpower specffum of the signal. The *Vocoders

to detect the can be insefted into internal portions of a network if the interfaces are equipped them accordingly' process presencc of voiceband modem or fax signals and

142

votcEDtGtlzATtoN tin {611 T

cln{?.drl

Einl0rrl + lin(?or+fl

Figure3.38 Effectof phase shiftin thesuperposition of twotones. ual frequencycomponents is perceptually muchlessimportant.one of thebestexamplesof theear'sinsensitivityto phaseis demonstrated whentwo notesareplayedsimultaneously,or nearly simultaneously,on a piano. The compositeJound, as perceivedby a listener,is seeminglyno differentif onenoteis struct< stightlylaterthan theother.In essence, theearsenses how muchenergyis presentat variousfrequencies in the speechspectrumbut doesnot sen$ethe phaserelationshipbetweenindividual frequencycomponents. Theeffectof a phaseshiftin onecomponent of a compositesignalis shownin Figure 3.38.The first compositewaveformis producedwhentwo individualfrequenry component$ haveidenticalstartingphases.The secondcompositewaveformoccurs whenthetwo frequencytermshavestartingphasesshiftedby-90owith respectto each other.Noticethat the compositewaveformsaremarkedlydifferenteventhoughthe differenceis imperceptibleto the ear.For thesereasonsthe time waveformproduced by a vocodergenerallybearslittle resemblance to the originalinput waveform.In_ stead,theemphasisof a vocoderis in reproducingthe short-term fower specrumof theinput. 3.8.1 ChannelVocoder channel vocoderswere firsr developedin l92g by Homer Dudley [40]. Dudley's originalimplementation compressed speechwaveformsinto an *utog signalwith a total bandwidthof about300 Hz. Basedon the originalconcept,digital channelvo_ codershavebeendevelopedoperatingin therangeof l_? kbps. A majorpart of the encodingprocessof a channelvocoderinvolvesdetermining the short-termsignalspectrumas a functionof time. As indicatedin Figure3.3g,; bankofbandpassfilters is usedto separate the speechenergyinto subbands that are full wave rectifiedand filtered to determinerelativepower levels.The individual powerlevelsareencodedandtransmittedto thedestination. Noticethatthis muchof a channelvocoderis very similarto the subbandcoderdiscussed previously.A sub_ bandcoder,however,typically useswider bandpass filters, whicir necessitate sampling the subbandwaveformsmoreoften (determininga waveforminsteadof just a powerlevel)' Sincea subbandcoderencodeswaveforms,it alsoincludesphase informationthatis ignoredby a channelvocoder. In additionto measuringthesignarspectrum, modemchannelvocodersalsodetermine the natureof speechexcitation(voiceor unvoiced)andthe pirch frequencyof

s€ FE

L.:

o

E E U (t\ t t rti

H 00

E o 6

.9 :: E E g E E 5g

FF

o

143

144

votcEDtctlzAloN

voiced sounds.The excitation measurementsare used to synthesizethe speechsignal in the decoderby passingan appropriatelyselectedsourcesignal through a frequencydomain model of the vocal tract transfer function. Voiced excitation is simulatedby a pulse generatorusing a repetition rate equal to the measuredpitch period. unvoiced excitation is simulated by a noise generator.Owing to the synthesizednature of the excitation, this form of a vocoder is sometimesreferred to as a pitch-excited vocoder. As indicated in Figure 3.39, a decoder implements a vocal tract transfer function as a bank of bandpassfilters whose input power levels are determined by respective subbandpower levels in the encoder.Thus outputs of each bandpassfilter in the clecoder conespond to outputs ofrespective bandpassfilters in the encoder.superposing the individual bands re-creates,in a spectral sense,the original signal. Maly variations in the basic channel vocoder have been developed involving the nature of the excitation and the meansof encoding the power levels. Recentadvances in digital technology have introduced the useof digita-lsignal processingto determine the input spectrumby way of Fourier transform algorithms in lieu of the bank of analog filters. All forms of vocoders that measurethe power spectraldensity are sometimes referred to collectively as spectrum channel vocodersto distinguish them from time-domain vocoderssuch as the LpC describedlater. The most difficult aspect of most vocoder realizations involves determining the pitch of voiced sounds. Futthermore, certain sounclsare not clearly classifiable as purely voiced or purely unvoiced. Thus a desirableextensionofthe basic vocoder involves more accuratecharacterizationof the excitation. Without accurateexcitation information, vocoder output quality is quite poor and often dependenton both the speakerand the particular soundsbeing spoken.Some of the more advancedchannel vocodershave produced highly intelligible, although somewhat synthetic sounding, speechat 2400 bps [231.

3.8.2 FormantVocoder As indicated in the spectogram of Figure 3.26, the short-term spectral density of speech is rarely distributed across rhe entire voice band (200-3400 Hz). Insread, speechenergy tends to be concentratedat three or four peakscalled formants. A formant vocoder determinesthe location and amplitude of thesespectralpeaksand transmits this information insteadof the entire $pectrumenvelope.Thus a formant vocoder produceslower bit ratesby encoding only,the most significant short-termcomponenrs in the speechspectrum. The most important requirement for achieving useful speechfrom a formant vocoder involves accurately tracking changes in the formants. once this is accomplished, a formant vocoder car provide intelligible speechar less than 1000 bps tl0l.

3.8.3 LinsarPredlctiveCodlng A linearpredictive coderis a popularvocoderthatextracts perceptually significant featuresof speechdirectly from a time waveform rather than from frequency specrra, as does a channel vocoder and formant vocoder. Fundamentally, Lpc analyzes a

3.8 vocoDFRS 145

Synthesis

Analvsis

relationship' andsynthesis Figure3.40 Basicmodelof LPCanalysis

speechwaveformto producea time-varyingmodelof the vocaltractexcitationand by passthespeech in thereceivingterminalre-creates transferfunction.A synthesizer perihact' By of the vocal model ing the specifiedexcitationthrougha mathematical the excitation, of of themodelandthe specification odicallyupdatingtheparameters interval,howin either.Duringanyonespecification adaptsto changes thesynthesizer process. a lineartime-invariant to represent ever,thevocaltractis assumed is shownin Figrelationshipbetweentheencoder/decoder The analysis-synthesis the A(z) matrixto minithe coefficientsof ure 3.40.Theanalysisprocessdetermines knowsboth If the decoder x(n)' mizetheerrore(n)with a givensetof speechsamples In the mostbasic x(n). the input samples A(z) and e(n),it canre-create(synthesize) specifled is indirectly to thedecoder.Instead,e(rr) form of LPC e(n)is not transmitted of the synexcitation asthe excitationof a vocalhact model.Noticethat e(n)is the thesizerin Figure3.40. is shownin Figure3.41' A blockdiagramof thebasicmodelfor speechgeneration The equawhich is alsoa modelof the mostbasicfbrm of LPC decoder/synthesizer' tion of thevocaltractmodelshownin Figure3'44 is definedas:

Pirch period

,l

@

"Voiced"

{

Excitationmodel

Vocal tract model

Figure 3.41 Speechgenerationmodel of linear predictive coding.

146

votcEDtctlzAloN

y(n)=Eooy@*k)+Gx(n)

(3.14)

FI

where)(n)= zth outputsample er = lcthpredictorcoefficient fi = gain factor x(n) * input at sampletime n p = orderof the model Noticethatthespeechoutputin Equation3.14is represented asthepresentinputvalue plus a linearcombinationof the previousp outputsof the vocaltract.The modelis adaptivein thattheencoderperiodicallydetermines a newsetof predictorcoefficients corresponding to successive speechsegments. BasicLpC doesnot measureandencodedifferencewaveformsor errorsignals.Instead,the errorsignalsareminimized in a mean-squated sensewhenthe predictorcoefficientsaredetermined. The ability to avoid encodingthe predictorerrorscomesfrom the fact that LPC usesprediction parametersbasedon the actualinput segmentsto which the parametersare applied (forwardestimation).In confiast,mostof thepredictivecodersmentionedpreviously base their prediction on past measurements only (backwardestimation).Rapid changesin the vocaltractor excitationcausemodelsbasedon pastmea$urements to be lessaccurate. Thenegativeaspectof forwardestimationis thedelayit insertsinto theencodingprocess,which,depending on theapplication,mayor maynotbe a consideration. The information that a LPC encoder/analyzer determinesandtran$mitsto the decoder/synthesizer consistsof l. Natureofexcitation (voicedorunvoiced) 2. Pitch period (for voiced excitation) 3. Gain factor 4. Predictor coefficients (parametersof vocal tract model)

The natureof the excitationis determined,as in other vocoders,by determining whetherstrongperiodiccomponents exist in the waveform.pitch is determinedby measuringperiodicitywhenit exists.In additionto measuringpitch with techniques similarto thoseusedby othervocoders,anLpc encoder/analyzer hasparticularpropertiesthataid in pitch determination [41]. Thepredictorcoefficientscanbe determined usingoneof severaldifferentcomputationalprocedures. All procedures useactualwaveformsamplesasthe desiredoutputs of the synthesizer. using thesesamplevalues,a setof p linearequationsin p unknowncoefficientsis produced.Thusthecoefficientsaredeterminedby inverting a p-by-pmatrix.sincetheorderof p may varyfrom 6 to rz, dependingon the *peecir quality desired,straightforward matrix inversionrepresents a significantamountof computation.Dependingon specificassumptions madein the model,however,the

3.8 vocoDEHS 147 matriceshave specialpropeltiesthat greatlysimpliff the solutionof the equations

t4lI.

of speech, Althoughlinearpredictivecodersprocesstime-domainrepresentations peaks of the speech good of the provide estimates they their operationis suchthat in gradual changes tracking of effectively spectrum.Furthermore,an LPC is capable natural sounding provide more thespectrumenvelope.Theoverallresultis thatLPCs vocoders[23]. Most LPC research speechthanthe purely frequency-domain-based on encodingspeechin therangeof 1.2-2.4kbps' hasconcentrated a 2400-bps,I0th-orderLPC (LPCof Defenseestablished The u.s. Departrnent over dial-uptelephonelines [42' 431. communications for secure l0) as a standald msec. The 54 bits areallocatedas every 22.5 of 54 bits This systemtransmitsblocks (gain for 10coefficients(voiced factor),41 bits 7 bitsfor pitch,5 bitsfor energylevel excitation),and I bit for framing. of the coderwith respectto naturalspeechis providedin Table The performance 3.4.A diagnosticrhymetest(DRT) tul4lis a meansof determiningthe intelligibility of correctword recognitionsfrom a standardized of a speechsystemasa percentage list of pairsof rhymingwords(e.g.,goatvs. coator thick vs. tick). The DRT-withwhenbackgroundnoiseapprothe word recognitionpercentage noisetestmeasures before encoding.The diagnostic the test words priateto the applicationis addedto usacceptability and subjective measure(DAM) [45] ratesintelligibility acceptabiliry ofthe preferences personal on thateliminatemuchofthe dependence ing procedures listeners.The scoresarenormalizedfrom 0 to 100' 3.8.4 Enhanced-Excitation Llnear Predlctive Codlng speechin the The basicLPC afgorithmdescribedin theprevioussectionsynthesizes decoderusinga very simplemodelfor theexcitationthatrequiresonly aboutlOToof datarate.Becausevoicedsignalsaremodeledby a simplepulsegenerthe aggregate an LPC coder is sometimesreferredto as a pulse-excitedlinear prediction ator. (PELP)coder.The simplicityof the modelinevitablyproducessyntheticsounding havebeendevelopedto To overcomethis shottcoming,numeroustechniques speech. are multipulseLPC algorithms enhancethe excitation.Threeenhanced-excitation linearpredic(RELP), mixed-excitation (MPLPC),residualexcitedlinearprediction (CELP). Because prediction linear tion (MELP),andvariousversionsof code-excited LPC algorithmsinvolveencodinga residualpredictionerror the enhanced-excitation andLPC'10 SPeech of Natural Comparlson TABLE3,4 Performance NaturalSpeech(3.6kHz) DHT DRTwithnoise DAM

95o/" 92"/" 65

LPC-10 90o/o

82"/" 48

148

VoIcEDIGITIZATIoN

in some form or another, these algorithms are often referred to as hybrid codecs; a combination of vocoding and waveform coding.

Multlpulse LPC As shownin Figure3.42,whereit canbecompared withaconventional Lpc system, MPLPC[46]is a conceptually simpleexrension of conventional Lpc. whereLpc

usestheresidualto determinethepitchperiodof voicedsignals,MpLpC usestheresidualto determinea sequence of pulsesto morecloselymatchtheresidual.In thesimplestcaseMPLPCusesa fixed numberof pulsesin a pulsekain anddetermines only the startingphaseof the train andthe amplitudesandpolaritiesof eachpulsewithin the train. A significanrside benefitof MPLPC is that it doesnot havero identify whethera speechsegmentis voicedor not andconsequently determinethepitchperiod of voiced signals.Instead,the multipulsedeterminationin eachanalysissegment automaticallyadaptsto thenatureof theactualexcitation.MPLPCat 96fi) bpsis usedin AT&T's l4A Arurouncement systemfor reco'rded messages to telephoneusen [46]. MPLPC is also the algorithmchosenfor the skyphoneAeronauticalTelephone serviceusing9.6-kbpschannelson the INMARSAT maritime$atellite[47]. Another applicationof an MPLPC algorithmis the panEuropean digital mobileradio system (GsM) [48] thatallocatesa l3-kbpsdatarareto voiceencoding.Thespeechcodecthar hasbeenstandardized by CEPTfor this applicationis referredto asregular*pulse excitationLPC with long-termprediction(RpE-LTp) [4g]. This sysremtransmits260 bit blocksconsistingof 72 bits of predictionparameters and188bitsof excitationencoding.Becausethe codecdoesnot $upportvoicebanddataratesat 1200bps and above,the systemhasprovisionsfor databypassofthe codec.

Mlxed-Excltation LPC As the nameimplies,mixed-excitation Lpc (MELP) [50] usesa moresophisticated model for the excitationthan either LPC or Mpl-pc. An MELp algorithmhasbeen

ENCODER

DECODEH Volcsd/unvolced

ffil*

4*q'Fs

Cocfficirntr

Input I

Vocsl

#n'l -'5;* Figure 3.42 Multipulse linear predictivecoding. comparisonof Lpc (a) and Mpl-pc (D)encoder/decoder.

3.8 vocoDEFs

149

selectedby the U.S. govemment for $ecurevoice applicationsat 2.4 kbps [51]. Developmsnt of this algorithm has paid particular attention to background noise (as might arise in a tank). Enhancedexcitation featuresof this version of LPC are;

1. Mixed pulseandnoiseexcitation 2. Periodicor nonperiodicpulsegeneration (to improveformantreproduction) 3. Adaptivespectralenhancement theimpulsesacrossmoreof a pirchinterval) 4. Pulsedispersion(spreading 5. Fourieranalysisof residual(to improvepitchdetermination)

ResidualExclhd LPC The APC algorithm describedin Section 3.6 transmits the encodeddifference signal (prediction error or residual) determinedin the encoderso the decodercan integrateit to recover the original input. Notice that the decodedresidual waveform in the APC decoderof Figure3.34actsasthe excitationof the predictionloop.If thepredictionloopuses an LPC formulation, APC essentiallybecomesresidual-waveform excited LPC. The formal term residual excited IIC (RELP) refers to a structure that is identical to the APC block diagram shown in Figure 3.34 but differs in the mannerin which the residual is encoded.An RELP encoderdoes not encodethe residual directly but preprocessesit to allow transmissionat a lower datarate.The fundamentalpremiseof the preprocessingis that the residual consistsof a fundamental frequency and multiples (harmonics)of the fundamental.Thus, an RELP encoderencodesonly the fundamental. The decoderreconstructsthe residual (in a frequency-domainsense)by decoding the fundamental and adding in the harmonics.In the sensethat the processof extract* "residual encoding" and that of decoding and ing and encoding the fundamental is "residual decoding," the diagram in Figure 3.34 servesas a basic adding harmonicsis diagram of RELP. As commonly implemented, an RELP encoder [5?] extracts the fundamental by low-pass filtering the residual and down sampling to reduce the sampling frequency to the Nyquist rate. As an example,if a (DSP-based)low-passfilter has a cutoff of 8ff) Hz, only every fifth sample of the filtered residual is neededto meet the 1600-Hz Nyquist rate. In this manner, an RELP decoder is excited by samplesoccurring at a 1600-Hz rate-approximately the samerate as in MPLPC. Thus the residual decoder in an RELP receiveris similar to the pulse generatorof an MPLPC synthesizer.In fact, the regular pulse excitation LPC of the Europeandigital mobile systemdeterminesits pulse excitation by testing each of four sequencesobtained by taking every fourth sample of a filtered residual and selectingthe sequencewith maximum correlation to the residual sequence.Thus even the encoding algorithms of MPLPC and RELP are sometimessimilar. In a comparison of three specific coders (subband,MPLPC, and RELP), MPLPC codersprovided the best performance [53]. RELP encoding has been used in various mobile radio and satellite applications. An example of the latter is a 9.6-kbps RELP codec designedto supportconversion of analog FM voice channelsto digital voice channels[54]'

150

vorcEDrGtTtzATtoN

Mux

Figure 3.43 CELp encoderblockdiagram.

Code-ExcitedLPC code-excited LPc (CELP)[55]is yetanother example of enhanced-excitation Lpc whosebasic block diagram can be loosely representedby Figure 3.34. As indicated in the GELP specific block diagram of Figure 3.43, cELp usesa codebookfor encoding residuals.Instead of encoding a residue waveform on a sample-by-samplebasis and using that as excitation in the decoder,CELP a$$umesresidualsare not random waveforms with independentsamplesbut rather that a block of residuesamplescan be represented by one of a manageable number of waveform templates. ,,Excitation encoding" in this caseinvolves selectinga codeword in a codebookthat minimizes the overall error in the reconstructed(synthe$ized)signal. "Residual decoding" in this context implies using the received codeword as an index into the table of codewords to obtain the residual sequencechosenin the encoder.Becausea block ofresidue samples can be consideredas a residue vector, this form of coding is also referred to as vector quantization (VQ) excited LPC t561. Maximum coding efficiency is achievedby encoding long sequences(i.e., vectors with many dimensions),which, of course,implies large codebooks.Thus, much of the researcheffort of this class of codecsinvolves establishinga large enough set of vectors in the codebook that all residue vectors can be adequatelymatched.Next, an efficient procedure for searching the codebook is determined to find the vector with the best match. Because the number of vectors is necessarily large, tree-structured searchesare required, which implies the enhies in the codebook are grouped into hierarchical families of vectors. BecauseGELP coders provide good quality at low bit rates, an extensive amount ofresearch has been undertakento produce a variety ofCELp algorithms. A parlicularly efficient implementationdeterminesthe excitation codeword as a sum of sequentially determined codewords called vectors. Thus, this technique is referred to as vector sum excited LPC (VSELP) [56]. The following list identifies prominent ver-

151 3.9 ENCODER/DECODERSELECTIONCONSIDERATIONS sionsof VSELP andCELPin North Americanapplications.Otherver$ionsof CELP codersareincludedin thelist of ITU standardcodingalgorithmsin Section3.10. VSELP(vectorsumexcitedLPC);usedin NorthAmericandigitalcellularsystems (IS-54/IS-136) at 7.95kbps[57] (Qualcomm vector $um excitedLPc): usedin CDMA digital cellular QSELP (IS-95) at l3 kbPs systems DoD-CELPFederalstandard(FS-1016)at 4.8 kbps[58]

SELECTIONCONSIDERATIONS 3.9 ENCODEH|/DECODER for digitizingvoicesignalsandhasinThis chapterhasdescribedseveraltechniques thevariousaldicatedthatmanyothervariationsarepossible.This sectioncompa.res for nonvoicesignals,(3) gorithmsin termsof (l) voice quality, (2) transparency effors,and(4) delay. toleranceof transmission 3.9.1 Volce Quality quality levelshavebeentraditionally Voice codingapplicationsand corresponding quality, and synthetic classifiedas broadcastquality, toll quality, communications quality. Thesecategoriesand their respectiverangesof dataratesare qualitatively shownin Figure3.44asobtainedfrom referencetl 11.Therelativelynewcategoryof hybrid coding(e.g.,MPLPCor cELP) hasbeenadded.Table3.5,obtainedfrom refDRT variouscodersin termsof thequalitymeasures: erences[38] and[59],compares reasonably coders do t441,DAM [45], andMos t9, 11, 181.Noticethatthe low-rate well on intelligibility(DRT) but fall off significantlyin termsof subjectivemeasure$. of speechqualityof variouscodersis shownin Figure3'45' obAnotherpresentation tainedfrom referencet60l (whichcontainsa goodoverviewofITU standardcoding algorithms).This figureessentiallycontainsspecificMOS ratingsof specificcoders.' of low-ratecoders(APC,RELP,LPC) areavailablein referAdditionalcomparisons and ences[61] [62]. The caregoryof toll quality,which is traditionallyusedfor public telephonenetis no longera well-definedcategory.Toll qualityin the work voicequalityobjectives, analognetwork could be quantifiedin terms of parameterslike frequencyresponse andnoiselevel.Becausenew,low-ratecoders(e.g.,CELPs)do not reproducewaveforms,a quantitativeanalyticalmeasureof qualityis notpossible.Suppliersof highly like digitalcellularandvoiceovertheIntemetcomcompressed voicefor applications "toll quality"underidealhansmonlyuseDRT or MOS scoresto supporttheclaimof (e.g.,no bit errorsor packetloss).This practicehascontributed missionenvironments "toll quality" standard' a to a relaxationof *MOS

sco.e* pay vary by 5% ot so from one study to another'

152

vorcEDtetlzATtoN

Waveform coding ) Broadcast quality

MOS 4

Toll quality

yntheticquhlity

8

1

6 3 2 Data rare(kbps)

6

4

Figure 3.44 Generalspeechquality versusffansmissionrate.

3.9.2 Transparency for Nonvoice Slgnals Theseparation betweencommunications qualityandtoll qualitycodersin Figure3.44 also separatesthosealgorithmsthat do not perform well on nonvoicesignalsfrom thosethatdo. Thelowerdataratesof communications qualirycodersareachievedby tailoringthe algorithmsto the specificsof voicesignals,therebysacrificingmodem and facsimile performance.For a comprehensivetheoreticaltreatmentof voiceband

TABLE3.5 SpeechOualityComparleons of Common Goders 64-kbpsPCM 14.4-kbps OCELPls 3Z-kbps ADPCM(c.721) (c.728) 16-kbpsLD-CELP (c.723.1) 6.4kbpsMP-MLO 13-kbpsRPE-LTP (GSM) 9.6-kbps MPLPC(Skyphone) 9.6-kbps OCELP 8-kbp$CELP 4,8-kbps DoD-CELP Z.4-kbps LPC

95 94 94

68 70

93 99 90

68 67 54

the lowerperformance corrosponds to 3qlopacketloss.

4.3 4.2 4.'l 4.0 3.9/3.44 3.5 3.4 3.4 3.7 3,0 2.5

CONSIDERATIONS153 3.9 ENCODERiDECODERSELECTION

S.6cft AJdly

lxrcfn$ffi

"ThreeNew Figure 3.45 Speechqualityof standardencodingalgorithms.(FromR. V. Cox, SpeechCodersfrom the ITU Cover a Range of Applications,"IEEE Communications 1997.) Magazine,September data signal digitization, see reference [63]. As end-to-end digital channels become more available, there will be less need to suppolt nonvoice applications as analog voiceband signals. (Seethe discussionof V.90 PCM modems in Chapter I I ') In addition to end user applications, coders installed in the intemal portions of a network must provide acceptablequality for network-related signaling tones such as DTMF, MF, and SF. DTMF tones,in particular, must be accuratelyreconstructedbecausethey are used for end-to-endcontrol by users.G.729 at I kbps has special provisions for carrying DTMF tones. Support for MF and SF signaling is less of a problem becausetheir use is confined to specific transmission links that have been mostly replaced by newer transmission Systemsusing common-channel signaling. An additional considerationfor voice quality is the performanceof fiomevery low bit rate vocoders in the presenceof audio background noise. If a coder is tuned too tightly to processvoice signals and voice signals only, it can go off into left field when speech is superimposedon background sounds such as loud music. The low-bit-rate codersusedin digital cellular applicationshave to be pafticularly sensitiveto this sinaation.

3.9.3 Toleranceof TransmlssionErrors Of the waveform coders,the differential systems(DPCM, ADPCM, DM) are the most tolerant of transmissionelror$ and PCM the least. The threshold of perceptibility of random error$ on delta modulation is 10-3. For PCM the threshold is 104. Delta modulation i$ intelligible at random error rates as high as lOVo,but PCM is unintelli-

154

vorcEDrcrrzATtoN

gibleat l7d elror rates.In bursterrorenvironments thetolerances of PCM anddifferentialsystemsaremorenearlythesame.(If the most significantbit of a pcM codeword is in error,it doesnot matterif the leastsignificantbits arealsoin error.) Thoseparameters of syntheticqualitycodersthatarecriticalfor voicereconstruction areusuallyredundantlyencoded. Errorcorrectionofcritical parameters in digital mobilesystemsallowsintelligiblevoiceat errorratesashigh as l Zo. 3.9.4 Delay The effectof encodinganddecodingdelayof a voicedigitization/compression algorithm mustbeconsidered in thecontextof theparticularapplication.If theapplication involvesinsertingartificial delayof morethan 10 msecinto local (analogrconnections,echo/singing controlwill haveto beadded.Furthermore, if a significantamount of delayis addedinto a long-distance circuit,existingechocancelersmay not have enoughdelaycapacityto accommodate thedelayinsertion. Experience wittr satellite-ba.sed voiceconnections indicatesthatroundtripdelaysonthe orderof 2ff) mseccanbe toleratedwitlroutsifficant userdissatisfaction. As indicated in the following, coding/compression algorithms,in themselves,do not approach this limit. ADPCM (G.726) LD-CELP(G.728) CS-CELP(c.729)t64l ACELP(G.723.1)

0.125msec 2.5msec l0 msec 30 msec (plus 7.5 msec of look-ahead)

Although the roundtrip encoding/decodingdelays (which are double rhe above numbers) do not approachthe 200-msecthreshold, they can add to oflrer systemdelays to exceed the maximum desirable delay. System factors such as interleaving for error correction and packet delay on TDMA mobile systemscan add another30 msec or so in eachdirection. If individual compressedvoice packetsare carried through an ATM or packet-switchednetwork, delays in excessor 200 msec are easily produced. (see Chapter I 0 for a discussionof theseapplications.)

3.10 ITU.TCODINGSTANDARDS Thefollowing list identifiesvariousdigital voicecodingstandards of theITU: o.7ll standardfor speechcodecsthat providestoll quality audio at 64 kbps usingeitherA-lawor p-lawPCM. G-721 standardfor speechcodecsthat providestoll quality audio at 32 kbps usingADPCM. G'722 standardfor speechcodecstharprovideshigh-quality(program)audioat 64 kbpsusingsubbandADPCM (SB-ADPCM).Thealgorithmusesa l6kHz samplerateto captureaudiofrequencies between50 and 7000Hz. Two ADPCM subbandsare used by this srandardto give audio performancesuperiorto a single-bandADPCM algorithmoperatingat the samebit rate.

REFERENcES 155

qualityaudioat20or40kbps thatprovidestoll Standardforspeechcodecs usingADPCM. G.723.1 Standardfor speechcodecsoptimizedfor modems.It providestoll quality audioat 6.4kbps(MP-MLQ) or 5.3kbps(ACELP). at 16,24,32,or40kbps(using?-,3-,4-,ot5G.726 AdaptivedifferentialPCM bit samples). G.726 for use in packetizedspeech G.727 An extensionof Recommendation TheADPCM samplesaredividedinto 2, 3, or 4 corebits and0, $ystems. bits.Thecorebitsprovidefor thebasicfunctioning l, 2, or 3 enhancement bits addqualityto thatprovided of the algorithmwhile the enhancement bits can be by the core bits. In overloadsituations,the enhancement discardedwhile thecorebits providebasicquality. for speechcodecsthatprovidesneartoll qualityaudioat 16kbps G.7ZB Standard usinglow-delayCELP (LD-CELP).G.728encodesfive p-law or A-law PCM samplesinto 10-bit,linearpredictivecodewordsat 1600codewords per second. for speechcodecsthatprovidestoll qualityaudioat 8 kbpsusing G.729 Standard CELP. G.723

REFERENCES I 2 3 4 5 6

"The Rolm Computerized Branch Exchange: An Advanced Digital J. M. Kasson, PBX," Computer Magazine, June 1979, p' 24. "speech Technology in Consumer Products," Speec* G. Frantz and K.-S. Lin, Technology,Apr. 1982, pp. 25-27 . "Customieed LPC Vocabulary L.ets Cars "lalk," Speech Technology, S. Finkelstein, Sept./Oct.1983,pp. 65-69. "The Compact Disc Digital Audio M. G. Carasso, J. B. H, Peek, and J. P. Sinjou, System," Phillips TechnicalReview,Vol. 40, No. 6, 1982,pp. 151- 155. L, Rabiner and B.-H. Jtang, Fundamentals of Speech Recogrlition, hentice-Hall, Englewood Cliffs, NJ, 1993. "speech Synthesis from Text," IEEE Communications Magazine, Jan. Y. Sagiska,

pp,35-41. 1990, 7 I

J.L. Flanagan, Speech Analysis, Synthesis,and Perception, 2nd ed. Springer-Verlag' New York. 1972. "An Overview of the SPHINX Speech K. F. Lee, H. W. Hon, and R. Reddy, Recognition System," IEEE Transactionson AcousticsSpeechand Signal Processing, Jan. 1990,pp.35-45.

N. S. Jayant and P, Noll, Drgital Coding,of Wavefotms: Principles and Applications to Speechand Video, Prentice-Hall, Englewood Cliffs, NJ, 1984. "Speech l0 J. Flanagan, M. Schroeder,B. Atal, R. Crochiere, N. Jayant, and J. Tribolet, 'll0-737 Coding," IEEE Transaction$on Communicatiorar,Apr. 1979, pp. "Quality Assessmentof Speech Coding and Speech 11 N. Kitawaki and H. Nagabuchi, SynthesisSystems,"IEEE CommunicationsMagazine, Oct. 1988, pp. 36-214. 9

156

tz t3 I4

l5 l6

t7

r8 r9 20

21 22

24

)\ 26

27 ?8

VOIGEDIGITIZATION B. J. McDermott, C. Scagliola, and D. J. Goodman, "Perceptual and Objective Evaluation of Speechhocessed by Adaptive Differential PCM," Bell SystemTechnical Journal, May/June 1978,pp. 1597-1618. "The D3 Channel Bank Compatibility Specifrcarion-Issue 3," Technical Advisory No. 32, American Telephone and Telegraph Company, Basking Ridge, NJ, Oct. lgi7. M. B, Akgun, "Transmission Objectives for a Subscriber Loop System Digital Voice Codec," paper presented at the International Symposium-Subscriber Loops and '7.L1-7.1.7. Services,Ottawa, 1974,pp. H. Nann, H. M. Straube, and C. P. Villars; "A CompandedCoder for an Experimental PCM Terminal," BeIl SystentTechnicalJoumal, Jan. 1962, pp. 173-226. K. E. Fultz and D. B. Penick, '"The Tl Carrier System," BelI SystemTechnical foumal, Sept. 1965,pp. 1405-1451. H, H. Henning and J. W. Pan, "D2 Channel Bank: System Aspects," Bell System TechnicalJournal, Oct. 1972,pp. 1641-1657. W. R. Daumer, "Subjective Evaluation of Several Efficient Speech Coders," IEEE Transactions on Communicarrions,Apr. 1982,pp. 655*662. N. S. Jayant, "Step-Size Transmitting Differential Coders for Mobile Telephony," Bell SystemTechnicalJoumal, Nov. 1975,pp, 1557-1581. D. L. Duttweiler and D. G. Messerschmitt, "Nearly InstantaneousCompanding and Time Diversity as Applied t0 Mobile Radio Transmission," IEEE International Communications Conference, I 975, pp. 4*12-40-15. I. M. McNair, Jr., -'Digital Multiplexer for ExpandedRural Sewice,,,Eelt l_absRecord, Mar. 1972,pp. 80-86. H. K. Dunn and S. D. White, "Statistical Measurementson Conversational Speech," Joumal of Acoustic Societyof America, Jan. 1940,pp. 278-288. J. W. Bayless, S. J. Campanella,and A. J, Goldberg, "Voice Signals: Bit by Bir," IEEE Spectrurn,Oct. 1973,pp.28-34. M. Decina and G. Modena, "CCITT Srandardson Digital Speech hocessing," IEEE Journal on SelectedAreas in Communicarions,Feb. 1988, pp. 227-234. G. Williams and H. Suyderhoud, "subjective PerformanceEvaluation of the 32 kbit/s ADPCM Algorithm," I EEE Globecom P roceedings, I 984, pp. 23.2.1-23.2.1.8. J, A, Greefkes and K. Riemens, 'Code Modulation with Digitally Controlled Companding for SpeechTransmission," Phillips Technical Review,Vol. 31, 1970, pp. 335-353. J. E. Abate, "Linear and Adaptive Delta Modulation," Pro ceedingsof IEEE, Mar, I 967, pp. 298-308. N. S. Jayant and A. E. Rosenberg,"The heference of Slope Overload to Granularity in the Delta Modulation of Speech," Bell SystemTechnical Journal, Dec. l97l, pp.

3rr7-3125. 29 J. A. GreeEkesand F. de Jager,"ConrinuousDelta Modularion,"Phillips Research Reports,Vol. 23, 1968,pp. 233-246. 30 J. A. Greefkes,"A Digitally CompandedDelta Modulation Modem for Speech Transmission," Proceedings of IEEE InternationalCommunitations,lgT0,p. 7-37to 7-48.

1$7 REFEHENCES "AdaptivePredictiveCodingof SpeechSignals,"Bell B. S. AtaI andM. R. Schroeder, SystemTechnicdI Joumal, Oct. 1970,pp. I 973- I 986. 32 P. Noll, "On Predictive Quantizing Schemes,"BelI SystemTechnicalJoumal, May/June1978,pp. 1499-1532. 33 M. R. Schroeder,"Linear Predictive Coding of Speech:Review and Current Mag,azine,Attg.1985,pp. 54-61. Directions,"IEEE Communications 'The GovemmentStandardAdaptive hedictive Coding Algorithm; 34 T. Tremain, 3l

Feb./IVIar. 1985,pp, 52-62. APC-04,"SpeechTechnology, "An Dynamic of 16 kb/s Sub-BandCoderPerformance, R. E. Crochiere, Analysis 35 Journal'Qct, andChannelErrors,"Bell SystemTechnical Range,TandemConnections, 1978,pp. 2927-2951. of 36 P. Mermelstein,"G.722A New CCITT CodingStandardfor Digital Transmission Magazine,Jan' 1988,p' 815. WidebandAudio Signals,"IEEE Communications Areasin 3'7 X. Maine,"7 kHz Audio Codingwithin 64 kbit/s,"IEEEJoumal on Selected Feb.1988,pp. 283-298. Communications, 38 N. S,Jayant,"High-QualityCodingof TelephoneSpeechandWidebandAudio"'IEEE Magazine,Jan.1990,pp. 10-20. Communications "Low-Bit-Rarc Voice Digitizer Hits the Market at 2,400 bls," Speech 39 L. Radin, Technology,Sept./Oct.1983,pp. 92-95. 40 H. Dudley,"RemakingSpeech,""Ioanralof the AcousticSocietyof Amerfua,Oct. 1939' pp. 169-177. of SpeechSignals,Prentice-Hall, 4l L. R. RabinerandR. W. Schafer,Digital Processing EnglewoodCliffs, NJ, 1978. 42 T. E. Tremain, "The Govemment StandardLinear Predictive Coding Algorithm: Apr. 1982,pp. a0-49. LPC-10,"SpeechTechnolagy, 43 J. P. Campbelland T. E. Tremain,"Voiced/UnvoicedClassificationof Speechwith Applicationsto the U.S. GovemmentLPC-I0E Algorithm," IEEE Intemational 1986,pp. 473476. on Acoustics,Speech,andSignalProcessing, Conference "Evaluating hocessedSpeechUsing the DiagnosticRhyme Test," 44 W, D. Voiers, SpeechTechnolagy, Jan./Feb.1983,pp. 30-39. 45 W. D. Voiers, "Diagnostic Acceptability Measurefor SpeechCommunication

46 47

48

49

Systems,"ProceedingsIntemational Conferenceon Acoustics,Speechand Signal Processing, May 1977,pp.204*207. "TheTelephoneVoiceof theFunre," Bell l-absRecord, B. S.Atal andR. Q. Hofacker, pp. 1985, 4-10. July "SpeechCodecfor the I. Boyd,C. B. Southcoft,D. P, Crowe,andP. J. Bolingbroke, Engineering, Service,"BritishTelecommunications SkyphoneAeronauticalTelephone July 1986,pp. 83-91. "speechhocessingin thePanEuropeanDigital J. E. Natig,S. Hansen,andJ. deBrito, MobileRadioSystem(GSM)-systemOverview,"IEEE GlobecomProceedings,1989, pp.29B.l. l-298.1.1.5. "SpeechCodec J.P.Petit,C. Galand,andM' Russo, K. Hellwig,P.Vary,D. Massaloux, for the European.MobileRadio System,*IEEE GlobecomProceedings,1989' pp' 298.2.1-298.2.r.s.

158 50

vorcEDrcrrzATroN A. V. McCree and T. P. Bamwell. "A Mixed Excitation LPC Vocoder Model for Low Bit Rate SpeechCoding," IEEE Transactions on Speechand Audio Processing,July

1995,pp. 242-250. 51 "Specificationsfor the Analog to Digital Conversionof Voice by 2,400Bir/Second Mixed ExcitationLinear Prediction,"Deparffnentof DefenseDraft, May 28, lggS (www.plh.af.miUddvpc/melp.hnn). 52 C. K, Un and D. T. Magill, "The Residual-Excited Linear PredictionVocoderwith Transmission RateBelow 9.6 kbot/s,"IEEE Transactions on Communications, Dec. 1975,pp, 1466-1474. "A Studyof Three 53 K. K. PaliwalandT. Svendsen, Coders(Sub-band, RELP,andMpE) fbr Speechwith Additive White Noise,"Proceedings IEEE IntemationalConference on Acoustics, Speech and SignalProcessing, 1985,pp.43.5,I-43.5.1.4. 54 A. J. Morris and S. Kay, -'TelephonyEnth Station,"IEEE Globecomproceedings, 1989,pp,48.I. I -48.1.1.4. 55 M, R. Schroederand B. S. Aral, "Code-ExcitedLinear prediction (CELP):

56 57 58

59 60 61

62

63 64

High-QualitySpeechat Very Low Bit Rates,"IEEE InternationalConferenteon Acoustics, Speech andSignalProcessing, 1985,pp, 25.1.1-?5.1.1.4. "Gain-Adaptive J. H. ChenandA. Gersho, Vector Quantizationwith Applicationto SpeechCoding,"IEEE Transactions on Communicarrons, Sept.1987,pp. gl8-930. "Vector SumExcitedLinear hediction (VSELP)7950Bit PerSecondVoice Coding Algorithm,"TechnicalDescription,Motorola,Shaumburg, IL, Nov. 14, 1989. J.P. Campbell,V. C. Welch,andT. E, Tremain,'An Expandable Error-hotected48fi) BPSCELPCoder(U.S.FederalStandard 4800BPSVoiceCoder\,"IEEEIntemational Conference on Acoustics, Speechand SignalProcessing,1989, pp. 735-737. T, S. Rappaport,WirelessCommunications: Principlesand Practice,Prentice-Hall, UpperSaddleRiver,NJ, 1996. R. V. Cox,'*IhreeNew SpeechCodersfrom theITU Covera Rangeof Applications," I EEE Communications Magasrne,Sept.1997,pp. 4O47. R. D. HoyleandD. D. Falconer,'AComparison of Digital SpeechCodingMethodsfor Mobile Radio Systems,"IEEE Journal on SelectedAreasin Communications, June 1987,pp.915-920. N. Kitawaki, H. Nagabuchi,and K. Itoh, "Objective euality Evaluation for Low-Bit-Rate Speech Coding Systems,"IEEE Journal on Selected Areas in Communitatio,ns, Feb,1988,pp. U2*24'7. J.B. O'Neal,Jr.,"WaveformEncodingof VoicebandDataSignals,"proceedings of the IEEE,Feb.1980,pp. 232-247. C. Schniderand M. H. Sherif, "The Road to G.729..ITU g-kb/s SpeechCoding Algorithm with WirelineQuality,"IEEE Corwnunications Magazine,Sept.1997,pp. 48*54.

PROBLEMS 3.1 Assume a signal con$istsof three tones:one at I kHz, one at 10 kHz. and one at 2l kHz. What tones will be presenrar rhe ouFut of a PAM decoderif the sam-

PRoBLEMS 159

3.2

3.3 3.4

3.5 3.6

3.7

pling rate is I ? kHz and the input is unfiltered? (Assume the output filter cutoff frequency is 6 kHz.) Derive an expressionfor the averagequantization noise power that occurs when the decoder output samples are offset from the center of a quantization interval by adistance equal to 25Voof the interval. (The outputvalues are aLtheTSVa point insteadof the 507opoint.) How much degradationin decibelsdoesthis offset represent(assuminguncorrelatedoffsets)? How much does the signal-to-noise ratio of a uniform PCM encoder improve when I bit is addedto the codeword? A black-and-whitetelevision signal has a banilwidth of about 4.2 MHz. What bit rate is required if this signal is to be digitized with uniform PCM at an SQR of 30 dB? Use a sampling-rate-to-Nyquist-rateratio comparableto that used for PCM voice encoding. How much dynamic range is provided by a uniform PCM encoderwith l2 bits per sampleand a minimum SQR of 33 dB? What is the signal-to-quantizing-noiseratio produced by a segmentedp255 PCM coder when the input signal is a full-range triangular wave? (Assume the repetition frequency is low enough that the bandlimiting filter does not change the waveform significantly.) Given a samplevalue of 420 mV for a p255 PCM encodercapableof encoding a maximum level of 2 V, determine each of the following: (a) The compressedp255 codeword (b) The linear representationof the compressedcode (c) The A-law code obtained by converting from the p255 code (d) The pJaw code obtained by converting back from the A-law code

3.8 Given the following p255 codewords, determine the (noninverted) codeword that representsthe linear sum (0 I 10 1001),(l 01 I 0I I I ). 3.9 Generatean encoding table (i.e., list the quantieationintervals and conesponding codes) for the magnitude of a piecewise linear code with segmentslopes 1, j, i, and f. Assume four equally spacedintervals are in each segment.Assume all intervals in the first segmentare of equal length (as in A-law coding). What is the signal-to-noiseratio of a full-amplitude samplefor the coder of Prob3.10 lem 3.9? 3.11 What is the signal-to-noiseratio of a maximum-amplitude samplein the first linear $egmentof the preceding coder? 3.12 What is the dynamicrangeimplied by Problems3.10 and 3.11? 3.13 For the encoderin hoblem 3.9, how many bits are required for unifotm encoding of the samedynamic range and sameminimum quantization interval? 3.f4 A uniform PCM system is defined to encode signals in the range of -8159 to +8159 in quantizationintervals of length 2. (The quantizationinterval at the origin extendsfrom -1 to +I.) Signals are encodedin sign-magnitudeformat with a polarity bit = 1 denoting a negative signal. (a) How many bits are required to encodethe full range of signals?

160

votcEDlcrrzAloN (b) How manyunusedcodesarethere? (c) Determinethe quantizationnoise,noisepower,andsignal-to-noise ratio (in decibels)of eachof thefollowing samplevalues;30.2,123.2,-2336.4,and 8080.9.

3.15 Repeatpart(c) in Problem3.l4 for piecewiselinearp255PCM. 3.16 Given two A-law piecewiselinear(noninverted) codewords(00110110and 00101100),determinetheir linearrepresentations, addthem,andconvefiback to compressed representation. 3.17 A D3 charrnelbankuses"robbeddigit" signalingwhereintheleastsignificantbit of everysixthframeis stolenfor signaling.Determinetherelativeincreasein overall quantizationnoiseproducedby this processunderthefollowing conditions: (a) The decodermerelytreatsthe signalingbit as a voicebit and decodesthe PCM sampleaccordingly. (b) The decoderknows which bits are signalingbits and generates an output samplecorresponding to themiddleof thedouble-length quantizationinterval definedby the7 availablebits.(Thisis theactualtechniquespecifiedfor D3 channelbankdecoders.) 3.18 If 2 bitsper sampleareaddedto a PCM bit stream,how muchcanthedynamic rangebeincreased if thequantization intervalsareadjustedto improvetheSQR by 3 dB? 3.19 How muchcantheSQRof a digitizedvoicesignalbeincreased if thebandwidth is reducedby 30Voandthedynamicrangeis unchanged? 3.20 AnA-law PCM encoderwith a perfectzero-levelthresholddetect$anarbitrarily smallinputsinewavebecause thefirst quantization intervaldoesnot straddlethe origin.Whatis thepowerlevelof thedecodedoutputsignal?Assumethepower level of a full amplitudesinewaveis +3 dBm. 3.21 Determinethe sequence of four p255 PCM samplesfor a 2000-Hztoneat -6 dBm whenthefirst sampleoccursat a phaseof 45". 3.?? RepeatProblem3.21forAlaw signals. 3.23 DeterminetheA-law codewordsthatcorrespond to thep-law codewordsof the digital milliwatt signalgivenin Example3.3.Assumeboth systemsarescaled to the samemaximumsignalpower.

DIGITALTRANSMISSIONAND MULTIPLEXING A fundamental considerationin the design of digital transmission $ystemsis the selection of a finite set of discreteelechical waveforms for encoding the information. In the realm of digital communications theory these discrete waveforms are referred to as signals. The sameterminology is used in this chapter with the understandingthat signals in the present context refer to intemal waveforms (pulses) of a transmission systemand not the control information (signaling) usedto set up and monitor connections within a network. In communicationstheory terminology, signalprocessing refers to filtering, shaping, and transformations of electrical waveforms, not the interpretation of conffol signalsby the processorof a switching machine. A secondaspectof digital transmissioninvolves defining certain time relationships between the individual transmissionsignals. The source terminal transmits the individual signals using predefined time relationships so the receiving terminal can recognize each discretesignal as it arrives. Invariably the establishmentof a proper time baseat the receiver requirestransmissioncapacity abovethat neededfor the digital information itself. Over relatively short distances(as within a switching system or a computer complex), the timing information (clock) is usually distributed separately from the information-bearing signals. Over long distances,however, it is more economical to incorporate the timing information into the signal format itself. In eittrer casethe timing information requireschannelcapacity in terms of bandwidth, datarate, or code space. This chapter discussesthe most common digital signaling techniquesfbr wireline transmissionsystems.Thus the subject of this chapteris often referred to as line coding. Thesetechniquesaregenerallyapplicableto any transmissionsystemffansmitting digital signals directly in the form of pulses (such as coaxial cable or optical fiber). The fact that thesetechniquesinclude low-frequency componentsin their transmitted spectrum leadsto their also being called low-pass or basebandtransmissionsystems. In Chapter 6 we discuss bandpasstransmission sy$tems,that is, radio systems or voicebandmodemsthat require modulation and carrier frequencytransmission.Chap-

161

162

DtcrrALrRANSMtsstoN ANDMULTIpLEXtNc

ter 8 discusses someadditionalbaseband transmission formatscommonlyusedin optical fiber system$. Thefollowing discussions concentrate on systemandapplicationlevel considerations of digital transmission. Analytic detailsof pulsetransmission systemsarenot emphasized sincematerialof thisnatureis availablein all workson digitalcommunicationstheory.somefundamentals of pulsetransmission systemsarepresented in AppendixC, wheremanyof theequationspresented in this chapterarederived.

4.1 PULSETRANSMISSION All digitaltransmission systemsaredesigned aroundsomeparticularform of pulseresponse. Evencarriersystemsmustultimatelyproducespecificpulseshapesat thedetectioncircuitry of the receiver.As a first step,considerthe perfectlysquarepulse shownin Figure4.1.Thefrequencyspectrumcorresponding pulse to therectangular is derivedin AppendixA and shownin Figure4.2. It is commonlyreferredto as a sin(x)/xresponse: -_ sin(roll2) aTl2

1'((0= 1(Il+

(4.1)

where co= radian ftequency 2nl, 7 = duration of a signal interval Notice that Figure 4.2 also provides the percentageof total $peckumpower at various bandwidths. As indicated, 90Voof the signal energy is contained within the first spectralnull at f = l/T.The high percentageof energy within this band indicatesthat the signal can be confined to a bandwidth of l/T and still passa good approximarion to the ideal waveform. In theory, if only the samplevaluesat the middle of eachsignal interval are to be preserved,the bandwidth can be confinedto IlZT.From this fact the maximum basebandsignaling rate in a specified bandwidth is determined as

R** = 2BW

- 1r

T

Z

Time -**

Figure 4.1 Definitionof a squarepulse.

(4.2)

4.1 PULSETFANSMISSION

163

t *ro Frtrtion of out-of-bmd pomr (
.E E

t E fitt

fl -ao

u

-30 Froqumcy (Hrl

pulsewithdurationT. of square Figure4.2 Speckum whereR = signalingrate,= IlT BW = availablebandwidth Equation4.2 statesa fundamentalresultfrom communicationstheory creditedto Harry Nyquist;The maximumsignalingrate achievablethrougha low'passbandwidth with no intersymbolinterferenceis equalto twice thebandwidth.This rateRô is sometimesreferredto asthe Nyquistrate. ofthe pulsesareeasiestto visualize,preservation Althoughdiscrete,square-shaped squareshaperequireswide bandwidthsand is thereforeundesirable.A more typical shapefor a singlepulseis shownin Figure4.3.Theringingon bothsidesof themain accompaniment to a channelwith a limitedbandwidth. partof thepulseis a necessary

l(tl

I I

\ ,''i*t t

I I I I I I

-3r -2

+--rlT

2T

4T

Time.*

Figure 4.3 Typical pulseresponseof a bandlimitedchannel.

164

DTctTALTRANSMtSStoNANDMULTtpLEXtNG

Normally, a digital transmissionlink is excitedwith squarepulses(or modulated equivalents thereofl,but bandlimitingfiltersandthetransmission mediumitselfcombineto producea response like theoneshown.Figure4.3 showspulseoutputin negative time so the centerof the pulse occursat f = 0. Actually, the durationof the preringingis limited to thedelayof thechannel,the filters,andtheequalizers. An importantfeatureof thepulseresponse shownin Figure4.3 is that,despitethe ringing,apulsecanbetransmitted onceeveryT seconds andbedetected atthereceiver withoutinterference from adjacentpulses.Obviously,the sampletime mustcoincide with the zerocrossingsof the adjacentpulses.Pulseresponses like the oneshownin Figure 4.3 can be achievedin channelbandwidthsapproachingthe minimum (Nyquist)bandwidthequalto one-halfof the signalingrate.AppendixC describes pulsetransmission designin moredetail. 4.1.1 Intersymbol Interference As the signalingrateof a digital transmission link approaches the maximumratefor a givenbandwidth,boththechanneldesignandthe sampletimesbecomemorecritical.Smallperturbations in thechannelresponse or the sampletimesproducenonzero overlapat the sampletime$calledintersymbolinterference. The main causesof intersymbolinterferenceare; l. Timing inaccuracies 2. Insufficient bandwidth 3. Amplitude distortion 4. Phasedistortion

4.1.2 Timlnglnaccuracies Timing inaccuraciesoccuring in either the transmitter or the receiver produce intersymbol interference.In the transmitter, timing inaccuraciescauseintersymbol interference if the rate of transmission does not conform to the ringing frequency designedinto the channel.Timing inaccuraciesofthis type areinsignificant unlessextremely sharp filter cutoffs are used while signaling at the Nyquist rate. Since timing in the receiver is derived from noisy and possibly distorted receive signals, inaccurate sample timing is more likely than inaccuratetransmitter timing. Sensitivity to timing errors is small if the transmissionrate is well below the Nyquist rate (e.9.,if the transmissionrate is equal to or lessthan the channelbandwidth, as opposedto being equal to the theoreticalmaximum rate of twice the bandwidth; seeAppendix C).

4.1.3 InsufflclentBandwidth Theringingfrequency shownin Figure4.3isexactlyequaltothetheoretical minimum bandwidth of the channel. If the bandwidth is reducedfurther, the ringing frequency is reducedand intersymbol interferencenecessarilyresults.

TRANSMISSION 165 4,2 ASYNCHRONOUS VERSUS SYNCHRONOUS

theNyquistrate,but do sowith Somesystemspurposelysignalat a rateexceeding interference accounted for in thereceiver.These prescribedamountsof intersymbol the pafrial-response systems-so calledbecause systemsarecommonlyreferredto as pulse. The most the time of a single channeldoesnot fully respondto aninputduring arediscussed in a latersection. $y$tems commonformsof partial-response 4.1.4 AmplitudeDlstoilion systemsinvariablyrequirefiltersto bandlimittransmitspectrums Digital transmission in receivers. Overall,thefiltersaredesignedto proandto rejectnoiseandinterference ducea specificpulseresponse.When a transmissionmediumwith predetermined canbeincludedin theoverallfilter design. is used,thesecharacteristics characteristics However,the frequencyresponseof the channelcannotalwaysbe predictedadeis referredto asamplitude quately.A departurefrom the desiredfrequencyresponse pulsedistortions(reducedpeakamplitudesandimproperringing distortionandcauses for irregularitiesin thefrequencyrefrequencies) in thetime domain.Compensation sponseof thechannelis referredto asamplitudeequaliqation. 4.1.5

Phase Distortion

of asthe superposition Whenviewedin the frequencydomain,a pulseis represented If therelative frequencycomponents with specificamplitudeandphaserelationships. arealtered,amplitudedistortionresultsas amplitudesof the frequencycomponents arealtered,phasedistortionocof the components above.If the phaserelationships of a signalexcurs.Basically,phasedistortionresultswhenthefrequencycomponents link. Compensation of phase periencedifferingamountsof delayin thetransmission inFor a goodtutorialon equalization, distortionis referredto asphaseequalization. seereference cludinga descriptionof an automaticequalizerfor datatransmission, is availablein reference[2]. [1]. A gooddescriptionof DSPbasedequalizers

4.2 ASYNCHRONOUSVERSUSSYNCHRONOU$TRANSMISSION involvingtwo fundamentallydifTherearetwo basicmodesof digital transmission a timebase(sampleclock)in thereceivingterminal ferenttechniques for e'rtablishing transmisis asynchronous link. Thefirst of thesetechniques of a digitaltransmission Within an of groupsof bits or characters. transmissions sion,whichinvolvesseparate individualgroupa specificpredefinedtime intervalis usedfor eachdiscretesignal. timesof the groupsareunrelatedto eachother.Thusthe However,the transmission for receptionof eachgroup' sampleclock in the receivingterminalis reestablished transmission, digital signalsaresent With the secondtechnique,calledsynchronous continuouslyat a constantrate.Hencethereceivingterminalmustestablishandmainto theincomingdatafor an indefiniteperiod tain a sampleclock thatis synchronized of time.

166

DIGITALTHANSMISSION ANDMULTIPLEXING

Tran$fiitted waveform

Receivad waneform

l t t | i l 1

l l l l l l r l

ldeol Ea|npletimer for each8-bit character

Figure4.4 Asynchronous transmission.

4.2.1 AsynchronousTransmiesion Between transmissions anasynchronous lineis in aninactiveor idlestate. Thebeginningof eachtransmission groupis signifiedby a startbit.Themiddteof thestartbit is determined, and succeeding information bits are sampled at a nominal rate beginning at the middle of the secondbit interval. Following the information symbols, one or more stop bits are transmitted to allow the line to return to the inactive state.* Figure 4.4 shows an asynchronousmode of operation commonly used for low-speed data communications. As shown in Figure 4.4, the detection of each information bit is accomplishedby ideally sampling the input waveform at the middle of each signal interval. In practice, sample times depart from the ideal depending on how much the start bit is comrpted by noise and distortion. Since the sampletime for eachinformation bit is derived from a single start bit, asynchronoussystemsdo not pedorm well in high-noise environments. of course,more than one start bit could be usedto improve the accuracyof the starting phase of the sample clock, but this would complicate the receiver and add more overheadfor transmissionof timing information. Sample timing enors also arise if the nominal rate of the sample clock in the receiver is different from the nominal rate of transmission at the source.Even though the start bit might define the proper starting phase for the sample clock, an offset in the clock frequency of the receiver causeseach successivesampletime to drift faflher from the centerofthe respectivesignal intervals. Since the very use ofthe term'hsynchronous" implies a free-running clock in the receiver, a certain amount of drift is inevitable in all asynchronoussystems.The maximum length of each symbol group or characteris determined by the limits of the initial phase inaccuraciesand the maximum expected frequency difference between the kansmitter and receiver clocks.

-Originally,

stop bits were inserterl to allow electromechanical equipment enough time to reset before the next character anived. With elechonic equipment thc only purpose of stop bits is to allow a start bit to always be a fransition to a space(logic 0).

4.2 ASYNCHRONOUSVERSUS$YNCHRONOUSTRANSMISSION

167

The main attraction of asynchronoustransmissionis the easewith which it determines the sampletimes in the receiver. In addition, asynchronousffansmissionautomatically provides characterframing and is inherently flexible in the range of average data rate$ that can be accommodated. For high rates, one character after another is transmitted. l,ower data rates are automatically accommodated by increasing the idle time between characters.In contrast, a synchronousreceiver must track changesin a transmitter rate before it can sample the incoming signals properly. Normally the receive clock of a synchronoussystemcan be adjustedonly quite slowly and only over a nilrow range. Hence an asynchronoussystem is more naturally suited to applications where the data rate varies. Synchronoustransmissionsystemscan suppofi variable information rates,but the task of adjusting the information rate falls upon higher level processes(data link protocols) that insert null codesinto the bit stream.The null codesare used as filler when a source has nothing to send. This form of transmissionis sometimesreferred to as "isochronous." An isOChrOnous mode Of Operationis required whenever a SynChronous line carries data from an asynchronoussource. The major drawbackof asynchronoustransmissionis its poor performancein terms oferror rateson noisy lines. Thus asynchronoustransmissionevolved for applications where implementation costs dominated performance considerations.Asynchronous transmissionhas been usedin voicebanddata sets(modems)for transmissionratesup to 1200 bps. For digital telephony, something similar to asynchronoustransmission was proposedfor two-wire digital subscriberloops. Thesesystemsprovided a full-duplex (four-wire) circuit by transmitting bursts of data alternately in each direction on "ping-pong" a single pair of wires. Thus these systemsare sometimesreferred to as ffansmissionsystems;they are not ffuly asynchronoussinceeachtransmissionin each direction occurs at prescribedtimes allowing timing information in one burst to carry over into the next burst. The use ofasynchronous tran$missionon long-distancetransmission links is obsoletebecausethe cost of electronicsfor betterperforming synchronous transmissionis no longer a consideration'

4.2.2 Synchronou$Transmis$ion links usedin thetelephonenetTl linesandall otherinterofficedigital transmission Thus the line-codingformat for exclusively. transmission works use synchronous thateachregenerative to en$ure special considerations thesesystemsmustincorporate incomingsignaling clock to the a local sample repeateror receivercan synchronize imply thata certainminirequirements rate.Generallyspeaking,the synchronization indicationof sigprovide continuous mum densityof signaltransitionsis requiredto relied upon to provide a naling boundaries.Often, purely randomdatacannotbe be provisions mu$t madeto insufficientnumberof transitions.In thesecasesceftain sertartificialtransitionsinto thetransmittedwaveforms.Althoughtheseextratransioverhead,the lossin capacitycan be tions imply a certainamountof transmission for ensuringthe existof f,rvetechniques relativelysmall.Followingaredescriptions enceof signaltransitionsfor timing recovery:

168

l. ?. 3. 4. 5.

DrcrrALTRANSMtsstoN ANDMULTIpLEXtNG

Sourcecoderestriction Dedicatedtiming bits Bit insertion Datascrambling Forcedbit errors

A sixthtechnique-insertingthetransitionsinto the signalwaveformsthemselvesis discussed in Section4.3. Source Code Restrlction Onemeansof ensuringa sufficientnumberof signaltransitionsis to re$trictthecode setor datapaftemsof thesourcesothatlong,transition-free datasequences do not occur.Historically,PCM channelbanksin theNorth Americantelephonenetworkprecluded all-0's codewordssince the original line code of rl lines producedno transitionsfor 0's. In thecaseof 8-bitPCM codewords, theexclusionof a singlecodeword represents a lossin transmission capacityof only onepartin 256.Relyingon sourcecodingto ensuresufficienttransitionsin the line codehasone very significantdrawbackThetransmission link cannotbeusedfor newapplications wherethesourcedoesnotexcludetheunwanteddatapattems.Forthisreason,thetotal capacityof a Tl line couldnot be usedfor randomdatauntil a new line code[binary eight-zerosubstitution(B8ZS)]wasintroduced. Dedlcated Timing Blts As an altemativeto excludingtransition-freedatapatterns,the line itself can periodicallyinsetttransition-bearing bits into the datastream.Thesebits areinsertedat regularintervals,independently ofthe sourcedata,to ensuretheexistenceof a minimumnumberof signaltransitions. Thussomefractionof thechannelcapacityis dedicatedto timing bits. As an example,the original DataphoneDigital service(DDS) offering for dara communications over Tl lines providesa maximumcapacityof 56 kbps for each channel.when carriedwithin a 64-kbpsTl channel,only 7 of the 8 bits in eachrime slot of the channelareavailablefor the user.Amongotherfunctions,the unusedbit in eachtime slotprovidesan assurance thatall 8 bits of a time slot arenot 0. Several fiber systemsdescribedin Chapter8 alsouseexplicit timing bits asinherentaspects of their line codes.The densityof timing pulsesin thesesystemsrangesfrom I in S bitsto I in 20bits.Noticethatinsertionof a dedicated rimingbit is essentially thesame procedureusedto establi$htiming for asynchronous transmission. In a synchronous receiver,however,a sampleclock is obtainedby averagingthe timing information overa largenumberof timing transitions,notjust one. Blt lnsertion In theprecedingDDS exampleI bit of every8 bits in a time slot is dedicatedto ensuringsufficienttiminginformationin thebit stream.Anotherpossibilityfor preclud"From

an inf'ormation theory point of view the loss in voice information is even lower since the probahility of occunence of the all-()'s codeword is much less than I in 256.

TRANSMISSION 169 VEHSUS SYNCHRONOUS 4.2 ASYNCHRONOUS

"bit insertion"only whennecessary. As anexaming unwantedline patternsis to use ple, the sotucedataovera Tl line couldbe monitoredfor all 0's in the first 7 bits of a time slot.Wheneverthe0's occur,a I couldbe insertedinto the datastreamasthe eighthbit of thetime slot.At theotherendof theline the I following seven0's is removedautomatically.Eachinsertiondelaysthe sourcedataby the time of I bit, but otherwisethefull capacityof the channelis available. to the "zero-bitinsertion"algorithmusedin This procedureis directlyanalogous the High kvel DataLink Control(HDLC) protocol.In this protocola specificdata patterncalleda "flag" is usedto indicatetheendof a datablock.Thetransmittermust beprecludedfrom inadvertentlysendingtheflag aspartof thesourcedata.Themeans flagsis to inserta 0 followinga stringof five L's in theuser of preventinginadvertent data.Sincea flag containsa stringof six I's, zero-bitinsertionprecludesunintended Thereceivingnodeof anHDLC datalink removesany0 following flag transmissions. five I's. The receipt of six 1's, however,can only be part of a legitimateflag it neverproducesall 0's (ex(01I I 1I l0). If HDLC dataareinvertedfor transmission, ceptduringtheidle state). Althougha bit insertionalgorithmallowsfor moreefficientuseof a channelthan First,this dedicatedtiming bits, the insertionprocedurehasa numberof drawbacks. proce$$ causesthesourcedatato be delayedeverytime aninsertionis made.Hencea application(suchasvoice)requiressmoothingthe real-timetransmission continuous, Second,thebit insertionprocesscauses arival ratewith databuffersatthedestination. anycharacter$tructurein theuser'sdatato bebecomeunrelatedto thetime slotstruclink. Thusif userdataconsistof 8-bit tureof a timedivisionmultiplexedhansmission cannotbe maintainedwith (like PCM voicesamples), character boundaries characters link. respectto 8-bit time slotsin a Tl transmission Date Scrambling to randomizethedatapatterns systemsusedatascramblers Many digitaltransmission aresimilarto thoseused these data scramblers links. Although on their transmission is to preventthetransmispurpose of these scramblers for encryption,thefundamental datapattemsgenthe traffic. Repetitive patterns, notto encrypt sionof repetitivedata point interference from an erateline spectrathatcanbe significantlymoredegrading patterns. data produced by random of view than continuouslydistributedspectra Voicebanddatamodems,for example,areallowedto operateat higherpowerlevels to randomizethedatatraffic.Also, digitalradiosystemsare if theyincludescramblers requiredby theFCC to not ffansmitline spectra,whichessentiallymeansthatrepetitive datapatternsmu$tbe excluded. areusefulin transformingdatasequence$ Evenwhennot required,datascramblers Scramwith strongtiming components. with low transitiondensitiesinto sequences (T1 andobsolete T2) but wasusedon bling is not usedon lowerrateT-carrier$ystems the274-MbpsT4M coaxialhansmissionsystem[3] andcurrentopticalfiber systems. (with equalinput andoutputbit rates)do not preventlong sffings Datascramblers Theymerelyensurethatrelativelyshortrepetitionpatterns of 0's in anabsolutesense. If purely traffic with a minimumdensityof transitions. aretransformed to randomized

17O DrcrrAL TRANSMtsstoN AND MULT|pLExtNc randominput dataarescrambled,theoutputdataarealsopurelyrandomandtherefore havea certainstati$ticalprobabilityofproducingarbitrarilylong stringsof 0's. The probabilityof a randomoccurrence, however,is acceptable whencomparedto the probabilityof nonrandomsequences representing speechpausesor idle dataterminals.To determinewhichseeminglyrandomdatasequence producesall 0's attheoutput of a scrambler,applyall 0's to the corresponding descrambler. TheT4M coaxialtransmission systemuseda datascramblerasthebasicmeansof producingadequatetiming information.This systemcould toleratemuch longer stringsof 0's because the timing recoverycircuitsin the regenerative repeaters used phase-locked loopsthatmaintaintiming overrelativelylong periodsof time.In contrast,theoriginalTl systemsrecovered timingwith tunedcircuitsthatresonated at the desiredclockfrequency(1.544MHz) whenexcitedby a receivedpulse.Becausethe tunedcircuitshavelowereffectiveQ's thana phase-locked loop,theoscillationsdrift from theproperfrequencyanddie out morerapidly.Hencethe originalrl receivers couldnot tolerateaslong a stringof 0's ascouldT4M receivers.PhaseJocked loop clockrecoverycircuitsareusedon all latergeneration wirelineandfibertrmsmission $y$tems so muchlongerstringsof 0's canbe tolerated. Forced Blt Errorg A fifrh methodof maintainingsufficienttiming informationin the line signalsinvolveshavingthetransmission terminalat the sourceforcean occasionalbit errorin orderto intenupta long, transition-free datapattern.If the transition-freelrequences arelong enoughanduncommonenough,the intentionalbit errorsmight be lessfrequentthanrandomchannelerrorson thedigitaltransmission link. Thustheintentional errorsmay not representa significantdegradationover andabovethat that alreadyexists.Nevertheless, forcedbit errorsarenot generallyrecommended aspartof a linecodingprocedurebut arementionedin theinterestof completeness. As mentionedpreviously,NorttrAmericanchannelbanksforce a bit errorin the secondleastsignificantbit of an all-0's transmission codeto ensuresufficientsignal transitions.An importantaspectof thisprocedureis thatit is performedby the source wherethesignificanceof thebit erroris known.If thetransmission link itself inserted thebit errors,theeffectswouldnot be ascontrollable,particularlywhena varietyof traffic typesarebeingserviced. A moresubtleproblemwith forcedtransmission errorsarisesif the digital transmissionlink is usedfor automaticrepeatrequerrt (ARQ) datatransmission. An ARe datacommunication link is designedto provideerror-freetransmission, despiterandomchannelerrors,by insertingredundancy into thedatastreamandcheckingthereceived data for error-freereception.If errors are detected,a retransmissionis requested. when the erTorsarenot random,but forcedby the transmission link, the ARQ systemwill becomefrustratedif it everencounters thereshictedsequence, no matterhow unlikelyit is.* once again,if forcederrorsareused,theyshouldbe incorporatedinto thesourceaspartofthe $ourcecoderestrictionprocessandnot aga function of thetransmission link. *If

thetransmission link usesa scrambler, theunlikelysequence will not berepeatecl.

4.s UNEcoDrNc 171

4.3 LINE CODING In the precedingsectionvarioustechniquesfor establishingtiming informationare describedin generalterms.The choiceof anyparticulartechniqueis dependenton the mostcommonline codes the specificline codein use.This sectiondescribes usedfor digital transmissionandindicateswhat additionalsteps,if any,areneeded to maintainsynchronizationbetweenthe transmitterandreceiver.Someline-coding techniquesprovideadequatetiming informationthemselvesanddo not require requirepreviously.In additionto synchronization discussed anyofthe procedures ments,other considerationsfor choosinga line codeare the spectrumof the line noiseandin' codeandthe availablebandwidth(particularlyat low frequencies), and performance monitoring, acquisition times, terferencelevels,synchronization implementationcosts.

4.3.1 Level Encodlng Conceptually, the simplestform of line codingusesa differentsignallevel to encode Within a computersystemthe mostcommonform eachdiscretesymboltransmitted. "1" andnear0 V for a "0." Overa of codingis an on-off codeusinga 3-V levelfor a in termsof powerto encodebinarydata link, however,it is moreeff,rcient transmission with an equivalentdifferencein levelsbut symmetricallybalancedabout0 V. For example,theaveragepowerrequiredfor equallylikely +3- and0-V encodingsis 4.5 W thesameenor distance (assuming1 fl resistance). With +l.5- and-1.5-V encodings, (2.25 engineers W). Communications power requirements with half the is achieved the balanced codeasa as a unipolar code and code commonlyreferto theunbalanced balits corresponding of binary data and polar code.A representativesequence that the level in Figure 4.5. Notice levelencodingareshown ancedandunbalanced of eachsignalis maintainedfor the durationof a signalinterval.For this reason

n[,]n n I

Unipoler(unbalanced) rigneling

Figure 4.5 Unipolarandpolar(NRZ) line codes.

172

DIGITALTRANSMISSIoNAND MULTIPLEXING

Figure 4.6 Direct-current wanderof NRZ signal. the balanced(polar) encodingis alsoreferredto asa nonreturn-to-zero(NRZ) code.* As indicated in Figure 4.5, an NRZ signal contains no transitions for long strings of I's or 0's. Hence one of the proceduresdescribedpreviously to en$uretiming transitions must be employed if NRZ encoding is used on a synchronoustransmission link. An NRZ line code is a pulse transmissionsystem wherein the pulse (before filtering) lastsfor the duration ofa signaling interval 7. Hence the frequency spectrum(assuming random data) of an NRZ code is the sin(x)/x specfrum of Equation 4.1 ancl shown in Figure 4.2. As indicated, the frequency $pectrumis significantly nonzero at zero frequency (dc). Most wireline transmissionlinks, however, do not passdc signals by virtue of their being altemating current coupled with transformersor capacitorsto eliminate dc ground loops. Furthermore, some systemspurposely remove dc components from the signal to allow line powering ofrepeaters or to facilitate single-sidebard transmission. The elimination of the low-frequency components in the waveform causeslong strings of I's or 0's to decay gradually in amplitude. Hence a receiver not only would lose timing information during these strings but also would lose its amplitude referencefor optimally discriminating betweena "r" lever and a "0" level. The effect of low-frequency cutof'f; called dc wander, is shown in Figure 4.6 for a typical transmissionsequence.Notice that following the long string of I's, the output of the link is such that 1-to-0 effors are more likely than O-to-l errors. Similarly, long strings of0's increasethe likelihood ofa O-to-1error. This problem arisesnot only for long strings of I's or 0's, but wheneverthere is an imbalancein the number of I's and 0's. Hence periodic timing pulsesare not sufficient to remove dc wander. The existenceof low frequenciesin a random data signal is the basic reasonwhy modems are needed for data communications over the analog telephone network. (Analog telephonecircuits do not pass direct current.) It is also the reasonthat NRZ coding is not often used for long-distancetransmission.Direct-current wander is not unique to data transmission systems.It is a phenomenonthat must be reconciled in television receivers,radar receivers,or radiation detectors. One technique of offsetting dc wander is referred to as dc or baseline restoration [a]. As illustrated in Figure 4.7, dc restoration involves passing received pulses through a capacitor,detectingthem, and then removing the chargeon the capacitorbefore the next pulse arrives. charge on the capacitoris removed by driving the voltage *Some

communications theorists ref'et to a balanced two-level code as a "bipolar code." The North American telephone industry, however, uses the term bipolar torefet to a t}reeJcvel code described in thc next section,

\

4.s L|NEcoDrNc 173

+v

t

Switch on

Figure4.7 Direct-current restoration for unipolarpulses. to a specificthreshold(0 V in Figure4.7)andthenremovingthedrivingvoltagewhen thethresholdis reached.Sinceall chargeon thecapacitoris removedaftereachpulse, thebaselineor decisionreference levelis constantat thebeginningofa signalinterval. An obviousdisadvantage technique of this is thatthesignalinputmusthavezeroarnplitudeor be disabledduringtheresettime. A generallymoreusefultechniquefor overcomingbaselinewanderis to usedecisionfeedback,alsocalledquantizedfeedbackequalization[5*7]. In contrastto dcrestoration, which drives the capacitor voltage to a constant,predeterminedlevel, quantizedfeedbackcompensate$ for dc wanderby locally generatingthe unreceived low-passresponse andaddingit to ttrereceivedsignal.To accomplish this,theoriginal datastreamis reconstructed. As shownin Figure4.8,thereconskucted datastreamis passedthrougha low-passfilter that generatesa pulseequalto the tail or droopcharacteristicof the channel.Thefeedbacksignaladdsto thereceivedsignalto eliminate the droop(intersymbolinterference). Using a frequency-domain analysis,the feedbackresponse is complimentary to the channelresponse. Quantizedfeedbackis used in ISDN basicrateline interfaces[8]. 4.3.2 Bipolar Coding The dc restorationtechniquesmentionedin the precedingsectionsimplify pulsedetectionby creatinga low-passpulseresponse in thereceiver.Therearenumerousline

Figure 4.8 Decisionfeedbackequalization.

174

DGTTAL THANSMtsstoN ANDMULTIpLEXtNc

Figure4.9 Bipolar(AMI)coding, codesthatarespecificallydesigned to notcontaindc energyandtherebybeunaffected by dc removal.Bipolarcodingsolvesthedc wanderproblemby usingthreeleversto encodebinarydata.specifically,a logic 0 is encodedwith zerovoltagewhile a logic I is alternatelyencodedwith positiveandnegativevoltages.Hencetheaveragevoltagelevel is maintainedat zero-to eliminatedc components in the signalspectrum. sincebipolarcodingusesalternatepolaritypulsesfor encodinglogic 1's,it is alsoreferredto asalternatemark inversion(AMI).Bipolarcodingis thebasicline-codingprocedureusedby Tl linesin thetelephone network.Ratherthanusingfull-periodpulses,however,TI linesusea 50zadutycycle pulseto encodeeachlogic 1.Return-to-zero (RZ) pulses(Figure4.9)wereselected to minimize intersymbolinterferenceand simplify timing recoveryin the regenerative repeaters of a Tl line [9]. Thepowerspectrumof a bipolarcodeis obtainedfrom I l0] AS

-

I cosror s(or)='Plc(co)|2 | 2(2p l)coswT+ (2p- r)z

(4.3)

wherep= probability of a 1 G(rrl)= spectrum of anindividualpulse fr\ sin(roTl4) j

c(oro)=|* |

[rJ

ar/4

forSOVIdurycyclepulses

Equation4.3 is plottedin Figure4.10 for variousvaluesof p. For pureryrandom data,p = j. Recall,however,thatsourcecodingfor p255PCM codecsproducesmore I's than0's in theinterestof establishing a shongclock signal.Hencetheappropriate valueof p for a Tl voiceline is normallysomewhatlargerthan0.5 anddependson theamplitudeof the voicesignal.Low-levelsignalsthatremainin thefirst encoding segmentproducea valueof p approximatelyequalto 0.65.on the otherhand,fullscalesinewavesproducea valuefor p thatis somewhatbelow0.5 sincemostof the samplesoccurnearmaximumamplitude. Becausea bipolarcodeusesaltematingpolaritiesfor encodingI's, stringsof l 's havestrongtimingcomponent$. However,a stringof 0's containsno timing information andthereforemustbe precludedby the source.The specifications for Tl line re*A

mark is a term arising from telegraphy to refer to the active, or l, state of a level encoded transmission line,

\

4.3 L|NECoDING 175

oStT Frgquency

Figure 4.10 Specnaldensityof bipolarcoding. peatersstatethat the repeaterswill maintain timing as long as no sfing of greaterthan f,rfteen0's is allowed to occur I1 11.A string of fifteen 0's can only occur if a 0 framing bit falls between a 10000000code in time slot 24 and a 0fi)00001 code in time slot l.

Coda SpaceRedundancY butonlytwoof thelevelsduring bipolarcodingu$esa temarycodespace In essence, any particular signal interval. Hence bipolar coding eliminates dc wander with an inefficient and redundantuse of the code space.The redundancyin the waveform also provides other benefits. The most important additional benefit is the opportunity to monitor the quality of the line with no knowledge of the nature of the traffic being transmitted.Since pulseson the line are supposedto alternatein polarity, the detection of two successivepulsesof one polarity implies an error. This error condition is known as a bipolar violation. No single error can occur without a bipolar violation also occuning. Hence the bipolar code inherently provides a form of line codeparity. The terminals of Tl lines are designed to monitor the frequency of occurrence of bipolar violations, and if the frequency of occurrenceexceedssomethreshold,an alarm is set' In T-carrier systems,bipolar violations are u$edmerely to detectchannelerrors.By adding $omerather sophisticateddetectioncircuitry, the sameredundancycan be used for conecting errors in addition to detectingthem. Whenever a bipolar violation is detected, an effor has occurred in one of the bits between and including the pulses indicating the violation. Either a pulse should be a 0 or an intervening 0 should have been a pulse of the opposite polarity. By examining the actual sample values more closely, a decision can be made as to where the error was most likely to have occurred.The bit with a samplevalue closestto its decision thresholdis the most likely bit in error. This technique belongs to a general class of decision algorithms for redundant signals called maximum likelihood or Viterbi decoders[12]. Notice that this method of error coffection requires storageof pulse amplitudes.If decision values only are stored,error correction cannot be achieved(only error detection)' An additional application of the unused code spacein bipolar coding is to purposely insert bipolar violations to signify special sifuationssuch as time division mul-

176

DIGITALTRANSMISSIoNANDMULTIPLEXING

tiplex framing marks, alarm conditions, or specialcodesto increasethe timing content of the line signals. Since bipolar violations are not normally part of the source data, these special situations are easily recognized. of course, the ability to monitor the quality of the line is compromisedwhen bipolar violations occur for reasonsother than channel errors.

4.3.3 BlnaryMZero Substltution A majorlimitationof bipolar(AMI) codingis its dependence ona minimumdensity of 1's in the sourcecodeto maintaintiming at theregenerative repeaters. Evenwhen stringsof0's greaterthan 14 areprecludedby the source,a low densityofpulseson theline increases timingjitter andthereforeproduceshighererrorrates.BinaryN-zero substitution(BNZS) [ l3l augmentsa basicbipolarcodeby replacingall skingsof N 0's with a specialNJengthcodecontainingseveralpulsesthatpurposelyproducebipolarviolations.Thusthedensityof pulsesis increased while theoriginaldataareobtainedby recognizingthebipolarviolationcodesandreplacingthemat thereceiving terminalwith N 0's. As an example,a three-zerosubstirutionalgorithm(B3zs) is clescribed. This particular substitutionalgorithmis specifiedfor the standardDS-3 signalinterfacein NorthAmerica[14].It wasalsousedin theLD-4 coaxialmansmission systemin canada[15]. In theB3ZSformat,eachstringof three0's in thesourcedatais encodedwith either 00v or BOv. A 00v line codeconsisrsof 2-bit intervalswirh no pulse(00) followed by a pulserepresenting a bipolarviolation(v). A BOv line codeconsistsof a single pulsein keepingwith thebipolaralternation(B), followedby no pulse(0), rurdending with a pulsewith a violation(V). With eithersubstitution, thebipolarviolationoccurs ' in the lastbit positionof thethree0's replacedby the specialcode.Thustheposition of thesubstitutionis easilyidentified. Thedecisionto substitutewith 00V or BOv is madesothatthenumberof B pulses (unviolatedpulses)betweenviolations(v) is odd.Henceif an oddnumberof I's has beentransmittedsincethe lastsubstitution,00V is chosento replacethree0's. If the interveningnumberof I's is even,BOv is chosen.In this mannerall purposefulviolationscontainan odd numberof interveningbipolarpulses.Also, bipolarviolations alternatein polaritysothatdc wanderis prevented. An evennumberof bipolarpulses betweenviolationsoccursonly asresultof a channelerror.Furthermore, everypurposefulviolationis immediatelyprecededby a 0. Henceconsiderable systematicredundancyremainsin the line codeto facilitateperformancemonitoring.Table4.1 summarizes the substitutionalgorithm. Example4.1. Determinethe B3zs line code for the following data sequence: 1010001 10000000010001 . use + to indicatea positivepulse,- to indicatea negarive pulse,and0 to indicateno pulse.

4.3 L|NECODTNG 177

TABLE4.1 B3Z$ SubstitutlonRulee Numberot BipolarPulses(1's) SinceLastSubstitution Polarityof Pulse Preceding 0000+

+0+ -0-

Sotution, There are two possible sequencesdependingon whether an odd or even number of pulseshas been transmitted following the previous violation:

Substitutions \ l0l CaseI (odd): +0-

000 00-

Case2 (even): +0*

+0+

Violations

ll

+-

\

000 +0+

000 -0-

001 000 00+ 00+

-0-

+0+

00-

1

00-

JJ

Example4.1 indicatesthattheprocessof breakingup stringsof 0's by substituting the minimumdensityof pulsesin the line with bipolar violationsgreatlyincreases is 337o whiletheaveragedensityis just over607o' code.In fact,theminimumdensity Noticethat HencetheB3ZSformatprovidesa continuouslystrongtimingcomponent. no restricwith information continuoustiming all BNZScodingalgorithmsguarantee in a completely tions on sourcedata.HenceBNZS codingsupportsany application manner, transparent AnotherBNZScodingalgorithmis the862,5algorithmusedon obsoleteT2 transmissionlines [16]. The 8625 algorithmis definedin Table4.2. This algorithmproducesbipolar violations in the secondand fifth bit positionsof the substituted sequence. ITTJrecommends anotherBNZS codingformatreferredto ashigh-densitybipolar in theEl primarydigitalsignal,HDB codingre(HDB) coding[l7]. As implemented containinga bipolarviolationin the lastbit placesstringsof four 0's with sequences position.Sincethis codingformatprecludesstringsof 0's greaterthanthree,it is referredto asHDB3 coding.Theencodingalgorithmis basicallythe sameastheB3ZS thebasicalgorithm.Noticethatsubstialgorithmdescribedearlier.Table4.3presents substitutions tutionsproduceviolationsonly in thefourthbit position,andsuccessive produceviolationswith altematingpolarities'

178

DIGITAL TRANSMISSIoN ANDMULTIPLEXING

TABLE 4.2 B€ZS Subetltution Rulee

Polarity ol Pulse lmmediately Preceding Six0'sto be Substituted

Substitution 0-+0+0+-0-+

+

Hxample: 1 0 0 0 0 0 0 1 0 1 1 0 0 0 0 0 00 0 0 0 0 0 0 0 0 1 + * ( 0 - + U + - ) + 0 - + ( o * - o - + ) ( u + - U - ++)) 0 0 0 - + (0+ - 0 - + ) - 0 (0- + 0 + - ) (0- + 0 T ) 0 0 0 +

A fundamental feature of end-to-end digital connectivity as provided by IsDN is 64-kbps transparentchannelsreferred to as clear-channelcapability (ccc) [18]. Two aspectsof a bipolar/AMl line code as used on Tl lines preclude ccc; robbed signaling in the least signilicant bit of every sixth frame and the need to avoid all-O's codewords on the channel. Bit robbing for signaling is avoided with commonchannel signaling (also an inherent requirement for ISDN deployment). Two meansof augmentingTl lines Io allow transparentchannelshave beendeveloped. The first method is zero-bytetime slot interchange(ZBTSI) describedin reference il 91.ZBTSI was used for upgrading some Tl facilities for ccc but is not a godd long-term solution. The secondmethod, which is most desirablein the long run, involves the use of a BSzs line code for Tl lines. As such, the BBZS line code becamethe new line code standardfor Tl lines with the introduction of D5 channel banks. (D5 channel banks can also operate with bipolar/AMl line codes for backward compatibility.) As indicaredin Table 4.4, the BSZS algorithm is similar to the B625 algorithm in that each individual substitution is dc balanced.Notice that there are two bipolar violations, in positions 4 and 7, in every substitution.The purposeful introduction of bipolar violations requires replacementof any equipment that monitors all violations for performancemonitoring.

TABLE4.3 HDB3SubstltutlonHutes Numberof BipolarPulses(1's) $inceLastSub$titution Polarity of Preceding Pulse +

odd

Even

000000+

+00+ -{0-

4.3 L|NECODTNG 179

TABLE4.4 BBZSSubetltutionRules Substitution

Pulse Polarity of Preceding

000-+0t.000+-G-+

4.3.4 Pair SelectedTernary in theprecedingsectionareexamplesof TheBNZS substitutionalgorithmsdescribed thetimingcontentof a binarysigspace to increase selectingcodesin theternarycode example. is another nal. PairselectedternaryGST) I20l The PST codingprocessbeginsby pairing the binary input datato produceseinto two ternarydigarethentranslated Thesecodewords quences of Z-bitcodewords. but only four ternary codes its for transmission.Since there are nine two-digit in selectingthe available two-digit binary codes,there is considerableflexibility Table 4.5. This paris shown in codes.Themostusefulof thepossiblecodingformats prevents dc wander but also a strongtimingcomponent ticularformatnotonly ensures pulses. The positive and negative by switchingmodesto maintaina balancebetween time the pulse is At that transmitted. from onecolumnuntil a single codesareselected single column until another encoderswitchesmodesandselectscodesfrom theother pulse(of theoppositepolarity)is transmitted. Exampte4.2. Encodethe following binary data streaminto a PST line code: 100. 0t 001l 10101 Soluti.on. Therearetwo possiblesolutionsdependingon whethertheencoderis in thepositiveor negativemodeat thebeginningof thesequence: 0l CaseI (+ mode): 0+ Case2 (- mode): -+

00 -+ -+

11 ++-

l0 -0 +0

ll ++-

10 +0 -0

00 -+ -+

Onepotentialdrawbackof thePSTcodingalgorithmis thatthebinarydatastream must be framedinto pairs.Hencea PST decodermustrecognizeandmaintainpair TABLE 4.5 Pair Selected Ternaly Encoding

BinaryInput 00 01 10 11

+ Mode

- Mode

*+ 0+ +0

0-0 +-

180

D|G|TAL THANSMISSIoN ANDMULTIPLEXING

boundaries.Recognition of the boundaries is not difficult if random data are being transmitted since a pairwise misframe eventually producesunallowed codes(00, + +, --). Furthermore, time division multiplex formats typically provide character and pairwise framing automatically. The power spechum of a PST line code with equal probabilities for l's and 0's is obtained from reference [20] and ploted in Figure 4.11. Also shown is the 8625 power spectrum [16] and the conventional bipolar power specffum. An important point to notice in Figure 4.l l is that bipolar coding and its extensions require equal bandwidths.Their only significant difference is that B625 and PST have higher energy levels as a result of greater pulse densities.The higher energy levels have the undesirableeffect of increasing crosstalk interference in multipair cables. However, the degradation from the increasedcrosstalk is somewhat offset by improved accuracyofthe recovered sample clock (assumingall signals in the cable are using the sameline code).

4.3.5 TernaryGoding SincebipolarandPSTcodingusea temarycodespaceto transmitbinarydata,they do not achieveashigh aninformationrateasis possiblewith moreefficientuseof the codespace.For example,an eight-element ternarycodeis capableofrepresenting 38= 656I differentcodes.In contrast,I bitsof binarydaraproduceonly 28= 256differentcodes.Thepreviouslydescribedline codesdo not takeadvantage ofthe higher informationcontentof ternarycodes;they selectcodesfor their timing contentand spectralproperties. one temaryencodingprocedureinvolvesmappingsuccessive groupsof 4 bits into threeternary(483T) digits.sincebinarywordsof 4 bitsrequireonly t6 of the27 possiblethree-digitternarycodewords, considerable flexibility existsin selectingtheternarycodes.Table4.6presents onepossibleencodingprocedure. Ternarywordsin the middlecolumnarebalancedin their dc content.Codewordsfrom the first andthird

0.5tT Frequ€ncy Figure4.ll

ltT

Specftum ofbipolar, B3ZS, and PST line codes for equally likely I's and 0's,

4,3 LINECODING 181

TABLE4.6 EncodlngTablefor 4B3TLine Code Disparity) TernaryWord(Accumulated BinaryWord 0000 0001 0010 0011 0100 0 10 1 0110 0 11 1 1000 1001 1010 1 0 11 11 0 0 11 0 1 1110 1111

--0 -00-- - + - + f - -

-00 0-0 00-

+++ ++0 +0+ 0++ ++* +-+ -++ +00 0+0 00+

0+0-+ +0-0+ +-0 -+0

columnsareselectedalternatelyto maintaindc balance.If morepositivepulsesthan Whenthe disparitybenegativepulseshavebeentransmitted,column I is selected. column3 is chosen.Noticethattheall-O's tweenpositiveandnegativepulseschanges, codewordis not used.Hencea $trongtiming contentis maintained'Becauseof the is sacrihigherinformationefficiency,however,the ability to monitorperformance used on the 4B3T codingis ficed,andframingis requiredon three-digitboundaries. Tprovides by ITT Telecommunications [21].This system TI48 spanline developed for two DSI signals(48 channels)usinga bandwidththatis only carriertransmission of discussion 50Vogrcatmthana Tl bandwidth(carrying24 channels)'A generalized is containedin reference[22]. 4B3T codingandotherternarycodingtechniques 4.3.6 DigitalBiphase BNZS andPSTuseextraencodinglevelsfor flexiBipolarcodingandits extensions no dcwander,andperfeaturessuchastimingtransitions, bility in achievingdesirable thecodespaceand by increasing formancemonitorability.Thesefeaturesareobtained sofar, discussed (The null of all codes first spectral norby increasingthebandwidth. rate 1/2.) includinganNRZ code,is locatedat the signaling Many varietiesof line codesachievestrongtiming andno dc wanderby increasing the bandwidthof the signalwhile usingonly two levelsfor binarydata'One of the mostcommonof thesecodesprovidingboth a strongtiming cornponentandno dc "diphase"or a "Manchester" wanderis the digital biphasecode,alsoreferredto as code.

182

DtctrALTRANSMtsstoN ANDMULTtpLEXtNc

'FH IJ

'tfr

Figure 4.12 Digital biphase(Manchester) line code. A digital biphasecode usesone cycle of a squarewave at a particular phaseto encode a I and one cycle of an opposite phaseto encode a 0. An example of a digital biphasecoding sequenceis shown in Figure 4.12. Notice that a transition exists at the centerof every signaling interval. Hence strong timing componentsare presentin the spectrum.Furthermore,logic 0 signalsand logic I signalsboth contain equal amounts ofpositive and negative polarities. Thus dc wander is nonexistent.A digital biphase code,however, doesnot contain redundancyfor performancemonitoring. Ifin-service performance monitoring is desired, either parity bits must be inserted into the data stream or pulse quality must be monitored. (A later section of this chapter discusses performance monitoring in more detail.) The frequency spectrum of a digital biphase signal is derived in Appendix c and plotted in Figure 4.13, where it can be compared to the speckum of an NRZ signal. Notice that a digital biphasesignal hasits first spectralnill atLlT.Hence the exrrariming transitionsand elimination of dc wander come at the expenseof a higherfrequency signal. In comparison to three-level bipolar codes,however, the digital biphasecode has a lower error rate for equal signal-to-noiseratios (seeAppendix C). Examination of the frequency specffain Figure 4.13 shows that the diphasespectrum is similar to an NRZ spectrumbut translatedso it is centeredabout 1/I instead of direct current. Hence digital biphase actually represents digital modulation of a squarewave carier with one cycle per signal interval. Logic I 's causethe squarewave to be multiplied by +1 while logic 0's produce multiplication by -1. Diphase is pri-

\ \ \ \ 1/T

ztr

3tT

4tT

Ftgure4.13 Powerspectraldensityof digitalbiphase.

4.3 LINECODING 183 marily usedon shorterlinks where terminal costs are more significant than bandwidth "Ethemet" IEEE 802.3 local area data network uses digital biphase utilization. The (Manchester)coding.

4.3.7 DlfferentialEncoding One limitation of NRZ and digital biphasesignals,as presentedup to this point, i$ that the signal for a 1 is exactly the negative of a signal for a 0' On many transmissionmedia, it may be impossible to determine an absolutepolarity or an absolutephasereference.Hence the decodermay decodeall I's as 0's and vice versa.A common remedy for this ambiguity is to use differential encoding that encodesa I as a changeof state and encodesa 0 asno changein state.In this mannerno absolutereferenceis necessary to decodethe signal. The decodermerely detectsthe stateof each signal interval and comparesit to the stateof the previous interval. If a changeoccured, a I is decoded. Otherwise. a 0 is determined. Differential encoding and decoding do not change the encoded spectrum of purely random data (equally likely and uncorrelatedI's and 0's) but do double the error rate. If the detector makes ar effor in estimating the state of one interval, it also makes an error in the next interval. An example of a differentially encodedNRZ code and a differentially encodeddiphasesignal is shown in Figure 4.14. All signalsof differentially encodeddiphaseretain a transition at the middle of an interval, but only the 0's have a transition at the beginning of an interval'

4.3.8 CodedMarkInversion A variety of line codeshave evolved similar to the digital biphasecode describedpreviously. One of theseis referred to as codedmark inversion (CMI) in ITU recommendations [l7l. CMI encodes I's (marks) as an NRZ level opposite to the level of the

(d)

o

l

0

Figure 4.14 Differentially encoded NRZ and diphase signals: (a) differentially encoded NRZ; (b) differentially encoded diphase.

184

DIGITALTRANSMISSIONAND MULTIPLEXING

l o l ' l ' l 0 l oI ' l o I

J l

I Figure 4.I5 Codedmarkinversion.

previousoneand0's asa half-cyclesquarewaveof onepar"ticular phase.Figure4.15 showsa sampleencodingfor cMI. Thereis no dc energyin the signalandan abundanceof signaltransitionsexist asin diphase.Furthermore,thereis no ambiguity betweenI 's and0's-Eliminationof theambiguityactuallyleadsto a majordrawback * of cMI coding;Its errorpedormance is 3 dB worsethandiphase. cMI is thespecified interfacecodefor the fourrh-level(Bl) ITU multiplexsignalat 139.264Mbps.cMI is alsousedfor electical interfacesof soNET sTS-3csignalsdescribedin chapterg. 4.3.9 MultllevelSignating In the line codesdiscussed so far, two-lever(binary)signalinghasbeenassumed. In applications wherethebandwidthis limitedbut higherdataratesaredesired,thenumberof levelscanbeincreased whilemaintainingthesamesignalingrate.Thedatarate R achievedby a multilevelsystemis givenas

R- ros,(r) F)

(4.4)

where.L= numberof levelsthat canbe freely chosenduring eachinterval I= signalinginterval The signalingrate llT is often referredto asthe symbolrateand is measuredin bauds.within thedatacommunications industryit is commonpracticeto use..baud" as being synonymouswith bit rate.strictly speaking,however,the bit rate is only equalto thebaudrateif binarysignaling(l bit persignalinterval)is used.Figure4. I 6 showsanexampleof aneight-leveltransmission formatthatachieves3 bitspersignal interval(i.e.,3 bitsper baud). Multilevel transmissionsystemsachievegreaterdatarateswithin a given bandwidthbut requiremuchgreatersignal-to-noise ratiosfor a givenerrorrate.Oneaspect of wirelinetransmission thatfavorsmultilevelline codingis thelower baudratefor a givendatarate,which in turn reducesthe crosstalk.Hence,in crosstalk-limited svs*The

enor performanceof CMI is 3 dB worse than diphasewhen bit-by-bit detection is used.The inetficiencyarisesbecause for one-halfof an intewal a I lookslike a 0. BecauseCMI hasredundancv. someof the inefficiency canbe recoveredwith maximumlikelihood ffiterti) tletection.

4.3 LINE coDtNc 185

wittr 3 bits per signalinterva'' Figure 4.16 Multileveltransmission tems the signal-to-noiseratio penalty of a multilevel line code is not as significant. The TlG T-carrier systemdevelopedby AT&T [23] usesa four-level line code at the TIC baud rate (3.152 MHz) to double the capacity of a TlC $ystem(from 48 channelsto

e6).

Another example of multilevel transmissionof pafiicular significance is the ISDN basic rate digital subscriberline (DSL), which uses fbur-level transmission at a $i8naling rate of 80 kilobauds to achieve 160 kbps. The primary factors that led to selecting a multilevel line code in this application are ( I ) near-endcrosstalkthat cannot be eliminated by pair isolation* as in Tl systemsand (2) high levels of intersymbol interferencecausedby bridged tap reflections. Both of thesefactors are easierto control when lower frequency signalsare used t241.Additional aspectsof this application are describedin the ISDN section of Chapter 11.

$ignaling 4.3.10 Partial-Response Conventional bandlimiting filters of a digital transmissionsystem are designedto restrict the signal bandwidth as much as possible without spreadingindividual symbols so they interfere with samplevalues of adjacentintervals. One classof signaling techparniques,variously referredto as duobinary [25], correlative level encoding [26]' or of intersymbol amount prescribed a introduces tiai-responsesignaling [27], purposely interferencethat is accountedfor in ttre detection circuitry ofthe receivers.By overfiltering the encoded signal, the bandwidth is reduced for a given signaling rate, but the overlapping pulses produce multiple levels that complicate the detection process and increasethe signal power requirementsfor a given error rate' Figure 4.17 shows the pulse responseof a typical partial-responsesystem. If the channel is excited by a pulse of duration T, channel filters (defined in Appendix c) limit the spectrumsuch that the main part of the pulse extendsacrossthree signal intervals and contributes equally to two sample times. The reason for the term partial responseis now apparent:The output only respondsto one-half the amplitude of the input. *Pair

isolation involves separating go and return pairs into separate cables'

186

DIGITALTBANSMISSIONAND MULTIPLEXING

t

\\N6xt output puhe

Figure 4.17 Outputpulseof partial-response channel.

If theinputpulseof Figure4.17is followedbyanotherpulseof thesameamplirude, theoutputwill reachfull amplitudeby virtueof theoverlapbetweenpulse$.However, if thenextinputpulsehasa negativepolarity,theoverlapproduceszeroamplitudeat the sampletime.Thus,asshownin Figure4.18,a partial-response systemwith twolevelinputs(+1,-I) producesanourputwith threelevels(+l , 0, -l). In similarfashion, Figure4.19showsa systemwith four inputlevels( +3, +1, -1, -3) andseven outputlevels(+3,+2,+1,0, -1, -2, -3). Detectionof a partialresponse signal(PRS)is complicated by theadditionallevels producedby the channelfilters andthe fact that samplevaluesaredependenton two adjacentpulseamplitudes.Onemethodof detectinga PRSinvolvessubtractingthe overlapof a previouslydetectedpulsefrom the incomingsignalto generatea differencesignalrepresenting thechannelresponse to thenewpulse.only theoverlap(intersymbolinterference) at the sampletimes needsto be subtracted. The difference samplesare nominallyequalto one-halfthe amplitudeof the unknowninput pulse. This techniquedoublesthe errorratein the samemannerthatdifferentiallvencoded errorratesaredoubled. Anothermethodof detectingpartial-response systemsinvolvesa techniquecalled precodingat thesource.hecodingtransformstheinputdatain sucha mannerthatthe outputlevel at thedetectordirectlyindicatestheoriginaldatawithoutcomparisonto theprevioussamplevalue.In a binarysystem,for example,a I is encodedwith a pulse of the samepolarityasthepreviouspulse.Hencelogic I 's showup at thedetectoras eithera +l or a -l sample.Transmission of a 0 is encodedwith a pulseof opposite polarity to the previouspulse.Hencelogic 0's alwaysshowup at the detectoras a zero-levelsignal.similar precodingtechniques exisrfor multilevelsystems[zg].

Figure 4.I8 TfueeJevelpartial-response inputsandourpurs.

4.s LINE coDtNe 187

inputsandoutputs' partial-response Figure 4.19 Seven*level The partial-responsetechniquejust describedis actually a special caseof a more generalclassof signaling techniquesreferredto as correlative level encoding [26]' To describethe properties of generalizedpartial-responseor correlative level systems,it is convenient to introduce a delay operator D to denote delay equal to one signal interval Z. Physically, D can be thought of as a delay line of length ?- Two units of delay is implemented with two delay lines in seriesand is denotedas D2' Using this notation, the partial-re$ponse$y$temdescribedabove is referred to as l+D PRS: The output representsthe superpositionof the input with a delayedversion of the sameinput. Other forms of overlap are possible. These systemsdo not necessarily produce the overlapping pulses by over{iltering an input' An alternative approach is to overlap and add the pulsesdirectly in the encodingprocess(hencethe term correlative level encoding). An interesting special caseof correlative level encoding is the I - D system shown in Figure 4.20 and extendedin Figure 4.21 to show the effect of overlapping Pulses' The I - D encoderusesa single cycle of a squarewave acrosstwo signal intervals -+) produce dc to encodeeachbit. Since neither ofthe two individual signals ( +- or positive and negative the energy,the encodedsignal has no dc wander.Furthermore, coding' of bipolar leveis of the composite signal alternate in a manner reminiscent phase the same In fact, if differential encoding is used (i.e., if a 0 is encodedwith widrh of one signal interurl I

Overlap into sdiEcent interval

#-r

Logic I

Logic 0

Figure 4.20 Individual signalelementsfor I - D correlativelevelcoding'

188

DtcITALTRANsMtsstoNANDMULTtpLEXtNG

Figure4.21 Representative waveform of I - D correlative levelencodins. asthepreviousintervalanda I astheoppositephase),thisform of encodingis identical to bipolarcoding(assuming theNRZ levelsarereplacedby s}vo RZ pulses). Thus I - D correlativelevel encodingis usedto shapethe bandwidthratherthan to limit it. The spectraof unfiltered I + D, I * D, and | - Dz signalsareobtained from referencel2Tlandplottedin Figure4.zz.TheTlD T-carriersystemof AT&T usesprecodedI + D level encoding(alsocalledduobinary). Thespectrumof the 1 * D2signalis particularlyinteresting. Ithas no dc componenr andan upperlimit equalto ll2T: thesameupperlimit as a maximallyfilteredNRZ codewith no intersymbolinterference. The I - D2partial-response systemshavebeen

/

l/ \

\t Frequency {Hzl Figure 4.22 spectraof unfilteredI + D, I - D, ard I - Dz correlativeencodedsisnals.

189 PERFoHMANcE 4.4 ERHoR usedby GTE in a modifiedT-carriersystemproviding48 voicechannels[29] andfor frequenoveranalogmicrowaveradiosusingvery low baseband digitaltransmission spectrumexi$ts)[30]' Theobciesfor pilotsandservicechannels(whereno messaBe systemof AT&T also used| - Dz codingwith soletelA-RDs (data-under-voice) four-levelinputsto get 1.544Mbpsinto 500kHz of bandwidth[31]' Notice that the I - Dz spectrumis essentiallythe productof the I + D andthe I a I D enby coxcatenating D spectra.Indeed,aI - D2 systemcanbe implemented * = level conelative | Dz'Thus (1 + D) D)0 response: + D channel a I coderwith and correlations representing for simultaneously useful are very polynomials system necessary than more levels use level systems correlative Since all shaping. spectrum for encodingthe data,they are inferior to uncorrelatedor full-responsesystemsin termsof error perfofinance' systemsidentifyingvariousline codesis proA summaryof digital transmission in videdin Table4.7.Otherline codesusedspecificallyin fiber systemsaredescribed Chapter8.

4.4 ERRORPERFORMANCE thetiming andpowerspecffumreTheprecedingsectionsof this chapteremphasized fundamentalconsiderationin Another codes. transmission o1 various quirements pre$ence of noise.Excepton in the performance rate the error line code is choosinga requireperformance error insignificant, may be noise lines, where relativelyshofi rate is specielror minimum If a celtain cost significantly. system mentrtcanimpact lower signal-to-noise rate at error the desired providing schemes fied, thosecoding to be spacedfartherapart,therebyreducinginstalrepeaters ratiosallowregenerative

TABLE4.7 DigitalTransmiselonSystems Bit Rate CountrY or DesignationAdministration (Mbps)

Line Code

T1 E1 T1C T148 9148A

AT&T ITU-T ATAT ITT GTE

1.544 2.048 3,152 2.37,ternary 3.152

T.ID

AT&T

3.152

T1G T2

AT&T AT&T

6.443 6,312

AMYBEZS HDBs(B4ZS) Bipolar 4B3T 1 -D2, duobinary 1+D, duobinary Four-level B6ZS

LD.4 T4M

Canada AT&T

274.176 274.176

B3ZS Polar

Media

Repeater SPacing

Twistedpair Twistedpair Twistedpair Twistedpair Twistedpair

6000ft 2000m 6000ft 6000ft 6000fr

Twistedoair

6000ft

Twistedpair Low-capacitance twistedpair Coax Coaxbinary(NRZ)

6000ft 4800ft 1900m 5700ft

190

DtctrAlTRANSMlsstoN AND MuLTtpLEXtNc

lationandmaintenance.+ Repeater spacingis animportanteconomicfactorin wireline transmission, in opticalfiber transmission, andon point-to-pointradiolinks. Theerrorrateresultsandcomparisons presented in this sectionarebasedon white Gaussiannoise.This is the mostcoilrmonform of noiseandthe bestunderstood. In someapplications, aswithin theolderelectromechanical telephonenetwork,impulse noisemaybemoreprevalent.Thustheseanalyses do notprovidea completeerrorrate analysisfor someapplications. Theintentof thissectionis to presentrelativeerrorperformancecomparisons of variousline-codingtechniques. For this purposea white noiseanalysisis mostappropriate. If impulsesarelargeenough,theycauseenorsindependent of thecoding$chemein use. For themostpart,thefollowingsectionspresentonly theresultsof theerroranalysesin theform of graphsof errorrateasa functionof signal-to-noise ratios.Appendix C derivesthebasicequationsusedto producetheresults. 4.4.1 SignalDetection Invariably,the detectioncircuiuryof a digital receiverprocesses incoming signal waveformsto mea$ureeachpossiblediscretesignal.In mostcasesthe mea$ures are nothingmorethansamplesof a filteredreceivesignal.Dependingon thesignalshape andthelevelof performance desired,thereceiverusesmoresophisticated piocessing. In anycase,theendmeasurement of a binarysignalnominallyproducesonevotmge level for a 0 andanothervoltagelevel for a I. A decisionof which signalwastransmitted is madeby comparingthemeasurement (at the appropriatetimeJto a threshold locatedhalfwaybetweenthesenominalvoltages.Naturally,theerrorprobabilitydependson thenominatdistancebetweenthevoltagesandtheamountof fluctuationin themeasurements causedby noise. Sincesignalmea$urement$ arenormallylinearin nature,theerrordistancebetween I ' s and0's is proportionalto thereceivedsignalvoltage.Equivalently,theamountof noisepowerrequiredto produceanerroris a directfunctionof thesignal-to-noise ratio. Theoptimumdetectorfor a particularsignalsetmaximieesthesignal-to-noise ratio at theinput to thedecisioncircuit. 4.4.2 Nolee Power white Gaussian noiseis characterized ashavinga uniformfrequencyspectrumacross anarbitrarilylargebandwidthandanamplitudedistributionthatvariesaccordingto a normal(Gaussian) probabilitydistribution.A parameterN,l conventionally represents the powerspecfraldensityof white noiseandis the amountof powermeasured in a bandwidthof I Hz. Hencetherms powerof whitenoisecominf out of a filter with a bandwidthBW is (Nr)(BW).t 'On

Tl lines the rcpeater spacing was predetermined by the locations where loading coils needed to be .removed. Thus the error rate performance essentially determined the maximum oata iate. TThe power spcctral density of white noise is also specifred * oensity. as a fro r* " ir"**ia*a rfr-t4 praclical matter, there is no difference in the specificarions,'since a rcal filter has the mathematical equivalent of identical positive- and negative-frequencybands.Thus the measuredpower corDng through a filter with a one-sided (positive-frequency) bandwidth BW is No BW in either case.

4.4 ERRORPEHFORMANCE I91

To determinetheamountof noisepowerpresentata decisioncircuit,it is necessary to determinely'pandthe effectivebandwidthof the detectioncircuitry' Whenthe detectioncircuiky consistsof nothingmorethana filter, theeffectivebandwidttris usually very closeto the 3 dB bandwidthof the filter. With othermore sophisticated detectors.the effectivebandwidthcanbe moredifficult to determine'As derivedin AppendixC, theeffectivebandwidthis usuallyreferredto astheeffectivenoiseband*idth (FIBW)of the receiver.Hencethe noisepowerat the decisioncircuit is (No) (NBW). 4.4.3 Error Probablllties to cross An enor in detectionoccurswhenevernoisecau$esthe signalmeasurement Appencircuitry' thethresholdbetweenthetwo nominaloutputlevelsof thedetection noiseas dix C derivesthis probabilityfor white Gaussian ,

;

\ I I -:-DfOD(erroD= lVZ?t6 "|

(4.s)

nf rzd4,

v

where v = nominal distance(voltage) to a decision threshold 02 = noise power at detector,= (No)NBW) Equation 4.5 is nothing more than the area under the probability density function of a normal distribution. As shown in Figure 4'23, the equation representsthe error probability as the probability of exceedingv/o standarddeviations in a normal distribution with zero mean and unit varianceN(l' 0). The error rate is completely determinedby the ratio of v to o. Since v is the noisefree sample voltage and o2 is the rms noise power, v?/O2is a signal-power*to-noisepo*"r rutio at the detector.This ratio is sometimesrefened to as a postdetectionSNR, iince it is measuredafter the detectioncircuitry. It is usually more important to express error rates in terms of an SNR at the input to the receiver. Figure 4'24 depicts a basic channel. a basic detection model, and the relation between a predetectionSNR and a postdetectionSNR. For reasonscliscussedin Appendix C, the most appropriatepredetection SNR for comparing line codesand digital modulation formats is an energy-

Dlo

l*

Figure 4.21t Probabilityof errorfor binarysignaling.

192

DGITALTRAN$MIssIoN ANDMULTIPLEXING

(';m;) (*) F#=(;rs) v1 i

Noire Ep€ctral denrity Ns

EquivalBnt noi* bendwidth : NBW

Noi* power = o ' = N B I / I ',N o

Figure 4.24 Signaldetectionmodel. per-bit-to-noise-densityratro E6/N0. The relationship betweenE6/N0Nrdthe signalpower-to-noise-powerratio is

g111p:siEpalPower norsepower _dEs1tr) NoNBW dErlog2l (llT) NoNBW where d= E = Ea= logyL = l/7= NBW =

(4.6)

pulsedensity energyper symbol energyperbit numberof bits per symbol(i.e.,t = numberof levels) signalingrate effectivenoisebandwidthof receiver

In determiningthe signalpower in Equation4.6, noticethe dependence on the pulsedensityd. In a balancedNRZ line codethepulsedensityis I, but in manyof the otherline codesthepulsedensityis dependent on thedataandany substitutionalgorithmsthat may be in use.In thesecodes,increasingthe prrlsedensityincreases the sNR but doesnot reducethe errorrate.The errorrateis determinedby the energyper-bit-to-noise-density ratio. In fact, increasingthe pulsedensitymight worsenthe errorrateifintederencebetweencablepairsis a significantconsideration. (Interferenceis directlyproportionalto thesignalpower.) Antipodal Signaling The optimumsignalingformatfor binarysignalingmaximizesthe errordistancefor a given receivedsignalpower and simultaneously minimizesthe noisebandwidth. This conditionarisesonly whentwo signallevelsareallowedandonly whenonesig-

4.4 ERRORPEHFOBMANCE 193

nal is theexactnegativeofthe other.Sincethesignalfor a I is theexactopposite(the antipode)of thesignalfor a 0, optimumsignalingis oftenreferredto asantipodalsignaling.Sinceno otherbit-by-bitsignalingschemeis anybetter,antipodalperfortnance Ofthe line codesdescribedpreviously,only is oftin usedasa basisfor comparisons. (NRZ) digital biphasecan provideantipodalperencoding and twolevel balanced providedby antipodal errorperformance the optimum 4.25 shows Figure formance. and the SNR. of E/Ne as a function signaling Error Bate of Level Encoded $ignale As alreadymentioned,a balanced,two-levelline codeis capableof providingoptiIf an unsymmetriclevel codeis used,suchasunipolar, *u* "ooi rateperformance. asin symmetriclevelencoding.Theonly differenceis is used basic detector the same "on" be movedfrom zeroto half theamplitudeof the must threshold thatthedecision signal.To maintainthe sameerrordistance,the averagetransmitpoweris increased penaltyof 3 dB with reby a factorof 2. Hencea unipolarcodecarriesa performance spectto antipodalPerformance. of a unipolar(on-oft) codewhereit is Figure4.25 showsthe idealperformance for all SNRsthe enor rateof the onNotice that performance. comparedto antipodal systemwith 3 dB lessSNR' antipodal rate of an the error equal to off systemis exaitly Bipolar Slgnaling bipolarsignalingis basicallyidenticalto a unipolar With respectto errorperformance, code.During any particularsignalinterval,the receivermustdecidebetweenoneof polarity.Hencethe decision two possiblelevels:zeroor a pulsewith the appropriate halfway betweenzeroandthe lies interval particular signal thresholdpertinentto a of thesignalinterval pulse to 507o the pulse. Nanowing amplitudelevelof theallowed (with to averagepulseenrespect performance doesnot changethe theoreticalerror can be usedto deterFigure 4'25 keying in on*off ergy).Thustheerrorratecurvefor mine theoreticalbipolar effor rates. in a bipolarline codecontributesto a slightlyhigherenor rate Oneconsideration bothpositiveandnegativenoise thanin on-offkeying. This increaseoccursbecause In zero-level signalis transmitted. when a cancausean erroneousthresholdcrossing level signal a lower positive noise when by contrast,a unipolarcodeis affectedonly is transmittedandonly by negativenoisewhenthe upperlevel signalis transmitted' If the bipolardetectorffeatsthe elroneouspulseas a I (despitea bipolarviolation)' theerroiprobabilitywhen0's aretransmittedis doubled.Thustheoverallerrorprobby 50Voif 0's and I's areequallylikely. ability is increased ofthe curve,anincreasein theerrorrateof 50Vodoesnot Becauseofthe steepness penalty.For example,if the errorrateis increased representmuchof a performance from 1 x l0-6 to 1.5x 10-6,the sourcepowerof a line codeneedsto increaseby only 0.2dB to getbackto a I x 10-6enor rate.At higherenorrates,a largerpenaltyoccurs in Figure4.25,where thecurveis not assteep.Theseeffectsaredemonstrated because (with be 507o0's) can comparedto the perof a bipolarcode the idealperformance formanceof a unipolarcode,

194

DIGITALTHANSMISSION AND MULTIPLEXING

SisElll|olrlr-lo-nqllF-pffi rrfi q LdE) 8 9 l 0 l t 1 2 t 3

6 lo-t F

E

E >

ll

F

2

3

f

E

8 7 B S r En rlv-FoFbit-b{ol$dilrity

0

l

t

1

2

1

rsflo, 4/IVo (dB)

Figure 4.25 Enor rates of polar (NRZ), unipolar, and bipolar line codes.

3

4.4 ERROHPEBFORMANCE 195

Thefactthatbipolarcodingincursa 3.2-dBpenalty(at 10-6errorrate)with respect to digital biphaseis indicativethat timing anddc wanderproblemsaresolvedby in*r*uring the numberof signallevels.In contrast'digital biphaseincursa bandwidth BNZS and penalry.Not all of the3.2-dBpenaltyfor bipolarcoding,andits extensions pattiming adding a consistent current and PST,"an be attributedto removingdirect or performance monitoring for redundancy tem.Bipolarcodingcontainsconsiderable * possibleerrorcorrection. cable systemon Z2-gauge Example4.3. Assumeeachsectionof a T1 transmission is near-endcrosstalklimitedandoperatingwith a 10-6errorrate.Whatdesignchanges areneededto reducethe error rateto 10-8? Solution. Sincethe systemis crosstalklimited,the enor ratecannotbe improved Thesolutionis to spacethe repeaters. by increasingthepowerout of theregenerative ofthis solution).FromFigure repeaters closertogether(ignoretheimpracticalaspects by I '6 dB to improve thatthesignalpowermustbeincreased +.i5 it canuedetermined the main lobe of the the error ratefrom 10-6to l0-8. In Figure4'10 it is seenthat the energyin the bipolar spectrumextendsup to 1.544MHz. However'most of of the attenuation spectrumliesbelow 1 MHz. UsingFigureL14, we determinethat = reduction ft is the cableis 5 dB/kft at 1 MHz. Hence1.6/5 '32 kft or 320 Z-Z-gauge in repeaterspacingrequired. a numberof importantaspectsof digital transmission Example4.3 demonstrates qualitycanbe obtainedwith only a small good transmission arbitrarily iirst, systems. plnalty in transmitpoweror repeaterspacing.tThus,as mentionedin Chapter2' a andswitchingnetworkcanbe readilydesignedto impalt no degdigitaltransmission converanddigifal-to-analog quality, exceptwhereanalog-to-digital to voice rJation place. take sions Second,the dramaticimprovementin errorratefor a relativelysmallincreasein the SNR impliesextremesensitivityin the oppositedirectionalso.A slightincrease in noisepoweror signalattenuationwould causea largeincreasein the error rate. betterperformHencethenominaldesignof a digitallink oftenprovidesconsiderably necessarY. than normallY ance formatusing Third, thesolutionto Example4.3 appliesto anydigitaltransmission all line Since per mile. attenuation the same hence, and, of frequencies the sameband change relative rate, the a 10-6 error at same steepness the codeshaveapproximately in SNR is the samefor all systems.Thus a Tl systemwith suboptimumdetection weremoved320ft improvementif therepeaters wouldexhibitthesameperformance 10-6 errorrateto begin less than a provide (Tl to are designed lines closertogether. on Tl lines') rate surveys of error repofl for a thorough with. Seereference[32] "In

pulse is removed to terms of logicJevel decisions, a bipolar code carrrot provide effor conection, If a error is made. another of the time efior is correctedbuthalf of the time the half eliminate abipolar violation, a bipolar Enor corection is possible only with a Viterbi-like detectot described previously. In this case code' performance a unipolar than better code orovides tsome systems, such as the tadio systems discussedin Chapter 6, require a greater decreaseifl repeatel spacing to achieve the same improvement in performance'

196

DIGITAL TRANSMISsIoN ANDMULTIPLEXING

E

t

s

I

8

9

t

0

t

l

1

2

1

3

t

4

r

b

1

8

Arrsrrgl snrqy-p€Fblt-to-ndire-dondty iltio, Ebftro (dB) Figure 4.26 Error rate of balancedmultilevel signals (all systemsproviding an identical data rate).

4.4 ENHOHPERFORMANCE 197

8

e

l

0

tl

l3 14 1? EulVo ort tlrtdunncl

ID

ru

Figure 4.27 Error rates of I + D partial-responsesystems.

198

DG|TALTRAN$MtsgtoN ANDMULT|PLEXING

Multllavel Error Eatee Themultilevelkansmissionsystemshownin Figure4.16doesnot requirebandwidth in excessof a binarysystemusingthe samesignalingrate,yet it achievesthreetimes thedatarate.Thepenaltyfor multileveltransmission comesin theform of greatlyincreasedsignalpowerfor a given error rate.For example,the averagepowerof the eight-levelsystemin Figure4.l6 is 8.7 dB abovethe averagepowerof a symmetric two-levelsystemwith thesameerrordistance.To makematter$worse,somesysrems arepeakpowerlimited,in whichcasetheeightJevelsystemhasa 12.4-dBdisadvantagewith respectto a two-levelsystem.Theerrorratesof multilevelsystemsarederivedin Appendixc andplottedin Figure4.26asa functionof E6/N11.For thosecases wherethe peakpoweris of interest,the relationshipbetweenthe peakand average powerof a multilevelsy$temis derivedin AppendixC as Peak-to-average(dB)= l0log,o

t.

tlt

(u L)E'/: At - rf

(4'7)

whereI is thenumberof equallyspacedlevelscenteredaboutzero[e.g.,+1, +3, +5, . . ., J (r -1)1. The enor ratesof I + D partial-response rrystems arealsoderivedin Appendixc andplottedin Figure4.27.Theseerrorrate$arederivedundertheassumption thatbir by-bit detectionis used.Sincepartial-rerrponse systemscontainredundancy (conelation) in adjacentsamples,betterperformance cal be achievedwith Viterbi decoders ll 21.

4.5 PERFORMANCEMONITORING Two basictechniques existfor directlymonitoringthequalityof a digitaltransmission link: redundancychecksand pulsequality measurements. Both techniquesare designedto providean indicationof thebit errorrate(BER)of the channel. 4.5.1 RedundancyChecks Redundancy canbeincorporated into a digitalsignalusingoneof two commonmethods.First,the line codeitself may containredundancy asin bipolarcoding.In a random,independent errorenvironmentthefrequencyof bipolarviolationsis verynearly equalto the channelBER (exceptfor extremelyhigh BERs).Second,logic-levelredundancycanbe insefiedinto thedatastreamin theform of overheadbits.For example, parity bits are insertedinto DS3 and DS4 signalsfor the expre$$purposeof monitoringthechannelerrorrate.(Theframestructures of theseandotherhigherlevel multiplexsignalsareprovidedin chapter7.) cyclic redundancy check(cRC) codes arealsoincorporated into a numberof transmission systemsasa mean$of monitoring BERsandvalidatingframingacquisition.Two prevarent examplesof cRC useare(l) extendedsuperframe (ESF)onT I lines[33] introducedwith theD4 channelbanksand (2) opticalfiber transmission usingthe soNET standarddescribedin chapterg.

MONITORING 199 4.5 PERFORMANCE

The useof logic-levelredundancy(eitherparity bits or CRC codes)is generally (suchasbipolarviolations)because thelatbetterthantheuseofline coderedundancy redunlink itself. Logic-level transmission on the technologyof the ter is dependent or radio. pairs fiber from cable to dancy,on theotherhand,is unaffectedby a change ESF into TI As describedlaterin thischapter,a majorimpetusfor theintroductionof of bipolar monitorabilityindependent transmission systemswasto get performance line codes.DSI signalsareoftenmultiplexedinto higherleveldigitalsignalsandcarried on a varietyof ftansmissionsystemssuchasfiber andradio.The CRC codesin theESFframingformatprovidethemeansfor end-to-endpedormancemonitoringinsystemscarrythedatastream. of whatevertransmission dependent pafity bits andCRC codesdo not providea line code redundancies, In contrastto equationrelatestheparityerrorrate The following one-to-oneindicationof theBER. (PER)to thechannelBBR:

pER= I [T]p(r * p)N-i -t'J

(i odd)

(4.8)

t=l "

whereN = lengthof a parityfield (numberof bits overwhichpantyis generated) errors p = BER assumingrandom,independent The relationshipbetweenthe PER andthe BER is plottedin Figure4.28for DS3 N timestheBER andDS4signals.Noticethatat low errorrate$thePERis essentially because anyoddnumber (Np).At high errorrates,however,thisrelationshipchanges '\ilhen the BER is high from a singleerror. of errorsin a frameis indistinguishable enoughthatmorethanoneerrorin a parityfield is likely, thePERis uselessasanabsoluteestimateof theerrorrate.In thesecasesthePERindicatesonly thattheBER is equalto 1/N.Becausea DS4frameformatcontains abovea thresholdapproximately for highererrorratesthan pariry bits,DS4signalscanbe measured a higherdensityof canDS3signals. it is unlikelyfor multiple Determination of a CRCerrorrateis simplifiedbecause errorsto notproducea CRCerror.Thustheprobabilityof a CRCenor (CRCER)is I minusthe probabilirythat no errorsoccur; CRCER=1-(l-p)N

(4.9)

whereN = lengthof theCRCfield (includingCRCbits) enors p = BER assumingrandom,independent Again,at low errorratesEquation4.9 revertsto N timestheBER (Np).Equation4.9 is plottedin Figure4.28for ESFframeswheretheCRCfleld is 4614bits.(TheCRC field doesnot includeF bits in Table4.8 excepttheC bits themselves.) mUStbe long thesample$equence in enOrratemeasurements, To haveCOnfidence enoughto allow anaverageof about10enorsin thesamplesize(e.g.,thesamplesize

d :Y E o .f g

b

= 0

.$ CN (4 rTl

t

a

b F

;

s

F

;

f

E

o

v

S m

E e bI)

qa

tr &

d

a d

U}

a

E t t E}

h q

'-Fa ff (b F!

+

{) b!

EI

eler r0r,$ cHs#til-|ed

4.5 PEBFORMANCE MONITORING 201 TABLE 4.8 Extendsd Superframe Framlng Channel Formata

F-BitAssignment ESFFrame Number 1 2 3 4 5 6 7 I

s 10 11

12 13 14 15 16 17 1B 19 20 21 22 23 24

ESFBitNumber 0 193 386 579 772 965 11 5 8 1351 15,14 1737 1930 2123 2316 2509 27Q2 2895 3088 3281 3474 3667 3860 4053 4246 4439

FPS

FDL

cRc

m

cB1 m 0 m

c82 m 0 m

c83 m I

m

c84 m 0 m

cB5 m 1

m

; m 1

aFPS,framingpatternsequonc€(. . .001011.. .); FDL,a kbpsfacilitydata link(messagebits m); CRC' GRC'6 cyclic redundancycheck (check bits CBl-CB6).

must be I0/BER). Hence, when trying to measurelow BERs (e.g., 10*6or l0-7), the measurementtime may be too long to respondto changing channelconditions such as radio channelfading.

4.5.2 SlgnalQualityMeasurements The secondbasic technique for monitoring digital ffansmission quality is to process the digital signal directly and measure ceftain properties related to lhe error rate. A simple approach involves merely measuring the received signal power, a coillmon technique in analog systems.In a fixed-noise environment this approachis adequate. However, on transmissionlinks where the noise level can vary or where signal distortions can arise, the quality of the pulsesthemselvesmust be measured.

202

DIGITALTHANSMISSION AND MULTIPLEXING

Docirion threshold

Pmudoerror rcgion

Figure 4.29 Pseudoerrordetection.

Figure4.29demonstrates theoperationof a'!seudo" errordetectordesigned to detect receivedpulseswith abnormalamplitudesat the sampletimes.In the example shown,binarydatais detectedby useof a singlethresholdlocatedmidwaybetween the normalpulseamplitudes.Two additionalthresholdsare includedto detectthe presenceof pulseswith abnormalamplitudes.Samplevaluesfalling into the central decisionregiona.renot necessarily dataenors,but a high pseudo-error rateis a good indicationthatthechannelis not performingproperly. In a random(Gaussian) noiseenvironmenttheoccuffencerateof pseudo*effors is directlyrelatedto theactualerrorrate.Figure4.30showsa Gaussian noisedistribution anddecisionthresholds chosento producepseudo-errors at I 00timesa 10-6errorrate. Hencean attractivefeatureof this error ratemeasurement is that it can measurevery low errorratesusingcomparativelyshorttestintervals.Note,however,thattheerror multiplicationfactoris dependent on theerrorrate.Thetechniqueof estimatinga vsry low error rateby extrapolatingfrom an artificially generatedhigh error rate is sometimesreferredto asa Q-factortechnique[34]. In essence, (meanand the parameters variance)of a Gaussian distributionareestimatedwith easilyobtainedmea$urements sothemuchlowererrorratescanbe estimatedfrom theparameters. This techniqueis particularlyusefulin opticalfiber systemswhereextremelylow errorratesareto be determinedthatotherwiserequireextremelylongmeasurement times.

-1

-0.2 0 0'2

1

Figure 4.30 Pseudoerrordecision thresholdsfor error multiplication of 100.

4 . 5 P E R F O H M A N C E M O N I T O H I N G 203

ratesarenot of pseudo-error detectionis that pseudo-elror A majordisaclvantage Channeldisaccuratelyrelatedto actualerrorrateswhenthe noiseis non-Gaussian. rateswhenthe actualerrorratesare tortions,in particular,causehigh pseudo-error small.Eventhoughabnormalpulseamplitudesarepresent,enorsdo not occurunless noiseaddsto thedistortion.On theotherhand,impulsenoisemightproducehigh acrates. tual errorratesbut relativelylow pseudoerror anddeducederrorratesis improvedwhen Thecorrespondence betweenmeasured decisionrethedecisionthresholdis narrowed.Unfortunately,narrowerpseudo-emor "error" rate multiplication gions are more difficult to implementand provide lower factors. 4.5.3 Framing Channel Errors As describedin Section4.6.2.identificationof individualchannelsof a TDM data sffeamrequiresoverheadbits with a uniquedatapattemfor frameidentification.After a receivingterminalhas"locked" ontothe ftamingpattern,it ignoresoccasionalbit errorsin theframingpatternbut declaresan out-of-frame(OOF)conditionwhenthe Thusframingbit errorscanbeusedto determinelow errorrates erTorrt areconsistent. lossof frambut not higherrorratesthatcausefalseOOF indicationsandsubsequent sohigh are usually ing.Errorratesthatarehighenoughto causefalseOOFindications thatthelink is effectivelyout of service.

4.5.4 Performance Obiectives link is its BER.If theerrors of qualityof a digitaltransmission Themostbasicmeasure (i.e., manner theyconformto a simplePoisoccurin a truly randomandindependent Unsonprobabilitymodel),the averageBER completelyspecifiestheperformance. fortunately,elror rateobservations[32] showthat ertorsoften occurin bursts,and becauseburstsaffectdifferentservicesin differentways,specifyingthe qualityof a link requiresmorethanjust the averageBER. For example,datatraffic transmission is generallytransmittedin blocksthat areretransmittedno matterhow manyerror$occur in a block.Thus,a bursthaslittle moreeffectthana singlebit error.The rateof of datablocksis morea functionof thefrequencyof theburststhan rekansmissions of the long-teflnaverageBER. Voice traffic, on the otherhand,is increasinglydegradedby all bit errorswhetheror not theyoccurin bursts. to certainopChannelerrorsthatoccurin burstscanalsobemuchmoredeleterious erationsof a networkthanaredistributederror$at a similaraveragerate.A lossof framing,for example,occursmoreoftenin a bursterrorenvironment.'Certaintiming (pulsestuffing)described in Chapter7 arealsosimilarlyaffectedby bursts. operarions erbothproduceoutages(e.g.,continuous synchronization Lossesof framingandtime *Errors

in a burst are not necessarily contiguous. They merely occur in a short time intewal at a rate that i$ too high to be r€asonably explained as chance independent events.

204

DIGITALTRANSMISSION AND MULTIPLEXING

rors) in the associated haffic channelsor eveninadvertentdisconnects if the conditionslasttoo long. Commonperformanceparameter$usedto characterizeerTorratesare: L Error-FreeSecond(EFS); Becausedatablockstransmitredat 64 kbpsrequire much less than a second,the percentageof EFS essentiallyspecifiesthe percentageof time that the channel is available for data transmission applications.(Whentheblock transmissions aremuchshorterthana second.a shortermeA$urement intervalis moreappropriate.) 2. BurstErroredSecond(BES): An erroredsecondwith at least100errors. 3. Enored Secand(ES): A secondwith at leastoneerror. 4. SeverelyErroredSecond(SES);A secondwith a BER grearerthan(10)-3. 5. DegradedMinute(DM): A minutewith a BER greaterthan 10-6. Thelastthrceparameters aredefinedin ITU recommendation G.821for performance objectivesof a digitalnetwork.Theobjectivesfor a 27,5ff)-kmreference connection(implying errorcontributions frommany$ources) areES< 87o,SES<0.ZVo,and DM
..5 PERFORMANCE MONITORING 205

Reed-SolomonCodes Although a wide variety of block-codingalgorithmshavebeendeveloped,ReedSolomon(RS)codingis themostpopularform. Whena blockof sourcedatasymbols of lengthM is inputto anRScoder,an outputblock of lengthN symbolsis produced, is whereN - M is thenumberof checksymbolsR. An RScodewith theseparameters = R) code can N ,R) code. An RS(N, ry referred to as a RS(N, M) RS(N, commonly a symbolconsistsof an 8-bit byteof data,*so conect+R symbolerrors.In mostcartes no worsethana singlebit enor. Digital video in a single byte are multiplebit errors (DVB) use RS(204,188) codes,which meansthat asmanyas broadcasting $y$tems and be corrected. eightbytescanbe coffupted Exampte4.4. Determinethe probability of failure of an RS(204, 188) code operatingin a randomerrorenvironmentwith a bit errorprobabilityof 10-3'Assume eachsymbolis an 8-bitbyte. Solution. Becausetheprobabilityof multiplebit errorsin a singlebyteis small, the probabilityof a symbol error pr is very close to 8 x l0-3. Using p,, the probabilityof failure (theprobabilityof morethaneight symbolerrors),pp canbe determinedas zo4

-P,)'*' o,=EfI)'*' I

=l-T Lt

r'4

-Pr)to*t

f?-)"r'

= 0.00005 GonvolutlonalCoding Block diagramsof two basic convolutional encodersare shown in Figure 4.31. Both of theseencoder$are referred to as rate l/2 encodersbecausethe sourcedata rate is half the channel data rate. During each bit time of a sourcetwo output bits are generated.As illustrated in Figure 4.32, the constraint-zlength coder in Figure 4.314 outputs odd parity over bits A and B along with the value of B directly. In Figure 4.31b, odd parity acrossoverlapping fields (A, B, C and A, C) are generated.In the caseof odd parity over A, B, C an output value is a I if and only if an odd number of A, B, C are 1. An important considerationof a convolutional coder is the constraint length,wlich specifiesthe number of instancesthat a particular sourcebit getsmappedinto a chan*Reed*Solomon symbol in bits.

codes are sometimes denoted as RS(N' N - R, B), wherc ,B designatesthe length of a

206

DIGITALTRANSMISSIONAND MULTIPLEXING

(o)

(b) Figure 4.31 Rate l/2 convolutionalencoders:(a) sourceconstraintlength 2; (b) source constraintlength3. nel bit. In Figure 4.3la the length of the shift registersis 2 bits, which meansthat this coder has a sourceconstraint length of 2. Becausethe output clock is twice the input rate, the channel constraint length is 4. Similarly, the encoder shown in Figure 4.31b has a sourceconsffaint length of 3 and a channel constraint length of 6. Longer constraint lengths produce better performance. Because the encoder shown in Figure 4.31a has the shortestpossibly constraint length, it is not commonly used but is convenient for illustrating the basic operationofa convolutional decoder.Notice that each

Sourcedata

0

F ,.. h 1

0

1

1

0

0

1

0 -{ I

01 11 0 1 1 0 11 00 01 1 1 -

Channeldata

Figure 432 Exampleinput and outputdatasequences for convolutionalencoderin Figure 4.31b.

MULTIPLEXING ?;O7 4.6 TIMEDIVISIoN

sourcebit in Figure 4.3la is mappedinto 3 channelbits. It is this redundancythat gets processedto recover the original data. Example 4.5. Determine the decoding logic to decode received data for the convolutional encoderof Figure 4.31a. Assume bit-by-bit decisionsare made on each received channelbit and only consider isolated channel errors. Solution. Two casesare possiblefor isolated channelerrors: an error in a sourcebit B or an error in a parity bit. If a sourcebit is receivedin error, two parity errors result. Thus, when two adjacent parity errors occur, the most likely scenario is that the implied databit needsto be inverted. If a single parity bit is receivedin error, the most likely scenariois that the parity bit is itselfin enor and no sourceerrorshave occurred. Thus, isolated parity error$are ignored. Example 4.5 illustratesthat a convolutional encoder/decodercan easily correct isolated channel errors-at the expenseof doubling the data rate. In contrastto the previous example and the normal implementation of an RS decoder, a convolutional decoder usually processe$the received signal with maximum-likelihood sequence (Viterbi) detectors[36]. In essence,a sequenceof receivedsignal samplesis compared to all possible channel input sequenceswith the closestallowable sequencechosento determine the sourcedata. The addedcomplexity provides signif,rcantbenefits. Constraint Iength 7 convolutional coding, for example, provides better than 5 dB of improvement in error performance for a given SNR. Since a doubling of the necessary bandwidth (for rate l/2 coding) implies a relative noise increaseof 3 dB, a net gain of 2 dB is achieved- Notice further that a convolutional encoder can correct a much greater number of comrpted bits than can an RS code-as long as the comlpted bits are separatedby something greaterthan the channelconstraint length. Hence, convolutional coding is well suited to correct independentenors while RS codeswork well in burst error environments.For thesereasonscombinationsof the two codesare often used. Maximum-likelihood detection of convolutional code sequencertis very similar to trellis-coded modulation (TCM) detection discussedin Chapter 6. The TCM discussion provides some example measurementsthat involve the same basic processrequired for Viterbi detectionof convolution codes.

MULTIPLEXING 4.6 TIMEDIVISION andis usedin somespecial signals is possible AlthoughFDMof digitaltransmission situations,TDM is by far themotitcolnmonandeconomicalmeansof subdividingthe capacityof a digital transmissionfacility. One applicationwhereFDM techniques lines,where havebeenusedfor digital signalsis on multidropdatacommunications of thedataaredistributedalongtheline. Most telephone the sourcesanddestinations network applications,however,involve clu$tersof channelsin the form of trunk groupsbetweenswitchingoffices.In local digital accessapplications,wheresub-

208

DrGrrAL TRANSMtsstoN AND MULTIpLEXtNe

scriberlocationsare distributedthroughouta servicearea,channelsare sometimes addedanddroppedon a distributedbasis,but only with a limited numberof access pointsbecause of maintenance andreliabilityconsiderations.* Therearetwo basicmodesof operationfor TDM: thosethat repeatedlyassigna pofiionof thetransmission capacityto eachsourceandthosethatassigncapacityonly asit is needed.Thefirst form of operationis referredto assynchronaus time division multiplexing(STDM) whennecessary, to distinguishit from the "as-needed" mode of operation.Otherwise,TDM is generallyunderstood to imply the synchronous variety.Theas-needed form of TDM is variouslyreferredto asasynchronous time division multiplexing(ATDM), AsynchronousTransferMode (ATM), srarisricaltime division multiplexing(Stat-Mux),or packetswitching.Circuit-switchedtelephone networksuse STDM, whereasdatanetworkstypically use ATDM. Discussionof theselattertechniques is deferredto Chapter10. Thetermasynchronous is alsousedin anotherTDM contextto referto themultiplexingof multiple,independent tributarysignalsinto a singlehigherratesignal.In this context,"asynchronous" is totallyunrelatedto eitherasynchronous transmission asdescribedat thebeginningof thischapteror packetlikemultiplexingmentionedin the previousparagraph. In this third contextasynchronous refersto multiplexingof multipleunsynchronized tributariesinto a singlehigh-rateserialstream.This chapter is concerned only with multiplexingof synchronized tributaries.TDM of unsynchronizedtributariesis describedin Chapter7. 4.6.1 Bit Interleaving Versus Word Interleaving Two differentstructures of (synchronous) time divisionmultiplexframesareshown in Figure4.33.In thefirst instanceeachchannelis assigned a time slotcorresponding to a singlebit-hence the term bit interleaving.In the secondinstanceeachchannel is assigned a longertime slotcorresponding to somelargernumberof bits,referredto asa word-hence the tennword interleaving. Thedecisionasto whichstructureto usein a particularapplicationis primarilydependenton the natureof the sources.[n T-carriersystemseachchannelproducesa complete8-bit word at a time.Henceword interleavingis desirableso all bits canbe transmittedasgenerated. ln conffast,higherleveldigitalTDM multiplexerstypicallyusebit interleavingof the lower level multiplexsignalssincethe lower level signalsrepresentcontinuous, I-bit-at-a-time,datastreams. Thespecificformatsof thehigherlevelmultiplexsignals aredescribed in Chapter7 whensynchronization of bit streamsis considered. Onenotableexceptionto higherlevelbit interleavingis theword-interleaved structureof fiber-based SONETor SDH multiplexingdescribedin ChapterL 'The

digital network has evolved with ever-increasinglevels ofmultiplexing creating serial bit streams with ever-increasing data rutes. The use of WDM on optical fiber systems represeflts a deparhrre from the single-streamparadigm stimulated by two factors: (l) it is impracticalto tansmit the ultrahigh speed serial channels and (2) separate wavelengths provide transparency for diverse applications.

4.6 TIME DIVISIONMULTIPLEXING

209

1 Frame

f+ Bit int€rlervlng I Freme

lilord intcrlearing

'

r'ffl =i L =P r=F r

1

2

3

4

1

TDM multiplexers, Figure 4.33 Bit interleavingandword interleavingof four-channel

4.6.2 Framlng To identify individual time slots within a TDM frame, a receiving terminal uses a counter synchronizedto the frame format of the transmitter. Just as for synchronization of sampleclocks, a ceftain amount of hansmission overheadis required to establish and maintain frame synchronization.In fact, most of the techniquesusedfor frame synchronizationare directly analogousto clock synchronizationtechniquesdiscussed previously. Specifically, the basic meansof establishingframe synchronizationare; l. Added-digit framing 2. Added-channelframing 3. Unique line signal framing 4. Statistical framing The main considerationsin choosing a framing procedureare time required to establi$h framing, effects of channel erTor$in maintaining frame synchronization,relationships between the line clock and sample clocks derived from the line clock, transmissionoverhead,and complexity of the framing circuitry. The severiry of a loss of framing and the time required to reestablishsynchronization dependon the nature of the traffic. Since a loss of framing implies a loss of data on all channels,the mean time between misframes must be as long as possible' For voice traffic infrequent misframes can be tolerated if frame synchronization is rees"glitch" in the output speech' tablishedrapidly enoughto minimize the duration of the For data traffic the duration of reframe times is not as critical as the frequency of occurrence since most data communicationsprotocols initiate recovery proceduresand retransmit a mes$ageno matter how many data bits are lost. A critical requirement for reframe time in the telephone network comes from the possibility that various in-channel control signals may be lost and interpreted as disconnects.Thus the maximum refiame time on a particular digital transmissionlink is often determinedfrom analognetwork signaling conventions.A loss of framing is also used as a performancemonitor for the transmissionlink and usually setsalatm condi-

210

DrcrrAL THAN$MrssroN ANDMULTIpLEXtNc

tions, which in turn causeautomatic switching to spareterminals or transmissionfacilities. Generally speaking,terminals serving 6(X)or more channelsrequire automatic protection switching [37]. Out-of-frame conditions occur in two ways. First, the locally derived sampleclock may lose synchronization with the line clock and produce a slip in the counter sequence.Normally, the timing information in the line clock is sufficiently strongto prevent misframes of this type. An exception occurs on radio links when deep fades reducethe signal power to the point that clock synchronizationis impossible. Receiving terminals can also lose clock synchronizationthrough no fault of their own or of the transmissionlink. If the sourceclock at the transmitting terminal has too much jitter or generates abrupt phase shifts, receiving terminals may be unable to track the clock signal. Both phenomenaarerelatedto network synchronizationpractices,asdiscussedin Chapter 7. A second source of mi$frames is from channel errors creating false out-of-frame framing pattems. Thus considerableredundancyin the framing pattem is required to minimize the probability of false misframes.A loss of framing is determinedwhen the occurrenceof framing pattem violations exceedssome shoft-term density threshold. In all of the framing techniques discussedin the following paragraphs,special framing bits or codesare insertedinto the information skeam. Theseinsertionsdo not have to occur with each frame. lnstead, they can be sent only once for a predetermined number of information frames. In this manner the transmission overhead for framing can be reduced,which is particularly important in bit-interleaved systems.An individual information frame containedbetween framing indicators is sometimesreferred to arra "subftame,"

Added-DigitFramlng Onecommontechnique of framinga digitalTDM informationstreamis to periodically insert a framing bit with an identifiable data sequence.Usually the framing bit is addedonce for every frame and alternatesin value. This particular format is the procedureused to establish framing in the original Dl channel banks. When the Tl line carriesonly voice traffic, this framing format is pafiicularly useful since no information bits can sustainan alternating l, 0 pattem. (An alternatingpattern representsa 4-kHz signal component, which is rejected by the bandlimiting filter in the PCM codecs.) Framing is establishedin a receiving Dl channelbank by monitoring frrst one bit position within a 193-bit frame and then another, until the alternating pattern is located. With this framing strategy, the expected framing time from a random starting point with random data i$ derived in Appendix A as

I averagenumberof) bitsto I I determine thatan I I linformationpositionis I not a framingnositionJ I

MULTIPLEXING 21 1 4.6 TIMEDIVISION

= (+N)(2N+l) = M + |N bit times

(4.10)

where Nis the number of bits in a frame including the framing bit. For D I channel banks, N = 193 so that the framing time is 37,346 bits, or 24. I 88 msec.Also of interest is the maximum framing time. Unfoftunately, fhere is no absolute maximum framing time for a Tl system with random data. It is very unlikely, however, that the framing time would ever exceedthe averagesearchtime fbr all bit positions, or 48.25 msec.This latter measureof framing time is refered to asthemaximum(werageframe time.It i$ the averagetime required to establishframing, but with the assumptionthat all bit positions must be testedfor the framing sequencebefore the actual framing bit is found. Obviously, the maximum averageframe time is twice the averagevalue from a random statting point defined in Equation 4. I 0. The framing time can be reducedby using more sophisticatedframe searchstrategies. One approach examines one bit at a time, as before, but during a reframe the searchbegins a few bit positions in front of the presentposition under the assumption that short lapsesof clock synchronizationcausesmall counter offsets. A secondapproach [38, 39] uses a parallel searchby monitoring all bit positions simultaneously for the framing pattern.With this framing procedure,framing is establi$hedwhen the last of the N* I information bit positions finally producesa framing pattern violation. The probability that all information bit positions produce a framing violation in n or less frames is derived in Appendix A as follows:

prob(frame time < n) = [ * (*f]t-t

(4.1l)

's whereN is thenumberof bits in a frameandtheprobabilityof I i sj. UsingEquation 4. 11, we determinethe medianframingtime by settingprob(frametime < n) = i. Hence

n = -logzll- (+)r/(N-')l

{4.r2)

Setting N = 193 for the frame length of the Dl channel bank producesthe result that n = 8.1 frames, or approximately I msec. An even more sophisticatedframing strategyinvolves continually monitoring all bit positions, even while the system is synchronized.Then, when a misframe is detected,a new frame position is immediately establishedor the frame searchtime is significantly reduced.Fufthermore, the continuous searchfor framing patternsprovides additional information for declaring an out-of-frame condition. (There is little point is discarding the presentframe position unless anotherbit position exhibits a more consistentfiaming pattern.)

212

DtctrAlTRANSMtsstoN AND MULTIpLEXINc

Theframingpatternof second-generation channelbanks(D3, D4) from AT&T was changedfrom thealtematingl, 0 patternto establisha longersequence for identification of signalingframes.As mentionedpreviously,thesechannelbanftsprovideI bits of voicefor all time slotsexceptin everysixthframe,whichusestheleastsignificant PCM bit for signaling.The signalingchannelsthusderivedaredividedinto an A and a B subchannel, implyingeachsubchannel sendsa bit in everytwelfth frame.Hence a l2-bit framingsequence is neededto identifythesignalingbitsandthetwo signaling subchannels. The lZ-frame superframe(SF) structureand associated framing sequenceareshownin Figure4.34. Whendescribingor processingthe framing$equence shownin Figure4.34,it is convenientto divide the framingbits into two separatesequences. During the odd framesthe framingbit altemates,while duringthe evenframesthe framingbit sequenceis Offi111000111000. Figure4.34indicatesthattheA signalingframecanbe identifiedby a 0-to-l transitionin the even-numbered framesequence. Conespondingly, a 1-to-0transitionin theeven-numbered framesequence signifiesa B signaling frame.Frameacquisitionbeginsby findingthealternatingbit sequence (with 385interveningbits). Then,the 000111framing patternis located.Anotherframing sequence,extendedsuperframe (ESF),for DSI signalsis describedin Section4.6.3.

Added-Channel Frdmlng Added-channel framingis basicallyidenticalto added-digitframingexceptthatframing digits areaddedin a groupsuchthat an exha channelis established. Hencethe Framing bir

Freme no, 1 2

OFil-T]

3

offi

l--::]

4

OFilTl

I-I

5 6

rlilTl lFI

Ftr-I 0FI*--*--T--*---]

-l f T--TI |

I

1 2 0 Figure 4.34 Twelve-framesuperframesfucturc of DSI signalwith robbeddigit signaling, Framealignmentsignal(F) = l0l0l0; multiframealignmentsignal(M) = ffil ll0.

MULTIPLEXING 213 4.6 TIME DIVISION transmissionrate of individual channelsis integrally related to the line rate of the entire multiplex. The fact that the frame boundariesare identified by whole codewordsaddsconsiderableperformanceand flexibility to the framing process.First, framing can be established more rapidly since random 8-bit codewords are very unlikely to appear as f'raming codes. (See the problems at the end of the chapter.) Second,the larger code spacesimplifies identification of auxiliary functions such as superframeboundaries, parity bits, or equipment$tatus.In most systemsthe addedchannelcontainsmore than framing bits. The first-level digital multiplex signal of ITU (El) is an example of a system using added-channelframing. The El standardestablishes32 channelsper frame with one channelproviding framing and one more channeldedicatedto signaling. Thus, 30 of 32 channelsare availablefor messagechannels.Figure 4.35 showsthe frame structureof the El signal. The frame alignment signal (FAS) is insertedinto the framing channel of every even-numbered frame and a I bit inserted into the secondbit position in every odd-numberedframe (to preclude FAS generationin those frames). The first bit of every FAS frame may carry a cyclic redundancy check (CRC-4) for additional frame alignment integrity and eruor rate monitoring. The remaining bits of the framing channel are used for CRC-4 alignment or alarm indications or re$ervedfbr other uses.(See ITU recommendationG.704 for more details.) The signaling channel ofFigure 4.35 depicts the use ofchannel-associatedsignaling (CAS), which implies that 4 bits per 16-frame multiframe are allocated to each messagechannel. The positions of the associatedsignaling channelsare determined with respectto the multiframe alignment signal (MAS). The 4 bits of each CAS signaling channel should never be 0000 to preclude inadvertent generation of the

Framing channel

Mesage channelr

0

1

FramooTTEET-I Framer f'lil-T

l

Framez fl- FAsT-l

Metcoge chennols

Signsling ch6nn6l

1

6

3

1

"'

lMAsTl

"'

"'

f rTG-]

"'

l-l

...

I ,Tu-]

"'

|_*--l

l-]

a

,rPreclurion of FAS

FrEmB rb

I trt

Ftame o

[EsT--*]

|

-l

"'

| fi'T3t-l

"'

T---l

...

liifrHT-l

...

l-l

Figure 4.35 Channelformatof ITU primarydigital signal(El).

l

214

DIGITAL TRANSMISSIoN ANDMULTIPLEXING

MAS. The use of thesebits is similar to the "bit-robbed" ABCD birs of the North American DS I signal.When common channelsignalingis carriedon an E1 link, the CAS channelsare replacedby a 64-kbps HDLC signaling link in time slot l6 of the El frame. The averageframe acquisition time of a multibit frame code is derived in Appendix Aas

:#j.+ Frame rime

(inbits)

(4.13)

whereN is the length of a frame including the frame code, /, is the length of the frame code,and it is assumedthat I's and 0's are equally likely. From Figure 4.35 it can be seenthat for the El signal N = 512 and L = 7. Thus the averageframe time from a random starting point is determinedfrom Equation 4.13 as 0.5 msec.Again, the "maximum average"frame time is twice the averagefrom a random starting point, or I msec.Notice that theseframe times are much faster than DSI added-digit frame times becausea higher percentageof bits are allocatedto framing.

Unique Line CodeFramlng Bipolar coding managesto shapethe spectrum(remove the dc component)of the line code by adding extra signal levels to provide more flexibility in selectingsignals.The sametechnique can be used to establishframe synchronization.In fact, with bipolar coding and added-digit framing, bipolar violations can be usedto identify the framing boundariesuniquely and rapidly. A particularly significant exampleof using line code violations for framing is the ISDN S/T basic rate interface describedin Chapter 11. Even without added-digit framing, unique line codescan be used to carry information while simultaneouslyconveying frame positioning. If the number of signal levels is doubled for the framing bit only, the extra levels provide sufficient code spacefor the information but uniquely identify that bit position as a frame boundary. This procedure does not necessarilyincreasethe effor rate, since for any particular bit (information or framing), half of the levels can be disallowed. The main advantageof using unique line codesfor framing is that the information bit positions cannot generateframing pattems. Hence framing is establishedas soon as a frame bit occurs, and misframes are detectedalmost immediately. The main disadvantagesofunique line codes are the added signal processingrequirements(if new levels are establishedjust for framing) and the dependenceon the transmissionterminals to locate framing. With the other framing techniquesdescribed in this section, the framing pattems are representedin the data of the bit stream. Hence the transmission equipment can be changed independently of the multiplex equipment. For additional information on framing techniquesand performanceanalysessee references[40] and [41].

4.6 TIMEDlvlsloNMULTIPLEXING 215

Statlstlcal Framing of datawithinindividualbitsof a transmisframingreliesonthestatistics Statistical sionscheme. Assumingthesourceof thedatais known,it maybepossibleto ascertain informationsuchaswhichbit is a mostsignificantbit (MSB) of a PCM codewordand therebyrecoverbyteframingwithoutan explicitframingbit of anytype.Obviously, Onesuchapplication thismethodof framingis only applicableto specialapplications. is determiningwordalignmentin ADPCM7-kHzaudio[42],asin ITU-T RecoiltmendationG.722. 4.6.3 DSI Extended Superframe link, the operatingtelephone When a Tl line is usedas an interofficetransmission and performance companieshave accessto both endsof the line for maintenance monitoring.Furthermore,most installationsincludeprotectionswitchingfacilities to that thatcanalsobe usedto routinelytesta line while thetrffic normallyassigned line is divertedto a spareline.A significantlydifferentsituationariseswhena T1 line is usedby a customerasaccessto thepublic$witchednetworkor a$a leasedline in a privatenetwork.The explosiveuseof Tl linesin suchapplicationscreatedthe need TheESFasprovidedin theD5 maintenance featuresfor T1 customers. for enhanced features. channelbankprovidesthedesired Foremostamongthecustomerneedsis noninfiusiveperformancemonitoringof an end-to-endTl circuit. Monitoring bipolar violationsat the customerserviceunit (CSU)providesnonintrusivemonitoringof thereceivesignalbut providesno informarionregardingthequalityof thesignalattheotherendof theline (whichterminates leasedTl circuitsgenerallyinvolveintermeat theprovider'sfacilities).Furthermore, Because theinterfaces equipment. multiplexing,andcross-connect diatetransmission, removebipolarviolations,thecustomerhasno accessto of eachof theseequipments of end-to-end performance statisticsintemalto theprovider'sfacilities.Determination portion at least a of theTl requires taking performance in this environment errorrate (for is usually no spare). circuitout of service whichthere in-servicepelformancemonitoring ESF allowscu$tomefsto achieveend-to-end, by redefiningthe8-kbpsF bit of eachDSI frameto consistof a 2-kbpsframingchana checksum nel, a 4-kbpsdatalink channel,anda Z-kbpsCRC channelrepresenting theCRCchannelis carriedintactfromonecustomer overall informationbits.Because facility produceCRC locationto another,channelenorsoccurringin anyintermediate errorsat the far end. oneof whichis theability to interrogate Thedatalink supportsnumerou$$ervices, the far end,or any intermediateequipment,for performancestatistics.Thus the data link providesboththecustomerandtheserviceproviderwith anextremelyusefultool for isolatingfaulty spanlinesor equipment. are As indicatedin Table4.8 obtainedfrom reference[33], the threesubchannels "extending" 24 DS t frames. formatto encompass the D4 superframe established by Because theactualframingbits occuronly onceeveryfour DSI frames,thereare771 (FPS)00101I ' bits interveningbetweenbits carryingthe framepatternsequence

216

DrcrrAl THAN$MrssroN AND MULTIpLEXtNc

The6 CRCbits (CBI to CB6)of eachextendedsuperframe represenr a CRCcheck of all4608informationbitsin theprevioussuperframe.* Besidesprovidingend-to-end performance monitoring,theCRCvirtuallyprecludesthechances of falseframingon a databit po$ition.Eventhoughstaticuserdatacaneasilysimulatethe FPS,it is extremelyunlikelythatuserdatacanspuriouslygenerate valid CRCcodesin successive (Sixbitsof randomdatamatcha 6-bitcodewith a probabilityof I in 64.) superframes. The performanceparametersmeasuredandreportedby the 4-kbpsdatalink (DL) areframingbit errors,CRCerrors,out-of-frame(OOF)events,line code(bipolar)violations,andcontrolledslip events(describedin Chapter7). Individualevent$arereportedaswell asevent$ummaries. Thefour performance summaries reportedare: I. 2. 3. 4.

Enoredseconds(ESs)(ES= at leastoneCRC event) (BSs)(BS= 2-3lg ESs) Burstyseconds Severelyerroredseconds(SESs)(SES= >319ESsor OOFs) (FSs)(l0consecutive Failedseconds SESs)

ESF CSUs typically determine the above parameterson 15-min intervals and store them for up to 24 hr for polling by a controller [43]. The sES report conforms to ITU recommendationG.821. In addition to suppofting remote interrogation of performance statistics,the data link carriesalarm information, loopback commands,and protection switching commands. In addition to the previously mentioned features,EsF introducesa new option for per-channel signaling via the robbed signaling bits in every sixth frame. Becausean ESF is 24 frames long, there are four signaling bits in every channel in every superframe as opposedto 2 bits in sF format (Figure 4.34). whereas the two signaling bits in the sF format are designatedas A and B bits, the four bits in the ESF caseare designated A, B, C, and D. Three signaling modes are defined: z-state where all bits are A bits, 4-state where the signaling bits are ABAB, and 16-sratewhere the signaling bits are ABCD. The SF format provides the first two signaling modes but not the last.

4.7 TIMEDIVISION MULTIPLEX LOOPSANDRINGS In Chapter 2 it is mentionedthat TDM is not a$amenableto applicationswith disrributed $ourcesand sinks of traffic as is FDM. In this section a particular form of a TDM network is describedthat is quite useful in interconnectingdiskibuted nodes.The basic structureof interestis referredto as a TDM loop or TDM ring and is shown in Figure 4.36. Basically, a TDM ring is configured as a seriesof unidirectional (two-wire) links arrangedto form a closed circuit or loop. Each node of the network is implemented with two f'undamentaloperationalfeature$.First, each node acts as a regenerativerepeater merely to recover,the incoming bit stream and retransmit it. Second,the net*Calculation

ofthe CRC actually includes F bits that are setto I for purposesofCRC calculation only. Thus, charurel errors in the F bits do not create CRC errors (unless they occur in the CRC bits themselves).

MULTIPLEX LOOPS ANDHINGS 4.7 TIMEDIVISION

217

Figure 4.36 Time divisionmultiplexloop'

work nodesrecognize the TDM frame structure and communicateon the loop by removing and insefting datainto specific time slots assignedto eachnode. As indicated in Figure 4.36, a full-duplex connectioncan be establi$hedbetweenany two nodesby assigninga single time slot or channelfor a connection.One node insetts information into the assignedtime slot that propagatesaround the loop to the secondnode (all intervening nodes merely repeat the data in the particular time slot). The destination node removes data as the assignedtime slot passesby and inserts return data in the proce$s.The retum data propagatesaround the loop to the original node where it is removed and replacedby new data, and so forth. Since other time slots are not involved with the particular connection shown, they are free to be useclfor other connection$involving arbitrary pairs of nodes. Hence a TDM Ioop with C time slots per frame can suppoft C simultaneousfull-duplex connections. If, as channelsbecomeavailable,they arereassignedto different pairs of nodes,the transmissionfacilities can be highly utilized with high concentrationfactors and provide low blocking probabilities betweenall nodes.Thus a fundamental attraction of a loop network is that the transmissioncapacity can be assigneddynamically to meet changing traffic patterns.In contrast, if a star network with a centralized swirching node is usedto intercorutectthe nodeswith four-wire links, many of the links to particular nodeswould be underutilized since they cannotbe sharedas in a loop configuration. Another feature of the loop-connectednetwork is the easewith which it can be reconfigured to accommodatenew nodes in the network. A new accessnode is merely inserted into the nearestlink of the network and the new node has complete connectivity to all other nodesby way of the TDM channels.In contrast,a $tarstructurednetwork requires transmission to the central node and expansion of the centralized switching facilities.

?18

DIGITAL TRANSMI$sIoN ANDMULTIPLEXING

The ability to reassignchannelsto arbiharypairsof nodesin a TDM loop implies thattheloop is muchmorethana multiplexer.It is, in fact,a distributedtansmission andswitchingsystem.The switchingcapabilitiescomeaboutalmostasa by-product of TDM transmission. TDM loopsrepresentthe epitomeof integratedtransmission andswitching. TDM loopshavebeenusedwithin computercomplexesto providehigh capacity andhigh interconnectivity betweenprocessors, memories,andperipherals[,14].The loopstructurein thisapplicationis sometimes moreatffactivethanmoreconventional bus structuressinceall transmissionis unidirectionaland thereforeavoidstiming problemson bidirectionalbusesthatlimit theirphysicallength.Furthermore, asmore nodesareaddedto a bus,theelectricalloadingincreases, causinga limitationon the numberof nodesthat canbe connectedto a bus.Loops,on the otherhand,haveno inherentlimits of transmission lengthor numbersof nodes. The loop structureof Figure4.36 is topologicallyidenticalto the token-passing ring developedby IBM and standardizedby the IEEE asa 802.5local areanetwork. However,a token-passingring operatesdifferently than a TDM loop in that thereis only onechannel.when a nodeon a ring becomesactive,it usesthe entirecapacity of the outgoinglink until it is throughsendingits message. In contrast,a nodeon a loopusesonly specifictime slotsin theTDM $tructure, allowingothernodesto be si"connected"usingothertime slots.[n essence, multaneously a TDM loop is a distributed-circuitswitchandan 802.5ring is a distributed-packet switch. A particularlyattractiveuseof a loopwith high-bandwidth linksis shownin Figure 4.37. This figure illustratesthe use of add-dropmultiplexers(ADMs) that access whateverbandwidthis neededat a local nodebut passthereston to othernodes.In fypicalapplications theamountof bandwidthallocatedto eachnodeis quasi-static: It is changedonly in respon$e to macroscopic changesin traffic patterns,possiblyasa functionof thetimeof day.Thisbasicoperationis genera-lly referredto asa cross-con-

Figure 4.37 Functionalmesh,fiber loop,andADMs.

REFERENcES 219

Figure 4.3E Useof reverseloop to circumventlink failuresin TDM loops. nect function as opposedto a swirching function, which involves call-by-call reconfigurations. Both types of switching operationsare discussedin the next chapter.An important point to note about Figure 4.37 is the ability to utilize a general-purpose physical topology but define an arbitrary functional topology on top of it. One obvious limitation of a loop is its vulnerability to failures of any link or node. The effect of a node failure can be minimized by having bypasscapabilities included in each node. When bypassed,a node becomesmerely a regenerativerepeater,as on T-carrier transmissionlinks. Link failures can be circumventedby providing alternate facilities. Figure 4.38 showsone particular structureusing a second,reverse-direcflon loop to provide backup capabilitiesin the caseof failures. When fully operational,the network can use the reverse loop as a separate, independent network for traffic as needed.Whenever a failure occurs, the nodes adjacent to the break establish a new loop by connectingthe forward path to the reversepath at both places.Hence all nodes continue to have full connectivity to any node on the new loop' A particular example of the useof the dual reverseloop for both protection and distributed queuedaccessto the channelsis the distributedqueueddual-bus(DQDB) [45] system developedby QPSX in Aushalia and standardizedby the IEEE as an 802.6 metropolitan area network. Further examples of loop structures and applications are provided in Chapter 8 where SOI.IET rings are discussed.

REFERENCE$ "A FastAutomatlcEqualizerfor DataLinks,"Pftillps F, deJagerandM. Christiaens, TechnicalRevielu, Vol. 36,1977,pp. 10-24. 2 K. Azadet and C. J, Nicole, "Low-PowerEqualizerArchitecturesfor High-Speed Magaeine, October,1998,pp' I I 8- 126. Modems,"IEEE Communicatinns 3 F. D. Waldhauer,"A Z-Level, 274 Mbls RegenerativeRepeaterfor T4M"' IEEE Conference, I 975,pp. 48-13-48- t 7. Intemational Communications "An Investigationof the Count RatePerformanceof T. V. Blalock, 4 N. Karlovac and I

Washington, Symposium, paperpresented at theNuclearSciences BaselineRestorers," DC.1974.

22O 5 6 7

I

9 l0 ll 12 13 14 l5 16

DtctrAl TRANSMtsstoN ANDMULIpLEXtNG F. D. Waldhauer, "Quantized Feedbackin an Experimental 280-Mb/s Digital Repeater for Coaxial Transmission," IEEE Trttnsacfionson Communications,Jan, 1974,pp, l-5. J. Salz, "Optimum Mean Square Decision Feedback Equalization," Bell System TechnicalJoumal, Oct. 1973,pp. l34l- I 373. J. M. Cioffi, W. L. Abbott, H. K. Thapar, C. M. Melas, and K. D, Fisher, "Adaptive Equalization in Magnetic-Disk Storage Channels," IEEE CommunicationsMagazine, Feb.1990,pp.15-29. S. Sugimoto, K. Hayashi, and F. Mano, "Design of ZBIQ Transceiver for ISDN SubscriberLoops," IEEE International Conferenceon Communications,Itrne 1989,pp. 8.1.1-8.1.5. M. R. Aaron, "PCM Transmission in the Exchange Plant," Bell SystemTethniml Joumal, Jan. 1962,pp. 99*141. Technical Staff, Bell Telephone Laboratories, Transmission Systems .for Communiuttion,r, Westem Electric Co., Winston Salem, NC, l97l , p. 668. "1.544 Mhps Digital Service," Bell System Technical Reference Publication No. 4l45l,May 1977. G. D, Forney, '"The Viterbi Algorithm," Praceedingsof IEEE, Mar. 1973,pp.268-2i8. V. l. Johannes,A. G. Kaim, and T. Walzman, "Bipolar Pulse Transmission with Zero Extraction," IEEE Transactions on Communications,Apr. 1969,pp. 303-310. "The D3 Channel Bank Compatibility Specification-lssue 3," Technica.lAdvisory No. 32, American Telephone and Telegraph Company, Basking Ridge, NJ, Oct. lgj7. B. Johnstonand W. Johnston,"LD-4 A Digital Pioneer in Action," Telesis,Vol. 5, No. 3, June 1977,pp.66-72. J. H. Davis, 'T2: A 6.3 Mb/s Digital Repeatered Line," IEEE Inrernational ComrnunicationsConferenu, I 969, pp. 34-9-34- 16.

17 RecommendationG.703, CCITT Orange Booft, Vol. 3, No. 2. 18 R. M. Wienski, "Evolution to ISDN within the Bell Operating Companies," IEEE CommunicationsMagafine, Jan. 1984, pp. 33-41. 19 American National Standard: Digital Hierarchy-Formats Specifications, ANSI Tl.l07-1988, Editorial RevisedApr. 1989. 20 J. M. Sipress,"A New Class of SelectedTernary Pulse Transmission Plans for Digital Transmission Lines," IEEE Transactions on Communitation Technology, Sept. 1965, pp. 366-372. 2l E. E. Schnegelbergerand P. T. Griffiths, "48 PCM Channelson Tl Facilities," National Elettronics Conference, 1975, pp. 20 1-205, 22 P. A. Franaszek, "Sequence-StateCoding for Digital Transmission," Bell System Technical Journal, Dec, 1967, pp. 143-157. 23 J. O. Azaret, J. F. Graczyk, W. M. Hauser, and J, V. Mahala, "New Improved T-Canier 24 25 26

Doubles Capacity, Cuts Costs," BelI l-abs Rewrd, July 1985, pp. 26-31. J, W. Lechleider, "Line Codes for Digital Subscriber Lines," IEEE Communit:ations Magazine,Sept. 1989,pp.25*31. A. Lender, -'The Duobinary Technique for High Speed Data Transmission," IEEE Transactionson CommunicationElectronits, May 1963,pp.2l4-218. A. t ender, "Correlative Level Coding for Binary Data Transmission," IEEE Spectrum, Feb. 1960,pp. I 04- I I 0,

PFOBLEMS 221 "Partial ResponseSignaling,"IEEE Transactionson 21 P. Kabal and S. Pasupathy, Sept.1975,pp. 921-934. Communications, "Multilevel Partial ResponseSignaling," 28 A. M, Genish and R. D. Howson, 1967,p' I 86. Conference, InternationalCommunicafions "DuobinaryPCM SystemDoublesCapacityof Tl 29 D, W. JurlingandA. L, Pachynski, l9TT'pp.32.2-297-32-2-301. Facilities,"IntemationalCommunicationsConferente, 30 T. Seaver,"An Efficient 96 PCM ChannelModemfor 2 GHz FM Radio,"National 1978, pp. 38.4.t -38'4'5' Conference, Telecommunications "special Joumal'Mar. 1984. BeII SystemTechnical Issue:IARDS," 3l "Observations of Errors and Error Rateson Tl Digital Repeatered 32 M. B. Brilliant, 33 34 35 36 37 38 39 40 4l 42

43

Joumal,Mar. 1978,pp' 7ll*746. Lines,"Bell SystemTechnical ..Carrier-to-Customer krstallation-DSl Metallic Interface," ANSI T1.403-1989, Institute. AmericanNationalStandards N. S,Bergano,F. W. Kerfoot,andC. R. Davidson,|EEE PhotonhsTechnologyLetters' Vol, 5, 1993,pp. 304-306. S. B. Wicker, Error Control Sysferns for Digital Communicationand Storage, EnglewoodCliffs, NJ, 1995. Prentice-Hall, and Applications,Artech L. H. CharlesI*e, ConvolutionalCoding: Fundamentals House.Norwood.MA. 1997. and Operationsin theBelI Eng,ineering TechnicalStaff,Bell TelephoneLaboratories, 1978. Electric Co., Indianapolis, System, Western "RobustFrameSynchronization for Noisy D. E. Dodds,L, R, Button,andS.-M.Pan, May 1985,pp. 465-469. on Communicatians, PCM Systems,"IEEE Transactions "StatisticalDistributionof PCM Framing D. E. Dodds,S.-M.Pan,andA. G. Wacker, Nov. 1988,pp' 1236-'1241. on Communications, Times,"IEEE Transactions "Frame SynchronizationTechniques,*IEEE Transacfionson R. A. Scholtz, Aug. 1980,pp' 1204-1212Communications, "On FrameSynchronization of D. T. R. Munhoz.J. R. B, deMarca,andD. S, Arantes, Aug. 1980'pp' 1213-1218. on Communicarions, PCM Systems,*IEEE Transactions "Apparatusand Methodsfor RecoveringAlignment M. Andersonand O. Petruclrka, Sept' from a Non-IdeallyEncodedMulti-bit Digital Signal,"U.S.PatentNo.4,955'037' 4. 1990. K. Stauffer and A. Brajkovic, "DS-l ExtendedsuperframeFormat and Related Magazine,Apt. 1989,pp' 19-23' Performance Issues,"IEEE Communications

GeneralDescription,Radio company ReportNo, 523-0561697-20173R, C-System CollinsRadioCompany,Dallas,TX, May I, 1970. Feb. 45 "New hoposal Extendsthe Reachof Metro Area Nets,"Data Communicuriorls, 1988,pp, 139-145. 44

PHOBLEMS 4.1 If the transmitter and receiver of an asynchronoustransmission system utilize clock sourcesthat are accurateto one part in 103,determinethe maximum number of bits in a codeword if the maximum sampletiming enor is 207oof a pulse

222

DIGITAL TRANSMISSION ANDMULTIPLEXING

interval. Assume the sampleclock is eight times the bit rate and that the stalt bit is determinedby counting four sampletimes from the beginning of the start bit. 4.2 Using the symbols +, 0, and - to representa positive pulse, no pulse, and a negative pulse, respectively, determinethe following line code $equencesof the binary data sequence

0l l0 r0000t0001100000001 0 (a) Bipolarwith mostrecentpulsebeingpositive (b) Bipolarwith mostrecentpulsebeingnegative (c) Pairselectedtemarybeginningin thenegativemode (d) B3ZSwith a +0+ substitutionhavingjust beenmade (e) 8625 with themostrecentpulsebeingpositive 4.3 Assumethat two identicalcablesystemsareusedfor digital transmission with equalpulseamplitudes. Onesystemusesconventional bipolarsignalingandthe otherusesPST.comparethecrosstalklevelsof thetwo $ystems. (Assumethat l's and0's areequallyprobable.) 4.4 A digitaltransmission systemoperatingat anerrorrateof 10-6is to haveits data rateincreased by 50Vo.Assumingthatthe sametransmittedpoweris to be used in the secondsystem,whatis thenewerrorrate? 4.5 Whatis theaveragereframetime of a D3 channelbank(usingSFframing)from a randomstartingpoint?What is the maximumaveragereframetime of a D3 chalnelbank?(AssumeI's and0's in theme$sage traffic areequallylikely.) 4.6 Repeathoblem 4.5 for theprimaryTDM multiplexsignalspecifiedby CCITT. 4,7 A TDM systemoperatingat 2 Mbps is to havea maximumaveragereframe time of 20 msec.What is themaximumpossibleframelengthif framingis establishedwith a bit-by-bitframesearch?(Assumethat I's and0's in message channelsareequallylikely.) 4.8 A Tl transmission systemusinga Dl frameformatis to haveanaveragereframe time (from a randomstartingpoint) of l0 msec.How largea block of bit positionsmustbe examinedin parallelto achievethedesiredresult? 4.9 what is theexpected framingtimefor a Tl line (D3 frameformar)if theframing strategyis bit by bit andthedatasheamhas60% I's and40VoO's? 4.10 What is the averagepulsedensityof 4B3T coding(AssumeI's and0's are equallylikely.) 4.11 A TDM transmission link using4B3T codingcantransmit32 64-kbpsvoice channelsusingthe samesymbolrateas a Tl line (1544kbps).Assuminga fixed Gaussiannoise environment,how much must the averagetransmit powerof the 4B3T systembe increased to providethe equivalenterrorrate of a bipolarcode? 4.12 Assumethatcrosstalkinterference in a multipaircablesystemproducesaneffect equivalentto Gaussian noiseat anequalpowerlevel.Usinganerrorrateof 10-6 asa designobjective,determinetheeffectivedegradation of thecrosstalkon binary$olar) NRZ codingundereachof thefollowingconditions.(Theeffective

PROBLEMS

2?3

astheincreased transmitpower,in decibels,required is determined degradation rate.) to achievethe desirederror (a) Thecrosstalklevel is I 6 dB belowthe averagesignallevel,but thecrosstalk is to be overcomeon only onepair (i.e.,all otherpairs$tayat a powerlevel for 10-6BER with no crosstalk). (b) Thecrosstalklevelis 16dB belowtheaveragesignallevel,but theeffectsof the crosstalkare to be overcomeon all pairs' (Flfnt:Use signal-power-tonoise-power ratios,not E/N^0.) 4.13 RepeatProblem4.12for bipolarcoding. of a po4-14 How muchdoes-1 8 dB of crosstalkdegradetheerrorrateperformance areat (Assume transmitters that all Iar binaryNRZ signalfor a BER of l0-7? per 15dB decadein frequency,what equalpowerlevels.)If crosstalkincreases is the relativecrosstalklevel of a four-levelNRZ codecarryingthe samedata penaltyof thefour-levelsystemcompared rate?Whatis theoverallperformance to thetwo-levelsystem? from temarysym4.15 How manydistinctcodewordsof length4 canbe constructed bols?How manyof thesecodewordscontainan equalnumberof positiveand canbeusedto ensurea minimumof one negativepulses?How manycodewords timing pulseoccursin eachcodeword? to encodebinarydata 4.16 Cana setof ternarycodewordsof length8 be constructed numbersof posicontaining equal per pulses 8-bit word and four usingexactly pulses? tive andnegative +1, -3, +1, -1, +3, +3, -3 of signallevels,determine 4.17 Giventheinput$equence of outputsignallevelsfor eachof thefollowingcorrelativeencodthesequence ings. (a)l+Dencoder (b)1-Dencoder (c) I -.d encoder 4.18 Whatis theprobabilityof a CRCerrorin a DSI signalwith ESFframingif the randomBER is 10-7? burstof errorsin an 4.19 Whatis theminimumandmaximumlengthof a correctable RS(204,188)codewhereeachsymbolis an 8-bitbyte? 4.20 Determinetheprobabilityof failureof anRS(7,2) codeoperatingin a randomerrorenvironmentwith a bit errorprobabilityof lQa. Assumeeachsymbolis an 8-bitbyte. 4.21 Determinethe probabilityof failure of an RS(31,15) codeoperatingin a ranwith abit errorprobabilityof l0-3.Assumeeachsymbol dom-errorenvironment is an 8-bit byte. 4.22 Determinethe parity error patternthat resultsin a bit-by-bit constraintlength convolutionaldecoderif two data(B) bitsin a row arecomrptedanddetectedin for errorbut no otherreceivedbits arein error.Usethe following bit sequence whereP0is thefirst receiveddatabit andB 1and82 arein enor.POB0, reference, P181,P282, P383.

DIGITALSWITCHING '"'

''

t'

transmisrion*lnetwork(terminals, in a communications Of thethreebasicelements themost dia, andswitches),switchesarethe mostinvisibleto theusersyet represent impoftantelementsin termsof availableserviceofferings.As mentionedin Chapter controlswitching in 1965whenstored-program wasestablished l, a majorr.nilestone control wasfirst introducedinto the U.S. publictelephonenetwork.Stored-program providesthe meansfor implementingmanyinnovativeuserservicesandfor greatly simplifyingswitchadminishationandmaintenance. The useof computersto controlthe switchingfunctionsof a centraloffice led to the designation"electronic"switching[e.g.,electronicswitchingsy$tem(ESS)or electronicautomaticexchange(EAX)]. However,the switchingmatricesof these in nature.Thefirst electronicswitchesareactuallyelectromechanical first-generation useof electronicswitchingmatriceso.cqrnedin Francein 1971,whendigitalswitching wasappliedto an endoffice envfriinment.Ironically,thesefirst digital switches control.Digitalelectronicswitchingmatriceswerefirst indid notusestored-program troducedinto the U.S. public networkin 1976with AT&T's No. a EsS-digitaltoll switch. In thelate 1970sdigitalclass5 switchesbeganto be installedin the UnitedStates, switches.At thattime for step-by-step mo$tlyin smallerend,officesasreplacements controlfrom switchingofficesalreadyhad stored-program mostmajor_ mettoi_oJif4n Because the digitaltoll andendoffice switching No. I ESSor No. I EAX machines. environments, analogtransmission machineswereinitially installedin predominantly The motivation their digitalmatrix providedno directbenefitto networkcustomers. lowermain-for the digitalmachineswa$reducedcostsfor the operatingcompanies: costs tenance,reducedfloor space,simplifiedexpansion,and lower manufabturing

tu.

By the mid-1980s the interoffice hansmission environment had changedto be almost exclusively digital. Thus analog-to-digital conversion costs moved from being associatedwith digital transmission links to being associatedwith analog switches, thereby further sealing the fate of analog toll or tandem switching technology. At this time frame, end office ffansmissionenvironmentsalso begantJqeswing to a digital environment. Interoffice trunks were alreadydigital, digital loop barrier systemsbecame

226

DIGITALSWITCHING

cost effective in metropolitan applications, remotely conmolled switching modules with digital fiber interconnectbecamecommon for service to outlying communities, digital cross-connectsystem$(DCSs) were being deployed, and the feeder portion of the subscriberloop plant beganto use fiber. The cost penaltiesof interfacing to a digital transmissionenvironment and higher maintenancecosts led to the older electromechanicalclass 5 switches being replaced by digital machineswhen expansionor consolidationof an office occurued.The ability to offer ISDN services was a lessermotivation for changing to digital switches becauselow-cost fiber transmissionin conjunction with DCS sy$temsallows provisioning of these servicesfrom other offices. This chapterdescribesthe basic operation and implementation of digital tifiie division switching as applied to PBXs, end offices, toll switche$,and crossconnects.Be' fbre digital switching is discussed,certain basic switching conceptsand terminology are introduced.

5.1 SWITCHINGFUNCTIONS Obviously, the basic function of any switch is to set up and releaseconnectionsbetween transmissionchannelson an "as-neededbasis," The structureand operation of a switch dependon particular applications.Three switching categoriesfor voice circuits are local (line-to-line) switching, transit (tandem) switching, and call distribution. The most common switching function involves direct connectionsbetween subscriber loops at an end office or between station loops at a pBX. These connections inherently require setting up a path through the switch fiom the originating loop to a specific terminating loop. Each loop must be accessibleto every other loop. This level of switching is sometimesreferred to as line switching. Transit connectionsrequire sening up a path from a specific incoming (originating) line to an outgoing line or trunk group. Normally, more than one outgoing circuit is acceptable.For example, a connection to an interoffice trunk group can use any one of the channelsin the group. Hence transit switching sffucturescan be simplified becausealternatives exist as to which outgoing line is selected.Furthermore, it is not even neces$arythat every outgoing line be accessiblefrom every incoming line. Transit switching functions are required by all switching machines in the telephone network. Some machines such as remote concentrators and toll or tandem switches serviceonly hansit traffic (e.g., do not provide local connections).Theseconceptsare illustratedin Figure 5. l. call distributors are often implemented with the samebasic equipment as pBXs. The mode of operation (software) is significantly different, however, in that incoming calls can be routed to any available attendant.Normally, the software of an automatic call distributor (ACD) is designedto evenly distribute the arriving calls among rhe artendants.Although it is not an inherent requirement that every incoming line (trunk) be connectableto every attendant,call distributors are normally designedto provide accessibility to all attendants.Furthermore, it is often desirablethat nonblocking op-

l\

5.2 SPACFDIVISIONSWITCHING

227

Figure 5.1 Local andtransitftaffic switchingexamples. erations be provided. (No matter what switch paths are in use, a new requestcan be servicedby any available attendant.)

5.2 SPACE DIVISIONSWITCHING as arrayof crosspoints, thesimplestswitchingstructureis a rectangular Conceptually, of Ninlets any one to connect matrix can be used This switching shownin Figure5.2. to two-wirecircuits,only to anyoneof M outlets.If theinletsandoutletsateconnected is required'* per connection onecrosspoint ale designedto provideintergroup(transit)connecRectangular crosspointEIIrays group an inlet to anoutletgroup.Applicationsfor thistypeof tionsonly,thatis, from in the following: an operationoccur 1. Remoteconcentrators 2. Call distributors 3. Portionof a PBX or endoffice switch that provides transit switching switches 4. Singlestagesin multiple-stage thattheinletsbe connectable it is not necessary In mostof theforegoingapplications, savingsin outlets, considerable groups of involving large to everyoutlet.In situations of outlimited number access only a if each inlet can be achieved can totalcrosspoints "limited By overlapis said to exist. availability" occurs, lets.Whensucha situation "grading" pingtheavailableoutletgroupsfor variousinlet groups,a techniquecalled An exampleof a gradedswitchingmatrixis shownin Figure5.3.Notice is established. *ln

fact, two (and sometimes thrce) switching contacts are associated with each ctosspoint of a two-wire switoh, Since these contacts are pa-rtof a single unit and opetate in unison, they are considered a single crosspoint.

228

DIGITALSWITCHING

,y Ourlerr Figure 5.2 Rectangular crosspointalray. that if outlet connectionsare judiciously chosen,the adverseeffect of limited availability is minimized. For example, if inlets I and I in Figure 5.3 requesta connection to the outlet group, outlets I and 3 should be choseninsteadofoutlets I and 4 to avoid future blocking for inlet 2. Graded switching structure$were often used for accessto large trunk groups in electromechanicalswitches where crosspointrrwere expensiveand individual switching modules were limited in size. Gradings were also used in individual switching stagesof large multiple-stageswitcheswhere more than one path to any particular outlet exists. Becausevery large digital matrices can be implemented with full accessibility, graded switch structuresare no longer necessaly. Intragroup switching, as in line-to'line switching, requireseach line to be connectable to every other line. Thus full availability from all inlets ro all outlets of rhe switching matrix is required.Figure 5.4 showstwo matrix structuresthat can be usedto fully interconnect two-wire lines. The dashedlines indicate that coffesponding inlets and outlets of two-wire switching matricesare actually connectedtogetherto provide bidirectional transmissionon two-wire circuits. For purposesof describingswitching matrices, however, it is convenient to consider the inlets and ouflets of two-wire switching matrices as being distinct.

I

2 3

{ lnlCli

_ E

6 I I

r

2

3

{

Outlctr Figure 5.3

Graded rectangular switching matrix.

5.2 SPACEDIVISIONSWITCHING

?,29

Inlrtil out|.t

(t)

ft)

(folded)' (b)triangular (a)square; matrices: Figure5.4 Two-wireswirching by selectinga in Figure5.4allow anyconnectionto beestablished Both structures However,thesquaremahix,whichis alsocalledatwo-sidedmatrix, singlecrosspoint. in two ways'For example,if input allowsanyparticularconnectionto be established link i is to beconnectedto input linkj, the selectedcrosspointcanbe at theintersection of inlet i andoutletj-or at the intersectionof inlet j andoutlet i. For simplicity these In a typicalimplementation, arereferredto as(f,i) and(J, i), respectively. crosspoints service,andcrosspoint(j' i) is usedwhen crosspoint(i,7) is usedwheninputi requests service. inputj reque$ts In the triangularmatrix of Figure5.4 theredundantcrosspointsareeliminated.The however.Beforesetting crosspointreductiondoesnot comewithoutcomplications, the switchcontrolelement inputi, input i and swirch switch between up a connection (i,i) is selected'If i is larger, crosspoint orj. If i is is larger: d which mustdetermine swirching,the computer*controlled (.7, With must be selected. i) smaller,crosspoint conholled electromechanically In the older, is trivial. comparison line number is more significant. control of the swirch complexity the added switches,however,

I

2

Inlorr i

i N

t

2

3

'

i

Outlrtt

Flgure 5.5 Four-wire switchingmatrix'

230

DIGITALSWITCHING

Switching machinesfor four-wire circuits require separateconnection$for the go and return branchesof a circuit. Thus two separateconnection$must be established for each service request.Figure 5.5 depicts a square-matrixstructureused to provide both connections.The structure is identical to the squarematrix shown in Figure 5.4 for two-wire switching. The difference,however, is that correspondinginlets and outlets are not connectedto a common two-wire input. All of the inlets of the four-wire switch are connectedto the wire pair carrying the incoming direction of transmission, and all of the outlets are connectedto the outgoing pairs. When setting up a connection betweenfour-wire circuits i andj, the matrix in Figure 5.5 must selectboth crosspoints (i, j) and (j, f). In actualoperationthesetwo crosspointsmay be selectedin unison and implemented as a common module.

5.2.1 Multlple.$tageSwltching In theswitching $tructures described to thispoint,aninletis connected directlyto an outlet through a single crosspoint. (Four-wire switches use two crosspointsper connection, but only one for an inlet-to-outlet corurection.)For this reason,theseswitching structuresare referred to as "single-stage" switches. Single-stageswitches have the property that each individual crosspointcan only be used to interconnectone particular inlet-outlet pair. since the number of inlet-outlet pairs is equal to N(N - lyz for a triangular array, and N(N - l) for a $quare array, the number of crosspoints required for a large switch is prohibitive. Furthermore,the large number of crosspoints on eachinlet and outlet line imply a large amount of capacitiveloading on the message paths. Another fundamental defrciency of $ingle-stageswitches is that one specific crosspoint is neededfor each specific connection. Ifthat crosspoint fails, the associated connection cannot be e$tablished.(An exception is the square,two-wire switch that has a redundant crosspoint for each potential connection. Before the redundant crosspointcould be usedas an alternatepath, however, the inlet-oriented selectionalgorithm would have to be modified to admit outlet-oriented selection.) Analysis of a large single-stageswitch revealsthat the crosspointsare very inefficiently utilized. Only one crosspointin eachrow or column of a squareswitch is ever in use, even if all lines are active. To increasethe utilization efficiency of the crosspoints, and thereby reduce the total number, it is necessarythat any par-ticularcrosspoint be usablefor more than one potential connection.If crosspointsare to be shared, however, it is also necessarythat more than one path be available for any potential connection so that blocking does not occur. The alternatepaths serve to eliminate or reduce blocking and also to provide protection againstfailures. The sharing ofcrosspoints for potential paths through the switch is accomplished by multiple-stage switching. A block diagram of one particular form of a multiple-stageswitch is shown in Figure 5.6. The switch of Figure 5.6 is a three-stageswitch in which the inlets and outlets are partitioned into subgroupsofNinlets and Noutlets each.The inlets ofeach subgroup are serviced by a rectangular array of crosspoints.The inlet arrays (first stage) are n x ft arays, where each of the ft outputs is connectedto one of the ft cenrer*smgear-

SWITCHING 5.2 SPACEDIVISION

231

M rt!il

maffix. switching Figure5.6 Three-stage areoftencalledjunctors.Thethird stageconsistsof rays.Theinterstageconnections provide corxrectionsfrom eachcenter-stagearrayto the artays that ft x n rectangular groupsof n outlets,All center-stagearrayswe N/n x N/n arraysthat provideconnections from any first-stagealray to any third-stagearray.Notice that if all affaysprovide full availability,thereareft possiblepathsthroughthe switchfor anyparticular centerconnectionbetweeninletsandoutlets.Eachof the ft pathsutilizesa separate paths through the provides altemate structure Thus the multiple-stage stagearray. to link is connected each switching since Furthermore, failures. swit"6 to circumvent minimized' loading is capacitive crosspoints, a limited numberof The total numberof crosspointsN1Erequiredby a three-stageswitch, as shownin Figure5.6,is

Nx=2Nft.-[#l

(s.1)

whereN = numberof inlets-outlets n= sizeofeachinlet-outletgroup ft = numberof center-stagearrays definedin Equation5.I can $hortly,thenumberof crosspoints As is demonstrated be significantlylower than the number of crosspointsrequiredfor single-stage arraysareneeded how manycenter-stage First.however.we mustdetermine switches. to provide satisfactoryservice.

232

DIGITAL SWITCHING

Nonblocklng Switches Oneattractivefeatureof a single-stage switchis thatit is strictlynonblocking.If the calledpartyis idle, thedesiredconnectioncanalwaysbe established by selectingthe particularcrosspointdedicatedto the particularinput*outputpair.Whencrosspoints are shared,however,the possibilityof blockingarises.In 1953charlesclos [2] of Bell Laboratoriespublishedan analysisof three-stage switchingnetworksshowing how manycenter-$tage arraysarerequiredto providea strictly nonblockingoperation. His resultdemonstrated thatif eachindividualarrayis nonblocking,andif thenumber of centerstagesft is equalto 2n -I, theswitchis strictlynonblocking. Theconditionfor a nonblockingoperationcanbe derivedby first observingthata connectionthroughthe three-stageswitchrequireslocatinga center-stagearraywith ar idle link from the appropriatefirsr $tageand an idle link ro the appropriatethird stage.sincetheindividualarays themselves arenonblocking,thedesiredpathcanbe rtetup any time a centerstagewith the appropriateidle links canbe located.A key point in the derivationis to observethat sinceeachfirst-stagearrayhasn inlets,only n - I of theseinletscanbebusywhentheinletcorresponding to thedesiredconnection is idle. If ft is geater thann - l, it follows that,at most,n - I links to cenrer-srage irrray$canbe busy.Similarly,at mostn - I links to theappropriate third-stagearray canbe busyif theoutletof the desiredconnectionis idle. The worst-casesituationfor blockingoccurs(asshownin Figure5.7) it alln - r busylinksfrom thefirst-stagearrayleadto onesetofcenter-stage arays andifall n I busylinks to the desiredthfud-stageErrraycomefrom a separatesetof center-stage arrays.Thusthesetwo setsofcenter-stnge arraysareunavailablefor thedesiredconnection.However,if onemorecenter-stage arrayexists,theappropriate inputandoutputlinksmustbeidle,andthatcenterstagecanbeusedto setup theconnection. Hence if k=(n * 1) + (n - l) + 1 =2n- 1,theswitchissrictlynonblocking. substituting

t l l r L--J ArnlhblG pffh

Flgure 5.7 Nonblockingthree-srage switchingmatrix.

SWITCHING ?33 5.2 SPACEDIVISION

this value of t into Equation 5.1 reveals that for a strictly nonblocking operation of a three-staseswitch

Nx=2N(2n-

(5.21

t)+(2n-t,[rOI

in a nonblockingthreein Equation5.2,the numberof crosspoints As expressed into subgroups partitioned are and outlets on how theinlets stageswiichis dependent resultingexsetting the to n and with respect of iize n. DifferentiatingEquation5.? (for the optimum large M) that reveal pressionequalto 0 to determinethe minimum an provides 5.2 then into Equation value of n valueof n is (N/z)Ltz.Substitutingthis three-stage of a nonblocking crosspoints expressionfor the minimum numberof switch:

N*(min)=4N({2N - l)

(s.3)

whereN = total numberof inlets-outlets. nonblockingthreeTable5.1 providesa tabulationof Nx(min) for various-sized in a single-stage of crosspoints the number thevaluesto stageswitcheiandcompares capabilities,a provide four-wire inherently *qu*" matrix.Both switchingstructure$ circuits. four-wire implies digitization voice for digital switchesbecause requirement resignificant provides matrix switching As indicatedin Table5.l, a three-stage of crossnumber the However, particularlyfor largeswitches. ductionsin crosspoints, switchesis still quiteprohibitive.Largeswitchestypically pointsfor largethree-stage For example, to providegreaterreductionsin crosspoints' o*" -o.* thanthreeStages up to 65,000lines' can service switchingmatrixthat theNo. 1 ESSusesaneight-stage muchfrom so not are achieved Themostsignificantreductionsin crosspointnumbers probabilities low acceptably additionalsiagesbut by allowingthe switchto introduce of blocking. Swltches of Nonblocklng Requlrsments TABLE5.1 Croeepoint Numberof Lines 128 51? 2,048 8,192 32,768 131,072

fot Numberof Crosspoints Switch Three-Stage

for Numberof CrossPoints Switch Single*Stage

7,680 63,488 516,096 4,2 million 33 million ?68 million

16,2s6 261,632 4.2 million 67 million 1 billion 17 billion

234

DIGITAL SWIT0HING

5.2.2 Blocklng Probablilties: Lee Graphs Strictly nonblockingswitchesarerarely neededin most voice telephonenetworks. Both the switchingsystemsandthenumberof circuitsin interofficetrunkgroupsare sizedto servicemostrequestsastheyoccur,but economicsdictatesthatnetworkimplementations havelimitedcapacities thatareoccasionally exceeded duringpeak-traffic situations.Equipmentfor the public telephonenetworkis designedto providea cerlainmaximumprobabilityof blockingfor thebusiesthourof theday.Thevalueof this blockingprobabilityis oneaspectof the telephonecompany'sgradeof service. (other aspectsof gradesof serviceareavailability,tran$mission quality,anddelayin 'tettingup a call.) A typical residentialtelephoneis busy 5,l0va of rhe time duringthe busyhour. Businesstelephones areoftenbusyfor a largerpercentage of theirbusyhour (which may not coincidewith a residentialbusyhour).In eithercase,network-blockingoccurrences on the orderof l7o*duringthebusyhour do not represent a significantreductionin the ability to communicatesincethe calledparty is muchmorelikely to havebeenbusy anyway.under thesecircumstances, end office switchesand,to a lesserdegree,PBXscanbe designedwith significantreductionsin crosspoints by allowing acceptable blockingprobabilities. Therearea varietyof techniques that canbe usedto evaluatethe blockingprobability of a switchingmatrix.Thesetechniquesvary accordingto complexity-, accuracy,andapplicabilityto differentnetworkstructures. Oneof the mostversatileand conceptuallystraightforwardapproachesof calculatingblocking probabilitiesinvolvestheuseof probabilitygraphsasproposedby c. y. Lee [3]. Althoughthistechniquerequire$severalsimpliffing approximations, it canprovidereasonably accurate results,particularlywhencomparisons of altematestructures aremoreimportantthan absolutenumbers.The greatestvalueof this approachlies in theeaseof formulation andthefactthattheformulasdirectlyrelateto theunderlyingnetworkstructures. Thus theformulationshelpprovideinsightinto thenetwork$tructures andhow thesestructuresmight be modified to changethe performance. In thefollowing analyses we aredeterminingtheblockingprobabilitiesof various switchingrrtructure$ usingutilizationpercentages, or ,,loadings,"of individuallinks. Thenotationpwill beused,in general,to represent thefractionof timethataparticular link is in use(i.e.,p is the probabilitythata link is busy).In additionro a utilizarion percentage or loading,p is also sometimesreferredto as an occupancy.The probability thata link is idle is denotedbyq = | - p. when any oneof n parallellinks canbe usedto completea connection,the com* positeblockingprobabilityB is theprobabilityrhatall links arebusyt; B=pn

'Transmission

(5.4)

and switching equipment in the public network is normally designedfor even lower blocking provide for growth in the traffrc volume. to .probabilities 'Equations 5.4 and 5.5 assumeeach link is busy or idle independently of other links.

swlrcHlNc 205 DlvlsloN s.z $PAoE Whena seriesof n links areall neededto completea connection'theblockingprobas I minustheprobabilitythattheyareall available: ability is mosteasilydetermined B=l-{

(5.s1

networkis shownin Figure5.8.This graphreA probabilitygraphof a three-stage pafiiculaf connectioncanbe establishedwith ft different pathsl latesthe fact that any alray.Theprobabilitythat anypalticular interstagelink onetlrough eachcenter-stage networkcanbe is busyis denotedby p'. Theprobabilityof blockingfor a three-stage determinedas ff = = = =

probability that all pathsarebusy (probabilitythat an arbitrarypathis busy)e (probabilitythat at leastonelink in a pathis busy)ft (l-qa)k

(5.6)

whereft = numberof center-stagealrays link is idle, - | - P'' q'= probabilitythat aninterstage If the probabilityp that an inlet is busy is known, the probabilityp' that an interstagelink is busycanbe determinedas

p,=fr tr
(s.7)

thefactthatwhensomenumberof inlets(or outwherep = Mn.Equation5.7presents Iets)arebusy,the samenumberof first-stageoutputs(or third-stageinputs)arealso busy.However,thereare F = Vn times asmanyinterstagelinks asthereareinlets or of intersggelinks thatarebusyis reducedby thefactorp' outlets.Hencethepercentage The factor p is defined as thoughft > n, which implies that the first stageof the switchis providing$paceexpansion(i.e.,switchingsomenumberof input links to a largernumberof outputlinks).Actually,p may be lessthan 1, implyingthatthefirst

P ' i e

Figure 5.8

Probability graph of three-stage network'

236

DtctrAL swtrcHtNc

stageis concenffating theincomingtraffic.First-stage concentration hasbeenusedin endoffice or largePBX switcheswheretheinletsarelightly loaded(S_l}vo).In tan_ demor toll offices,however,theincomingtrunksareheavilyutilized,andexpansion is usuallyneededto provideadequately low-blockingprobabilities. substitutingEquation5.7into Equation5.6providesa completeexpression for the blockingprobabilityof a three-stage switchin termsof the inlet utilizationp:

'=L'['-fii] I

r

rz'1ft

(s.8)

Table 5.2 tabulatesnumbersof crosspointsobtainedfrom Equation5.g for the sameswitchsizespresented in Table5.1.Thenumberof centerarrayswaschosenin eachcaseto providea blockingprobabilityon theorderof 0.002.Theinlet urilization in eachexamplewasassumed to be 107o.Noticethatthedesignswith smallbut finite blockingprobabilitiesaresignificantlymorecosteffectivethannonblockingdesigns. Theswirchdesignsin Table5.2assume thattheinletsareonly l0zobusy,asmight be thecasefor anendoffice switchor a PBX. Thedramaticsavingsin crosspoints for largeswitchesis achievedby introducingsignificantconcentration factors[yp; into the middle stage.when the inlet utilizationis higher(astypicallyoccursin tandem switches),high concentration factorsarenot acceptable, andthe crosspointrequirementsthereforeincrease.Table5.3 lists corresponding crosspointrequirements and implementationparaffreters for inlet loadingsof 'l\Vo. The resultspresented in Tables5.2 and5.3 indicatethat very largeswitchesstill requireprohibitivelylargenumbersof crosspoints, evenwhenblockingis allowed.As mentionedpreviously,very largeswitchesusemorethanthreestagesto providefurtherreductions in crosspoints. Figure5.9showsablockdiagramof a five-stageswitch obtainedby replacingeverycenter-$tage arrayin Figure5.6 with a three-stage array. Thisparticularstructureis notoptimumin termsof providinga givenlevelof performancewith the fewestcrosspoints, but it is a usefuldesignbecauseof its modularity. (Furthermore, it is a lot easierto analyzethansomeotherfive-stagesfucrures.) If themiddletlree stagesof a five-stageswirchasshownin Figure5.9 arestrictly nonblocking(kz= Znz- I ), thedesignprovidesa savingsof 9704crosspoints in each TABLE5.2 Three-Stage SwitchDeslgnsfor BlockingProbabititlee of 0.002andIntet Utlllzation of 0.1

Switch Size,N 't28 512 2,048 8,1S2 32,768 131,072

Numberof Crosspoints I 16 32 64 128 256

5 7 10 15 24 41

0.625 0.438 0.313 0.234 0,188 0.160

2,560 14,336 81,920 491,520 3.1million 21.5 million

Numberof Crosspoints in Nonblocking Design 7,680 63,488 516,0e6 4,2million 33 million 268million

(k= 15) (k = 31) (k= 63) (k= 127) (,(= 255) (k = 511)


237

TABLEE.l Three-StageSwitch Deslgnsfor Blocking Probabllltlesof 0.002end lnlet UtllltationBof 0.7 Switch Size N

128 512 2,048 8,192 32,768 131,072

n

k

I 16 32 64 128 256

14 22 37 64 11 6 215

B

Numberof Crosspoints

in Numberof CrossPoints Design Nonblocking

1.75 1.38 1.16 1.0 0.91 0.84

7,168 45,056 303,104 2.1million 15.2million 1 1 3m i l l i o n

7,680 63,488 516,096 4.2million 33 million 268million

( k =1 5 ) ( k= 3 1 ) (k = 63) (k = 127) (k = 255) ( k= 5 1 1 )

switch designspresentedearlier' array of the 32,7681ine,three-stage center-stage in the 32,768-linetandemswitch saved are crosspoints I million Hencea littl* one, not introduceblocking,theperformstages do middle the designof Table5.3. Since design' of thethree-stage performance to the is identical switch anceof thisfive-stage amounts smgll allowing by be obtained could design Naturally,a morecost-effective of bloclcingin themiddlestages.Theprobabilitygraphof thefive+tageswitchis shown in Figure5-.10.Fromthis graph,theblockingprobabilityis determinedasfollows:

(s.e)

B = ll - (sr)z[t - (1- d)t']]r'

11 1il1

fl1 r 11

azx*z ;frx ffirr,,,

2

3

4

Flgure 5.9 Five-stageswirchingnetwork'

238

DIGITAL $WITCHING

rz=rrfir rfir

2

Ft

,-1/3-1i.15--\,^.

Pr

Flgure5.10 probability graphof five_stage network. where4r - | - h and,qr= I - Or. Even greatercrosspointreductionscan, of course,be achievedby using more stagesto replacethe ratherlarge first- and third-stageaffays.For example,the total numberof crosspoints in the 32,0001ineswitchcanbereducedto lessthan3 million. The 130,000-lineswitchis not practicalwith electromechanical switchingmakices but is well within thecapabilitiesof a digiraltime divisionswitch. 5.2.3 Blocklng Probabiliiles: Jacobaeus Theformulationsof blockingprobabilityobtainedfrom probabilitygraphsenrailseveralsimplifyingassumptions. one of theseassumptions involvesexpressing thecompositeblocking probability of the alternatepathsas the produci of the blocking probabilitiesof eachindividual path. This stepassumesthat the individual probabilitiesare independent. In fact, the probabilitiesare not independent, particutarty whensignificantamountsof expansionarepresent.Considera switchingmatrixwith k=Zn - 1.Equation5.8producesa finite blockingprobabilityeventtroughtheswitch is knownto be strictlynonblocking.Theinaccuracy resultsbecause when2n * z paths arebusy,theremainingpathis assumed to be busywith a probabilityof I - (qjr. In fact,theremainingpathis necessarily idle. In general,whenspaceexpansion exists,theassumption ofindependent individual probabilitiesleadsto an erroneouslyhigh valueof blocking.Ttrelnaccuracyresults


239

becauseasmoreandmorepathsin a switcharefoundto be busy,the remainingpaths links canever arelessandlesslikely to be in use(only a subsetofn ofthe interstage at any one time). be busy A moreaccuratebut not exactanalysisof multistageswitchingmatriceswaspresentedin 1950by C. Jacobaeus [4]. Althoughtheanalysisis conceptuallystoaightforhere. amountof manipulationthatis notpresented involve a considerable ward,it does as from reference is obtained switch [5] for a three-stage equation Theresulting

u=ffipk(2-il,*

(s.10)

where n = number of inlets (outlets) per f,trst-(third-) stagearray 11= number of second-stageIIIrays pinletutilization In the interest of comparing the two methods, Equations 5'6 and 5.10 have been evaluatedfor three-stageswitches with varying amount$of spaceexpansion.The results were obtained for inlet utilization of 0.7 and are presentedin Table 5'4' Table 5.4 revealsthat the two analysesare in close agreementfor near-unity expansion factors. In fact, if p = l, the two formulations produce identical results' As expected,the Lee graph analysis (Bquation 5.8) producesoverly pessimisticvalues for the blocking probability when p > 1. As anothercomparisonbetweenthe two approaches,Table 5.5 is included to demonstrate the use of Equations 5.8 and 5.10 for switches with significant amounts of concenfrationmade possible by a relatively low inlet utilization of 0.1. Table 5.5 reveals that a Lee graph analysis (Equation 5.8) consistently underestimates the blocking probability when concentration exists. Actually, the Jacobaeus analysis presentedin Equation 5.10 also underestimatesthe blocking probability if large concenhation factors and high blocking probabilities are used.When necessary, more accuratetechniquescan be used for systemswith high concentrationand high TABLES.4 Comparisonof Blocking ProbabilityAnalyses(P= 0'7)" Numberof Center Stages,k 14 16 20 24 2B 31D

SpaceExpansion, F

Lee 5.8 Equation

0,875 1.0 1.25 1.50 1.75 1.94

0.548 0.221 0.014 3 . 2x 1 0 + 3 . 7x 1 0 4 8.5x 104

aswitchsize N = 512; inletgroup8iz€n = 16i inletulilizationp = 0.7. Di,,lonblocking.

Jacobaeu$ Equation5.10

0.598 0.221 0.007 2 . 7x 1 0 + 7.7x 104 0 , 1x 1 0 - 1 2

240

DIGITAL SWITCHING

TABLE5.5 Comparisonof BlockingprobabllltyAnatyses(p= 0.1)" Numberof Center Stages,k

10 12 14 16

SpaceExpansion,

F 0.375 0.5 0.625 0.75 0.875 1.0

Lee Equation 5.8

Jacobaeus Equation 5.10

0,0097 2 . 8x 1 0 4 4.9 x 10+ 5 . 7x 1 0 4 4.0 x 10-10 2 . 9x 1 0 * 1 2

0.Q27 8 . 6x 1 0 + 1 . 5x 1 0 + 1 . 4x 1 0 - 7 7 . 8x 1 f l o 2.9 x 10-12

aswitchsize N = 512; inletgroup siz6 n = 16; Inl€rutilizationp = 0.1

blocking. However, switcheswith high blocking probabilities normally have no practical interest so they are not consideredhere. Usersof PBXs sometimesexperiencedhigh blocking probabilities, but blocking in thesecasesusually arisesfrom too few tie lines to other corporatelocations or too few trunk circuits to the public network. The subjectof blocking in trunk groups is heated in Chapter 12. up to this point, the blocking probability analyseshave assumedthat a specific inlet is to be connectedto a specific outlet. Also, it has been assumedthat the requests for service on the individual lines are independent.These assumptionsare generally valid for swirching one subscriberline to another in an end office switch or for connecting one station to another in a PBx. Neither of these assumptionsapply to connections to or from a fiunk group. when connectingto a trunk circuit, any circuit in a trunk group is acceptable.Thus the blocking probability to a specif,rccircuit is only as important as its significance in the overall blocking to the trunk group. The blocking probability to any particular circuit in a trunk group can be relatively large and still achievea low compositeblocking probability to the trunk group as a whole. If the blocking probabilities to the individual trunks are independent,the compositeblocking probability is the product of the individual probabilities. However, the paths to the individual trunk circuits normally involve {iome common links (e.g., the junctors from a first-stage array to all second-stagearrays). For this rea$on the individual blocking probabilities are normally dependent,which must be consideredin an accurateblocking probability analy-

sis.

As an extreme example, consider a casewhere all trunks in a trunk group are assigned to a single outlet array in a three-stageswitch. since the paths from any particular inlet to all trunks in the group are identical, the ability to selectany idle 61nk is useless.In practice, the individual circuits of a trunk group should be assignedto separateoutlet arays. Another aspectof fiunk groups that must be consideredwhen designing a swirch or analyzing the blocking probabilities involves the interdependenceof activity on the

swlrcHlNc s.z sPAcEDtvtstoN

241

individual circuits within a trunk group. In contrast to individual subscriberlines or PBX stations,individual circuits in a trunk group are not independentin terms of their probabilities of being busy or idle. lf some number of circuits in a trunk group are testedand found to be busy, the probability that the remaining circuits are busy is increased.The nature of these dependenciesis discussedmore fully in Chapter 12. At this point it is only necessaryto point out that these dependenciescause increased blocicingprobabilities if the individual ffunks are competing for common paths in the switch. ,l,gain, the effect of thesedependenciesis minimized by assigningthe individual trunks to separateinlet-outlet arraysso that independentpathsareinvolved in connections to and from the trunk group. This process is sometimes referred to as decorrelatingthe trunk circuits. One last aspectof the blocking probability as a gradeof serviceparameterthat must be mentionedinvolves variations in the loading of the network by individual users.In the design examplesfor an end office presentedearlier, it was tacitly assumedthat all subscribersarebusy l07o of the time during a busy hour' In fact, some subscribersare active much more than I 07oof the time, and other subscribersare active lessthan l07o of the time. In terms of haffic theory, some subscriberspresentmore than 0.1 erlangs* of traffic to the network, whereasothers presentless' When a switch is partitioned into subgroups(as all Iarge switchesmust be) and the haffic is concentratedby first-stage switching alrays, a few overactive subscribers in one subgroup can significantly degrade service for the other subscribers in the subgroup. It does not matter that the subscribers in some other subgroup may be explriencing lower than averageblocking. Their essentiallynonblocking service is no compensation for those subscribers experiencing a relatively poor grade of service. Operatingcompanieshave traditionally solved the problem of overactivesubscribersby specifically assigningthe most active lines (businesses)to separateinlet groups of the switctr. Sometimesthis procedurerequires making traffic measurementsto determine which lines are most active and reassigningtheselines to different palts of the switch. These proceduresfall into the general category of line administration. If the subgroupsare large enough, or ifthe designsprovide adequatemargin for overactive ure.s, thir aspectof line administration can be minimized. One feature of a digital switch that can be utilized in this regard is the ability to design economical switches with very low nominal blocking probabilities so that wide variations in traffic intensities can be accommodated. Even a modern digital switch can experienceloading problems when confronted with extreme traffic conditions. An Internet service provider (ISP) in a metropolitan area may atffact an extremely large amount of traffic that all passesthrough a single class 5 switch. Although the connectionsto the ISP are lines, as far as the switch is concerned,they actually representa trunk group with very heavy haffic, so much so that $pecialline adminishation is required.

*An isbusy'A circuit oftimethatadevice theproporlion specifying oftrafficintensity is ameasure erlang is saidtocarry0.1erlangif it is busyl0oloof thetime.

242

DIGITAL SWITCHING

5.2.4 Folded Four-Wire Switchee Multiple-stageswitchescanbe usedfor eithertwo- or four-wireswitchingoperations. Figure5.ll depictsa four-wireconnectionthrougha four-stageswitch.Noticethat two pathsmustbeestabli$hed for thecompleteconnection. A two-wireconnectionrequiresonly onepathsinceeachoutletis extemallyconnected to its corresponding inlet. Thetwo pathsshownin Figure5.1I demonshate a particularlyusefulrelationship; Onepathis a mirror imageof the otherpath.If the switchdiagramis folded aboutthe verticalcenterline,thepathscoincide.Hencethismethodof settingup connections is sometimesreferredto as a folded operation.When all connectionsin the switch are setup with a foldedrelationship,severalbenefitsresult. Firstofall, only onepathfindingoperationis neededsincethereversepathis automaticallyavailableasa mirror imageof theforwardpath.In essence, everycrosspoint on onesideis pairedwith anothercrosspointin a corresponding arrayon theopposite sideof theswitch.Wheneveronecrosspointof a pair is usedin a connection, theother crosspointin thepair is alsoused.For example,thethird inlet arrayin the first stage usescrosspoint(6,4) to connectits sixthinlet to its fourthoutlet(leadingto thefourth artay of the secondstage).The correspondingcrosspointin the third outlet array of thelaststageconnectsits fourth inlet (comingfrom thefourth arrayin thefour-thstage) to its sixth outlet.[n general,crosspointf,7 in onearrayis pairedwith crosspointj,I in the conespondingarrayon the oppositesideof a switch.Sincethe availabilityof onecrosspointin a pair ensuresthe availabilityofthe other,therever$epathis automaticallyspecifiedandavailable. A secondadvantageof the folded four-wire operationresultsbecausethe amount of informationspecifyingthestatusof theswitchcanbe cut in half. Only thestatusof

grth Inldt of thlrd arEy

of rfiird rnv

a t I

a I a

(7 llt El6|rrfr of liftdnfi

inlit my

Eldudnfior|tl|t of flftrnfi

Figure 5.11 Four-wireconnectionthroughfour-stageswitch.

ffr.y


245

each pair of crosspoints or associatedjunctor is needed to find an available path through the switch. A third benefit of the folded structure occurs becausethe blocking probability is one-half of the probability of finding two pathsindependently.It might seemthat pairing the crosspointsin the describedmannerwould reshict the pathsavailablefor a particular connection. On the contrary, the crosspoint pairings merely guaranteethat a reverse path is automatically available for any selected path in the forward direction' The folded operation in the preceding paragraphs referred to a switch with an even number of switching stages.An even number was chosenbecausethe conceptis easiest to demonstratewhen no center stage is present. The basic approach can be extendedto odd numbersof switching stagesif the centerstagecontainsan even number of anays and is folded about a horizontal line at the centerof the stage.In this manner, crosspoint i,7 in the top center-stagealray is paired with crosspointj, i in the bottom center-$tagearray, and so on.

5.2.5 Pathfinding switchis virtuallyautomaticsincethenecDetermininga paththrougha single-stage In conessarycrosspointis uniquelyspecifiedby theinlet-outletpairto beconnected. thepath switchcomplicates trast,availabilityof morethanonepathin a multiple-stage of theswitchmustkeeptrackof whichpotential selectionprocess.Thecatlprocessor pathsfor a particularconnectionare availablein a statestore.A pathfindingroutine processes thestatestoreinformationto selectanavailablepath.Whenevera newconthestatestoreis updatedwith theapproor anold onereleased, nectionis established priateinformation. Pathtlndlng Times Pathfindingbperationsrequirethe useof commonequipmentand must thereforebe Thetime analyzedto determinetherateat whichconnectrequestscanbe processed. requiredto find an availablepath is directly dependenton how manypotentialpaths cantesta numberof pathsin pararetestedbeforeanidle oneis found.Somesystem$ time.Sincetheexpectednumberof potential allel andtherebyshortentheprocessing pathsthatmustbe testedto find anidle pathis a functionof link utilization,pathfinding timesunfortunatelyincreasewhenthe cornmoncontrolequipmentis busiest. Assumethattheprobabilityof a completepaththroughtheswirchbeingbusyis denotedby p. If eachof k possiblepathsthroughtheswitchhasanequalandindependent probabilityof beingbusy,theexpectednumberof pathsNothatmustbe testedbefore in AppendixA asfollows: anidle pathis foundis determined

N,:H

( s.lr )

Example5.1. Whatis theexpectednumberof potentialpathsthatmust.betestedto 8192Jineswitchdefinedin Table5.2? find anidle pathin thethree-stage,

244

DretTAL swtrcHtNc

solution. As indicatedin the table,a $paceexpansionfactor of 0.234is usedto providea blockingprobabilityof 0.002.Hencetheutilizationof eachinterstagelink is 0.1/0.234=0.427. The blockingprobabilityof eachpaththroughthe switchis merelythe probabilitythat oneof two links in seriesis busy.Hencetheprobability = 0.672,andtheexpectednumberof pathsto be resredis p = | * (I - O.427)z

Np=

I - (0.672)15

| -0.672

=3.04 Example 5. I demonstratesthat, on average,only 3 of the 15 potential paths need to be testedbefore an idle path is found. However, when the switch is experiencing greaterthan normal ffaffic loads,the averagenumber of paths testedincreases.For example, if the input line utilization increasesfrom 10 to l|Vo, the blocking probability increasesfrom 0.002 to 0.126, and the expectednumber of paths to be testedin the pathfinding operation increasesfrom 3 to 4.g. Although this simple pathfinding example is, by itself, not particularly important, it demonstrates a very important aspect of the loading of common equipment in a switching system(or a network as a whole): Linear increasesin ttre offered traffic produce geometric increasesin the utilization of the network resources.If a systemis designed and analyzedunder nominal traffic conditions and the results are extrapolated to determinetotal capacity,greatly optimistic estimatesoften occur. In the pathfinding example the processingrequirementsincreasejust when the call processor(storedprogram conhol) is already loaded with greaterthan averagetraffic volumes. For further discussionof the effects of traffic loads on the common equipmentof a switching sy$temseeSchwartz [6].

5.2.6 SwitchMatrixControl When an available path through a cofilmon control switching network is determined, the control element of the switch transfers the necessaryinformation to the network to selectthe appropriatecrosspoints.Crosspoint selection within a matrix is accomplished in one of two ways. The control may be associatedwith the output lines and thereforespeciff which inputs areto be connectedto the associatedoutputsor the control information may be associatedwith eachinput and subsequentlyspeciff to which outputs the respectiveinputs are to be connected.The first approachis referred to as output-associatedcontrol while the secondis called input-associatedcontrol. These two control implementationsare presentedin Figure 5.12. Input-associatedcontrol was inherently required in step-by-stepswitches where the information (dial pulses) an-ivedon the input link and was used to directly select the output links to each successivestage.[n common control system$,however, the addressinformation of both the originating line and the terminating line is simultaneously available. Hence the connection can be establishedby beginning at the desired

5.2 SPACEDIVISIONSWITCHING

245

(b) inputassociated. Figure 5.12 Switchmatrixconffol:(a) outputassociated; outlet and proceeding backward through the switch while selecting inputs to each stage. The implementation of both types of digital crosspointarraysusing standardcomponents is shown in Figure 5.13. Output-associatedcontrol uses a conventional data selector/multiplexer for each matrix output. The number of bits required to control eachdata selectoris log2 N, where N is the number of inlets' Thus the total number of bits required to completely specify a connection configuration is M log2 N. Input-associatedcontrol can be implemented using conventional line decoders/de"wired-or" logic function. Thus the multiplexers. The outputs are conmoned using a output gatesof each decodercircuit must be open-collectoror histate devices if transistor-transistor-logic (TTL) is used. The total number of bits required to specify a connection configuration in this caseis N log2 M. A significant drawback of input-associatedcontrol arisesfrom the need to disable unused inputs to prevent cross connectswhen another input selectsthe sameoutput.

Dfir rclictot/mu ltiPloxor

loga N

Linr dccoder/demultiplexer

log?rY

ft) Figure 5.13 Standard component implementation of digital crosspoint aray: (a) outputassociatedcontrol; (b) input-associatedcontrol.

246

DIGITALSWITCHING

with output-associatedconffol, unused output$ can remain connected to an input without preventing that input from being selectedby another output. For this reason and for generally greaterspeedsof operation,digital switching networks typically use output-associatedcontrol. Notice, however, that the total amount of information neededto specify a connectionconfiguration with input-associatedcontrol is lessthan that with output control if the number of inputs N is much smaller than the number of outputsM (N log2 M < M log2Il). Furthermore,input-associatedconffol is more flexible in terms of wired-or (e.g., bus) expansion.

5.3 TIMEDIVISION SWITCHING Asevidenced by multiple-stage switching, sharing of individual crosspoints for more than one potential connectionprovides significant savingsin implementation costs of spacedivision switches.In the casesdemonstrated,the crosspointsof multistagespace switchesare sharedfrom one connectionto the next, but a crosspointassignedto a particular connection is dedicated to that connection for its duration. Time division switching involves the sharing of crosspointsfor shofier periods of time so that individual crosspointsand their associatedinterstagelinks are continually reassignedto existing connections.When the crosspointsare sharedin this manner, much greater savings in crosspointscan be achieved.rn essence,the savings are accomplishedby time division multiplexing the crosspointsand interstagelinks in the samemannerthat transmissionlinks are time division muttiplexed to shareinteroffice wire pairs. Time division switching is equally applicableto either analog or digital signals.Ar one time, analog time division switching was attractive when interfacing to analog transmissionfacilities, since the signals are only sampled and not digitally encoded. However, large analog time division switches had the samelimitations as do analog time division ffansmission links: the pAM samples are particularly vulnerable to noise, distortion, and crosstalk.Thus, large electronic switching matriceshave always incorporatedthe cost of digitizing PAM samplesro maintain end-to-end signal quality. The low cost of codecsand the prevalenceof digital trunk interconnectionsimply that analog switching is now usedin only the smallestof switching system$(e.g.,electronic key systems).

5.3.1 AnalogTimeDlvlsionSwltching Although analog time division swirching has become obsolete,it is a good starting point to establishthe basic principle of time division switching. Figure 5.14 depicts a padicularly simple analog time division switching structure.A single switching bus supportsa multiple number of connectionsby interleaving pAM samplesfrom receive line interfacesto ffansmit line interfaces.The operation is depicted as though the receive interfacesare separatefrom the respectivetransmit interfaces.When connecting two-wire analog lines, the two interface$are necessarilyimplemented in a common

DrvrsroN swtrcHtNc 247 s.s TIME

[-l r r i - iI |-l

-tlr t-l

f\H

-1

t - l l

i-1

I r_r, I

a

i

F\ t-l Lin6 inGrfe6

L-l/ I -$,vltchlng i bur Tlme _ I Cvclls I divirion I qqqqql I controt

Figure5.14 Analogtimedivisionswirching' in somePAM-PBX systems,analogsamplesweresimultanemodule.Furthermore, in bothdirectionsbetweenthe interfaces[7]. ouslytransferred Includedin Figure5.14aretwo cycliccontrolstores.Thefirst controlstorecontrols gatingof inputsonto the bus one sampleat a time. The secondcontrol storeoperates in synchronismwith the first and selectsthe apPropriateoutput line for eachinput sample.A completesetof pulses,onefrom eachactiveinput line, is referredto asa frame.Theframerateis equalto the samplerateof eachline. For voicesystemsthe samplingraterangesftom 8 to 12 kHz. The highersamplingrate$weresometimes filtersin theline interfaces. usedto simplifythebandlimitingfilter andreconstructive 5.3.2 DlgltalTimeDlvlsion Swltchlng The analogswitchingmatrixdescribedin theprecedingsectionis essentiallya space for shortperiods divisionswitchingmatrix.By continuallychangingtheconnections of time in a cyclicmanner,theconflgurationof thespacedivisionswitchis replicated oncefor eachtime slot. This modeof operationis refenedto as time multiplexed swirching.While thismodeof operationcanbequiteusefulfor bothanaloganddigital signals,digital time divisionmultiplexedsignalsusuallyrequireswitchingbetween time slotsaswell asbetweenphysicallines.Switchingbetweentime slotsrepresents a seconddimensionof switchingandis refenedto astime switching. unle$s In thefollowing discussionof digitaltime divisionswitchingit is assumed, otherwisestated,thattheswitchingnetworkis interfaceddirectlyto digitaltime divi-

248

DIGITALSWITCHING

sionmultiplexlinks.This assumption is generallyjustified$ince,evenwhenoperating in ananalogenvironment, themostcost-effective switchdesignsmultiplexgroupsof digital signalsinto TDM formatsbeforeany switchingoperationstakeplace.Thus mostof thefollowing discussionis concerned with the intemalstructures of time divisionswitchingnetworksandpossiblynot with the structureof an entireswitching complex. The basicrequirementof a time division switchingnetworkis shownin Figure 5.15.As anexampleconnection, channel3 of thefirst TDM link is connected to channel l7 of thelastTDM link. Theindicatedconnectionimpliesthatinformationarriving in time slot 3 of thefirst input link is transferred to time slot l7 of thelasroutput link. Sincethevoicedigitizationprocessinherentlyimpliesa four-wireoperation,the returnconnectionis requiredandrealizedby transferringinformationfrom time slot 17 of the lastinput link to time slot 3 of the first ouFut link. Thuseachconnection requirestwo transfersof information,eachinvolving translationsin both time and space. A varietyof switchingsffucturesarepossibleto accomplishthetransfersindicated in Figure5.15.All of thesestructuresinherentlyrequireat lea$ttwo stages:a space divisionswitchingstageanda timedivisionswitchingstage.As discussed later,larger switchesusemultiplestagesof bothfypes.Beforediscussing switchingin bothdimensions,however,we discussthecharacteristics andcapabilities of timeswitchingalone. A Digital lllemory Swltch Primarilyowing to the low cosrof digital memory,time swirchingimplemenrarions providedigitalswitchingfunctionsmoreeconomically thanspacedivisionimplementations.Basically,a time switchoperates by writing datainto andreadingdataout of a singlememory.In theprocess, theinformationin selected time slotsis interchanged, asshownin Figure5.16.when digital signalscanbe multiplexedinro a singleTDM format, very economicalswitchescan be implementedwith time switchingalone. However,practicallimitationsof memoryspeedlimit thesizeof a time switchsothat someamountof spacedivisionswitchingis necessary in largeswitches.As demonstratedin later sections,the mosteconomicalmultistagedesignsusuallyperformas muchswitchingaspossiblein thetime stages. Thebasicfunctionaloperationof a memoryswitchis shownin Figure5.17.Individualdigital message circuitsaremultiplexedanddemultiplexedin a fixed manner I FNAffiH

I I t

t

Figure 5.15 Time andspacedivisionswirching.

SWITCHING 249 5.3 TIMEDIVISION

operation. Figure5.16 Timeslotinterchange to establisha singleTDM link for eachdirectionof travel.The multiplexinganddeaspartof the swirchitself, or they may be multiplexingfunctionscanbe considered form terminals.In eithercase,a byte-interleaved in remotetransmission implemented transmission hierarchy(DSz, DS3, of multiplexingis required.The asynchronous DS4) usesbit interleavingand thereforerequiresback-to-backdemultiplexingand multiplexing operationsbefore switching c€urbe accomplished.In contrast,the in Chapter8 canprovidebyteinterleavingspeSONETnansmission formatdescribed cifically soit canbe moredirectlyinterfacedto a digital switchingsystem. by Theexchangeof informationbetweentwo differenttime slotsis accomplished (TSI) circuit.In theTSI of Figure5.17datawordsin incoming a time slotinterchange locationsof thedatastorememory.Datawords time slotsarewritteninto sequential obtainedfrom a control for outgoingtime slots,however,arereadfrom addresses between store.As indicatedin the associated controlstore,a full-duplexconnectron TDM channeli andTDM channelj impliesthat datastoreaddressi is readduringouttwiceduringeach goingtime slotj andvice versa.Thedatastorememoryis accessed link time slot.First, somecontrolcircuitry (not shown)selectsthe time slot number

MUX

T I M E S L O TI N T E H C H A N G E

Figure 5.17 MIDVTSI/DEMLIXmemoryswitch.

250

DIGITAL SWITCHING

as a write address.Second,the content of the conffol store for that particular time slot is selectedas a read address. since a write and a read are required for each channel entering (and leaving) the TSI memory, the maximum number of channelsc that can be supportedby the simple memory switch is

r25 c:4

(5.r?)

where I ?5 is the frame time in microsecondsfor 8 kHz sampled voice and f" is the memory cycle time in microseconds. As a specific example, consider the use of a t 5.2 nsec of memory. Equation 5. 12 indicates that the memory switch can support 4096 channels(2048 full duplex connections) in a strictly nonblocking mode of operation. The complexity of the switch (assumingdigitization occurselsewhere)is quite modest:The TSI memory storesone frame of data organizedas 4096 words by 8 bits each.The control store also requires 4096 words, but eachword has a length equal to log2(c) (which is 12 in the example). Thus the memory functions can be supplied by a096 x 8 and 4096 x lz bit randomaccessmemories (RAMs). The addition of a time slot counter and some gating logic to selectaddressesand enablenew information to be written into the conffol $torecan be accomplishedwith a handful of conventional integratedcircuits (ICs). This switch should be conrrastedto a spacedivision design (Equation 5.3) that requires more than 1.5 million crosspoints for a nonblocking three-stageswitch. Although modern IC technology might be capable of placing rhar many digiral crosspointsin a few ICs, they could never be reachedbecauseof pin limitations. As mentionedin Chapter 2, one of the main advantagesof digital signals is the easewith which they can be time division multiplexed. This advantagearisesfor communication between integratedcircuits as well as for communication between switchine offices.

Time Stagesin General Time switching stagesinherently require some form of delay element to provide the desiredtime slot interchanges.Delays are most easily implemented using RAMs that are written into as data arrive and read from when data are to be transferredout. Ifone memory location is allocated for each time slot in the TDM frame format, the information from each TDM channel can be stored for up to one full frame time without being overwritten. There are two basic ways in which the time stagememoriescan be controlled; written sequentiallyand read randomly or written randomly and read sequentially.Figure 5.18 depicts both modesof operation and indicateshow the memories are accessedto translateinformation from time slot 3 to time slot 17. Notice that both modes of operation use a cyclic control store that is accessedin synchronism with the time slot counter.

s.4 TWo-DTMENS|oNAL swrrcHrNc 251 DEtd Stors

'ims glot Countdr

(rl

{bl

Figure 5.18 Time switch modules: (a) sequential writes/random reads; (b) random writes/ sequentialreads.

The first modeof operationin Figure5.l8 impliesthat specificmemorylocations channelsof theincomingTDM link. Datafor eachincomarededicatedto respective locationswithin the memoryby incrementinga ing time slot arestoredin sequential modulo-ccounterwith everytimeslot.As indicated,thedatareceivedduringtime slot 3 areautomaticallystoredin thethird locationwithin the memory.On output,inforfor mationretrievedfrom the controlstorespecifieswhich addressis to be accessed word of thecontrolstoreconthatparticulartime slot.As indicated,the seventeenth to tainsthenumber3, implyingthatthecontentsof datastoreaddress3 is transferred theoutputlink duringoutgoingtime slot 17. Thesecondmodeof operationdepictedin Figure5-18is exactlytheoppositeof the frrstone.ftrcomingdataarewritteninto thememorylocationsasspecifiedby thecontrol store,but outgoingdataareretrievedsequentiallyundercontrolof an outgoing time slotcounter.As indicatedin theexample,informationreceivedduringtime slot retrieveddur3 is writtendirectlyinto datastoreaddress17,whereit is automatically ing outgoingTDM channelnumber17.Noticethat the two modesof time $tageopcontrol and €ration depicted in Figure 5.18 are forms of output-associated designexamplepresented control,respectively. In a multiple-stage input-associated later,it is convenientto u$eone modeof operationin one time stageand the other modeof operationin anothertime stage.

5.4 TWO.DIMENSIONALSWITCHING Largerdigital switchesrequireswitchingoperationsin botha spacedimensionanda thatcanbe used time dimension.Therearea largevarietyof networkconfigurations To beginwith, considerthesimpleswitchingstructo accomplish theserequirements. a time stageT foltureshownin Figure5.19.This switchconsistsof only two $tages: Iowedby a spacestageS. Thusthis $tructureis refered to a time-space(TS) switch. Thebasicfunctionof the time stageis to delayinformationin arrivingtime slots until thedesiredoutputtime slotoccurs.At thattimethedelayedinformationis trans-

252

DIGITALSWITCHING

f,'lgure5.19 Time-space(TS)swirchingmatrix. ferred through the space stage to the appropriate output link. In the example shown the information in incoming time slot 3 of link I is delayed until outgoing time slot l7 occurs. The return path requiresttrat information arriving in time slot 17 of link N be delayed for time slot 3 of the next outgoing frame. Notice that a time $tagemay have to provide delays ranging from one time slot to a full frame. Associated with the $pacestage is a control store that contains the information needed to speciff the space stage configuration for each individual time slot of a frame. This conhol information is accessedcyclically in the samemanner as the control information in the analogtime division switch. For example,during eachoutgoing time slot 3, control information is accessedthat specifresinterstagelink number 1 is connectedto output linkN. During other time slots,the spaceswitch is completely reconfigured to support other connections. As indicated, a convenient meansof representinga conffol store is a parallel endaround-shift register. The width of the shift register is equal to the number of bits required to specify the entire spaceswitch configuration during a single time slot. The length of the shift register conforms to the number of time slots in a frame. Naturally, $omemeansof changingthe information in the conhol storeis neededso that new connections can be established.In actualpractice, the control store$may be implemented as RAMs with counters used to generateaddressesin a cyclic fashion (as in the time stagecontrol storesshown previously).

lmplementationComplexltyof Time Dlvlslon Switchee In previous sections, alternative space divisionswitching structures werecompared in terms of the total number of crosspointsrequired to provide a given grade of service. In the caseof solid-stateelectronic switching matrices, in general,and time division switching, in particular, the number of crosspointsalone is a lessmeaningful measure of implementation cost. Switching structures that utilize ICs with relatively large numbersof internal crosspointsare generally more cost effective than other $uuctures

5.4 TWO.DIMENSIONAL SWITCHING 253

Hencea morerelevantdesignpabut morepackages. thatmayhavefewercrosspoints If alternate total number switcheswouldbethe of IC packages. rameterfor solid-state packages of ICs, the may numberof designsare implementedfrom a comlnonset closelyreflectthenumberof crosspoints. In additionto thenumberof crosspoints in spacedivisionstages,a digitaltime divisionswitchusessignificantamountsof memorythatmustbeincludedin ane$timate of theoverallcost.-Thememorycountincludesthetime stagememoryarraysandthe In largeswitches,thenumcontrolstoresfor boththetime stagesandthespacestages. expense canbe reducedat the ofincreasingtheamount berof spacestagecrosspoints implementationcomplexityrequires of of memoryrequired.Thus,athoroughanalysis knowingthe relativecostof a crosspointrelativeto thecostof a bit of memory.Becausea crosspointis closelyassociated with an extemalconnection,it costssignifiintegrationICs cantlymorethana bit of memory.Theuseof standardmedium-scale leadsto onscrosspointcostingaboutthesameas I fi) bits of memory.tUseof custommemorycanchangethisfactor,particularly ICswith integrated or application-specific perpin. For provideaccess to manymorecrosspoints IC packages because large-scale purposes of implemenof illustratingvariousdesigntrade-offs,thefollowinganalyses tationcomplexitycontinueto considercrosspointcoststo be lfi) timesthe costof a memorybit. Dependingon theimplementationapproach,this factormay not be accurate,but minimizingthecostof a matrixis no longermuchof a concern,exceptin extremelylargeswitchingsystems(e.g.,in switchesexceeding100,000voicechannels). asfollows: complexityis expressed Theimplementation No

=N**ffi ComPlexitY where

(s.l3)

Nx = number of spacestagecrosspoints NB = number of bits of memory

complexityof theTS switchshownin Example5.2. Determinetheimplementation N = 80. Assumeeachinput line of TDM input lines where the number Figure5.19 (24 matrix assumea one-stage channels). Furthermore, single DSl signal containsa space stage. is usedfor the in the spacestageis determinedas Solutinn. Thenumberof crosspoints Nx = 8d: 6400

*lt

is worth noting that digital memories are inherently implemented with at least two crosspoints per bit. In this case the crosspoiflts are gates used to provide write and read accessto the bits. These crosspoiflts, howcver, are much less expensive than messagecrosspointsthat are accessedfrom extemal circuits, TSeethe first or second edition of this book.

254

DIGITALSWITCHING

(Thecrosspoints on themaindiagonalareincludedsincetwo channelswithin a single TDM input may haveto be connectedto eachother.)Thetotal numberof memorybits for the spacestagecontrolstoreis determined as Nnx = (numberof linksXnumberof controlwords)(number of bits per control word) = (80x24X7) = 13.440 Thenumberof memorybitsin thetime stageis determined asthesumof thetime slot interchange andthecontrolstorebitsl Nsr = (numberof links)(numberof channels)(number of bitsper channel) + (numberof links)(numberof control words)(numberof bits per controlword) = (80X24X8)+ (80)(24)(s) = 24,96O Thustheimplementation complexityis determinedas Complexity= Nx *

W

: 6114equivalent crosspoints

The implementationcomplexitydeterminedin Example5.2 is obviouslydominatedby thenumberof crosspoints in thespacestage.A significantlylowercomplexity (andgenerallylower cost)canbe achievedif groupsof input links arecombined into higherlevel multiplexsignalsbeforebeingswitched.The costof the front-end multiplexersis relativelysmallif the individualDSI signalshavealreadybeensynchronizedfor switching.In thismanner,thecomplexityof thespacestageis decreased appreciably whiletheoverallcomplexityof thetimestageincreases only slightly.(See the problemsat the end of this chapter.)The implementationcostsare reducedproportionately,up to the point that higherspeedsdictatethe useof a moreexpensive technology. A significantlimitationof theTS structureof Figure5.19occurswhena connection hasto bemadeto a specificchannelof anoutletasopposedto anychannelof anoutlet. A connectionto a specificchannelis blockedwheneverthe desiredtime slot of the inlet TSM is alreadyin use.For example,if time slot 17of thefirst inlet is in usefor someconnectionfrom the first inlet to somelink otherthanlink N, the connection from channel3 of inlet I to channel17of inletNcannotbemade.Because of thislimitation,theTS structureis usefulonly if theoutletsrepresent trunk groups,which implies any channelof an outletis suitable.*Applicationsthat requireconnectionsto specificchannelsrequireadditionalstagesfor adequate blockingprobabilityperformance. 'A

full-duplex connection requires the reverse connection to be established, which adds restrictions to which outlet channels can lre used. See the problems at the end of the chapter.

5.4 TWO-DIMENSIoNAL SWITCHING 255

Multlple*StageTime and SpaceSwitching As discussed in theprecedingsection,aneffectivemeansof reducingthecostof a time divisionswitchis to multiplexasmanychannelstogetheraspracticalandperformas muchswitchingin thetime stagesaspossible.Time stageswitchingis generallyless expensivethan spacestage$witching-primarily becausedigital memoryis much themselves cheaperthandigital crosspoints(AND gates).To repeat,the crosspoints it is thecostof accessing andselectingthemfrom externalpins arenot soexpensive, thatmakestheir userelativelycostly. Naturally,therearepracticallimits asto how manychannelscanbe multiplexed into a commonTDM link for time stageswitching.Whenthesepracticallimits are complexitycanbeachievedonly by reached,furtherreductions in theimplementation usingmultiplestages.Obviously,somecostsavingsresultwhena singlespacestage matrixof a TS or ST switchcanbe replacedby multiplestages. thespacestagesby a time involvesseparating A generallymoreeffectiveapproach stage,or, conversely,separating two time stagesby a spacestage.Thenext two secThefirst structure,consistingof a time stage tionsdescribethesetwo basicstructures. (STS)switch.Thesecis referredto asa space-time-space betweentwo spacestages, ond structureis referredto asa time-space-time(TST) switch.

5.4.1 $TS Swltching A functionalblock diagramof an STS switchis shownin Figure5.20.Eachspace (nonblocking)switch.For verylargeswitches, switchis assumed to be a single-stage Establishing it maybe desirable to implementthespaceswitcheswith multiplestages. a paththroughan STSswitchrequiresfinding a time switchanay with an available write accessduringtheincomingtime slotandanavailablereadaccessduringthedesiredoutgoingtime slot.Wheneachindividualstage(S,T, S) is nonblocking,theopspaceswitch.Hence erationis functionallyequivalentto theoperationof a three-stage a probabilitygraphin Figure5.2I of anSTSswitchis identicalto theprobabilitygraph of Figure5.8 for three-stage spaceswitches.Conespondingly,the blocking probabilitv of an STSswitchis

(STS)switchingstructure. Figure 5.20 Space-time-space

256

DIGITALSWITCHING

2 a a I

Figure 5.21 Probabilitygmph of STSswitch wirh nonblockingstages.

B=(1 -q'?)k

(5.14)

where4'=l - p'=l - pl\ (F= fr/llt ft = number of center-stagetime switch arrays Assuming the spaceswitchesare single-stageEuray$and that eachTDM link has c messagechannels, we may determine the implementation complexity of an STS switch as*

Complexity= numtterof spaceslagecrosspoints * (numberof spacestagecontrolbits + numberof time stagememorybits + numberof time stagecontrolbits)/100

=2H,{+

2/cclog, N + tc(8) + ftclog, c 100

(5.15)

Example5.3. Determinethe implementationcomplexityof a 2048-channel srs switchimplementedfor 16 TDM links with 128channelson eachlink. The desired maximumblockingprobabilityis 0.002for channeloccupancies of 0.1. solution. The minimum number of center-stagetime switches to provide the desiredgradeof servicecanbe determinedfrom Equation5.14as fr = 7. Using this valueof ft, thenumberof crosspoints is determinedas(2X7Xl6) = 224.Thenumber of bitsof memorycanbe determined as(2X7X128)(4) + (7X128X8)+ (7)(lZB)(7)= *This

derivation assumesoutput-associatedcontrol is used in the first stageand input-associatedcontol is third stage, A slightly different result occurs if the space stages are controlled in different ffiar3l"

swtrcHrNc 257 s.4 rwo-DrMENsroNAL 20,608. Hence the composite implementationcomplexity is 430 equivalent crosspoints. complexityobtainedin Example5.3 shouldbe comThevalueof implementation paredto thenumberofcrosspointsobtainedfor anequivalent-sized switch three-stage listedin Table5.2.Thespaceswitchdesignrequires8l ,920crosspoints whiletheSTS Thedramaticsavingscomesaboutas designrequiresonly 430equivalentcrosspoints. having already result of the voice been digitizedandmultiplexed(for transsignals a first insertedinto an analogenvironpurposes). switches mission were Whendigital ment,thedominantcostof the switchoccurredin theline interface.Digital interface basis. costsaremuchlower thananaloginterfacecosts,particularlyon a per-channel 5.4.2 TST Swltchlng time-spaceswitchis shownin Figure5.22- the The secondform of multiple-stage TST switch.Informationarrivingin a TDM channelof anincominglink is delayedin paththroughthe space$tageis available.At the inlet time $tageuntil an appropriate that time the informationis transferredthroughthe spacestageto theappropriateoutlet time stagewhereit is held until the desiredoutgoingtime slot occurs,Assuming the time stagesprovidefull availability(i.e.,all incomingchannelscanbe connected any spacestagetime slotcanbe usedto establisha connecto all outgoingchannels), In a functional sense the spacestageis replicatedoncefor everyintemaltimeslot. tion. by This conceptis reinforced theTST probabilitygraphof Figure5.23.

Figure 5.22 Time-space-time(TST) switchingsfiucfure.

258

DIGITAL SWITCHING

2 I a a

Figure 5.23 Probabilitygraphof TST swirchwith nonblockingstages. An important feature to notice about a TST switch is that the spacestageoperate$ in a time-divided fashion, independentlyof the external TDM links. In fact, the number of spacestagetime slots I does not have to coincide with the number of external TDM time slots c. If the spacestageis nonblocking, blocking in a TST switch occurs only if there is no intemal $pacestagetime slot during which the link from the inlet time stageand the link to the outlet time stageare both idle. obviously, the blocking probabiliry is minimized if the number of spacestagetime slots I is made to be large. In fact, as a direct analogy of three-stage$paceswitches,the TST switch is strictly nonblocking if I = 2c -1. The generalexpressionof blocking probability for a TST swirch wirh nonblocking individual stages(T, S, T) is

B =lr _ q?lt

(s.16)

w h e r e 4=l l - p r = l * p l a cr,= rimeexpansion(//c) I = numberof spacestagetime slots Theimplementation complexityof a TST switchcanbe derivedas

Complexity =N2 +

N/ log, N + 2Nc(8)+ 2Nl log, c

100

(s.17)

Example5.4. Determinethe implementation complexityof a 2048-channel TST switchwith 16TDM links and 128channelsper link. Assumethe desiredmaximum blockingprobabilityis 0.002for incomingchanneloccupancies of 0.1 Solution' UsingEquation5.16,we candeterminethenumberof internaltime slots requiredfor thedesiredgradeof serviceas25.Hencetimeconcentration of 1/g = 5.I ? is possiblebecauseof the light loadingon the input channels.The implementation complexitycannow be determined from Equation5.17as656.

25S 5.4 TWO-DIMENSIONAL SWITCHING is Theresultsobtainedin Examples5.3 and5.4 indicatethattheTST architecture that the TST than the Notice, however, switch opermorecomplex STSarchitecture. ateswith time concentration whereasthe STSswitchoperateswith spaceconcentration. As the utilization of the input links increase,less and less concentrationis acceptable. If the input channelloadingis high enough,time expansionin the TST switchandspaceexpansionin the STSswitcharerequiredto maintainlow blocking probabilities.Sincetimeexpansioncanbe achievedat lesscostthanspaceexpansion, a TST swirchbecomesmorecosteffectivethanan STSswitchfor high channelutilization.Theimplementation complexitiesof thesetwo system$ arecomparedin Figure 5.24asa functionof theinput utilization. advanAs canbeseenin Figure5.24,TST switcheshavea distinctimplementation tageoverSTSswitcheswhenlargeamountsof traffic arepresent.For smallswitches, complexitiesfavor STSarchitectures. the implementation The choiceof a particular architecture on otherfactorssuchasmodularity,testability, may be moredependent thatgenerallyfavorsanSTSstructureis its relaOneconsideration andexpandability. tively simplercontrolrequirements[8]. For very largeswitcheswith heavytraffic of a TST switchis dominant.Evidenceof this loads,the implementation advantage fact is providedby the No. 4 ESS,a TST structureintroducedin 1976to servicean excessof 100.000voicechannels.

SlSrsotprmlonf .i

STEcomFhxlty .2

{

{

,.,f

lm

$

d

E

g

."d"'

E

I

rsrtlil.rDililofid

.B

6m

A

ztr

,z Udllrdoft ot lnFut drnndr p

Figure 5.24 Complexity comparisonof STS and TST switching structuresfor a blocking probabilityof 0.002.

260

DrerrAL swtrcHtNc lnlct tifte rtrgtc

Spacr BtNg€

$pece stEaE

OuttGt tlm6 rtage

Figure 5.?5 Time*space-space-space-time (TSSST)switchingsfllrcture.

TSSSTSwltches Whenthe spacestageof a TST switchis largeenoughto justiff additionalcontrol complexity,multiple spacestrages can be usedto reducethe total crosspointcount. Figure5.25depictsa TST architecture with a three-stage spaceswitch.Becausethe threemiddle stagesare all spacestages,thi$ structureis sometimesreferredto as a TSSSTswitch.TheEWSD switchof Siemens[9] usesa TSSSTstructure. Theimplementation complexityof a TSSSTswitchcanbe determinedas* Complexity= Nx *

Nu* *Nu, *Nrrc

(5.18)

100

= ZNft+ k (Nln)z whereNyg= numberof crosspoints, = Nsx numberof spacestagecontrolstorebits, = 2k(Nln)llog2(n) + k(Nln)l log2(N/n) NBr = numberof bits in time stages,= 2Nc (8) ffnrc = numberof time stagecontrolstorebits,= 21Vllogz(c) Theprobabilitygraphof a TSSSTswitchis shownin Figure5.26.Noticethatthis diagramis functionallyidenticalto theprobabilitygraphof a five-stagespaceswitch shownin Figure5.10.Usingtheprobabilitygraphof Figure5.26,we candetermine theblockingprobabilityof a TSSSTswitchas

B = { 1 _ (qr)tlr_ 0 _ d)ollt

The assumedconnol orientationsby stagesareoutput,output,output,input, andinput,

(s.l9)

s.4 rwo-DrMENsroNAL swtrcHrNc 261

r

P, = Pla pr=pllodl c= Uo

u-*''

Figure5.26 hobabilitygraphof TSSSTswitch. where4r*I-pt=l-pla Qz=r-Pz=I-plu9

Example5.5. Determinethe implementationcomplexity of a 131,072-channel TSSSTswitchdesignedto providea maximumblockingprobabilityof 0.002under channeloccupancies of 0.7.Assumethe switchservices1024TDM input links with is usedon thespace 128channelson eachlink. Also assumethatunity timeexpansion stages. Solution. The spaceswitchcanbe designedin a varietyof waysdependingon how manylinks areassignedto eacharrayin thefirst (andthird) spacestages.A valueof 32 is chosenas a convenientbinarynumbernearthe theoreticaloptimumpresented earlier(N/2)12.With thisvaluefor n, theonly unknownin Equation5.19is thenumber of center-stage arraysft, which is determinedto be 27 for a blockingprobabilityof 0.0015.Thecomplexityis determinedfrom Equation5.l8 as + (27)QDz=82,944 Nx = (2X1024)(27) = 1,658,880 I 28X5)+ (27)(32)(r28X5) Nnx = 2(27X32X

=2,097,152 Nsr= 2(10?4X128X8) = 1,835,008 Nnrc* 2(1024X128)(7) = 138,854 82,944 + 5,591,040/100 equivalent crosspoints

DIGITALSWITCHING

5.4.3 No.4 ESS Toll Swltch As shownin Fi gtxe5.27, thebasicstructureof a No.4 ESSmaFixis time-space-time with four stagesin thespaceswitch(i.e.,a TSSSST)tl0, I ll. Theinputsto thematrix arel20-channelTDM links formedby multiplexingfive DSI signalstogether.Before theseinputsareinterfacedto the matrix, they arepassedthrougha decorrelatorto distributethe channelsof incomingTDM funk groupsacrossmultipleTDM links into thematrix.Decoruelation is usedbecausetheNo. 4 ESSis a toll switchin which the incomingTDM links represent trunk groupsfrom otherswitchingmachines. In contrastto TDM links of an endoffice switchformedby multiplexingindependent subscriberchannels,the channelsof a TDM trunk grouparenot independently busy.In fact,theactofconcentratingmultiple,independent sourcesontoa fiunk groupcauses highlevelsof correlationbetweentheactivityon theindividualchannels. If thechannels of a trunk group were not decorrelated, they would experiencemuch higher blockingprobabilitiesbecause theywouldall be vying for thesamepathsthroughthe matrix. Decorrelationshufflesthetrunk groupssothealternatepathsavailablefor any particularconnectionaremorelikely to be statisticallyindependent. Noticethat,besidesshufflingchannels, providesspaceexpansion(7 to 8) andtime the decorrelator (l?0 to I28). expansion Themaximumsizeof theNo.4 ESSuses128deconelators with seven12O-channel inputseach.Thusthe maximumchannelcapacityis (128X7X120)= 107,520channels.Thespacestageis a 1024x 1024matrixwith four possiblepathsprovidedduring eachof the 128spacestagetime slots.Theprobabilirygraphof theNo. 4 ESSswirch is shownin Figure5.28,from which thefollowingblockingprobabilityis derived: B = [l - (l -p,Xt -pzXl *pr)]rz8

(5.20)

wherep is theextemalchannelutilization,p1= Q l8)(1201128)p, andp2is theblocking probabilityof thefour-stagespaceswitchderivedin AppendixA as

pz=zplql+rlplfi + s0p1q1+ szplfi+zlp\q\,+Bfrqr+ p!

(5.21)

Example5.6. Determinetheblockingprobabilityandnumberof pathsrhathaveto be searched in a No. 4 ESSwith inlet channeloccupancies of 0.7 and0.9. Solution. The occupancies of the interstagelinls (pzr)aredeterminedto be 0.574 and 0.738,respectively.Using Equation5.21, the corresponding valuesofp2 are determinedas 0.737 and 0.934.Equation5.20 can now be usedto determinethe blocking probabilitiesas 0.fi)2 and 0.56 for inlet occupanciesof 0.7 and 0.9, respectively. The probability that all four spacepathsare busy in a parriculartime slot is I erezh (he baseterm in Equation5.20),which evaluatesto 0.952forp = 0.7 and

R p - 9 ura oO

!1

(} I (t D( +

g.

E

l A x{

r A iA

d

.i E

L

o

i5

:'-(

|J.{ * A

z F FI

vi P EO tl

iE

o F

5 U E E

Ir

o

263

7 264

DIGITAL SWITCHING

+

pt

Figure 5.28 Probabilitygraphof No. 4 ESSmatrix.

0.9955forp - 0.9. Equation5.11now determines the averagenumberof time slots (four pathspertime slot)thatmustbe testedto find anidle pathas2l and9g,respectively. Example5.6 demonstrates that low blockingprobabilitiescan be achievedeven whenindividualpathsarehighly utilizedif thereareenoughalternatives-a feature thatis mostpracticalwith timedivisionswitching.Thisexamplealsodemonstrates the sensitivityof the call procs$singtime asa functionof loadin thatpathsearchtimes morethanquadruplewhenthenodebecomesoverloaded with traffic. Reference[l l] reportsthat the pathsearchtime is only 72voof the total processingload at p = 0.7, but thisexampleimpliesthatit wouldbecomemorethan50zoof theloadatp = 0.9 if otheraspects of call processing werelinearfunctionsof thetraffic volume(a risky assumption). 5.4.4 Sy$rem 75 Dtgitat PBX As originallyreleased, the system75 PBx of AT&T wasa midrangepBX that can suppofr400 stations,200datainstruments, and200trunks[12].Theswitchingmatrix consistsof two bytewideTDM busesa$shownin Figure5.29.Becausethebusesoperateat 2.048MHz, thereare256time slotsperbusperframe.Thusthetotalnumber of timeslotsonthebusesis 5 12,whichsupport256full-duptexconnections.* All ports of theswitchhaveaccessto all time slotson bothbuses.Thedual-busarchitecture allows theuseof slowerelectronicsandprovidesredundancy in thecaseof failure. As long asthereis an idle time slot on eitherbus,any sourcecanbe connected to anydestination. Thustheblockingprobabilityis zerofrom rhestandpoint of matching *Be"uur-

some of the time slots are used for overhead and continuous distribution of various tones, the maximum traffic capacity is 7200 CCS [12], which relates to 200 connections.

SYSTEMS 265 5.5 DIGITAL CROSS-CoNNECT

Switch BUBA $Yvitch Bu$ B

75matrixarchitecture. Figure5.29 System are ifmore Connections lOss.BlockingCAnoccur,hOwever,asa reSUltofcongeStiOn: of this by thenumberof time slots.Blockinganalyses requested thancanbe supported type requirea different form of mathematicsreferredto ascongestiontheory.As pretheorydealswith theprobabilitythattheofferedtrafsentedin Chapter12,congestion fic load exceedssomevalue.Matchingloss,on the otherhand,is concernedwith averageor busy-hourtraffic volumes. As an example,congestionanalysesin Chapter12 showthat if 800 stationsare 37.5Vobwy,on average,theprobabilitythat a stationrequestsservicewhen400 stationsarealreadybusyis lO-s.Thusthe systemis virtuallynonblockingfor voiceapthe blocking areoften full-periodconnections, plications.Becausedataconnections involvin applications maybecomemoresignificant probabilityfor voiceconnections 12 for some problem in Chapter set ing intensivevoicebanddataswitching.Seethe examples.

5.5 DIGITAL CROS$.CONNECTSYSTEMS A DCS is basicatlya digital switchingmatrixwith an operationsinterfacefor setting betweeninput andoutputsignalsor chamels.Instead up relativelystaticconnections response to signalinginformationpertainingto call-byin connections of establishing to networkconin response areestablished DCSconnections requests, call connection patch act as an electronic figurationneeds.The mostbasicfunctionof a DCS is to frameswere facility. Manualcros$-connect panelin lieu of a manualcross-connect points fatransmission between typicallyinstalledin switchingofficesasdemarcation points between cilities and switching machinesand in wire centersas demarcation feedercablesanddistributionfacilities.In bothcases,themajorpurposeof thecrossfor testcircuitsandaccess of transmission connectframewasto allowrearrangement ing thecircuitsin bothdirecfions. environmentwhile Figure5.31 Figure5.30showsa typicalmanualcro$ri-connect with a DCS.Manualcrossconnectsareimpleshowsthesamefunctionimplemented within theDCS mentedwith bridgingclipsandpunch-downwires.Crossconnections console(eitherlocally or reareestablishedby enteringcommandsat themanagement

7

266

DIGITAL SWITCHING Menutl (voicefrequencyl Cros-Connect Freme MGIEllicFociliti$ t

l

\ / \ / -\ rr

-

ltr.r\

Alr iul I I

Local Switching System

Figure5.30 Manualcross-connecr sysrcm. motely).As indicatedin the two figures,a majorfeatureof a DCS is its elimination of back-to-backmultiplexingfunctionswhen crossconnectingindividualchannels within TDM transmission links.Additionaladvantages of a DCS systemwith respect to manualcross-connect $vstems are: r. Automatic RecordKeeping.Becausethe crossconnectsare under'processor control,circuit connectionreportsarereadilyavailablethroughthemanagement interface.In contrast,recordsin manualsystemswereinherentlyerrorproneand oftenout of date.

Conlole

Clnnnrl Benk

Chann€l Switchcd Treff ic

'*l

Frcilitiet

Ti6 linrs Fortign Exchange

IDLC

M13

DCS lllol

::

I

Local (DigitBll Swltching Syrtem

Circuit SwitchedTraff ic

Flgure 5.31 Elecnonicdigitalcross-connect system.

SYSTEMS 267 CROSS-CONNECT 5.5 DIGITAL 2. Remoteand Rapid Provisioning. Provisioning is the basic procertsof providing (or discontinuing) service to a particular customer. The basic operations involved are outside-plant cross connections, inside-plant cross connections, configuration changes in switching system data base, and customer record updatesin business(billing) $ystems.Obviously, the more theseprocesseswere automated, the faster and more accurately they could be performed. 3. AutomatedTest Access.Testing analogcircuits at a manualcross-connectframe involves physically breaking the connection (by removing bridging clips) and attaching the test equipment to the side of the circuit to be tested.All manual operationsare eliminated with an electronic patch panel by entering commands at the management console to connect the desired test tone$ and (DSP) measurementchannelsto the circuit under test. Figure 5.31 depicts two type$ of network traffic: circuit-switched traffic and channel-switched trffic. Circuit-switched traffic representslocally switched fraffic (DSO circuits typically) and channel switched-traffic refers to leased line equivalents of digital channels.Channel-switchedhaffic might terminate at another public network switching office as in a foreign exchange(FX) circuit or at a distant cu$tomerpremise as a tie line of a private network. In the latter case,more than one DSOchannel might be concatenatedtogether to form a single higher rate channel referred to as a fractional TI circuit or M x 64-kbps channel. Channel-switchedservicestypically account for over one-half of the ffansmissionbandwidth betweenU.S. metropolitan offices [13]' The processof separatingchannel-switchedservicesfrom circuit-switched servicesis "grooming." Figure 5.30 also showsthat universaldigital loop caroften referred to as rier (UDLC) becomesintegrated digital loop carrier (IDLC) in a digital environment (Figure 5.31). (Chapter I I describesDLC $y$temsin more detail.)

5.5.1 Consolidationand Segregation Two basically different DCS grooming functions are depicted in Figures 5.32 and 5.33: consolidation and segregation.When multiple-accesslines carrying traffic destined to a coillmon distant node are partially filled, the per-channelco$tsof transpon to the remote node can be reducedby consolidating the traffic. Conversely,when different types oftraffic originate at a single location, it is desirableto allow a single faFrrthlly Filled Locsl Acc$6r

Totally FillGd Nstvvork

50t6

mt\*l-l 30e6--{

*E;

30!_.+{_l

rm*

o"'F-

_

t0096

Figure5.32 Consolidation.

268

DIGITALSWITCHING DDS Nttrvork Tio Llnal Fortign Exchenga Losil Swltching

Figure5.33 Distribution. cility (e.g.,a T1 line)to carryall typesof traffic andsegregate it at theDCS.Examples of suchtraffic arecircuit-switched channels, DDS channels,tie lines,multidropdata circuits,FX circuits,or otherspecialservicesttratmight be availableonly at another office. 5.5.2 DG$ Hierarchy The cross-connect systemdepictedin Figure5.31terminatesDSI signalsandinterchangesDSOsignals,which leadsto the designationDCS l/0. similarly, a digital cross-connect systemthatterminatesDS3 signalsandrearranges DSl signalswithin theDS3sis referredto asa DCS3/1.If a DCS,sucha$a DCS3/0,providesrearrangementof lowerlevelsignals,rruchasDSOs,it doesnotnecessarily meanthatit alsoprovides cross-connect servicesfor intermediate-level signals,such as Dsls. cross connectionof DSls requirestransparent mappingof theentirel.5zt4Mbps,whichincludestheframingchannel-a capabilitythatmaynorbeprovidedwhen64-kbpsDSO channelsarethe primaryfocus.when a DCS providesmultiplelevelsof crossconnects,theintermediate levelsaresometimes designated-asin DCS 3/l/0 for terminatingDS3sand rearrangingDSls and DS0s.In most cases,it is not necessarJ to providecross-connect servicesat all levelsofthe digitalsignalhierarchyfor all ofthe terminations.Figure5.34 depictsa DCS hierarchythat providesrearrangement of lowerlevelsignalson only subsets of higherlevelsignals.Thehigherlevel(e.g.,Ds3) signalsthatarenot crossconnected to a lowerlevelDCSmaybeunchannelized highspeed(44.736-Mbps) signalsor transitDS3sthatterminateon a DCS 3/l or 3/0 at a

Figure 5.34 DCS hierarchy.

SYSTEMS 269 5.5 DIGITAL CROSS-CONNECT

distantlocation.Theprimarypurposeof theDCS3/3is to providenetworkrestoration or protectionswitchingof the DS3 signalsandto possiblyprovidetime-of-dayrearin Chaplinks described SONBTtransmission rangement of trunkgroupassignments. at muchhigherratesproducingadditionalDCS layers. ter 8 arecrossconnected 5"5.3 IntegratedCrose-ConnectEquipment Figure5.31 depictsdistinctmultiplexingequipmentand two distinctand colocated swirchingsystems:the DCS and the local digital switch.AlthoughDCS functions within thepublic networkaretraditionallyimplementedwith separateequipment,privatenetworksoftenutilize equipmentthatprovidesintegratedmultiplexingandcrossconnectfunctions.This equipmenthas evolvedfrom CPE-basedTl multiplexers becauseTSI circuitsare insignificanthardwareadditionsto Tl multiplexinghardware.Suchequipmentarevariouslyrefenedto asintelligentmultiplexers[14],nodal functionsarealsoincolpoprocessors, or networkingTl multiplexers.Cross-connect again,thecostof theTSI functionis vidually ratedinto newerDLC systemsbecause, nil. functionswith higherleveldigitalsignals(suchas Integrationof cross-connect a DS3) hasnot occurredbecausethe mannerin which the higherlevel signalsare multiplexed(describedin Chapter7) is not amenablefor direct terminationon a switchingsystem.The newerform of (synchronous)multiplexing,as specifiedby the SONET standarddescribedin Chapter8, allows integrationof higher level systemsfor both public andprivateequipdigital multiplexerswith cross-connect ment. functionsare basicallynothingmore than "pegged"or Althoughcross-connect "nailed-up"circuit-switchedconnections, the two functionsaretraditionallyimplementedseparatelybecauseof the following differencesin the applicationrequirements: l. A DCS needsto be strictly nonblockingat the DSOlevel,which is generally in largepublicnetworkswitches. uneconomical of a 1.544-Mbpssignalrequirestransportof the cross-connection 2. Transparent framingbit that is not possiblein typicaldigital circuit switchesbecauseDSI with framingin the DSI signalsarechannelized interfaceequipmenta$sumes the 193rdbit. 3. DCS functionality does not involve processingof signalingbits so fully functionalcircuitswitchinterfaceshaveexcesscostswhenu$edasinterfacesto systems. cross*connect DSOchannels(fractionalTl channels) 4. Crossconnectingmultiple,concatenated datastream.Althoughthe order requiresmaintainingorderin the concatenated can always be maintainedby careful mappingof the individual 64-kbps connections,it i$ a function that is typically not provided in DSO circuit switching software.Figure 5.35 shows how the order of bytes in the by ill-chosen64-kbpsconnections' datastreamcanbe transposed concatenated

27O

DIGITAL SwITcHING

lffimlng Oflr

Ditr Outgolng

Ch{nnel ConnGtloni Ero At to 8t , Az to Bil As to Bl

Figure 5.35 Transposition of data in concatenatedtime slots.

5.6 DIGITALSWITCHING IN AN ANALOGENVIRONMENT whendigitalendofficeswitches (or PBXs)areinstalled in ananalogenvironment, theanaloginterfacesarenecessarily unchanged. Althoughthedigital switchmay interfacewith digital subscribercanier or digital fiber feedersystem$,thesesysrems merelyextendthe analoginterfacepoint closerto the subscriber.This sectiondescribesthebasicconsiderations ofusing digital switchingin suchan analogenvironment.Chapterl l describes digiralendoffice switchingwith digital subscriber loops in relationto the integratedservicesdigital network. 5.6.1 Zero-LossSwitchlng As alreadymentioned,a well-designeddigital transmissionand switchingsystem addsno appreciable degradation to the end-to-endquality ofdigitized voice.In particular,thederivedanalogsignalcomingout of a decodercanbe adjustedto thesame levelasthatpresented to theencoderat thefar end.Curiously,zero-losstransmission presentssomesignificantproblemswhendigitalswitchingis usedin theanalogfiansmissionenvironmentof a class5 centraloffice. Analogendofficeswitchesaretwo-wireswitchesdesigned to interconnectbidirectionaltwo-wirecustomerloops.The voicedigitizationprocess,however,inherently requiresseparation of the go andreturnsignalpathsinvolvedin a connection.Thus, a class5 digital endoffice or digitalPBx mustbe a four-wireswitch.when inserted into a two-wireanalogenvironment,hybridsarerequiredto separate the two directionsof transmission. As shownin Figure5.36,hybridsateachendof ttreinternaldig! tal subnetworkproducea four-wirecircuit with thepotentialfor echoesandsinging. (Amplifiersareshownin conjunctionwith theencoders to offsetforwardpathlossinherentin thehybrids.)Instabilityarisesasa resultof impedance mismatches at thehybridscausingunwantedcouplingfrom the receiveportionto the transmitportionof the four-wireconnection.Impedance mismatches occurbecauseof the variabilityin the lengthsandwire sizesofthe subscriber loops.In pafticular,loadedandunloaded wire pairshavemarkedlydifferentimpedance characteristics. Theinstabilityproblemsarecompounded by certainamountsof artificialdelaythat are requiredin a digital time division switch.Althoughthe delaythrougha digital switch(severalhundredmicroseconds, typically)is basicallyunnoticeable to a user,

ENVIRONMENT 271 INANANALOG SWITCHING 5,6 DIGITAL

I I \

underirNbh\ trEnfhybrid coulllns

I /

interfaces. digitalswitchwithtwo-wireanalog Flgure5.36 Four-wire

delay theequivalentof asmuchas30 to 40 milesof wire.This increased it represents be outside might otherwise frequencies that oscillation hasthe effectof loweringthe filters' thevoicebandandeffectivelybe removedby encoder/decoder used at the interfaceto the As mentionedin Chapterl, hybridsweretraditionally mediumis basicallyfour-wire.In theseinstances, toll networkwherethetransmission amount theinstabilityof thefour-wirecircuitwascontrolledby designinga prescribed pathof shofiertoll network (by wayof netloss)into thetransmission of netaftenuation circuits.On thelongercircuits,theechoesandsingingareeliminatedby echosuppresClass5 digital switchescouldpreventinstabilityin the same $orsor echocancelers. process' manner:by designinga certainamountof lossinto the encoding/decoding path (approximately in each 2-3 dB amount attenuation of signal While thenecessary to some toll added connections loss the [5]) couldbe toleratedon localconnections, wouldbe unacceptably large. attenuation Fromtheforegoingdiscussionit canbe seenthatthe useof selectable lossis insefiedinto thetalkis onesolutionto theinstabilityproblems.Thenecessary calls,whichalreadyhave but not u$edon long-distance ing pathfor localconnections atthehyinto them.A secondsolutioninvolvesmatchingtheimpedance lossdesigned bridsmoreclosely.Beforethe adventof DSPnearlycompleteeliminationof the unwantedcouplingwas prohibitivelyexpensive.However,adequateisolationof the pathscouldbe accomplished with just two differentmatchingnetworks: transmission matchinghasbeen onefor loadedloopsandonefor unloadedloops[61. Impedance (DSP) that containtrainable simplifiedby the useof advancedinterfaceelectronics matchingcircuitsessenmatchingcircuitry.Theimpedance andautomaticimpedance echocancelers[17]. tially representshort-delay avoidsinstability Notice that an all-digital network(with four-wiretelephones) problemsbecausethereareno two-wireanaloglines.Voice is digitizedat the telepathfrom the receivesignalall the way to the destination phoneandusesa separate telephone.Thus an endrto-endfour-wire circuit completelyeliminatesechoesand to operateon a zero*lossbasis' singing,allowingall connections

272

DIGITALSWITCHING

5.6.2 BORSCHT In ChapterI thebasicfunctionalrequirements ofthe subscriber loop interfacearedescribed.Theserequirementsarerepeatedherewith two additionalrequirementsfor a digital switch interface:codingand hybrid. The completelist of interfacerequirementsis unaffectionately knownasBORSCHT[18]: B : Batteryfeed o: Overvoltageprotection R: Ringing S : Supervision C; Coding H: Hybrid T: Test As mentionedin chapter l, thehigh-voltage,low-resistance, andcurrentrequirementsof manyof thesefunctionsareparticularlyburdensome to integrated circuitimplementations. First-generation digital end office switchesreducedthe termination costsby usinganalogswitching(concentrators) to commoncodecs.The DMS-100 [19] of NorthernTelecomandtheNo. 5 ESSof AT&T [20] useanalogconcentration at the periphery.Integratedcircuit manufacturers haveworkeddiligentlyto implement the BORSCHTfunctionsin what is calleda subscriberloop interfacecircuit (sl-rc). Perline sLICs allow implemenrationof perJine BORSCHT functions. sLICs can be usedin PBx applicationswith a minimumof otherextemalcomponents.In cenffaloffice applications,wherelightningprotectionand test accessare moredemanding,sLICs typicallyneedothercomponents for a completeinterface. 5.6.3 Conferencing In ananalognetworkconference callsareestablished by merelyaddingindividualsignalstogetherusinga conference bridge.Iftwo peopletalk at once,theirspeechis superposed.Furthermore,an active talker can hearif anotherconfereebeginstalking. Naturally,thesametechniquecanbeusedin a digitalswitchif thesignalsarefirst convertedto analog,added,andthenconvertedbackto digital. As describedin chapter3,1t?55andtheA-law (ITU) codeweredesignedwith the specificpropertyof beingeasilydigitally linearizable(EDL). with this property,the additionfunctioncanbeperformeddigitallyby first convertingall codesto linearformats,addingthem,andthenconvertingbackto compre$sed formats.To theuser,the operationis identicalto thecustomaryanalogsummation. For a conference involving N conferees,N separatesummationsmust be performed,one for eachconfereeand containingall signalsbut his own.For a descriptionofthe conferencing algorithmin the system75 PBX of AT&T, seereference[12]. For moregeneraldescriptionsof conferencing implementations in digital switches,seereference[21].

REFERENCES273 Another conferencing technique involves monitoring the activity of all conferees and switching the digital signal of the loudest talker to all others. Although this technique is functionally different from a customaryanalog conferencebridge, it is advantageous for large conferencesbecausethe idle channel noise of the inactive talkers does not get addedinto the output of the conferencebridge. High-quality conference circuits also include echo cancelersso higher signal powers can be provided.

REFERENCES I

M. R. Aaron, "Digital Communications-The Silent (R)evolution?" IEEE Magazine,Jan.1979,pp. l6-26. Communications "A Study of Non-Blocking Switching Networks," Bell SystemTechnical C. Clos, Joumal, Mar. 1953, pp.4O6-424. "Analysis of Switching Networks," Bell SystemTechnical Journal, Nov. C. Y. Lee,

pp.1287-1315. 1955, "A Studyof Congestion No. 48, EricssonTechniques, in Link Systems," C. Iacobaeus, Stockholm, 1950,pp. 1-70.

I

A TimeFor Innovation,Merle Telecommunications, A. A. CollinsandR. D. Pedersen, TX, 1973. Dallas. CollinsFoundation. NetvvorkProtacols, Modeling and Analysis' M, Schwartz, Telecommunications ReadingMA, 1987. Addison-Wesley, "New Time DivisionSwitchingUnits T. E. Browne,D. J. Wadsworth,andR. K. York, Joumal,Feb.1969,pp.443-476, for No. l0l ESS,"Aell SystemTechnical "Telephones 1979,pp.51-60. Go Digital,"IEEESpectrurn,Oct. S. G. Pitroda,

"EWSD;WhereIt Is," 'lieruen's TelecomRepon,Vol. 12,No. 2-3' 1989' 9 N. Skaperda, pp.56-59. l 0 J. H. Huttenhoff.J. Janik.G. D. Johnson,W. R. Schleicher,M. F' Slana,and F' H' TechnicalJournal,Sept.1977' Bell System Tendick,"No. 4 ESS:PeripheralSystems," pp. 1029-1042. u A, E. RitchieandL. S.Tuomenoksa,"No. 4 ESS:SystemObjectivesandOrganization," Journal,1977,pp. 1017-1027. BeIlSystemTechnital t2 L. A. Baxter,P. R. Berkowitz,C. A. Buzzard,J. J. Horenkamp,and F' E' Wyatt' "system 75: Communications and Control Architecture,"AT&T TethnicalJoumal, Jan.1985,pp. 153-173. "hoviding and Managing l 3 R. K. Berman,R, W. Lawrence,and P. C. Whitehead, ChannelSwitchedServicesin an IntelligentISDN Environment,"IEEE Clobecom, 1987,pp.4.6.1-4.6.5. t4 J. L, MelsaandH. R. Scull,"The Applicationof lntelligentT1 Multiplexersin Hybrid 1989,pp. 15.3.1-15'3'6. andISDNNetworks,"IEEEGlobecom, "Transmission Planningfor anEvolvingLocalSwitchedDigital Network," 1 5 J. L. Neigh, July 1979,pp' 1019-1024. on Communications, IEEETransactions "Tnro in Digital Class5 loss Considerations l 6 R. Bunker,F. Scida,and R. McCabe, July 1979,pp' 1013-1018. on Communications, Office,"IEEE Transactions

274

DIGITAL SWITCHING

l 7 D, L. Duttweiler, "A Twelve Channel Echo Cancellor,', IEEE Transactions on Communications,May 1978, pp. 647-653. 1 8 F. D. Reese,"Memo to Management-You Must Appraise How New Technology Fits Customers," TelephoneEngineering and Management, Oct. I, 1975, pp. I 16-121. l 9 J. B. Terry, D. R. Younge, and R. T. Matsunaga, "A SubscriberLine Interface for the DMs-l00Digitalswitch," IEEENationalrelecommunicationsconference,lgTg,pp. 28.3.1-28.3.6. 20 D. L. Camey, J. I. Cochrane, L. J. Gitten, E. M. Pretl" and R. Staehler. ..The 5 ESS Switching System: Architectural Overview," AZ& T Technical Jorrrual, Aug. 19g5. 2 l R. J. D'Ortenzio, "Conferencing Fundamentalsfor Digital PABX Equipments,',IEEE International Conferenceon Communications, 1979, pp. 2.5-2g-2.5-36.

PROBLEMS 5.1 How manyfour-wirevoice circuit connections can be providedby a bidirectionalPAM switchingbusif theminimumachievable pulsewidth is 250nsec? 5.2 TheTS switchof Figure5.19usesDSI signalson eachrDM rink. what is the implementationcomplexityif groupsof five DSI inputsarefirst multiplexedto form 16input links with 120channelson eachlink? 5.3 DeterminetheLeegraphandJacobaeus blockingprobabilitiesof thefirst switch in Table5.5 (ft = 6) if two inletsout of 16 becomeconnectedto l-erlangsubscribers.(Possiblytheselinescouldbe dial-upporrsto a compurer.)(Hint:Two inletsarepermanently busybut the remaininginletscontinueto be O.l-erlang porrs.) 5.4 RepeatProblem5.3but insteadof two inletsbeingbusyassumethattwo of the outputlinks of thefirsr-stagemodulehavefailed. 5.5 How manycrosspoints areneededin a l024line, three-stage spaceswirchif the inputloadingis six common-channel signalsper line andthemaximumacceptableblockingprobability(usinga Leegraphanalysis)is 0.005? (a) If n = 16,N/n = 64 (b)If n =32,N/n=32 (c) tf n = 64,N/n = 16 5.6 what is the(Leegraph)blockingprobabilityof theTS switchin Figure5.l9 for channelloadingof 0.2erlang? (a) AssumeeachTDM inputis a 24-channel interofficetrunk group. (b) AssumetheTDM inputsarederivedfrom24-channel bankswith eachanalog interfaceconnected to a dedicated0.2-erlangline. 5.7 DesignansTS switchfor 128primaryTDM signalsof theccITT hierarchy(30 perinput).Blockingshouldbelessthan0.002andtheloadingis voicechannels 0.2 erlangper channel.How manytime slot interchange modulesareneeded? Whatis thecomplexiryof the switch? 5.8 RepeatProblem5.7 for a TST design.

PROBLEMS 275

5.9 Determinethe numberof crosspointsNx and the total numberof memorybits Ns requiredfor a TST switchdefinedasfollows;numberof linesis 32, single$tagespaceswitch,numberof channelsper ftameis 30, andtime expansionis 2. 5.10 Whatis theblockingprobabilityof theswitchin Problem5.9if thechannelloading is 0.9 erlangperchannel? 5.ll How manybits of memoryareneededin a time slot interchangecircuit for a 60channelsignalwith 9 bitsper time slot? controlin the first stageand 5.12 DeriveEquation5.15 assuminginput-associated controlin thethird stage. output-associated 5.13 DeriveEquation5.18assumingall controlorientationsareoutputassociated.

DIGITALMODUL,ATION ANDRADIOSYSTEMS variousmeansof encodingdigital informationfor transmission Chapter4 discusses links.In the contextof this chapterthe dataen(or fiber) transmission overwireline to asbasebandcoding.To impressthe same 4 is referred in Chapter codingdescribed signalis commonlyused a baseband informationontoa carrierfor radiotratrsmisSiOn, sigof abaseband specfum the dc-centered shifts carrier. modulate Modulation the to binary transemphasizes 4, which to Chapter In contrast frequency. nal to thecarrier multileveldigital modulationto achievehigh data mission,this chapteremphasizes of a radio channelor analogvoicebanddata defined bandwidth rateswirhin therigidly can most multilevelmodulationtechniques chapter, later in this circuit.As described baseband multilevel one of the with a carrier modulating by directly be implemented signalsdescribedin Chapter4. describedin this chaptercover Applicationsfor the digitalmodulationtechniques systems, andvoicebandmodems. cellular digital point-to-pointmicrowavesy$tems, alongheavyffaffic longer used are no systems Atthoughpoint-to-pointmicrowave applicationsin in thin-route utilized are still public they network, routesof ttreU.S. prohibitive' Furis right-of-way of a fiber where cost the U.S. andaroundthe world private networks in used is still commonly point-to-point microwave thermore,digital sucha$digitalvideo (alongwith someold analogsystems)andin specialapplications basestation. digital cellular from a transportfrom a studioto a transmitteror to and modulation uses mobile units stations and betweenbase Digital cellulartransmission point-to-point sysdigital original techniquesthat areidenticalto that usedin some TEInS.

theFCCestablished Whendigitalpoint-to-pointradioswerefirst beingdeveloped, in the UnitedStatesto encertaindatarates[1] thatmustbe achievedby theseradios Basically, spectrum. surea minimum utilizationefficiencyof the radio frequency of same number the theseminimumrates(shownin Table6.1)specifyapproximately time in use at the radios voicecircuits(at 64 kbps)aswereavailablein theanalogFM (19?4).Althoughindividualvoicechannelsin a FDM signaloccupy4 kHz of bandwidth, frequencymodulationtypically expandsthe signalbandwidthby a factorof 277

278

DIGITAL MoDUI.ATIoN ANDRADIo SYSTEMS

TABLE6.1 InformatlonDensltlegFequiredby Fcc for common-carrlerMicrowsve ChanneleUsing64 kbps per VolceCircuit Band (GHz)

2.110-2.130 3.700-4.200 5.9?5-6.425 6.52ffi.87s 1Q.7-11.7

ChannelBW(MHz)

BitRate(Mbps)

Density(bps/Hz)

3.5 20 30 30 40

6.144 73.7 73.7 73.7 73.7

1,8 3.7 2.5 2.5 1.8

about4, dependingon the amountof FM deviationapplied.Thusthedigital systems wererequiredto competewith an equivalentvoicechannelbandwidthof about 16 kHz' If the digitalradioswereforcedto competewith analogsingle-sideband radios [2, 3] introducedin theearly 1980s,theycouldnot havedonesowithoutusinglower bit ratevoiceencoding. over andabovetheminimumbit ratesestablished by theFCC,competitionamong digital microwavemanufacturersand the economicsof manimizing the numberof voicecircuitsper radio stimulatedthe development of moreadvanceddigitalmodulationtechniques to achieveevengreatertransmission rates.As an exampleof signal processing advances, the6G150MB digitalradioof NEC carriesa 155-Mbpspayload in 30 MHz of bandwidth-an informationdensityofjust over5 bps/Hz.Thefirst part of this chapterdescribesbasicmodulationtechniquesandthe transmission efficienciestheyprovide.The lattersectionsdescriberadiosystemdesignconsiderations for point-to-pointmicrowaverelaysystem$. lnformation Density A usefulparameter for characterizing thebandwidthefficiencyof a digitalmodulation systemis theinformationdensity,definedas

s-#

(6.1)

where R = data rate in bits per second BW = bandwidth of digital signal in hertz The units of information density are sometimesreferred to loosely as bits perhertz. However, as defined in Equation 6.1, the units shourdbe bits per secondper hertz or simply bits per cycle. Since bits per secondper hertz conveysthe natureof information density more completely, it is the preferred unit. The bandwidth factor in Equation 6.1 can be defined in a variety of ways. In theoretical studiesthe bandwidth of a signal is usually determinedas the width of the ideal filter used to strictly bandlimit the signal (i.e., the Nyquist bandwidth). hr practical systems,where the spectrumcan never be strictly bandlimited, the bandwidth is more

279 MODULATION 6.1 DIGITAL difficult to define.In radiosystemsa channelis usuallyspecifiedwith a certainminiatthebandedges.In thiscase,theinformationdensityis easily mumsignalattenuation determinedasthe bit rateof a signalmeetingtheserequirementsdividedby the allotted bandwidth. areprovided,however,thebandwidth Whenno particulalemissionspecifications moreor lessarbitrarymanner' in another requiredby a digital signalmustbe defined sincethe signalspectrummay drop using a 3-dBbandwidthis usuallyinappropriate ratherslowly andtherebyallow.significantamountsof energyto spill overinto adjacent channels.Using a greaterattenuationlevel to definethe bandedgeswould be more appropriatebut is still somewhatarbitrary.A generallymorerelevantcriterion for practicalsystemsdefinesthebandwidthof a signalasthechamelspacingrequired into identicaladjacentchannels. to achievea specifiedmaximumlevelof interference used in Europe,is the 997opower commonly of bandwidth, Yet anotherdefinition bandwidth. The theoreticalmaximuminformationdensityfor binarysignalingis 2 bps/tlz for signal'If a foura two-levelline codeor 1 bps/flz for a modulateddouble-sideband levelline codeis usedto achieve2 bits per signalinterval,thetheoreticalinformation carriersignal. densityis 4 bps4lz for theline code,or ZbpslHzfor a double-sideband signalscanbe doubledby usingsingleTheinformationdensifl of amplitude-modulated sidebandtransmissionto effectivelyachievethe sameefficiencyas the line codes (baseband signals). line datarateon a dial-upanalogtelephone practical example,a representative As a rate). signaling 2400-Hz and a per signal interval (usually l2 bits with is 28,800bps Sincetheusablebandwidthof a telephonechannelis approximately3 kHz, a typical informationdensityof a dial-upline is 9.6 bps/H2.. As a startingpoint for digital microwaveradios,Table6.1 includestheminimum bands informationdensitiesrequiredin common*carrier bit ratesandcorresponding of the UnitedStates.Competitionandcompatibilitywith the digital hierarchyled to efficiencies. realizationsof evengreatertransmission

6.1 DIGITALMODULATION in two imporon a radiolink differsfrom wirelinetransmission Digital transmission mannerto in some a carrier must modulate information tantregards.First,the digital proce$s can modulation (RF) the In many cases signal. producethe radio frequency (NRZ) by a nonreturn-to-zero be viewedasa specialform of amplitudemodulation signalthatamplitudemodua baseband line codesignal.Thustheline coderepre$ents latesthe carrierin the transmitterandis reproducedby demodulationin the receiver. Representingthe modulationprocessin this mannerhas the advantagethat the RF spectracanbe determinedby merelytranslatingtheline code(baseband)spectrato the carrierfrequencY. $elected *Slightly

higher rates are sometimes achieved with V.34 modems. Rates approaching 64 kbps are also possible in special (V.90) applications described in Chapter 1I .

280

DtctrALMoDULATtoN ANDRADIosysrEMS

second,a radio link differs from wireline tansmissiondue to the necessityof strictly bandlimitingthe transmittedsignalsto preventinterferenceinto otherchannels.Althoughwirelinetransmission links automatically filter theline signalto some extent,explicit filter requirements sometime$ occuronly in the receiversto rejectas muchnoiseaspossible.*Sinceradiolinks arebandlimitedinthehansmitterandzpise filter functionmustbe partitionedbetweenthe filtered in thereceiver,theend-to-end two ends.Figure6.I showsa block diagramof a radio link showingrepresentative basebandand RF waveformsalong with corresponding frequencyspectrums.For modulation,Figure6.I showsmultiplicationof thecarrierby thebaseband waveform. Modulationin a digitalradiomustultimatelybe designedin conjunctionwith the firlterfunctions.For easeof description,however,modulationtechniques areconsideredfirst. Later on, the filtering requirements of eachtype of modulationare discussed.Figure6.1alsoshowsthebaseband encodingproce$s to beindependent ofthe modulationprocess.For mostof themodulationtechniques describedin this chapter thisview is appropriate. In thecaseof trellis-coded modulation(TCM) andcarrierless amplitudeandphase(cAP) modulation,described later,encodingis incorporated into themodulationprocess. 6.1.1 AmplitudeModulailon Historically,the simplestform of modulationto generateand detectis amplitude modulation(AM). A conceptual illustrationof amplitudemodulationis shownin Figure 6.2.Themathematical definitionis x(t)=[l +am,(r)]cosro"t

(6.2)

wherea = modulationindex(0 < a S l) mn$)=n-level, symmetricNRZ baseband signalnormalizedto maximum amplitudeof I 0[ = is the radiancarrierfrequency,= 21;1 Amplitudemodulationis an exampleof a classof specialmodulationtechniques referredto as"linear modulation."Linearmodulationimpliesthat the spectrumof the modulatedsignalis obtainedby ffanslatingthebaseband spe{trumto theselected carrier frequencyband.As shownin Figure6.2, amplitudemodulationby a two-level digitalbaseband signalessentiallytranslates thesin(x/x baseband specrrumup to the carier frequencyf,.other linearmodulationtechniques aredouble-sideband modulation,single-sideband modulation,andvestigial-sideband modulation. Inspectionof Equation6.2or Figure6.2indicatestharif lfi)7o modulationis used (a = I ), no carier is producedfor a logic0. Forobviousreasons, thisform of amplitude modulationis oftenreferredto ason-off keying,oramplitudeshifi keyrng(ASK). As 'Adherence

to EMI compatibility specifications defined by the FCC (Paxt 15) often requires some amount of transmit filtering in on-premiseswireline transmission systems.

!

c CI -g

tE

j j

,f

6

t

f

?i o

B ,6

E!

o

E

ct

{)

f

bI) 6

E ,t4

o

j

o E

EE F F

E

(h

\a P HO

4

rE

FB

281

282

DIGITALMODULATIONAND RADIOSYSTEMS

*kl = [t *an,(rll coro.r

] -.t

Figure 6.2 Digital binaryamplitudemodulation. shown in Figure 6.3, on-off keying can be obtainedby direct multiplication of a carrier with the two-level (unipolar) line code describedin Chapter 4. Amplitude-modulated signalsare usuarly demodulatedwith a simple envelopedetector. The cost effectivenessof this detectoris the basic reasonthat commercial analog broadcastingusesamplitude modulation. unfortunately, the eruorperformanceof digital amplitude modulation in general,and envelopedetection in particular, is inferior to other forms of digital modulation and detection. For this reason, amplitude modulation is used only where the cost of the receiver is a significant consideration. Digital microwave links and digital cellular systemsuse other forms of modulation and demodulation to minimize the enor rate for a given signal-to-noise ratio. Conventional amplitude modulation provides suboptimum error performance for two basic reasons.First, if a < r, a discrete (informationless) spectral line occurs at the carrier ftequency. Although the existenceof this spectralline simplifies canier recovery' it increasesthe transmitter power without aiding discrimination between information signals. with 1007omodulation (on-off keying) no line specfraare produced, but the system is still inefficient in its use of transmifted power. As discussedin chapter 4 for two-level line coding, the maximum use of transmittedpower is achievedwhen one signal is the negativeof the other. Thus, a seconddeficiency of amplitude modulation arisesbecausea 0 signal is not the exact negative ofa I signal. To achieve optimum performance,a symmetric twolevel basebandsignal should directly modulate (multiply) the carier. As shown in Figure 6.4, this form of modulation producestwo iden-

6.1 DtctrALMoDULATIoN283

o

frF+

#'q#q##t+t# +Sffiffiffi Figure 6.4

Phase-reversal keying.

retical signalsexceptfor a 180"phasereversal.Hencethis modulationis sometimes keying phase shift ferredto asphasereversalkeying(PRK),or moreoften,two-level (2-PSK),f-Ioticethat PRK cannotbe detectedby envelopedetection'Instead,a PRK signalmustbe detectedby comparingit to a coherentcarier reference. Coherentdetectioninvolvescomparingthe incomingsignalto a local carriersynchronizedin phaseto that usedat thetransmitter.With conventionalamplitudemodulation thereis no informationin the phase.With PRK signaling,however,all of the informationis in the phase.The useof a coherentreferencefor PRK signalsallows Thebasicequationdefininga PRK signalis optimum(antipodat)elror rateperformance. oct x(t1= r77r111cos

(6'3)

signallm2$)= +1 for a I andm2(t)= -l for wherern2(t)is thebinarydigitalbaseband equationis a 01.Thedemodulation y(t) =.r(t)[2cosoct] = lmr(t) cosoctl [2 cosro"tl * m?(t)+ m2(t)cos2ro.t

(6.4)

term is removedwith a low-passfilter' wherethe double-frequency proce$$ As indicatedin Figure6.4 andEquation6.4, the coherentdemodulation the datadetect processed further to be produces -Since a symmetrictwo-level signalthat can it must code, to a line equivalent essence, signalis, in baseband thedemodulated in Chapter4. [n particular,theremustbe asdiscussed includetiming considerations enoughsignal transitionsin the basebandsignal to allow recoveryof a sample clock. However,the basebandsignalneednot precludedc wandersincedc levels

284

DIGITALMoDULATIoNANDHADIoSYSTEMS o

r

r

0

0

t

0

t

t

0

0

l

o

t

t

0

0

l

trffi#hirfut\ddill Eq# Phsre dctoclol

Modulator

Low pssg fllur

Denrodulstol

Figure6.5 Binaryfrequency shiftkeying. aretranslatedto thecarrierfrequencyandpassedadequately by thedouble-sideband system.6.1.2 Frequency $hift Keytng In additionto an inefficientuseof signalpower,conventionalamplitude-modulated signals(not PRK) haveoneotherundesirable characteristic. By definition,an amplitude-modulated signalusesmultiplesignallevels,which impliesAM is quitevulnerable to si8nal saturationthat narrowsthe distancebetweenamplitudelevels and producesspectrumspreading. A commonsourceof saturation in a radiosyslemoccurs in the outputpoweramplifierof the transmitter.In mostcasesoutputamplifiersare operatedat lessthanmaximumpowerto eliminatesaturationandothernonlinearities, sotheycanaccommodate amplitude-modulated signals[4]. pM u.leconstant-amplitude systems FM or signalsnot adversely __Angle-modulated affectedby signalsaturation.HenceFM and pM canbe transmitteaat higherpowei levelsthanAM systems.The ability to usesaturatingpoweramplifiersis oneof the rea$ons why FM wasoriginallychosenfor analogmicrowaveradios.This sectiondiscussesdigitalfrequencymodulation,commonlyreferredto asfrequencyshift keying (FSK).Thenext sectiondiscusses digitalphasemodulation,commonlyrefenedto as phaseshiftkeying(PSK).Both systemsprovidea consrant-amplitude signal.systems usingconstant-amplitude carriersarealsoreferredto asconstant-envelJpe sysrems. Thegeneralexpression for ann-aryFSK signalis

.(,):*,[[*..9J'] 'A

(6.5)

single-sideband system does not pass dc energy. Thus if single-sideband modulation rs used, the basebandsignal must exclude.dc energy from its spectrum. Some double-sidebandsystems might also require the elimination of baseband dc energy so that a carrier tone can be inserted intl the center of tlre passband without affecting the signal.

MODULATION 285 6,1 DIGITAL

wherero"= radiancenterfrequency signal ln, = n-levelNRZ digitalbaseband signals = between frequency radian difference Aro A typicalbinaryFSK signalis shownin Figure6.5 alongwith a simple(but lowperformance)meansof implementingthemodulatoranddemodulator.Themodulator oscillator(VCO)thatis biasedto producethecenterfrequency is a voltage-controlled whenno modulationis applied.Theamplitudea of thesymmetrictwo-levelbaseband signalproducesa frequencydeviationof +Arr/2for a I and -L{dy?for a 0. loop:a VCO, a phasedetector, asa phase-locked is implemented Thedemodulator thedifferencein phasebetweentheFSK anda loop filter. Thephasedetectormeasures signalandthe VCO output.A positivevoltageis producedwhenthe receivesignal leadsthe VCO, anda negativevoltageis producedotherwise.After beingfilteredto minimizetheeffectsof noise,thephasedetectoroutputdrivestheVCO in sucha way asto reducethephasedifference.Ideally,theinput controlvoltageof thedemodulator VCO will be identicalto the input of the modulatorVCO' The loop filter, however, to minimizethe effectsof noise. response slowsthe demodulator necessarily thanPSK,parFrequencyshiftkeyinggenerallyprovidespoorererrorperformance moticularly for multilevelsignalingin a confinedbandwidth.Older asynchronous FSK modulation' networkused overtheanalogtelephone demsfor datatransmission comparedto betthe simplerimplementation Newersystemsdo not useFSK because is no longersignificant. ter performingtechniques

Mlnimum Shift KeYing attention Oneparticularform offrequencyshift keyingthathasreceivedconsiderable FSK is binary MSK (MSK) Basically, [51. in raaio systemsis minimumshift keying phase in 180o difference exactly selectedso that with the two signalingfrequencies in one signalinterval'In this mannerMSK shift existsbetweenthe two frequencies producesa maximumphasedifferentialattheendof anintervalusinga minimumdifan MSK signalmaintainscontinuous Furthermore, i"r"n"e in signalingfrequencies. phaseat signalingtransitions.For thisreasonMSK belongsto a classof FSK signals frequencyshiftkeying(CPFSK).Figure6'6 depictsa iefenedto ascontinuous-phase MSK waveform.Noticethatthereis exactlyone-halfcycledifference representative betweena I signalanda 0 signal(? cyclesversus1'5 cycles). for MSK signalingcanbe derivedfrom Equation6.5, expression A mathematical = n/7. The result for wherenn(t) = mz(t)is a symmetricbinary NRZ signal and ACO anyparticularsignalintervalis

fo- 1,6tT

h*2lt

rrtr t/r- shnrllno

Figure 6.6 Minimum shift keYing.

286

DIGITAL MoDULATtoN ANDRADtosysTEMS

f ( T *E*u q o ) ) (logic1) lcosl'cr x(r)={ \ l ( T E t . \

f'o'[*'- a+qo)

(6.6)

(logic 0)

where1/7is thesignalingrateandS6is thephaseatthebegiruringof thesignalinterval (+n in Figure6.6). Themainattractionof MSK is its comparatively compactspectrum.Furthermore, with anappropriatemeansof detection,MSK canprovideoptimumerrorperformance in termsof the energy-per-bit-to-noise-density ratio (E/No). The expressionfor the powerspectraldensityof theMSK signaldefinedin Equation6.6 is

s(rrr):t6rt7[4"==f -Ot')

(6.7)

[n,

wheres = l(D- to.lr. The frequencyspectrumof an MSK signalis plottedin Figure 6'7, whereit is comparedto the frequencyspecrrumof pRlt(2-psx; signatingwittr the samedatarate.Noticethat the MSK spectrumis morecompactandhasits first spectralnull at 3/4Iinsteadof l/Ifor pRK.* Minimum shift keyingis actuallyjust oneexampleof a generalclassof continuous-phase modulation(cPM) techniques thatmaintaina constantamplitude,narrow power spectrum,andgoodenor performance.For a goodoverviewandbibliography of this classof modulationschemesseereferencet6l. cpM schemes havenot been usedin point-to-pointmicrowaveapplicationsbecausethey do not providehigh informationdensities.Many versionsof CPM havebeenusedin sat"llit" applications wherenonlineartransponder amplifiersprecludethe useof modulationiechniques with multipleamplitudelevels[7]. Gausslan MSK Like MSK, Gaussian minimumshift keying(GMSK) producesa constant-amplitude and continuous-phase RF signal.GMSK differs from MSK throughthe use of a Gaussian baseband pulseshapein praceof a squarepulseshapefor MSK. Because the pulserisesanddecaysasymptoticallywith respectto a zeroresponse Gaussian level, it hasa muchmoreconstrained bandwidth.Althoughit is conceivable that a GMSK signalcouldbe generated by filteringthemodulatedsignal,a rypicalimplementation [8] utilizesbaseband filtering,asshownin Figure6.8,whereit is contrasted to unfilteredMSK' TheMSK signalis generated by directFSK modulationof a carrierwith a baseband signalthatis scaledin amplitudeto producea modulationindexor 0.5.A modulationindexof this valueproducesthe differenceof l80o of phaseshift for the two datavalues.one complicationof the bandwidthJimitingfilter of GMSK is the *MSK

i. achrally more closely related to 4-PSK and therefore rs compaxed to it in a later section of this chapter.

MODULATION 287 6.1 DIGITAL

Sigmlingr.to = D|ta rute= l/T

g

fttctioh

,f,-

E

E

of outd+and

Pdvir

t

a

I I

& ol out*o{-hnd ponnr IMSKI

Fruqrrency(Hrl

Figure 6.7 Power spectraof unfiltered MSK and 2-PSK signals'

MSKWaveform Unfiltered NRZ Baseband Waveform ModulationIndex * 0.5 (a)

Filter PulseResPonse Gaussian

A Index= 0.5 Modulation (b) Figure 6.8 Comparisonof (a) directmodulationMSK and(b) GMSK'

288

DtctrAL MoDULATtoN AND RADto sysrEMS

creationof intersymbolinterferencesimilar to partial-response sy$temsdiscussed laterin this chapter.In essence, GMSK tradesa smallamountof adjacentchannelinterference for a smallamountof intersymbolinterference. GMSK modulationis utilizedin GSM digital cellular[9] discussed in Chapterg andin cellulardigitalpacket data(CDPD)applications. 6.1.3 Phase $hift Keying The secondcategoryof angle-modulated, constant-envelope signalsis referredto as phaseshift keying(PSK).Actually,oneform ofpsK hasalreadybeendiscussed as phasereversalkeying(PRK),morecorlmonly referredto as2-psK, indicatingthat eachsignalintervalusesoneoftwo phasesthatare lg0" apartto encodebinarydata. Multiple-phase shiftkeyingis alsopossible.Four-psK(alsocalledepsK) andg-psK pSK. arethemostcommonexamplesof multiple-phase Phaseshift keying (which includes4-eAM describednext) is the mosrpopular modulationtechniquefor intermediate informationdensityhigh-performance applications.The popularityis primarily dueto its constantenvelope,lnsensitivity to level variations,andgooderrorperformance. Both z-pSK and4-psK providetheoretical optimumerrorperformance in termsof a signal-to-noise ratio (E/ d. A generalexpression for n-aryphaseshiftkeyingis providedin Equation6.g.This expression assumeslfi)7a modulationis employed.That is, thephaseshift from one intervalto the nextcanrangeanywherefrom -190" to +1g0..It is po$sibleto devise PSKsystemswith lowermodulationindicesthatallow only transitionsto neighboring phases:

"tt=cosfco'.O9O)

(6.8)

whereAf =Znln is theseparation betweena-djacent signalphasesandmn$)is a symmetricn-levelNRZ baseband signalwith levels+1, t3, . . . . Examplesof typical 2-PSKand4-psK waveformsare shownin Figure6.g. Ttre signalingratefor the4-PSKsystemis shownto be exactlyone-halfthez-psr signal_ ing ratesothatequaldataratesareprovided.Thesamefigurealsoshowscorresponding phasordiagramsof the signalingphasesof a cosinewaveasdefinedin Equation 6'8. other phaseorientations arepossible.Theparticurarphases$hown,however,are convenientfor laterdiscussions relating4-psK systemsto othertypesof digitalmodulation.

Quadrature Signal Bepresentailone Despitethe somewhatexoticsoundingname,quadrature signalrepresentations area very convenientandpowerfulmeansof describingpsK signalsandmanyotherdigitally modulatedsignals.Quadrature signalrepresentations involveexpressing an arbitraryphasesinusoidalwaveformasa linearcombinationof a cosinewaveanda sine


Timrwrwform

Ph$e disgrim h)

Timewwgform

Fharndirgrem ft)

Figure 6.9 Phaseshift keying; (a) 2-PSK (b) 4-PSK.

is providedby wavewith zerostartingphases.The derivationof this representation the trigonometricidentity:

(6.e)

cos(to"t+ 0) = cosQcosro./- sin $ sin ro"t

Notice that cosQandsin Qareconstantsover a signalinginterval andhencereprecos(ro"t+ $) asa linearcombinationof thesignalscos sentcoefficientsfor expressing to each O"t and sin co.f.Sincecos rrr"tand sin or./ are90" out of phasewith respect "in quadrabe said to other, they are ofihogonalin a phasordiagramand henceare ture.tt In essence,co$ Oct and sin tDcfrepresentbasis vectors in a two-dimensional phasor diagram. The cosine signal is usually referred to as the in-phase or l signal, and the sine signal is referredto asthe out-of-phaseor B signal' Table 6.2 provides an example

TABLE6.2 QuadratureSlgnalCoefllclentelor 4'PSK Modulatlon Coefficients Quadrature DataValues

cos ocf

sin ool

CompositeSignal

01 00 10 11

0.7Q7 -0.707 4.707 0.707

4,707 -0.707 0.707 0.707

cos(otf+ n/4) cos(ru"f+ 3r/4) cos(
290

DGIALMoDULATtoN ANDHADto sysrEMs

TABLE6.3 OuadratureSignalCoefficientefor g-pSKModutatlon Quadrature Coefficients DataValues

cos ocf

sin o.f

CompositeSignal

011 010 000 001 101 100 110 111

0.924 0.383 -0.383 -{.924 -0.924 -{.383 0.383 0.924

-{,383

cos(cu"f + dB) cos(ro"f+ 3r/8) cos(o"t+ 5nl8) cos(ro"f + 7rl8) cos(ohf* 7rrle) cos(ocf- 5nl8) cos(ro.f-3nl8) cos(ro.f- dB)

4.924 -{.924 -o,383 0.383 0.924 0.924 0.383

of quadrature signalrepresentations for the4-psK signalspresented in Figure6.9.Table 6.3 providesa correspondingspecificationfor an g-psK systemusing signal phasesasprovidedin thephasordiagramofFigure6.10.Thephasordiagramassumes counterclockwise rotation,andhencethesinefunctionlagsthecosinefunctionby 90o. Most of the rest of this chapterrelies heavily on quadraturesignalrepresentations to describevariousmodulationconcepts, analyses, andimplementationr. Modulator Implementatlone A varietyof techniques arepossiblefor implementingpsK modulators. As mentioned whendiscussing PRK modulation,a 2-psK modulatorcanbeimplemented by merely invertingthecarier (multiplyingby -l) for a logic 0 andby nof invertingfor a logic l. someof the basictechniques psK signalsare usedfor generatingmultiple-phase thefollowins;

dn 0r.t

Figure 6.10 Phasordiagramof B-pSKsignal.


L synthesisof the desiredwaveformsusingdigital signalprocessingat suitably (asin voicebandmodems). low carrierfrequencies multiplephasesof a singlecarrierandselectingbetweenthephases 2. Generating dependingon the datavalues. to provide 3. Using controlleddelaysselectedthrougha switchingarrangement signals the desiredphaseshifts.Delaysareoften usedto generatethe separate in method2. signals. thePSK signalsasa linearcombinationof quadrature & Generating of multiphase a directimplementation represent Noneof theforegoingtechniques directly,a de6.8 Equation To implement PSKmodulationasdefinedin Equation6.8. proportion to the levelsin phase in direct shifts vice is neededthat producescarrier can a multiplier where signalmn(t).Exceptfor thespecialcaseof Z-PSK thebaseband be usedto produce+180ophaseshifts,suchdevicesdo not exist' Direct modulationby basebandsignalscanproducePSK signalsif a quadrature signalimplementationis used,as indicatedin method4' Actually, two multilevel onefor thein-phase(I) signalandonefor the signalsneedto be established: baseband (0) signalsarereferredto asr4(t) andtt?q(t)for baseband These out-of-phase signal. signalscorp levels chosenfor thetwo baseband The the1 and signals,respectively. combination a linear signal as a PSK repre$ent respondto the coefficientsneededto of the/ andQ signals.For example,Figure6.11 showshow an 8-PSKsignal,defined quadrature signals. by addingtwo amplitude-modulated in Table6.3,canbegenerated 6. I is providedin Figure 2. This implementation A blockdiagramof thecorresponding particularform of modulatoris chosen,not so much asa recommendationfor actual implementation,but becauseit demonstratesimportantmodulationconceptsand is of PSK signaling' usefulin analyzingthe spectrumrequirements Demod uI ator Irr p le m e ntati o n Owing to the constantenvelope,*all PSK systemsmustbe detectedwith the aid of a is coherentwith oneof thetwo theidealreference For Z-PSKsystems, localreference. (mixed) with an in-phasesignal,a possiblephases.Whenthis referenceis multiplied the oppositephase,a maximaximumpositiveoutputis obtained.Whenmultipliedby provides antipodalperformmum negativeoutputis obtained.In this manner,2-PSK The demodulationprocessfor ancewhena local coherentreferenceis established. in Figure6.4 andEquation6.4, wherethe mixing and filtering 2-PSKis presented signalruz(t). processeffectivelyrecoversthebaseband filter shownin Figure6.4 is generally a low-pass mixer and of a combination The mathematipropertyis represented phase detection phase The detector. referredto asa cally as *Discussions

to this point have not considered the effects of filtering on constant-envelopesignals. A heavily filtered PSK signal does not have a constant envelope. However, as long as filtering occurs after channel nonlinearities (e,g., power amplifiers), the most harmful effect of a nonlinearity, spectrum sprcading,is avoided.

29?

DIGITALMODULATION ANDRADIOSY$TEMS

tr(ll ' frr . cB @.r

rq {rl = ng ' rin o.r

t k l = r f ( r l + r Oi r l

ilt=fiQl ilO r ln9t

0r = FhrF ol Jrh Intflil

Figure 6.11 Generationof 8-pSK signals by superposition of quadrature-amplitudemodulatedsignals.

+ 0) (2)cosocr} #r; = 1o*Ouss{cos(ro"r = lowpass{cos S + cosQcos2ro"t- sin2qrl =COs0

(6.10)

wherelowpass{.}is a low-passfilter functiondesignedto removetwice carrier terms. Whendetecting2-PSKmodulation,a single-phase detectorindicateswhetherthe receivedphaseis closerto 0oor to 180".Thedesiredinformationis directlyavailable asthepolarityof thephasedetectoroutputcosQ.In multiphasesystems, however,the informationprovidedby a single-phase detectoris inadequate for two reasons.First,

Ampliturlc moduhtid Id{ml

Oonnrmerryrlopr PSK$nd

$n adc t

Amplltudt modrlrted 0SCul

Figure 6.12 GeneralizedPSK modulatorstructure.


of cos$ providesno informationasto whetherQis positiveor negative'Seca measure ond,theoutputof thephasedetectoris proportionalto the signalamplitudeaswell as unlessrefto cos$. Hencethemagnitudeof thephasedetectoroutputis meaningless erencedto the signalamplitude. Both of the aforementionedproblemsare overcomeif a secondmixer and filter thebestperformAs expected, thephasewith respectto a differentreference. measure first. Ify{r) is theoutto the is orthogonal reference anceis obtainedwhenthesecond phasedetector then the second 6. I 0, in put ofthe first phasedetectordefined Equation outputyQ(r)becomes + 0) (2) sin rrl.rl lB$) = lowpass{cos(o"r = _sin 0

(6.11)

phaseambiguity Thesecondphasedetectornot only resolvesthepositive/negative All decisionscanbe but alsoeliminatesthe needto establishan amplitudereference. basedon thepolarityof a phasedetectoroutputandnot on the magnitude.As a first example,considerdetectionof the4-PSKsignalsdefinedin Table6.2'Noticethatthe first databit in a pair is a 0 whenthephaseangleis positive(tt/4 or 3nl4) anda 1 otherwise.Hencethefirst databit is completelyspecifiedby thepolarityof sin Q:theoutput of the secondphasedetectoryA(f).Similarly,the seconddatabit is a I whenthe phaseis it/4, indicatingthat the polarityofyr(fl providesall informationnecessary of a 4-PSK (QPSK)demodulato detectthe secondbit. The basicimplementation tor/detectoris shownin Figure6.13.AIso shownis a 4-PSKmodulatorto emphasize the relationshipsbetweenthe modulatorand demodulator.A 4-PSK systemis preit is a usefulfoundation it is a popularsystemandbecause $entedspecificallybecause for describingothermodulationtechniques' An importantconceptto noticein *re 4-PSKsystemshownin Figure6.13is that two separatedatastreams.The modulatordividesthe incoming thereare,in e$$ence, bit streamso that bits are sentaltematelyto the in-phasemodulatorl andthe out-ofphasedephasemodulatorO. Thesesamebits appearat the outputof the respective tectorsin the demodulator,wherethey areinterleavedbackinto a serialbit $tream.In using binary PSK channelsareestablished this mannertwo essentiallyindependent andsin ocf. Thechannelsareusuallyreferredto asthe theorthogonalcarrierscosr.rlct two channelswithin an This techniqueof establishing I andSchannels,respectively. existingbandwidthis sometimesreferredto asquadraturemultiplexing' As long asthe carriersin the modulatorandthe referencesin the demodulatorare is maintainedfor bothchannelsin thereceiver),the1and truly orthogonal(coherence eachother.Any amountof misalignmentin theserewith not interfere channels do B lationshipscausescros$talkbetweenthe two quadraturechannels.Crosstalkalso channel. phasedistortionin thetransmission arisesif thereis unequalized the capacity multiplexingincreases At first thoughtit might seemthatquadrature that thebinary however, be remembered, 2. It must a factor of given by bandwidth of a thebandHence signal. is a double-sideband quadrature channel each signal on PSK in comutilized is only 507o quadrature multiplexing, without of the channel, width

?,94

DIGITAL MODULATIoN ANDRADIoSYSTEMS

parison to a single-sidebandsy$tem. when quadraturechannels are used, a singlesideband operation is no longer possible since the sideband separationprocess deshoys the orthogonality of the two signals.In essence,quadraturemultiplexing only recoversthe loss of capacity incurred by the double-sidebandspecfrum.In fact, $ome single-sidebandmodulators [10] possessa remarkable resemblanceto the epsK modulator shown in Figure 6.13. Demodulation and detection of higher level psK systemsare complicated by the fact that the use of only two referencesdoes not provide a simple meansof detecting all databits. There are two basic ways in which simple decisions(positive versusnega* tive) can be establishedto detect all data. One method is to establishmore references in the receiver and measurethe phaseof the received signal with respectto the aclditional references.The secondmethod is to use only two referencesand related phase detectorsand generateall additional measurementsaslinear combinationsof the first two.

Frtqudrcy HrcG

D|tr Input t 0 1 0 1 0

4-PSK$lgnrl

lflf Q Chrnnd

Dt$ iltput

4-PISK Signal

r 0 0

1 t

1 0

Q Glunml (h)

Figure 6.13 Four-PSK modulator-demodulator structure: (a) modulator; (D) demodulatordetector.


for 8-PSKdetection. references Figure6.14 Receiver

As anexampleof thefirst method,considerthe8-PSKsystemdefinedin Table6.3 andshownin Figure6.10.Optimumdetectionfor this systemcanbe obtainedif two additionalphasedetectoroutputsareprovidedwith respectto referencesat +ru/4and -tc/4.Thetwo newreferences A andB asshownin Figure6.14' aredesignated The ouputs of four phasedetectorscorrespondingto the four referencesaredeterminedasfollows;

+ 0Xz) sin to"f} = -sin S )0 = lowpass{cos(ro"r + OXz)sin(rrl"r+ i r) } yu = lowpass{ cos(co"t =O.707cos0 - 0.707sin$

(6.12)

+ 0X2) cosoct} = cosS yr = lowpass{cos(rrr.r + 0X2)cos(co"t * i n)} yA= lowpass{cos(ro"t = 0;707cos$ + 0.707sinQ As an aid in determining the appropriatedecision logic, theseequationshave been evaluatedfor each of the eight possible signal phasesand listed in Table 6.4. Examination of Table 6.4 indicatesthat the fust data bit is I whenever yg is positive. Similarly, the seconddatabit is 1 whenevery1is positive. The third databit is a I whenever bit is )a,lr, and y6 are all positive or when they are all negative.Hence the third data In summary, phase of outputs' detector a logical combination determinedas

D r =Q

Dz=I

Dt=AIB+AIB

(6.13)

positive whereDi is the lth databit and0, /, A, andB arelogic variablesrepresenting outputsfrom yg, /r, )4, andyr' respectively.

296

DtcrrALMoDULATtoN ANDHADtosysrEMs

TABLE6.4 Elght-P$KPhaeeDetactorOutpurs Data

Phase

011 010 000 001 101 100 110 111

nl8 3rll8 5nl8 7r18 -hil8 -5nl8 -3?r/8 4tlB

Yn

-0.383 4.924 -0.924 4,383 0.383 0.924 0.9?4 0.383

YB

0.383 4.383 {.924 -0.924 -0.383 0,383 0.924 0.924

Yt

0.924 0.383 -0.383 -0.924 -0.924 {.383 0,383 0.924

YA

0.924 0.924 0.383 4.383 -0.924 4.924 4.383 0,383

Anothermethodof detecting8-PSKsignals,whichavoidstwo extrareferences and phasedetectors, is revealedin thephasedetectorEquation6.12.The exhamea$urementsyd andy6 canbe determinedas h= 0.70'lyr O.707yn

+ 0.707y, t6= 0.7O7y,

(6.14)

Hencethe Ja and)r measurements can be obtainedas linear combinationsof the quadraturechannelphasemeasurements y7andJg, andno additionalphasedetectors arerequired.Theresulting8-PSKdemodulator/detector is shownin Figure6.15.Notice that implementation of the linearcombinationsin Equation6.14canignorethe magnitudeof the0.707multiplierssinceonly the signof theresultis needed. The linear combinationsin Equation6.14 essentiallyrepresenta rotationof the quadrature channelbasisvectorsthroughanangleofr/4 radians.By changingtheangle of rotation,thelinearcombinations neededfor otherphasemeasuremen6 areeas-

Figure 6.15 Eight-PSKdemodulator-detector usingonly two references,

MODULATION297 6.1 DIGITAL ily determined. Hence all phase-detector-baseddemodulators can be implemented with two phase detectors and as many linear combinations as needed.The general equation for the linear combinationsproduced by a rotation of a radiansis v"=v^coscL-y,slng

y. = v^ $rn fl, + V, cOSg,

(6.1s)

Beference and Clock Recovery all requirea local,phase-coin theprecedingparagraphs di$cussed Thedemodulators herent,carrierreferencefor their operation.Furthermore,multiple-pha$esystemsre* quire at leastone more referencein quadraturewith the first one. Recoveryof any -rapreferenceis complicatedby thefact thatphaseshift keyingis a double-sideband pressedcarier modulationsystem.In otherwords,thereis no discretespectralline at the carrierfrequencyasthereis with someothertypesof modulation.In fact, theFCC hasruledspecificallythatno spectrallinesareallowedin thefransmittedsignal.The absence of a spectralline at thecarrierfrequencyis overcomeby usingoneof several a nonlinearprocessingtechniques[1]. After one coherentreferenceis established, quadraturereferenceis obtainedby delayingor differentiatingthe flust. Clockrecoveryis obtainedin severalways.Oncecarrierrecoveryis accomplished occurs,theclockcanbe obtainedby lockingontotransitionsin the anddemodulation Baseband clockrecoveryoccursasthoughthesignalhasneverbeen signal. baseband In contrast,the clock can sometimesbe recovereddimodulatedand demodulated. rectly from the modulatedsignal.If a PSK signalis heavilyfiltered,the envelopeis not constantand,in fact, containsamplitudemodulationat the signalingfrequency. by envelope of signaldemodulation, independently Hencetheclockcanberecovered, filtered PSK signal detectinga heavily [2]. Clock transitionsareoften assuredby purposelyshifting thecarrierreferencehalfwaybetweensignalpointsin everyinterval.Using4-PSK,for example,thereferences wouldbe shiftedby 45o.Duringanyoneintervaltherearestill only four possiblesignal statesseparated by 90o.However,thesignalstatesalternatebetweenbeingaligned a phase with thel andQ axesandbeingalignedat45', asshownin Figure6'9' Because (filtered) andthe carier envelope the interval, both shiftis ensuredwith everysymbol reference shifts, Without the transitions. amplitude recoveredbaseband signalhave in as of described minimum data transitions properclockrecoveryrequiresa density Chapter4. D itfere ntial Detectl o n reference,somesystemsmerelycompafe AS an alternativetOrecoveringa Coherent the phasein the presentintervalto the phasein the previousinterval.The signalreceivedin thepreviousintervalis delayedfor onesignalintervalandis usedasa referenceto demodulate thesignalin thenextinterval.Assumingthatthedatahavebeen encodedin termsof phaseshift, insteadof absolutephasepositions,the datacanbe "differential dete{tion," inherdecodedproperly.Hencethis technique,referredto as ently requiresdifferential encoding.

298

DIGITALMODULATIONAND RADIOSYSTEMS

In general,PSKsystemsrequiredifferentialencodingsincethereceiversnormally haveno meansof determiningwhethera recoveredreferenceis a sinereferenceor a cosinereference.Furthermore, thepolarityof therecovered referenceis ambiguous. Thuserrorprobabilitiesfor PSKsystemsaredoubledautomatically because of thedifferentialencodingprocess.Differentialdetection,on the otherhand,impliesan even greaterlossof performance sincea noisyreferenceis usedin thedemodulation process.Typically,differentialdetectionimposesa penaltyof I to 2 dB in signal-to-noise ratio[10]. PSK Specta By far theeasiestwayto determinethespectrumof a PSKsignalis to analyzethebasebandwaveformsappliedto thequadrature channels. Owingto theorthogonalityof the two channels,the signalsareuncorrelated, andthecompositespectrumis merelythe sumofthe individual(identical)spectra. In either?-PSKor 4-PSK systemsthe baseband signalis a symmetrictwo-level NRZ waveform.Thecorresponding spectrumis thecommonsin(x)/xspectrumshown in Figure4.2. High-levelsysrems(8-PSKor greater)usesymmetricmultilevelNRZ baseband signalssimilarto thatshownin Figure4.16.ThemultiphasepSK baseband signalis somewhatdifferentsinceunevenlyspacedlevelsasdef,rned in Table6.3 are used. As mentionedin chapter4, a multilevelNRZ signalhasthe samespectrumasa two-levelsignal.Henceall conventionalPSK systemsproducea spectrumthat follows the sin(x)/xresponsedefinedin Equation4.1 but translatedto the carrierfrequency.tFigure 6.16 showsthe PSK spectrumfor two-, four-, and eight-phase systemsdesignedto providethesamedatarate.Hencethehigherlevelsystemssignal at lower ratesandhaveproportionatelynarTowerspectra. PSK Error Pertormance The errorperformatrce of any digital modulationsystemis fundamentally relatedto thedistancebetweenpointsin a signalspacediagram.For example,a z-psK sy$tem, asrepresented in the phasediagramof Figure6.9, is capableof optimumerrorperformancesincethe two signalpointshavemaximumseparationfor a given power level (radiusof thecircle).In otherwords,one2*psK signalis the exactnegativeof the other'Hence2-PSKmodulationprovidesantipodalerrorperformance asdefined in Chapter4. Theerrorperformance of a multiphasePSK systemis easilycomparedto a 2-psK systemby determiningtherelativedecrease in theerrordistance(voltageoutputof a properlyreferenced phasedetector).In additionto the error distance,however,the relativevaluesof thenoisebandwidthsmustalsobeconsidered. (Recallthatthenoise bandwidtheffectivelydetermines thevarianceof the noisesamples.) *Absolute

phase can be determined if a particular pattern in the data sream such as a framing pattern is related to some particular phase of the caxrier. ,unambiguously 'The baseband levels must be unconelated to each other to produce a sin(.r)/.rspectrum,If phasetransitions from one interval to the next are restricted in some manner, the baseband levels are correlated. and a different spectrum results.

MODULATION ?99 6.1 DIGITAL

6 3 F

E s E

E T

€

\ "o*\

I l - I 3T 2t

?

t T

3r

.l 3T

3 2T

_E3T

Figure 6.16 Spectrumof unfilteredPSK signalscarryingequaldatarates. The general expression for the distance between adjacent points in a multiphase PSK systemis

d= 2 ,i"

(6.r6)

[_N) whereN is the numberof phases. A generalexpressionof the bit error probability (or bit error rate) of an N-phase PSKsystemis derivedin AppendixC as

Pu=*fu erf(z)

(6.17)

where

,=,*F)u"*rt"'["+J Equation6.17revealsthat,with respectto EblNg,4-PSKprovidesthe sameerrorperformanceasdoes2-PSK.Thus,asmentionedearlier,both systemsprovideoptimum performance, but 4-PSKutilizeshalf asmuchbandwidth.The 4-PSKsystemhasan error distancethat is 3 dB smallerthan the error distanceof Z-PSK.However,the in thenoisebandwidth(indicating shortererrordistanceis offsetby a 3-dB decrease a 3-dB reductionin noisepowerat the detector).For the 4-PSKsystemto havethe samenoisepowerat the detectorit would haveto experiencea 3-dB greaternoise ratios(SNRs)can spectraldensity.Henceconventionalsignal-power-to-noise-power be misleadingparametersfor comparingdigital modulationsystem$.Howevet,as mentionedin Chapter4, error rateperformancesin termsof SNRsare desiredwhen

300

DtctrALMoDuLAloNANDRADto sysrEMS

E o

t =

E

t

8

e

1

0 r 1 1 2 t s 1 4 1 8 t 8 Enrqy-prrdt-E-noh-dindty mio A/Vo {dBl

Figure 6.17 Errorratesof PSKmodulationsystems.

1

7

MoDULATToN 301 6.1 DtcrrAL determiningthe effectsof interferenceor when specifyingerror rate$with respectto of 2-, 4-,8-, 16-, quantities.Figure6.17displaysthe errorperformance measurable neededfor and32-PSKasa functionof E6lNs.AppendixC providestherelationships rates in terms of SNR. enor 6.1.4 QuadratureAmplitude Modulation phaseshift keyingwith As describedpreviously,a convenientmeansof representing four or morephasesinvolvestheuseof quadraturesignals.In thecaseof 4-PSK,the quadraturesignalsrepresenttwo separate of channelsby virtue of the independence the basebandsignalsfor eachquadraturechannel.In higherlevel PSK systems,the level of thebaseband level of a baseband signalfor the1 channelis not independent signalshavebeen for theQ channel(seeTable6.3or Figure6.I I ). After thebaseband processes channelmodulationanddemodulation however,thequadrature established, for all PSK systems. areindependent amplitudemodulation(QAM) canbeviewedasanextensionof multiQuadrature independently phasePSKmodulationwhereinthetwo baseband signalsaregenerated (quadrature) channelsareestablished ofeachother.Thustwo completelyindependent In thespecialcaseoftwo levincludingthebaseband codinganddetectionprocesses. (fl) 4-PSK is andis oftenreferredto as els on eachchannel,the system identicalto such.Higher level QAM systems,however,aredistincflydifferentfrom the higher of a I6-QAM systemobFigure6.18showsa signalconstellation levelPSK $ystems. channel.Thedotsrepresent compositesigtainedfrom four levelson eachquadrature nal points while the hatchmarks on the axesrepresentamplitudelevels in each quadraturechannel.Figure 6.19 showsa basic QAM modulatorand demodulator waveformfor l6-QAM. structurealongwith,arepresentative Noticethat,in contrastto PSK signals,theQAM signalshownin Figure6.I 8 does with PSKmodulation nothavea constantenvelope. A constantenvelopeis maintained quadrature channels. on the A QAM system by restrictingthe combinationof levels doesnot restrictthe combinations sincethe levelson eachchannelareselectedindependenfly.ThusQAM modulationcannotbe usedwith saturatingamplifiers.

Figure 6.18 Signalconstellation of I6-QAM modulation.

30?

DIGITAL MoDULATIoN ANDHADIoSYSTEMS

Modulrtor

Oetnodulrtor

Ftgure6.19 QAMmodulator-demodulator. Thespectrumof a QAM systemis determined by thespectrumof thebaseband signalsappliedto thequadrature channels. Sincethesesignalshavethesamebasicstructure as the basebandPSK signals,QAM spectrumshapesare identicalto psK spectrumshapeswith equalnumbersof signalpoints.Specifically,l6-eAM hasa spectrumshapethatis identicalto I6-PSK,and64-QAM hasa $pectrumshapeidentical to tr-PSK. Eventhoughthe spechumshapesareidentical,the error performancesof the two system$ arequitedifferent.with largenumbersof signalpoints,eAM systems always outperformPSK systems.Thebasicreasonis thatthe distancebetweensignalpoints in a PSK $y$temis smallerthanthe distancebetweenpointsin a comparableQAM system.Figure6.20comparesthe signalpoinrsof a I6-QAM sysremwith the signal setof a 16-PSKsystemusingthesamepeakpower.

r6-oi.u

r6-P$K

Figure 6.20 Comparisonof I6-QAM and I6-PSK signalsets.

6,1 DIGITAL MODULATION303 The general expressionfor the distance between adjacent signal points in a unit peak-amplitudeQAM systemwith /, levels on each axis is

d= {2 L-l

(6.r 8)

overan Equations6.16and6.18revealthatan n-aryQAM systemhasan advantage n-aryPSK systemwith the samepeakpowerlevel.In termsof averagepowerlevels, the QAM systemhasan evengreateradvantage. The following equation,derivedin ratio of a QAM AppendixC, providesa generalexpressionfor the peak-to-average svstem: Peakpower -

L(L - I)2

(6.re)

Averagepower zZ!!l 1zi- t1? of a I6-PSK systemrelativeto a Example6.1. Determinethe enor performance I6-QAM systemwith the samepeak power level. Also determinethe relative performance with respectto identicalaveragepowers. Solution. Sincethe two systemsprovideidenticalnumbersof signalpoints,they signalat the samerateandrequirethe samebandwidthfor a given datarate.Thusthe relativeerrorperformanceis completelydeterminedby therelativedistancesbetween signal points. (When different signalratesare used,the effect of different noise EvaluatingEquations6.16and bandwidthsin thereceiver$mu$talsobe considered.) over 16-PSKfor a givenpeak 6.18indicate$that 16-QAMhasa 1.64-dBadvantage powerratio of power.Equation6.19indicatesthat I6-QAM hasa peak-to-average ratios,the advantageof a 2.55 dB. SincePSK $ystemshaveunity peak-to-average powers. 16-QAMsystemovera 16-PSKis 4.19dB for equalaverage Theresultsof Example6.I showthat I 6-QAM is a significantlybettermodulation formatthan 16-PSKfor applicationslike voicebandmodelsthathaveno technology of I6-QAM is advantage limits.- In digital microwaveapplicationsthe performance it is at 16 signal diminishedby availableamplifier technology,but, nevertheless, points that PSK modulationdefersto QAM modulationfor performancereasons. digital radiosused8-PSKmodulation[13] and someused Somefirst-generation 'Voiceband

modems are not resticted by technology but by signal power limits to prevent intetference with other signals in the network. It is ironic that these rcstrictions arise primarily from old FDM analog radio systems in which a high-powered signal in one channel could create inter{erence (crosstalk) into other channels. In a predominantly digital network the main signal limitation would be the saturation point of PCM encoders(and possibly crosstalk in subscriberpairs). It is also ironic that increasing the signal power ofa voiceband modem (V.34 and earlier) to falljust short ofthe PCM saturation point would not improve performance in a mostly digital network. Performance of high-speed modems in this case is primarily determined by multiplicative noise crcated by PCM companders[18].

304

DIGITAL MoDULATIoN ANDHADIoSYSTEMS :21.96d8

Figure 6.21 256QAM andsteppedsquare256QAM.

I6-QAM [a] with comparable performance. Second-generation radiosuse64-eAM followed by third-generation radios with 256-QAM [16, t7]. [5] Reliableoperationof high-densitymodulationformarslike 256-eAM requiresextremelylinearamplifiersto toleratethewide rangeof signalamplitudesandextremely accurateadaptiveequalizersto removesmallpercentages of intersymbolinterference that arisewhen a high-amplitudepulse is kansmittedadjacentto a low-amplitude pulse.Amplifier andequalizerperformance requirements arebasicallya functionof thedynamicrangeof thesignalamplitudes*;

Dynamic roilg€ =

lorog,ot"Tl

(6.20)

The dynamicrangeof a squareQAM modulationformat canbe improvedby modifying thesignalsetto eliminatecornerpointsthatproducehighpeakpowers.one such techniquereferredto as 256-stepped squareQAM (256-sseAM) tlgl is shownin Figure6.21,whereit canbe comparedwith the signalsetof conventional256-eAM. Example6.2. Determinethe dynamicrangeof both the conventional256-eAM signalsetandthe 256-SSQAMsignalsetshownin Figure6.21. Solution. Usingintegralunitsalongeachquadrature axisof l, 3,5,7,9,1 l, 13,and 15for 256-QAM,thedynamicrange(DR) is / t'q 2 - r - t^" s2\ I DR(256-QAM):lOlosro |l 1 2 + 1 2| \ , / = 23.52dB 'Ifthe

quadrature carriers are perfectly recovered, intersymbol interference would only be a functio n ofthe dynamic range along each axis. To make an allowance for quadrature carrier phase enor, which causes interfercnce between I and p channels, the composite signal power is used.

. 6,1 DIGITAL MODULATION 305

Thehighestamplitudesignalpoint of 256-SSQAMis the (5,17)point.Thusthe dynamicrangeof the256-SSQAMsignalis

= lolog,o DR(256-ssQAM) l'Hl \-

-

J

:21.96 dB The resultof Example6.2 showsthatthe peakpowerof 256-SSQAMis 1.56dB 256-QAMfor the samesignalpoint sepalowerthanthepeakpowerof conventional ration.Thusin a peak-powerlimited application,256-SSQAMcanoperatewith 1.56 256-QAM.Whenother,lesssignifithancanconventional dB moresignalseparation cantfactorsareconsidered, reference[20]reportsthata 140-Mbps256-SSQAMradio advantage. at6 GHzhasa 2-dB performance The theoreticalenor rateequationfor QAM modulation,derivedin AppendixC, enorrateequationwith thesamenumberof levis identicalto themultilevelbaseband produce equation effor is usedto ratecurvesfor 4-, 16-,64-,and256-QAM els.This modulationin Figure6.??.Table6.5 comparesthe most coillmonforms of digital modulationusedin point-to-pointmicrowaveradiosystems. Offeet Keylng signalsin highJevelPSK modulaof the baseband Becauseof the interdependence coincide.However, channelsnecessarily tion, the signaltransitionson thequadrature modulatethe quadraturechannels,they are not sinceQAM systemsindependently Whenthesignalintervals to alignthesignalintervalson thetwo channels. consffained overlapeachotherby 507o(Figure6.23),themodeof operationis referredto asoffset keying.Offsetkeying is commonlyusedon 4-QAM systems(morecommonlyreof offsetkeyingliesin the Themainadvantage ferredto as4-PSKor QPSKsystems). to theincomingcarrecoverycircuitryto becomesynchronized abilityof thereference rier at lower signal-to-noise ratiosthanconventional(aligned)QAM andQPSKsystems[21,22], QAM Representation of Mlnimum Shlft Keying haveassumed signals theuseof baseband Theprecedingdiscussions of QAM sy$tems with NRZ levelencoding.A moregeneralview of QAM allowsarbitrarypulseshapes in generating signals.Oneparticularlyinterestingpulseshapeis a halfthebaseband pulse; sinusoid

zl A(t):cosffF I r < / <" + \ -

where? is the durationof a sienalinterval.

)

(6.21)

306

DGtrALMoDUt-ATtoN ANDRADtosysrEMs

10-l

\ \

\

\ \

t0-?

\ \

\

\ \ \

\

l0-3

\ \

\

\

\

zEE-Ievel

\ \ \

-

e

\ \

1o-4

\

E o

\

= 5

to d

10-6

\

\

\ \

\

\

\ \ \

to4

\ \

\

\ 0l

\

\

\

\

l(r7

\l \

+ l0-€ 9

10

lt

32-lcvil

t6-i

\

\ 12 13 t4 t5 t8 r7 18 rg Avrage rnergy-per-bit-to - mis - dsrrlty rutio Eb/,Vo (dBl

Figure 6.22

Error rates of QAM modulation systems.

\ 2{t

2l

6.1 DIGITAL MODULATION307 TABLE6.5 Comparisonol VarlousDigitalModulatlonTechnlqussBasedon Equal Data Rstes Signal-to-Noise^Ratios forBER= 10* (dB) System Designation 2-PSK 4.PSK,4-QAM QPR 8-PSK 16.QAM 16-QPR 16.PSK 32.QAM 64.QAM

Information Density(bp#Hz) 1 2 Zb q

4 4b 4 5 6

EblNo on the Channel 10,6 10.6 1?,6 14.0 14.5 16.5 18,3 17.4 18.8

SNHat Decision Peak-to-Average Hatio(dB)e Circuit 13.6 13,6 17.6 't8.8 20.5 24.5 24.3 24.4 26,6

0.0 0.0 2.0 0.0 2.55 4.55 0.0 2.3 3.68

aRetioof maximumsteady-state Bignalpowerto averag€signalpowerwith randomdata.MBasured on a liltorpartitloning. channelwithsquare-rool Dlna strictB€ns€,th6 signalbandwidthof a panial-response syst€mis no narrowerthan the th6orBtical (Nyquist)bandwidth fulFresponse of a conesponding sybtem.As a practicalmann€r,however,the partialre8pons€ systemsrequireabout17ol,lsssbandwidth [21].

pulseshapingof duration7 is usedon bothchannelsof an offset If half-sinusoidal as keyedQAM system,thequadrature channelsignalsp(r) andq(r) canbe expressed follows;

= ai*,[r$J""p(t) +r f +t=,=+t) Q(t)= aq'-[ff)-i" *"t ( 0 < r < I )

(6.22)

tr(rl=Dr(,|'cE

l ^^l l ^ l , l . l ' l".llo" l . l ' l ^ l rln(dJl

Figure 6.23 Offsetkeyed4-PSKsignaling.

308

DGITAL MoDULATtoN AND RADto sysrEMS

wherea; andc, aredatavalues(tl) for the/ and0 channels, respectively,cos(ntlLT) is shapingon the.lchannel,andsin(nrl?T)is shapingon thep channel.(Thesinefunction is usedto accountfor offsetkeying.)Addingthetwo quadrature signalstogether andapplyingsomefundamental trigonometricidentitiesproducestheresult

x(t): p(t)+ q(t) = a i c o l's [ + r +nr) 7J (a,=no) ( m \ = 4i cos rrj [*t

(a,+au)

(6.23)

Equation 6.23 is essentiallyidentical to Equation 6.6 defining minimum shift keying. Hence, except for a logicJevel transformation of data values, offset QAM with sinusoidalpulse shapingis identical to MSK 15,241.The relationship is further demonstrated in Figure 6.24, which shows how an MSK signal is generatedby offset keyed quadraturechannel modulation with sinusoidalpulse shaping.

I-Channelbrrohnd: Dr(rl

,5rltl = Df(tl ' cor (o.rl

Q-Chennelbftsbrnd: Dg lrl

.$a(tl = Dq (rl . rin (
$(tl = sr{tl +.So(rl

^ l r , ^l I t w, =2tf,

l r l r . l r " l r " l ^l lôl ^ l o lo * 2'l5lt

fi =a.isrt

Figure 6.24 MSK signalingproducedby offsetkeyedQAM with sinusoidalpulseshapes.

MoDULATToN309 6.1 DterrAL

6 3 F

T E t

f,

I

Frnumcrr {Hzl

Figure 6.25 Powerspectraof MSK and4-PSKsignals.

TheseforegoingresultsshowthatMSK is very closelyrelatedto offset-keyed4pulseshapesfor baseband PSK. The only differenceis in the useof half-sinusoidal MSK andsquarepulseshapesfor 4-PSK.Becauseof this closerelationship,it is interestingto comparetheamplitudespectraof the two systemsin Figure6.25.As indicated,theMSK spectrumhasits first spectralnull at a 507ohigherfrequencythan 4-PSK'sfirst null. Otherthanthis,theMSK spectrumis morecompactthanthe4-PSK outsidethemainlobe spectrumandis significantlylower in amplitudeat frequencies of thespectrum. ForthigreasonMSK is anattractivemodulationtechniquewhereconandlittle or no filteringaredesired.Onesuchapplicationis on digital stantenvelopes requiringtranssatellitelinks with frequencydivisionsubchannels [7]. In applications mitter filters, MSK hasno particularadvantageover 4-PSK and usuallyrequiresa morecomplicatedmodulator. 6.1.5 CarrlerlessAmplitude and Phase Modulatlon Carrierlessamplitudeand phase(CAP) modulation[25] is a specializedform of QAM. As indicatedin Figure6.26,no explicitmodulationof a carrier(or carriers)ocwith DSP filter curs.Instead,two separate modulatedsignalsaredirectlygenerated quadrature functions:onefor anin-phasechannelandonefor a channel.Thein-phase filter conveftsf channeldatadirectlyinto a filteredDSPcosinewavewhile thequadraturefilter converts0 channeldatainto a filteredDSPsinewave.In this mannerthe into thefilter algorithm.(Typisymbolrateis lockedto animpliedcarrierembedded carier frequencyis equalto thebaudratesothereis onecycleof cally,theembedded signals a "carrier"in eachsymbolinterval.)After fhein-phaseandquadrature-phase

310

DIGITAL MoDULATIoN ANDHADIoSYSTEMS

Figure6.26 CAPmodulator blockdiagram. areadded,the result is convertedto an analogsignaland f,rlteredto smooththe DSp samplingfrequency. CAP modulationis used in one versionof asymmetricdigital subscriberline (ADSL) [26] and for a 51.84-Mbpsunshieldedtwisted-pairATM srandard[27]. ADSL applicationsaredescribedin Chapter11. 6.1.6 Partlal-ResponseQAM Another popular modulationtechniqueis quadraturepartial-response signaling (QPRS).As commonlyimplemented[28], a QPRSmodularoris nothingmorethana filter that "over filters" the quadraQAM modulatorfollowed by a narrow-bandpass ture signalsand producescontrolledintersymbolinterferencein eachchannel.The mostcommonapplicationof QPRSinvolvestwo levelson eachchannelbeforefiltering andthreelevelsafterward(seeChapter4). This systemis essentiallya 4-PSK systemwith partial-response filtering to increasetheinformationdensity.As shownin Figure6.27,theeffectof partial-re$ponse fiIteringis to produceninesignalpointsfrom theoriginalfour. In a similarmanner,a 16-QAMpartial-response system,with four levelson eachchannelbeforefiltering, hassevenlevelsafterwardand49 signalpointsin all t291.

Edtor" filtfiing

Afti. tlltfiing

Figure 6.27 QPRSsignalconstellations.

\-

MODULATION 31 1 6.1 DIGITAL

Heiredcodne chonncl

NBw= 1/r

Truncetion loll = OdB (dueto prkingf

Sigr|nl sspsr|tiod

=,/Ve

Cdinc chrnnol

-i

r

Truncrtlon lon = 2dB

NBW= Z/tT (2dBbolortl/n

Signrl rGpdratlon

=t/Iatz

withequaldatarates. of QPSKandQPRSsystems Figure6.2E Comparison

Althoughthemodulators of a QPRSsystemcanbeconventional QAM modulators, must be modified to accountfor the extra levels in the the demodulator/detectors for eachchannel the detectionprocesses waveform.After the signalis demodulated, areindependentandidenticalto the basebandPRSdetectionproceduresdescribedin Chapter4. a QPRSsystemto a QPSK(4-PSK)systemin termsof equal Figure6.28compares peakpowerout of the modulator.The averagetransmitpowersaredifferent,howsystemoverfiltersthe signalto reducethe transever,becausethe partial-respon$e practice, the informationdensityis increasedby about 177o mitted bandwidth.In

t231. filteringcutsthedistancebetweensigin Chapter4, partial-response As discussed in error performance. However,the reduction nal pointsin half, indicatinga 6-dB of thecorbandwidth than filter the noise is lower noisebandwidthof a PRSreceive is redistance of the degradation that enor systemso some respondingfull-response performance loss of a 6.28 the net covered.For the filter systemsshownin Figure power channel the respect on the to QPRSsystemis 4 dB. Notice,however,thatwith performance lossis only 2 dB. 6.1.7 Trellls-Coded Modulation of PRS mentionedin the previoussectionasThe 2-dB pedormancedisadvantage sumesthatdataaredetectedby makingharddecisionsonebit at a time.As described thatoverlapstwo bit intervals.If in Chapter4, however,a PRSsignalhasredundancy the signalspanningtwo intervalsbeforemakinghard the detectioncircuitry processes penaltycanbe recovered. a redundantsignalin Processing decisionstheperformance

312

DrcrTAL MoDULATtoN AND RADto sysrEMS

this way is an exampleof a maximumlikelihoodor viterbi decodingalgorithm[30] for redundantsignals.If theredundancy is extendedto morelevelsandmoreintervals andoptimallyprocessed, evengreatercodinggainsarepossible.As typicallyimplemented,theseextensions useextrasignalpointswith reskictedsequences (transitionrr betweenpoints).Whenthe allowedsequences are represented with stateffansition diagrams, theyform a trellis.Hencethetermtrellis-coded modulation(TcM) t3l, 3zl. As mentionedin chapter4,TCM is very similartoconvolutionalcoding.Themajor differenceis that convolutionalcoding addsredundantsymbols(increasesthe bandwidth)but TCM addsredundantsignallevels(increases the signalspace).Both $ystemsaredetectedin similar fashionsusingtrellis diagrams.The decodingalgorithmsof both$ystems essentiallydeterminethemostlikely sequence of statesof the tran$mitted codes,Thetransitionsbetweenstatesthendeterminethedata. As a TCM example,considertheexpansionof a 4-PSKsignalsetto an B-psK signal setas shownin Figure6.29.Althoughthereareeightsignalpointsin the g-psK constellation, only four pointscanbe freelychosenin any oneinterval.which four pointscanbe chosenis dependent ofthe signalpointschosenin previousintervals.A TCM demodulator/decoder thendetermineswhich of the allowedsignalsequences mostcloselymatchesa received$equence beforemakingharddatadecisions. Figure6.29 showsthat if datadecisionsaremadeoneintervalat a time. without processing the redundancy, the errorperformance of ttreexpandedsignalsetis 5.33 dB worsethantheoriginalsignalset(a consequence of thenoisepowermarginof adjacentsignalpointsbeingreducedfrom 0.5 to 0.146).Also shownin Figure6.29are thenoisemarginsof morewidely separated signalpoints.TCM recoversthe5.33-dB penalty,andmore,by ensuringall allowedsequences includesomeof thelargerdistances.Furthermore, noisein only oneintervalis notlikely to producea decisionerror. when four-statecoded8-PSKusesthetransitionrellis shownin Figure6.30,theperformanceapproaches a 3-dB improvementwith respectto uncoded4-psK t3ll. To understand Figure6.30,it is importantto realizethatit represent$ statetransitionsin theencodingprocess,not signalwaveforms.Theuseof thestatesin thisman-

4 (157.5)

= 0.146 or2= sinlae.s") 5 (202.5)

d3=sint+s")=0.S dsz=sintoz.b") = 0.854 doz=sintggo)=1.0 Figure 6.29 EighrPSK signalpointsanderrordisrances.

MODULATION 313 6,1 DIGITAL

Figure 6.30 Four-statetrellis for 8-PSKmodulation.

pasthistory.(All pasthistoriesareenner is just a convenientmeansof representing capsulated intojust four states.)Transitionsfrom onestateto anotherarelabeledwith the8-PSKsignalpointchosenfor encoding2 bits of data.Noticethatonly four of the 8-PSKsignalpointscanbe freelychosenandthat thesepointshavethe sarne$eparation asuncoded4-PSK.The significantaspectsof the decodingproces$aresuflrmarizedasfollows; 1 . An allowed transition between two statescan occur with either of two signals, which implies sequenceinformation does not help discriminate between those two particular signals.However, the two signals in question are chosento have maximal separation(noise margin 1) so redundancyis unnecessary.

2. Transitions that originate in different statesand terminate in any particular state are encodedwith signalshaving noise power margins of 0.5. 3 . All sequencesthat begin and end in common statesare at least three intervals long with minimum noise power margins of 0.5, 0.146, and 0.5. Thus the total noise margin betweenany two minimum-length sequencesis 1.146.

that arelongerthanthreebut It may be necessaryto discriminatebetweenseguencerl in all casesthe endpointsinvolvetransitionswith noisepowermarginsof 0.5' Thus thenoisemarginis alwaysgreaterthan l, whichis 3 dB greaterthanthenoisemargin of uncoded4-PSK.Determinationof the exacterrorrateof a TCM systemis much detectionis used.Reference moreinvolvedthanwheninterval-by-interval [31] shows

314

DGTTAL MoDULATIoN ANDHADto sysrEMS

that the bit error rate of coded 8-PSK asymptotically approaches3*dB improvement with respectto 4-PSK as the noise margin would indicate. Example 6.3. A receiver for a trellis-coded modulation system as shown in Figure 6.30 detects a phase sequenceof 20o, 220", and 10o. Determine the allowabre $equenceof three signals with the closest distance to the given received sequence. Assume the starting stateis stateB. solution. To begin with we can determine the closest encoder signal phasesas 22.5",202.5o,22.5",whichcorrespondtosignalpoints l,5, l.ExaminationofFigure 6.30 indicates that this is a disallowed sequencebecausea 5 signal cannot follow a I signal. If data were decidedon a symbol-by-symbol basis,an error would certainly be made. By tracing the trellis from state A, a list of allowable state sequencescan be determinedas provided in Table 6.6. For eachtransition, the most likely signal of each pair of signals that can produce a particular transition is indicated. The symbol errors (in degrees)for eachof thesesignalsis then determinedfollowed by the total sequence error. As indicated, the most likely statesequenceis cAA (signal sequence:1,6,0). Although the result of Example 6.3 indicatesone particular sequenceis more likely than any other, there are two other allowable sequences(l4l and 162) that are fairly close to the most likely signal sequence.Notice that these two sequenceshave end states(C and B) that are different than end stateA ofthe selectedsequence.Thus, there is more information to come as to which is the most likely sequence.Becausestates A and B have no coillmon allowed signals, the next symbol will provide additional discrimination between 160 and 162. The very next signal will not help discriminate between 160 and 141 but the signal following that will. A thorough determinationof the most likely transmitted signal sequenceneedsto consider other possible starting phasesand, consequently,previous signal values. (This is an exerciseleft to the student.) Coded 8-PSK TCM hasbeenusedin satellitecircuits [33] where nonlinearitiesdictate the use of a constant-envelopesignal. TCM with higher density eAM signal sets

TABLE6.6 Sequenceand Error Det€rmlnatlonfor Example6.3 States

cBc DDD CAA DCB CBD DDC CAB DCA

Signals 141 351 160 370 143 353 16? 372

SymbolEnors ?,5,62.5,12.5 9?.5,17.5, 12.5 2.5,27.5,32.5 92.5,72.5,32.5 2.5,62.5,102.5 92.5,17.5,102.5 2.5,27.5,57.5 92.5,72.5,57.5

TotalEnor 77.5 122.5 62.5 197.5 167.5 212.5 87.5 222.5

6.1 DIGITAL MODULATI0N 315

voicebandQAM modemssuchasthe 14.4-kbpsV.33 [3a] and is usedin high-speed 28.8-kbpsV.34 [35]. TCM can alsobe usedin conjunctionwith MSK modulation t361.TCM hasbeenutilizedin point-to-pointmicrowaveradiossuchasthe 155-Mbps 6G150MB radioof NEC.The modulationformatis 128-QAM.NEC refersto TCM asmultilevelcodedmodulation(MLCM). 6.1.8 MulticarrlerModulation to thispointit hasbeenimplicitly assumed In all of themodulationformatsdescribed carriersopthatmodulationoccurson a singlecarrier,or possiblyon two quadrature eratingat the samefrequency.Multicarriermodulation(MCM) involvesdividingthe datastreaminto multiple, lower ratesffeamsthat aretransmittedin parallelon multiple carrierfrequencies. AlthoughMCM hasbeenutilizedin somespecialapplications of DSPtechnologythatallowseconomicimplein the past[37],It is theemergence The for commercialapplications. mentationof multiplemodulatorsanddemodulators mostcommonmeansof implementingMCM utilizesfastFouriertransforms(FFTs), into separate datastreamsfor asshownin Figure6.31.Sourcedataaredemultiplexed arethenindependently encodedinto The datafor eachsubchannel eachsubchannel. the amplitudesof quadraturechannel complexnumbersrepresenting discrete-level into a timecarriers.An inverseFFI transformsthe complex$pectrumcomponents domainwaveformthat is convertedto analogandtransmitted. in thereceiverwheretheincomingtimeThebasicstepsof theprocessarereversed domain waveformis sampledand fed to an FFT that recoversa complexspectrum' asindividualQAM (or possiblyPSK) The spectrumcomponents arethenprocessed signalsto decodethe data,which arethen multiplexedbackinto a compositedata stream.Not includedin Figure6.31areancillaryfunctionsof equalizationandclock recovery.Normally,onefrequencycomponentis dedicatedasa pilot signalthatcarAmplitudeequalizationinvolvesmerely ries clock informationfbr all subchannels. scalingtheFFT componentamplitudesat theoutputof theFFT.Amplitudeequalization may not evenbe requiredif PSK modulationis utilized because,eventhough the theremaybe amplitudedistortionin thechannel,within eachnarrowsubchannel

Encode

InvstEe FFT

Figure 6.31 Multicarrier modulationutilizing fast Fourier transforms.

316

DtctrALMoDULAIoN ANDRADto sysrEMS

distortion is essentiallyflat, which implies that PSK data can be recoveredby merely determining the phaseof each complex frequency term. Multicarrier modulation with an FFT implementation is commonly referred to as discrete multi-tone (DMT) in North America and as orthogonal frequency division multiplexing (OFDM) in Europe. (The term "orthogonal" occurs becausethe frequency components of an inverse FFT are harmonically related and therefore have zero-valuedcross-correlationproducts.)The terms DMT and OFDM are interchangeable with the exception that in some OFDM applicationsit is understoodthat all subchannels utilize the $ame form of modulation with the same number of bits per channel. In DMT systemsthe modulation of the subchannelsis more general so that different dataratescan be carried on different subchannelsdependingon the transmission quality of the respectivesubchannels.A principal application of DMT is ADSL [38] standardizedby ANSI committee TlEl.4. OFDM is specifiedby EuropeanTelecommunications standards Institute (ETSD as the modulation format for Digital Video Broadcasting(DVB) t391. One of the most attractive featuresof DMT is the inherent ability to match an information signal spectrum to a channel response.An example of such a system is shown in Figure 6,32 that is representativeof an ADSL application on subscriberwire pairs. A significant impairment of using existing wire pairs for high-bandwidth clata is the possiblepresenceof bridged taps.ISDN basic rate installationsrequire removal of bridged taps. In an ADSL implementation bridged rap$are accommodatedby detecting their presenceduring channel characterizationand then transmitting only as much information in the affected subchannel(s)as can be reliably supported.Notice that a conventional wideband data signal would experiencesignificant distoftion if ffansmitted on the channel of Figure 6.32. Thus, a significant amount of amplitude (and probably phase)equalizationwould be required.

Bridgedtap notch Narrowband interference

Channel response

Information densityper subchannel Frequency Figure 6.32 Use of DMT modulationto matchsubchanneldatacapacityto transmission channel.

PARTITIONING 317 6.2 FILTER

Figure6.32showsthata DMT systemassignsinformationdensitiesto individual ratios determinedduring channel basedon respectivesignal-to-noise subchannels that Includedin theprocessis detectionofnanowbandinterference characterization. wideband Noticethata conventional to be eliminatedaltogether. causesa subchannel For moredetailson systemmightbetotallydisabledby thenarowbandinterference.* in ChapterI l. seetheADSL discussion DMT implementations

6.2 FILTERPARTITIONING Thetransmittingandreceivingequipmentof a digitalradiosystemtypicallycontains severalfilters that limit the signalspectrumto somedegreeor another.Sincethe endpulsereto-endfrequencyresponse of thechannelmustconformto certainbaseband sponseobjectives,thedesiredcompositefilter functionmustbepartitionedamongthe individualfilters.Normally,a numberof the filters canbe designedto providetheir For respectivefunctionswithoutsignificantlyimpactingthe channelpulseresponse. input frequencies. process of the produces sum and a difference a a mixing example, Only the sumis wantedwhenmixing upward,or only the differenceis wantedwhen termscanbeeliminatedby a filter thatdoes mixing downward.Usuallytheundesired not significantlyaffectthepulseshapeof theunderlyingsignal.Thefollowingdiscuspul$eresponse; thatonly two filters significantlyinfluencethebaseband sionassumes onein the transmitterandone in the receiver. 6.2.1 Adlacent-Channel Interference One basic pil?o$e of the radio channelreceivefilter is to minimize the amountof noisepresentat the detector.A secondpurposeof this filter is to rejectenergyin adjacentradio channels.Energyfrom an adjacentchannelthat doesnot get rejectedis interference.In this discussion,we assumethat the referredto as adjacent-channel This situin theadjacentchannelsareidenticalto thedesiredspectra. spectrumshapes number of frequency-division-multi depicts a is shown 6.33, which in Figure ation plexeddigitalchannels. occursasa resultoftwo interference As indicatedin Figure6.33,adjacent-channel power passes P1 becausethe adjacent phenomena. unwanted First, the receivefilter prevent into overlap thedesiredchannel.Thesecsignalis notcompletelytruncatedto thereceivefilter doesnot provideinf,tond sourceofinterference,P2,occursbecause nite attenuationof power properly belongingin the adjacentchannel.Unwanted powerP1is minimizedby narrowingthetransmitfilter, while P2is minimizedby nar*Totally

avoiding the effects of high-energy narrowband interference in a DMT sylttem is not as simple as it might seem, If the interference is present at the input to the A,/D converter in the recei ver, the interference may causeA/D saturation(orincreased quantizationnoise ifa compandedconverter is utilized). Thus, total avoidance of narrowband interference requires front-end norching of the signal. A front-end notch could also be used in a conventional wideband system followed by decision feedback equalization to accommodate the inserted distonion [401.

318

D|GITAL MoDULATIoN ANDRADIo SYSTEMS

rowing the receive filter. Since channel pulse responseconsiderationsconstrain the compositefilter function to someminimum width, one componentof the interference cannotbe reducedwithout increasingthe other. Hence the total filter function must be partitioned in some marlner to minimize the sum of Pr and p".

6.2.2 Optlmum Partitionlng The optimumfilter designfor anyparticularapplicationmay dependon a numberof factorsincludinglegislatedemissionspecifications, availabletechnologyfor power amplifiers,theavailabilityofcrosspolarizationfor adjacent*channel isolation,andthe relativeeffectsof noiseversusadjacent-charurel interference. In theabsence of external constraints, a classicalresultattributedto Sundet4ll andalsopresented in references[42] and [43] determines optimumpartitioningasonemarchingthe outpurof the transmitfilter to the squareroot of the desiredchannelresponse.The desiredoutput spectrumY(or)is obtainedas

Y(or)= Holo);I1r*(ol)X(ro)

whereX(ro) = channelinput spectrum flrx(rn) = transmitfilter response /1nx(ot)= receivefilter response Thentheoptimumpartitioningis obtainedas

Adi*rnt rlgnrl

Rffihrfi

De$rrd tignrl

Adleent rl0nrl

filtrr.E|pofite

Figure 633 Adjacent-channelinterference.

(6.24)

6.2 FILTEHPARTITIONING 319

lflo*(o)l =lY(rt)llt?

lar*(ru)l=H

(6.25)

to minimizetheadjacent-channel thefilter amplitudere$ponses Equation6.25def,rnes interferenceunder the condition tttat the adjacentchannelscontainidentical signals is the with identicalpowerlevels.In addition,if thetransmittedspectrumffty(ro)X(ro) Hs;q(o),thereceivefilter is matched complexconjugateof thereceivefilter response to the channelspectrum,andthe bestpossibleerror performanceis obtainedwith respectto signalpoweron thechannel. Partitioningasdefinedby Equation6.25is shownin Figure6.34for a pulseinput signalspecanda raisedcosineoutput(seeAppendixC). Eventhoughonly baseband passband filter functions. is to easily tra areshown,theconcept extended AlthoughEquation6.25providesa soundtheoreticalbasisfor determiningtheopof a particularapplicationmay retimum filter partitioning,practicalconsiderations "optimum." quire deviationfrom the Oneproblemthat arisesin digital microwave radiosis relatedto thepeakat thebandedgesof thetransmitfilter in Figure6.34.With a passivefllter this peakcanbe obtainedonly by insertinglossinto the midbandreno ill effectsresult. If the transmitpowercanbe increasedascompensation, sponrte. However,theoutputlevelof manydigitalmicrowaveradiosis limitedby thetechnology of microwavefrequencypoweramplifiers(a few wattstypically).Sincetheseradios are device power limited, midbandinsertionloss cannotbe overcomeand thereforesubtractsdirectlyfrom thereceivedsignallevel. applicationsthe optimumhansmitfilter is onehavinga In device-power-limited at thedetector, in thepassband. To achievethedesiredchannelresponse flat response insertion Channel lossesat peaked band edges. at receive must then the filter be the attenunoise ate performance signal and both the fhe $ince this point do not degrade the of increasing receiver peaks havetheundesirable effect atedequally.However,the interference. Although noisebandwidthand the P1componentof adjacent-channel theseincreasesrepresenta poorerperfonnancethanthat obtainedby theoreticallyoptimum partitioning,the degradationis not a$ Sreatas the insertionloss neededto achievethe "theoreticaloptimum."As onefulther note,it shouldbe mentionedthat

ilob. Nnd Infirfafdftot

excitationand raised-cosine Flgrrre 6.34 Theoreticaloptimum filtering for square-pulse output.

320

DrcrrALMoDULATtoN ANDHADtosysrEMS

not all theoretically optimum transmit filters havea peak at the band edges.In particular, partial-responsesystemsdo not require band edge peaking (seeAppendix C). Another aspectof the optimum designto be kept in mind is that optimum error performance is achievedwith respectto signal power on the channel,and not with respect to unfiltered transmitterpower. It is this featurethat allows any amount of attenuation to be inserted into the charurelat the transmitter and not degradethe optimum performance. If performanceis measuredwith respectto unfiltered transmit power, the best transmit filter may be different from that defined in Equation 6.25. Not only would less midband attenuationbe desirable,as discussed,but it might also be desirable to widen the transmit filter response.This decreasesthe truncation loss in the kansmit filter so the receive filter can be narowed to decreasethe receiver noise bandwidth. As an extreme example of how widening the transmit filter can improve error rate performance,consider removing the transmit filter and incorporating it into the receiver. For a given output at the power amplifier, the signal at the detector is un* changed.The receiver noise bandwidth, however, is reduced becausethe composite filter is nilrower than the original receive filter. Hence a higher signal-to-noiseratio is presentat the detector. The penalty incurred for removing the transmit filter, of cour$e, is a greatly increasedP1componentofadjacent-channelinterference.Ifadjacent channelsdo not exist or if they are adequatelyisolated by cross-polarization(i.e., if the systemis noise limited), the performance can be improved by moving some of the transmitter ffuncation lossesto the receiveruntil adjacent-channelinterferencematchesthe noise. Keep in mind, however, that if the sy'rtemis adjacent-chamel interferencelimited, the optimum partitioning is indeed defined by Equation 6.25.

6.3 EMISSION SPECIFICATIONS Oneunavoidable consideration whendetermining thefilterfunctions of a transmitter and receiver is the out-of-band emission specificationsestablishedby the FCC in the united states or the ITU-R in other parts of the world. In many casesthe emission specificationsdictate a narower transmit filter than the theoreticaloptimum. Thus the partitioning aspectof the filter designsmay be predetermined. It is somewhatironic that the FCC emissionspecificationswere intendedto control adjacent-channelinterferencebut, in somecases,actually causethe interferenceto increaseby forcing the use of a wider than optimum receive filter. The FCC specifications, however, were intended to protect adjacent analog radio channels from out-of-band digital emissions.They were not selectedwith adjacentdigital channels in mind. In January 1975, when the FCC establishedthe out-of-band emission specifications, two separatespecificationswere established;forradios operatingbelow 15 GHz and for radios operating above l5 GHz. The emission limitations for operationbelow l5 GHz are more restrictive than those for operation above 15 GHz becausethe lower frequencieswere heavily used for analog FDM-FM radios, which are more sensitive to interference.The higher frequencieshave not beenusedextensively becauseof vul-

SPECIF|CATIONS321 6.3 EMISSTON

nerability to rain attenuation.Thesefrequenciesare usedfor short-distance,specialInitially, the putposeapplicationsandarenot ascongested asthelower frequencies. mostpopularbandsfor digitalmicrowavewere I I and6 GHz. Theemissionlimitationsfor operationbelow 15GHz are A = 35 + l0log,oB + 0.8(P- 50)

(P > 50)

(6.26)

whereA = powerin 4-kHzbandrelativeto meanoutputpower(dB) B = authorizedbandwidth(MHz) P = percentremovedfrom carrierfrequency(50 is thebandedge) outsidethebandbut A mustbe at least50 dB everywhere In addition,the attenuation limitationsare emission point. Notice that any the 80 dB at need to exceed doesnot transmitted relative to power but the only of levels, absolute in terms not $pecified power.Thustheseemissionlimitationsdo not constraintheoutputpowerlevelof the by microwaveamplifiertechradio.Microwaveoutputpowersareoftenconshained power when technologycanreachit') does exist (A of 10 W of ouryut limit nology. usinga40MHz of bandI 1-GHz mask radios the for emission 6.35 displays Figure radiousinga signaling 8-PSK power of a 90-Mbps spectrum shown is the width.Also the $-PSKpower specification, the emission with To be compatible rateof 30 MHz. betweenthesigper The difference power 4-kHz band, in terms of spectrumis shown requiredof the attenuation the minimum represents the FCC mask nal spectrumand provides densityof an information thereby signal filtered 8-PSK transmitfilter. The used by Col= frltering format and basic This modulation is the 90/40 2.25bpslHz. MDR-I1 radio, the digital in their 9O-Mbps lins/Rockwell [13]' Theemissionlimitationfor microwavebandsabove15GHz is definedas

PorYrrin {kHr bmcl rclstir'c to mcln outFut potw {dBl

FCCrmidon limitrtiom SX $.ct.um Fowr/{ kHr (3OirHt $gilllne .rtrl

-60 *70 -H)

h Frnqucncy(MHrl

Figure 6.35 FCC emissionmaskat I t GHz anda 90-Mbps8-PSKspectrum.

322

DIGITALMODULATION ANDRADIOSYSTEMS

A = l l + 1 0 1 o g , o B+ 0 . 4 ( p - 5 0 )

(p>50)

(6.27)

where A = attenuationin l -MHz band below mean output power (dB) B = authorizedbandwidth (MHz) P - percentremoved from carrier frequency The attenuationA must be at least I I dB but does not have to exceed56 dB. In conjunction with the emission limitations, the FCC has stipulated that no discretelines exist in the transmittedspectrum[44]. Thus no carrier componentscan exist and no repetitive data patterns are allowed to occur. The repetitive data pattems are effectively eliminated by using a scramblerin the transmitter and a descramblerin the receiver. As mentioned already, PSK and QAM are forms of double-sidebandsappressed carrier modulation so that the carrier terms are eliminated automaticallv as long as modulation is continuous.

6.4 HADIOSYSTEMDESIGN Theforemost design requirement of apoint-to-point radiosystem for telephony is operational dependability,usually referredto as availability. Availabitity is expressedas the percentageof time that a systemprovides a specified minimum quality of service. with analog systems,minimum performanceis determinedby the noise power in the receivedsignal.The performanceof a digital systemis determinedby the bit error rate. Typical objectives for bit error rates range from l0-3 for voice trafTic to 10-6or l0-7 for data traJfic. Recall that a bit error rate of l0+ correspondsto the thresholdofperceptibility fbr errors in a PCM voice signal. Typical designobjectivesfor microwave radio systemsspeciff availability on rhe order of gg.g\vo [45, 46]. Hence the maximum acceptableaccumulation of outage,due to all causes,is on the order of z hr per year. Radio systemavailability is dependenton equipmentreliability and path propagation characteristics.The high-availability objectivesof a typical radio systemmandate redundalt equipmentand often require redundantpath$.The needfor redundantpaths is determinedby the likelihood of atmospheric-inducedoutages(rain attenuationor multipath interference).Rain is a dominant considerationat higher carrier frequencies (above lt GHz), and multipath interference must be considered at all frequencies. Multipath fading is dependenton prevailing climate and terrain. Redundantradio equipment typically operatesin either a backup or a hot-standby mode with automatic protection switching. The transmissionpath is backed up with sparechannels(freguency diversity) or sparepaths (spacediversity receivers).In extreme ca$e$a backup route may even be utilized.

6.4.1 FadeMargins The main technique usedto circumvent atmospheric-inducedoutagesis to provide an exffa $trong signal at the receiver during normal path conditions. The difference be-

DESIGN 323 6.4 RADIO SYSTEM tween the normal receivedpower and the power required for minimum acceptableperformance is referred to as the fade margin. Greater fade margins imply less frequent occurrencesof minimum performance levels. Radios operating in higher frequency bandsgenerally require greaterfade margins becausethey are more susceptibleto rain attenuation.A 50-dB fade margin is typical for a digital radio at I I GHz, while a 40dB fade margin is typical for lower microwave frequencies. The amount of fade margin actually required for a particular route dependson the probability of multipath-induced fades and heavy rainfall occurrences.Thus drier climatespermit lower fade margins, thereby allowing greaterrepeaterspacing.In some mountain-basedmicrowave links in the western United States,microwave hops can be 100 miles long. By comparison,the averagehop in other parts of the country is less than 30 miles long. When large fade margins are provided, the received signal power during unfaded conditions is so strong that bit errors arevirtually nonexistent.Nevertheless,problems with extra strong signals do exist. Namely, automatic gain control in a receiver must operateover a wide dynanric range.If the maximum signal level into the demodulation and detection circuihy is not controlled, saturationis likely to degradepetformance, especially in high-density modulation formats such as 64- or 256-QAM, where information is encodedinto the signal amplitudes. To minimize dynamic rangerequirementsin a receiver and reduceinterferencebetween systems,adaptive transmitter power control (ATPC) is sometimesused [47]. ATPC usesa feedback data link from a receiving station to control the output power of a transmitting station.Thus when excesspower is unnecessary,it is not used' ATPC is commonly used in digital mobile telephone$ystemswhere interferencecontrol is a primary concern.

6.4.2 SystemGain One of the most impoltant parametersused to characterizedigital microwave system performanceis the systemgainA,. Systemgain is defined to be the difference,in decibels, of the transmitteroutput power and the minimum receivepower for the specified error rate;

Ar=

fr,\ r0ros,o ['p;j

(6.28)

where P.1= transmitter output power Pn = r€ceive power for specified error rate The minimum acceptablereceive power is sometimes referred to as the threshold power and is primarily dependenton the receiver noise level, the signal-to-noiseratio required by the modulation format, and various system degradationssuch as excess noise bandwidth, signal distortions, intersymbol interference,carrier recovery offsets,

324

DIGITAL MoDULATIoN ANDRADIoSYSTEMS

timingjitter, andcouplingandfilter lossesthateitherattenuate the signalor increase thenoiselevel. Noisepowerin a receiveris usuallydo:ninatedby thermalnoisegenerated in thefrontendreceiveramplifier.In thiscase,thenoisepowercanbe determinedasfollows: pN = (FXNfl(B) = F(kTo)B whereF= No= B= ft = 7o=

(6.2e)

thereceivernoise figure thepowerspectraldensityof thenoise thereceiverbandwidth' 1.38(10)-23 is Boltzmann's constant theeffectivereceivertemperatures in degreesKelvin

Equation6.29 essentiallystatesthat the receivernoisepower is determinedby the spectralnoisedensityof the receiverinput resistance andthe additionalnoiseinhoducedby the amplificationprocess(noisefigureF). Normally,a referencetemperatureof 290 K is assumed sothatthethermalnoisedensity(ftTo)is 4 x 10-21wFIz. The noisefigure of any deviceis definedasthe ratio of the input signal-to-noise ratioto theoutputsignal-to-noise ratio; (S/Mi, r_= (s/N)out

(6.30)

In effect,thenoisefigurespecifiestheincreasein noisepowerrelativeto theincrease in signalpower.sinceall physicaldevicesinhoducenoise,thenoisefigureofany system is alwaysgreaterthan I andis usuallyexpressed in decibels.If a systemhasno gainor attenuation, thenoisefigureis exactlyequalto theratioofoutputnoi$eto input noise.Noisefiguresof low-noisemicrowaveamplifierstypicallyrangefrom z to 5 (3-7 dB). Radioreceivernoisefiguresaretypically6-10 withouta low-noiseamplifier. combiningEquations6.28and6.29andincorporaring a termD for thedegradation from idealperformance producesthefollowing generalexpression for systemgain;

A,=ro**,,[ffi*rJ-"

(6.3 r)

whereSNRis thetheoreticalsignal-power-to-noise-power ratiorequiredfor themaximum acceptable errorrateandD includesall degradations from idealperformance. -Typically,

B is assumcd to be the minimum theoretical bandwidth for the particular modulation format in use. Excess bandwidth required by practical implementations is then incorporated rnro a sysrem degradation factor.

DESIGN 325 6.4 RADIOSYSTEM

raNoticethatthe SNRtermin Equation6.31refersto signal-power-to-noise-power tios andnot EblNO.Therelationshipbetwe€nSNRandE6lNsis providedin AppendixC. the Thesystemgain,in conjunctionwith antennagainsandpathlosses,determines fademargin: Fade margin = A, + 6r + GR + 20 log,n fu-Af - A0

(6.32)

where 6i1 = transmitter antenna gain (dB) Gq = receive antennagain (dB) L = tran$mittedwavelength Af = antennafeeder and branching loss (dB) Ao = free*spaceattenuation(distanced must be in the sameunits as l,) = 20logro (4nd)

Thedirectivity, andconsequentlythegainof, an antennais directly proportionalto the sizeof its apertureandinverselyproportionalto the squareof the transmittedwavelength.In determiningthe receivepower,however,it is actuallyonly the areaof the antennathatis importantandnot thedirectivityor gain.Thusradiosystemdesigners convenientlyconsidertransmitand receiveantennagainsas contributingto signal powerbut includea wavelengthnormalization(20log1sl,) to relatethegainof thereceivingantennabackto the sizeofits aperture. In additionto providingincreasedantennagain,greaterdirectivitiesalsoreduce paths,whicharisefrom greateremanation multipathproblems.Thelongersecondary Unfortunately,anangles,havelower powerlevelswhenthedirectivityis increased. in severalregards:Economically tennagainsarelimited by practicalconsiderations mechanicalalignmentis diffisizedtowerscansupportonly limited-sizedantennas, cult, anddirectionalstabilityof boththe antennaandthepathis limited. single Thefeederandbranchinglosses41includedin Equation6.32arisebecause reFurthermore, radios. for separate channels carry several antennasystemstypically availreceivers be and spare transmitters dictate that usually liability considerations or ablefor protectionswitching.The processof combiningsignalsfor transmission of attenuation amounts introduces inherently various reception distributingthemafter or splittingof signalpower. Example6.4. Determinethe systemgainof a l0-Mbps, z-GHzdigitalmicrowave repeaterusing4-PSKmodulationandan outputpowerof 2.5 W. Assumethe excess bandwidthof the receiveris 307oand ttrat other departuresfrom ideal performance Assumea noisefigureof 7 dB for the receiver,andthe amountto 3 dB degradation. desiredmaximumerrorrateis 104. Also determinethefademarginassumingantenna gainsof 30 dB eachanda pathlengthof 50 km. Thebranchingandcouplinglosses are5 dB. Solution. FromFigure6.17,therequiredvalueof E6lN0for 4-PSKmodulationcan as10.7dB. UsingEquation3.42(in AppendixC), it canbe determined bedetermined

326

DIGITAL MoDULATIoN ANDRADIoSYSTEMS

that the signal-power-to-noise-power ratio at the detectoris 3 dB higherthanE6lr/6. Thus,therequiredSNRis 13.7dB. since4-PSKmodulationprovides2 bps/rlz,the signalingrateis 5 MHz, whichis thetheoreticalminimum(Nyquist)bandwidth.Equation6.31cannow be usedto determinethe systemgain: 13.7*7-3*10losl.3 ) :l16dB At a carrierfrequencyof 2 GHz,thewavelengthis 3 x 108/2xlOe= 0. l5 m. Thusthe fademargincanbe determined from Equation6.31: Fademargin= I I 6 + 60 + 20 log,o(0.l5) - S- Z0 log,n(4ru5 x lOa) = 38.5dB

6.4.3 FrequencyDiverslty As mentioned previously, neither the transmitting and receiving equipment nor the path is normally reliable enoughto provide an acceptablelevel of systemavailability. Frequency diversity is one means of providing backup facilities to overcome both types of outages.A deep multipath-induced fade occurs when a signal from a $econdary path arrives out of phasewith respectto the primary signal. Since the phase shift produced by a path is proportional to frequency, when one carrier fades, it is unlikely that another carrier fades simultaneously. Frequency diversity involves the use of a sparetransmitter and receiver operating in a normally unusedchannel. Since separatehardware is used, frequency diversity also provides protection againsthardware failures. The simplest meansof implementing frequency diversity is to use one-for-one (l : l) protection switching as indicated in Figure 6.36. one-for-one protection switching implies that one sparechannel is provided for each assignedmessagechannel.when high-spectrumefficiency is required, it is generally necessaryto have only one spare for a group of N channels( I ; N protection swirching). In fact, the FCC has stipulated that, in somefrequencybands,a systemmust be implementablein one-for-Nconfigurations. The main impact of a I ; Nprotection systemis the complexity of the switching unit and the needto switch back to the assignedchannel in the first available hop so that a single sparechannel can be reusedrepeatedlyon a long roule. Frequency diversity normally does nothing to alleviate rain outages since all channels in a particular frequency band are simultaneously affected. when rain is a pafticular problem, it can be overcome only by using higher transmit powers or shorter hops.

6.4 RADIoSYSTEMDESIGN

327

-'Wrt

Figure 6.36 Protectionswitchingwith one-for-onefrequencydiversity.

6.4.4 SpaceDiverslty two asshownin Figure6.37,by verticallyseparating Space diversityis implemented, receive antennason a single tower. The resulting difference in the two paths is normally sufficient to provide independentfading at the two antenna$.Spacediversity is the most expensivemeansof improving the availability, particularly if separatereceivers for multiple channelsare used for each antenna.The cost can be minimized, however, by combining the two received signals in a phase-coherentmanner for input to a common receiver t451. This technique provides less hardware backup than when completely separatereceiversare used for each antenna.

6.4.5 Angle Diverslty Becausemultipath fading is producedby multiple incident rays alTiving at slightly different angles,protection from fading can be achievedby discriminating on the angle

Figure 6.37 hotection swirchingwith spacediversityreceivers.

328

DIGITAL MoDULATIoN ANDRADIoSYSTEM$

of arrival' Angle diversityutilizestwo side-by-side receivingantennaswith slightly differentangularelevationsto providethediscrimination. Althoughanglediversityis generallynot aseffectiveasspacediversity,it canimprovethe availabilityofdigital signals[48, 49] in applicationswherethe towerrequirements of spacediversityare impracticalor disallowed. 6.4.6 Adaptlve Equallzation Sincemultipatheffectsare frequencydependent, not all frequencies in a particular channelsimultaneously experience thesameamountof fading.Thusmultipathfading canproducenot only a generalattenuation ofthe receivedsignalbut alsotheequivalentof in-bandamplitude(andphase)distortion.Distortionof thespectrumamplitude producesa generaldegradation in the errorperformance overandabovethefenalty incurredby the generalattenuationof the signal.In widebanddigital radios,frequency-selective fading (as opposedto flat fading)hasprovedto be the dominant sourceof multipathoutage$.Fortunately,betterperformance canbe achievedif the spectrumamplitudeis equalized(adjustedto a uniformlevel). Sinceatmospheric-induced multipathinterferena varieswith time,removalof multipath-inducedamplitudedistortionrequiresadaptiveequalization.A commonapproachto adaptiveamplitudeequalieationmerely samplesthe energyat selected frequencies in the receivedsignalspectrum.A compensating filter shapeis theninsertedinto the signalpathto adjustall energysamplesto a commonlevel.This basic techniqueis usedin digitalradiosmanufactured by Bell Northern[45] andRockwell International[50].Reference[50] reportsa 7-dB improvementin effectivefademargin canbe achievedwith an adaptiveequalizer. Because channeldistortionsproduceintersymbolinterference in thereceivedbasebrurdsignal,time-domainequalizationwith adaptivetransversalfilters or decision feedbackequalizersis alsoused.Reference[51] contraststherelativemeritsof both approaches and providesan extensivelist of references.The use of high-density modulationformatssuchas256-QAMplacesverystringentrequirements ontheadaptive equalizerssotheseradioscommonlyusebothtypesof equalizers[47]. Frequency-selective fading,which comrptsonly a portion of a wide-bandwidth channel,is particularlydifficult to equalizebecauseit changesrapidly andrequires complicatedequalizerstructures(in the frequencydomainor the time domain).The problemcanbegreatlysimplifiedif, insteadof usinga singlecarrier,multiplecarriers (i.e.' MCM) within theallottedbandwidthareusedto createmulriplesubchannels. With thisapproach, fadingwithin anysubchannel is relativelyflat andeasyto equalize.Thedisadvantage of this approachis thecostof replicatedcircuiry for eachsubchannel. As reportedin [5?],however,atmospheric-induced outagesaregreatlyreduced. 6.4.7 Route Design Thelayoutof a point-to-pointmicrowaverelaysysteminvolvesnumerousconsiderations of thelocalterrain,prevailingatmospheric conditions,radiofiequencyinterferencewith othersymbols,andinterferencefrom onehop to anotherwithin a singlesystem.

REFERENCES 329

Path Length The foremost considerationin setting up a single hop of a microwave system is that line-of-sight transmissionis required. Using antennaheights of about 60 m, the curvature of the earth limits the transmission distanceto approximately 50 km. Longer line-of-sight distancesare possible if taller antennatowers are used.However, plsctical considerationsof mechanicalstability limit the height of a tower, particularly since longer distancesimply larger and more directive antennas. When antennasare rigidly mounted on the top of a building or the side of a mountain, mechanicalstability may not be a problem but the $tability of the path itself may becomea limiting factor. Under some atmosphericconditions, radio waves can be refracted to the point that a narrow beam completely misses the receiving antenna. Bending of the propagationpath can also causea f,rxedobstacleto intermittently obstruct transmission, implying that clearance between the notmal path and nearby physical obstructionsis required.Thus there is always a practical limit on how narrow the beam can be. Local teruainalso influences the prevalenceof multipath fading. Nearby bodies of water contribute significantly to multipath conditions during late evening or early morning hours when thereis litfle wind. Direct transmissionovor water is usually very difficult becauseof reflecfions off the water surface.Transmission over water often requireshigher antennasand some meansof blocking direct reflections.

Cochannellnbrterence

Since there are a limited number of channelsavailable for point-to-point microwave systems,the same channelSmust be used Over and over again. ReuSeof microwave fiequency bands is enhancedby the directivity of the antennasand the general need for line-of-sight reception.In many metropolitan areas,however, thereis so much traffic converging into one particular areathat it is impossible to completely isolate two systemsusing the samechannel. This type of interferenceis refened to as cochannel interference. Cochannel interf'erenceresults from converging routes or from overreach of one hop into anotherhop of the same systemreusing a channel. Sometimesreflections or atmosphericrefractions can contribute to overreach,even when direct line of sight is not present.Overreach is usually controlled by zigzagging the hops so that a beam from one tran$mitter misses all subsequentreceive antennasin a route. The reduced use of point-to-point microwave for long-distanceftaffic has reducedthe cochannel intetference in point-to-point systems.However, cochannel interference is a major considerationin the design and deployment of digital cellular systems,as discussedin Chapter 9.

REFERENCES I 2

Federal Communications Commission, Report and Order, Docket No. 19311' FCC 7 4-985, releasedSePt.27, 1974. ..SpecialIssue: AR6A Radio System," Bell systemTethnical Joumal, Dec. 1983.

330

DIGITAL MODULATION ANDRADIOSYSTEMS W. A, Conner, R. J. Girvin, and W. R. McClellan, ,.FrequencyControl of the MSR_6 single sideband system," IEEE International conference on communicarions. 19g3, pp, E2.5.1-E2.5.5. Y' Yoshida, Y. Kitahara, and s. yokoyama, "6G-90 Mbps Digital Radio system with I6-QAM Modulation," IEEE International conference on comrnunicafiozs, 19g0, pp,

52.4.r-52.4.5.

l0 l1

l2 l3

t4

l5

r6 t1

t8

l9

s. Pasupathy,"Minimum shift Keying: A spectrallyEfficient Modulation,"/EEE Comrnunications Magaline,July 1g79,pp. l4-ZZ. "continuous phaseModulation,"IEEE communications c.-E. sundberg, Magafine, Apr. 1986,pp.25-38, Y. Morihiro, s. Nakajima,andN. Furuya,"A 100Mbir/s hotorype MSK Modemfor satellitecommunications,"IEEE Transactionson communications,oct, 1979,pp. l512-1518. "hactical GMSK Data Transmission," MX-coM Application Note, rggg (www.mxcom.com). K. MurotaandK. Hirade,"GMSK Modulationfor Digital Mobile RadioTelephony," IEEE Transactions on Communications, July lg8l. R. E. Ziemer and w. H. Tranter,principlesof communications, HoughtonMifflin, Boston,1976. E. PanayirciandE, Y. Bar-Ness,"A New Approachfor Evaluatingtheperformance of a symbolTiming RecoverysystemEmployinga GeneralTypeof NonlinEarity IEEE ,,' Transailionson Communicarrans. Jan.1996. c' R. Hogge,Jr.,'carrier andclock Recoveryfor g-psK synchronous Demodulation," IEEE Transactions on Communications, May 1979,pp. 529_533. P' R. HartmannandJ. A. crosset,"A 90 MBS Digital Transmission systemat I I GHz Using8 PSKModulation,"IEEE Internationalconferenceon communications.lg16. p p .l 8 - 8 - 1 8 - 1 3 . I. Horikawa,T. Murase,and y. saito, "Designand performances of 200 Mbivs 16 QAM Digital Radio system,* IEEE Transactionson computers,Dec. 197g, pp, r953-1958. J. A' crosserand P. R. Hartmann,"64-eAM Digital ttadio Transmissionsysrem IntegrationandPerformance," IEEE Internationalcomrnunications conference.lgg4. pp. 636-641. w. A' conner,"DirectRF Modulation256eAM Microwavesystem,"lEEEGlobemm Proteedings, 1988,pp. 52.7.1,52.7.6. H. Matsue,T. shiraro,and K. watanabe,"256 eAM 400 Mb/s MicrowaveRadio system with DSP Fading countermeasures," IEEE Intemational conferenceon Communications, June1988,pp.41.5.l-41.5.6. K. PahlavanandJ. L. Holsinger,"A Modelfor theEffectsof pcM compandorson the Petformanceof High speed Modems," IEEE Grobecomproceediigs, 19g5, pp. 28.8.1*28.8.5. T. Ryu' J. uchibori, and y. yoshida,'A steppedsquare256 eAM for DigitarRadio system,"IEEE Intemationalconferenceon communicarions, 19g6,pp. 1477-l4gl.

331 HEFERENCES "6 20 M. Tahara,T. Deguchi,S.Mizoguchi,andY' Yoshida, GHz 140MbpsDigitalRadio

Systemwith 256-SSQAMModulation,"IEEE GlobecomProceedings,1987' pp' 38.3.1-38.3.6. "Performance with PartiallyCoherent of Offset-QPSKCommunications 7.1 S. A. Rhodes, ConferenceRemrd, Nov. 1973' pp' Detection." National Telecommunications 32A-l-32A-6. "Offset QuadranrreCommunicationswith 2? M, K. Simon and J. G. Smith, IEEE Transadionson Communftations, CarrierSynchronization," Decision-Feedback Oct.1974,pp. 1576-1584. "High Density Digital Data 23 J. W. Bayless,R, D. Pedersen,and J. C. Bellamy, Conference Record, 1976, pp. Transmission," National Telecommunications 51.3-l-51.3-6.

25

30

3l

33

"MSK and Offset QPSKModulatton,"IEEE S. A. Oronemeyerand A. L. McBride, pp. 1976' 809-820. Aug, on Communicafians, Transactions "MethodandApparahrs forWideband R. D. Gitlin, S.K. Rao,J.Werner,andN. Zervos, CentralOfficeand of Digifal SignalsBetween,for Example,a Telephone Transmission U.S.PatentNo.4,9U'492' May 8' 1990. CustomerPremises," "comparison of single-carrierand Multitone Digital Modulationfor B. Saltzberg, Magazine,Nov' 1998,pp. l14-122' ADSL Applications,"IEEE Communication't ,,Mid-rangePhysicalLayer Specificationfor Calegory3 UnshieldedTwisted-Pair"' ATM Forum,Sept.1994' afohy-0018.000, "Modulationconsiderations for a 9l Mbit/sDigital c. w, Andersonands. G. Barber, May 1978,pp. 523-528' on Communications, Radio,"IEEE Transactions .,FadeMargin flnd outage computationof 49-QPSRRadio D. P. Taylor andM. Shafr, Equalization,"IEEE Intemational Conferenteon Feedback Employing Decision 1983,pp. 1453-1458' Communications, A. J. Viterbi,"Error Boundsfor ConvolutionalCodesandanAsymptoticallyOptimum DecodingAlgorithm," IEEE Transactionson lffirmation Theory,Apr' 1967' pp' 260-269. "Trellis-CodedModulation with RedundantSignal Sets Fart G. Ungerboeck, Feb.1987'pp. 5-l l' Magaeirre, Introduction."IEEE Communicatktns "Trellis-codedModulationwith Redundantsignal setsPartII: state G. Ungerboeck, Magarine,Feb.1987'pp. t2-21 . IEEE Communications of theArt," "A R. J. F. Fang, Coded8-PsK systemfor 140-MbivsInformationRateTransmission over 80-MHz NonlinearTransponders,"Proceedings7th IntennationalConferenceon 1986,pp' t2-16' DigitalSatelliteCommunications, '"TrellisCoding l4'4kb/s Data T. Kamitake,K. Uehara,M. Abe, and S. Kawamura, with a single-chipHigh-speedDigital signalhocessor,"/EEE ModemImplemented GlobemmProceedings,I 987'pp. I 2.8.I - I 2.8.6' .,A Modemoperatingat DatasignalingRatesup to 28,800bpsfor use on theGeneral swirched TelephoneNetwork and on LeasedPoint-to-PointZ-wire Telephone-Type Circuits,"ITU-T Rec.V.34,Sept,1994. "GeneralizedMinimum shift-Keying Modulation Ramin sadr and J. K. omura, Jan.1988,pp. 32-40' of IEEETransactions Communications, Techniques,"

332

DIGITAL MoDULATIoN ANDRADIo SYSTEMS

37 J. A' C' Bingham,'-MulticartierModulationfor DataTransmission: An IdeaWhose Time HasCome,"IEEE Comtnunitations Magazine,May1990,pp. 5_14, 38 J. M. cioffi, "Asymmetric Digita-l subscriberLines," in The cornmunications Handbook,J. Gibson,Ed., CRCpress,BocaRaron,FL, 199g,pp. 450_479. 39 Digital Vidco Broadcasting (DVB): Framingstucture,channelcodingandmodulation for digitalterresrrialtelevision,EN 300 744,vl.Z.l, ETSI. lggg_01. 40 B. Saltzberg,"Comparisonof Single-Canierand Multitone Digita.lModulationfor ADSL Applications,"IEEE Comrnunications l4agaline,Nov. lg9g, pp. llL-llL. 41 E. D. sunde,"IdealBinarypulseTransmission by AM andFM," Eel/ systemTechnical Joumal,Nov. 1959,pp. 1357-1426. 42 w. R. BennetandJ. R. Davey,Data Transmission, McGraw-Hill.New york. 1965. 43 R. w. Lucky, J. salz, and E. J. weldon, Jr., principlesof Data communications, McGraw-Hill.New york. lg68. 44 FederalCommunications Commission,MemorandumOpinionand Ordermodifying FCCReponand Orderof Docket193I l, released JanuaryZg,1975. 45 G. H. M. de witte, "DRS-B:systemDesignof a Long Haul 9l Mb/s Digital Radio," I EEE NationalTelecommunicat ions Conference, I g7g, pp. 3g,l l _3g.1.6. 46 H. Kostal,D. R. Jeske,andv. K. prabu,"AdvancedEngineeringMethodsfor Digital Radio Route Design,"IEEE Intemational communitationsconference,l9g7, pp. l9B.6.l_198,6,6. "Digital 47 c' P. Bates,P.L. Penney,andK. L, seastrand, RadioTechnologyin theAT&T Network," IEEE Intemationar communications conference, 19g7, pp, l 9 8 . 5 . l- 1 9 8 . 5 . 7 . 48 E. A. SweedykandP. Balaban,"A MultipathFadingModelfor TenesrrialMicrowave Radio,"IEEE Glohecomproceedings,lgB8,pp. 52.6.1_52.6.1 . 49 s. H. Lin, "openingNew vistas for MicrowaveRadio,"gel/core Exchang,e,July/Aug. 1988,pp.9-12. 50 P. R. HartmannandE. W. Allen, "An AdaptiveEqualizerfor Correctionof Multipath Distortion in a 90 MB/s B-psK system," IEEE Intemational conferenceon Communications, 1979,pp,5.6.l-5.6.4. 5l J. K. chamberlain,F. M. clayton, H. sari, andp. vandamme,,.Receiver Techniques for MicrowaveDigital Radio," IEEE cownunitations Magazine,Nov. 19g6,pp. 43-54. 52 Y, Nakamura,H. ohtsuka,s. Aikawa,and H. Takanashi,'Advancedrechniquesfor super Multi-carrier Digital Microwave Radio with Trellis-coded 2s6 eAM proceedings,lggg, Modulation," IEEE Globecom pp. l l.3.l_l1.3.6.

PROBLEMS 6.1 To prevent the transmission of line spectra, digital radio terminals use data scramblers to randomize the data patterns. Furthermore, differential encoding is normally required for proper datadetection.Both functions inffoduce error mul tiplication. If the combined effect of theseoperationscausesan averageof five

PROBLEMS 333

decodedenors for everychannelenor, what is the effectivepenaltyin transmit powerfor a 4-PSKsystemat BER = 10-6?At BER = 10-3? errorsat a rate 6.2 If a digitalradioreceiveris experiencingthermal-noise-induced from of I per 106bits, whatis thenewerrorrateifthe pathlengthis decreased 30 to ?5 miles? 6.3 DeriveEquation6.18. 6,4 what is the minimumtheoreticalbandwidthof an 8-PSKsignalcarrying4800 bps? 6.s A 32-QAM signalsetis implementedby eliminatingthefour cornerpointsof a 36-QAMsignal.what is theminimumerrordistancein termsof thepeaksignal power?How doesthis answercompaleto theerror distancefor antipodalsignaling?(Expresstheanswersin decibels') ratio of a 32-QAM signal(Problem6.5) assuming 6.6 what is the peak-to-average of randomdata? transmission compareto 32-PsKenor 6.7 How does32-QAM (Problem6.5)eruorperformance performance? (a) In termsof averagesignalpowers? (b) In termsof peaksignalpowers? systemusing4-PSKmodulationprovidesanerrorrateof 6.8 A carriertransmission bandl0-6. tf themodulationis changedto 16-PSKto naffowthetransmission to how muchmustthehansmitpowerbe increased width (datarateunchanged), maintainthe sameerror rate? 6.9 RepeatProblem6.8 assumingthat the bandwidthis unchangedandthe datarate is doubled. What is theerrorrateof anidealI6-QAM signalwith a signal-to-Gaussian-noise 6.10 ratio(SNR)of 18dB? 6.11 What is the error rate of hoblem 6.10 if interferenceis also present at a level that is 21 dB below the received signal? (Assume the effect of the intefference is identical to Gaussiannoise at the samerms power.) 6.r2How much must the transmit power in Problem 6.1 I be increasedto off'setthe effect of ttre interference?(Assume the interferencelevel is fixed.) Problem 6.1?, but assumethat the interferenceincreasesin proportion to Repeat 6.13 the increasein transmit power. (Signal powers in the adjacentchannelsare increasedalong with the desired signal power.) Problem 6.3 assuminga starting stateof DRepeat 6.14

RONIZATION SYNCH NETWORK CONTROLANDMANAGEMENT systemsare of transmission requirements In Chapters4 and6 somesynchronization discussed.Theserequirementsinvolve carier recoveryfor coherentdetectionof modulatedsignals,clock recoveryfor samplingincomingdata,and framingproceduresfor identifyingindividualchannelsin a TDM signalformat.All of theseconsideratronsareinherentin digital transmissionsystemsand,for the mostpart, operate deof otherequipmentin a network.An instanceof one subsystem's independently pendencyon anotherwasnotedfor Tl lines.Tl sourcedatamu$tincludeminimum link. In contrast,other densitiesof 1's to maintaintiming on theoriginaltransmission independentlyof the synchronization clock line codesare describedthat maintain sourcedata. This chapterdiscussesnetwork-relatedsynchronizationconsiderationsfor interconnectingvariousdigital transmissionand switchingequipment.Foremost among theseconsiderationsis synchronizationof switching equipment'When digital switchingequipmentwasfirst installedin thepublic telephonenetwork,the interfaceswere analogtransmissionsystems.Hence,each switchingmachine could operatewith an autonomousfrequencysource(clock) that convertedall voice signalsinto digital signalswith preciselythe samedatarate (nominally 64 kbps).Theseswitchingmatriceswere designedto carry one channelrate and one channelrate only. The advent of subsequentdigital switch interconnectionrequired switchesto carry digital channelsoriginating someplaceelse in the network-from a different frequency source. Thus, network synchronization requirementsarose. transmissionand switchingequipmentsare interWhen individual synchronous thateithersynneedto beestablished to form a network,certarnprocedures connected chronize the clocks to each other or provide for their interoperabilitywhen each clocks.Followingthe discussionof networkclock synusesindependent subsystem is extendedto otheraspectsof network chronization,the conceptof synchronization control.

336

NETWORK SYNCHRONIZATION CONTROL ANDMANAGEMENT

7.1 TIMING All digital systemsinherentlyrequirea frequencysource,or ..clock,"asa meansof timing internaland externaloperations.operationstimed from a singlefrequency sourcedo notrequireparticularlystablesourcessinceall commonlyclockedelements experience timing variationsin common.A differentsituationoccurswhentransfers aremadefrom one synchronousequipmentto another(asfrom a ffan$mitterto a receiver).Evenif theclockof thereceivingterminalis "synchronized" to thetransmitting terminalon a long-termor average basis,short-termvariationsin eitherclockmay jeopardizethe integrityofthe datatransfer.Thusit is generallynecessary to usefrequencysource$(oscillators)in boththe transmitterandthereceiverthatareasstable asis economicallyfeasible. 7.1.1 Tlming Recovery: Fhaee-Locked Loop A commonmeansof synchronizinga receiverclock to a transmitterclock usesa phase-locked loop (PLL) asshownin Figure7.l. A phasedetectorcontinuouslymeasuresthe phasedifferencebetweenthe incomingclock anda locally generated clock. Thephasedetectorin Figure7.1merelymeasures thedifferencein thezerocrossings betweenthe two signals.Whenthe zerocrossingof the line clock precedes the zero crossingof thelocalclock,a positivevoltageis generated; otherwise,a negativevoltageis produced.The outputof the phasedetectoris filteredto eliminateasmuchreceivenoiseaspossible,andthenthephasemeasurement adjuststhe frequencyof the voltage-controlled oscillator(VCo) to reducethephasedifference.Someamountof noiseor interference inevitablypasses throughthephasedetectorandthefilter, causing erroneousadjustments in the vco frequency.As time passes,however,a frequencyoffsetproducesever-increasing phaseshifts.Whenthephasedifferencebuilds up,it is easierto detect,andtheappropriate changes in theVCo occur.Hencethelocal

Figure 7.1 Phase-locked loop clockrecoverycircuir.

rMrNG

337

clock maintainsthe desiredaveragefrequencybut inherenflyproduces$hort-termvari"hunts"theunderlyingfrequencyofthe line clock. ationsofphaseandfrequencyasit Theline clockin Figure7.1is shownto havea transitionin everyclockinterval,a line codes.With someline or diphase-type situationthatdoesoccurwith Manchester duringintervalsof transitions are no clock (bipolar particular) there in or AMI codes artificialtransiinserts circuitry clock extraction the binary0's. In thesecaseseither duringinis disabled phase previous detector or the intervals from tionsexffapolated pulse is detected. tervalswhenno 7.1.2 Clock InstabilltY The variationin the outputfrequencyof the VCO describedaboveis an exampleof clock instability.All clockshavea certainamountof instability-even free-running oscillators.An importantaspectof clockinstabilityis its frequency:therateat which the clock frequencychangesfrom beingtoo high to beingtoo low. The frequencyof the instability canbe directly observedasthe frequencyspectrumof the VCO control voltagein a PLL clockrecoverycircuit.WhentheVCO controlvoltagevariesslowly, the variationsarereferredto asclock wander.Whenthevariationsaremorerapid,the instabilityof the clock is refenedto asjitter. The mostgenerallyaccepteddividing pointbetweenwanderandjitter is 10Hz. Purewanderon a Tl line thereforeproduces positivephaseerrorsfor morethan77,200bit intervals(0.05sec)followedby more than77,200bit intervalsof only negativepha$eerrors.*Themainsourcesof clockinstability(bothwanderandjitter) in a networkare; 1. Noiseandinterference media 2. Changesin the lengthof transmission 3. Changesin velocityof propagation 4. Dopplershiftsfrom mobileterminals 5. Irregulartiming information

Nolseand lnErterence

with a verylow cutofffrequency,it canfilter If thelow-passfilter of a PLL is designed link thatwouldotherwise on a transmission outalmostall of thenoiseandinterference comrptthetiming recovery.Therearethreemain reasonswhy arbimarilylow passfil(e.9.,lock terscannotbeused.First,theability ofthe PLL to acquiresynchronization on to theunderlyingclock)is inverselyrelatedto thePLL bandwidth'If the VCO beginsoscillatingat thewrongfrequencyandthebandwidthis too narrow'thePLL may thisproblemis neverpull theoscillatorto thefrequencyof theline clock.Sometimes, and by usingtwo bandwidths:a wide oneto acquiresynchronization accommodated a narrowonethatis selectedafterlock is achieved' +As discussed later, observing wander in this way requires extremely stable clocks as a reference. The normal PLL clock recovery circuitry cannot be used to observe wandet because the VCO tracks the relatively long-term phase offsets, Thus the clock recovery citcuit does not filter out thc wanderbut passes it on.

338

NETWoRKSYNcHHoNIZATIoNcoNTHoLANDMANAGEMENT

A secondconsiderationthat generally precrudesthe useofvery nanow filters, even after acquisition, is that the sourcemay vary in frequency that cannot be tracked by a slowly respondingPLL. In this case,the recoveredclock doesnot track the ideal sample times leading to high error rates or, worse yet, the pLL loses synchronization altogether and has to reacquire lock. The third limiting factor for low-bandwidth PLLs is the instability of rhe VCO irself. If the vco begins to drifr in frequency, very low bandwidth filters preclude adjusting the VCO input voltage soon enough to prevent bit errors or possible loss of synchronization. Because operational considerations dictate certain minimum PLL bandwidths. noise and interferenceon the transmissionlink always causethe recoveredclock to be more impaired than the sourceclock. The pLL does, however, eliminate that portion of a disturbancewith frequency content abovethe bandwidth of the PLL. Thus disturbanceswith low-frequency content are the most difficult to deal with. systematicjitter, as produced by particular pattems of intersymbol interference, can have an arbitrarily low frequency content and is discussedmore fully in a later section. An important considerationin the design of a digitrrl transmissionlink is the accumulation ofjitter in tandemclock recovery circuits. If a recoveredclock is usedto time the transmissionof outgoing data, as in a regenerativerepeater,some amount of incoming jitter is imbedded into the outgoing clock. The clock recovery circuit in the next receiver tracksits incoming clock but introduceseven morejitter due to noise and interferenceon the secondsection. Thus jitter accumulatesat every regenerativerepeaterusing its received line clock as its transmit clock. If there is a large number of regenerativerepeaters,thejitter can accumulateto a point where subsequentclock recovery circuits have difficulty tracking the receive clock, produce sampling error$, and possibly lose lock.

Changeeln Lengthof TransmiesionMedia Path length changesoccur as a result of thermal expansionor conffaction of guided transmissionmedia or of atmosphericbending of a radio path. while a path is increasing in length, the effective bit rate at the receiver is reducedbecausemore and more bits are being "stored" in the medium. similarly, as the path shortens,the bit rate at the receiver increa$e$becausethe number of bits storedin the transmissionlink is decreasing.After the path has stabilized, the receive signal returns to the nominal data rate. The mo$t significant changesin path length occur with communications satellites' Geostationarysatellitesproduce path length variations of approximately ?fi) miles, which causepropagationtime variations of approximately I msec I I . path length changes I also occur in guided transmission media such as copper wire and optical fibers. These changesare referred to as diurnal changesbecausethey occur once a day.

Changesin the Vetocityof propagation Temperaturechangesnot only causeexpansionand contraction of wireline transmis* sion media but also can changethose propagation constantsof the media that deter-

7.1 rMrNG

339

mine the velocity of propagation.The resulting changein received clocked stability, however, is much less than that produced by the changein path length [2]' The propagation velocity of radio waves in the atmosphere also changes with temperature and humidity. Although these velocity changes are more significant than those occurring in wirelines, they are still smaller than the path-length-inducedvariations. Notice that a change in propagation velocity is effectively equivalent to a "stored" in the transmission path is change in path length since the number of bits changed.

Doppler Shlfts The mostsignificantsourceof potentialtiming instabilityin a receivedclock occurs asa resultof Dopplershiftsfrom airplanesor satellites.For example,a Dopplershift inducedby a 350-mphairplaneamountsto anequivalentclockinstabilityof 5 x 10*7' Dopplershiftsequivalentto Digital mobiletelephonereceiversmust accommodate clock instabilitiesof aboutonepartin 107.*Again,Dopplershiftsoccur,in es$ence, asa resultof pathchanges. Ineg u lar Tlm ing lnfo rmetlo n requirementof a digital line codeis thatit in Chapter4, a fundamental As discussed provide sufficienttiming informationto establishandmaintaina receiverline clock. jitter in therecovered durclockincreases If thetiminginformationis datadependent, ing periodsof relativelylow densitytiming marks.Themagnitudeof thejitter is de* pendentnot solely on the densityof timing marksbut also on the timing (data) patterns.In an ideal repeater,only the densitywould matter.In practice,however, jitter [3]. variousimperfectionsleadto pattern-dependent level digitalmultiplexersinsertoverhead higher in this chapter, later As discussed purpo$es. When the higherrate data for various stream bits into a compositedata individual channelsis irregular' within rate of data the arrival streamis demultiplexed, jitter generating new line clocksfor thelower produces when timing This inegularity jitter) and troublesome (waiting the most is often time ofjitter This source ratesignals. later. in more detail is discussed 7.1.3 Elastic Stores essentiallyrepresent The timing instabilitiesdescribedin the precedingparagraphs the caseof noise-and link. In changesin the numberof bits storedin a transmission "bits stored"occursbecausedataare samthe changein interference-inducedjitter, pleda little earlieror a little laterthannominal.Sincetheoutgoingdataof a regenerative repeateraretransmittedaccordingto the recoveredclock, a phaseoffsetin the clockmeansthedelaythroughtherepeateris differentfrom whenthereis nomisalignmentin timing. *If

you *e pa*noid about cooperation between cellular operators and law enforcement authorities, you may not want to use your cell phone while speeding.

340

NETwoHKSYNcHHoNIZATIoNcoNTRoLANDMANAGEMENT

Ifphase offsets in successiveregenerativerepeaterscoincide, a net changeof several bits of storagein a long repeatedtransmissionlink occurs. Sincetheseexffa bits enter or leave the hansmission link over relatively short periods of time, the accumulated jitter rnay representa relatively large, but short lived, instability in the receive clock. Becauseregenerativerepeater$use incoming sample clocks as output clocks, sustained timing differencesbetween inputs and outputs do not exist. The endpointsof a transmissionlink, however, may interface to a local clock. In this casea difference between a received and a relatively fixed local clock must be reconciled with an elastic store' An elastic store is a data buffer that is written into by one clock and read from by another.If short-term instabilities exist in either clock, the elastic storeabsorbsthe differencesin the amount of datatransmittedand the amountof datareceived.An elastic store can compensateonly for short-term instabilities that produce a Iimited difference in the amountsof data transmitted and received.If sustainedclock offsets exist. as with highly accuratebut unsynchronizedclocks, an elastic store will eventually underflow or overflow.

TDld-Swltch ln|errtace A typicalneedfor anelasticstoreoccurswhena digitaltransmission link is interfaced to a digitaltime divisionswitch.As shownin Figure7.2,theelasticstoreis placedbetweentheincomingdigitaltransmission link andtheinlet sideof the switch.In most instances thedigitalswitchprovidestimingfor all outgoingTDM links sotharno timing discrepancies existbetweentheselinks andtheswirch.For thetimebeing,assume that the far end of the digital link derivesits clock from thereceivesignalanduses thatclockto timedigitaltransmissions returningto theswitch.Thisis thesituationthat ariseswhena remotechannelbankis connected to a digital swirchthroughTl ]ines andis commonlyreferredto aschannelbank"loop timing."when looptimingis used, the line clock on the incominglink of the switchis nominallysynchronized to the switchclock.However,for rea$onsdiscussed previously,a certainamountof instability in the incomingclock necessarily exists.The elasticstoreabsorbstheseinstabilities sothatpurelysynchronized dataareavailablefor the swirch.

Figure 7.2 store.

Interface between TDM nansmission link and a digital switch using an elastic

TrMrNc 941

links andtheelasticstoremaintains In essence, theloopformedby thetransmission a constantand integral numberof clock intervalsbetweenthe inlet and outlet of the switch.Thus,from a timingpoint of view,theinletsandoutletsoperateasthoughdito eachotherusinga commonsourceof timing. rectlyconnected Removal of Accumulated Jitter Anotherapplicationfor an elasticstoreis shownin Figure7.3,whereit is usedto reNormally,aregenrepeater. timingjitter in a regenerahve movetransmission-induced erativerepeaterestablishesthe transmittiming directly from the locally derived local sampleclock.In Figure7.3,however,thetransmittimingis definedby a separate theshoft-terminstabilitiesin thereceiveclock,butthe clock.Theelasticstoreabsorbs "avlong-termfrequencyOfthetransmitclock is controlledby maintaininga cefrain to eragelevel of storage"in theelasticstore.Thusthetransmitclockis synchronized theline clockon a long-termbasis,but not on a short-termbasis.Ifthe elasticrltoreis all transientvariationsin thedatarate,high-frequency largeenoughto accommodate instabilityof theinputclockis removed. usedto recovertiming,deof themechanism regardless repeaters, All regenerative rive their outputclocksby averagingthe incomingtiming informationovera period of time.Tunedcircuitsaveragetheincomingclockfor relativelyfew signalintervals; phase-locked loopsdo sofor manyintervals.In all casesa certainamountof storage to increasethe available or delayis implied.An elasticstoreis merelya mechanism delay so that output timing adjustmentscanbe mademore gradually.Notice that an elasticstorealwaysinsertssignificantamount$of artificial delayinto the datapath' To removearbitrarilylow frequencyinstabilities(wander),aftihalily large elastic stores(andarbitrarilystableVCOs)arerequired.Thusa jitter-removingelasticstore the threatens only whenjitter accumulation shouldbeinsertedinto achainof repeaters ability of a regularregenerativerepeaterto maintainsynchronization' link but alsooccursin digiof a digitaltransmission Jitteris notjust a phenomenon For example,jitter removallike that shownin Figure7.3 is used tal storage$y$tems. in laserdiscptayersto eliminateblurringof thevisualimages[4]. Elastlc Store ImPlementdtiong The requiredsizeof an elasticstorevariesfrom a few bits to severalhundredbits for links.Figure7.4 showsonemeansof imcommunications long-distance high-speed, converter,a register,and plementinga smallelasticstoreutilizing a series-to-parallel

repeater. regenerative Figure 7.3 Jitter-removing

342

NETWoRK sYNcHRoNIzATIoN coNTRoLANDMANAGEMENT Input clock Input rignrl

Ortput clock OrrtDUtdrtr

Figure 7.4 Basicimplementation of an elasticstore. a parallel-to-seriesconverter.As indicated, incoming data are transferredinto the register as soon as each word is accumulatedin the series-to-parallelconvefier. Some time later, data in the register are transferredto the output parallel-to-serierrconverter as a complete word is shifted out. Notice that transfersto the parallel-to-seriesconverler are independentof the incoming clock. As long as ouFut transfbrsoccur between input transfers,no data are lost, and short-term jitter is absorbedby varying delays through the elastic store. Normally, somecontrol circuitry (not shown) is neededto initialize the elastic store so that the first transferinto the registeroccurs midway betweenoutput transfers.This processmeanssomeincoming dataare initially discardedby the series-to-parallelregister until the desiredtransfer time occurs. The relative times of the parallel transfersinto and out of the holding registerprovide a direct indication ofthe relative phaseofthe input and output clocks. Thus the parallel transfer clocks contain the information neededto generateVCO control voltagesif the elastic store is being used to remove accumulatedtransmissionjitter. The basic structureshown in Figure 7.4 can be extendedto implement larger elastic stores,as shown in Figure 7.5. The only changeinvolves the substitution of a f,rrst-in, first-out (FIFO) buffer for the holding register of Figure 7.4. This data buffer i$ designed specifically to allow input transfersunder the control of one clock while output$ are controlled by a different clock. Normally, the FIFO buffer is initialized by inhibiting output transfers until it is half firll. In fact, some commercially availabll FIFO buffers have an output signal specifically indicating when it is half full or greater.

7.1.4 Jitter Measurements A simplecircuitfor measuring timingjitteris shownin Figure7.6.As indicated, it is

nothing more than a phase-lockedloop (pLL) with the output of the phasecomparator providing the measurementof the timing jitter. Normally, the bandwidth of the lowpassfilter (LPF) is very small so the vco ignoresshort-termjiner in the timing signal. If there is no jitter at all, the output of the phasecomparatoris constantand no signal is passedby the high-passselectionfilter (HpF).

7.1 TrMrNc 343 Inrutfita Irput clalt

Outputclock Output (htr

of anelasticstore. Figure7.5 FIFOimplementation by thecircuit in Figure7.6 because Very low frequencyjitter ciurnotbe measured jitter maybe of no contheVCO tracksslowlychangingphaseshifts.Low-frequency cern,however,becauseit canbe hackedby a PLL. Higher frequencyjitter, on the otherhand,is moreaptto causesamplingerrorsor a lossof lock in theclockrecovery Thusthespectralcontentof thejitter aswell asits magnitudeis of interof a repeater. jitter, the circuit of Figure7.6 est.Besidesnot beingableto measurelow-frequency cannotoperateif it cannotlock ontothefundamentalclock signal,Section7.4.7describesanotherway to measurejitter andothertiming impairmentswithin a network. theoristsasa powermeasPhasejitter is commonlyspecifiedby communications urementin unitsof radianssquaredor cyclessquared.(Onecycle= Zruradians')As of thevarianceof thenumindicatedin Figure7.7,phasejitterPoweristhenameasure link. In a physical berof clockcyclesor unit intervals(UIs) storedin thetransmission timing variations,not sensejitter "power" haslittle meaningbecauseit represents jitter asa powercanbe obtainedby power.Somephysicaljustificationfor expressing observingthat theIms power o| of the phasedetectoroutput signalis propottionalto therms phasejitter ofr:

2 at

Jitt0f

Clock rlgnal

Phi|e lock loop

timingjitter' Figure 7.6 Circuittbr measuring

344

NETWORKSYNCHRONIZATION CONTROLAND MANAGEMENT Prohbility dkfibution of ryrtolr in trrnrlt

of nuftbGr

lntunbncou| numblr = /V Ar,6rgp - ilo

Trsn|ftrhtfi *ith it|bh rimingsurE+

Rtcdrnr

Figure 7.7 Phase jitter modeled as the variance in the number of symbols ',stored" in the transmissionlink.

ot,=K3siry2)

(7.1)

whereKa is thephasedetectorgainfactorin voltsperradian. Example7.1. Givenan rms phasejitter of 10.7dB relativero oneuI, what is the standarddeviationof thephaseoffset? Solutinn. Thevarianceof thesignalphaseis determined = asofr: l0 exp(10.7/10) 11.76UI squared. Hence,the standarddeviationis (ll.?6)u2 = j.4j UI (symbol intervals)'Since687oof a normalprobabilitydistributionlies within one standard deviation,thephaseof thissignalis within +3.43symbolintervalsfor 687oof thetime. Onepercentof the time the signalphasewill be outside2.6 standarddeviations.or +8.9symbolintervals. If phasejitter arisesas a resultof additiveGaussiannoiseon a stablesignal,the phasenoisecanbe approximated as

4:#

(rad2)

(7.2)

whereofr= additivenoisepower Pr = signalpower Equation7.2 is the basicequationofphasejitter producedby additivenoiseon a continuoussinusoid[5]. when timing is exhactedfrom a datasignal,the timing informationis usuallynot continuous.The distinctionis not importantbecause jitter producedby additivenoiseis normallyinsignificantcomparedto othersources [6]. For an analysisof phasejitter producedby regenerative repeaters operatingon randomlyoccurringtiming transitions,seereference[7].

TrMrNc 3rt5 TABLE7.1 Maxlmumlnstabllltyof D$l Customerlnterlace Variation Peak-to-Peak Maximum

Band Frequency

28 Ulsin 24 hr 5 U l si n 1 5m i n 5 Uls 0 , 1U l s

<10H2 < 10Hz 10 Hz-40 kHz I kHz-40 kHz

As an example of a particular instability specification, Table 7.I lists the wander and jitter specificationsof a DSI digital carrier (Tl) interface to the public network [8]. Notice that the higher frequencyjitter specificationallows deviationsin pulse centers of only 5Vo,which effectively specifies the accuracy of the data sample clock' Larger variations are allowed for lower frequency instabilities becausethe clock recovery circuits can track the changesto maintain a good sample clock.

7.1.5 $ystematlcJitter wasreportedby Byrne' An originalanalysisof jitter in a chainof digitalregenerators Karafin,andRobinson[3]. Figure7.8 showsthebasicmodelof their analysis'Each of theregenerativerepeatersin a T-carrierline extractstiming from thereceivedwaveasa transmitclock.Becauseof form andpasses thattiming on to thenextregenerator (primarily intetference)in the timing reintersymbol implementationimperfections jitter on the datapatterns'One is dependent produced repeaters by coverycircuits, pattemproducesan exphase lag. pattern produces exheme Another an worst-case to theother,a phase worst case tremephaselead.Whenthedatflpatternshiftsftom one thejitterprothesamebasicimplemenfadon, everyrepeaterha$ rampoccurs.Because natureofthisjitter tendsto be coherent.Thesystematic ducedby individualrepeaters line clockjitter at the end of a sourceof accumulated makesit the most signiflrcant chainof repeaters.

I I

Phuodritt $rilih rr Stvitch ln drtr ptt$7nt

indlvlilirl ttlpetif

Figure 7.8 Model of systematicjitter in a string of regenerativerepeaters;(D2= phaseshift datapattemfor phaselead;(Dt = phaseshift producedby worst-case producedby worst-case datapattemfor phaselag.

7

346

NETWoRKSYNCHRoNIZATIoNcoNTRoLANDMANAGEMENT

As indicated in Figure 7.8, the last repeaterin the chain experiencesa large phase ramp equal to the number of repeaterstimes the phaseshift of eachindividual repeater. This phaseramp representsan abrupt changein the clock frequency that may cause bit errors or a complete loss of synchronization.Thus there is a limit to the number of repeatersthat can be used without jitter removal. For more analysesof jitter accumulation including the combined effects of systematicjitter and rzurdomperturbations, seereferences[9J, [0], and [11].

7.2 TIMINGINACCURACIES In the precedingsectionthe natureof certain instabilities or transientvariationsin timing was discussed.Although thesevariations representshifts in the frequency of a line clock, the shifts are only temporary and can be absorbedby elastic stores.In some instancesdigital communicationsequipmentusing autonomousfrequency sourcesmust be interconnected.when this happens,the clock ratesof the two systemsarenever exactly the $ame,no matter how much accuracyis designedinto the frequency $ources. An offset in the two clocks, no matter how small, canaotbe reconciledby elastic stores alone. In the preceding section,channel bank loop timing was mentioned as an example of how remote terminals are synchronized to a digital switch. When the remote terminal is anotherdigital switch using its own frequency sourceas a reference,a difTerent situation results. As shown in Figure 7.9, the outgoing clock for each direction of transmissionis def,rnedby the local switch clock. Thus the incoming clock at each switch interface containsnot only tran$mission-line-inducedjitter but also a small and unavoidablefrequency offset.

7.2.1 Slips As indicated in Figure 7.9, the interface of eachincoming digital link necessarilycontains an elastic store to remove transmissionlink timing jitter. The elastic store at the first digital switch is written into by the recoveredline clock but read from the local rateR1.If the averagerate of the recoveredline clocklR2is different fromRl, the elastic

Figure 7.9 Connections betweenautonomously timeddigital switches.

7,2 TIMINGINACCURACIFS 347

storewill eventuallyunderflowor overflow,dependingon whichrateis larger.When R2is greaterthanR1,the elasticstoreat the first digital switchoverflows,causinga lossof data.If R2is lessthanR1,thesameelasticstoreunderflows,causingextraneous datato be insefiedinto the bit sffeamenteringthe switch.Normally,the extraneous dataarea repetitionof databits alreadytransferred into theswitch.Disruptionsin the datastreamcausedby underflowsor overflowsof an elasticstorearereferredto as "slips." Uncontrolledslipsrepresent very significantimpairmentsto a digitalnetworkbeThetefore,slips are alcausethey generallycausea loss of framesynchronization. lowed to occuronly in prescribedmannersthat do not upsetframing.One general approachto controllingthe slipsis to ensurethattheyoccuronly in theform of a repetition or deletionof an entireframe.Thusthe time slot counter$andframing logic asConholledslipscomprising sociatedwith the multiplexgroupremainsynchronized. entireframescanbe assured by usingelasticstoreswith at leastoneframeof storage. As a slip occurs,the storagelevel in the elasticstoreis effectivelyincreasedor decreasedby a full frame.Ratherthanactuallyinsertingor deletinga frameof information, the desiredeffect is achievedmore easily by indexingaddresspointersin a random-acce$$ memory.Sucha systemis shownin Figure7.10. writing input information by sequentially Theelasticstorein Figure7.10operates to individualTDM channels. into memoryaddresses corresponding Datafor individin thesamesequential ual outputchannelsareobtainedby readingthesameaddresses manner.Ideally,if thereis no offsetbetweenthe clock rates,the readtimesof each channels.The elastic channeloccurmidway betweenwrite timesfor corresponding delayvariationsup to one-half storethenhasthecapabilityof absorbingtransmission of a frametime.

Memory wrltS* Memory reodr

Double reEd

Figure 7.10 Hlasticstoreoperationwith a one-framememory,

348

NETWORKSYNCHRONIZATION CONTROLANDMANAGEMENT

The timing diagram in Figure 7.10 depicts an exaggeraredtiming offset in which the switch clock R2 is greaterthan the incoming clock lt1. As indicated, the read times carch up gradually with the write times until a "double read" occurs.At that time the information retrieved for eachcharurelis a repetition of the information retrieved for the previous outgoing frame. Although write and read times for only one channel are shown, the corresponding times for all other channelshave the samerelationship. Thus all charurelsslip together.Notice that R1 is greater than R2, a slip occurs when a "double write" on all channelscausesthe information in the previous incoming frame to be overwritten. The elastic store operation depicted in Figure 7.10 is very similar to the operation of a time slot interchangememory describedin Chapter5. This relationshipis exploitable in a TST switch where the inlet memory can provide both the elastic store function and the time switching function. When the two functions are combined, slips generally occur at different times for different channels. Nevertheless,individual charnels maintain proper frame alignment since each channel is transferredthrough the inlet memory using dedicatedmemory addresses. One attractivefeatureof using the inlet memory as an elastic storeis that, when setting up a new connection,an internal switching time slot can be chosenso that the inlet memory read is halfway betweeninlet memory writes for the particular channel.Thus a slip in that connection will not occur for a long time, probably not until long after the connection is released.(With a clock inaccuracy of one pafl in 108,ttre time between slips in any one channel is 3.5 hr.) One potential problem with the elastic store in Figure 7.10 occurs when write and read times nearly coincide. When both accessesto a single channeloccur one after the other, transient timing instabilities can causethe two accessesto cross back and forth with respectto each other. Thus slips causedby double readsmay follow slips caused by double writes and vice,versa.To remedy this situation, some amount of hysteresis is neededin the counter adjustmentprocess.The hysteresis,in turn, implies that additional storageis neededto defer the occurrenceof one type of slip after a slip of the other type has recently occurred. One meansof implementing an elastic storewith the desiredhysteresisis to usetwo framesof storageas shown in Figure 7.11. Forconvenience,the elastic storeis divided into an A-frame memory and a B-frame memory. The counterlogic again accessesthe memories in sequentialfashion except that frames are written alternately into the A and B memories.Under normal operation,the memoriesare accessedin the sameway for output data. When a slip is imminent, however, control logic causesthe output channelcounter to be reset so that the A memory is read twice in a row. This situation is depicted in the timing diagram of Figure 7.1I, which again assumesthat R2 is Sreaterthan R1.The imporrant point to be noticed in the timing diagram is that after the counter adjustmentproducesa double read of memory A, the write and read times of each individual memory are approximately one frame time apart. Thus anotheradjustment can be deferred until the write and read accessesagain drift one full frame time with respectto eachother. The structure and mode of operation shown in Figure 7.11 describethe elastic store used for DSI signal interfacesto AT&T's (now Lucent's)No. a ESS II2].

7.2 TIMINGINACCURACIES

n,$

349

VP

r

J I

I awntr I l \

I

A

H

B

J}

fi2

P/S A

ilhmory r$lt||

A

l,/hmofy rgrd

A Double t?ral

Figure 7.ll

Elasticstorewith a two-framememory.

9llp Bate Qbjactlves If the difference between an elastic store's input data rate and its output data rate is M,- the time between slips is

A7={ AR where N is the number of bits that get dropped or repeatedwhenever a slip occurs. Normally, a slip involves a full frame of data, in which casethe time between slips is determined as

47=-l^F

whereAF is the differencein framerates. As long asslipsarecontrolledsotheydo not disruptframing,ltheir only effectis aninfrequentrepetitionor deletionof theinformationwithin affectedTDM channels. The audibleeffectof slipson a digitizedvoice signalis an occasional"click." Only oneslip in 25 producesanaudibleclick in PCM voice[]. Voice signalscantherefore tolerateseveralslipsperminute[13]. *Typicatly,

clock offsets are specified in relative terms (e,g,, onepart in 106). A clock tlat is accurateto P parts per million (ppm) has a maximum offset of AR = ll . P/10". TWith single-frame slips, framing is disrupted in thc sense that framing pattem is shifted, Thus, the frame sequence has to be reacquired, but this process is simplified by the fact that the location of the framing bits and the integrity of the message channels are maintained.

350

N ETWORKSYNCHRONIZATION CONTHOLAND MANAGEMENT

A moresubtleandtroublesome aspectof slipsoccurswhena digitizedchannelcarriesvoicebanddata.High-speeddatamodemsfor the analogtelephonenetworkuse QAM modulationwith coherentdetectionin thereceiver.Sincethesemodemsareparticularlysensitiveto phaseshifts,theyareparticularlyvulnerableto slips.An 8-bitslip in a digitizedmodemsignalusinga carrierof 1800Hz causesaninstantaneous phase shift of 8l o.obviously,a phaseshift of this sizecausesa dataerror,but moreimportantly,it upsetsthecarrierrecoverycircuitryin thereceiverandcauses multipleerrors. A singleslip canupsettheoperationof somevoicebandmodemsfor severalseconds [ 1 4 ,1 5 ] . characterizations of theeffectsof slipson Group3 facsimileequipment[16]reveal thata singleslip cancausethelossof four to eightscanlineswithoutanerrorreport. Sometimes thelossof thelinesis not immediatelyevidentto thereconstructed image. Diagonallines,however,readilyrevealmissingverticalspace. Encryptedtraffrc(voiceor data)is moresusceptible to slipssincetheencryptiondecryptionprocessusuallyrelieson bit-synchronous scramblersand unscramblers. Whenthebit countis alteredby insertionor deletionof bits in a time slot,countersin the sourceanddestinationbecomeunsynchronized. At berrt,the decryptionprocess causeseveryslip to be audible.At worst,unintelligiblespeechor dataresultuntil the unscrambler is resynchronized. Anotherimportantaspectof encryptedcommunication is transmission of encryptionkeysor indexesto encryptionkeys.If synchronization is lost for somereason,it maybenecessary to resynchronize theencryptionkeys, therebycompromisingsecurity. Whena digitaltransmission link is beingusedto transmitdatadirectly,theeffect of a slip maynot be anymoresignificantthana singlechannelerror.Most datacommunicationsreceivingequipmentrequestsa completeretransmission of anyblock of datanot satisfyingcertainredundancy checks.Thusoneerroris asbadasmanyerrors or a completelossof data.Theeffectof the slip will be moresignificant,however,if the communications protocol[e.g.,the DDCMP of Digital Equipmentcorporation (compaq)lrelieson bytecountprocedures to delimitmessage blocks.Insefiionor deletion of databy the networkcausesthe receivecounterto becomeunsynchronized, andthenormalexchange of informationis disrupteduntil thelossof synchronization is recognized. Fromtheforegoingconsiderations for datatransmission, theslip rateobjectivefor theAT&T networkandadoptedby BellcoreandANSI for NorthAmericawasserar oneslip in 5 hr for anend-to-end connection[17-19]. Sinceslipscanoccurat multiple pointswithin a network,the objectivefor slipsat individualtrunk and switchinginterfaceswas$etat oneslip every20 hr. Example7.2. Determinethe relativeaccuracyrequirement$ of two independent clocksto maintaina mutualslip rateobjectiveof one slip in 20 hr. Assumea frame rateof I kHz asin PCM voicesisnals. solution. The slip rateobjectiveimpliesthattheframerateproducedby oneclock canbe different thanthe framerateproducedby the otherclock by no morethan

7.2 TIMINGINACCURACIES

*

=

I

r;

* * *

351

= 1.39x 10-5slipsper second

Sinceilrereare8000framesper second,therelativeaccuracyis determinedas 1.39xl0-5

ffi

n.-, = 1.7x "lo-e sliPs/frame

Hencetheclocksmustbe accurateto 1.7partsin lOe. only a maximumrelativeinaccuracy,the absoBecauseExample7.2 determines each lute inaccuracyof individualclockmustbe lessthan( I .7 x 10-e)/2,or 0.85parts in 10e. 7.2.2 Asynchronou$ Multlplexing In the precedingsectioncertainaspectsof networksynchronization werediscussed that implied the needfor clock synchronization to preventa lossof databy way of slips.In this sectiona procedure referredto as"pulsestuffing"is discussed thatavoids The termpulsestuffingcanbe somewhatmisboth slipsandclock synchronization. Ieadingsinceit impliesthatpulsesareinsertedinto the line codeto maketiming adjustments.Actually,pulsestuffinginvolvesonly the datastreamandis independent of theline codeor modulationsystemin use.Pulsestuffingis a termcommonlyused in North Americawhile the $ameconceptis refenedto as"justification"in Europe.Thebasicconceptofpulsestuffinginvolvestheuseof anoutputchannelwhoserate is purposelyhigherthantheinputrate.Thustheoutputchannelcancarryall inputdata plus somevariablenumberof "null," or "stuff," bits.The null bits arenot partof the incomingdata.Theyareinserted,in a prescribed manner,to padtheinputdataskeam to the higheroutputrate.Naturally,the extraneous null, or stuff,bits mustbe identifiablesothat"destuffing"canrecovertheoriginaldatastream. Thepracticeof pulsestuffingarosewhenthe initial digitalTDM hierarchy(Table 1.10)was defined.At this time therewere only isolateddigital transmissionlinks within the networkthat precludedsynchronizingthemto a commonclock.Whenit cametimeto combinelowerratetributaries(e.g.,DSls) into higherlevelsignals(e.g., DS2sor DS3s),themultiplexingprocedurenecessarily hadto accommodate tributaries operatingat $lightlydifferentrates.Thegenerictermfor combiningtheseunsynchronizedsignalsis asynchronous multiplexing.In this context,"asynchronous" refersto multiplexingofunsynchronized tributariesinto a higherlevel signal(using pulsestuffing).It doerrnot refer,in any way, to a meansof transmission. The higher link. level signalis alwayscarriedon a synchronous transmission As the amountof digital equipmentin the networkgrew andmore andmore of it becameinterconnected, boththemeansandthenecessityfor a differentform of mul*Justification

is the pnnting industry practice that aligns the right sirle oflines oftext by inserting a variable amount of spacewithin the line. As you will seein the following discussion,aligning individual tributaries to thc rate of a higher level multiplexer is conceptually the same process,

352

NETWoRKSyNcHRoNTzATToNcoNTRoLANDMANAGEMENT

tiplexing arose.The chosenapproachis refened to as synchronousmultiplexing (SONET)in North AmericaandSynchronous Digital Hierarchy(SDH) in therestof theworld.Theprinciplesof SONETandSDH aredescribed in conjunctionwith fiber systemsin thenextchapter. P ulse-Stuff i ng Concepts As a startingpoint for understanding theneedfor pulsestuffing,considerthe simple two-channel,bit-interleavedmultiplexerin Figure 7.12. As indicated,within any stringof even-numbered bits in the multiplexerouryut,thenumberof bits carriedin eachsubchannel is necessarilyidentical.Thus the ratesof the subchannels are also identical.If the two input clatastreamsarerunningat differentrates,the outputcanbe synchronized to oneofthe channelsbut not both.Thusslipswould necessarily occur in at leastoneof thetributaries. As a simplifiedexampleof pulsestuffing,Figure7.13 showsa two-channel, bitinterleaved formatasbeforebut with theadditionaldetailneededto allow adjustments of theinformationflow within eachsubchannel. As indicated,themultiplexedoutput is formattedinto 10-bitmasterframeswith 5 bits assignedto eachsubchannel. The frrst 3 bits in eachsubchannelof eachmasterfoamealwayscarrydatafrom therespective tributaries.Thefourthbit in eachsubchannel (C1andCj specifywhetherthelast bits (S1andSj carrydataor arestuffbits.WhenC1is a "1," a bit is stuffed;otherwise s 1carriestributarydata.Henceeachma$terframecancarry3 or 4 bitsfromeachtributary.Il on average!eachtributarysends3.5bits duringa ma$terframe,variationsof +l4Voin thetributaryclockratescanbe accommodated. An importantpoint to noticeaboutan asynchronous multiplexeris thattheoutput framestructureis unrelatedto theframeskuctureof thelower level inputs.As far as thehigherlevelmultiplexeris concemed, eachinprrtsignalis merelya serialbit stream with no particularstructureassumed. Framingbitsin thelowerlevelmultiplexsignals aretransmittedright alongwith the informationbits. After thehigherlevel signalis demultiplexedand the tributariesare unstuffed,framingof the lower level signals mustbe established for furtherdemultiplexing. Althoughpulsestuffingcanbeimplemented with a varietyof higherlevelframing formats,thegenerallymostdesirablefeaturesof a pulse-stuffingformatareidentified asfollows: 1. Theuseof fixed-lengthmasterframeswith eachchannelallowedto stuff or not to stuff a singlebit in themasterframe

ffi Figure 7.12 Two-channel multiplexer showing equal output data rates for each input,

7.2 TIMINGINAccURACIES 353 I Frame

E

D

c

B

A

example. Figure 7.13 Simplifiedpulse-stuff,rng

2. Redundant$tuffing specifications 3 . Noninformation bits distributed acrossa masterframe Timing offsetsaregenerallyquitesmall,soonly smalladjustments from a singleoccasionalstuff bit arerequired.Thuslargenumbersof tributarybits canbe combined into a masterframewith one specificbit positionidentifiedasthe S bit. Nominally, one-halfof the masterfrarnescontainthemaximumnumberof inforapproximately mationbitsNM,andtheotherhalf containNy - l informationbits. Thepurposeof a pulse-stuffingoperationis to preventa lossof datawhentwo inwith respectto eachother. terconnected digitaltransmission links areunsynchronized If singlebit errorscancausea stuffbit to be interpretedasinformation(or viceversa), the basicobjectiveis lost. Furthermore,noticethat an erroneousinterpretationof a stuff codecausesembedded, lower level multiplexsignalsto loseframing.For these reasons, theinterpretation of C bitsmustbeencodedredundantly. Underthe a$rtumption that channelerrorsarerandom,the probabilityof misinterpretinga stuff code is

Po:rfti,';,-',' - p)'

wherep = probability of a channelerror n = numberof correctable stuff codeerrors(2n + I bits in a stuff code) Informationbits shouldbe distributedacrossa masterframefor severalreasons. First,by sepamting thesebits asmuchaspossible,enorsin redundant$tuffingspeciIf thespecification bits aretoo ficationbits(C bits) aremorelikely to beindependent. closetogetherandbursterrorsareprevalent,theredundantencodingis of little use. Second,by distributingnoninformationbits, the irregularityof informationflow is minimized.Whenhigherlevel multiplexsignalsaredemultiplexed, a clock for each individuallower level signalneedsto be derivedfrom the irregularinformationrate in eachchannel.Generation of a suitablystableclocksynchronized to theinformation rate is simplif,redif informationburstsor gapsare minimized.Furthermore,elastic storesneededto smoothout the informationratesare smallerwhen the length of informafiongapsis minimized.

354

NETWORKSYNCHRONIZATION CONTROLAND MANAGEMENT

M12 Multiplexer An exampleof a higherlevel multiplexingformatis providedin Figure7.14.This is theformatusedfor 6.3l2-MbpsDsZ signalsin theNorthAmericandigitalhierarchy.* A DSz signalis derivedby bit interleavingfour DSI signalsandaddingtheappropriateoverheadbits. A DS2masterframeis 1176bitslong.Of thesethereare1148informationbits(287 perchannel),I I framingbits (I%, Mr, Fo,Fr), l2 stuffingconrrolbits (Cr, Cz,C:, C+), 4 S bits (Sr,Sz,Sr, S+),andan alarmbit X. Sincean S bit canbe a null bit or aninformationbit, eachchannelcan send287 or 288 bits in a masterframe.An S bit is designated asan informationbit if all threeof the corresponding C bits are0. The S bit is a null (stuff)bit if all threecorresponding c bits are 1.obviously,thisencoding procedureallowsfor singleerrorcorrectionin the stuffingcontrolbits. The first level of framingis established by the alternatingF6,F1,F0,. . . pattern. Noticethat exactly146bits separate theFs andFl bits.Anotherlevel of framingfor identifyingthe C andS bits is established by theMs andM1 bits.A fourthM bit (X) is not usedfor framingand thereforecan be usedas an alarmservicedigit. Similar framestructuresexist for otherhigherlevel digital signals.Figures7.15-7.18show thestructures for DS3,DS4,DSIC, andtheE2 second-level digital signalof theITU, respectively. Example7.3. Determine the minimum and maximum input channel rates accommodated by an Ml2 multiplexer.Also determinethe rateof DSI misframes causedby anerroneous interpretation of a stuffedbit. Assumethebit errorrateis 10-6. Solution. The maximuminformationrateper channelis determinedas 6.31?:?8s = 1.5458 Mbps lt76 Theminimuminformationrateper channelis determinedas 6.312x287 = 1'5404MbPs ffi6 Sincetherearethreepossiblecombinations of two errorsin theC bits,theprobability of misinterpreting an S bit is closelyapproximated by 3 x (10-6)2=3 x l0-r2. Thedurationof eachmasterframeis 117616.312 = 186Fsec.Thustherateof misframesper DSI sienalis 3 x 10-12

"#r":0'016x

persecond 10{ misframes

whichis equivalentto onemisframeevery2 years. 'DS2

signals are no longer transmitted as individual signals. They only exist as an intermediate level between 28 DSls and a DS3 sienal.

7.2 TIMINGINAcCURACIES 355 Zgt-Bit1;btilffi

+--'-

Erulf bitr

/tB

aE ilC-*r-rO

f !

C1

F l

E T

-E Figure 7.1C Frame format of DSZ digital signal. Stuffing occurs in channel i when the previous Ci bits = lll; X is an alarm bit that equals I for no-alarm condition. Framing is establishedby the F6F1F6. . . sequencewith 146 intervening bits.

Example7.3 demonstrates thatthetoleranceof a 1.544-MHzDSI clockis -3.572, to +1.796kHz. ThustherelativeaccuracybetweentheDSI andDS2 clocksmustbe I.79611544,or only I part in 860. This relativelylarge timing toleranceis much greaterthanwhatis requiredfor reasonable clockandline instabilities.Thetiming adjustmentcapabilitiesandunsymmetrictolerancerangewerechosenout of a desireto (l) minimizeDS2reframetimes,(2) providea line clockthatis a multipleof 8 kHz, and(3) minimizewaitingtimejitter [20]. C8O blt ilbfrfile

s

E t

ll

$E

s

Figure 7.15 Frameformatof DS3digitalsignal;P is evenparityoverall message bits in the previousmasterframe.Stuffingoccursin channeli whenthe previousCi bits = lll, The X (alarm)bits andthe P bits mustbe l l or 00 so the MoMrM6sequence canidentifythe endof the masterframe.

356

NETWoRKSYNCHRoNIZATIoNcoNTRoLANDMANAGEMENT Stuft bit

E

E tl o

gE o ti

{

Figure 7.16 Frame format of DS4: Ci bits = I l l implies stuff the eighth messagebit position for channel i following the last ci; Pr is even parity over the 192 previous odd-numbered messagebits; P0 is even parity over the 192 previous even-numberedmessagebits.

ElastlcSlore Size Requlrements A functional modelof anMl ? multiplexer is shownin Figure7.19.Associated with eachlower level (DSl) input is an elastic store to hold incoming data until it is transferred to the higher level (DS2) output. The elastic storesserve two purposes:to remove the arrival jitter of the incoming data and to hold data for the proper time slots. In addition to generatingframing, the control logic of the multiplexer monitors the

318 bir NUbfrEmG

rtro-i-c,-33-ro 5 ({

tl o

Mr

cr

q

Ml

6 q

E :E

Figure 7.17 Frame format of DSIC digital signal. Stuffrng occurs in channel i when the previous Ci bits = lll; X is an alarm bit that equals I for no-alarm condition. Framing is establishedby the FoFrFo. . . sequencewith 158 intervening bits.

7.2 TrMrNc rNAccuRActEs 357 21?Eitsblm Fflri

digffi$t

rign|l

200ilmgr

blt

irBtff tffir8{8 bltr

Figure 7.18 Frameformatof second-level digital signalof ITU-T (E2), Stuffingoccursin channeli whenthepreviousCi bits are lll.

$toragelevel (which servesa$ a phasecomparator)of eachelastic store and initiates a stuffing operationwheneverthe elastic storeis lessthan half full. Conversely,no stuffins occurs when the elastic store is more than half full.

Sincestuffingcanoccurat only certaintimesandonly at a certainmaximumrate, the elastictltoremustbe at leastas largeasthe peakjitter (peakphaseoffset)of the incomingsignal.As discussedat the beginningof this chapter,jitter accumulates alongthe entirelengthof a repeatered transmission link. Thus longerlinks require largerelasticstoresif slipsareto be prevented. As an exampleof therelationshipbetweenline length(numberof repeaters) and numberof bits of elasticstorageneededby anM12 multiplexer,referto Figure7;20. This figurewassubmittedby AT&T to the CCITT specialstudygrouponjitter [21]. jitter analysisof reference[3]. The analysisis an extensionof the systematic

DS-l inputs

DS-Zoutout Sdl6ctorcodtrol framing inrrtion dnd rtuffing

Framing gonaration Etuffing contfol

Storag6 l#el Eign|lr

Figure 7.19 Functionaldiagramof a Ml2 multiplexer

358

NETWoRKSyNcHRoNtzATtoNcoNTRoLANDMANAGEMENT

{ E

e

lio o o o E

fl. o E o

cl o E

e5

3.0

Jltur rlopq (kHr)

Figure 7.20 Maximumnumberof regenerativerepeatersasa functionof elasticstoresizeand jitter slope;,4 = numberof cellsin elasticstore. The abscissaof Figure 7.20 is the maximum-phaseslope produced by a clock recovery circuit in the presenceof a worst-caseshift in datapatterns(worst-casesystematic jitter). Since phase slope is nothing more than frequency offset, the required elastic store size can be determined as the maximum phaseslope times its maximum duration. Since the total phase slope is proportional to the number ofrepeaters, Figure 7.?0 displays the maximum number of repeaters per storage cell in an elastic store versu$ the jitter slope of an individual repeater when making worst-case timing transition. As an example, f,ust-generation Tl repeater$produce a worst-ca.seslope of 2.4 ktlz. Ml2 multiplexers allocate 5 bits of storageto input phasejitter (3 more bits are included for implementationeaseand waiting time jitter). From Figure 7 .20 it can be seenthat the ratio of N-* to A is 56, which implies that N*"* = 56 x 5 = 280 repeaters.*

*The

performance of long Tl lines is not as much of a concem as it once was becausemost long-distance DSl circuits ate now embedded in frber links, which have much greater repeater spacing,

rNAccuRAcrES359 7.2 TrMrNc

7.2.3 WaitlngTimeJitter to generatea Whendemultiplexinga higherlevel TDM datastream,it is necessary (or transtransfened Because are the subchannels clockfor eachderivedsubchannel. must Derivaclock be continuous. datastream,the derived mitted)asa synchronous data streamsof clocksis complicatedby the insertioninto TDM tion of subchannel overheadbits that creategapsin the bit anival times.Inegularityin the dataanival ratecausedby thesegapsis referredto aswaitingtimeiitter. Mostof theoverheadbits(e.g.,framingbits,paritybits, stuffingcontrolC bits)occur on a regularandpredictablebasis.The waitingtime jitter causedby thesegaps (sometimes canbeeliminatedeasilywith anelasticstore referredto asmappingjitter+) andanoutputclockderivedfrom theincomingline clock.For example,a singlePCM channelclock at 64 kbpscanbe derivedfrom a 1.544-MbpsTl line clockby multiplyingby I anddividingby 193.Mappingjitter asit occursin a EI signalmappedinto an E2 signalis depictedin Figure7.21. In this figure the phaseof the tributarydata mappedinto the higherlevel signalis shownrelativeto an unmapped(continuous) dataclock at the sameaveragerate.Notice that the phaseof the tributary falls behind the thereference duringperiodsof datagapsbut catchesup duringdatafieldsbecause averagetributary datarateduringthesefields is slightly higherthanthereference.Notice furtherthat the fill level of an elasticstoredoesnot get perfectlyreconciledat the on average, a nonintegral endof a singleE2 masterftame.This resultoccursbecause, master calried in frame. a numberof bits of a tributary are In contrastto mappingdataratevariations,waitingtime jitter producedby pulse waitstuffingis significantlymoredifficult to dealwith. Thedifficulty arisesbecause For this reason ing timesproducedby stuffedpulsesareirregularandunpredictable. TDM line mustbe derived the subchannel outputclocksderivedfrom a pulse-stuffed independently andonly fromtheaveragearrivalrafeof eachchannel'sdata-not from thehigherlevelTDM rate! usingjitter-removingelastic aregenerated Outputclocksfrom M12 demultiplexers stores,asshownin Figure7.22.If laryeelasticstoresandvery slowly adjustedoutput clocksareused,mostof thejitter canbe removed.Unfortunately,waitingtimejitter down to zerofrequencyso the jitter (e'g', wander)can hasfrequencycomponents to aslow a bandof neverbe eliminatedentirely.However,thejiner canbe conf,tned frequencies asdesiredby usinga largeenoughelasticstore.Figure7.23depictstime intervalerrorsresultingfrom thepulsestuffing-destuffingprocess.Thekey point to noticeis a full bit of offsetbetweenthesourcedataclockandthemappeddatacanoccur at anytime but an adjustmentcanonly be madewhenthenext stuff opportunity occurs-hencethetermwaitingtimejiner. Waitingtimejitter is basicallya functionof how oftenpulsesarestuffed,but it is If theinputclock alsodependent on theratioof actualstuffsto stuffingopportunities. "Some

references include mapping jitter to include waiting time jitter. Here, mapping jitter is used to represent repetitive data rate variations that occuf when both the tributary and the higher level transport signals are respectively at their pre,ciserates. Thus, waiting time jitter only occurs as a result of frequency adjustments,

360

NETWoHKSyNcHRoNtzATtoNcoNTRoLANDMANAcEMENT

Phffi of tn'hrrsry m cmpositc

rig*t

rcl8tivc b tqFt EigFrl iD Ulg

X: otrcrtad lr& Frfio I; lfrffilrim bit pilidm fn I rrfrltry sr Tffig a{iahflr (.||d) blr poridn ' Ewy_@lpilh AM c*df, Zoj hfffi|timtir p* ffirieyftr wDoilftb rd,fdwlirftd|thf, til !H hhEyin m S bit poiria strpaftrg o r!.d.

Ftgure7.21 Mapping Jitterof a CCITTEl in a CCITTE2sisnal.

is jitter free,theoutputjitter peakswhenone-halfof theopporhrnities areused.From thepoint of view of maximumtolerancefor clockoffsets,a stuffingratioof { is ideal. To reducethewaitingtimejitter, however,stuffingratiosof approximately {'areoften used.For a thoroughanalysisof waitingtimejitter, seereferencet2zl. As anexample of waitingtimejitter dependence on stuffing(iustification)ratios,seeFigure7.24 obtainedfrom reference[21]. The abscissa ofFigure 7.24represents the ratio of stuffs to opporhrnities while theordinateis jitter powerproducedby a singlepulse-stuffing process.Thejitter poweris expressed in decibelsrelativeto one slot squared(a slot is an olderterm for a unit interval,ul). curve A showsthe outputjitter pro-

Figure 7.22 Functionaldiagramof Ml2 demultiplexer/desynchronizer,

SYNCHRONIZATION361 7.3 NETWORK

AdJurtuent thrcshold

Tlms iileffsl

€rrorE

H lnput clock phase

waitingtimes. Figure 7.23 Time intervalerrorsproducedby adjustment (dBl 0dB = t slotr

-12 E o

E *rE .; 5

,A

'1 \

-h

7

JE

€ E h

B -20 o o

5

o

ol

0.4

A : Tributsryjitter - 0.10 tlot r.m.r. B I TributEry iitter - 0.25 rlot r.m.r

05

Juttification

0.6

0.7

0.s

cclTT.4773A 1.0 0.9

ratio

on justification ratio. Figure 7.24 Waiting timejiner dependence

ducedwhentheinputjitter is -20 dB (0.t UI rms).CurveB showsthe outputjitter whenthe inputjitter is -72 dE (0.25UI rms).Figure7.24 showsjitterproduced esby a singlemultiplexer.From measureddata 122],a good order-of-magnitude timateof the waiting time jitter accumulatedby N, tandemM12stuffing-destuffing operationsis

4=ffiruti'

(7.6)

7.3 NETWORKSYNCHRONIZATION in the precedingsection,whenevera digital transmissionlink is conAs discussed thetwo systemsby havingthe nectedto a digital switch,it is desirableto synchroniee

362

NETWoRKSYNCHHoNIZATIoNcoNTHoLANDMANAGEMENT

transmission link obtainits timingfrom theswitch.An obviousexceptionto thismode of operationoccurswhena digital kansmissionlink is connected to a digital switch onbothends.Generally,a hansmission link in anall-digitalnetworkderivesits timing from just oneof the switchesto whichit is connected. If theotherswitchis not synchronizedto thefirst in somemanner,anunsynchronized interfacenecessarily results. This sectionis concemedwith networksynchronization as a whole,not simply the synchronization of a singleinterface.Basically,networksynchronization involves synchronizingthe switchesof the network.The transmission links canthenbe synchronizedautomaticallyby deriving timing directly from a switchingnode. Therearetwo basicreasonsfor payingassiduous attentionto the timing requirementsof a digital network.First, the networkmustpreventuncontrolledslips that couldproducemisframes,inadvertent disconnects, andcrossconnects. It is generally very difficult or very expensiveto preventslipsaltogether. Thusa secondaspectof a networktiming plan requiresestablishing a manimumrateof controlledslipsaspafi of theend-to-endcircuitqualityobjectives. Synchronizing privatenetworksis sometimes difficult because thenetworktopologiesarenot designedwith networksynchronization in mind,andtheswitchingequipment (PBXs) me not designedto provide synchronizationto other nodes. Furthermore,the privatenetworksoften interfaceto multiple cardersin multiple locations.Determiningwhich signalsto synchronize to, particularlyon a dynamicbasis whena referencesignalbecomesunavailable, is exceptionallydifficult. Therearesix basicapproaches used,or considered for use,in synchronizing a digital network: 1. Plesiochronous 2. Networkwidepulse$tuffing 3. Mutual synchronization 4. Networkmaster 5. Master-slaveclockins 6. Packetization 7.3.1 Plesiochronous A pJesiochronous networkdoesnot synchronize the switchesbut merelyuseshighly accurateclocksat all swirchingnodessotheslip ratebetweenthenodesis acceptably low. This modeof operationis the simplestto implementsinceit avoidsdistributing timing throughoutthenetwork.A plesiochronou$ network,however,impliesthatthe smallerswitchingnodescarrythecostburdenof highlyaccurate andredundant timing source$' As a compromise, largenetworkscanbedividedinto subnetworks for timing purposes anduseplesiochronous operations for inter-subnetwork synchronization and someother,morecost-effective, meansof providingintra-subnetwork synchronization. As describedin section7.5, the public telephonenetworkin the united states usesplesiochronous synchronization at theupperlevels.

7.3 NETWORKSYNCI{HONIZATnN 363 Plesiochronoustiming is also usedto synchronizeinternationaldigital network interconnections.In recommendationG.811 [23], the ITU has establishedthe stability objectives for clocks of all intemational gateway digital switches.The stability objective of one part in l01l implies that slips betweeninternational gateway switcheswill occur at a rate of one per 70 days. (This assumesone clock is positive one part in 10lt and anotherclock is negative one pafr in 1011.)

7.3.2 NetworkwidePulseStuffing If all internallinks andswitchesof a networkweredesignedto run at nominalrates all voicesigslightlyhigherthanthenominalratesofthe voicedigitizationprocesses, nalscould propagatethroughthe networkwithout slipsby stuffingthe information to rateup to thelocalchannelrate.Noneof theclockswouldhaveto be synchronized be tolerated. At every intercould relatively clock accuracies coarse eachother,and facebetweensystemsrunningunderdifferentclocks,however,the individualchannelswouldhaveto be unstuffedfrom theincomingrateandstuffedup to thelocal or theTDM linksof thenetworkwouldprovideTDM channels outgoingrate.In essence, throughwhich userdataflows at lower andvariablerates,the differencesbeingabsorbedby internalpulsestuffing. operations of higherlevelmultiplexerswhereall chanln contrastto pulse-stuffing nelsin a lower level digital signalarestuffedasa group,switchingoperationsimply pulse-stuffing Theneedfor separate thateachchannelmustbe stuffedindependently. signals beingswitched is illustratedin Figure7.25,whichdepictstwo voice operations into a commonTDM outletlink. Obviously,thebit ratert3of bothoutputchannelsis identical.If the two channelsoriginatein portionsof the networkrunningunderdiffor mustbe madeseparately ferentclocksR1andR2,the pulse-stuffingadjustments eachchannel.The complexityof stuffingand unstuffingevery64-kbpschannelat everynetworkelementwould havebeenextremelyexpensivewhenthe digital network beganto takeshape.

Figure 7.25 Swirchingtwo channelswith different ratesonto a commonTDM output

364

NETWoFKSyNcHRoNtzATtoNcoNTHoLANDMANAGEMENT

Although the availability of low-cost logic could minimize the cost aspectof networkwide pulse stuffing, other problems would occur. First, the 64-kbps clocks for speechreconstruction would be required for each channel and would contain relatively large amountsof waiting time jitter. second, the network would no longer provide byte framing so the channel recovery processwould have to also include byte framing logic. Byte boundariesof PCM datacan be readily determinedfrom statistical data patterns in the bit positions (e.g., the polarity bit), but other applications for the channelsmay require explicit byte boundary identifiers.

7.3.3

Mutual Synchronizatlon

The two precedingsectionsdiscussmodes of operation for the network that do not involve synchronizationofindividual clocks. This sectionand the next two describenetwork timing plans that synchronizeeachindividual clock to a common frequency.The first method, mutual synchronization,establishesa coilrmon network clock frequency by having all nodesin the network exchangefrequency referencesas shown in Figure 'l .26.Each node averagesthe incoming referencesanclusesthis for its local and transmitted clock. After an initialization period, the network clock normally convergesto a single stable frequency. under certain conditions, however, the averaging process can become unstable[24]. The main attractivenessof a mutually synchronizednetwork is its ability to remain operationalin spite of a clock failure in any node. The main disadvantagesare the uncenainties of the exact averagefrequency and unknown transient behavior. Mutual synchronizationhas not been consideredfor the North American telephonenetwork. In Great Britain, however, a hierarchical timing structure was once consideredthat utilized mutual synchronizationwithin some portions of the network [25].

7.3.4 NetworkMaster Anothermethodof synchronizing thenetworkis shownin Figure7.27.with this

methoda singlemasterclockis tranrtmifted to all nodesenablingthemto lock ontoa commonfrequency.As indicated,all networknodesaredirectlyconnected to thenetwork master,implyingthe needfor a separate transmission networkdedicatedto the

SN

Flgure 7.26 Mutual synchronization: SN, switching node.

SYNCHHONIZATION365 7.3 NETWORK

node, synchronization: SN,switching Figule7.27 Networkmaster alsoimply thatalternatepaths Reliabilityconsiderations distributionof thereference. for the separate timing netbe providedto eachnode.Becauseofcost considerations with direct distribution,a networkmaster work andreliabilityproblemswith reference transmission to eachnodeis undesirable. networkis evolvingthroughthe use Somethingsimilarto a mastersynchronized of Global PositioningSystem(GPS) satellitesfor timing distributionto network networksaroundtheworld areusing of telecommunications nodes.Switchingsystem$ their switchingoffice clocks.As the GPSandothersatellitesystemsto synchroniee costsof theGPSreceiversandsuitablystableoscillatorsdropin cost,moreandmore networknodesate beingtimed from this master(highly accurate)source.Because systems CDMA digitalmobilesystemsalsolock to GPS,lowercostsynchronization stations CDMA in lieu base thatobiainGPStimingfrom the havealsobeendeveloped receive from of directly receivingGPSsignalsthat often requireoutsideantennasto multipleGPSsatellitessimultaneously. 7.3.5 Master-Slave Synchronizatlon The main drawbackto networkma$tersynchronizationasdescribedin the preceding facilitiesto everynode.Figure andreliabletransmission sectionis its needfor separate a masterreferenceby way of 7.28 showsa networkconfigurationthat disseminates A networkreferencefrequencyis transmittedto a few links themselves. themessage their clocksto selectedhigherlevel switchingnodes.After thesenodessynchronize the referenceand remove tansmission link-induced timing jitter, the referenceis passedonto lowerlevelswitchesby wayof existingdigitallinks.Thenextlowerlevel to anincominglink from thehigherlevelandpasstimswitches,in turn,synchronize ing on to anotherlevelof switchesby way of theiroutgoingdigitallinks.Theprocess "masof passingthe referencedownwardfrom onelevel to thenext is referredto as " ter-slavesynchronization. eitherdirectlyor indiSinceall switchingnodesin the networkaresynchronized rectly to the samereference,they all run at the samenominalclock rate.Thus slips of thedifferentpathsthroughwhichtimingis disshouldnotoccur.However,because short-termfrequencydifferencescanoccurbetweensomenodes'If these seminated,

366

NETWoRKSYNcHRoNIZATIoNcoNTHoLANDMANAGEMENT

Indlrcctly rynchronlrrd linkr

.---;_

Figure7.2t Master-slave synchronization. nodesaresynchronized indirectly,asshownin Figure7.28,infrequentslipsmightoccur. Furthermore,reliability considerations imply that backupclocksmust be provided in all switchesshouldthe clock distributionsystemfail. when this happens, slips becomemore likely, but only after relativelystablebackupclockshave had enoughtime to drift from the commonreferencefrequency. AT&T andthe united statesIndependent TelephoneAssociation(usITA) originally selectedmaster*slavesynchronization for the switcheddigital networkin the united states[26]. Thereferenceftequencywaslocaredin Hillsboro,Missouri,from whichselected No. 4 ESSswitchingcentersreceivedtheirtimingby wayof dedicated transmission facilities.synchronization of all otherswitchesoccunedby way of existingdigitaltransmission links.As discussed in Section7.4,theoriginalplanhasbeen changedto useplesiochronous synchronization at the highestlevel.Furthermore, as mentionedin the previou$section,the availabilityof Gps timing sourcesis leadingto moreandmorenodesatthehigherleveland,consequently, fewerandfewerslavenodes. 7.3.6 Packetlzation Thesynchronization discussions of thefive precedingsectionshaveassumed implicitly thata synchronous, circuit-switched networkwasbeingconsidered, sinceprevailing digitalvoicenetworksoperatein thatmanner.Forcompleteness, however,another form of networkmustbe mentioned-a packet-switched network. As discussedin chapter 10,packet-switched networksbreakup messages into identifiableblocks(packetsor cells)of data.In betweentheblocks.thehansmission linkscarryeitheridle codesor controlmessages. If all messages (controlanddata)are separated by a nominalintervalof idle transmission, elasticstorescan be resetin preparationfor the nextblock.As long aseachblock is limited in length,theelastic storescanabsorbclock differencesandavoidlossesof data.(In essence, slipsoccur in theidle codes.) 7.3.7 Network Timing Performance Measuremcnts After choosinga synchronization architecture for a network,it is necessary to be able to measure thequalityof thetiming signalswithin thenetworkto qualifyequipment,

367 7.s NETwoRK sYNcHRoNrzATloN links, and margins,isolatefaulty equipmentor transmission determineperformance usetheconceptof time possiblyevaluatedesignaltematives. ANSI andITU standards interval error (TIE) andmaximumtime interval error {MTIE) for thesepurposes'An is a time variance(TVAR). additionalmeasureof clockperformance Maxlmum Time lnterval Error A TIE is the differencein delay betweena timing signal and an ideal timing signal measured at the endof a particulartime period(i.e.,theerrorat the endof a time invarjationin TIE valuesthat occur terval).An MTIE is the maximumpeak-to-peak within a specifiedtime interval.Theseconceptsareillustratedin Figure7.29.Figure 7.29adisplaysthe TIE that occurswhenmeasuringa perfectlystablebut inaccurate thefrequencyofthe signalundertestdiffersfromthereference timing signal.Because frequencyby a constanfvalue (A/), the TIE is directly propol'tionalto the measureby countingclockcyclesin both signalsand mentintervalS.Theenor is determined expressingthe differenceas the time requiredby the test signal to catch up (or fall back)to thecurent referencecount.ThustheTIE producedby a constant-frequency offsetis

T''=T _ s(/+4fl - s(/) f

J

='H)

(7.7)

differencein clockcycles,A/is thefrequencyoffset,and whereAn is theaccumulated intervalin seconds. S is themeasurement monotonicallywith S,theMTIE andthe Because theTIE in Figure7.294increases TIE areidentical.In contrast,Figure7.29bdepictstheTIE producedby a timing signal with no long-termfrequencyoffsetbut someshott-terminstabilitiesfiitter andwander)aswouldoccurif thereferencewasrelayedthrougha network.As indicated,the difTIE variesasa functionof Sbut is bounded.TheMTIE is thelargestpeak-to-peak MTIE with interval S. The determinationof ferencein TIE valuesin a measurement no long-termfrequencyoffsetsis alsoshownin Figure7.30,whereperiodicsamples of time interval errorsaredepicted. Figure7.29cdisplaysthe moregeneralcasewherethe signalundertestcontains bottrinstabilityanda frequencyoffset.In ttriscasethechoiceof thevaluefor S is critical. of theoffset.If Sis too large,thefrequency If S is too small,jitter will maskthepresence needto Thusperiodicmeasurements theTIE andMTIE measurement$. offsetdominates thetiming imperfections. be recordedfor a time historyto fully characterize with theuseof a "perfect"reference in Figure7.29assume Thefirst threeexamples "imperfect" If an referenceis used,meaningful which to makethe mea$urements. areobtainedaslong astheTIE variationsaresignificantlygreaterthan measurements

NETWOHKSYNCHRONIZATION CONTROLAND MANAGEMENT

PErfEGt Source

StablE lnaccilratG Sourcl

Time + PErfoct Soutcs

Time +

(d)

-rF Timc Tim6 -r>

ss

Figure 7.29 TrE andMTIE for variousclock imperfections:(a) pureoffset;(D)purejitter; (c) offset andjitter; (d) TE differencemeasuremenr.

7.3 NETWORK SYNCHRONIZATION369

timeintervalerrorexample, Figure7,30 Maximum betweentwo therelativeperformance theimperfectionsof thereference.Sometimes to two different timing signalsis of interest.For example,a PBX with Tl connections betweenthetwo transplaces(or carriers)is moredependent on thetimingdifferences Figure7.29ddepictsmeasuringthe missionlinks thanon the absoluteperformances. aremeaningful, relativeperforrnance of two suchlinks.In thiscase,all measurements evenarbitrarilylow frequencywander,whichcancauseslipsbut is difficult or imposindicatetiming probsibleto measurein an absolutesense.If relativemeasurements maybe neededto isolatetheproblemsource. lemsexist,absolutemeasurements Time Varlance conveyany informationregardingthe freNeither TIE nor MTIE measurements jitter interval,f). (other thanthatconveyedby themeasurement quencycontentof the thejitter of thejitter requiresrepresenting A moregeneralstatisticalcharacterization magnitudeasa functionof frequency,or equivalently,asa functionof time between TVAR valuesaredeterThe time variance(TVAR) is sucha measure. TIE samples. betweenTIE samplesthat differences of second-order minedastheexpectedvaJiance by a time t, wheret variesfrom zero(or somefractionof a second)to areseparated in units period.TVAR valuesarecustomarilymeasured somemaximumobservation TVAR valTheformulafor calculating squared). of time squared(e.g.,nanoseconds = uesfromTIE samples;;(i 1, . . . ,19 is TVAR(I) = o:(r) :.tgltltt)tl N*3n+l fn-r

-ffi-

r . -+L1) r L v ' ' r 3n

I Fl

lE

'2

-zxv*k**r*) {"r*r"*o |

(7'8)

l lm where t = nto (to is the sampling interval) and the obserrvationperiod is Nto' The use of second-orderdifferencesremovesthe effect of a dc offset or of a linear phaseramp in the TVAR samples.Thus, there is no need to ulie a synchronizedrefer-

370

NETWoHKSYNCHHoNIZATIoNcoNTHoLANDMANAGEMENT

enceto deterrninethe TIE samplesbefore determining the TVAR values.However, if TVAR values for long observationperiods T are to be determined,a stablereference is neededto preclude referencewander from influencing the measurementvalues.

Time Devlatlon Timedeviation (TDEV)measurements aremerelythesquare rootof theTVARmeasurements. Thus,TDEVandTVARvalues havethesamerelationship astheclassical standarddeviation and variance of a probability distribution. TDEV is customarily measuredin nanoseconds.

7.4 U.S.NETWORK SYNCHRONIZATION Theoriginalplanfor synchronization of theu.s. networkformulated by theBellsystemandtheu.s. Independent Telephone Association wasto usemaster-slave syn-

chronization with a single master clock [26]. Due to the breakup of the network into multiple independentcompaniesand to difficulties in reliably distributing a highly accurate master,the synchronizationarchitecturefor the United Stateswas changedto a plesiochronous/hierarchicaldesignin the lare l9g0s [13, lg, I g]. Although the hierarchical design is still in use,it is gradually changing to a ..flatter" design by incorporating top-level functionality in more and more nodes.

7.4.1 SynchronizationRegions As shown in Figure 7.31, the public network is partitioned into synchronization regions that are intemally synchronizedwith a master-slave timing hierarchy that establishesdifferent levels of timing quality: sffatum I to stratum 4. stratum I clocks have the highest quality while stratum 4 clocks have the lowest. Timing for each region is establishedby a primary referencesource(pRS) at stratum l. stratum I clocks are free-running clocks with inaccuraciesno greater than one part in 10il. Some regions may have their own PRS while others may use a synchronization signal from anotherparty (e.g., an interexchangecarrier such as AT&T). For the most part, the synchronizationregions correspondto LATAs. Every region must have at least one stratum 2 clock, which is typically associatedwith an access tandem switch. Toll offices within LATAs may also have stratum 2 clocks. All toll switches within AT&T network contain stratum ? clocks [13]. Except when they use a coilrmon PRS, the synchronization regions are independently syncfuonized. Thus connectionsbetween the regions (using interexchangecarriers like AT&T) are typically plesiochronouslytimed. within a single region nodesare synchronizedin a master-slave hierarchy as indicated. The accuracyrequirementsof the four levels of stratum clocks are provided in Table 7 .2. These accuraciespertain only to situations in which the nodes are operating in a free-running mode. Normally the nodes are synchronizedto higher level clocks so the long-terrn accuracyis haceableback to the respectivePRS. In addition to listing free-running accuracies,Table 7 .2lists accuraciesthat must be met durine holdover

o 0

o .9

o _th

o

E(J l;6

F(, U O

E t

xa oqJ r4 f ; tc E6(JE

:i;t q i

* [

6 N

ti h

+ F

\ i

I

E

o

{ -

o

z (t t* I uo

371

572

NETWoRK syNcHHoNtzATtoN coNTRoLANDMANAGEMENT

TABLE7.2 StratumClockAccuracyRequlrementa Level Stratum1 Stratum2 Stratum3E Stratum3 Stratum4

Free-RunAccuracy +1 x 10-11 + 1. 6x 1 0 + t4.6 x 10+ f4.6 x 104 i32 x 104

24-HrHoldoverAccuracy Notapplicable t1 x10-loperday +1 x 10-s(day1) < 255slipsin day1 Notrequired

states.Theholdovermodefor stratum2 and3 clocksrequiresthesenodesto gradually transitionto thefree-runningstatewhentheylosetheirreferencesignals.*A stratum 2 clock,for example,losesthepreviouslyestablished frequencyat therateof I x I 0-r0 per day,implyingit would take 160daysto reachthe worst-case free-runningaccuracy'Stratum4 clocks,however,haveno holdoverrequirements sotheyenterthefreerunningstateimmediatelyuponlosingtheirreference(s). Networksynchronization requirements alsospecifyhow thevariousnodesrespond to degradationor completefailure of their references. Stratum2 and 3 clocksmust bridgeshortintemrptionsin thereferencewith minimumspecifiedtime-keepingerrors [l9]. Stratum3 clocks,whichtypicallyhavea primaryreferenceanda secondary reference,mustperforma very gradualswitchoverto the secondary referencewhen theprimaryfails. Abruptswitchovers, asoftenoccurin stratum4 nodes(pBXs),can causephasetransientsin the outputclock,which canin turn disruptsynchronization in all downstream devices(e.g.,within a privatenetwork). 7.4.2 Primary Reference Sources TheANSI MTIE specification for theaccuracyof a primaryreference sourceis shown in Figure7.32.Noticethatjitter (instabilitythatcanbeobservedin lessrhan0.05sec) is not specified.wander,asmeasured over500-sectime intervals.is limited to 3000 nsec.The asymptotefor long-termtiming errorscorresponds to inaccuracies of 1 x 10*1r. In additionto beingdesignedwith highly accurate(cesiumbeamor rubidium) clocks, all PRSsmust be continuouslyverified with universalcoordinatedtime (urc). such verificarioncanbe achievedby usinga UTC-basednavigationsysrem suchasLoran-c or GPS.MCI usesLoran-c andGps to directlysynchronize its pRS. AT&T usesGPSto monitor(verify)thelong-termaccuracyof eachpRSnodeestablishedin NorttrAmerica.tThetypicalaccuracyof thesenodesis muchbetterthanthe ANSI or ITU requirement [13]. *Holdou-t

operations are typically implemente
7.5 NETWoRKCONTROL 373

........ MTtE aTrT, 10-il

105

ô r d

E

E '* t

(to-?s + X - 3000nsec (proritionrlly)

t0?

l0l

t

l0-l

104

10? Ob86NationPBriod(sec)

Figure 7.32 PermissibleMTIE versusobservationperiodat the outputof a primary reference source.

7.4.3 1996 AT&T Synchronlzation Architecture ar' Beginningin 1996AT&T begana moveawayfrom a two-tieredsynchronization chitectureto a single-levelarchitecturewith a sfiatum2 clock in everyoffice synchroare: nizedto GPS[?7]. Significantfeaturesofthis architecture distributionnetworkhasbeeneliminated. l. The synchronization 2. Synchronizationis independentof the networktopologyso the traffic-carrying networkcanbe changedwithout affectingsynchronization. 3. Eachnodeis monitoredby two adjacentnodesand eachnodemonitorstwo verification' adjacentnodesfor performance 4. Performanceverificationinvolvesthe useof both MTIE andTDEV. 5. DSI timing signalsarederivedfrom SONETopticalsignals.

7.5 NETWORKCONTROL The synchronizationproceduresdescribedin theprecedingsectionrepresentmethods In this section andswitchingsystems. for controllingthetimingbetweentransmission

374

NETWoBK$yNcHBoNIzATtoNcoNTHoLANDMANAGEMENT

synchronizationis discussedin a more generalsense.Insteadofjust time or frequency control, the synchronizationconcept is extendedto higher level functions ofconnection control and network control as a whole. Fundamentalto the control conceptis the interactionbetweentwo processes(e.g.,the exchangeof information from one switching machine to anotherto set up or monitor a connection). A particularly useful meansof defining the interaction of two processesis a state transition diagram. The main purposeof the statetransition diagram is to abstractthe operational state$of a process and indicate what events causeffansitions from one $tateto another.when theseevents are messages(signaling tones or ccs messages) from another process,the statetransition diagram efTectivelydef,rneshow two communicating proce$$esinteract.

7.5.1 HlerarchicalSynchronizationprocesses Asanexample of amoregeneralized concept of synchronization, Figure7.33has been

included to demonstratethree distinct levels ofcontrol for a conventional telephone connection using digital transmission and multiplexing. The lowest level process shown in Figure 7.334 depicts nothing more than the clock synchronizationproces$ required to transmit and receive digital information. There are only two statesto the processin both the transmitterand the receiver.The purposeof clock synchronization is to causeffansitions between the two statesin the receiver to coincide with transitions in the clocking processof the transmitter. To accomplish this, a certain amount of transmissioncapacity is required in the form of line code transitions. Figure 7.33b depicts a higher level synchronizationprocessinvolving the framing of a time division multiplexer. Both processesrepresenta modulo-N counter, where N is the number of channelsin the TDM frame. The two processesare synchronized (framed) by utilizing someof the transmissioncapacityto sendframing patterns. once the receiver acquiresframing, the counter in the receiver counts in synchronismwith the counter in the transmitterso that individual TDM channelsare properly identified. Figure 7.33c provides state transition diagrams of a somewhatmore complicated but easily understoodprocess.The figure depicts the connectioncontrol ofa conventional telephonecall. The statetransition diagram ofthe first processrepresentsa subscriberplacing a call (going off-hook). The secondstatetransition diagram represen6 the sequenceof statesthe control element in the local switch goes through to set up the connection. As indicated, the processbegins by the originating subscribergoing off-hook and waiting for the dial tone. When the switch recognizesthe off-hook signal (current flow in the line), it connectsthe subscriberline to a digit receiver that returns a dial tone. The subscriberthen dials the addressofthe desiredtelephoneand entersanotherwait state' Upon receiving the last digit of the address,the switch control processesthe request.once the statusof the called party is determined,a busy tone or a ringback tone is returned to the originating subscriber.A busy tone prompts the subscriberto hang up (go on-hook) while a ringback signal causesthe subscriberto stay in the wait state until the called party answersor until the caller "times out" and abandonsthe call.

coNTRoL 375 7.s NETWoRK Rectlvtr

Traffimlttor

fl H fl} (a)

Suvltchcofitrol

Originatingrub,ecriber (c)

(a) timing process;(b) processes: Figure 7.33 Statetransitiondiagramof synchronizatiQn (c) telephoneconnectionprocess, tiaming proce.qs;

When the called party answers,both processesenter the connectedstateand communication between the end users begins.The end usersthen get involved in yet an"synchronization." Voice telephone users begin by exchanging other level of "connection" betweentheir thought processesto greetingsand identities to establisha communicateon a mutually understoodsubject.The messageexchangeprocessalso requires synchronizationso that only one person talks at a time. Hence various forms "turn the line around." Although being somewhatsubof control signalsareneededto tle in nature, thesecontrol signalsrepresenttransmissionoverheadin the same sense as control signals within the network. A talker may indicate his end of transmission

376

NETWoRKSyNcHRoNtzATtoNcoNTRoLANDMANAGEMENT

by asking a question,by inflections in the voice, by the messageitself, or more com* monly by a pause. Data communicationsequipment goesthrough the samebasic proceduresin order to establishconnection$and exchangeinformation. In this case,the proceduresaredefined more formatly and, consequently,are more restrictive. The formal rules of communication between data communications equipment are usually referred to as a -'protocol." Data communications protocols typically include a definition of certain control codes,code interpretations,messageframing, turn-aroundproceduresfor half_ duplex lines, error control, messagesequencing,fault control, and recovery. Automated fault control and recovery proceduresfor communications networks can becomequite involved and difficult to implement reliably. When individual voice circuits malfunction (e.g.,becomenoisy or disconnected),the recoveryproceduresare left to the users.They merely redial the connection and take up where they left off. However, Iarge trunk groups or switching systemsmust be designedfor higher levels of dependabitityand maintainability. The dependabilitycriterion ensuresthat failures or malfunctions rarely occur or that they are circumventedautomatically by protection switching' High levels of maintainability ensure that failures are repaired quickly when they occur. within switching systems,most of the instructions and memory words of the processorare dedicatedto hardwareand software performancemonitoring, recovery procedures,and maintenancediagnostics.

7.6 NETWORK MANAGEMENT In addition to controlling individual connectionsand equipment, a communications network must also manageits facilities on more macroscopiclevels. The basic goal of network managementis to maintain efficient operations during equipment failures and traffic overloads.The main considerationsare routing conffol and flow control.

7.6.1 RoutingControl Routing conffol refers to proceduresthat determine which paths in a network are assigned to particular connection$.If possible, connectionsshould use the most direct routes at the lowest levels of the network. The direct routes are obviously desirable becausethey use f'ewer network facilities and generally provide better ffansmission quality' However, economic considerations often limit the capacities of the direct routes so that alternate routes are needed to maintain suitably low blocking prob_ abilities between one switching machine and another. lf a t.unk group between two switching machinescontainsenough circuits to provide an acceptablylow blocking probability, a significant number of the circuits in the group are idle during averagetraffic loads.A more economicaldesign allocates a limited number ofheavily utilized trunks in the direct route and provides alternateroutes for overflow (alternately routed) traffic. In this manner the users are able to share larger portions of the network. chapter 12 presentsbasic examplesof how a network

377 7.6 NETWoHKMANAGEMENT can be engineeredto minimize the transmission facilities while providing a given grade of service (blocking probability). As discussedin Chapter l, the use of cenmalized control for the network, with common-channelsignaling, provides signiflcant efficiencies of operation in congestednetworks.

7.6.2 FlowGontrol ln the precedingsection,altematerouting is discussedas one aspectof managingtraffic in a communications netwolk. Routing algorithms are concernedonly with the utilization of paths or directions of travel within a network. Another requirement of netWork management is to control the amount of traffic in a network' Managing the rate at which traffic enters a network is referred to as flow control. A network without effective flow control proceduresbecomesvery inefficient or ceasesto function entirely when presentedwith excessivelyheavy traffic conditions' The generalizedperformanceof a large, uncontrolled network is shown in Figure 7 .34 as a function of the offered traffic. As indicated, when light traffic conditions exist, the network caffies all traffic requestspresented.As the load increases,however, some of the offered haffic is rejected becauseno appropriate circuits are available for particular connections;that is, blocking exists. As the input load increaseseven further. a network with no flow control eventually begins to carry less traffic than it does when Iighter loads are presented.If the offered load increaseseven more, the network may even ceaseto carry any tralfic at all. The reason that the volume of carried traffic decreaseswhen the offered traffic exceedssomecritical value is that partially completedrequeststie up network resources while trying to acquire other resourcestied up by other partially completed requests' Thus a form of dynamic deadlockoccurs.Prior to the developmentof centralizednetwork control, this situation would often arise on busy calling days (e.g., Mother's Day). In a network with distributed control all sourcesof traff,rcare serviced by successively seizing trunks to intermediate switching nodes until the destination is reached.If heavy traffic exists, request$emanating from all sidesof the network en-

,9

E

E I

Figure 7.34

Traffic caffied versus traffic offered for a network with no flow contol.

378

NETWoRKsYNcHRoNIzATIoNcoNTRoLANDMANAGEMENT

counter congestionsomewherenear the middle. At that time, the partially completed connections are tying up facilities neededby other requests.In the limit, when extremely heavy traffic exists, all network resourcesare held by partially completedrequestsand no complete connectionscan be established! Another exampleof the needfor flow control is automobilehaffic at a metropolitan intersection' Have you ever encounteredan intersection in which your lane of traffic was blocked by cross traffic backed up into the intersection?In essence.the driver blocking your direction of travel seizeda cornmon resource(the middle of the intersection) without being able to obtain the next resourcerequired (the other side ofthe intersection).Bumper-to-bumpertraffic in one direction can significantly degradethe throughput in other directions. With heavy traffic in all directions, total throughput can grind to a halt until the congestionis relieved from the periphery inward. The fundamentalprinciple demonstratedby theseexamplesis th;t efficient use of the common resourcesof a heavily loaded network requires some form of flow control. In the automobile example, smooth operation of an inter$ectiondepends on each driver looking ahead and not entering an intersection unless he can g*t ull tt * *uy across.A telecommunicationsnetwork must use the samebasic principle (it is hoped with more discipline). The control elementsat the periphery or ihe networt must be aware of the intemal status of the network and conhol the flow of haffic from its sources. More than one level of flow control is sometimesimplemented within a network. In a data communicationslink, some form of flow conhol is required to keep a source terminal from overloading the terminal at the other end of the link. The receiving terminal usesa reversechannel (sometimeswith a lower bit rate) to inform the source when to ceaseand when to begin transmissions.This level of flow control in a circuitswitched network involves the terminals themselvesand is of no concern to the network since the traffic flows within an establishedconnection. of more concem ro a circuit-switched network is how to control the flow of connectionrequestsinto the interior of the network. In setting up a long-distanceconnection, the first few circuits required should not be seizedunlessthere is a reasonablechancethat all ofthe circuits necessaryto complete the connection can be obtained. Partially completed circuits only degradethe network capacity by increasingcongestionwithout satisfying a serv_ ice request.Network flow control i$ greatly simplified with common-channel signal_ ing support for cenffalizednetwork control. The following paragraphsdescribe basic, uncentralizedflow control techniquesand how cenhalized conffol simplifies their implementation.

Trunk Directionalization Theoperation of tuunl< circuitscanbeclassified according to two differentwaysof controlling seizures for particular calls.Two-waytrunkscanbeseizedat eitherend.

one-way trunks, on the other hand, can be seizedonly at one end. (Notice that this has nothing to do with the direction of messagetransferon establishedconnections. which is always in both directions.) when one-way trunking is used,the trunk group is usually partitioned into one group that can be seizedat one end and one group that can be

379 7.6 NETwoRKMANAGEMFNT seizedat the other end. Two-way trunk gfoups areobviously more flexible in servicing fluctuating traffic patterns, but they are more difficult to control since the possibility of simultaneousseizures(called glare) at both ends must be resolved' A useful feature to incorporate into two-way trunks is the ability to directionalize them by marking them busy at one end and effectively creating one-way trunks. With this mechanism,a distant,overloadedswitching node can be relieved of additional incoming traffic while providing it sole accessto the trunk group for outgoing traffic' Thus the overloadednode relieves its congestionwhile inhibiting new anjvals. When the network as a whole experiences heavy haffic loads, trunk directionalization can be used to reduce the flow of connect requestsinto the interior of the network while establishing one-way trunks from the interior to the periphery. Thus connect requests that manage to get to the interior have a much better chance of obtaining all facilities required for the complete connection'

Cancellation ol Alternate Routing localizedoverloadsby transferringtraffic Alternaterouting of traffic accommodates however,altemateroutingis overloads, routes. networkwide During underutilized to undesirablefor two reasons.First, alternateroutesimply that a greaternumberof If thesesamefaciliandswitchingfacilitiesareneededfor a connection. transmission numberof links per the total connections, more direct two or to be a.rsigned tiescould more traffic. carry could network the reduced and be call could Second,the probability that an alternatelyroutedcall canacquireall the necessary is relativelylow. Trying to setup a connectionwith a largenumberof faresources ifthe probabilityofgettingall ofthe facilitiesis low (particularly cilitiesis undesirable alepending)' requests are fruitlesslytiedup whilelessdemanding facilities so,if some Code Blocking Codeblockingrbf"rt to artificially blockingcalls intendedfor specificdestination codes.If the calls areblockedat originatingend offltcesbeforethey acquireinternal arerelievedof incomingtraffic withouttying up networkfacilities,the destinations facilitiesthatmaybe neededfor outgoingrequestsfrom the specifiedareas. The methodof flow conhol is particularlyusefulin times of naturaldisasters, which typically stimulatelarge numbersof calls both into and out of the areaof the disaster.[n theseeventsa networkcontrolcentercaninitiatecodeblockingfor all, or to the of thecallsinto thearea.Theprincipleof givingpreference a largepercentage, there unless seized are First,no networkfacilities outgoingcallsservestwo pufposes. out of into or runks It is the is a reasonablechanceof obtainingall facilities necessary. fiunks one of these Once thatarethefocalpoint of networkcongestion. thedisasterarrea Second,codeblockingis is seized,therestof theconnectioncanprobablybeestablished. outgoingcallsareprobablymoreimpoftantthanincomingcalls. usefulbecause Centralized Connectlon Eontrol to eliminatesei previouslyaredesigned described All of theflow controlprocedures probability. completion has a low connection the desired if resoulces of common zures

380

NETwoRKsyNcHHoNtzATtoNcoNTRoLANDMANAGEMENT

Becauseof the distributed natureof network control implied by theseoperations, these control proceduresare necessarilyprobabilistic. To maintain a certainamount of network efficiency, the network is purposely operatedat less than maximum capacity. A more desirablemode of operation,from a throughputpoint of view, is to allocate network facilities from a single centralized conhol node. Since this central node has accessto the statusof all network resources,no facilities are assignedto a pafiicular requestunlessall facilities neededfor the completeconnectionare available. Network hansmission links are assignedin a manner that is analogousto the assignment of in_ temal links of common control switches. Complete centralizeclconfrol of a network as large as the public telephonenetwork is obviously infeasible from the point of view of maintaining statusoi all interstage links within end office switchesand from the point of view ofiurvivability of the ne1work when the control node tails. However, many aspectsof centralizedcontrol have beenimplementedin North America and around the world with common-channel sig_ naling (CCS)' For example, INWATS call requestsare routed ro a cenrral node that determinesif the destination is busy or not. If the destinationis busy, the originating end office is instructedto return the busy tone without any of the int#al mansmission links ever being seized'This mode of operationis particularly useful for 800 numbers oNwATS) that occasionally experiencevery heavy traffic flow becauseof national television announcemenrs. without ccs, the previous mode of operation was to return a busy tone ail the way through the network from the place at which the busy circuit or subscriber is located. Thus the path through the network was tied up during the time the busy tone was being returned' CCS allows the originating office to return the busy tone so internal network facilities can be releasedand reassignedimmediately upon detecting the busy condi_ tion.

REFERENCES I 2 3 4 5 6 7

8

M. Decinaand U- deJulio,"InternationalActivitieson NetworkSynchronization for Digital communication,"IEEE Intematiotutlcommunications conference,lg7g, J. R, Pierce,"synchronizingDigiral Networks,"Bel/ systemTechntualrournal, Mar. 1969,pp.615-636. c. J' Byrne,B. J. Karafin,andD. B, Robinson,"systematicJitterin a chain of Digitar Repeaters," Bell System TechnicalJournal,Nov. 1963,pp.2679_2714. "Digitally DejirteringLaserDiscplayers,"IEEESpectrum,Feb. 1990,p. 14. F. M. Gardner,PhaselockTechniques,2nd york, ed.,Wiley, New 1979. E. D. sunde,"self-Timing Regenerative Repeaterc,,, Bell systemTechnicalrournal, July 1957,pp.891-938. D. L. Duttweiler,"The Jitterperformanceof phase-Locked t oopsExtractrngTiming from BasebandData waveforms,"BerI systemTechnicalJournal, Jan. 1976,pp, 37-58. "carrier-to-customer Installation-DSlMetallic Interface," ANSI Tl. 403_l9gg, AmericanNationalStandards Institute,New york, 19g9.

REFERENCES381 "synchronizationFailuresin a chain of PLL 9 H. Meyer, L. Popken,andH. R. Mueller, May 1986,pp' 436-445' on Communicatiazs, Synchronizers ," IEEE Transadions Artech House, 1 0 P. R. TrischittaandE. L. Varma,finer in Digital TransmissionSystems, Norwood,MA, 1989. "The Accumulationof Pattern-Dependent Jitter for e l l P, R. Trischitta and P. Sannuti, June Communications, on Transattions IEEE Regenerators," Fiber Optic Chain of pp.76r-765. 1988,

"No' 4ESS: 12 J, F. Boyle, J. R. Colton,C. L' Dammann,B' J' Karafin,and H' Mann, Interfacesand Toll Terminal Equipment,"Bell System Transmission/Switching TechnicalJountal,Sept.1977, pp' 1057-1097C' D' Near' P. Greendyk,A' M. Montenegro, r3 J.E. Abate,E. W. Butterline,R. A. Carley, "AT&T's of Synchronieation to the Approach New S.H. Richman,andG. P.Zampetti, pp. 1989' April Magazine, IEEE Communications Networks," Telecommunication 35-45. "The Effects of Slips on Data Modems," IEEE H. Drucker and A. C. Morton, 1987,pp. 12'4.I-12'4'3' on Communitations, InternationalConference ..Identification of Digital Impairmentsin a voicebandchannel,"IEEE t 5 J. F. Ingle, 1989,pp. I 2'3'I - I 2'3'5' InternationalConference on Communicarions, "The Effect of slips on FacsimileTransmission," IEEE l 6 J. E. Abateand H. Drucker, 1988,pp' 32'3.1-32'3'4' e on Communicalions, Intemational Conferenc "The switchedDigital J. c. Lawson,andw. L, Ross, t7 J. E. Abate,L. H. Brandenburg, Joumal,Sept.1977,pp. 1297-1320' NetworkPlan,"Bell SystemTechnital ..Digital synchronizationNetwork Plan," Bellcore TA-M1,000436, Issue 1, Bell l8 Morristown,NJ' Nov' 1986. Research, Communications ,.synchronization Interfacestandardsfor Digital Networks,"ANSI Tl. 101*1987' 19 Institute,New York, 1987. AmericanNationalStandards Technical staff, Bell Telephone Laboratories, Transmission systems for NC' 197I' Winston-Salem' Bell TelephoneLaboratories, Communications, ,,Impact of Jitter on the second order Digital Multiplex at 6312 kbit/s," AT&T zl Submittalto CCITT studygrouponjitter, GreenBook,Vol. 3, pp. 861*869' Joumal,Jan.1972,pp. 22 D. L. Dutrweiler,..waitingTime Jitter,"Bell systemTechnical 165*207. 23 "PlesiochronousOperationof InternationalDigital Links," CCftT Recommendation G.LLI, OrangeEook,Geneva,Switzerland'1976. "Performance Clocks," Bell of a Systemof Mutually Synchronized 24 J. P. Moreland, System Tethnical Joumal, Sept.197l, pp. 2449-MM' "synchronization of theDigitalNetworkin theUnited ZS p, A. MitchellandR. A. Bourler, on communications, 1979, conferences Intemational Kingdom," IEEE pp.l1.2.1-11.2.4. in a SwitchedDigital 26 C. A. Cooper, "synchronizationfor Telecommunications pp' 1028-1033' I9?9, July on Communications, Network."IEEE Transactions "AT&T SynchroniaadonNetwork Architectureand Operations," 2':. C. Olszewski, hesentation to 1999 NIST-TD{I Workshop on Synchronization in Systems,Boulder,CO, March9-ll' 1999' Telecommunications

382

NETWoRKsYNcHRoNIzATIoNcoNTHoLANDMANAGEMENT

PROBLEIT,IS 7.1 Determinethesizeof anelasticstoreneededto accommodate a velocityshift of *10fi) km/hrthatlastsfor l0 secif thedatarateis l0 Mbps.(Thespeedof light is3xl08m/sec.) 7.2 How manybitsareneededin anelasticstoredesignedto interfacetheEl digital signalof ITU-T to a digital switch?what is rhemaximumslip rateif the line clock andswitchclock differ by +50 to -50 ppm (themaximumrecommended offsets)? 7.3 what is the maximum(ggvoprobability)phaseoffset(in signalintervals)pro_ ducedby ajitter powerof + I 0 dB relativeto I radz? 7.4 Determinethe rateat which DSr signalsin a DS2 multiplexloseframing be_ causestuff codesof a DS2 signalareincorrectlyinterpreted. Assumethechan_ nel BER is l0-3.AssumetheBER is 10-6 7'5 What wouldthe rateof incorrectDSZstuff codeoccurrences be if a 5-bit stuff codewereusedinsteadof the3-bit stuffcode?AssumethechannelBER is l0-3. 7.6 A digitaltransmission link is to beusedto hansmitblocks(packets) of datawithout slips.If thetransmission link is autonomously timedwith,"spe"t to there_ ceivingterminal,whatis themaximumallowableblocklengthif theclocksvary by +50ppmeachandan elasticstoreof 16bits existsin thereceivingterminal? Assumetheelasticstoreis initializedto half full betweeneachblock. 7.7 Assumethesystematic jitter from a singlerepeaterproducesa symmeffic,worst casephaseslopeof 300 rad/secfor I msec.what is the peak-to-peak jitter in decibelsrelativeto a unit intervalat theendof a line with 200 suchrepeaters? 7.8 RepeatExample7.j for DSZsignalsin a DS3. 7.9 A jiuer powerof 20 dB relativeto I radzis observedat the receivingend of a digitalmicrowavelink. Whatis theprobabilitythatthephaseoffsetwill exceed 14.0 symbolintervals? 7.10 Determinetheamountof phaseshift injectedinto a 2400-Hzcarier signal by a slip of onePCM sample. 7.tl Determine the TIE and the MTIE at the end of a 10-secinterval produced

by a DS3 signal that has a constantoffset of one part in 106. 7.12 RepeatProblem 7. I I but assumea l -MHz jitter component is added that has a peak-to-peakamplitude of 8 uls. Assume the starting phaseof thejitter component is 0o.

7.13 Comparethe slip rateof $tratum3E with stratum3 clocksin holdover conditions. 7.14 what is theoscillatoraccuracy(in ppm)impliedby theholdoverrequiremenr of a stratum3 clock?

OPTIC FIBER SYSTEMS TRANSMISSION The evolutionof the commonequipmentportionsof the public telephonenetwork from analogtechnologyto digitaltechnologybeganandendedwith digitaltransmissionsystems:Tl systemsfor short-haultrunksandfiber optic systemsfor long-haul kunks.If fiber optic technologyhadnot emerged,extensivedigitizationof thelongdistancenetworkwould not haveoccurredasrapidly.High-bandwidthcoaxialcable andanasystemsaretoo expensiveto universallyreplaceradiosystems, transmission log radiosare more efficient than digital radiosin termsof voice circuitsper previouslyavailablebandwidth.Eventhoughhigh-densitydigitalmodulationtechniques and sophisticatedvoice compressionalgorithmscanmakeup for the bandwidthinefof digital radiosoveranalograficienciesof digitizedvoice,anyultimateadvantage of the analogradio diosis not nearlygreatenoughto walTantwholesalereplacement on theotherhand,providesuchdramaticsavsystems, backbone.Fibertransmission ings in equipmentand operationalcoststhat wholesaledeploymentof high-density justified' routeswith fiber systemswaseconomically "photophone," patAs a noteof historicalinterest,Figure8.1showsa diagramof a entedby AlexanderGrahamBeIl in 1880.BeIl developedseveraltechniquesfor of a selenium modulatinga light source(thesun)so asto directlyvary theresistance cell detectorandtherebycreateananalogelectricalcurrentfor the speaker-Themodulation mechanismshownin Figure8.1 is a mirror thatis vibratedby acousticenergy to deflectmoreor lesslight to thereceiver.It wasalmostexacfly100yearslaterthat practicalmediumandsuitablesourcesmadeopticaltransmission a suitableuansmission The par"ticularcharacteristicsof optical fibers that makethem so useful for fransarelow loss,high bandwidth,smallphysicalcrosssection,EMI immissionsy$tems security. and munity, Fiber Attenuation As a resultof the inventionof the laserin 1960,materialsscientistsbegansearching sysmediathatcouldbeusefullyappliedasa communication for opticaltransmission years in 1970 later l0 announced fiber was practical optical tem.Thefirst instanceof a 383

384

FIBEHOPTICTRANSMISSIONSYSTEMS

l. d. DELr.

Apprntur for Bignrlhg rld 0onmulortll3,orlhl pLotopLoar. t{o. ?3$rl9g.

Frlrltrd

D.c_ r_ rlrn

j1 l

:

Figure 8.1 AlexanderGrahamBell's photophone. [l]. This announcementdescribeda silica-basedfiber with .,only" 20 dB/km attenu_ ation. In just a little less than l0 more years!commercially viable optical fibers with 0.2 dB/km attenuationhad been developed [2]. such remarkably low attenuation at_ tracted immediate attention becauseit meant intercity transmission links could be traversed with very few repeaters,implying a dramatic savings in equipment and maintenance. As shown in Figure 8.2, the attenuationof an optical fiber is very dependenton the wavelength of the light signal in use. Two wavelengthsthat experienceparticularly low attenuationin contemporaryfibers are 1300 and 1550 nm. Rlpresentative attenu_ ations at these wavelengths are 0.35 and 0.2 dB/km, respectively. with an easily achievablenet loss of 20 dB the latter attenuation value allows 100 km between re_ peatersor amplifiers. Because0.2 dB/km is close to the minimum theoretical attenuation at 1550 nm, additional reductions in attenuation cannot be realized without going to higher wavelengths.If fibers and associatedelectronicscan be developedto operateat higher wavelengths,much lower levels of attenuationmay be achieved in the future. The main application of such systemswould be in submarinecables where the avoidanceof repeatersis most desirable.

SYSTEMS 385 FIBEROPTICTRANSMIS$ION

to

to

5

5

E 2

I

{ i fA

E . 5

.5

.s

8 . 2 5

6.r

.t

d

8oo

looo Figure t.2

tt*

***l#r,

(n.l16ffi

18oo

Fiber attenuationver$uswavelength.

Bandwidth

media,thebandwidthof an optical transmission Whencomparedto elechomagnetic fiber is mind boggling:a singleopticalfiber operatingat 1300or 1550nm of wavelengthhasa potentialbandwidthof 20 THz (20 x totz Hz), which is enoughfor 312 sy$temsare million 64-kbpschannels.Bandwidthlimits of fiber optic transmission mostly determinedby the electro-opticdrivers and receiversor the electronicinter* facesto thesedevices.As describedin Section8.I . l, multimodefibershaveaninherbandwidthlimitation,but this particularlimitationis avoided ent distance-dependent in single-modefibers. Single-modefiber systemsalso have a distance-dependent bandwidthlimitation.but this limitationis asmucha limitationof theopticalsources asit is a limitation of the fiber. Small Physical Croee Sectlon One of the most beneficialfeaturesof Tl systemsis their ability to relieve overto two-wireanacrowdedcableductswith a 12+o-I savingsof copperpairscompared log transmission.Fiber systemsprovide the sameadvantagewith respectto Tl thesame only to a greaterdegree.A 25-paircoppercablehasapproximately $ystems, line-powered crosssectionasa fiber cablewith 24 fibersand somecopperpairsfor or amplifiers.The latter can easilycarry 100,000voice circuits.Furtherrepeaters for morecapacityby merelyinstalling beupgraded cansometimes more,fibersy$tems higher speedelecrronicsor wavelengthdivision multiplexing(WDM). Thus, after coppercableis replacedby a fiber cable,duct congestionis unlikely to reappear unlessextensiveuse of broadbandservicesgreatlyexpandsthe needfor fiber facilities. Electromagnetic lnterference lmmunity to electricity,the f,rberitself cannotpick up Becauseglasshasa very high resistance interferingsignalsor noiseor propagateharmful or damagingtransientsto personnel andequipmentat the endpoints.The immunity to interferencealsomeanscrosstalkis

386

FIBERoPTIcTRANSMIssIoN SY$TEMS

nota problembetweenmultiple-fibercables.In systems utilizingline-powered repeaters,however,someamountof copperis includedin the cableto carry power,implying that the immunity, particularlywith respectto hazardousvoltages,is compromised. Theimmunityof fibersto externalinterference suchasnoiseandcrosstalkimplies thereis no crosstalk-induced limit from high poweredtransmitters or moresensitive receivers'Receiversensitivityis ultimatelylimited by internalnoisein thephotodetectors,and transmitpowerhastechnologylimits due to spectrumspreadingin the sourcesandnonlinearities in thefiber.Until theselimits arereached, however,theabsenceof crosstalkbetweenfibersallowsthecapacityor transmission distanceof a fiber systemto be increasedby merely upgradingthe electronics,not the fiber. Althoughcrosstalkfrom one fiber to anotheris nonexistent, crosstalkcanarisebetweenseparate channelswithin WDM channelsof a singlefiber. Security Becauseopticalfibersradiateno energy,noninvasiveeavesdropping ofthe signalis impossible.Furthermore,invasivetapsaremorediffrcult to implementthanarewirelinetaps,whichmerelyrequirebridgingacrosstheconductors with a sufficientlyhigh impedanceto removea usablebut unnoticeable amountof signalenergy.A similar processis possiblewith opticalfibers,but it requiresbendingthefiber a veryprecise amountto allow a smallamountof energyto escapeandbe amplifiedby a tap.Not only doesthis processallow passivelytappinga fiber,but alsosignalscanbeinjected into thefiber throughsucha bend.This techniquehasbeenusedasa meansof locally testingthe effectiveness of a fiber spliceandhasbeenconsidered asa meansof implementingdistributed,passivetapsfor fiber distributionto thehome [3]. 8.1 FIBEROPTICTRAN$MISSIONSYSTEMELEMENTS As shownin Figure8.3,thebasicelementsof a fiber optictransmission systemarethe electrical-to-optical transducer in thetransmittingterminal,theopticalfiter itself,the optical-to-electrical transducer in thereceiver,andsignalprocessing circuitryfor am_ plifitcation,clockrecovery,anddatadetection.Regenerative repeaters requireopticalto-elechicalconversionfor the signalprocessingfunctionsand conversionback to opticalfor transmission. Direct optical amplificationwith erbium-dopedfiber amplifiers (EDFAs) areusedin lieu of repeaters,particularlyfor wDlftransmission links.

Flgure 8.3 Fiberoptictransmission systemelements.

ELEMENTS SYSTEM 8.1 FIBEHOPTICTRANSMISSION

387

Jsket

Figure 8.4 Opticalfiber construction. The systemdepicted in Figure 8.3 is not necessarilya digital one. Although all of the fiber applications for telephony have utilized digital signals,analog hansmission is possible.Analog FM modulation of optical signalshas been successfullyapplied to feeder applicationsof CATV systems[4]. Furthermore, sy$temsthat carry 80 channels of AM video on fiber to residenceshave been developed, a capability that allows analog television set$to receive fiber-based cable TV without a digital video decoder.

8.1.1 OpticalFiberFundamentals of anopticalfiberis shownin Figure8.4.Boththecoreand Thebasicconstruction the cladding aretran$parentto the desiredlight signal but the cladding is designedwith a lower index of refraction, which causesmost light waves in the core to be reflected back into the core. As shown in Figure 8.5, whether or not a ray is reflected back into the core is dependenton the angle at which it strikes the sore-cladding boundary. If the angle is too sharp, the ray is not reflected but passesthrough the cladding and is absorbed by the opaque, protective jacket. The sharp angles can occur at two places; (l) near the sourcewhere all of the source'$output is not focusedinto the center of the fiber and (2) at bends, splices, or other imperfections in the fiber.

tllultimode Flbere to thereto deliverasmuchsignalenergyaspossible Althoughit is usuallydesirable ceiver. waves that reflect back and forth before reaching the receiver may be undesirable if they experiencetoo much delay with respectto the primary ray traveling down ttre center of the frber. An optical fiber that allows rays to arrive at the receiver via multiple paths is referred to as a multimode fiber. Multimode fibers have core diameters

Jack6t Chdding Core

Figure 8;5

Fiber with multimode propagation.

388

FTBER oplc rRANSMtsstoN sysrEMs

that are large comparedto the wavelengthof the signal.A typical multimode fiber will have a 50-pm core diameter and a 125-pm cladding diameter (such a fiber is designated as a 5oll25 fiber). The delayedrays causepulse spreadingreferred to as multimode dispersion.The significanceof the spreadingdependson the width of the pulse or, conversely,the datarate being used.Multimode dispersionthereforecreatesan inherent operationallimit defined as a bandwidth-di$tance product (BDp). The BDp of a typical step-indexmultimode fiber is 13 Mbps-km [5]. The previously describedmultimode fiber is referred to as a step-index fiber becausethe index of refraction in the core is constantwith a stepchangein the index occurring at the core-cladding boundary. Multimode dispersion can be significantly reducedby varying the index of refraction within the core so that a high value occurs in the center and a low value occurs at the edge.Becausethe speedof propagationof light is higher in lower indices of refraction, rays that reflect back and forth within the core ffavel at an average speed that is greater than a primary ray that remains entirely within the center. Thus, if the index of refraction is carefully graded wirhin the core, all rays can be made to arrive at the receiver with the same amount of delay. such a fiber is refened to as a graded-indexfiber. A representativeBDp of a graded-index fiber is 2 Gbps-km [6]. Notice that this is an improvement of more than two orders of magnitude over typical step-indexmultimode fibers. Example 8.1. Derermine the loss limir and the multimode dispersion limit of a graded-indexmultimode fiber systemoperatingat 820 nm and providing a bandwidth of 90 Mbps (enough ro carry two DS3 signals). Assume that the difference between the available output power from the source and the input power required by the receiver for an acceptablemaximum error rate is 42 dB. solution. From Figure 8.2, the attenuationof a multimode fiber operatingat g20 nm is approximately 3 dB/km. Thus,

.. 42 Losshmit=8.0=t+ttt't Using2 Gbps-kmasa typicalBDP of a graded-index multimodefiber,themultimode dispersiondistancelimit is determinedas

Dispersion limit =

#

= ZZ.2km

Theresultsof Example8.I arerepresentative capabilitiesof first-generation fiber systemssuchastheFT3c systemof AT&T [7]. In actualpractice,repeaterspacing wouldbe lessthanthe l4-km losslimit to allow marginfor, for example,componenttolerances, splicing,andaging.The first FT3c system,which beganservice betweenNew York city andwashington,DC in February19g3,hadrepeaterspac-

8.1 FIBEROPTICTRANSMISSIONSYSTEMELEMENTS

ing of 7 km-the system.

389

location of repeatersin a coaxial cable system replaced by the fiber

Slngle-Mode Fibars werebeingbtoughtinto servicein thelongfiber systems As thefirst-generation distancenetwork,opticaltechnologyhadalreadyadvancedto the point that deploysystemswasunderway.Onekey technologyof thesecond mentof second-generation fibers(SMFs)thatprovidetwo distinctadvangeneration wastheuseof single-mode tages.First,SMFshavesmallerdiametercores(8 pm typically)thatrestrictpropagation to a fundamentalmodeandthereforeeliminatemultimodedispersion.Second, by about whichreducestheattenuation SMFshavelessinternal(Rayleigh)scattering, nm is about2 at 820 operating an SMF the of 50Vo16,81.For example, attenuation fiber. dB/km,asopposedto 3 dB/kmfor a multimode involvedtheuseof longer $ystems of second-generation Anotherkey development significantlyless (1300nm), which,asshownin Figure8.2,experience wavelengths systems.Use usedin first-generation thanthe 800-900-nmwavelengths attenuation requirednewtechnologyfor sourcesandreceivers,asdescribed ofthesewavelengths 8.1.2and8.1.3. in Sections

ChromaticDlsperaion

The combinationof eliminatingmultimodedispersionandusinga wavelengthwith muchlower signalattenuationrevealedanotherffansmissionlimitation referredto as chromaticdispersion.Chromaticdispersionariseswhen a photonicsignalcontains more than one wavelengthand the individual wavelengthspropagateat different Thus,chromaticdispersionis thephotonicequivalentofphasedistortion(also speeds. Dispersionlevels (wireline)propagation. calleddelaydistortion)in electromagnetic perkilometer' per nanometer picoseconds of arequantifiedby a dispersioncoefficient psec/nm km. is 16 nanometers A representative valuefor the SMF at 1550 Theeffectof chromaticdispersionis minimizedby usingoneor moretechniques. First, operationcan be centereclabouta wavelengththat exhibits a small amountof chromaticdispersion.Typical silica fibers,for example,produceapproximately15 at 13fi) nm thanat 1550nm.Second,anopticalsource timeslesschromaticdispersion shouldbe chosenthat is as $pectrallypureas possible(e.g.,hasa naffow spreadof wavelengths). Third, nalrow retum-to-zero(RZ) pulsescan be usedto preventinfibercanbeincludedin the Fourth,a dispersion-compensating interference. tersymbol themainfiber.Whenjust to in opposition slope path has a dispersion transmission that possible[6]' are of 250 GHz-km BDPs areused, thefirst two techniques Exampte8.2. Determinethe loss limit and the chromaticdispersionlimit of a high-performanceSMF optical fiansmissionsystemoperatingat 1300 nm and providinga bandwidthof 417Mbps (enoughto carry nine DS3 signals).Assumea narrowband$ourceis usedwith an output power that is 42 dB gteaterthan the receivepower(i'e., assumethe systemgainto be 42 dB). minimumacceptable

390

FIBEH oPTIcTRANSMISSIoN $YSTEMS

solution. As indicated in Figure 8.2, the attenuation of a single-mode fiber operating at 1300 nm is approximately 0.35 dB/km. Thus, .. 42 = 120km Lossllmit = 0j5 Using 250 Gbps-km as the BDp of a silica, single-modefiber,

Chromaticdispersion limit =

#:600

km

Whenthedispersionlimit is muchhigherthanthelosslimit, asis thecasein Exunple 8.2,it indicatesthattheopticalsourcesareprovidinga niurowerspecffumthanis nece$sary. Thus,the systemdesigncouldbe changedto uselessexpensivesourcesor the poweroutputof the sourcecouldbe increased to get a longerlosslimit. (Increasing thepoweroutputof the$ourcemay widenits spectrumandtherebyreducetheBDp.) Thesystemparameters of Example8.2arerepresentative of thesecond-generation FT seriesG systemsdeployedby AT&T [9]. ttre Fr seriesG systemshadmaximum repearerspacingof 48 km (29 miles),indicatingthat theinitial systemscouldbe installedwith significantmarginfor reliabilityandfutureupgrades. Many of theroutes were,in fact, upgradedto 1.7Gbpswithoutchangingthe fiber tIO, I il. Evenat the higherdatarates,the FT seriesG routeshadsignificantlossmarginfor splicingand the insertionof wavelengthdivisionmultiplexersasanothermeansof upgradingthe system.wavelengthdivisionmultiplexingis describedin section9.3. Example8.2 showsthatattenuation is thelimiting distancefactorin 1300-nmsystemsoperatingat low andintermediate datarates.To operatewith evenlongerdistances,Figure8.2 indicatesthat 1550nm shouldbe used.with typical silica fibers, thelimiting factorat 1550nm wouldbechromaticdispersion. To ouirco-" this limitation,two othertypesof fiber havebeendeveloped;a dispersion-shifted singlemodefiber (DS-SMF)that hasminimal dispersionar 1550nm and a dispersionflattenedsingle-mode fiber (DF-sMF) that haslessdispersionacrossa rangeof wavelengths.Figure8.6contraststhedispersionof bothof thesetypesof fiberswith a conventional SMF' EventhoughFigure8.6indicateszerodispersionexistsat 1300 nm on thesMF andat 1550nm on theDS-SMF,someaffountof dispersionhasto be assumeddue to fiber manufacturingvariationsand uncertaintyof the actualcenter wavelengthproducedby a source.valuesof l-3.5 psec/kmnm aretypicallyusedas thedispersioncoefficientat the "zero"dispersionpoints. 8.1.2 Electrical.to.Optlcal Transducers Two basictypesof semiconductor devicesconvertelectricalsignalsinto opticalsignalsandcanhavetheoutputcoupledinto anopticalfiber:laserdiodes(LDs)andlightemittingdiodes(LEDs).LDs generallyprovidebefterperformance in termsof higher outputpower'Breaterbandwidth,andnarrowersignalspectrum.LEDs, on the other

sYsrEMELEMENTS391 oprc rHANsMlsstoN 8.1 FIBER

20 E

Ero

J

t

E

.E 0 -to .t a -20

Wrydength(nml Figure 8.6 Dispersion of SMF and DS-SMF fibers.

hand,arelessexpensive,requireonly simpleinterfacecircuitry, afe moretolerantof conditions,andaregenerallymorereliable.ThusLDs areusedfor long environmental andinterfaceco$tsare transmissiondistancesandLEDs areusedwhenruggedness moreimportantthanperformance. to minimize it is necessary To achievemaximumdistancesbetweenrepeater$, LDs forms havebeen of chromaticdispersionby usingnarrowbandsources.Various 8.7 developedthatcomecloseto achievinga spectrallypureoutput.Figure showsthe "single-ftequency" devices: typicaloutputspectrumof onethemostpopulartypesof a Fabry-Perotlaserdiode.Anothertype of diodewith a very nalrow spectrumis a LD (DFB-LD). As indicated,the dominantmodeof a Fabrydiskibuted-feedback Perotdiodeis typically10dB strongertltantheadjacentmodes.To maintainthespecto tightly confol the operationalenvironment trum indicated,it is generallynece$sary Otherwise,modulation-dependent (biaspoints,modulationlevel, and temperature). effectsbroadenthe spectrumor evencausevery shoftdurationmodesshiftsthatpro-

6

s

$ D

-2 d

E .E

-

2

Ftgrrre 8.7

l o + t + lrom Cenurwarbltngth (nml Wwsler|g8h Representative laser output $pectrum.

2

392

FIBEROPTICTRANSMISSION SYSTEMS

parametersof VarloueOpticalSources TABLE8.1 Repreeentatlve DeviceType

(nm) Wavelength

Si LED Ge LED InGaAsP LED DFBLD DFBLD IUDFBLD

Launched Output Power(dBm) -16 -19 -10 -5 -5 +2

850 1300 1300 1300 1550 1550

FWHMSpectrum Width'(nm) 50 70 120 1.0 0.4 0.8

ducea dominantmodeat oneof theadjacentwavelengths. This latterphenomenon is referredto asmodepartitionnoise[12], which cancausea halvingof the BDp of a system[6]. Thespectralwidth of an opticalsourceis commonlyspecifiedasthefullwidth at half-maximum(FWHM), which representsthe spreadbetweenthe wavelengthsat whichthe specrrumis arhalf themagnitudeof thepeak. To achievegreaterdistancesin loss-limitedsystems,eithera more sensitivereceiveror a higherpoweredsourceis necessary.* Laserdiodesprovidethe highest availablepowers,asindicatedin Table8. l, which listsrepresentative parameters of variousLD andLED source$. Systemsthataredispersionlimitedcannoibeimproved by higher output powersbecausethe intersymbolinterferencedominatesthe imperfectionsin thereceivedsignal. The effectivetransmitpower dependson the couplingefficiencybetweenthe sourceandthe fiber. The couplingefficiencymay be aslow as l%oin thecaseof an unfocusedLED connecred to a single-mode fiber. High-efficiencycouplingsof 6ovo arepossiblewith focusedsources.The launchedpowervaluesprovidedin Tableg.I includethecouplingefficiencies. Bandwi dth-Dlstance Prod ucte Thebandwidth-distance factorof a fiber systemresultingfrom chromaticdispersion is determined from thefiberdispersion coefficientandthespectralwidthof thesource. Themaximumacceptable amountof pulsespreadingwith an NRZ line codeis typically specifiedto be one-foufthofa bit intervalT. Thus. DL AL!0.257 whereD = Z= A.l,= 7=

(8.1)

dispersioncoefficientof fiber (psec/kmnm) distance(km) spectrumwidth of source(nm) durationof a bit interval(psec)

Transmit power cannot tle increased arbitrarily without encountering nonlinear phenomena in the fiber itself. The optical power threshold where nonlinear effects begin to occur may bJas low as l0 mW [6],

ELEMENTS 393 SYSTEM 8.1 FIBEROPTICTHANSMISSION

Using the relationshipthat the bandwidth(B) is the reciprocalof the bit interval, the BDP is determinedas NRZ-BDP=Bi

-< 250 DAl.

Gbps-km

(8.21

because energyin oneinA50VoRZ line codeallowstwiceasmuchpulsespreading interval. Thus tervalis half a bit periodfrom energyin the next

RZ-BDP. #

Gbps-km

(8.3)

but areinfesy$tems TheRZ line codesareobviouslyprefenedin dispersion-limited thereceivermustdetectpulse rior to NRZ line codesin loss-limitedsystemsbecause with half asmuchenergy(assumingthepeaktransmitpoweris heldconstant). Example8.3. Determinethe BDP of a 1550-nmSMF systemand a 1550-nm DS-SMF$ystemusingthe 1550-nmDFB LD sourceidentifiedin Table8.1.Assume an NRZ line codeanda "zero" dispersionvalueof 3.5psec/kmnm to allow relaxed tolerancesfor the fiber andthe sourceoperatingwavelength. Solution, From Figure8.6,the dispersioncoefficientof an SMF fiber at 1550nm is 16psecflrmnm. FromTable8.1,the spectralwidth of the sourceis 0.4 nm. Thus, ?qn

NRZ-BDP=ffi=39GbPs-km Using the given mined as

"zero" dispersionvalue,the BDP of the DS-SMFsystemis deter-

Gbns-km NRZ-BDP=5ft= 179 8.1.3 Optical-to-ElectrlcalTransducers Two basicfypesof photodetectorsare availableas transducersto convertthe optical energyat the receiverto electricalenergyfor amplificationandotherprocessingsuch asclockrecoveryanddatadetection.Siliconbasedp-i-n diodeswerethefirst photodeof 800-900nm. Theseditector$to be used,operatingin systemsusingwavelengths odes are inexpensiveand reliable and provide good performance.Their major drawbackis that they do not operateat higher wavelengthswherefiber loss is mini-

394

FIBEROPTICTHANSMISSIONSYSTEMS

TABLE8.2 RspresentativeParametersof OptlcalDetectors(BER= 1g-tt1a DeviceType

Wavelength (nm)

SipFn Si APD InGaAspr:n InGaAsAPD InGaAsAPD InGaAsAPD InGaAspr-n InGaAsAPD

850 850 1310 1310 13 1 0 1550 1550 1550

ReceivePower(dBm) -48.0 -58.0 -35.0 -43.0 -26.0 -37.5 -37.0 -33.0

DataRate(Mbps) 50 50 420 420 8000 678 1200 4000

aThedetectorsensitivities assume an NF|Zline code. The sonsitivitiesdecrBaseby g dB for (S0%)BZ codes.

mized. Germanium devicesfunction at the higher wavelengthsand are more sensitive but are more temperaturedependentand less reliable. The secondbasic type of photoderectoris an avalanchephoto diode (ApD), which enhancesthe receiver sensitivity becauseit operateswith internal gain. (A p-f-n diode does not have internal gain and therefore requires all amplification to be externally applied, which raisesthe noise level.) The amplification inherent in the ApD's conversion from an optical signal to an electrical signal is useful becauseit meansthe ApD can be 10-15 dB more sensitivein detectinglow-level signals at a given error rate. A major drawback of an APD is ttrat it needsa high supply voltage to operateand is very sensitiveto temperature.APDs tend to have somewhatlower GBps thanp-i-n diodes, which restricts their use in very high data rate systems.Table g.2 lists various types of photodetectorsand provides representativeperformanceparametersfor each.

6 E .E

,E 6

a

-60L .001

Figure 8.t

.01

.1 .2.4 OEtrR{te (Gbm}

t

2

4

Receiversensitivityversusdatarate.

8.2 LINEcoDESFORFIBERoPTICTHANSMISSIoN 395

Bothtypesof receiversutilizedirectdetection,whichmeansthattheymerelymea$of opticalenergyto detectdata.Because higherdatarates urethepresence or ab$ence imply lessenergyper bit, the sensitivitydecreases with datarate.As a first approximation,a doublingof the datarateimpliesthereceiverbecomes3 dB lesssensitive. Thegeneralformulafor determiningthechangeof sensitivity(in decibels)asa function of a changein thedatarateis SensitivityR = SensitivityRo+ where

tot"*,.[ftJ

(8.4)

R = new data rate Ro = data rate at given sensitivity

Figure 8.8 plots the receiver sensitivity ofrepresentativep-i-n diode detectorsand APD detectorsas a function of the data rate for both NRZ and RZ line codes.

8-2 LINE CODESFOR FIEEROPTICTRANSMISSION (FOT)systemsmustsatisfythe samebasicreLine codesfor fiber optictransmission quirements in described sy$tems astheline codesin electrical(wireline)transmission Chapter4. Somefiber systemsusethe line codesdescribedpreviously.A few line here.Opticalsources codesdeveloped specificallyfbr fiber applications aredescribed anddetectors areprimarilyusedin nonlinearmodesof operationwith significantgain andthresholdvariations,which impliestheyarebestsuitedto operatingin only two states:on or off. Hence,a two-levelline code(on-off keying)is mostnatural.Consideringthe wide bandwidthavailable,multilevelline codesareusuallyunnecessary andextratiming transitionscanbe includedin theline codewithoutsignificantpensystems alty.An exceptionto theavailabilityof "free"bandwidthoccursin submarine wheremaximumrepeaterspacingis achievedby minimizingthebandwidthof theoptheneedfor ultratical signal.Thesesystemsaretypicallydispersionlimitedbecause high reliabilityprecludesthe latestsingle-frequency laserswith unprovenlongevity. of A particularlyusefulpropertyofdirect detectionopticalreceiversis theabsence any polarity ambiguity,which meansthe transmitterandreceiverare inherentlydc Eventhoughthe opticalsignalis incoupledsodifferentialencodingis unnecessary. beherentlydc coupled,the dc balanceof the line codemay still be a consideration (particularlyAPDs)is sometimes on dependent causethe gain of somephotodiodes it is usuallydesirableto accoupleamplifier thedc levelof theline code.Furthermore, stagesin thereceivers[3], which meanselectricalpulseamplitudesat thedetecting aretypi comparator$ aredependent In all, theseconstraints on opticalpulsedensities. in wirelinesy$tems, so line codesfor cally easierto dealwith thantheircounterparts fiber systemsgenerallytolerateBreateramountsof dc variation.Noticethat,in fiber and a CMI line codehaveunvaryingdc levelsequalto systems,both a Manchester pulse amplitude.A numberof otherline codesusedin fiber systems one-halfof the allow somevariationin theaveragedc level.Becausethesevariationsarecontrolled,

396

FIBER opTtcrRANsMtsstoN sysTEMS

theline codesaresometimes refemedto as"dc-constrained" line codesasopposedto "dc-balanced" beingperfector near-petfect line codes.Thedc variationsthatdo exist in theseline codeseitheraretoo smallto causeproblemsor areaccommodated by providing someform of dc restoration[14]. For applications thatdo notrequiremaximumrepeaterspacingor maximumdata ratestypically useManchester(diphase)or cMI line codesbecauseboth of these codesprovidestrongtiming contentanda completelyconstrained dc level (502oon). An early exampleof the useof cMI is a high-speedintraofficelink developedby NTT of Japan[5]. Manchester andCMI line codesexperience theworstcaseof boththelosslimit and thedispersionlimit. The worst-case losslimit arisesbecause thereceivermustdetect opticalpulseswith durationequalto half thebit interval(i.e.,with a sensitivityof an RZ code).The dispersionlimit of a Manchester or CMI line codeis actuallyhalf of theNRZ dispersionlimit because thesymbolrateis essentiallydoubled.(Eachbit intervalis dividedin half, with the needto derecra pulseor no pulsein both halves.) Thus,theManchester or CMI BDP canbe determinedas

=# MC-BDP

Gbps

(8.s)

Fromthepreviousdiscussion it canbe seenthatwhenperformance is moreimpoftant thanlow cost,someotherline codeis needed.Threebasicline codesarecommonly used:scrambledNRZ, scrambledRZ, or somevariationof a codereferredto as an mBnB line code,described below.Thechoiceof anNRZ or RZ line codeis primarily basedon whetherthe systemis losslimited or dispersionlimited.Loss-limitedsystemsuseNRZ line codeswhereasdispersion-limited systemsuseRZ codes.A scrambleris typicallyusedto enhance timingrecoveryandcontroldc wanderin a statistical sense(but not in an absolutesense).Often,overheaddatapatternsin thedatastream providessomeassurance thatthedatastreamcannotexactlymatchthe scramblersequenceso a ceftainminimumnumberof transitionsare assured.This is the casein SONETtransmission systems described in Section8.5. 8.2.1 mBnB Line Codee As an alternativeto relying on scramblers for ensuringtiming transitions,a classof line codeshasbeendevelopedthatencodes rn binarydatabits into blocksofn binary line bits (mBnB),wherenr < n. Becauseonly 2* datacodesmustbe selected from 2, codesin eachblock,thereis codingflexibility for conhollingtiming anddc wander. For example,if m = 4 andn = 5, sixteendatacodesmustbe chosenfrom the setof thirtytwo 5-bit line codes.A specificchoiceof thesecodesis providedin Tableg.3, whichdepictsthe datacodesandcontrolcodesselectedfbr thefiber-distributed data interface(FDDD standard[6]. The significanta$pects of thecodingassignments in Table8.3 are:

g,z LINEcoDEsFoRFIBEH oFTrcTRANSMtssroN397 TABLE8.3 FDDI4BSBLlneCodee LineGode Decimal

Binary

0

00000 00001 00010 00011 00100 00101 00110 00111 01000 01001 01010 01011 01100 01101 0 11 1 0 01111 10000 10001 10010 10011 '101 00 10101 10110 10111 11000 11001 11 0 1 0 11011 11100 1 11 0 1 11110 11111

1

2 3 4 5 6 7 E

I 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31

Symbol

o V V V H L V R V 1 4 5 V T

Quiet Violation Violation Violation Halt Violation Re$et Violation

Violation

E -l

Violation

K B q

2 3 A B J S

Set Set

c D E F

o I

Assignment Functional

Name

ldle

Linestatesymbol Oisallowed Disallowed Disallowed Linestatesymbol Startdelimiter Disallowed Controlindicator Disallowed Datasymbol0001 Datasymbol0100 Datasymbol0101 Disallowed Enddelimiter Datasymbol0110 Datasymbol0111 Disallowed Startdelimiter Datasymbol1000 Dalasymbol1001 Datasymbol0010 Datasymbol0011 1010 Datasymbol Datasymbol1011 Startdelimiter Controlindicator 1100 Datasymbol Datasymbol1101 Datasymbol1110 Datasymbol1111 Datasymbol0000 Linestatesymbol

L The code spaceis used for control as well as data. 2. When transmitting data,the minimum pulse density is 40Voand there can never tre more than three intervals without a pulse. 3. The dc componentis constrainedto fall between a minimumof 40Vopulsesand a manimum af 60Vopulses, a range that is one-fifth of an unconstrained random data stream.

398

FIBERoPTIcTHANSMISSIoN SYSTEMS TABLE 8.4 Digital Blphaee (Manchester)lB2B Line Code

LineCode Binary 0

00 01 10 11

1

2 e

Functional Assignment Disallowed Data symbol 0 Data symbol 1 Disallowed

The featuresofthe 4B5B line code describedabove are achievedat the expenseof a25vo inweasein the line data rate. By way of comparison,a digital biphase (Manchester)line code and the cMI line code describedin chapter 4 can both be representedas lB2B line codeswirh the coding assignmentsprovided in Tables 8.4 and g.5, respectively.Notice that both of thesecodescome at the expenseof a lfi)7o increase in the line datq rate. Becausethe 4B5B line codedefined for FDDI is intendedfor a specific application, it contains codes for control as well as for data. The 5B68 codes given in Table 8.6 areintendedfor transmissiononly and thereforedo not allocatecode spacefor control. Notice that the 5B68 code assignmentsare made in such a way that the dc level is fixed at 507opulsesbut the maximum run length of no pulsesis 6. Thus, tighter control of the dc level comes at the expenseof increasing the worst-caseduration between pulses.5868 and TBBB line codeshave been used exten$ively in Europe. Examples of 7B8B usageare a 565-Mbps rerrestrial system developedby British Telecom [17] and a 280-Mbps NLI submarinesysremdevelopedby src of Great Britain [18]. The 4B5B and 5B6B examples use output blocks that are only I bit greater than the input blocks. considerably more coding flexibility is achieved when the ourput block is more than I bit longer than the input block. As an example, a 6B88 code allows all 64 input codesto be encodedwith an output code containing exactly four I's in every code. (The number of combinations of four I's in 8 bits is 70.) Thus dc wander can be maintained without having to altemate between low-density and high-density codesbut coding efficiency is sacrificed.

TABLE8.5 CodedMarkInversion(CMl)1B2BLineCode LineCode Binary 0 1 2 3

00 01 10 11

Functional Assignment Datasymbol1 (if 11 previously sent) Datasymbol0 Disallowed Datasymbol1 (if00 previously sent)

8.2 LINEcoDE$ FoB FIBEHOPTI0THANSMISSIoN 399 TABLE 8.6 5868 Line Code Aesignrnentss

58 lnput 00000 00001 00010 00011 00100 00101 0 0 11 0 0011'l 01000 01001 01010 01011 0 11 0 0 01101 01110 0 11 1 1

68 Output 011101/100010 101110/010001 010111/101000 000111 101011/01 0100 001011 001101 001110 110101/001010 010011 0 10 10 1 010110 011001 0 11 0 1 0 011100 110110/001 001

5B Input 10000 10001 10010 10011 10100 10101 10110 1 0 111 11000 11001 11010 1 1 011 11100 11 1 0 1 1 1' t1 0 11111

68 Output 111010/000101 100011 100101 100110 101001 101010 101100 0l1011/100100 110001 110010 110100 101101/010010 111000 110011/001100 101101/0't0010 100111/011000

aA code 1/code0 output must alternatebetweenthe code with four 1's and the code wlth two 1's to mainlain dc balanc€.

An 8B 108 coding affangementfl 9, 201hasbeen defined asthe fiber channeltransmission standard(ANSI X3.230-1994). As indicated in Table 8.7, each input byte is separatedinto a 5-bit field and a 3-bit field that are respectivelyencodedusing a 5B68 and 3B4B algorithm. In some cases,two codewords are provided for encoding particular input data. The choice of one code in a pair (which are complementsof each other) is made to maintain dc balance.[f the previously transmitted unbalancedcode had more I 's than 0's, the code with fewer 1's is chosen.If the previously transmitted unbalancedcode had more 0's than I's. the code with fewer 0's is chosen.*

8.2.2 Bit InsertionCodes The mBnB block codesdescribedin theprevioussectionhaveonesignificantdisadvantage:They aredifficult to implementon very high speeddatastreams.(In lower speedapplications, decodelogic or tablelookupin a smallread-onlymemoryis trivial.) Very high speedlinks thereforeuseanothertypeof codereferredto asbit insertion codes.Thesecodesareactuallyspecialcasesof mBnB codeswith n = m + 1 and thecodesselectedsothecoding/decoding logic is greatlysimplified. Thefirst bit insertioncodeto beconsidered is theMB lP code,whichmerelyinserts anoddparitybit aftereveryrz bitsof sourcedata.Oddparityensuresthatat leastone *Two

special cases exist that may alter the rule when 001 I I is 5B6B encoded or 0t I is 3B4B encoded.

400

FIBEBOPTICTHANSMISSIONSYSTEMS

I is includedin theparity field of m + I bits.As a point of reference, noticethat the digitalbiphase(Manchester) line codeis a 18lP line code.An exampleof theuseof a 24BlP line codeis theTransPacificsubmarinecable system(Tpc-3) put into servicein December 1988by AT&T andKDD of Japan[21]. An evensimplerbit insertioncodedescribedby engineers of NTT in Japan[22] is the mB I c code,which merelyinsertsa bit aftereveryru sourcebits-the valueof whichis thecomplement of theiilrmediatelyprecedingbit. Thus,theaddedbit always forcesa datatransitionandis veryeasyto implement.An obviousdisadvantage of the simplicityis thelossof performance monitorabilityoverthefirst rn- 1bitsof a block. Againnoticethata diphaseline codeis a degenerate caseof an mBlc line codewith ru = L An exampleof theuseof a l0BlC line codeis theF-1.6Gsysremof NTT in Japan[23].

TABLE8.7 88108 FlberChannet Codtng 58 Input 0 (00000) 1 (00001) 2 (00010) 3 (00011) 4 (00100) 5 (00101) 6 (00110) 7 ( 0 01 1 ) I (01ooo) I (01001) 1 0( 0 1 0 1 0 ) 1 1( 0 1 0 1 1 ) 1 2( 0 1 1 0 0 ) 1 3( 0 11 0 1 ) 14(01110) 1 5( 0 1 1 1 1 ) 38 Input 0 (000) 1 (001) 2 (010) 3 ( 0 11 ) 4 (100) 5 (101) 6 (110) 7(11)

68 Output 100111/011000 011101i10001 0 101101/010010 110001 110101/001010 101001 011001 111000/000111 111001/000110 100101 0 10 10 1 110100 001101 101100 0 111 0 0 010111/101000 48 Outout 0100/1011 1001 0101 1100/0011 0010/1 101 1 0 10 0 11 0 0001/1 110

58 Input 16(10000) 1 7( 1 0 0 0 1 ) 1 B( 1 0 0 1 0 ) 1 9( 1 0 0 1 1 ) 2 0( 1 0 1 0 0 ) 2 1( 1 0 1 0 1 ) 2 2( 1 0 1 1 0 ) 2 3( 1 0 1 1 1 ) 24 (11000) 25(11001) 2 6( 1 1 0 1 0 ) 2 7 ( 11 0 11 ) 28(11100) 2e(11101) 3 0( r 1 1 1 0 ) 3 1( 1 1 1 1 1 )

68 Outpul 011011/100100 10001 1 0 1 0 011 110010 001011 101010 011010 111010/0001 01 110011/001 100 100110 010110 110110/001 001 001110 101110/010001 011110/100001 101011/010100

DIVISION MULTIPLEXING 401 8.3 WAVELENGTH

8.3 WAVELENGTHDIVISIONMULTIPLEXING Wavelengthdivision multiplexing(WDM) is the basictechniqueusedto establish multiple,independent opticalchannelson a singlefiber. The conceptof WDM is ilsourcesinto a lustratedin Figure8.9 showingthecoupling(multiplexing)of separate (demultiplexing) of the signalsout of the fiber. Prismaticrefiber andtheseparation fractionis depictedasthebasiccouplingmechanism at bothendsof thelink. Refraction is usable only for channelswith relatively large wavelengthseparations. Diffractive gratingdevices,on theotherhand,canoperatewith channelspacingon the of being orderof 1-2 nm [24]. All WDM devicessharethe commoncharacteristics purelypassiveandbeingreversibleso anyparticulardevicecanperformeithermultiplexingor demultiplexing functions.Otherthansomeopticalinsertionlossandsome crosstalkfrom imperfectseparation,the multiplexing/demultiplexingoperationsare transparent to the individualchannelsignals.The insertionlossesof the diffraction gratingdevicescanvaryfrom I to 7 dB dependingon fiber size(multimodevs. single mode)andnumberof channels[24]. When the f,rrstfiber routeswereinstalled,WDM wa$not usedbecausea single pairs high-speed transmitterandreceiverwasgenerallylessexpensivethanseparate and receiversand accompanying WDM devices.Thus, of lower speedtransrnitters upgradeto increasethecapacity WDM wouldtypically be installedasa subsequent of a systemwithout laying new fiber. For example,the original 90-Mbps,820-nm FT3C systeminstalledin 1983by AT&T in the northeastcorridorwasupgradedin 1984by addingan additional180Mbpson a 1300-nmcarrier[25].In addition,referby ences[9] and [0] describehow the FT SeriesG canhaveits capacityincreased multiplexinga 1550-nmsystemonto an existing 1300-nmsystemand/oradding higherspeedelectronicswithouttakingtheexistingsystemout of service. The passivepropertyof WDM devicesis a dominantattractionfor someapplications suchasfiber to the home.Allocatingindividualfibersfrom a cenffaloffice to impJyingtheneedfor someform of fiber sharing.Theoutside eachhomeis expensive, plantenvironment(-40-70"C) andtheneedfor remotepowerfor activecomponents imply thatpassiveWDM devicesareprefened.Theuseof WDM in this application is sometimes referredto asthepassivephotonicloop (PPL) [26]. identicalto FDM asusedon electrical(copper)or electroWDM is fundamentally In fact,usingthefollowingrelationshipallows systems. magnetic(radio)transmission wavelengths to be relatedto frequency: v =,fl

Flgure 8.9 Wavelengthdivisionmultiplexing.

(8.6)

4OZ FIBER oplc rHANsMtsstoN sysrEMs where y= velocityoflight,= 3 x 108m/sec ,f = frequency(Hz) L.= wavelength(m) optical systemsaredefinedin termsof wavelengthasopposedto frequencybecause optical$ources aretraditionallyspecifiedin termsof thewavelengths of theiroutputs. Nevertheless, systemswith extremelyclosewDM channelspacing(on the orderof 0.04nm) aresometimes referredto asFDM systems[27]. systemsthat useelectronicmultiplexingto producedelectricalFDM signals for modulationof a singleopticalcarderarereferredto assubcarriermultiplexing systems.Becausethe individualchannelsof thesesystemsaretypicallyclosetogetherin frequency,andhencein wavelength, passiveseparation is usuallyinfeasible. Example8,4. Determinethe differencein wavelengthof two optical signals separated by 2GHz andcenteredat 1500nm. Solution. The frequency(in freespace)ofa 1500-nmsignalis

f=#-=2oo,oooGHz 1500x l0-o Thusthe upperandlower frequencies aredeterminedas200,001and 199,9g9GHz, respectively. Thecorresponding wavelengths are

Ir= ^

n':

3xlOE = 1500.007 nm 199.999 x l0e 3x108

too.oot ^, ,t

= 1499'992 nm

andthediff'erence in wavelengths is 0.015nm. Example8.4 showsthatevenwhensubcarriers areseparated by zGHz,the resulting wavelengthdifferenceis small-too smallfor passivedemultiplexing. Wavelengthdemultiplexingwith passivediffuactiondoesnot inherentlycausea lossof signalpowerin the individualchannels.An altemativeapproachto demultiplexingis shownin Figure8.10,whichinvolvespowersplittingof thereceivedsignal followedby wavelengthfilteringto extractindividualchannels. This approachis passive and functionallyidenticalto diffractive separationbut is wastefulof optical power.Its mainadvantage is thatpowersplittingcanbe implemented aspassivetaps distributedalongthefiber roure.Thus,it is a usefultechniquein applications (like local areanetworks)wheredistancesarelessimportantthanflexibility in networktopology.

DESIGN 403 8.4 FIBEHSYSTEM

Figure 8.10 WDM with powersplining/filtering.

DenseWavelengthDivisionMultiplexing primarilyin response in WDMtechnology occurred Majoradvances in thelate1990s, expand of theirfiber thebandwidth to theneedsof common carriers to dramatically facilities for high-bandwidth data networking. In contrastto previous systems,which carried a small number of WDM channels,the newer systemscarried a large number of closely spacedwavelengths.Thesesystemsare generallyreferredto as densewavelength division multiplexing (DWDM) systems. An example of a first-generation (1996) DWDM systemis the MultiWave 1600 $ystemfrom Ciena Corporation. This systemprovided 16 channelsspaced0.8 nm apafr in the region of 1550 nm. DWDM systems are enabled by optical amplifiers (EDFAs) that transparently amplify all wavelengthsin the band and by fiber Bragg gratings fabricatedinto glassfiber for demultiplexing and filtering in a receiver. DWDM usefulnessis not confined to retrofitting of existing plant. DWDM may be the only possibleway to achieveextremely high bandwidths.A 4O-channelsystemoperating at 2.488 Gbps (OC-48) per channel provides an aggregatebandwidth of 100 Gbps-a difficult speedfor a single TDM channel,particularly on dispersionlimited fibers and with other high-speedlimitations such as polarization mode dispersion.An additional advantagesof DWDM is the inherent transparencyof individual wavelengths, which allows transmission and interoperability of mixed types of services. Yet another advantageof DWDM is the inherent reliability provided by separateelectronics for each wavelensth interface.

8.4 FIEERSYSTEMDESIGN The examplesin the previoussectionshighlightonly the mostbasicaspectsof fiber for thedesignof considerations optictechnology.This sectionprovidessystem-level practicalFOT systems.As is the casewith anytransmission technology,the system on systemavailabilityrequirements andcostof maintedesignis stronglydependent f,rberconnections to nance.At the low endof theserequirements areeasilyaccessible individualworkstationswithin a singlebuilding.At theotherendof the spectrumof cablesthatareobviouslyveryexpensiveto rereliabilityrequirements aresubmarine pair. Repairsareminimizedby usingonly ultrareliablecomponents. Eventhoughsubgophers, must contendwith oceanic they marinecablesareimmuneto bacftfioes* and 'Some

outside plant enginecrsofthe telephoflecompanieswouldprobablynot be too surprisedifabackhoe could somehow get to a submarinecable.

404

FTBER oplc rRANsMtsstoN sy$rEMS

counterparts:fishing trawlersandsharks(which seernattractedto electromagnetic radiation).AT&T developed a "seaplow"for buryingsubmarine cablesin areasof largescalefishingactivity[28].Abouttheonly advantage thatsubmarine applications have in their favor is theconstant,relativelylow temperature environmentfor repeaters/amplifrers. 8.4.1 Fiber Connectors and Sptices Threebasictechniques ofjoining fibersareconnectors, mechanical splices,andfusion splices[29]. connector$areusedfor terminalconnectionrr, patchpanels,or otherinstanceswhereloss(0.4-l dB) is lessimportantthaneaseof maintenance andreconfiguration. Mechanicalsplicesinvolvemechanicalalignmeutandclampingof two adjoining fibers.Therearemanydifferenttypesof mechanicalsplicingequipmentthatprovide variouslevelsof lossdependingon thefiber size(multimodeor singlemode)andinstallationtime.Splicescanbemadeon individualfibersor in bulk on fiber bundlesor ribbons.Two basicsplicingprocedures arepossible:passiveor active.Passivesplices arefasterbecausethe craftsperson merelyplacesthe fibersin the alignmentsleeves andclampsthefibersin place.Activealignmentinvolvespassinga signalthroughthe spliceandmakingfine mechanicaladjustments to minimizethe lossbeforethe final clampis applied.A singlefiber splicedescribedin reference[30] providesanaverage of 0'2 dB losswhenpassivelyinstalledand0.05dB losswhenactivelyinstalled.The installationtimesare5 and7 min, respectively. Fusionsplicesprovidethelowestlossesbecause theyessentially form onecontinuousfiber that is almostasgoodas a single-drawn fiber. In fact, submarinefibersare pretestedat the factory for tensilestrengthandfusedat the weakpointswherebreaks occur[28].Thisprocessensures thatthefiberscanwithstandthetensilestressencounteredwhenlayingandrecoveringthefiber cablebut introducesslightlyhigherlevels of averageattenuation in the fiber. For example,the I 04-kmrepeaterless systeminstalledbetweenTaiwanandtheislandof PengHu has0.24dB/kmar 1550nm including all splices [28]. Fusion splicing equipmenris availablerhar is completely automated with high-resolution TV usedto alignthefibersandestimatethespliceloss by measuring themisalignment afterfusion[29].Fusionlossescanvarybetween0.01 and0.I dB. For undersea applications, only valuescloseto 0.0I dB areaccepted. 8.4.2 Protection Switchlng Protectionswitchingwithin an FoT systemis basicallythe $ameasin otherhighcapacitytransmission $ystemswhereinonespareline, includingtransmitters, repeaters,and receivers,can be switchedinto serviceif one of Nmain lines fails (l : N protection).Thereare,however,a few uniqueconsiderations for protectingfiber systems,arisingprimarilybecause of thehighcapacityof thesy$tems. First,a systemcan be installedwith lower cost(lowerperformance) optoelectronic deviceswith an upgradefor higher performanceanticipatedat a later date.In this case,the protection

DESIGN405 8.4 FIBEH SYSTEM upgrading switchingprocedures andpackagingconceptsshouldallow one-at-a-time of electronicswithoutdisruptingservice[11]. fiber A moresignificantanddifficult aspectof protectionswitchingarisesbecause routestend to carry large-traffic volumesrepresentingaggregationsof traff,rcthat facilitieson $eparate route$.One-for-Nprowould otherwisebe carriedby separate tectionof mostfiber systemsis of no usewhena backhoecutsanentirecable.Protection from the lossof a completefacility requiresnetwork-levelroutingdiversity,a featureof a networkthattendsto disappear whenusingfiber andlargeswitchingsysmaximum advantage-large hubsfed by largetrunk groups. temsto their economic provides is a 1 ; I routediversifyandno hub dependence A networkarchitecture that SONETring describedlaterin this chapter.

8.4.3 System Galn The fust stepin determiningthe repeaterspacingof an FOT sectionis to determine betweenthe launchedoutputpower of a hansmitterand the receive the difTerence fashionto thesameterm powerrequiredfor a designated minimumBER.In analogous introducedfor radio systems,this parameteris referredto asthe systemgain. Notice thatthesystemgainincludescouplinglossesat theopticalsourceandopticaldetector. The systemgainmustbe greaterthanthesumof all thelossesin thepath.Sourcesof losses,bending loss include inherentattenuationof the fiber, splicing/connector losses,WDM devicelosses,andphotonicswitchlossesif any.Splicinglossesthatoccur atregularintervalsduringtheinstallationof thefiber areoftenaddedinto thefiber parameterinvolvedin determining attenuationsothereis only onedistance-dependent repeaterspacing.Thedifferencebetweenthesy$temgainandthesumof all thenominal lossesis the lossmargin.In anyparticularinstallation,the lossmarginallowsfor powerpenalties unanticipated dueto dispersion, tolerances, equipmentmanufacturing splices,componentaging,andpossibleWDM upgrades. Example8.5. Determinethesystemgain,theBDP,thedispersionlimitedrepeater data spacing,andthe lossmarginfor an FOT systemwith the following parameters: -5 = = = = dBm rate 565 Mbps, line code 5BdB RZ, wavelength 1550nm, source DFB-LD with 0.4nmFWHM, fiber= SMF,detector=InGaAsAPD,repeaterspacing = 65 km, andsplicinglo$ses= 0.2 dB/km. Solutinn, Theuseof the5BdBline codeimpliestheIinedatarateof 565(6/5)= 67t Mbps.Thereceiversensitivityfor 678Mbps is determinedfrom Figure8.8or Table 8.2as-34.5 dBm.Thus. Systemgain= -5 - (-34.5)= 29.5dB an RZ line codeis in use. Because

406

FIBEH oplc rRANSMtsstoN sysTEMS

BDP: :;so?, =73.6Gbps-km BDPspacing =m: 17xO.4 Pathloss= (0.2+ 0.2X65)= 26 dB

l09km

Lossmargin=29.5- 26 = 3.5dB

Example8.5is representative of the565-Mbpssystemdevelopedby Fujitsuof Japan for carryinganE5 digital signal(565.148Mbps)[3 I ]. Thefactthattherepearer spacing is 607oof theBDP limit impliesthatdispersionpenaltiesaresmall(estimatedto be 1.2dB).

8.5 SONET/SDH Thefirst generations of fiber opticsystemsin thepublictelephone networkusedproprietaryarchitectures, equipment,line codes,multiplexingformats,andmaintenance procedures. Somecommonalitywith othersystemsin thenetworkcamefrom supplierswho alsosupplieddigital radio system$. In thesecases,the multiplexingformats andmaintenance protocolsemulatedcounterparts in theradio$ystems, whichalsohad proprietaryarchitectures. Theonly thingin commonwith all of theradioandfiber systemsfrom all of the supplierswasthat the interfaceto the networkwa$$omenumber of DS3cross-connect signals.Proprietarymultiplexingformatsfor multipleDS3signalsevolvedbecause therewasno higherlevelstandardcompatiblewith theapplications. A DS4 signal, which is composedof six DS3 signals,requirestoo much bandwidthfor radiosystemsandcarriesa largercrosssectionof channels(4032)than neededin manyapplications. TheRegionalBell operatingcompaniesandinterexchange carriers(IXCs),theusersof theequipment,naturallywantedstandards sotheycouldmix andmatchequipment from different suppliers.This becameparticularlyimportantas a result of competitionamongtheIXCs who desiredfiber interfacesto thelocalexchange carriers(LECs)but did not wantto necessarily buy from the samesuppliersastheLECs. (It mightbe necessary for an lxc ro interfacewith a differentsupplierat eachLEC.) To solvetheseproblems,andothers,Bellcoreinitiatedan effort that wa$latertaken up by the Tlxl committeeof theExchangecarriersstandardsAssociation(ECSA) to establisha standardfor connectingonefiber systemto anotherat theopticallevel (i.e.,"in the glass").This standardis referredto asthe synchronous opticalnetwork (soNET) t32, 331.In the laresragesof thedevelopment of this srandard, ccITT becameinvolvedso thata singleinternationalstandardexistsfor fiber interconnect betweentelephone networksof differentcounties.Internationally, thestandard is known asthe synchronous digitalhierarchy(sDH) t341.ThesoNET srandard addresses rhe following specificissues: l. E$tablishes a standardmultiplexingformatusingsomenumberof 5l.84-Mbps (STS-1)signalsasbuildingblocks. 2. Establishesan optical signal standardfor interconnectingequipmentfrom differentsuppliers,

8.5 SONET/SDH 407

TABLE8.8 SONET$ignal Hlerarchy NorthAmericanDesignation Electrical Signal STS-1 STS-3

srs-12 STS.24 STS-48 STS-S6 STS-19?

OpticalSignal

oc-1 oc-3 oc-12 oc-24 oc-48 oc-96 oc-192

DataRate(Mbps) 51.84 155.52 622.08 1244.16 2488,32 4976.64 9S53.28

ITU-TDesignation STM.1 STM-4 STM-8 STM-16 STM-Sz STM.64

3. Establishes extensive operations, administrations, maintenance, and provisioning(OAM&P) capabilitiesaspartof thestandard. 4. Defines multiplexing formats for carrying existing digital signals of the asynchronous multiplexinghierarchy(DSl, DSlC, DSz, DS3). 5. SupportsCCITT (ITU-T) digitalsignalhierarchy(E1,E2, E3, E4). 6. Definesa DSOidentifiablemappingformatfor DSI signals. otherapplications capableof accommodating 7. Establishes a flexiblearchitecture of transmission rates. Wide-bandwidth with a variety suchasbroadband ISDN (greater by multiple signals than5l.84 Mbps)areaccommodated concatenating signalthat is STS-1signals.A STS-3csignal,for example,is an 155.52-Mbps treatedby the networkasa singleentity. The dataratesandsignaldesignationsof the SONEThierarchyareprovidedin Table trans* 8.8.At thelowestlevelis thebasicSONETsignalreferredto asthesynchronous port signallevel I (STS-I).Higherlevel signalsarereferredto asSTS-Nsignals.An STS-1signals.TheopticalcounterSTS-Nsignalis composedof Nbyte-interleaved partof eachSTS-Nsignalis atropticalcarrierlevel N signal(OC-N).Table8.8 also includesITU nomenclaturefor the SDH, which refersto signalsassynchronoustransport modulesN (STM-N).Becausecorlmon applications of theITU signalhierarchy lowest STM signalis a 155.52signal, level the cailrot efficientlyusea 51.84-Mbps Mbps(STS-3c)signal.* is primarilyconcerned with OC-Ninterconnect AlthoughtheSONETspecification STS-1andSTS-3electricalsignalsrwithin theSONEThierarchyareuseful standards, networkelements(e.g.,multiplexers, within a switchingoffice for interconnecting systems)t35, 361. switchingmachines,andcross-cormect .A

referredto asa STM-OSDH signal. 51,84-MbpsSTS-I SONETsignalis sometimes TAnSTS-1electical signalusesa B3Zs line codeand a STS-3electricalsignalusesa CMI line code.

408

FIBERoPTIcTRANSMISsIoN SYSTEMS

44.738 Mbpe

51,84OMbpB

sTs-t 81"840Mbpl

Byte Interlosrrgd Mux CEPT'4 139,?64Mhpr

Figure 8.11 Functional block diagram of SONET multiplexing.

8.5.1 SONET Multiplexlng Overview The first stepin the soNET multiplexingprocess(shownin Figureg.ll) involves generationof a 51.840-MbpssTS-l signalfor eachtributary.The sTS-l signalcontainsthetributary(payload)trafficplustransportoverhead. As indicatedin rhefigure, a variety of tributary type$areaccommodated; 1. A single DS3 per srs-l that can be a srandardasynchronous DS3 signal generated by anM13 or M23 multiplexer.Asynchronous DS3inputsarepassed transparentlythroughthe systemto a DS3 output.Becausethis transparent option exists, arry4.736-Mbps signal can be carried within the payload envelope. 2. A groupof lowerratetributariessuchasDSl, DSlc, DS2,or El signalscanbe packedinto the STS-I payload. 3. A higher rate (wideband)signal can be packedinto a multiple numberof concatenated srs-l signals.Prevalentexamplesof higher rate signalsare 139.?64-Mbps fourth-levelmultiplexesof ITU or a broadband ISDN signalat 150Mbps.Eachof theseapplications requiresthreeSTS-1signalsconcatenared togetherto form an sTS-3c signal.Higher levels of concatenation (to form srs-Nc signals)are possiblefor higherratetributaries.concatenared srs-l signalscontainintemalcontrolbytesthatidentifythe signalasa componentof

8.5 soNET/$Dn tt09 a higher speed channel so the integrity of the concatenateddata can be througha network. maintainedasit passes An STS-Nsignalis createdby interleavingbytesfrom N STS-l signalsthataremueachof All timing (frequency)adjustmentis donein generating tually synchronized. SONET node the individualSTS-I signals.STS-I signalsthat originatein another with a possiblydifferentfrequencyarerateadjustedwith theequivalentofbyte stuffing (described to theclockof thelocalnode.No matter later)to becomesynchronized is, whatthenatureof thetributarytraffic all STS-I kibutariesin a STS-Nsignalhave the samehigh-levelformatanddatarate. by first scramblingthe STS-Nsignal Opticalcarrierlevel-Nsignalsaregenerated (exceptfor framingbytesandSTS-IDbytes)andthenconvertingtheelectricalsignal with direct to an opticalsignal.Otherthanscrambling,the OC-N signalis generated conversionto anopticalsignal.Thus,thedatarates,formats,andframingof theSTSN andOC-N signalsareidentical. A SONET$ystemis definedas a hierarchyof threelevels-sections,lines,and dedipaths-as indicatedin Figure8.12.Eachof theselevelshasoverheadbandwidth level. in Figure As indicated catedto administeringand maintainingthe respective 8.11, oneof theoverheadfunctionsprovidedwithin an STS-Nsignalinvolvescalcu* lationandtransmission of a paritybytefor theentireSTS-Nsignal.Parityis alsodeasdescribedin the following section. fined for theotherlevelsofthe architecture 8.5.2 SONET Frame Formats Theframeformatof anSTS-1signalis shownin Figure8.13.As indicated,eachframe consistsof9 rowsof90 byteseach.Thefirst 3 bytesofeachrow areallocatedto transpoft overheadwith thebalanceavailablefor pathoverheadandpayloadmapping.The

Path Terminating Equipment

Line Terminating Equipment

Scction Torminatirg Equipment

S€ction Terminating Equipfiont

Line TBrminsting Equipment

Figure 8.12 SONETsystemhierarchy.

PEth Terminating Equipment

410

FIBERoPTIcTRANSMIsSIoN SYSTEMS

Tranrpon Ovshord Sction O\rrrhrld

Llno Owrheld

sTs-l lnformstionPayload

At 81 Dt

ta. El 02

c1 FT D3

J1 B3 c2

Hl 82 D4 07 D10 z1

HZ K1 D6 D8 Dlt t2

H3 K2 D6 D9 D12 E2

Gl Path F2 Owrhear H4 B

3 Columns

I Rowe

z4 z5 87 Columns

Figure 8.13 STS-I frameformat. tran$port overhead is itself composed of section overhead and line overhead. Path overheadis contained within the information payload as indicated. The 9 rows of 87 bytes (783 bytes in all) in the information payload block is referred to as the envelopecapacity.Becausethe frame rate is 8 kHz, the compositedata rate of each STS-I signal can be representedas the sum of the transportoverheadrate and the information envelopecapacity: STS-1 rate = overheadrate + information enveloperate = 9 x 3 x I x 8000+9 x 87 x 8 x 8000

= I-728x 106+ 50.112x 106 = 51.840Mbps

(8.7) The internalformatof theenvelopecapacityis dependenton thetype of tributarytraffic beingcanied.one aspectof theenvelopeformatthatis commonto all typesof traffic is the 9 bytesof pathoverheadindicatedin Figure8.13.The actuallocationand purposeof this overheadaredescribedin thenexttwo sections. As a specificexampleof a higherlevel(srs-N) signal,Figure8.14depictsthedetailsof an srs-3 signalthatalsorepresents theSTM-I signalformatin rru terminology. Transmission of the bytesoccursrow by row andleft to right. Thus,the first 3 bytesof an srs-3 framearethethreeframingbytesAl, Al, Al. Most of thesecrion andline overheadfunctionswithin an sTS-3 signalarecarriedin the srs-l number I overhead. Thusmanyof thecorresponding bytesof theothersrs-l signalsareunusedandaresodesignated with anasterisk.Notice,however,thatpathoverheacl is includedin theinformationenvelopefor eachof theSTS-I signals. After a frameof an srs-N signalis scrambled, a paritybyte (BIp-8) is generated thatprovidesevenparityovercorresponding bitsin all bytesof thesTS-Nframe.This paritybyte is insertedinto the sectionoverheadof the first STS-I signalof the next STS-Nframe.

8,5 SONET/SDH 411

TrunsportOverhead

ST$"3 lnformrtlon Payload

\ l A r A r A 2 A 2 A e C l C l C l 81 El FI D l * r D 2 D 3

Jl 83 c2

Jl 83 c2

Jl 83 c2

H3

Gl F2 H4 z3 z4

Gl F2 H4 z3 7A

Gl F2 H4 23 24

HI 82 D 07 D 1

Hl 82 4

Hl 82 D

H2 H2 H2 r + t(1 S r r D D8 0 " D l l r * p 1 2

21

z1

21

zt

:1)

(2)

(3)

(1)

z2

22

(2) (3) I Columns

H3 l(2 e D9 t

H3

z5

E2 (1)

(2)

(3)

z5

(1) (2) (3) (1) 261 Columns

z5

(2)

Flgure 8.14 STS-3frameformat.

8.5.3 SONETOperations,Adminietratlon,and Malntenance The SONET standaldplacessignificant emphasison the need for operations,adminishation, and maintenance(OAM) of an end-to-endsystem.As shown in Figure 8.15, the OAM architecture is based on the section, Iine, and path Iayers described previously. OAM standardizationis a requirementfor mixing equipment from multiple vendors and easeof managementof all levels of a system(an individual repeatersection or an end-to-endpath).

SectlonOverhead in eachSTS-1frame Thefunctionalallocationof the 9 bytesof sectionoverhead shownin Figure8.13are; Al Framingbyte= Fd hex(l1110110) A2 Framingbyte= 28 hex (00101000) Cl STS-l ID identifiestheSTS-lnumber(1, . . . ,1f)for eachSTS-lwithin an STS-Nmultiplex B I Bit-interleavedparity byteprovidingevenparity overpreviousSTS-N frame after scrambling PCM orderwire(localorderwire) El SectionJevel64-kbps Fl A 64-kbpschannelsetasidefor userpurposes D1-D3 An 192-kbpsdaia communications channelfor alarms,maintenance, control,andadministration betweensections Thefactthatthereis sucha richnessof maintenance supportat thesectionlevel (from onerepeaterto another)is indicativeof therecognizedneedfor extensiveOAM facilities andthe availability of economicaltechnologyto provideit.

E

a? $ (t

p u h

G

o

ts

o F rrl

z, (n ln od P AD ll

412

8.5 sONEr/SDn 413

Llne Overhead Thefunctionalallocationof the l8 bytesof line overheadin eachSTS-I frameshown in Figure8.13areasfollows: Hl*H3 Pointerbytes usedin frame alignmentand frequencyadjustmentof payloaddata;thefunctionsof thesebytesaredefinedin detailin Section 8.5.4 82 Bit-interleavedparity for line-level error monitoring Kl, K2 Two bytes allocated for signaling between line-level automatic protectionswitchingequipment D4*D12 A 576-kbpsdata communications channelfor alarms,maintenance, conffol, monitoring,andadministrationat the line level ZI.ZZ Reservedfor futureuse E2 A 64-kbpsPCM voicechannelfor linelevel orderwire Notice that the line-level OAM facilities are similar to thoseavailableat the section levelwith the additionof theprotectionswitchingsignalingchannelandHl, H2, and H3 pointerbytesusefor payloadframing andfrequencyadjustment. Path Ovarhead As indicatedin Figure8.I 3, thereare9 bytesof pathoverheadincludedin everyblock (9 x 87 bytes)of informationpayload.Theimportantaspectof thisoverheadis thatit is insertedwhenthe tributary dataarepackedinto the synchronouspayloadenvelope (SPE)andnotremoved(processed) Thus,it prountil thetributarydataareunpacked. netofthe paththroughthesynchronous videsend-to-endOAM supportindependent work, which may involve numerousintermediatemultiplexers,cross-connect Theexactlocationof these9 byteswithin thepayswitches,or add-dropmultiplexers. Thefunctions onpointervaluesdefinedin thenextsectron. loadenvelopeis dependent of the path overheadbytesare: Jl A 6a-kbpschannelusedto repetitivelysenda 64-bytefixed-length stringso a receivingterminalcancontinuouslyverify the integdtyof a path;thecontentsof themessage areuserprogrammable parityat thepathlevel 83 Bit-interleaved CZ STS path signallabel to designateequippedversusunequippedSTS STSpayloadmappingthat signalsand,for equippedsignals,thespecif,tc mightbe neededin receivingterminalsto interpretthepayloads Gl Status byte sent from path-terminating equipment back to path-originating equipmentto conveystatusof terminatingequipment andpathenor performance(receivedBIP error counts) F2 A 64-kbpschannelfor pathuser H4 Multiframe indicatorfor payloadsneedingframesthat arelongerthana single STSftame; multiframe indicatorsare usedwhen packinglower ratechannels(virtualtributaries)into the SPE Z3-ZS Reservedfor future use

414

FIBEROPTICTRANSMISSIONSYSTEMS

8.5,4 PayloadFramlngand FrequencyJu$tification Payload Framing The locationofthe 9 bytesofpath overheadin the sTS-l envelopeis not definedin termsof the srs-l transportframing.Instead,thepathoverheadis considered to be thefirst columnof a frameof datareferredto asthespE, whichcanbeginin anybyte positionwithin the sTS-l payloadenvelope(seeFigure8.16).Theexactlocationof the beginningof the sPE (byteJl of the pathoverhead)is specifiedby a pointerin bytesHl andH2 of theSTSline overhead. NoticethatthismeansthatanSPEtypically overlapstwo STS-1frames. Theuseof a pointerto definethelocationof the SPEframelocationprovidestwo significantfeatures. First,SPEframesdo nothaveto bealignedwith higherlevelmultiplexframes.It maybethatwhenfirst generated, anSPEis alignedwith theline overheadat the originatingnode (i,e., the poinrervalue is 0). As the frame is carried througha network,however,it ardvesat intermediatenodes(e.g.,multiplexersor crossconnects)havingan arbitraryphasewith respectto the outgoinghansportframing. If the sPE hadto be framealignedwith theoutgoingsignal,a full spE frameof storageanddelaywouldbe necessary. Thus,theavoidance of framealignmentallows SPEson incominglinks to be immediatelyrelayedto outgoinglinks withoutartificial delay.Thelocationofthe sPEin theoutgoingpayloadenvelopeis specifiedby setting theHl, H2 pointerto thepropervalue(0-782). The secondadvantage of the pointerapproachto framingspE signalsis realized whendirectaccessto subchannels suchasDSls is desired.Becausethepointerprovidesimmediateaccessto the startof an SPEframe,any otherpositionor time slot within theSPEis alsoimmediatelyaccessible. If thetributaryusesa byte-synchronous mappingformat,individualchannelbyteshavefixedpositionswith respectto thestart of thesPE.This capabilityshouldbe comparedto theprocedures requiredto demultiplex a DS3 signal.In a DS3 signalthereis no relationshipbetweenthehigherlevel framingandthe lower level DS2 and DSI framingpositions.In essence, two more framerecoveryprocesses areneededto identifya DSOtime slot.Theuseof pointers 87 Columnr--+ .fl H;

Frame0 9 Row*

F

{ h 'Psth

Frame I I Eowe

Overherd

=

],".87 Columnr

Figure 8.16 Representative locationof SPE.

8.5 SONET/SDH415 in the SONET architecture eliminates the need for more than one frame recovery processwhen accessingbyte-synchronouslower level signals.

FrequencyJ uetiflcatlon to each be synchronieed Althoughit is generallyintendedthatSONETequipment otheror to a cofilmonclock,allowancesmustbe madefor theinterworkingof SONET equipmentthat operateswith slighfly differentclocks.Frequencyoffsetsimply that an SPEmay be generatedwith oneclock ratebut be carriedby a SONETtransportruna frequencyoffsetis to accept ning at a differentrate.Themeansof accommodating in the SPEpointers.PointeradvariableSPEframeratesusingdynamicadjustments justmentsallow SPEframesto float with respectto thetransportoverheadto maintain a nominallevel of storagein interfaceelasticstores.Figure 8.17 showsthe basic a slow incomingSPE.If the elasticstorebeginsto empty, meansof accommodating positivebyte stuffingis invokedto skip oneinformationtime slot (the slot immediincrementing thepointertodelaythe atelyfollowingtheH3 byte)andsimultaneously SPEframeby onebyte. a fastSPEclock,requiressendinganextra Negativebytestuffing,to accommodate SPEbytewhenevertheelasticstorebeginsto fitl. As indicatedin Figure8.18,theH3 deslotcarriestheextrabyteof data,whichrequiresthepointerto be simultaneously cremented,therebyadvancingthe SPEframeby I byte. To protectagainsterrorsin theneedto incrementor decrement a pointer misinterpreting byte-stuffingoperations, is redundantlyencoded,*andthe new pointervalueis transmiffedfor a minimumof occurs.This implies threeframesfollowingtheframein whichthepointeradjustment that a l-byte adjustmentcanbe madeonceeveryfour frames(onceevery500 psec). For an analysisof the effectsof channelerrorson SONETpayloadpointers,seereference[37]. by Example 8.6. Determinetherangeof SPEdataratesthat canbe accommodated thebyte-stuffingoperationdescribedabove. Solution. FourSPEframesnominallycontain4 x 9 x 87= 3132bytesof data.Thus, thenominalSPErateis 8 x 3132x 2000= 50.112Mbps.Whenpositivebytestuff,rng a slowincomingSPErate,313I bytesof dataaretransmitted is usedto accommodate in four frames.Thus,the lowest,slip-freerateis Minimum sPE rate= 8 x 3131x 2000= 50.096Mbps a fast SPErate,3133bytesof Whennegativebyte stuffingis usedto accommodate dataaretransmittedin four frames.Thus,the highest,slip-freerateis MaximumSPErate= 8 x 3133x 2000= 50.128Mbps .During

the a jusrnent frame only, 5 even-numbered bits of the pointet value are invefied to indicate a negative stuff (data in byte H3). When a positive stuff occurs, 5 odd-numbercd bits of the pointer arc inverted,

416

FIBER oplc rBANSMtsstoN sysrEMS STS-Frsme -----l*--

I I

-

0p

Strn of ST*t SynchronourEmretope

Frlme z

FredGn + |

Pwlth 1 bi$ invtrted

Fromen + 2

Pffi

-P+

|

Ftlme n * 3

50Orr

Figure8.17 Positive justificationoperation. STS-I pointer Example8.6 demonstrates that the SONETclock accuracyrequiredfor maintaining sPE datais 50.112t 0.016Mbps-a very wide torerance of f320 ppm.In comparison,a soNET nodeis specifiedto maintaina minimumtimingaccuracyof z0 ppmif it losesits reference. Thus, the frequencyof timing adjustmentswas chosenmore from a desireto simplify theprocessthanfrom just assuringa marginfor slip-free operations. The useof byte stuffingto accommodate timing differencesintroduceswaiting timejitter into SPEpayloads, just asbit stuffingintroduces waitingtimejitter intoDSI signalsbeingasynchronously multiplexedinto DS2or DS3 signals.If thespE is carrying DSl traffic, the effectof byte sruffingat the spE rateis aboutone-fourrhof a bit intervalattheDS1rate.(Because theSPJE rateis 32timestheDSI rate,theduration of an sPE byte is one-fourththe durationof a DSI bit.) Noticethat this amountof

B.s $oNET/SDH STS-l

Poimgr V r l u e( P |

-l*

417

Framd

I

I Start of $TS-l

0rtg Synchronous

Fremea

Frame il + I

Pwith D bits ifiYErt€d

Fremea * 2

Frerne r + 3

500 gr

justification operation. Figure8.18 Negative STS-I pointer phasejitter is comparableto the amountintroducedby bit stuffinga DS2 signalinto a DS3 signal. 8.5.5 Virtual Trlbutaries usessubTo facilitatethetransportof lowerratedigital signals,the SONETstandard asshownin STS-I payloadmappingsreferredto asvirtual tributary(VT) structures, or subframes Figure8.19.ThismappingdividestheSPEframeinto sevenequal-sized accountfor VT blockswith 12 columns(108bytes)in each.Thus,the subframes bytes 7 x 12= 84 columnswith thepathoverheadandtwo unusedcolumns(reserved R) accountingfor the remainderof the 87 columnsin an SPE.The rateof eachVT structureis determinedas 108x I x 8000= 6.912Mbps.

(\l

c{

o

(l

rJt

FI

+

ct

(\|

gr

f

E

5

6

lDE

q

.o F

bs

ro

c V

iEn

f;a

f

c

5

6

,

(

llr

#

So .F + F

9

l-

E

rrl

. 5

h :

E

H

v== 5 EL t ! E

0!l

H

{

sE 'r Fr

E ro

L

i

c

E

n

c\ I

al

J

F

o

F

i

O

E

.8 E-3 E F

(o ro

tr b 5

o

+

C'

F

(rI

(\|

qEB6flTffifiH 418

FD i

S.5 SONET/SDH 419

TABLE8.9 SONETVirtualTributarles TributaryType DS1 E1 DSlC D52

Numberof VT Designation Columnsper VT

wl.5 w2.0 VTs.O VT6.O

Numberin VT Group

Maximum Numberin SPE

b

4 3 2

12

1

28 21 14 7

3 4

to carryoneoffour typesofsignals. canbeindividuallyassigned TheVT structures Dependingon thedatarateof a particularsignal,morethanonesignalmaybe caried within a VT structureas a VT group.All signalswithin a VT groupmustbe of the samet)?e, but VT groupswithin a singleSPEcanbe differenttypes.The particular lower ratesignalsaccommodated asVTs arelistedin Table8.9.The lastcolumnindicateshow manyof thelowerrate$ignalsarecarriedin a singleSPEif all sevenVT groupsarethe sametype. VT-SPEpayloadsareallowedto float within an STS-I SPEin thesamefashionas pointersto SPEpayloadsareallowedto float at theSTS-1level.Thus,a secondlevel of pointerlogic is definedfor VT payloads.Again,a floatingVT-SPEallowsfor mininodesandfor frequencyjustificationof VT-SPEs mal framingdelaysat intermediate High-rateVT-SPEsareaccoillmotransitionsbetweentimingboundaries. undergoing datedby insertingan informationbyteinto V3 while slow-rateVT-SPEsareaccommodatedby stuffing into the informationbyte immediatelyfollowing V3 when necessary. The mappingformatfor a VTl.5 is shownin Figure8.20.EachVTI.5 usesthree columnsof datato establish108bytesin a VT1.5 payload.Therearefour suchpayloadsin a l2-columnVT group.TheVl, V2, V3, V4 bytesof thepayloadhavefixed positionswithin theSTS-I payload.Theremaining104bytesof theVT1.5 signalconstitutetheVTI.5 payload,the startof whichis theV5 bytepointedto by Vl andV2. Figure 8.21depictstwo differentmappingsfor a VTl.5 payload:an asynchronous mapping. mappinganda byte-synchronous Aeynchronous Mapping The asynchronous operationis identicalin conceptto the bit-stufflngoperationdescribedin Chapter7. TheDSI bit streamis insertedinto the informationbits (I) with no relationshipto the VT-SPEframeor byteboundaries. As indicated,therearetwo (Sr and52)availablein everyfour-framesuperframe. Thus,the stuffingopporhrnities VTl.5 superframe carries771,772,or773 informationbits dependingon thevalueof the stuff controlbits C1andC2.The nominalnumberof informationbits in eachframe is 193x 4=772. Nominalframescarryinformationin 52whilestuffingin 51. network, Because theasynchronous operationis compatiblewith theasynchronous of the asynit is the formatusedin mostSONETapplications. The majoradvantage of fransmission chronousmodeof operationis thatit providesfor totallytransparent

42O

FIBERoPTIcTRANSMISSIoN $YSTEMS

l--fr4.l

eCoUmfiB

rooooffi

ilEeR.un )offiool

nmootl0

ffin tr ffin tr

I Rorva

ffil1

g Rorys

l) A Wl.f rm lhreemluffir of m SPE fr f 08 bytcain a j00 Dre. f,perFrffio. ?) Vl , V2, V3, V4 bytcr hsveffrfid locdiilB h I $ptsidmified by tho hf t$o b&BdH4. 3) Vt ed V2 poid' to V5 qftirh is rhe fr* tryteofba flocing WiJ Sff. 4) SEEoncrtod bl,tff V5, .E!,26, Z7 mr h idediaal rd*ive porilima of m SFE. Figure E.20 Superframe sffucture for VTI.S kibutaries.

the tributary signal in terms of information and in terms of information rate. The major disadvantageof the asynchronousmode is that 64-kbps DSO charurelsand signaling bits are not readily extracted.

Eyte-Synchronous Multip I exlng In contrastto the asynchronous mapping,the byte-synchronous payloadmapping shownin Figure8.21ballocatesspecificbytesof rhepayloadto specificbytes(channels)of theDSI tributarysignal.Hence,this modeof operationovercomes themain drawbackof theasynchronous modein that64-kbpsDSOchannelsandsignalingbits within the payloadareeasilyidentified.In fact, whenthe DSl tributaryarisesfrom legacyapplications, the signalingbits of a DSI aremovedfrom the leastsignif,rcant bit (LSB) of everysixthframeof respectivechannelsandplacedin dedicatedsignaling bit positionswithin rhevr-sPE. Thusbyte-synchronous multiplexingoffersan additionalfeatureof convertingfrom in-slot signalingto out-slotsignalingfor DSI signals. Al importantaspectof thebyte-synchronous formatshownin Figure8.21bis the absence of timingadjustments for thesourceDSI signal.Thus,theDSI interfacenecessarilyrequiresa slip bufferto accommodate a DSI sourcethatmaybe unsynchronized to the local soNET clock.Althoughslipsin byte synchronously mappedDSI

s.5 soNET/sDn421

v5 R R R R R R I R lffi I htts (246) J2

c.Go o o o I R lga I bftB (246)

z6

c.Go o o o I R 194 | bltl

(2/sd) z7

c . c ' RR R q S R lm I bfts (24d)

t"24 Eyteg 2 Byte8 24 BytEs 2 Eyteg 24 Byt€B 2 B!,toB 24q,tss

\15 R P n$ S " S n $ r F R ClS0clHnneh (1-24) J2

R P ^s r s " s " g , F R DSOchannels (1-24

ze eP"S'S"9"$rF R DSo chann€l8 (1-24't

z7 F. P" Sr Sr8.g,

F R

DSOchsnnolB

(r.24)

1(}{Bytee No, oflixod informdim bils = 4(19?) +3(1, = 77 I $1,$1- Surfflinfrnuatim bitr Cr, c! -Sutrmgor*rot bits R bitc ffe SxEdrtlfftte (uupdcifiod) O bits arsrctcrved fm futurupurpo*e

(a)

Sr, Sr" $r, 8r - Chmel-asso+iEtrd sigtalingbiB P1,Ps= Si8nrlingphuc indicdsr bits F=DSl *aftinebit R bih $e ItuEdsrtrbitr (uspociffed)

(b)

Figure 8.21 DSI mappings in VTl.5 SPE: (a) asynchronous;(b) byte synchronous.

signalsmayoccurattheSONETnetworkinterface(e.g.,SONETgateway),slipscannot occurwithin the SONETnetworkbecauseinternalnodesrateadjustthe VT1.5 payloadswith pointeradjustments.-

El Mappinge usedfor DSls. El signalsaremappedintoVT2 signalswith thesamebasicprocedures of four columnsof bytesin an As shownin Figure8.22,the VT2 signalis composed STS-l thatproducea totalof 144bytes.After removingtheVl, V2, V3, andV4 bytes, mappedEIs andbytesyntheVT2 payloadhas140bytes.Formatsfor asynchronously chronouslymappedEls areshownin Figure8.23.Noticethat thebyte-synchronous signalingin slot 16-the mappingfor a 30-channelEl carrieschannel-associated form of out-slotsignalingdesignedinto El signalsat theirinception.The samebasic format supportscommon-channelsignaling,which is sometimesreferredto as a '

The original specifications for SONET included a locked timing format for VT-SPEs that eliminated the VT pointers so DSO channels could be identified directly within the STS-1. This mode of operation has since been abandoned,

422


I H{-l

4Columne

toooootr I Rorrs

noom0l

n tr ffin il E ffiI n H

gRUfYE

EoomdO

9RsrYs

)ffi|1

Ndcr:

eRorw

l) A VTz'|H fo|f oolumr o,fm $pE fu 144bytesitr r Jfi) rxe, speftfi€. 2) Vl, VA V3, V4 byta havefi xedlocd,iqs in r SPEi.ft'niffd by th€ lrd two tfrs of H4. 3) Vl sd V2 psid' to Vi wtid is the fir* byAaofihe f,o*hg Vfit $fB 4) SFEovcrfad b]ics V5, J2,26, Z7 ooon h iders,icnlrddilo porttime of m SpE . Figure 8.22

Superframe skucture for VTZ tributaries.

3l-channelEl format.In this casechannel16is theccs channelandchannelsl-15 and 17-31 arethebearerchannels.Thus,themultiplexmappingis not changed, just the nomenclature of the channelsandthe spE typedesignation in the vr pathoverheadbyte V5. 8.5.6 DS3 Payload Mapping Theprevioussectiondescribes severalalternatives for packingvirtualtributariesinto an sTS-l envelope.when all sevenvrs in an envelopearevrl.Ss, a total capacity of 28 DSls is provided-rhe sameas a DS3 signal.Thusonemethodof carryinga DS3 signalinvolvesdemultiplexingit into its consrituentDSI (or DSZ signals)and packing the constituentsas virtual fributaries.This approachis attractivein that the virtual tributariesare individually accessible for cross-connect or add-dropmultiplexersystems.If theapplicationdoesnot needto accesstheindividualtributaries.it is simplerto packtheDS3 signaldirectlyinro an sTS-1,asindicatedin Figureg.24. The payloadmappingin Figure8.24treatsthe DS3 signalsimplyasa 44.736-Mbps datastreamwith no impliedintemalstructure.Thus,this mappingprovidestransparent transportof DS3-ratedatastreams. Eachrow ofa nine-rowsPE envelopecontains87 x 8 = 696bits, whichcancarry 621or 622DS3databits dependingon thevalueof theC bits.Noticerhatthisformat

8.5 SONET/SDH 423

v5 R R R R R R R R 256| btts @6) R R R R R R R R R R R R R R R R

t** Saqf€s SByt€

c.Go o o o R R 258I blt$

(sa6)

R R R R R R R R R R R R R R R R

T SBytE€

C.GR R R R R q $tt I I I | | |

248| bltE (31)€) R R R R R R R R

R R R R R R R R R R R R R R R R

Chanreb l-31

R R R R R R R R R R R R R R R R E P ^ R R R R R R

Unus€dalot0 SaBytes

R R R R R R R R

R R R R R R R R

Chahnels1-31

R R R R R R R R Unus€dsk*O

c.Go o o o R R 256 | blte (ffi|

\,8

H B R R R R R R UnusedelotO

4Bytee

Chanrtelst.91 R R R R R R R R R R R R R R R R R R R R R R R R

Unu*dstoto flq€B

lgE-

Channcls 1+1 R R R R R R R R

1/0 B!/t68 No.offixed informdim btc : 4t256) -l - 1023 $r, Sr - Strffinfqfidiqr bits Cr, Q = Stuffhrgodrol bit$ R bits arcfillcd friffbiu (uspecificd) O bits arer€Bffvedfm finm PuPocee

(a)

CA$bitgindund 16 Pl, Po: SiEnrlinEFa$Ebdiodor bits R biteffcfixed *uffbttc (lE+€Eifiod)

(b)

(b)bytesynchronous' in VT2SPE;(a)synchronous; Figure8.23 El mappings hasfive C bits,whichallowsfor singleanddoublebit errorcorrection.Thepathoverhead(POH)bytescarrythe9 bytesof POH asdefinedin Section8.5.3. of theminimumandmaximumDS3ratesis left asanexerciseat the Determination endof thechapter. 8.5.7 E4 Payload Mapping one exampleof a soNET supenatemappingis shownin Figure8.25for a 139.264Mbps fourth-levelITU-T signal(E4). This signalis packedinto a 155-52-Mbps payloadenvelope STS-3c(or STM-I) signal.Figure8.25showsonly thesynchronous

424

FIBEROPTICTRANSMISSIONSYSTEMS

87(8) -096 Bit$

POH: R; C: l: S; O;

POH

8 R

1

c

I

R

1 1 1 1 1 111 CCRBRRRR

R

ccRRooRs I

POH

R

c

I

R

CCRRRRRR

R

ccRRooRs I

I

l 11rr11

Path overheadhê I Fixed etuff bit (realed) 69 Stutf controt bit s lnformstion (paytoad) bit 621 Stuff opportunity bit l Ovsrheadcommunications bit 2 Totalbie 6S

NotF: $ contains information when CCCCC - 00000. S is a stuff bit wh€n CCCCC= fl 111. Figure 8.24 Asynchronous 44.736-Mbps (DS3) payload mapping.

l+87(3)=261

x

121

1

POH

I

FOH

Bytes Y

r2l

Y

t2l

Y

121

131

x

121

Y

12r

z

tzl

131

x

t2l

Y

121

z

121

\

POH: Prth ovrhead bVte t : Information (payload) byte X: CRRRRROO Y; R R R R R R R R Z: iiiiiisR Toral bytec

l 241 5 13 1 IEI

C: Stuft control blt R: Fix€d stuff bit

o;

Overheadcommunlcationsbir S; Stuff opportunity bit i: Infomation bit

Noto: S contslne informstion when CCCCC:00000. $ ie s atuff bit whon CCCCC* t1 tlt. Figure 8.25 Asynchronous 139.264-Mbps (E4) payload mapping.

B,s soNET/sDu 425 (SPE-3c),not the 9 bytesof sectionandline overheadin eachrow. Noticethat there is only onecolumnof POH within the SPE-3cenvelope.ThePOH bytescarrythe 9 bytesof overheadasdefinedin Section8.5.3. signala$a transpar* Thepayloadmappingin Figure8.25treatsthe 139.264-Mbps entdatasfteamwith no impliedinternalstructure.Eachrow of a nine-rowSPE-3cenvelopecontains87 x 3 =261 bytes,whichcancarry 1934or 1935databits depending on the valueof the C bits. Noticethatthis formatalsohasfive C bits, which allows for singleanddoublebit errorcolrection.

8.5.8 SONETOpticalStandards "mid-span-meet" of $ONET equipThe optical interfacestandard[36Jdefinedfor of the fibers.Generation mentallowsfor eitherNRZ or RZ line codeson single-mode showninFigure8.26.The OC-N signalfromtheSTS-Nsignalrequiresa scrambleras to eachSTS-Nframeby presettingthe shift registerto all scrambleris synchronized I's immediatelyafter transmittingthe lastCl byte of the STS-Nsectionoverhead. A miniThus,the framecodes(A1,A?) andSTS-I ID (Cl) codearenot scrambled. mum level of timing contentis assuredby the Al, A2, andCl bytesalongwith the with thescramblersestaticoverheadbits of theSTS-Nframethatareanticoincident quence.Because is presetatthe$amepointof everyframe,everybit pothescrambler sition in successive framesexperiencethe samescramblervalue.Thus,when static overheadis "exclusiveored" with the scrambler,the samedatavaluesarise.(The scramblermerelyconvertsfixed overheaddatainto a different patternof fixed data.) The BER objectiveis I x l0-I0 fbr opticalsectionsof 40 km or less.Equipment for applicationswith discan be freely interchanged from separatemanufacturers (Transmitters and mayrequirejoint engineering. tancesup to 25 km. Longerdistances receiversfrom separatesuppliersmust be jointly specifiedto supportlonger distancefr.) at 1310nm with SONETsystemsarespecifiedto operatewith centralwavelengths sMF fibersor at 1550nm with Ds-sMF fibers.operationat l3l0 nm with Ds-sMF fibersor at 1550nm with SMFfibersis not disallowedbutmustbejoinfly engineered. A rangeof laserwavelengthtolerancesand maximumallowablespectralwidthsis DEIEIn

Figure t.26 SONETscrambler.

4?;6

FIBEHOPTICTRANSMISSIONSYSTEMS

TABLE8.10 FepreeentativeMaximumSpectratWidthsof SONETSources",0 SONETSource

oc-1 oc-3 oc-9 oc-12

Bate (Mbp$)

<25km

51.84 155.5? 466.56 622.08

30nm 1Snm 1Onm Bnm

<40km 25.0 nm 1 0 . 0n m 7.5 nm 6.0 nm

eThe spectral width of a $ource is determinedas the wavelBngthdif{erencebetweenth6 p6ak mod€ and the farthBstmode thet iB 10 dB b6low the peak. DThese specificationsare for lasBrsop€ratingwithin 10 nm of th6 c€ntral wav€lBngths(1310 end 1550 nm). Lasef,swith greater deviation lrom the c€ntral wavelength - are allowed but haG nariower sp€crratwldth sp€cificationsto compensatetor or€aterfib€r di8perslon.

specifiedfor both 1310and1550nm.Table8.l0 providesrepresentative valuesof the specifications. 8.5.9 SONET Networks A basicblock diagramof a soNET networkis shownin Figureg.27.Gatewaynetwork elements(GNEs)provideinterfacesto external(asynchronous) digital signals. Thesesignalsare mapped(synchronized) and unmapped(desynchronized) by the gatewayusingtheappropriate mappingformar.At this point only bit stuffingis used to synchronize theasynchronous tributariesto SONET.No pointera-djustments occur in theGNE.As thesTS-Nsignalspropagate tluoughthenetwork,pointeradjustmenrs in pointerprocessing(PP) interfacesnury be appliedat internalnetworkelements (NEs),but thelower level interfacemappingsthatoccurat theGNEsareuntouched. If a particularNE accesses vr payloads,vr payloadsin rhesamevr grouprharpass throughthenodemayexperience vT pointeradjustments. otherwise,vlr pointeradjustmentsdo not occur(onlythesTS-l levelsignalsarerateadjusred). Thefottowing paragraphs summarize pointerprocessing aspectsof a SONETnetwork: I' Pointerjustificationevents(pJEs)neveroccurin an originatingGNE. 2. A desynchronizerexperiencescontinuouspJEs only as a result of a synchronization differencebetweenthe originatingGNE and the terminating

Figure t.27 soNET network elements:s, synchronizer;pp, pointer processor;D, desynchronizer,

8.5 soNEr/sDn 427 at intemal nodesof a SONET GNE.* SynchronizationdifferenceVfailures pointer but thesegetremovedwhen adjustments, networkproducecontinuous to the sourceGNE. the SPEpassesthrougha nodethatis synchronized 3. PJEburstsoccurfor two possiblereasons.The first is a resultof a reference of a node'slocalclockto alignit with phaseadjustment switchanda subsequent Burstscanalsooccurasa resultof clocknoise the phaseof the newreference. In orderfor pointeradjustments. in multiplenodesproducingnear-simultaneous gateway, all of the propagate a desynchronizing to these to adjustments all of This can only threshold. path appropriate be must at the elastic storesin the produced behavior some abnormal happenif the sourceGNE haspreviously a ratherlargeamountof wander. suchasa lossof a referenceor sustained 4. A pointeradjustmentat the SPElevel doesnot affecta VT signalunlessit is haPpens to theVT andthatparticularadjustment passedto a nodethataccesses VT the when this level. occurs, the Even movement VT at causea pointer pointeradjustmentmustpassthroughthe network(withoutabsorption)to the gatewayto affectthe outgoingtributarysignal.On average' desynchronizing oneof every30 PJEsat the STS-1levelproducesa PJEat theVTl.5 level. circuit (PP)depictFigure8.28showsa blockdiagramof anSPEsynchronization (desynchronizes) theSPEpayexhacts processing: one half ing two halvesof pointer to the localSTS-I the SPE loadfrom a receivedsignalandtheotherhalf synchronizes payload from the redata frame rate. The RX pointer processingblock extractsthe blockmoniit to theelasticstore.TheTX pointerprocessing ceivedsignalandpasses tors the fill level of the elastic store and makespointer adjustmentsto maintain a nominallevelof storage.The sizeof theelasticstoreonly needsto be on theorderof 8 bytesin length,not a full frame.Theability to usea relativelysmallelasticstore(as network)is oneof thefeacomparedto frame-lengthelasticstoresin theasynchronous The payloadsare allowedto architecture: synchronization turesof a pointer-based float with respectto theSTS-l frameboundaries' Frequency of Palnter Justlllcation Events to a commonprithatis traceable If all NEsof a SONETislandusea timingreference clock source(PRS),PJEsoccuronly asa resultof distribution-induced maryreference frequencyoffset.Thus,whenall NEs aresynchrowanderthatproducesno sustained PJEsoccurat randomtimesandhaveequalnumbersof nizedto the samereference, positiveandnegativevaluesoverthelongrun. ContinuousPJEsoccuronly whenthereis a referencefailureat someNEs within a SONETislandor theislandis intentionallydesignedto operatein a plesiochronous mode.If thereferencefailureoccursat someinternalnodeof the SONETisland,the resultingPJHsareremovedat the next nodein the paththat is still lockedto the same at a GNE mustdeal referenceas the gatewayNE. Thus,a tributarydesynchronizef *This

statement assumesthat the terminating GNE synchronizes incoming SPEs to a local clock before they urrive at the desynchronizer.

428

FIBEHoPTIcTRANSMIsSIoN SYSTEMS Odrr SPEe

Ovtrterd datr

Flgure t.28 Block diagramof SPEsynchronizing equipment:RX, receiver;TX, transmitter. with continuous PJEs only when there is a referencefailure in either the originating GNE or in the rerminating GNE.

SONETDetynchronizers soNETdesynchronizers arenecessarily designed withverylowclockrecovery band-

widths to smooththe effects of (l) isolatedpointer adjustments,(2) continuouspointer adjustments,(3) pointer adjustment bursts, or (4) combinations of the latter two. A pointer burst is defined as the occurrenceof multiple pointer adjustmentsof one polarity occurring within the decaytime of the desynchronizercircuit (i.e., the reciprocal of the desynchronizerclosed-loop PLL bandwidth). Thus. it is ironic rhat as the clock recovery bandwidth is narrowed to smooth the effect of a burst, the probability of a burst occurrenceis increased(by definition only). Extremely nrurow pLL bandwidths are easiestto implement using digital filtering techniquescommonly referredto as bit leaking. Bit leaking is essentiallya mechanismfor converting byte-sizedpointer adjustments into bit- (or fractional-bit-) sized timing adjustmenrs. Figure 8.29 shows a block diagram of a microprocessor-controlledDSl desynchronizer. The microprocessoris used to perform long-term averaging ofphase ad-

DSI ertrrctlon

Ml3 multlplerer

Figure 8.29 VTl.5 desynchronizer hardwarefunctionalcomponents.

8.6 SONETRINGS

425

justmentsin lieu of dedicatedlogic thatrequireslargecountersandwide word sizes is to deterfor low-bandwidthDSPfiltering.Thefirst functionof thernicroprocessor by all frequencyadjustmine theaverageDSI payloadfrequencyoffsetrepresented After andsTs pointeradjustments). mentevents(bit stuffs,vT pointeradjustments, a stuff ratio valueis calculatedthat the averagefrequencyadjustmentis determined, in theM13 allowsinsertioninto a DS3signalasshown.(TheM12 stageis embedded in the multiplexer.)The elasticstorefill level is usedfor very long term adjustments andfor outputfrequencythatarisefrom finite precisionlimits of theDSPcalculations to the variationsin theDS3clock,whichis typicallynotsynchronized accommodating line clock. SONET

8.6 SONETHINGS of largeswitchingmaAs hasbeenmentionedearlierin this book,the development large sectionshasimpacted cross with extremely systems transmission chinesand leva trend fewerhierarchical toward with architectures network telecommunications on the operadependence is increased of this trend consequence els.An undesirable tional statusof individual switchingmachinesand fiansmissionpaths.A SONET thatspecifiring, or moresimplya SONETring,is a networkarchitecture self-healing rings areshown networksurvivability.Two basictypesof self-healing cally addresses in Figure8.30:a unidirectionalring anda bidirectionalring. Themaindifferencebetweenthetwo typesof ringsis how thetwo directionsof a duplexconnectionaree$tablished. to bothhalves ring a singletimeslotof theentirering is assigned In a unidirectional of a connection.As indicatedin Figure8.304,traffic is normallycarriedonly on the pathusedfor protection.In the (unidirectional) workingpathwith thecounterrotating directlyfrom A to B, but the (out be carried of an might OC-48) an STS-1 example, C and D to A. A bidirectionalring, B through from be carried would returningSTS-1 overthe shortest duplex connection ofthe both halves on the otherhand,establishes pure fiber andanotherasa working as a identified pathin the ring. Thus,no fiber is trip delaysfor provide round shorter pureprotectionfiber.Because bidirectionalrings prefened modeof it andallowreuseof time slotson thering, is the mostconnections operationfor interofficenetworks.Rings for subscriberaccessapplicationsdo not carry muchtraffic betweenADM nodesand thereforearemore suitedto a unidirectionalmodeof operation. 8.6.1 Unidirectlonal Path-Switched RIng ring (UPSR)[38] transmits As shownin Figure8.31,a unidirectionalpath-switched the sameinformationfrom A to B in both directionsaroundthe ring. Normally, only by thereceivingnode;If a failureoccurs,a nodecan$etheworkingpathis accessed lectthedataon theprotectionchannel.Noticethatin theexampleshownselectionof theprotectionpathactuallyleadsto a shorterpathfor theconnectionfrom A to B'

430

FIBEHoPTIcTRANSMISSIoN SY$TEMS

Prot€c{ionpsth

(*)

(b)

Figure 8.30 (a) Unidirectionaland(b) bidirectionalrings.

8.6.2 Bldirectlonat Line-$witched Ring Bellcoredefinesrwo versionsof bidirectionalline-switchedrings (BLSRs) I39l: a two-fiberBLSR anda four-fiberBLSR.on a two-fiberBLSRprotecrionis provided by reservingbandwidthin eachof two counterrotating fiber paths(Figureg.32).If all trafficis to beprotected, only 507oofthe totalsystemcapacitycanbeused.Undernormal conditionsconnections betweentwo nodesutilize the shortestpathbetweenthe nodes'If a fault in eitherdirectionof hansmissionoccurs,the nodesadjacentto the faultperformring switchesasindicated.A ring switchinvolvesswitchingtrafficfrom working channelsof thefailed facility to sparechannelsof theotherfacility on the side of the nodeon which the fault occurs.The protection-swirched traffic propagate$all theway aroundthering,beingignoredby interveningnodes,until it is switchedback

8.6 SONETRINGS

431

ProtoctionPath

Wofting Path

PathTerminatlon

swirching. FigureE.31 UPSRprotection to theworkingchannelsby the othernodenextto thefault.Noticethatall nodes(including thenodesadjacentto the fault) communicateon working channelsin the same mannerasthey did beforethe protectionswitching.That is, the pathterminationsare not partof theprotectionpath.Themainimpactof theprotectionswitchis anincrease in delay for affectedtraffic (and a momentaryinsertionof exkaneousdatawhen the switchoccurs). On a four-fiberBLSR(Figure8.33)two pairsof fibersareprovidedfor eachdirection of kansmission-one bidirectionalworking pair and anotherpair for protection of the first pair. Thus,working andprotectionchannelsarecarriedon differentphysiarenormallysetup to usethe shortestdistanceof cal facilities.Again,connections

ion Switches

.if:-*-

West nodes

Esst nodes

West nodes

East nodes

Figure 8.32 Two-FiberBLSR protectionswitches

432


Protgdion

*\ \

Working Pair/

Prt""tion p"i,

RlngProtsdlon$rritches

\

wo*ing e"ii

Figure 8.33 Four-FiberBLSRprotectionswitches.Notethata ring switchanda spanswitch cannotcoexistwithoutchannelswitchinganda reductionin capacity, travel for each side of a connection.If a failure occurs on only a working facility, protection switching occurs similar to "span switching" of a point-to-poini system:The traffic is merely switched to and from the protection facility by nodesadjacent to the fault. However, if a fault affects both the working and the proiection facilities, a ring switch is neededas shown. Again, protection-switchedtraffic propagatesall the way around the ring without being accessedby intervening nodes.eil traffic accessesstill occur on the working channelseven though the sameinformation is passingthrough the nodesin the protection path. A four-fiber BLSR obviously requires more facilities rhar a two-fiber BLSR but has numerousadvantages.First, the protectedcapacity of the systemis twice as large. second, fiber failures on only the working pair can be accommodatedby a spanswitch with minimal disruption to traffic. Third, multiple separatefailures can tccur on working pairs and be accommodatedby multiple span switches.Fourth, the presence of a spare pair simplifies maintenancetesting and possible upgrading of facilities. For thesereasons,a four-fiber BLSR is generally favored.

REFERENCES F' P. Kapron,D. B. Keck, and R. D. Maurer,"RadiationLossesin Glassopticar Waveguides," Applied Physics l*rrers, Nov. I g70,pp. 4Z34ZS. T' Miya, Y. Terunuma,T. Hosaka,andT. Miyashita,"urtimateLow-LosssingleMode Fibreat 1.55pm," Electronicl*tter.r, Feb.1979,pp. 106-10g. D. Large,"The Star-BusNetwork:Fiber Opticsto the Home,"CED Magazine,Jan. r989. R. olshansky,v. A. Lanzisera,and p. M. Hil, "subcarrierMurtiplexedLightwave systemsfor BroadbandDistribution,"Journalof LightwaveTethnilogy, sep-t.19g9, pp. 1329-1342.

REFERENoES430 An HistoricalPerspectle,"IEEE 5 T. Li, ,'Advancesin optical Fibercommunications: 1983'pp. 356-372, Apr. in Communicatfons, Areas Joumalon Selected "Lightwave Dec' 1985'pp' himet" IEEElourrutl of QuantumElectronics' 6 P. S.Henry, 1862-1877. 7 J. R. Stauffer, ',FT3C-A Lightwave system for Meffopolitan and Intercity Apr. 1983'pp. Applications,"IEEE Journal on SelectedAreasin Communications, 413-419. "Introductionto LightwaveSyst'ems," in 8 P. S. Henry,R. A. Linke, andA. H. Gnauck, l l, S.Mitler andI. P. Kaminow;Eds.,AcademicPress,San OpticalFiberTransmission Diego,1988,pp.781-832. "TenesnialIntercityTransmissionSystems,"in Optical 9 D. C, Gloge and I. Jacobs, II, s, Miller andI. P. Kaminow,Eds.,Academichess, san Diego, Fiber Transmission 1988. "High Capacity Lightwave TechnologyComes of Age," AT&T l0 D. L. Howells, No.4, 1988,pp.l0*15' Technology,Vol.3, "FT SeriesG Lightwave 1l W. C. Mara, J. S. Linnell, N. M. Denkin,and K. Ogawa, Digital Transmission system Architecture and upgrade capability," IEEE ions,1987,pp. I 9B'3.I - I 98.3'5. InternationalConferenuon Communicat "Analysisof Mode PartitionNoisein LaserTransmission Systems,*IEEE l2 K. Ogawa, pp' 1982, 849-855. May Electronrcs, of Joumal Quantum ..An OpticalRepeaterwith High-Impedance InputAmplifrer,"Bell,$ystem 13 J. E. Goell. pp. 629-&3' TechnicalJoumal,Apr. 19?4, "mBlC CodeandIts Performance in an Optical 14 N. Yoshikai,K. I. Katagiri,andT. Ito, communicationssystem,"IEEE Transactionson,communications,Feb. 1984,pp. 163-168. "High-speedcMI optical Intraoffice 15 K. Hagishima,Y. Kobayashi,and K. Aida, Areasin IEEE Joumal on Selected System:DesignandPerformance," Transmission 1457' pp. 145 I Dec. 1986, Comtnunications, 16 F. Ross."Overviewof FDDI: The FiberDistributedDataInterface,"IEEE .lournalon Sept.1989'pp. 1043-1051. in Communications, SelectedAreas R. Dawes, "A High-Reliability565 Mbit/s Trunk and Cochrane, R. Brooks, l7 P. Dec. TransmissionSystem,"IEEE Joumal on SelectedAreasin Communicatiorts, 1986,pp. 1396-1403. "NLl Submarine System,"/EEEfoumal on Selected l8 R. L. WilliamsonandM. Chown, Apr. 1983'pp' 454-458' Areasin Communications, "Byte-orientedDc Balanced(0.4) 88/108 19 P. A, Franaszekand A. x. widmer, Dec.4' 1984. Code,"U.S.Patent4,488,'739, PartitionedBlock Transmission and.I/OforComputerNetworks, 20 A. F. Benner,Fibre ChannelGigabitCommunications 1996. McGraw-Hill,New York, "The First TranspacificOpticalFiberSubmarine 21 Y, Niro, Y. Ejiri, andH. Yamamoto, June1989'pp. on Communicafians, CableSystem,"IEEE IntemationalConference 50.1.r-50,1.5. "Line CodeandTerminalConfigurationfbr 22 N. Yoshikai.S. Nishi. andJ. L Yamada, System,"IEEE Journalon SelecteilAreas OpticalTransmission Very Large-Capacity pp. 1432-1437. Dec. 1986, in Communicafians,

434

FTBEH oplc rRANSMtsstoN sysrEMs

23 H. Kimura and K. Nakagawa,"F-1.6Gsystemoverview," Reviewof the Electrical Communications LaboratoriesNTTJapan,Vol. 35,No. 3, I 9g7, pp. ZIV_ZZS. 24 s' s. wagner and H. Kobrinski, "wDM Apprications in Broadband Telecommunications Networks,"IEEE communicationsMagazine,Mar. 19g9,pp. 22_30. 25 R. J. sanferrare,"TerrestrialLightwavesystems,"Ar& T TechnicalJournal, Jan7qeb. 1987. 26 S. S. Wagner,H. L, Lemberg,H. Kobrinski,L. S. Smoot,andT. J. Robe,,.A passive PhotonicLnop ArchitectureEmployingwavelength-DivisionMultiplexing,',IEEE Globecom Proceedings, l gSS,pp.48.l, t -48.1.5. "[I-v 27 N. K. Dutra, Device Technologiesfor Lightwave Applications,l'AT&z TechnicalJoumal,Jan./Feb.1989,pp. 5-l g. 28 P. R. TrischittaandD. T. s. chen, "Repeaterless underseaLightwavesystems,- IEEE Communications Magafine,Mar. I 989,pp. I 6_2l. 29 s. c. Mettlerandc. M, Miller, "optical Fibersplicing,,,in optical Fiber Transmission I I ' s' Mller andI. P. Kaminow,Eds.,Academichess, sanDiego,r 9gg,pp.263-300. 30 J. B, HaberandJ. W. Rogers,..Fiberin rheLoop," AT&T Technology, Vol. 3, No. 4, 1988,pp.2-9. 31 N. Yamaguchi,K. Yarnane,and T. Kihara,"565-Mbit/soptical Fiber Transmission System,"Fujitsu Science and Technology Journal,Mar. 19g9,pp. 60_63. 32 "SynchronousOptical Network (SONET)-Basic DescriptionIncluding Multiplex Structure,RatesandFormats,"Tl.l05-lgg5, ANSI, New york, 1995. 33 "synchronousoptical Network (soNET)-payload Mappings,"T1,105.02-1995, ANSI, New York. 1995. 34 Recommendation G.707, "Network Node Interfacefor the synchronousDigital Hierarchy(SDH),"ITU-T, Geneva,Switzerland,Mar. 19g6. 35 N. B. sandesara, T. H. Jones,and A. G. Edwards,..soNET Intra-officeInterconnecr Signal,"IEEE Globecomp roceedings, I 9gg, pp. 30.5.I _30.5.7. 36 "synchronousoptical Network (soNET) Transportsystems:common Generic Criteria,"GR-ZS3-Core, Bellcore,Morristown,NJ. Dec. 1995. 37 N. Zhang, K,-H. Liu, and E. c. posner,"Reriablepayroadpointer protocol for SynchronousOptical Network," IEEE InternationalCommunicatittnsConference Proceedings, I 989,pp. l4.Z.l-l 4.2.j. "soNET 38 Dual-Fedundirectionalpath swirchedRing (upsR) EquipmentGenedc Criteria,"GR-1400-Core, Bellcore,Jan.1999. 39 "soNET Bidirectional Line-swirched Ring Equipment Genenc criteria," GR-l230-Core, Bellcore,Dec.1993.

PROBLEMS 8.1 Determine tle attenuationg, in decibelsper kilometer such that the loss limit of a Sfi)-Mbps fiber system is exactly equal to the dispersion limit. Assume the transmitteroutput is 30 dB abovethe receiverthresholdfor the desirederror rate and that the systemhas a BDp of g0 Gbps-km.

PROBLEMS 435

multimode,So-Mbpsfiber systemwith 2 dB/kmloss 8.2 An 850-nm,graded-index, buslocalareanetwork.Assumethe in thefiber is to beusedfor a token-passing pair. If systemusesthe first entriesin Tables8.1 and 8.2 asa source-detector of loss passive with 0.4 dB many taps theBDPof thefiberis 500Mbps-km,how transmitters between canbeinsertedperkilometerwithoutaffectingthedistance andreceivers. 8.3 Whatis thedistancelimit (withouttaps)of thesystemin hoblem 8.2if thedata rateis reducedto l0 MbPs? 8.4 A graded,multimodefiber exhibits 100psec/kmnm of chromaticdispersionat 850nm. Determinethe BDP of a NRZ systemusingan LED having40 nm of specrralwidth. 8.5 A WDM SMF systemusing 1300and 1550nm is dispersionlimited at both be related? How mustthe spectralbandwidthsof the two $ources wavelengths. = psec/kmnm at 1550nm') (AssumeD= 3.5pseclkmnm at 1300nmandD 17'5 and mBlC line 8.6 What is the minimum and maximumpulsedensityof mBlp codes? 8.7 What is the longeststringof no pulsesthat canoccurin mBlP andmBlC line codes? 8.8 A 486B line codecanencodeall4 bitsof sourcedatawith 6-bit line codeshavcodesare left ing exactlythreepulsesin everycode.How manydc-balanced overfor link control? mapped 8.9 Determinetheminimumandmaximumdatarateof an asynchronously DSI signalthatcanbe caried within a VTI-5 envelope. mapped 8.10 Determinetheminimumandmaximumdataratesof anasynchronously El signalthatcanbe carriedwithin a VT2 envelope. DS3pay8.11 Determinetheminimumandmaximumdataratesof anasynchronous loadthatcanbe carriedwithin an STS-1envelope. 8.12 Determinethe minimumandmaximumB[ dataratesthat canbe accommodated within an STS-3cenvelope. 8.13 Determinethe averageframeacquisitiontime of an STS-1signalassumingall nonframingbits haveequallylikely randomvaluesof 0 or 1' 8.14 RepeatProblemL13 for an STS-3csignal' 8.15 Determinethepayloadcapacityof a STS-3c(STM-1)signal.

DIGITALMOBILETELEPHONY capacityby dividing A basicconceptofa cellularsystemis to provideever-increasing thecell frequencyreuse.Unfortunately, cellsinto smallerandsmallersizesto increase divisionconcepthasprovento be impracticalin termsof findingsuitablelocationsfor basestationantennasandfor gettingrepeatedconstructionauthorizationsfrom govin theearly1990s Theexplosivedemandfor mobiletelephones erningorganizations. thedevelopment helped stirnulate world in elsewhere the States the and United within havebeenpurapproaches Two basic the demand. to of new Sy$tems accommodate new freand allocating of $y$tems existing capacity channel the sued:expanding quencybandsto cellularmobilephoneservice.All of thenew systemsutilize digital usedby analogcellularsystems in lieu of theanalogFDM transmission transmission andNMT in scandinavia. Britain, TAcs America, in Great in North AMPS suchas theavailabilityof loware enabledby systems cellular digital viable Commercially to problemsin several provide solutions processing to technology signal costdigital into a low enough signal speech a digital to compress necessary it is key areas.First, penalty with respect a spectrum not impose does bit ratethatdigitalvoicetransmission haveadalgorithms compression speech 3, in Chapter As discussed to analogsystem$. efspectrum provide cases, can, point in some thatdigitization vancedto the Sreater to a transmission of digital the application Second, ficiency than analogsy$tems. of dyeffects the to overcome equalizer a sophisticated mobileenvironmentrequires namic multipathtransmissionimpairment$.Third, the susceptibilityof the speech compressionalgorithmsto channelerrorsrequiressophisticatederror conection and Lastly,the low-bit-ratevoicecodersandthe digital systemarchicontrolmeasures. tecturesintroducesignificantartificialdelayinto thevoicechannel,whichimposesthe needfor echocancelersfor acceptablevoice quality'

9.1 NORTHAMERICANDIGITALCELLULAR NorthAmericanDigital Cellular(NADC),alsoknownasUS Digital Cellular(USDC) a digital upgradefrom the previouslydeor Digital-AMPS(D-AMPS),repre$ents to as AdvancedMobile Phone Service referred ployed analog cellular system 437

438

DIGITAL MoBILETELEPHoNY

(AMPS).The D-AMPS sysremis designedto be compatiblewith AMps. In fact. a D-AMPS $ysteminstallationcancoexistwith an AMPS installation,thusallowinga gracefulmigrationfrom an all-analogserviceto an all-digitalservice.An analog_todigital migrationis supportedby a dual-modephonethat can operateas an AMps phonein onecall andasa D-AMPSphoneon thenextcall. D-AMpS is standardized by EIA/TIA asInterimStandards IS-54and15-136. 9.1.1 D.AMPSTransmlssion Format The most significanraspectof maintainingcompatibilitywith AMps is the needto adhereto theAMPSFDM channelstructure.Thischannelstructureuses30-kHz-wide channelsin therange824-8g4MHz.within each30-kHzFDM channelIS-s4defines six digitalchannelsoperatingin a timedivisionmultiple-access (TDMA) modeof op_ eration,asshownin Figure9.1.Transmission from a basestationto themobilesis accomplishedwith a continuousTDM streamwith six time slots.Transmission from eachof themobilesoccursin databurststhataretimedto arriveat thebasestationin separate, nonoverlapping time slotssynchronized to theoutgoingtime slots.Associ_ atedwith eachburstfrom a mobileis a guardtime to preventoverlapandprovidea traxsmitterramp-upprecedingthe data.The guardtime betweentime slotsis minimizedby adjustingthetransmittime of the mobileswith controlmessages from the basestation.Theseadjustments aredynamicto accommodate mobility. The TDMA digital transmission frameformat within each30-kHzchannelcontainssix time slotswith a total of l9zl4 bits. The repetitionrateof eachframeis 25 framesper second,whichleadsto an aggregate bit rateof 49.6kbpsin the30 kHz of bandwidth.The modulationformat is r/4 shifted,differentially encoded,quadrarure phaseshift keying.This formatis essentially4-psK modulationwith two four-point constellations offsetfrom eachotherby nl4 radians.By alternatingbetweenconstellations,a symboltransitionof at leastn/4 radiansis alwaysassured-a propertythat helpsin symbolclockrecovery. Full-ratevoicecodingutilizestwo time slotsin eachframefor the voiceinformation. Thus,the systemcapacitywith full-ratevoicecodingis threetimesthat of an AMPS systemsincetherearethreeTDM voicechannelswithin eachFDM channel.

Baee

station*-

*-Fig're 9.1

,

fJ

ff

|

ilrobiles

r

r-f

f-J

f-r

+-

l+E

TDMA transmission format of NorJh American dieital cellular.

CELLULAR 439 OIGITAL 9.1 NORTHAMERICAN

If half-ratevoicecodinggetsimplementedin the future,the capacityexpansionwill be sixfold. 9.1.2 D-AMFS$peech Godlng Thespeech-coding algorithmis vectorsumexcitedlinearpredictive(VSELP)coding the is in Chapter3. The D-AMPS VSELP algorithmprocesses described which [], of 20 msecduration.Eachspeechsegmentis represpeechwaveformin segments senredby 159bits.Sincetherearetwo VSELPframesin eachTDMA frame,theraw datarateof the voiceis 2 x 159x 25 = 7950bps.To the raw bit dateis added5050 bpsofredundancyencodingfor errorcorrectionanddetectionto producea composite, datarateof 13 kbpsfor a voicechannel.As shownin Figure9.2, thereare aggregate the 260 datachannelbits and64 overheadbits in eachtime slot.Table9.1 describes basicuseof eachdatafield within thetime slots, 9.1.3 D-AMPSControlGhannel thedigitalnatureof D-AMPS In additionto providinga threefoldcapacityexpansion, providesotheradvantages thatarenotpossibleor atleastdifficult to achievein ananalog systemsuchasAMPS.The first of theseis useof theCDVCC channelto ensure that a basestationmaintainsconnectionswith intendedmobiles.AMPS utilizes a similarfeaturewith supervisoryaudiotones(SATs).A SAT is a toneat 5970,6000' or 6030Hz thatis insertedandremovedfrom theaudiosignalspecificallyfor detectconnectionintegrity.The availabilityof only threetones ing fadesand ascertaining and the complexityof inserting,detecting,andrepeatingthesetonesare significalt limitationsof AMPS. providedby thedigitalnatureof D-AMPSinvolvesthe A moresignificantadvance thischannelis always in eachtime slot.Because useof theSACCHchannelembedded present,it is quite useful for communicatingconkol and supervisioninformation usesofthis contol chanwhile speechis activelyin progress.Specific,advantageous

rs0

t2

DATA

HFVD

It0

sYl{c

EACG}I

DATA

CDYCC

Brrt thtlon to Mobl|t

tt

n

F G DATA 8YI{G

t2 DATA

SACCH

12

eDvcg

lfr DATA

MobllotoBrrt sftd|on f igure9.2 Timeslotformatof NorthAmericandigitalcellular.

440

DtctrALMoBtLE rELEpHoNy

TABLE9.1 DataFietdFunctionsof D-AMPSTImeStots CDVCC DATA

G R BSVD SACCH SYNC

codeddigilalv9rifhationcolorcode.A uniquecodesentby a basestationand returnedby eachmobilefor basestationconlirmation of connection intogrity. Application bearerchannelbit$(voiceor data).can alsobe usedfor a fast associated controlchanne_l (FAcc) whenthereis no activeapplication or a situation ariseswhenapplication transmission needsto be usurped. Guardtime.Mobiletransmitter is off. Ramptim€.Mobiletransmitter rampsup to assigned powerlevel. (unused). Reserved slowassociated controlchannel. A continuous channelusedto sendconhol andsuperuisory information. synchronization channel. Usedfor synchronization, equalizer training, andtime slotidentification.

nel involve authentication, additional connection integrity, transmit power control, channelquality mea$urementreports,handoffs to a new cell, keypad depressions,and calling number identification. The SACCH control channelis also usedfor timing adjustments specific to the TDMA operation. A particular example of the usefulnessof the SACCH channelis its supportof mobile assistedhand off (MAHO). If a mobile with an establishedconnection moves from one cell to another,the processof handing the mobile off is performed with much more control and reliability than is possible in AMps. The MAHO processbegins by the base station telling the mobile to make channel quality mea$urementson the current chatrneland on candidatechannelsfor a potential handoff. Channel quality measurementsinvolve received signal power levels and bit error rates(BERs). The TDMA nature of D-AMPS facilitates measurementsof candidatechannelsby tuning to the candidatefrequency during an inactive time slot. After each set of measurementsthe mobile sendsthe results to the basestation (via SACCH) whereuponthe basestation can determine if a handoff is justified.

9.1.4 D.AMPSErrorControl There are three mechanismsincorporated into D-AMPS for mitigating the effects of channelserrors: elror correction, error detection,and interleaving. Enor correction is implemented with a half-rate convolution coder for the perceptually most significant bits of the voice. There are 77 such bits out of a frame size of 159 bits. The half-rate convolution coding processthereforeadds77 more bits to the channel.Of the 77 bits, 12 are particularly important. A 7-bit CRC check sum is addedfor rhesebits to determine if any of these 12 bits are received in error. when a cRC enor occur$.certain critical parametersfrom the previous error-free frame are used to reconsg.uctspeech to avoid use of aberrant values. If several CRC errors are received in successive frames,the reconstructedspeechis muted. The generalterm for theseoperationsis bad frame masking. The third error control mechanism involves separatingthe data in a single speech frame, interleaving it with datafrom adjacentspeechframes,and transmitting it in two

sysTEMFoRMoBtLE s.a GLoBAL coMMUNtcATloNS441 time slots. This processreducesthe possibility that a burst of errors will circumvent the error corection capabilitiesofthe convolutional coding. A drawback to interleaving is the delay it addsto the channel,which must be accountedfor in echo cancelers.

9.2 GLOBAL SYSTEMFOR MOBILECOMMUNICATION$ (GSM) [2] is a cellularmobilecommuGlobalSystemfor Mobile Communications standardized by theEuropeanTeleconrmunicationssystemdeveloped in Europeafld (ETSI). beenadoptedworldwide Institute GSM hassubsequently nicationStandards beon Initial work GSM standardization astheintemationaldigitalmobilestandard. occuffed in 1991. ganin 1982.Thefirst field trial of a GSM system Union (ITU) allocatedfrequencyspectrum The InternationalTelecommunication (base stationto mobile)and890-915MHz for the at 935-960MHz for thedownlink uplink (mobileto basestation).Eventhoughsomeof thisspectrumwa$beingusedby established analogsystems,therewasno attemptto be compatiblewith the existing frequencyplan. (The incumbentanalogsystemsin the variousEuropeancountries wereincompatiblewith eachotherandamaingoalof GSMwasto establisha coflrmon standardsobackwardcompatibilitywasnot a consideration.) 9.2.1 G$MChannelStructure with morefreeprovidedGSM systemdesigners Startingwith anopen-spectrumplan most significant The to designers. was available D-AMPS system than domin design the 200-kHz-wide digital RF use of to is with respect D-AMPS differencein GSM GSM Each RF chanD-AMPS RF channels. as to the channels opposed 30-kHz-wide in 6, mentioned Chapter As modulation. Mbps GMSK nel operatesat 270.833 using in as D-AMPS. modulation used to 4-PSK fairly related closely GMSK modulationis GMSK modulationdoes,however,requiremorebandwidttrthantightly filtered4-PSIE = 1.35bps/Hzfor GSM and asevidencedby an informationdensityof 270.8831200 = modulation formatof GSMprovides The GMSK D-AMPS. for 48.6/30 1.6?bps/Hz power generation thanis RF is more for that efficient RF signal a constant-envelope important efficiency is most by This D-AMPS. filtered modulation used tightly 4-PSK for hand-heldbatterylife.. As shownin Figure9.3,a GSM RF channelutilizesdigitalTDMA with eightfullby a single rate voicechannels,in contrastto threefull-rate voicechannelssuppor.ted on channels a single more TDMA D-AMPS RF channel.The ability to terminate (The of a 270'833stations. cost providesa costadvantage for GSM base transceiver kbps TDMA transceiveris no differentthan the cost of a 48.6-kbpsTDMA transceiver.) The GSM systemcarrieseightfull-ratevoicechannelsin 200 kHz of bandwidth, whichamountsto 25 kHz per voicechannel,a specffumefficiencyidenticalto Euro*As

discussed below, GSM uses Iess speech compression than does D-AMPS, which Ieads to a higher rate digital channel for voice and in tum requites more ftansmit power.

442

DIGITALMOBILETELEPHONY

Baee

|

+-

r-l

|

ff

.

G l{-

Mobilee

Figure 9.3 TDMA transmission formatof GSM. pean analog FDM systemsof the time. Thus, the introduction of the GSM systemdid not provide spectrumefficiency improvementsas D-AMps did. A GSM systemdoes, however, provide cellular systemefficiencies in that digital transmission,in general, and strong error correction, in particular, allow operation at lower signal-to-noiseratios' Greater noise or interference tolerance leads to longer transmission distances and/or greateramounts offrequency reuse. The burst period of a GSM systemis l?0/26/8 = 15126ms. This burst period is derived from a 120-ms superframe consisting of 26 TDMA frames and g bursts per TDMA frame' Twenty-four frames of a 26-friune super frame are allocated to traffic (e.g.' voice) transmission while one of the frames is allocated to a SACCH control channel for each traffic channel. The last TDMA frame of a superframe is reserved. A unique aspectof GSM, with respecrro D-AMps, is that a TDMA burst format is usedin both directions of transmission,as opposedto only on the uplink from the mobile to the basestation.The format of thesebursts is shown in Figuie 9.4, where it can be seenthat there are 148 bits ofdata and an idle guard time correspondingto the period of 8'25 bits. The burst transmissionrate of a taffrc channel can now be determined as 156.25/r5D6= 270.833kbps.The fields within the burst areidentified in Table 9.2. The use of multiple bursts in the downlink direction, as opposed to continuous ffansmission,is advantageousin that it inherently allows tuming off the base station transmitterduring idle channels,which in turn reducesthe total amount of interference betweencells in a widespreadand congestedinstallation. An advantageof continuous transmission,as used in D-AMPS, is the relative easeof implementation and greater performanceof the digital receiver in the mobile.

15/26=0.577msec

Figure 9.4 Time slotformatof GSM burst.

FORMOBILE 9,2 GLOBAL SYSTEM COMMUNICATIONS443 TABLE 9.2 Data Field Functlons of G$M Tlme Slot Flag Guard Tail TCH Train

A singlebitusedto signifyvoiceor FACCHcontentin an associated TCHfield timingmarginbetweenbursts ldleperiodof 8.25bitsinteruals 3 "0"bitsfor equalizer training Fieldfor transporting bearerdataor FACCHdata Fieldof fixeddatapatternusedto trainequalizers andacquirea dataclockfor theentireburst

9.2.2 GSM Speech Coding GSM usesregularpulseexcited-linearpredictivecodingwith a long-termpredictor in Chapter3. Speechis diloop (RPE-LTP)[3]. TheRPE-LTPalgorithmis described vided into 2O-msec samples,eachof which is encodedas260 bits, giving a total bit rateof 13kbps.This is the original,full-ratespeech-coding algorithm.An enhanced full-rate (EFR) speech-coding algorithm has been implementedby some North qualityusing AmericanGSM 1900operators. EFRis saidto provideimprovedspeech theexistingl3-kbpsbit rate. 9.2.3 GSM ChannelCoding and Modulation similarto D-AMPS.First of a]l. the 260 bits GSM utilizeserrorcontrolmechanisms of a speechframeatedividedinto threeclasses: ClassIa, 50 bits-most sensitiveto bit errors Classlb, 132bits-moderately sensitiveto bit errors ClassII, 78 bits-least sensitiveto bit errors the ClassIa bits havea 3-bit CRCaddedfor errordetection.If anerroris detected, versionof the previouscorrectlyreceived frameis replacedby a slighfly attenuated (a total of 189 frame.The50 ClassIa, 3 CRC, 132ClassIb, anda 4-bit tail sequence bits) areprocessed by a half-rateconvolutionalencoderfor errorcorrection.The output of theconvolutionalencoderis addedto the 78 ClassII bits to producean aggregatespeechframe of 456 bits. Thus, the redundantlyencodedspeechrate is = 22.8kbps. 456tO.OZ0 To furtherprotectagainstthebursterrors,eachsampleis interleaved. The456bits of outputby the convolutionalencoderaredividedinto eightblocksof 57 bits, and time slotbursts.Sincea timeslotburst theseblocksarespreadacrosseightconsecutive carriestwo 57-bitblocks,eachburstcontainsraf,fic from two differentspeechsamples. 9.2.4 G$M Mobile Station The GSM mobilestation(MS) consistsof themobileequipment(theterminal)anda smartcardcalledthe SubscriberIdentityModule(SIM). The SIM providespersonal of a spemobility,sothattheusercanhaveaccessto subscribed servicesirrespective cif,rcterminal.By insertingthe SIM cardinto anotherGSM terminal,theuseris able

444


to receivecallsat thatterminal,makecallsfrom thatterminal,andreceiveothersubscribedservices. The mobile equipmentis uniquelyidentifiedby the InternarionalMobile Equipment Identity 0MEI). The SIM card containsthe IntemationalMobile Subscriber Identity(IMSI) usedto identifythe subscriber to the system,a secrerkey for authentication,andotherinformation.The IMEI andthe IMSI areindependent, therebyallowing personalmobility. The sIM cardmay be protectedagainstunauthorized use by a passwordor personalidentitynumber. 9.2.5 GSM FrequencyHopplng Themobilestationis inherentlyfrequencyagile,meaningit canmovebetweena transmit, receive,andmonitortime slotwithin oneTDMA frame,all of whicharenormally on differentfrequencies. GSM makesuseof this inherentfrequencyagility to implementslow frequencyhopping-the mobileandBTS transmiteachTDMA frameon a differentcarderfrequency.The frequency-hopping algorithmis broadcaston the broadcast controlchannel.sincemultipathfadingis dependent on carier frequency, slow frequencyhoppinghelpsalleviatetheproblem.In addition,cochannelinterferenceis morerandomthanwhenfixed-frequency allocationsexist. 9.2.6 GSM Short Message $ervice ShortMessageService(SMS)is an integrated bidirectionalmessaging servicethatallows GSM cellularsubscribers, andvariousPCSofferings,to sendandreceivedata. Individualmessagerr (with GSM)canbeup ro 160bitsin length.Because thesAccH is usedfor sMS datatransmission, messages canbe receivedor tran$mitted duringa voicecall. Initial applications of SMSfocusedon alphanumeric pagingserviceswith fundamentaldifferences:SMS is bidirectionaland message deliveryis guaranteed. subsequent applicationsservedby sMS arevoicemail notification,e-maildelivery, stockquotes,anddownloading/updating of SIM cards.

9.3 CODE DIVISIONMULTIPLE.ACCESS CELLULAR The two digital cellularsystemsdiscussedin the previoussections,D-AMps and GSM,utilizea combinationof frequencydivisionmultiplexingandtimedivisionmultiplexing+asa methodof partitioninga blockof allocatedfrequencyspectruminto individual communication channels. This section discussessystems with a fundamentallydifferentapproachto channeldefinition-code divisionmultiple access(CDMA)-that belongto a classreferredto asspreadspectrumcornmunications systems[a]. The term spreadspectrumrefersto thefact thattransmission bandwidth *Within

a particular GSM cell it is conceivablethat a single FDM channel supportingeight TDMA channels is sufficient for sufficiently low traffic situations. Irr this case, the particular cell utilizes only TDMA. Nevertheless, the mobiles still suppott FDM operations so they can move to different cells and switch freouencies.

445 9.3 CODE DIVISIoN MULTIPLE.AG0ESS CELLUI.AR usedby anindividualchannelis muchwiderthantheinherentbandwidthrequiredby themessage beingtransmitted. Spread$pectrumsystemshavetraditionallybeenused wheretheincreased in militaryapplications complexityof implementation isjustified by two particularfeatures.First, it is relativelydifficult to detectthe presenceof a spreadspectrumsignalbecause the signalenergyis spreadacro$$a wide bandandis oftenmaskedby backgroundnoise.Second,it is moredifficult to jam a spreadspectrum signalbecause thejammingsignalenergymustbe spreadacrossa wide bandas opposedto beingfocusedinto a relafivelynarrowband.Theperformance of a spread spectrumreceiveris comparableto the pelformanceof receiversfor traditional narrowbandsignalsaslong asthe spreadspectrumreceiverknowsandcansynchronize to themethodbeingusedto spreadthespectrum. Two generalcategoriesof spreadspectrumcommunications arefrequency-hopping systemis one systemsanddirect-sequence modulationsystems. A frequency-hopping in whichhansmission at anyparticularinstantis confinedto a relativelynarow band Insteadof of frequencies commensurate with theinherentbandwidthof themessage. particular does,a within as a conventional communications system staying one band jumps of within frequency-hopping betweennarrowbands frequencies a large system blockof spectrumin someprescribedmanner.As mentionedin theprevioussection, a GSM cellularsystemhastheability to operatewith frequencyhopping(specifically, slowfrequencyhopping). 9.3.1 CDMA Channel Establishment A contemporarycellular CDMA system,as developedby Qualcommand standrtpectrum ardizedby EIA/TIA in Interim StandardIS-95 [5], usesdirect-sequerce spreading.One particularmethod of implementinga direct-sequence spectrum spreadingsystemis shownin Figure9.5.In this systemthe sourcedatais "exclusive ored"with a relativelylong digital codeword.In essence, the "exclu$iveor" process replacesa I of thesourcedatawith thegivencodewordwhile a 0 of thesourcedatais of thecodeword.If a codewordcontainsn bits, replacedwith thebitwisecomplement theoccupiedspectrumof thehansmittedsignalis n timesaswideasif thesourcedata Chip Rste Clock

Oanier generation of spreadspectrumsignal. Figure 9.5 Direct-sequence

446


were directly transmitted.Signals of other CDMA channelsoccupy the sameband of frequenciesbut do so with different speckum-spreadingcodes,which allows separation of the signals in the signal processingcircuinry of a receiver. The basic processof separatingCDMA channelsin a receiver involves correlating a received signal with each of the various codewords (i.e., channels)assignedto the cell. The correlation processproduces a correlation measurementby subtractingthe number of mismatchesin a codeword from the number of matches.Table 9.3 lists an example set of codewordswith particularly useful correlation properties.As indicated in the table, a codeword has sevenmatcheswith itself and no mismatchesfor a net correlation of +7. A sourcedata value of 0 producesno matchesand sevenmismatches for a net correlation of -7. The measurementweights for all other codewordsareeither + 1 or - I , dependingon the valuesof the sourcedata. If all sevenchannelsdefined in Table 9.3 are active in a single cell, the worst-caseinterferencebetweenthe codesproducesan interferencevalue ofeither +6 or -6. Thus, the desireddata can be recovered with a discrimination threshold of 0 for each channel. (The worst-casenet measurement valuesare +l for a I and *l for a 0.) Notice that samecorrelation propeftiesexist for all channel codeswith respectto the other codesin the table. Example basebandwaveforms for the seven-channelCDMA sy$temof Table 9.3 are shown in Figure 9.6. Notice the channel0 receivermeasurementis positive, which implies a data value of l. In this example interferencefrom the adjacentchannelsactually enhancedthe channel 0 measurementfrom an expectedvalue of 7 to 9. Discriminating between nominal measurementvalues of +7 for a desired signal and +6 for worst-caseinterferenceis obviously very tenuous,particularly becausethe individual channels will be received with different power levels. It i$ importanr ro note, however, that an interferencevalue of +6 can arise only if the data values of all intedering channelsdestructively coincide. on average,the composite interference hasan averagevalue of0. The length ofa spectrum-spreadingcode in IS-95 is actually 64 bits long, which meansthat the example worst-caseinterferencewould be 63 in relation to a desiredchannel value of 64. The chancesof destructiveinterferencefrom all 63 channels,or even a large number of channels,is astronomically small.

TABLE9.3 ExampleDlrect-$aquenceSpectrum-Spreading Godesa ChannelNumber

0 1 2 3 4 c

6

ChannelCode

111 0 0 1 0 0111001 1 0 111 0 0 0101110 0 0 1 0111 1 0 0 1 011 11 0 0 1 0 1

Numberof Matches

Numberof Mismatches

7 3 3 3 3 3 3

0 4 4 4 4 4 4

aspectrum-spreading code of d€sirBdchannel: 1110010.

Net Correlation

+7 -1 -1 -1 -1 -1 -1

9.3 CODEDIVISIoNMULTIPLE-ACCESS CELLUTAR 447 Chf,nn€loCdi Chnffid Numbcr

(0)

ftti Vf,hr$

{

(1) 0 fr)

'l

(3) 0 ({)

"-.'

o

*i--j

(5) 0 (6)

|

Clilnn€l0 Mtsrruttffil

Figure 9.6

Example seven-channelCDMA encoding and decoding.

of a 64-channel Example9.1. Determinetheprobabilityof maximuminterference CDMA system with 64-bit spreadingcodes. Also determine the effective powerratio of the sameCDMA sy$tem.Assumeall channels signal-to-interference operateatthesameeffectivepowerlevelatthereceiverandthatall channelcodeshave a cro$scorrelationof+l bit. Solution. The probabilityof 63 destructiveinterferersis merelythe probabilityof occurrenceof 63 equallylikely binary events:prob(maxinterference)= (0.5)63= of a codeword I x 10-1e.The valueof a desiredreceivesignalis the autocorrelation with itself andcanthereforeberepresentedasa valueof 64. The interferencelevel is the sumof 63 binaryrandomnumberrtwith equallylikely valuesof tl. Eventhough probabilitydistribution,the sumof a a singleinterfererdoesnot producea Gaussian large numberof independent randomvmiablesapproaches a Gaussiandistribution (centrallimit theorem).The meanand varianceof an individual interfererareeasily determinedto be 0 and l, respectively.Themeanandvarianceof a sumof 63 such ratio is now variables are 0 and 63, respectively.The signal-to-interference determinedas

F'42

S I R =l 0 t o g r o t i = l s d B

As presented in Chapter6, the 18-dBSIR resultof Example9.1is quitesufficient to supportan acceptableelror rateif the effective powerlevel of all channelscanbe

1

448

DIGITAL MoBILE TELEPHoNY

maintained to be equal. In actual practice, a CDMA deployment does not use all possible codesin a single cell, just as an FDM sy$temdoes not use all frequenciesin one cell. Thus, the amount of interFerencein a cell is limited by the number of codes assigned to the cell and a lesseramount of interferencefrom adjacent cells (assuming the adjacentcells do not use identical, synchronousspectrum-spreadingcodes).The effect of varying power levels in the interfering channelsis coveredin someproblems at the end of the chapter.

9.3.2 CDMA Multlpath Tolerance A primaryadvantage of a CDMA transmission systemis its robustness in thepresence of multipathconditions.The basicreasonfor multipathtolerancecanbe appreciated by examiningtheexamplecodesgivenin Table9.3.Noticethateachcodeis a cyclic shift of all othercodesin the table.Becausethe selectedcodeshavelow correlation with eachother,a delayedversionof anyparticularcodehasthesamelow correlation with anundelayedversionof itself.Thus,theeffectof a multipathdelayof morethan one spreadspectrumbit (referredto asa chip) is no morethanthe effect of the interferencefiom anothercDMA channel,evenif thedelayedversionis at the$amepower level astheprimarysignal. Theeffectof a multipathconditionon a D-AMPSchannelmaybemuchmoredeleterious.Becausea D-AMPS systemoperatesat a relativelynEurowbandof frequencies,it is possiblethat a completefadeoccursfor a particularchannelat a particular physicallocation.If theusermovesthroughthephysicallocation,theeffectof thefade is a momentarydropout.If theuserdwellsat thelocation,theconnectioncanbe lost unlessa handoffoccur$(to a new frequencyand/oran adjacentcell). The slow frequency-hopping featureof GSM ameliorates the effectof a completemultipathfade to a greaterdegree,but not aseffectivelyasa cDMA system.In GSM theremay be momentarydropouts,but theyarenot longenoughto causea droppedconnectionand do not requirea handoff,evenif a userdwellsin a physicallocationwhereoneparticularfrequencyis totally lost.In essence, thefrequency-hopping processofGsM is equivalentto repeatedandautomatichandoffsto different frequencies. Froma somewhat philosophicalpointof view,thebasicintentof a GDMA system is to equalizetheperformance of all channelsin thesystem.In a FDIvI/TDMAsystem it is likely thatsomechannels operatewith veryhighperformance whileothersoperate at very low performance or cannotbe usedat all. Theexistenceof high-performance channelsdoesnot compensate for the existenceof low-performance channels.Thus, a systemthat equalizesthe performance of all channelshasa greatertotal capacity. The primaryreasonfor variablechannelperformance in FDIWTDMA is multipath fading.An FDM/TDMA transmittertypically operateswith a certainamountof excesspowerreferredto asfademargin.Theexcesspoweris notmuchof aproblemwith adjacent-channel intetferencebecauseit is relativelyeasyto isolateFDM channels with frequencyguardbandsandTDM channelswith time guardbands.cochannelin* terferencefrom onecell to anotheris the crux of theproblem.If a particularchannel

9.s coDEDrvrsroN MULTTpLE-AccEss cELLUIAR 449 is operating at an exce$$power level, that channel cannot be reusedexcept at a rela' tively larger distance.

9.3.3 CDMAPowerControl Effectivetransmitpowercontrolof themobileunitsin a CDMA systemis botha requirementanda benefit.It is a requirement because a mobiletransmitterthatis close to a basestationreceiverwill obliterateothermobilesin the samecell thatarefarther away.Thisis referredto asthenear-farproblemof CDMA. If thetransmitpowerlevels of all mobilesin a CDMA sy$temarecontrolledto be no higherfhanabsolutely necessary, thechannelscanbe reusedmoreoften.Althoughpowercontrolis usedin FDM/TDMA systems,it is not possibleto operatewith bareminimumpowerlevels thesystemcannotrespondfastenoughto adjustthepowerlevelsfor fastmulbecause tipathfades.A sidebenefitof usingminimumpowerlevelsin a CDMA mobileis increasedbatterylife. The powerlevelsof a mobile are controlledin two ways;openloop and closed loop.In theopen-loopmode, amobilecandetermineits transmitpowerlevelbymeaspower its received level uring undertheassumption thattransmission lossesareequal directions. in both This assumptionis reasonablefor a CDMA systembut not for a FDM/TDMA systembecausethe latter aremuchmore$usceptibleto independent frequency-selective fading(multipath).Closed-looppowercontrolinvolvesbasestation measures of mobilereceivedpowerandadjustments to themobilepowerlevels with thecontrolchannel.Powerconftolin IS-95CDMA is described in references [6*8]. Table9.4lists thebasicparameters of theIS-95CDMA digitalcellularsystemfor the 800-MHzcellularband.The speechcompression algorithmusesQCELP(Qualcommcodeexcitedlinearprediction).Thefirst commercialinstallationof CDMA occurredin HongKong in 1995. 9.3.4 CDMA Soft Handoff reA uniquefeatureof a CDMA systemis the ability of a mobileto simultaneously ceive from more than one source.Becauseeachcell in a CDMA cellularnetwork TABLE 9.4 lS-95 CDMA Moblle Telephone Sy8tem Parameters

Channelbandwidth Rate Voice-Coding Errorcontroloverhead Aggregatechannelrate Codelength $preadspectrumchannelrate format Modulation

1.25MHz 9.6kbps(maximum) 9.6kbps(downlink) 19.2kbps(uplink) 19.2kbps(downlink) 28.8kbps(uplink) 64 chips 1.2288Mbp$ otfsetQPSK(mobileto base) QPSK(baseto mobile),

450


transmits in a common frequency, a common RF receiver inherently receives the spreadspectrumsignal from all adjacentbase$tations.Signalsfrom multiple basestation$ can then be acquired with multiple basebandcode correlatorsor by time sharing a single correlator with the separatecodes.In a soft handoff operation the samevoice signal is distributed to selectedcells adjacentto a currently active cell. An active mobile can then comparethe quality of the signals and switch to the best one before disconnecting from the establishedbase station. A "make-before-break" operation is not feasible in an FDIIfITDMA 'rystem wherein adjacentcells utilize separatefrequenciesthat require separateRF receivers. FDIVf/TDMA system$inherently usehard handoffs, which require disconnectingfrom one basestation before connecting to a new one. Notice, however, that a soft handoff operation in a CDMA system increasesthe background interferencebecauseof the multiple active signals for a single connection.

9.4 PEHSONALCOMMUNICATION SYSTEM A personalcommunication system(PCS)is a cellularsystemoperatingin a bandof frequencies at 1.9GHz.Theoriginalconceptfor PCSincludedmultiple,newfeatures andservicesbeyondthoseofferedby a basiccellularsystem.Someof theenvisioned featuresweresingletelephone numberfor multipleservice$ (voice,data,fax) anduser mobility for homeor office useandlocationdetermination. Althoughsomecommercial PCSofferingsprovidesomenew userfeatures,initial North AmericanFCSsystem$afebasicallycellularsystemsutilizing a newbandof frequencies. when the FCC allocatedthe PCSfrequenciesfor the united states,they did so withoutstipulatingwhichtypeof systemshouldbe deployed.Thus,anyorganization that bids for andreceivesa franchisefor PCSspectrumis free to choosewhatever type of systemit want$for providingserviceto the public.As a result,North American PCS systemshave beendevelopedwith threedifferent transmissionformats: D*AMPS,GSM,andCDMA. TheD-AMPSimplemenration followsEIA/TIA standard 15-136,which is basicallya revisionof IS-54 that incorporatesdigital control channels.(IS-54definesthe useof an analogcontrolchannelfor compatibilitywith AMPS.)

9.5 VOICE PBIVACYAND AUTHENTICATION Ensuringprivacyofconversations andpreventingfraudaretwo criticalaspects ofcellular telephonesystems thatareaddressed morecompletelyin thedigitalsystems than in the originalanalogsystems.An FDM analogsystemis particularlyvulnerableto casualeavesdropping becausea relativelysimplescannercanbe usedto identify an activechannelandtune into the conversation. If the scannerhastwo receivers.the eavesdropper canlistento bothsidesofthe conversation, assumingtheeavesdropper is in a high enoughlocationor closeenoughto the activemobilethat it canreceive theuplink signal.

e.6 rRtDruM 451 Digital signalsare inherently more complicatedto intercept becausean eavesdropper may needto monitor the connectionestablishmentproces$and not just tune in after a conversation has started. Explicit encryption parameters for optional voice privacy is establishedduring call establishment,or pos$ibly during a conversation,by exchanging privacy control messagesin the control channel. An eavesdroppermust receive the relevant information $entin both directions before eavesdroppingis possible. Even without explicit encryption, an eavesdroppermust be located somewhere near the basestation to properly receive both sidesof a TDMA conversafion.At other locationswithin a cell the eavesdropperwill encounteroverlapping time slotsthat will inflict bit errors into the eavesdropper'sreceived data. A CDMA system produces a similar effect when the phaseof two specffum-spreadingcodesfrom two separatemobiles coincide. Authentication of a valid mobile stationis significantly $trengthenedin digital systems. A major problem in analog cellular systemsis the processof cloning, in which disreputable persons monitor call establishment handshakeprocedures to acquire valid mobile equipment electronic serial numbers (ESNs) and program them into counterfeit mobile units that are then used to place calls. Part of the improved fraud prevention involves maintaining a more up-to-date databaseof valid mobile ESNs. The strongestpnrt of the fiaud prevention involves determining authorization codes basedon past call history in both the base station and the mobile. The authorization code determined by the mobile and sent to the base station must coincide with the authorizationcode calculatedby the basestationbefore serviceis allowed. Additional stepsmay require entry of accesscodesby the user or, in the caseof GSM, the magnetic data card called the SubscriberIdentification Module (SIM).

9.6 lRlDlUM Iridium is a satellite-basedsystem for telephone and two-way paging services.The satellitesystemis a Low Earth Orbit Satellite (LEOS) $ystem,which meansthat signal powers and antennasizescan be reducedwith respectto conventional geostationary satellites.(The Iridium sy$temorbits are less than 500 miles, versusthe 23,000 miles for geostationary satellites.) In addition to enabling lower power ffansmission, a LEOS avoids the propagationdelay of a geostationarysatellite, which is a minimum of 500 msec.A basic disadvantageof a LEOS systemis the continuousmovement of the satelliteswith respectto ground locations.Becauseof this, Iridium provides a large number of satellites(66) so continuous coveragefrom at least one satellite is always available. Iridium phones are dual-mode phones. A phone first makes an attempt to place a call over a terresffial cellular system*but will default to the satellite network if local cellular coverageis not available. When communicating through the Iridium system, a user first gets connectedto the nearestavailable satellite. From there the communication might retum to the ground or be relayed through multiple satellitesbefore re.Multiple

versions of hanclhelds are available to operate on mMA,

GSM, or CDMA cellular networks,

452

DIGITAL MoBILETELEPHoNY

TABLE9.5 lrldlum SystemParameters Numberof satellites Satellite orbit Orbitperiod Transmission protocol RFmodulation Speechdatarate Radiofrequency

66 485 miles (780 km) 1 0 0m i n FDMA/TDMA QPSK 2,4 kbps 1.6 MHz handheldto satellite 23 GHz satelliteto $atellite 19 GHz uplinkto satellite 29 GHz downlinkfrom $atellite

turningto theground.Thesecondgroundlink mightbe directto anotherIridium user or it may involvea basestationwith interconnection to a public telephonenetwork. Systemaspectsof theIridium systemareprovidedin Table9.5.

9.7 TRUNKEDRADIO The term trunkedradio generallyrefersto PrivateMobile Radio(pMR) communicationsnetworks.Originally,usersof PMR equipmentwereallocatedspecificoperating frequencies dedicatedto eachuser(or organization). Suchallocationswereobviously inefficientin termsof bandwidthutilizationwhenthe usersdid not havecontinuous needfor radiocommunications. Significantimprovements in efficiencyareachieved whenthe$eparate channelsareplacedin a groupandsharedby a largergroupofusers on anas-needed basis.Whena userof a trunkedradiosy$temneedsservice,theradio equipmentaccesses anidle channelthatbecomes temporarilyassigned to thatuser.As soonastheusersof a particularchannelceaseto transmit,thechannelbecomes availablefor otherusers.Accessto a communications channelutilizesa controlchannel anda centralizedcontrollerfor resourceallocation. A trunkedradiosystemis not a cellularsystembut doesutilize a centralnodereferredto asa repeater. Therepeaterreceivesa signalon onefrequency,shiftsit to anotherfrequency,andtransmitsthesignalon thenewfrequency.Thus,endu$ersdo not communicate directlywith eachother.*Transmission throughtherepeateris moreeffectivebecause thetoweris locatedat a highpointin thecoverageareautilizinga relatively high transmitpowerthatfacilitatescommunications betweenenduserswho are likely to not havean adequate pathbetweenthem. In a generalsense,trunkedradiosystemsarePMR systemsthatprovideradiocommunicationsdirectly betweenuserswithout use of the public telephonenetwork. However,thedistinctionbetweena trunkedradiosystemanda cellularsystemhasbe*$ome

ffunked radio equipment does support a (special) two-way mode of operation in which the users cofiununicate directly with each other. This mode of operation is generally reserved for communication s in outlying areas where the rcpeater cannot prbvide seryice.

HEFERENCES 453

comeblunedwith U.S.offeringsfromcompanies like NextelandOeotek.Thesecompaniesutilize technologydevelopedby Motorolathat augments the useof spectrum previouslyusedfor privateradio servicereferredto as SpecializedMobile Radio (SMR).SMR applications typicallyinvolvefleetoperationssuchastaxi cabsanddelivery vehiclesin needof dispatchserviceswhereinmultiplemobilessimultaneously heartransmissions on a courmonchannel.SMRradiosutilizeanalogFM/FDM transmissionwith 25-kHzchannels. genericallyrefenedto asEnhanced Thenewequipment, MobileRadio Specialized (ESMR),upgrades analogSMR transmission to digitalTDM transmission in thesame protocolsallow for six mannerthat D-AMPS upgradesAMPS. ESMR transmission conventional digital TDM channelsin each25-kHzchannel.ESMR systemssuppor"t dispatchapplications suchasprivategroupcallandothersimilarspecialized $ervices ing. Most importantly,for this chapter,an ESMR systemcanprovideconnections to the public telephonenetwork-referred to as an interconnectfeature.With this feature,anESMR systemcanfunctionasa cellulartelephonesystem. 9.8 CELLULAR DIGITAL PACKET DATA CellularDigital PacketData(CDPD)[9] is a systemdesigned to providedataservices up to 19.2kbpsasan overlayof an AMPS installation.The primaryserviceconcept of theneedto upgradeanAMPS is to ofTerfixedandmobiledataservicesindependent sy$temto a D-AMPS system.CDPD usesAMPS channelsthat are not being used it is a packet-oriented for voice.Because dataservice,multipleuserssharea single 30-kHzchannel-a significantsavingsin spectrumusageartcomparedto the use of connection-oriented voicechamelsfor data.Principalapplicationsfor CDPDare mobileInternetaccessandcreditcardverification. Accessto the sharedchannelis accomplished with Digital SenseMultiple Access with CollisionDetection(DSMA/CD) which is similarto CSMA/CDof otherradio systemsandEthernetLANs. CDPDusesGMSK modulationwith RS(63,47)forward errorconection.

REFERENCES "Vector SumExcitedLinearPrediction(VSELP)7950Bit PerSecondVoice Coding Illinois, Nov, 14,1989. Algorithm,"TechnicalDescription,Motorola,Schaumburg, "Overview M. Rahnema, of the GSM Systemand Protocol Architecture,"IEEE Communications Magazine,Apr. 1993,pp. 92-100. P.Vary, K. Hellwig,C, Galland,M. Russo,J. Petit,andD. Massaloux,"SpeechCodec for the EuropeanMobile RadioSystem,"in IEEE GLOBECOM1989,Nov, 1989,pp. 29.8.2. R. Dixon,SpreadSpectrumSystemt with Commercial Applications,Wiley, New York, 1994. "Mobile Station-Base Station Compatibility Standardfor Dual-Mode Wideband SpreadSpectrumCellularSystem,"EIA/TIA./IS-95,Washington,DC, July 1993.

454

DIGITAL MOBILE TELEPHONY A. Salamasiand K. S. Gilhousen,"On the SystemDesignAspectsof CodeDivision Multiple Access(cDMA) Applied to Digital cellular andpersonalcommunications Networks,"Praceedingof the Forty-FirstIEE VehicularTechnology Conference, May 1991,pp.57-62. A, J. Viterbi and A. M, Viterbi, "Erlang Capacityof a Power Conrolled CDMA System,"IEEE Journal on SelectedAreas of Comtnunitations,Nov. lgg3, pp. 882-890, K. S. Gilhousen,L M. Jacobs,R. Padovani,A. J. Viterbi, L. A. Weaver,and C. E. Wheatley,"On the Capacityof a CellularCDMA System,"IEEE Transactionson VehitularTechnology,May 1991,pp. 303-312. "Packet A. K. Salkintzis, DataoverCellularNerworks:The CDpD Approach,"IEEE Communications Magazine, June1999,pp. 152-159,

PROBLEMS 9-1 what is theavailablebearerrareof a half-ratedigiralchannelin a D-AMps system? 9.2 what is thedatarateof the slow associated controlchannelin a D-AMps system? 9.3 what is theavailablebearerrarea full-ratedigitalchannelin a GSM sy$tem? 9.4 what is thedatarateof the slowassociated controlchannelin a GSM system? 9.5 Determinethereceiveroutputmeasurements for channelsI and2 for theGDMA exampleof Figure9.6. 9.6 what is theeffectivesignal-to-interference ratioof a singlecDMA uplinkchannel operatingat a distancethat is twice asfar from the basestationas62 other channels? Assumea codelengthof 64, crosscorrelationsof +1, andall transmitters operateat identicalpower levels. (a) Assumeall interferersare active. (b) Assumehalf theinterferersoperatewith a,25qodataratebecause of no voice activity. 9.7 In a CDMA systemwith a codelengthof 64 determinethesignal-to-interference ratioof a singleuplinkCDMA channelif thereare16activeinterferingchannels operatingat aneffectivereceivepowerlevelthatis 12dB higherthatthedesired channel.

10 DATAANDASYNCHRONOUS TRANSFER MODENETWORKS Theprimaryfocusof this bookis to describehow andwhy digitalelectronictechnolvoice,a fundamentally analogsignal.Naturally,thedigiogy is usedto communicate inherentlydigitalinformation(data),butonly tal telephone networkcanaccommodate for voice.This if datacanacceptor be adaptedto servicecharacteristics established chapterdiscusses digitaltechnologyandnetworksspecificallydirectedto supporting how Ironically,thelaterpartof this chapterdescribes seruices. datacommunications networks voice. to the data-oriented areadapted carry (data)is actuallyanancientpractice.In early of discretemessages Communication greater thanthe rangeof the humanvoicewas over distances timescommunication providedby sightor soundof discretesignals(e.g.,heliographs, smokesignals,flags, systemto useelectricity, the first practicalcommunication andhorns).Furthermore, thetelegraph, is inherentlydigital.As telegraphyevolvedfrom theoriginal,manually basedsystems to fully automated systems, theydeveloped into whatarecommonlyremessage in networks.Modem switchingis discussed ferredto asme$sage-switching the first sectionof this chapter. arose,it wasonly natural As theneedfor modemelectronicdatacommunications that thepublictelephonenetworkwouldbe usedfor datakansmissionservices.The of a networkdesigned availabilityovershadowed numeroustechnicalshortcomings primarilyfor voicecommunications Themaindeficiencies services. of a conventional telephonenetworkfor datatransmissionare: (modems)on analogaccesslines 1. Needfor signalhansducers 2. Limited datarates 3. High errorrates(in the olderanalognetwork) 4. Inefficientcircuitutilizations so did thejustificationfor more requirements increased, As datacommunications solutions.Onesolutionto reducingdatatransmiscost-effective datacommunications 455

456

DATAANDASYNcHHoNoUSTFANSFERMoDENETWoRKS

sion costswas to improve circuit utilizations through the use of packet-switchingnetworks. The technology of packet switching was pioneeredby the Advanced Research Projects Agency of the u.s. government. This agency developeda network referred to as the ARPANET []. In addition to ARPANET, which was used only by government, educational, and industrial research institutions, a number of public packetswitching networks were also developedin the United Statesand around the world. ARPANET developmentseventually evolved into what is now known as the Internet. A secondapproachto improving data communicationsinvolves developing separate networks specifically designedfor digital ffansmission (no analog circuits with modems).The first major enterpriseof this type in the United Stateswas a nationwide digital microwave network developed by Digital rransmission corporation (DATRAN). Becausethe data communicationsmarketplaceof the time could not support a separate,dedicatednetwork for data, DATRAN ran into financial difficulty and declared bankruptcy in 1977. Another, more successful,data-networking approach was the Dataphone Digitat Service (DDS) offering of AT&T. This serviceutilizes digital technology of the telephone network for strictly data applications.DDS circuits are dedicatedto data services, but the facilities and routes are sharedwith telephonenetwork facilities. A major hurdle for DDS is achieving digital accessto DDS circuits. If a subscriberis ourside the range of digital transmissionfacilities of a DDS serving office, a voiceband modem over an analog line is required. A fourth approachto satisfying datacommunicationsservicesinvolved developing meansfor directly accessingthe digital ffansmissionand switching equipment of the telephonenetwork. The first widespreadapproachof the telephonecompaniesfor providing universal digital accessis the Inregrated services Digital Nerwork (ISDN). ISDN provides digital accessto the digital facilities of the telephonenerwork for voice or data services on a call-by-call basis. ISDN digital subscriber lines and other methods of digital accessto digital networks such as the Internet are described in Chapter I L

10.1 MESSAGE SWITCHING As one telegraphsystemafter anotherwas installed in the countriesaround the world, nationwide communications networks evolved. A messagecould be sent from one point to anothereven if the two points were not servicedby a common telegraphline. In this case,telegraphoperatorsat intermediatepoints would receive a messageon one line and retransmit it on another.When a telegraphoffice had severallines emanating from it, the processof hansferring a messagefrom one line to anotherwas, in essence, a switching function. The processof relaying, or switching, a messagefrom one telegraphline to another becamesemiautomatedwhen teletypeswith paper tape punchesand readerswere developed. An incoming messagecould be punched automatically onto a paper tape by one teletypeand subsequentlyreadby anotherteletype for transmissionon the appro-

10.1 MESSAGESWITCHING457

priateoutgoingline.Theprocessof transferringa message from oneline to anotherin switches. this mannerled to thesesystemsbeingreferredto astorn-tapemessage in 1963 switcheswascompletelyautomated One of the world's largestme$sage when Collins Radio Companyof CedarRapids,Iowa, installeda computer-based switchfor theairlinecompanies of NorthAmerica.This systemandthemore me$$age direcfly recentsuccessors eliminatepapertapetransfers by storingincomingmessages into a computermemory(disk file) andforwarding themautomaticallyto the appropriateoutputline whenavailable.Hencethis modeof operationis oftenrefenedto as store-and-forwardmessage switching. is a headercontaininganaddress andpossiblyrouting Includedwith eachmessage proce$sor at eachnodecandetermineto whichoutputline informationsothemessage As indicatedin Figure10.1,theprocessor in eachnodemainto $witchthemessage. queuesfor eachoutgoinglink. Thesequeuesarenormallyservicedon tainsmessage be ina first-come,first-servedbasis.However,priority informationcan$ometimes cludedin eachheaderto establishdifferentclasses,or gradesof service,therebyalto be placedat theheadof a queue. lowing tirne-criticalmessages A message-switching networkis fundamentally differentfrom a circuit-switching networkin that the sourceanddestinationdo not interactin real time. In fact, most on a delayedbasisif a destinamessage-switching networkscoulddelivera message nettion nodeis busyor otherwiseunableto accepttraffic. In a message-switching nodebeforesendinsa work thereis no needto determinethestatusof thedestination message, asthereis in circuit switching.

Figure 10.1 Message-switching network.

458

DATAAND ASyNcHHoNous THANSFER MoDENETwoRKS

Message-switchingnetworks are also fundamentatly different from circuit-switching networks in their responseto traffic overloads.A circuit-switching network blocks or rejects excesstraffic while a message-switchingnetwork normally acceptsall traffic but provides longer delivery times as a result ofincreased queuelengths. Another importaxt distinction of a message-switchingnetwork is that the transmission links are never idle while traffic is waiting to use them. In a circuit-switching network, a circuit may be assignedto a particular connection but not actually carrying traffic. Thus, some of the transmissioncapacity may be idle while someuser$are denied service.In contrast,utilization of the transmissionlinks of a message-switching network is directly related to the actual flow of information. Arbitrarily high utilization efficiencies are possible if increased store-and-forwardqueuing delays are acceptable' Chapter 12 provides basic results of queuing theory that relate utilization efficiency to queuing delay.

10.2 PACKETSWITCHING The circuit-switched telephonenetwork is ill-suited to interactive data traff,rcbecause it is fundamentally designed for less frequent service requests with comparatively long holding times (3-4 min on average).Neither the control elementsin the switches nor the capacity of the signaling channelsare capableof accommodatingfrequent requestsfor very short messages.The result is that connection setuptime may be many times greaterthan the holding time of a data message.Obviously, more efficient utilization of the network requires greater control channel bandwidth and increasedcall processingcapacitiesin the switches. Beyond this, however, interactive data traffic with low-activity factor$ requires a network operation that is fundamenta-llydifferent from a conventional circuit-switched network, The most appropriatemode of operation for traffic that comesin bursts is more closely related to a message-switchednetwork than to a circuit-switched network. Figure 10.2 depicts both the conceptualstructure and the conceptualoperation of a packet-switchednetwork. A single messageat the sourceis broken up into ..packets" for transmissionthrough the network. Included in each packet is a headercontaining addressand other control information. Each packet is relayed through the network in a store-and-forwardfashion similar to a message-switchingnetwork. At the destination node, the packetsare reassembledinto the original contiguous messageand delivered. The main feature of a packet-switchingoperationis the marurerin which the tran$mission links are sharedon an as-neededbasis.Each packet is transmitted as soon as the appropriate link is available, but no transmission facilities are held by a source when it has nothing to send. In this manner, a large number of relatively inactive sourcescan sharethe transmissionlinks. In essence,link utilization is improved at the exPenseof storageand control complexity in the nodes. A circuit-switched network ha$ conffol overheadassociatedwith connection setup but very little control thereafter' In contrast,packet-switching nodes must processthe headerinformation in each packet as it arrives. Thus a long messagein a packet-switchednetwork requires more

459 ro,? PAoKETSWTToHTNG

fonsrrd nod*r

Flgure10.2 Packet-switching network. conhol overheadthanif it wereservicedin a circuit-switchednetwork.Considering thedecliningcostof digitalmemoryandprocessing, theincreased controlcomplexity As discussed laterin becomes lessandlesssignificantasdigitaltechnologyadvances. this chapter,one particularvariationof packetswitching,AsynchronousTransfer of conMode (ATM), is designedto specificallysupporthardwareimplementations trol intensivefunctions,therebysupportingvery hightraffic volumeswith low delay. networkincreases, the averagetransmisAs the traffic load in a packet-switched correspondingly. ln contrast,a circuit-switchednetworkeither sion delayincreases grantsserviceor rejectsit. Thereis no gracefuldegradation in service.Conversely, circuit-switched network, muchnetworktranswhenonly a few circuitsarein usein a packet-switched network,the missioncapaciryis idle.Whenthereis a light loadon a activeusersbenefitby shorterthanusualdelaytimes.Hencefrom a user'sgrade-ofservicepoint of view, the two networktypesarefundamentallydifferent. networks Using automaticrepeatrequest(ARQ error control,packet-switching (e.g.,ARPANET)traditionallyprovidedessentiallyerror-freetransmission for each transfer.Thisprocessrequiresthereceivingnodesto monitorredundant node-to-node checkbits appendedto eachpacketto determineif the packetwasreceivedcorrectly. (message NAK). Hencetransa retransmission is requested Wheneruorsaredetected, packets in memoryuntil a positiveresponse mitting nodesmusthold all transmitted (messageACK) is retumedby the receivingterminal.Fufihermore,an entire packet is usuallyreceivedandcheckedfor errorsbeforeforwardingit to anothernode. Customers typicallyaccesspacketnetworksby way of leasedlinesor dial-upconnections.Dial-upconnections areusedby infrequentusers,while leasedlinesarepreferredby heavyusersto achieveconstantavailability,higherdararates,andpossibly lower errorrates. Despitethe similarityto a me$sage-switching operation,a work is differentin two impodantre$pects:

460

DATAANDASYNCHRoNoUSTHANSFERMoDENETWORKS

l. The store-and-forwarddelay through a packet-switcheclnetwork is relatively short. Thus interactive communicationscan occur in much the samemanner as if a dedicated,end-to-endcircuit is established. 2. A packet-switchednetwork doe$not provide $torageof messages,except in an incidental manner while relaying packets from one node to another. The network is designedto provide switched communication between two nodes. both of which are actively involved in the communications process. A packet-switching network does not normally store a me$sagefor later delivery to an inactive or busv terminal. One reasonfor breaking messagesinto packetsis to allow transmissionof the first segmentof a long messagewhile other segmentsare in transit. If the entire me$$age had to be received at each node befbre forwarding it to the next node (as in message switching), the delaysthrough the notlesmight be too large. Another reasonfor breaking the mes$agesinto packets arises from operational simplifications derived from storing, processing,and transmitting smaller and possibly fixedJength blocks of data. In addition, if long message$are transmittedintact, short messagesexperienceexcessive delays when queuedbehind long messages.Packetizationallows shon messages to get through a transmissionlink without waiting behind long messages.This same principle occursin multiprogrammed computers,which use time slicing to allow short jobs the opportunity of finishing before previously startedlong jobs. one more motivation for packetizationis that when a ffansmissionblock is too long, it is unlikely that the entire messagewill be received correctly. Packetizationprovides a means of retransmitting only those portions of a messagethat need to be retransmitted.

10.2.1 PacketFormats The format of a packet in a packet-switchingnetwork can vary significantly f'rom one network to another. Some formats include numerou$ fields fbr control infbrmation while other system$rely more heavily on special control packetsto transmit control information. Generally speaking,the conffol information associatedwith a particular messageor link is included in the headerof a mes$agepacket.Less frequent,networkrelated control information is communicatedtlrough special control packets. As indicated in Figure 10.3, a packet contains three major fields: the header,the message,and the redundancycheck bits. some packet$may not contain a message field if they are being used strictly for control purposes.Although a variery of techniques for generating redundancy checks are possible, the most popular technique usescyclic redundancychecks (cRCs). Basically, a cRC is nothing more than a set of parity bits that cover overlapaing fields of messagebits. The fields overlap in such a way that small numbers of errors are always detectedand the probability of not detecting the occurrenceof 2 large number of errors is l in 2M, where M is the number of bits in the check code. A headertypically containsnumeroussubfieldsin addition to the necessarvaddress field. Additional fields sometimesincluded in a headerare:

IO,2 PACKETSWITCHING /161

Trxt

Hcdtr

rR To

iddtr|8

Op coda

From EddrG$

$tq

cRc

BytE count

Figure 10.3 Typical packetformat, I , An operationcode to designatewhether the packet is a message(text) packet or a control packet.In a sensethis field is a pat"tof the destinationaddress,with the addressspecifying the control element of a switching node.

z. A

source address for recovery purpo$es or identification of packets at a destination node that is capable of simultaneously accepting more than one message.

at the destinationnode,detect messages 3. A sequencenumberto reassemble faults,andfacilitaterecoveryprocedures. 4. A lengthcodeto indicatethelengthof a packetwhenlessthana standardsize packetis transmitted.Someprotocolsinsertspecialdelimiters(flags)at the end of a packetandthereforedo not usea lengthcount. 10.2.2 Statittical Multiplexlng The digital time divisionmultiplexingtechniquesdescribedin thepreviouschapters providemultiplechannelsby periodicallyassigninga time slot to eachchannel.The time slotsareassignedwhetheror not the respectivesourceshaveanythingto send. basis,but for the Channelassignments maybealteredon a connection-by-connection Bedurationof a "call" a particulartime slotis dedicatedto a respectiveconnection. causeof thecontinuousmannerin whichtime slotsoccurfor eachchannel,this form time divisionmultiplexing referredto as synchronous of multiplexingis sometimes (STDM).In this sectionwe describeanotherform of multiplexing,variouslyreferred time divisionmultiplexing(ATDM)- or statisticaltime division to as asynchronous (statmux). multiplexing This form of multiplexingis mentionedherebecauseof its closerelationshipto packet-switching techniques. Statmuxesoperatewith framingformatsthat arebasicallyidenticalto STDM framing formats.The majordifferenceis that a statmuxperiodicallyredefinesthe length of its framesto changethe numberof time slotsand,hence,the numberof channels. Whereasan STDM systempermanentlyassignsa time slot to eachof its sources,a "Here Il the contextof d igital is alother context-sensitive useof the termssynchronousandasynchronous. telephony,"synchronousmultiplexing" refersto combiningtributary signalsthat havebeensynchronized "asynchronousmultiplexing" to each other as in upper layer SONET multiplexing. Corrospondingly, refersto the useof pulsestuffing to accofimodateunsynchronizedtributaries,

462

DATAANDASyNcHHoNousTHANSFERMoDENETwoHKs

statmux assignsa time slot only when a sourcebecomesactive. A time slot is eliminated (the frame shoftened)when the respectivelrourcebecomesinactive. statmux $ystemshave been primarily used to provide line sharing for a multiple number of interactive terminals communicating with a host computer. when only a few sources(terminals) are active, each sourcehas a relatively high data rate channel available to it. As the number of active sourcesincreases,the frame length increases so individual channel rates decrease.some systems limited the number of active sourcesto ensurecertain minimum data rates. The purpose and performance of statmux systems are very similar to the purpose and performance of a packet-switching link. The main difference is that a packetswitching link hansmits larger blocks of data with a header included in each block. Each time slot of a statunux$ystem is shorter and contains only source data. Figure 10.4 contraststhe basic operation of messageswitching, packet switching, and statistical time division multiplexing. The messageswitch transmitseachmessagein its entirety in a f,rrst-come,first-served manner. packet switching breaks messagesup to allow interleaving of packets from other sources. Thus short mes$agesnever ger queuedbehind long messagessuch as file transfers.The statmux system breaks the messagesup into even finer blocks (words) of data and addsperiodic frame rlefinition messagesso that receiving terminals can properly identify the individual time slots and switch the incoming data accordingly. As indicated in Figure 10.4, a packet-switching operation becomesvery similar to a statmux operationif the size of the packetsis small. In fact, commercially available

irhrr|[e rwitohing

A

HI

-

B

B

B

B

8

A

8

C

B . C

trII trDtrDEDtrDEDEtrEDtrU -t

c

Ptrket sinching

iEdGflnltiod

f igure 10.4 comparison of message-switching,packet-switching,and statistical multiplexing.

10,?PACKET swrTcHrNc 463 netstatisticaltime divisionmultiplexerscanbe usedto build up a packet-switching nodeneedssuffiwork. In additionto the multiplexers,however,a packet-switching cient storageto accumulatean entire packet,checkit for errors,processit, and retransmitit. Strictlyspeaking,a statisticalmultiplexerdoesnot accumulate anentire message or packet.It switchesthe incomingdataasthey arereceived.Error control for retransmission) areimplemented befunctions(redundancy checkingandrequests tweenendpointsof a "connection"insteadof betweennodes.In summary,a statmux A packet-switching is strictlya multiplexer/demultiplexer. nodeprovidesmultiplexandnetworklevelconholfunctions. ing functionsaswell asmessage-level 10.2.3 Routlng Control regardingvariousprocehavebeenundedaken Muchdiscussion andexperimentation duresfor routing packetsfrom sourcesto sinksthrougha packet-switchingnetwork routing allowfor a certainamountof adaptation or alternate [2]. All routingtechniques differ, however,in how to circumventline or nodefailures.The varioustechniques fasttheyrespondto failuresandwhetheror not theycircumventnetworkcongestion arethefollowing: aswell asequipmentfailures.Thebasictechniques l. Dynamicrouting 2. Virrual circuitrouting 3. Fixed-pathrouting Eachof thesetechniquescanbe implementedin a varietyof waysand can assume someofthe characteristics ofthe otherroutingcontrolprocedures. Dynamlc Routing on a distributedbasiswith networknodesexamining Dynamicroutingis implemented outgoing ofeachreceivedpacketto determinetheappropriate thedestinationaddress processing locally storedinformationto deterlink. The outgoinglink is selectedby path provides mine which minimum delayto the destination.Therouting criteria are routinelyupdatedto includeboth the operationalstatus(health)andthe sizeof the queuesin the neighboringnodes.The routingdecisionsareadjustedrapidly enough may follow dffirent pathsthroughthenetthatindividual packetsof a singlemessage work. Dynamicrouting,with its ability to respondquicklyto changesin networktopology or traffic conditions,is one of the original featuresespousedfor packet consideredto be inherentin the switching.In fact, dynamicroutingwas $ometimes network. definitionof a packet-switching In spiteofthe obviousattractionofbeing ableto adjustto rapidfluctuationsin trafOneimplication fic patterns,dynamicroutinghasa numberof significantdrawbacks. packetsin a message of allowingsuccessive to follow differentroutesthroughthenetAlthoughsequence work is thatpacketsmay arriveat a destinationout of sequence. processis complinumbersareusedto rearrange thepacketsproperly,thereassembly

464

DATAANDASYNCHFIoNoUSTHANSFEFMoDENETwoFKS

cated, particularly since the destination does not know if a missing packet is merely delayedor lost entirely. Another drawback to dynamic routing is the possibility of oscillation occuring in the routing decisions.Ifthe bandwidth assignedto updating the routing control algorithms is too small, a lightly loaded node will attract more traffic than it can handle before neighboring nodes are informed of the change in the traffic flow. In this instance,a packet might even wind up at a node from which it has previously been sent. Purely distributed control, in general,and dynamic routing, in particular, also present difficulties with respect to flow control in a packet network. As mentioned in chapter 7, alternate routing in the switched telephone network is sometimes suspended when the network as a whole becomesoverly conge$ted(alternateroutes require more resources).obviously, the same principle applies to a packet-switching network' Flow control in packet networks is discussedin later sections. Dynamic routing is most appropriatefor small networks or in a military environment where survivability of the network in the presenceof muttiple-node failures is a requirement' A military network i$ typically more concernedwith reliable 4d timely completion of a few important messagesthan with achieving the highest possible throughput from a given amount of resources.

Virtual Circults A virtual circuit network embodiessome of the basic aspectsof both circuit switching and packet switching. The transmissioncapacity is dynamically assignedon an ,.asneeded"basis,but all packetsof a multipacket messagefollow the sarneroute through the network. Before interactive communication begins, a route is establishedthrough the network, and all participating nodesare informed of the "connection" and how to route the individual packets that follow. From then on, all packets flowing between the two end points follow the sameroute through the network. In essence,a virtual circuit is a logical concept involving addressesand pointers in the nodesofthe network but no dedicatedtransmission facilities. At the end of a connection (or ..session"in datacommunicationsterminology), a virtual circuit is releasedby a ..disconnect"me$sagepropagating through the network. Except in the caseofpermanent virnral circuits, separateconnections,or sessions, involving the sametwo endpoints do not necessarilyuse identical paths thrcugh the network' Each virtual circuit is establishedduring the call establishmentphase depending on the traffic pattern$at the time. Thus a virtual circuit network can respond to network failures or changing traffic patterns,but in a longer time frame than a dynamically routed network. When virtual circuits are changedfrom one connectionto the next, the mode of operation is sometimesreferred to as a switched virtual circuit networkby direct analogy to conventional circuit swirching. Virtual circuits can be establishedusing either distributed or cenffalized conffol. when distributed control is used,the call establishmentmessagepropagatesthrough the network with eachnode making a local decision as to which outgoing link should be selected.

10.2 PACKET swtrcHtNc

465

As discussedin Chapter7 concerningconventionalcircuit-switched networks, centralized call establishmenthas the basic advantageof being able to setup circuits with a networkwide view of network statusand traffic patterns.The TYMNET network of the United States[3] and the DATAPAC network of Canada[4] are examplesof virtual circuit, packet-switchednetworks with centralizedconffol. Sincethe call establishmentphaseofa virtual circuit representsexcessiveoverhead for single-packetme$sages,a virtual circuit mode of operationis obviously most useful when a network servicesa relatively large number of multiple-packet messagesor $essions.Thus a dual-mode network is suggestedrequiring vittual circuits for longer messagesand immediate transmission with dynamic routing for single-packetmessages.In this instance, the single-packetmessagesare usually referred ta as datagrams. One of the main advantagesof a virtual circuit operation is its ability to provide more orderly control of packet delivery. If a node in a virtual circuit never forwards a packet pertaining to a particular connection until the previous packet has been acknowledged,packetscannotarrive out oforder. A secondadvantageofa virtual circuit is the reducedaddressingrequirementsof individual packets.Once a virtual circuit has been established,complete destinationaddressesare no longer neededin the packets. In place of destination addresses,virtual circuit identifiers can be used that are local to each link. In essence,virtual circuit identifier$ are pointers to memory addressesin the call processorsof the packet-switching nodesor to look-up tables in ATM nodes. The designatedmemory addressescontain the pertinent information conceming the actual source,destination,and routing of the packets.Another important feature of a viftual circuit operationis its inherent ability to aid in flow control, as discussedin the next section. The main disadvantageof a virtual circuit operation is the possibility of greater transmission delays. When a path for a virtual circuit is established,it is chosen to minimize the delay through the network under the traftic pattems at that time. If the traffic pattems change,packets pertaining to a particular virtual circuit may experiencelong queuing delays on some links while alternatelinks are more lightly loaded. Yum and Schwartz [5] report that analysesof routing techniquesfor severalsmall network configurationsindicate a packetdelay improvement of lA-ZlVa is possiblewhen adaptiverouting is used in lieu of a fixed routing rule. When first considered,call establishmentmight seemto be a disadvantageof a virtual circuit network. Actually, however, flow control considerationsrequire sometype of query/responsemes$ageto determine the statusof the destination nodes before a $ourceis allowed to begin sendinga packet $trearn.Thus the control overheadand the "connection" is usually a fundamentalrequirement, delay associatedwith settingup a even in a dynamically routed network.

Fixed-PathRoutlng of a virnralcircuitnetworkexFixed-path routingembodies thesamebasicconcepts between alwaysusethe ceptsuccessive anytwoendpoints connections, or sessions,

466

DATAANDAsyNcHRoNousTHANsFEBMoDENETWoRKS

samepath' In essencea fixed-path network is one that assignspemanent virtual circuits to eachpair of endpoints. One attractive f'eatureof a fixed-path network is the absenceof the call establishment phaseof a virtual circuit network. However, unlessthe necessaryresources are perrnanently allocated, a "clear-to*transmit" messageis needed before sending packets.Inactive virtual circuits do not tie up network resource$in the samemanner as conventional circuits, but each virtual circuit does require transmission capacity and store-and-forwardbuffers in a probabilistic sense.If minimum gradesof service (delay times) are to be guarauteed,a network mu$t limit the number of virtual circuits existing at any particular time. Hence clear-to-tranrrmitsignals are neededbefbre a fixed-path circuit becomesactive. of course,a network can provide two modesof operation: permanentlyactive (hot line) virtual circuits and virtual circuits activatedand deactivatedas needed. An obvious disadvantageof a purely-fixed-path network, as describedso far, is its vulnerability to node or link failures. To circumvent this problem, a network control centerusually assignssemipennanentpathsthrough the network that arechangedonly when necessaryfor failure survivability or maintenance.Some older message-switching networks were implemented with fixed-path routing in the strict sense.This mode of operationwas more appropriateto merrsage-switchingnetworks becausemessagedelivery was less time critical and could be deferred while repairs were undertakenfor inoperative equipment.

10.2.4 Flow Control As discussedin ChapterT concerningconventionalcircuit-switched networks, routing and flow control are two closely related operationalrequirementsof any communications network. The same basic principlb fbr controlling the flow in circuit-switched networks also appliesto packet-switchednetworks. Namely, wheneverthe network is experiencingexcessivesffess,either from a loss ofcapacity due to failures or from an abnormally heavy demand fbr services,new service requestsmust be blocked at the periphery of the network before they tie up common reriourcesand compoundthe congestion. In a packet-switchednetwork, the common resourcesare store-and-forward buffers and transmissionlinks. Flow control in a packet network is primarily concernedwith buffer managemenr. For example, if all store-and-forwardbuffers in adjacent nodes become filled with packetsdestinedto eachother, the nodesare unable to receive additional packets,and a deadlockexists.Recall that, in pure packet switching, a node doesnot releasea buffer as soon as it transmits a packet.The buffer is releaseclwhen an acknowledgmentis returned from the adjacentnode. If a receiving node has no availablebuffers, it cannot accepta new packet and thereforecannot acknowledgeit. (Framerelay anclATM net_ works describedlater do not provide node-by-nodeerror control so they do not save copies of messagesin internal nodesof the network.) Flow control requirementsimply that interface nodesin a packet-switchednetwork are aware of overload conditions and refusenew requestsfor service until the conges-

10.2 PACKET SWITCHING467 tion is relieved. A parficularly attractive feature of a virtual circuit network is that the call establishmentphaseprovides an automatic meansof determining whether or not a particular requerttshould be serviced.If no path through the network can be established becausethe interior nodes are too conge$ted,the request is rejected. On the other hand, if a virtual circuit is established,there is reasonableexpectation that the entire requestwill be serviced in a timely manner. Unless network nodes are very conservativein acceptingrequest$for new virtual circuits, the ability to set up a circuit does not guaranteethe avoidanceof excessive congestion or deadlocks.A node acceptsa new virnral circuit based on an expected load and its capacity to service that load. If the tratfic volume and pattems happento exceedthe expectedload, excessivedelay or congestionis possible. Virtual circuits are an effective means of conrolling the flow of multiple-packet mes$ages,but they representtoo much overheadfor single-packetor datagramflow control. If a virtual circuit network must supporta significant number of single-packet mes$ages,it can allow immediate transmissionof thesemes$agesand forego the call establishment phase. In terms of transmission overhead, a datagram is not much greater than a call establitrhmentpacket. Thus, from this point of view, the packet might as well be sent immediately and be consideredits own circuit setup message. In terms of store-and-forwardbuffers, however, a single-packetmessageis much different from a call establishmentpacket.A call establishmentpacket requiresa certain amount of storage and processing by the call processor of each network node it reaches,but it does not compete for store-and-forward buffers as does a me$sage packet. Ifnecessary,a call establishmentmessagecan be ignoredby an overloadednode, and the originating node times out waiting for the network respon$eand reissuesthe request.The time out should be long enough that the network has had a chanceto rela,r. In confrastto call establishmentpackets,if a messagepacket is ignored by an overloaded node,congestionmigratesto the node that last transmittedthe packet,sincethis node is holding a copy of the rejectedpacketin a buffer. The buffer cannotbe released until an acknowledgmentis received.Hence datagram$cannot be allowed to enter the network unlessa reasonablechancefor complete passageexists.When the network is heavily congested,attemptsto set up virtual circuits might also be suspended. A conventional circuit-switched network is unconcerned with flow control between the endpoints of a connection since, once the circuit is established,the activity or inactivity ofthe endpointshas no effect on other connectionsor on the network as a whole. End users necessarily administer flow control between themselves so the sourcedoesnot overrun the receive buffers of a sink. Theseproceduresconcern only the endpoints. In contrast,the very nature of a packet-switching network implies direct involvement with endpoint activity. If a sourcepumps excessivetraffic into a network, other users experiencedegradedperformance.Hence interfacesto a packet-switching network necessarilyinclude flow control for respecfivesources.Source flow control establishesa maximum data rate for a network, If a sink acceptsdata at a lower rate for a sustainedperiod of time, this fact must be communicatedto the network node servins the sourcein order to slow the sourcedown.

468

DATAANDAsyNcHRoNoUsTRANSFEHMoDENETwoRKS

store-and-forward buffers of a packet-switching network are neededfor communications purposesand are not used as a storagemedium for messages.If a packetswitching network also provides message-switching services, message srorage functions should be implemented separatelyfrom the communications buffers. Then. the packet-switching network is used to transfer data to the message$toragefacility. Another implication of maximizing store-and-forwardbuffer utilization is the need to wait until a sink is ready to acceptdata before a sourcebegins sending.Ifa sink is not ready, packetsget stuck in store-and-forwardbuffers at the far end of a network and cause congestion. Thus some form of request-to-transmit/clear-to-transmitsequence is needed before messagetransmission begins. This requirement is independentof the routing algorithms employed in the network. Hence,in actual practice, the setting up of a virtual circuit may not representa time penalty. single-packet meri$ages(datagrams)can be an exception. If they are ffansmitted without a clear-to-transmitsignal, they may be discardedat the destinationnode if the sink is inactive or has no receive buffers available.The destinationnode then needsto return a rejection messageto the sourceindicating the statusof the sink. In this manner, the messageitself is a requerrt-to-transmitsignal. The ARpANET useda flow conhol strategy wherein single-packetmessagesserye a$ their own requestsfor buffer storagebut multiple-packet messagesrequire preallocatedbuffers in groups of eight at the destination[6, 71. Flow control in TYMNET is implemented in a different mannerbecauseof its exclusive use of virtual circuits for all messages.Before any node in the network can send a packet to a neighboring node, it must receive a clear-to-transmit signal from the neighboring node for the particular virtual circuit. The clear-to-transmit signal is an indication that a specified number of store-and-forwardbuffers are being held in reseryefor that particular virtual circuit. After a node sendsthe specified number of packets,it sendsno more until anotherclear-to-transmitsignal is received (indicating the previous packetshave been forwarded or more buffers have been allocatedto the virtual circuit). By using the samenode-to-nodemethod of flow control at the interface betweenthe network and the users,a very solid end-to-endflow control strategy is established. The TYMNET flow control strategyis somewhatconservativein that it may allocate store-and-forwardbuffers to one virtual circuit while another virtual circuit has more use for them (possibly causinga decreasein line utilization). This conservatism. however, provides a number of useful features: l. Networkwide flow control is established automatically by the flow control within each virtual circuit. 2. Under light traffic conditions, packet flow within each virtual circuit adjusts automatically to the maximum rate that the sourceand sink can support.If only a few virtual circuits exist, a relatively large number ofbuffers can be assigned to each circuit, allowing the retum of more frequent clear-to*transmitsignals. 3. If a sink stopsacceptingpacketsfor somereason,this condition propagatesback to the sourceby way of cessationof clear-to-transmit signals. Thus the source

SWITCHING 469 10.2 FACKET

stopssendingwhenall allocatedbuffersarefuIl. This principleof operationis sometrmes referredto asbackpressure. lockups 4 . As long as a nodeneverovercommitsits buffers,store-and-forward buffers cannotoccur.Ifseveralsinksstopacceptingdata,the$tore-ard-forward assignedto the particularvirtual circuits get filled and becomeunavailable. throughtheir own Othervirtual circuits,however,can maintaintransmission buffers. assigned 5 . The mechanism is fail-safe in the $ense that positive indications (clear-to-transmit signals) are needed before packets are forwarded to flow controlbandwidth neighboringnodes.Whennetworklinksareoverloaded, requirementsare minimal. If flow control signals stop altogether,packet transmission stops.

10.2.5 X.25 a public packetin 1976for accessing X.25 is an ITU standardprotocolestablished switchingnetwork.The datalink layer of X.25 is link acce$sprocedurebalanced (LAPB) usinghigh level datalink control(HDLC). HDLC is a bit-orientedprotocol DataLink Protocol(SDLC)established basedon thepreviouslydefinedSynchronous NetworkArchitecture(SNA) networks.HDLC hasbeconte by IBM for Synchronous The applications. thedatalink protocolof choicefor numerousdatacommunications balancedaspectof LAPB refersto a peer-orientedoperationbetweenthe two endsof modeof operationa primarynodecontrolsthelink for one thelink. In anunbalanced nodes. or moresecondary Thebasicformatof anHDLC packetis shownin Figure10.5.As indicated,packets aredelimitedby a staflingandan endingfiag (01I I I I 10).The datastreambetween the flagsis precludedfrom includinganinadvertentflag by azerobit insertionalgorithm. Whenevera stringof five ls occurin thedatastreama 0 is insertedby the source node.The receivingnoderemovesa 0 occurringafterfive ls. If a 1 is presentafter five ls, it mustbe the sixth I of a flag andis thereforenot removed. of 8 bits. in increments Theaddress field is typically8 bitslongbutcanbeextended typiThe informationfield canbe anynumberof bits long,but specificapplications cally define a maximumlength.The informationfield is nonexistentin a conhol packet.Theframecheckfield containsa l6-bit CRCbut canbe 32 bits long in some

Fleg

AddruE$

Control

lnformation

I Frums ch6ck

Flag 0 11 11 1

01111110

0

N(S)

rtf

N(R)

l-forffat mntrol fi6ld

Figure 10.5 HDLC frameformat,

47O

DATAANDASYNCHHoNoUSTFANSFER MoDENETWoRKS

applications.The control field is 8 or 16 bits long. Informarion in the control field is formatted in various ways dependingon whether the packet is an information packet or a control packet.control formats are designatedby a leading l. Figure 10.5 shows the format for a information packet (I-format) as designatedby the leading 0 in the control field. The P/F field is a single bit used for polling. The N(S) and N(R) fields contain sequencenumbersusedfor both error conffol and tlow control. The length of N(s) and N(R) is 3 or 7 bits dependingon the size of conhol field in use.* For illusftative purposes,the following discussionassumesthe use of 3-bit fields. Every time a source sendsa packet containing information, it increments N(S) modulo 8. When a destination receives a packet with a valid check sum and the next expectedsequencenumber, it returns that sequencenumber to the source in its outgoing N(R) field. Thus, a sourcenode knows that the destinationhascorrectly received every hansmitted packet up to and including the packet with the sequence number N(R) in its incoming packets.If one or more packetshave been received in error, the value of N(R) returnedto a destinationdoesnot changeuntil the sourcetimes out and reffansmits the packets beginning with the last received N(R) value plus I. [All packetsafter a lost or comrpted packet are retransmittedbecausethe destination ignores packetswith an out-of-sequencevalue in the incoming N(S)]. with a 3-bit length for N(s) a sourcecan have a maximum of sevenpacketsoutstanding at any particular time. (A particular sequencenumber cannot be reuseduntil an acknowledgmentfor the previous one has been received.) Beyond the maximum number of outstandingpackets determinedby the length of the sequencefields, particular applicationsmay be configured for a smaller number of outstandingpacketsfor flow control and reducing the amount of memory neededby a sourceto hold copies of unacknowledgedpackets.Whateverthe value of the maximum number of outstanding packets is, a destination is expectedto be able to receive that number as a burst. The destinationdoesnot have to immediately acknowledgea valid received packet if it is congestedand wants to free up rerrourcesbefore receiving more packets.If a system is configured with a maximum number of outrrtandingpacketsthat often exceeds the capacity of a receiver,the systemworks but the sourcespendsa lot of time retransmitting packets that were discarded by the destination becauseit had no resources (memory) to store them, x.25 permits a user on an x.25 network to communicatewith a number of remote locations simultaneously. connections occur on logical channels of two types: switched virtual circuits (svcs) or permanentvirtual circuits (pvcs). svcs require a connection establishmentprocessbefore data can be transferred.A PVC is similar to a leasedline in that the connectionis permanently establishedby network administration. Therefbre, data can be sent without connection setup. To establish a connection on an svc, the originator sendsan x.25 call request packet, which includes the addressof the remote destination.The destinationdecides whether or not to acceptthe call (the call requestpacket includes the originator's address and possibly other information that the destination can use to decide to accept -Use

ofHDLC on high-speedfiber links has led to the use ofeven larger sequencenumbers becausea large number of frames can be in fiansit on the link.

SWITCHING 471 10.2 PACKET

the call). A call is acceptedby returninga call acceptedpacketor rejectedby returning a clearrequestpacket. 10.2.6 Frame Relay functions As local areanetworksandfile $erversevolvedto providedataprocessing LANs at separate sitesbecame within a corporation,somemeimsof interconnecting LAN interconnection Figure10.6illustratesthreewaysof accomplishing necessary. to implementa colporatewideareanetwork(WAN). Thefust methodutilizesa public betweentheLANs. This approach (X.25)packet-switching networkto relaymessages is a usefulonefor limitedamountsof inter-LANtaffic. It alsofacilitatesinterconnecttokenrings,or tokenbuses)ing two or moredissimilarLANs (suchasEthernets,

Public packet network (a)

LAI{ I

Lsas€d digital lire

LA}-I2

(b)

Frame relay network (c)

methods:(a) through x,25 packet-switching Figure 10.6 LAN-Io-LAN interconnectron netowrk;(b) throughleasedline; (c) tfuoughframerelaynetwork.

472

DATAANDASYNCHRoNoUSTHANSFERMoDENETWoRKS

becausethe X.25 interface of each LAN acts as a data link layer protocol converter. The main drawback is the cost and slow responsefor high-bandwidth ffansfers. The secondapproachof interconnectingthe two LANs is to use a leasedline. obviously this approachis more cost effective if large amountsof dataare transferredbetween the two locations. The LANAVAN interface equipment used to connect to the leased line will vary depending on the nature of the individual LANs and how the LANs and WAN are administered.Functional possibilities for the LAN-WAN interface are bridges, routers, or switches.The speedchosenfor the leasedline inevitably requires compromising performanceobjectives with cost. The third method of interconnectingthe LANs is to use a frame relay service of a common carrier. Framerelay serviceswere developedspecifically to addressLAN in_ terconnection and are available from all major common carriers. Frame relay is a packet-rtwitchingprotocol but is faster than traditional X.25 networks becauseit does not provide error control. Error control at the data link layer (e.g., HDLC of x.25) requires receipt of an entire packet before it can be forwarded from one link to another. Frame relay also supports shortenedaddressprocessingwith a data link connection identifier (DLCI) field in rhe headerwhich identifies a pvc. The frame relay prorocol is defined by the ANSI Tl.618 standardand ITU Recommendation e.922. The PVC feature of frame relay permits the functional equivalent of a leased-line connection but much more cost effectively for high-rate, bursty traffic. Furthermore, a single frame relay accessdevice (FRAD) can achieve connectivity to multiple remote LANs by using a different DLCI for eachremote LAN. Frame relay nodescan processthe simpler addressformat and switch an incoming messagedirectly to an outgoing line as it is being received. The elimination of error controlat the datalink layer of a network protocol reflects the fact that the underlying transmissionfacility, fiber, is essentiallyerror free. In this environment it is more efficient to move effor control to a higher level (e.g., an application layer). when infrequent errors occur, the application can invoke error control as appropriate to the application. Example 10.1. Determine the amount of transmission capacity required to refransmitframesof1000bitsof datainanetworkwith l0tandemlinkseachof which has bit error probability IO"t. (a) Assume link-by-link enor conrrol. (b) Assume end-to-endenor control. (c) Repeatthe calculationsfor a bit error probability of l0-5. Solution, (a) with link-by-linkerrorcontroltheprobabilityof abit errorin a frameis 1000x

10-8= l0*s. Theexpectednumberof bits of transmission capacityrequiredto rerransmiris l0-5 x 1000= 0.01bit/link. (b) with end-to-end error conffol the probability of a comrpted frame is 10x tO-s = l0+. The expectedretransmission capacityrequiredis lO-a x 1000= 0.1biilink.

CRCsareincludedin framesof framerelay data,but a framerelay nodedoesnot requesta retransmission if an error is detected.

10.2 PACKET SWITCHING 473

(c) With a bit enor rare of 10-5, the respective calculations are 10 bits/link for link-by-link conffol and 100 bitsflink for end-to-endcontrol. Example 10.I illustrates that with very Iow bit error probabilities both forms of error conffol require insignificant amounts of transmissioncapacity for frame retransmissions.Thus, from a transmissioncapacity point of view there is no reasonto use linkby-link effor control. On transmission links with higher rates, a$ in Tl lines with marginal error perfbrmance,link-bylink error control is obviously desirable,pruticularly in a packet network where packetsmight be larger than 1000 bits. 10.2.7

TCP/IP

The original networking protocol of the ARPANET was designatedNetwork Control Protocol (NCP). While this protocol addressedthe needsof a single network, it could not be usedas a meansof communicating from one network to another.The desireto develop a protocol that would allow communication from a node in one network to a node in another network led to the eventual development of Transmission Control Protocolflntemet Protocol (TCP/IP) [8]. TCP/IP replacedNCP in ARPANET hostson JanuaryI, 1983. TCP/IP is a protocol that can be used within a network, but more importantly, it can be used as a prOtocolto Communicatehetween two netwOrks'"Thus, the Internet is composedof many networks that communicatewith eachother using TCP/IP. Foremost among these many networks are LANs. The Internet is yet another way for interconnecting LANs. ftr conftast to fhe creafion of a corporate WAN using frame relay, LAN interconnection via ttre Internet is primarily intended to crossco{porateboundaries. As defined by the Internet Engineering Task Force (IETF), TCP/F usesvariablelength packetsthat are processedwith basically one class of service:best-effort routing. A user datagramprotocol (UDP) replacesTCP when real time servicessuch as voice or video are carried. UDP has no reffansmissions.Before Voice on the Infernet (VoIP) can be realized for somethingapproachinguniversal service,the routers must be extencledto support a new class of service: one with controlled delay, delay variation, and delivery guarantees.A first attempt to addressthe quatity-of-service issue was a propo$edmodification of TCPIP with a Reservation Protocol (RSVP). This was determinedunsatisfactorybecauseit required changing all nodes in a ap-proac:h "differentiated service" network simultaneously [9]. A more practical approachis a (DiffServ) enhancementto TCP/IP. When considering the viability and expected quality of service of packetized speech,it is worthwhile to consider the difTerencebetween IP telephony and Internet telephony. IP telephony refers to carrying packetizedspeechin an IP network. Ifthis network is wholly containedwithin one organization,commonly referred to as an intranet, it is more likely that a relatively high level quality of service can be provided. Packetizedspeechcarried over the Internet that traverse$multiple, autonomou$net*Th"

m,r*t *pp*unt aspectof TCP/IP to a uscr is the Universal ResoutceLocator (URL), with the following format: filetype://www.address.tld/resource.The address.tld (.top level.domain) identifies a location on the web while everything after thc back slash identities a resource(e.g., a frle) inside the location.

474

DATAANDASyNcHRoNousTRAN$FEHMoDENETWoHKS

works is much more difficult to rnanage,particularly when economic considerations for allocation ofcosts and revenuesare considered.For a description ofmany ofthe technical considerationsforpacketized voice seereference [10].

10.3 ASYNCHRONOUSTHANSFERMODE NETWORKS In contra$tto TCP/IP,theAsynchronous TransferMode(ATM) networkarchitecture incorporated featuresfor supportingreal-timetraffic suchasvoiceandvideoin theinitial implementation. Theprincipalaspectsof ATM directedto real-time$uppofiare short,fixed-sizedpackets(cells),shortheaders, andno link-byjink errorcontrol.* ATM is a standardized architecture of packet-oriented transmission andswitching originallyproposedfor a BroadbandIntegraredservicesDigital Network(BISDN). ATM hassincebeenexpandedin scopeto supporta wide varietyof seruicetypes; wideband,narrowband, bursty,non-reartime,andrealtime.The synchronous TDM (circuit-switched) networkthatevolvedprimarilyfor voiceservicessupportsthesame serviceswith externaladaptations, but theadaptations comefrom u u*i"ty of suppliersnecessitating theneedfor multiple,nonintegrated, nonsiandardized equipmentand support.ATM standardizes thewide rangeof servicesby definingquality-of-service re_ quirementsfbr variousnaffic qrpes.The quality-of-servicepu*-rt ru specifically in_ tendedfar voiceservicesaremaximumdelay,delayvariation,andcell losi probability. 10.3.1 ATM Cells TheATM transmission formatconsistsof fixed-sizecellsof 53 byteseach.As shown in Figure10.7,therearefive overheadbyteswithin eachcell, wtrichteadsto 4g bwes of payloadper cell' The useof fixed-sizedcells for all applicationsfacilitateshardware-onlyimplementations of switchingfunctions(queuingandtransfer;. Cell Loss Prlority Thecell losspriority (CLP)bit in thecell headeris usedto identifyrwo basicclasses of servicewith respectto discardingof cellswithin a network.when a cell with a cl-p valueof I arrivesat a heavilycongested networkelement,thatcellcanbediscarded to relievecongestionfor higherpriority traffic (e.g.,cellswith cl-p = 0). cells with a cl-p valueof 0 areneverdiscardedunlessatl cellswith lowerpriority areaheadydiscarded. 10.3.2 ATM Service Gategoriee ATM servicesarecategorized (by theATM forum Il l, 121)into thefollowing caregories relatedto the statisticalnatureof thedataraterequirements of therespectivesourcesand thequalityof selvice(Qos) thatthenetworkcanprovidefor those$ervices: -TCP/IP

also defets eror control to the application but has variable-sized packets and a relatively complicated header format.

MODENETWOHKS 476 TRANSFER 10,3 ASYNCHRONOUS

8

7

6

Bits 5 4

3

2

1

VidualPathldentifier

FloruControl VirtualPath tdentifier

3

ViduafCharrnelldentifier PayloadType

CLF

4

HeaderEnorCheck

5

Payload

I

Payload

Byr"*

5S

FlgureI0.7 ATM cellformat;CLP,cell lossprioriry' (CBR) services Constant-bit-rate (VBR) services Variable-bit-rate Availablebit rate(ABR) services Unspecifiedbit rate(UBR) services Con stant-Bit-Bate Servl cee a specifieddatarateassuming An applicationrequestinga CBR serviceis guaranteed the servicerequestis accepted(i.e.,ttrenetworkcanpreallocatesufficientresources TDM service.CBR for therequest).Thus,a CBR serviceis similarto a synchronous servicesare specificallydesignedfo emulatecircuit switching.A CBR application send cannotexceedthe specifiedrate(i.e.,thepeakcell rate)but neednot necessarily dataat the specifiedrate.In this sense,CBR servicesaredifferentfrom synchronous TDM services-if a sourcehasperiodsof time in which thereareno dataavailable, the channelcapacitymay be releasedfor otherapplications' Va riable-Eit-Rate Se rulcee realtime (rt-VBR) andnoninto two distinctcategories: VBR servicesareseparated with tightly conbursty are for applications (fft-VBR). rt-VBR services The realtime or The main QoS as voice video. sqch variation requirements delay straineddelayand

476

DATAANDASyNcHRoNousTRANSFEFMoDENETwoRKS

parameterof an rt-vBR serviceis the maximumcell transferdelay(maxcTD). The nIt-VBR servicesarefor burstydatatransmission applications with no particulardeIay requirements. wheneveran rt-vBR or nrt-vBR servicerequestis accepted, the foltowingthree parameters areagreeduponfor sourcetransmission statistics: Peakcell rate(pCR) Sustainable cell rate(SCR) Maximumburstsize(MBS) As long asthe sourceadheresto the abovestatisticalparameters,the networkis committed to carryingthe traffic. Avallable Blt Rate Serulces ABR servicesareintendedfor non-real-time applications thatoperatebetweena minimum andmaximumdataratespecifiedby a minimumcell rate(MCR) parameter and a PCRparameter' An essential aspectof ABR servicesis theneedfor feedbacksignals within thenetworkto controltheflow of datafrom theABR sources. Thenetworkdvnamicallyallocatesadditionalresources to ABR services(beyondthosefor MCn,up_ port)whentheadditionalresources areavailableandsoinformstheABR sources with thefeedbackmechanism. Unspeclfled Blt Rate Seruices UBR servicesareinherentlyintendedfor non-real-time applications whereinno ripecific qualityof serviceis desiredor implied.Neitherdelayparameters nor cell lossratios (cLRs) are specified.If a networkbecomesoverly ctngested,cells from uBR applicationsarediscardedby the network(asopposedto confrollingthe flow of the UBR source)'when a UBR applicationnegotiatesa connection,a pcR parameter valuemay be providedby the network,whichprovidesthe applicationwith an indicationof thethroughputthatcanbe expected. Thefollowinginoicateswhicheos pa_ rametersarerelevantto theATM classesof service.

QoSParameter Gelllossratio(CLR) Celltransitdelay(CTD) Celldelayvariation (CDV) Peakcellrate(PCFI) Sustained cellrate(SCR) Bursttolerance(BT) Flowcontrol

CBH

N-VBH

X X X X

X X X X X X

nrt-VBR

X X X X

477 10.3 ASYNGHHONOUSTRANSFERMODENETWORKS

10.3.3 ATMGonnectlong ATM networksareconnectionoriented,which meansthat a sourcemustmakea rearoutethrough questfor servicebeforetransmittingcells.Networkcontroldetermines probabilistic in response (in sense) thenetworkandallocatesresources a statisticalor mode in a LANs typicallyoperate connectionless to eachservicereque$t.In cOntrast, need the arises' datato a destinationwhenever whereina $ourcesendsunconskained (VCCs)and virtualchannelconnections ATM supportstwo typesof connections: (VPCs).A VCC is a virtualcircuitbetweentwo user$reprevirtualpathconnections sentinga singlestreamof cells.WhentherearemultipleVCCsbetweentwo common a VPC canbecreatedasa bundleof VCCs.Theuseof VPCssimplifiesnetendpoints, andthe operationsof intermediatenodesof an ATM networkbemanagement work processed as a singleentity as opposedto individualVCCs.Thus,a causea VFC is network' to a trunk groupin a circuit-switched VPC is analogous 10.3.4 ATM Switching of an ATM switchingfabricis to transfercellsarrivingon one Thebasicrequirement beforean physicallink to anotherlink. Becauseincomingheadersmustbe processed incomingcell can be fiansferredthrougha switchand becausean outgoinglink is likely to bebusy,ATM sffucturesalwaysprovidesomeamountof delayin the switching process.Thus,ATM switchesprovideboth spaceandtime switchingfunctionsin a similarmannerasTDM circuit switchesdescribedin Chapter5. A significantdifference,and complication,of an ATM switch servicingstreaffr ATM cellscanartraffic suchasvoiceor videois that,in contrastto TDM channels, that a parthe greatly increases likelihood phenomenon rive at irregularintervals.This The network. of a switching link blockingon a sharedinterstage ticularcell encounters until cells queue for blocked basicsolutionto this problemis to addcell buffersa$a path is free. Simplefirst-in, first-out (FIFO) queuingoperationsare the necessary in hardwarebut betternetworkefficiencyoccursif theFIFOdisreadilyimplemented ciplineis modifiedto allow a cell in thequeuewith anopenpathto bypassablocked aredesirable. ennl fl3, 141.Thus,morecomplicatedhardwareimplementations The principalaspectsof an ATM switchthat are differentfrom a TDM circuit switchare: incomingtime slot soheaderinformationneedsto be t . Thereis no preassigned to know whereto transferincomingcells' extractedandprocessed outgoingtime slot for outgoingcells so the delay 2. There is no preassigned with waiting involvedin transferringcellsis a variablequeuingdelayassociated until theoutgoinglink is idle. The headerinformationin an outgoingcell is generallydifferentthanwhat is receivedsotheswitchmustinsertappropriateheaderinformationinto thetraffic stream.

478

DATAANDAsYNcHHoNoUSTHANSFERMoDENETwoRKs

ATM ltlemory Switch As is thecaseof TDM circuitswitches, ATM switches usememoryswitching to the maximum extentpo$sible. Aslongasthememoryspeeds support thetraffic,themost

economicaldesignis onein which all traffic is writteninto anareadfrom a cornmon memory.In additionto the switchingfabric,a high-speed ATM switchmustimple_ mentothercell processing functionsin hardware.Thesefunctionsareheadere*t action,headererrorchecking,tablelookupfor routeinformationusingreceivedvirtual pathindicator(VPD or vcc, recordingof traffic usagestatisti"r,""ll queuing/rejec_ tion, insertionof outgoingvpl/vcc, andoutgoingchecksumgeneration. Example10.?. Determinethe memoryspeedrequiredfor an ATM switchfabric using a shared-memory archite*urein supportof 12 srs-3 (oc-3) bidirecrional ports' Also determinethe numberof voice connectionsthat can-becarriedby this switch.Assumethatanactivevoicesignalrequires12kbpsandthattheactivity factor is 40Vo. solution. Exceptforthegbytesofpathoverhead, alrbyresofan srs-3 payloadcan be usedfor ATM cells. Thus, each srs-3 input rink provides 260 x g/s3 44 cells/frame.(A partialframeis not carriedsocell boundaries canbe alignedwith the startof thesrs-3 payload.)If a S3-byte-wide memoryis usedfor storingthecells,rhe total accessratefor l2 bidirectionalporrsis 12x z x,44 xg000 = g.++t rralt. sinceeachATM cell provides4g bytesofpayload,the numberofvoice connec_ tionsthatcanbe supported by onesTS-3 is a4 x 4g x g x g0fi)/12,000/0.4 = 2g,160 connections. Thetotalcapacityof the switchis 337,g20voicecalls. It is interestingto comparetheresultsof Example10.2with a circuitswitch design for 64-kbpsvoicechanners. using a standardsoNET multiprexingformat,a single srs-3 caniesthreeDS3 signals(2016voicechannels).sirnitarty,*asrs-3 (e.g., an srM-1) carriesa singleB$ signalwith 1920voicechannels. In eithercase,thecapac_ ity of eachATM link is over10timesthecapacityof theconventional circuit-switched architectureusingthe samedatarate.The advanfege comesfrom a combinationof compressing the speechandtakingadvantage of voiceinactivity. A memoryswitchfor 12x 2016= z4,rgzchannelsrequiresan accessrateof 3g7 MHz, whichis muchgreaterthantheg.44g-MHzaccess rateof theATM switch,even thoughthetotalcapacityis muchless.Thereasonfor thehighaccess rateofthe circuit switchis thatonly 8 bitsarebeingswitchedat a time.rr muttiptebyteswereaccumulatedand switchedthrougha wider memory,asin the eru case,ttreaccess rateof the circuit switchwouldbe reducedaccordingly.circuit switcheshavenot been im_ plementedwith wider ffansfersto preventthe insertionof extra circuit delay.Voice interfacesto ATM networksnecessarily requireanechocancelerto accommodate the delay. Example10'?is a nonblockingswitchasfar asinternaloperations of theswitchare concerned'However,if a largeburstof kaffic arrivesat oni particularoutputport, it maybe thatsomecellsarediscardedbecause thequeuefor thatparticularport is full.

4Ig 10.3 ASYNCHRONOUSTRANSFERMODENETWORKS

Noticethat if traffic is beingqueuedfor someports,otheroutgoingportsareoperating belowtheircapacity. The amountof variationin the traffic flow on outgoingATM links is a directfunction of the connectionadmissioncontrol(CAC) processinvolvedin settingup conareVPCsandtheseVPCshaverelativelyconstantarrival nections.If all connections rates(becausethe aggregatedatarateof multiple, independentvariable-ratechannels of thenetworkwill beefficientlyutilized resources thetransmission is fairly constant), traffic arehighly variable,largerqueues(and queue If arrivals sizes. the with modest link utilization. to largerdelays)arerequired achievehigh transmission consequently given priority so types are Whenthereis a mixtureof traffic types,the time-critical cells. cell lossandlong delaysareconfinedto non-real-time-critical Spac*Memory Switch The implementationof large ATM switchesrequiresmultiple stagesof spaceand timesswitching(e.g.,buffering)in the samemannerthatlargetime divisionswitches involves Oneparticularlyusefularchitecture in Chapter5 areimplemented. described queues' aS shown by output followed separate switch as a TDM $pace front-end bus a (S/P) ofthe circuits to series-to-Parallel is synchronized inputbus in Figure10.8.The must the bus Thus, every cell. for received input portssothat a time slot is assigned table via a is decoded header operateat rhetotal speedof all incomingsignals.Each for recells outputsotheoutputmemoriesstoreonly look-upto selecttheappropriate spectiveoutputports. The main attractionof the multiple-output-memoryarchitectureshownin Figure broadcasting 10.8is its ability to scaleto a largerangeof switchsizes.Furthermore, by havingtheheaderdecodelogic enablesimulandmulticastingareeasilysupported doesnotdo much writesinto multiplebuffer$.Unfortunately,thisarchitecture taneou$ theoutputport only to support need problems. Output accesses speed to solvememory slots on the bus. time successive suppofr must rate,but input accesses transmission acting caches high-speed with Memoryspeedrestrictionscanbe somewhatalleviated in Figure shown asspeedbuffersbetweenthebusandlarger,sloweroutputbuffels,as a sourceofcell losswhen 10.9.Thecachesdo notaddmuchcell delaybutdorepre$ent outputline' of a burstto therespective onefills up because

Figure 10.t Input bus/outputbufferATM swirch.

480

DATAANDASYNCHBONOUS THANSFEHMODENETWORKS

Figure 10.9 Inputbus/outputbufferATM switchwith caches.

Memory-SpaceSwitch Anotherversionof a two-stage ATM switchis shownin Figure10.10wherein all in-

coming cells are available to all output ports via separatepaths for each input. Each input path in each output port module is terminated in a separatecell buffer. Associated with each buffer is a table look-up function that selectsonly those cells destined to the respectiveoutput port$. Output data are selectedfrom the $eparatebuffers as a cornmon queue,thereby pedorming a spaceswitch function. The main advantageof the configuration shown if Figure I 0. I 0 is that all data paths and memories operateat the external link speed.The basic disadvantageis the need for AP buffers (and AP headerprocessors),where N is the number orplrts. cell loss occurs when an incoming cell encountersa full buffer. The probability of this occurring is conkolled by how much traffic for a particula-r output is accepted from a particular input. The use of multiple, individual queues leads to smailer group sizes, which leadsto more total memory for a given cell loss probability. However, because the memories operateat relatively low speeds,the cost impact is minimizea. Notice that delay probabilities are determined by the total traffic ur.*pt-d for an ourput port in the samemiutner as previous con_figurations.*

Mem ory-$ pace-l,ilemory Switch

ATM switcharchitectures with singlestagesof spaceandmemoryswitchingareimpracticalin termsof speedand/orhardwarerequirementswhenvery large$ysrem$ are desired.As is thecasewith circuit-switching architecture$, morepractic-aVeconomical structuresrequiremultiple memorystagesor multiple spacestages-or both. one such architectureis a counterpartto a TST switch, the memory-space-memory (MSM) implementation shownin Figure10.I L Theinpurmemorystageof anMSM bufferscompletecells of informationthat arequeuedfor transfeithroughthe space stageandthenbufferedagainfor transferfrom the outputmemory.successivecells of aparticularconnectionarenotnecessarily transfenedthroughthi spacesmgeat any palticularpredetermined time. As eachcell is receivedinto an inpui buffer,a queue entryis madefor rheappropriate transferthrough thespacestage.Supportormuitiple *This

assumesthat the separatebuffers associated with each output link are managed as a single FIFO queue for output purposes.

481 10.3 ASYNCHFONOUSTBANSFERMODENETWORKS

ATM swirch. Flgure10.10 Multiple-memory classesof servicereqUiremultipleinput queues(usuallysharinga coilunonmemory element). Beforea cell in an input buffer canbe transferredthroughthe spacestage,the desiredouput stagemustbe availablefor input.If the desiredoutputmemoryis busy, the cell carnotbe transferredandhead-of-lineblockingoccurs.If anothercell in the input queuecanbe selectedwhenoneor moreof the head-of-linecellsareblocked, the impactof the blockingis minimized.Suchoperationsareea$yto implementin computer-controlledstore-and-forwardswirchesbut must be implementedin hardware in a large,high-speedATM switch. An altemativeapproachto determiningif blockingexistsbeforea hansferis madeis to blindly fransfercellsthroughtheswitch andhaveblockedcells reenteran input polt though a recirculationpath,asshownin of sucha systemis very Figure 10.12.Determiningthe delay/celllossperformance complicatedanddependenton theprobabilityof blocking andthenumberof cellsthat canentertherecirculationbuffer(s).

ATM switch. Figure 10.11 Memory-space-nremory

482

DATAANDASYNCHRoNoUSTRANSFEFM0DENETWoRKS

Figurel0.l? Memory-space_memory switchwithrecirculation. For bestperformance, theinputcellsfrom all input linesof anMSM switchshould be distributedacrossall input buffersusing a front-endshuffledistributionswitch [15]' This processdecorrelates the enhiesin an input queuefrom the cell arrivalsof anyoneparticularinput line. Recallthat a similaroperationis advantageous in TST switchesto reduceblocking probabilitiesfor individual channelsof a busy trunk group. Thedesignof a spacestagefor anMSM switchmatrixcanfollow anyoneof a va. riety of approaches, includinga single-stage squarematrix or a multipie-stage clos network,asdescribedin chapter5. with eitherof thesedesignsthe probabilityancl treatmentof outputblockingcanbe minimizedby runningthe spacestagefasterthan the cell rateof the extemalports(i.e.,with the MSM equivalentof time expansion). The existenceof expandedopporhrnities for spacestagetransfersreducesthe probabilityof contentionfor anoutputbufferandminimizesthedelaywhencontentionoccurs.A very significantconsideration in the spacestagedesignis the complexityof hardwarecontrol logic, particularrywhen multiple service Irioriti", and multiple pathsthroughthe switch areaccomrnodated. Banyan Network Swltch one approachto implementingmodularspacestagestructuresis a banyannetwork U6l. As indicatedin Figure10.13,abanyannerworkis implemenred by inrerconnecring 2 x 2 switchingelementsin multiple,recursivestages.Noticethatthe g x g skuctureis implemented astwo 4 x 4 structures andanadditionalfront-endsrageto select eitherofthe two 4 x 4 structures from eachinput.Repeated replicationsandfirst-stage additionsallow implementation of arbitrarysizednetworks. A banyannetworkprovidesfull accessibilityin thatanyinputcanbe connected to anyoutput.Noticethatthereis a uniquepathbetweenanyinputandanyoutput.How-

483 10.3 ASYNCHRONOUSTRANSFEFMODENETWORKS

-l-

-Et*

r-t

Cross

Slraight (a)

(b) Figure 10.13 Banyanswirchingnetwork:(a) 2 x 2 switchingelementconnectionstatesl(b) 8 x I network.

to thefirst outever,thenetworksarenot nonblocking.If thefirst inputis connected to oneofthe bottomfouroutput,for example,thesecondinputcanonlybeconnected puts.Onemethodof overcomingtheblockingaspectsof thebanyannetworkinvolves cascadingtWosuchsectionsto producewhat aregenerallyreferredto asBenesnetworks[17]. A Benesnetworkis strictlynonblockingif rearangementof connections a Batcher of a banyannetworkis to concatenate is utilized.Anotheraugmentation the process orders The front-end sorting a network. banyan sortingnetwork[18] with modiAnother blocking. throughtheswitchwithout cellssotheycanall betransferred blockingproblemsis to addparallelnetworks ficationto a banyannetworkto address referredto as switch planesto provide more pathsto an output.This latter approach in Chapswitchdescribed is identicalto addingmorecenterstagesto a multiple-stage thesimplicityof uniquepathrouting. compromises thisapproach ter 5. Unfortunately, The basicattractionof a banyannetworkis the modularsimplicityof its control. When a cell it to be transferredthroughthe network,the table look-up processproofbits for internalroutingthroughthenetwork.Thefirst bit defines ducesa sequence theselectionof thefirst-stageelement.Thesecondbit deflnestheselectionof thesecond stageand so on. The routing information is appendedto the front of the cell and is transferred throughtheswitchonestageat a time,with leadthecompositemessage ing bits deletedasthey areused.Thus,a banyannetworkutilizesdirectprogressive in Chapter switch,described step-by-step controlreminiscentof anelectromechanical 5. Unfortunately,modificationsto the basicbanyanarchitectureto provide multiple pathscompromisesthe simplicity anddistributedaspectof the hardwarecontrol. A wide variety of otherATM switch architecfureshavebeenproposedanddeveloped[19-23].

484

DATAAND ASYNcHRoNoUS THANSFER MoDENETWoHKS

10.3.5 ATMApplicarions Although the basic ATM architectureis primarily directed to servicing data communications, the ATM architectureincludes provisions for other servicessuch as voice, video, and multimedia. Supportfor the various servicesis included in various versions of ATM adaptationlayers (AALs). service adaptationoccursat the periphery of a network in edgesswitches.Internal nodesof an ATM network (core switches)are unconcerned with the nature of the traffic other than supporting the eos objectives of the generalclassesof service.The role of AALs are shown in the functional layers of Figure 10.14 of the (ATM) Multi-service switching Forum (MSF) The generat [?4]. classesof service and the adaptationlayers to support theseservicesare identified as follows: AALI: Transport of cBR rraff,rc (program audio, video, and emulation of TDM-basedcircuits (i.e.,DSl, El). AAL2: Protocol standard for low-bit-rate and time-dependentvariable-bit-rate (VBR-rt) connection-orientedtraffic (compressedvoice anclpacketizedvideo). AAL3/4: Protocol standard for supporting both connectionless and connection-orientedvBR traffic [primary application, switched Multimesabit Data Service(SMDS)1. AALS: Transport of vBR traffic anclsignaling messages(packet data, Ip, frame relay, LAN emulation).

CircuitEmulatlon Serulces A circuitemulation service (cES)t25lis usedto transport TDM-based channels such asDSO,DSl, or El signals. cESsareprocessed according to anAALI protocol. A Applications Gontrol plf,ne Switching plane Adaptation plsne

Extemelinterfaces Figure 10.14 ATM functional implementation layers.

485 10.s ASyNcHFoNoUSTHANSFERMoDENETWoBKS

critical aspectof a CES serviceinvolvesrecoveryof the sourceclock at the egress point of the ATM network.Figure 10.15depictsthreedifferentclock recoverysceThe mostdesirable nariosdependent on the natureof the networksynchronization. situationis shownin Figure 10.154,in which the entirenetwork,includingthe data to a commonprimaryreferenceclock (PRC).In this casethe source,is synchronized output dataclock is derivedfrom the PRC andthereforehasthe sameaveragevalue asthe source.As long astheATM networkmeetscell delayvariation(CDV) criteria, the cell buffer in the egressAALI equipmentabsorbscell arrivalvariationsandno dataslipsmayoccur. dataarelost.If cell arrivalvariationsexceedCDV requirements, As is the casein TDM networksthe slipsshouldbe controlledto consistof integral numbersof framesof theunderlyingdata(e.g.,193bitsfor DSI or 256bitsfor El) to precludelossof framingin the underlyingcircuit. Figure l0.l5b showsa networkconfigurationwhereinthe ATM networkis synchronizedto a commonPRCbut the datasourceis timedfrom someotherreference. residualtime stamps(SRTSs). For this situation,ATM networksutilize synchronous overhead ofa cell datastreamat an ingress SRTSsareperiodicallyinsertedinto the AALI. TheegressAALI candeterminetheaveragedatarateof thesourceby dividing SRTSsby thetimeintervaldetheamountof sourcedatareceivedbetweensuccessive finedby the SRTSs.Noticethatthe useof SRTSsdependson the ingressandegress to a commonPRC. equipmentbeingsynchronized casein which the ingressand egress Figure 10.15cdepictsthe most undesirable (PRCI andPRCZ).In this case,adaptive references nodesarecontrolledby separate clockingis utilized to recoverthe sourcedataclock. Adaptiveclockinginvolvesa phase-locked loop (PLL) with the fill levelof thecell bufferprovidinga phasemeasA urement. very low bandwidthfilter is usedto determinethelong-termaveragefill the recoveredoutputclock of the VCO. Jitterin the level,which in tum establishes andnarrow-bandwidth derivedclockis easilycontrolledhy usinga large-cell-buffer (wander) caused by irregularcell arvariations PLL. However,low-frequencyclock clock. Dependingon the appli rival timesareunavoidablyransferredto thederived problems. If a DSI signal,for cation,thewandermay or may not causedownstream example,is immediatelydemultiplexedinto analogsignals,no impairmentsto voice voicebandmodemsmaybe adverselyaffected(bechannelswill arisebut high-speed phase).If theDSI signalis insteadpassed to thecarrierreference causeof impairments to a switchingmachine,thewandermaycontributeto slipsin theelasticstoreinterface of theswitch.Again,slipsdo not impairvoicesignalsbut significantlycomrptvoicebanddata. LAN Emulatlon theATM applications, To providecompatibilitywith LAN protocolsandestablished Forumcreateda protocol,calledLAN Emulation(LANE), which defineshow Ethernet(IEEE802.3)andtokenring LAN (IEEE802.5)datacanbecarriedoveranATM network.Figure10.16showhow anATM networkcanprovideLAN-to-LAN connec(ATM to thedesktop).Thecontivity aswell asconnectivitydirectlyto workstations nectionlessnature of contemporaryLANs is simulatedin a manner that first

DATAANDASYNCHRONOUS TRANSFERMODENETWOFKS

(a)

(b)

lPRcTl .;

(c) Figure 10.15 Synchronization of circuitemulationservices:(a) fully synchronized network; (D)unsynchronized source(sRTsderivedclock);(c) unsynchronized ATM gateways(adaptive clocking).

487 10.3 ASYNCHRONOUSTRANSFERMODENETWOBKS

EnterpriseLAtl

usingATM LAN emulation. Figure 10.16 LAN interconnection

stationalreadyexists.If so, if a connection(SVCor PVC)to an addressed determines andthen is sentto thestation.Otherwise,anSVCis first established anATM message in everyLAN atthe message is sent.A LAN Emulationclient (LEC) is established tachmentto the ATM network.Severaltypesof LAN Emulationserversareusedto LAN functions$uchasmulticastand configuretheLECsandprovidecontemporary broadcastservices. Low-Bit-Rdte Voice in a CESmode,greatsavings Althoughvoicecanbecarriedas64-kbpsPCM channels voice. Transportof comin bandwidthare achievedby transmittingcompressed to andfrom digital cellular pressedvoiceis particularlyappropriatefor ffansmission Similarly, or aboutto becompressed. basestationswherevoiceis alreadycompressed voicechannelsbetweenPBXsof a ptivatenetwork transportof a groupof compressed As shownin Figure10.17,individualvoicechannelscanbe serviced is advantageous. (AAL2) switchedvirtual circuitsby packingmultiple framesof low-bitas VBR-rt rate$peechinto ATM cells.The particularexampleshownin Figure 10.17assumes G.729ConjugateStructuredAlgebraicCodeExcitedLPC (CS-ACELP)voicecompressionwith 80 bits per lO-msecprocessing ftame.Noticethatcell packingputs4.8 thevoicelatencyby asmuchas40 ACELPframesinto anATM cell,whichincreases msecin eachdirectionof transmission. The voice latencycanbe reducedby forming a trunk groupandpackingmultiple voicepacketsinto individual48-bytecell payloads.A singlecell canthen compressed carry four lO-byteCS-ACELPpacketsin a frameandnot adda lot of delayto the in-

488

DATAANDAsyNcHRoNoUsTRANsFEHMoDENETwoRKS

Ftgure10.17 Packing low-bit-rate voiceintoATM. dividualchannels. This approachis particularlyusefulfor fixed networktransportof digital cellulartraffic, which inherenflyinvolvesmultiple connectionsto and from basestations. Example10.3. Dererminethe probabilirythat the delay of an ATM voice cell exceed$l0 msecif the cell is carriedon a DSI accesslink to an ATM network. Assumetheaccesslink is 907outilizedon averase. solution. The solutionto this problemutilizes queuingtheory as presentedin chapter12 wheremoreproblemsof this typeareprovided.It involvesdetermining . theprobabilitythattheDSl acce$s queuecontainsenoughcellsto representl0 msec of transmission time. Assumingall but the framingbit of the DSI signalis usedto carryATM cells,theservicetime (f.) of a singlecell is determined as 53x8 = 276Psec t = rgt; Booo Therefore,l0 msecof delayrepresents lo/0.2j6= 36.2cell times.Equation12.25of Chapterl2 is usedto geta solution: P(>t)= Ps-(l*P)t/rn p(>10msec): (0.9)e(14'e)36'2 = 0.025 Example10.3indicatesthat2.5vo of the cellswill be delayedby morethan l0 msec'Tenmilliseconds of delayon anacce$s link is quiteacceptable considering there arepresumablyonly two accesslinks in a connection.Delaysof this magnitudeare not acceptable on internallinks of an ATM network,but the internallinks havevery highbandwidthsandassuchvery shortservicetimessoqueuingdelayof voicecells

489 10.3 ASYN0HHONOUSTRANSFEHMODENETWORKS

priority andthe is not muchof a consideration-aslong as voicecellsget adequate CAC limits thenumberof voicecallson a link. Thefirst The solutionto ExampleI0.3 involvesa coupleof simplifyingshortcuts. of theseis the useof Equation12.25,whichis intendedfor exponentialservicetimes in as opposedto constantservicestimesimpliedby fixedJengthcells.As discussed generally not for cell analyses are appropriate ATM Chapter12,constant-service-time A Iargefile or extendeddutraffic because the cellsarenot statisticallyindependent. ration talkspurtcreatesa burst of cells that havean effect comparableto a variablelengthmessage-hencethe useof an exponentialservicetime model. of hoA secondsimplificationof the solutionof Example10.3is an assumption mogeneous traffic-all voicecells.Multiple voice sourcesare reasonablywell behavedfrom a traffic statisticspoint of view becauseno one sourcecanproducean instantaneous burstof cellsascana file transfer.Whena mixtureof traffic typeswith a mixtureof arrivalstatisticsandserviceprioritiesexist,theanalysisis quiteinvolved andtypicallyrequiressimulation(asopposedto time-honoredanalyticalsolutions). arethequeuesize,thequeuediscipline(priOtherfactorsthatinfluencethe analyses oritiesandorderof service),cell discardalgorithms,flow controlalgorithms,andthe of a link is allocatedto realtime versusdisspecificCAC algorithm(whatpercentage of cretionarytraffic). An extensiveamount researchhasbeenundertakento analyze andprocessing algorithms.Refertheperformance of variousATM switchstructures ences[26-28] aresomegoodexamples. The conditionof 907oloadingof voice channelsin Example10.3is requiredto loadingof voicechannelsis attempted, achievea controlledamountof delay.If 700Vo queuingdelayis unbounded. Thechannelcan asimpliedby Example10.2,theaverage non-real-time traffic. The still be 1007oloadedif theadditional107ois discretionary, solutionrequiresvoicetraffic to be givenpriority in a mannerthatneversendsdiscretionarytraff,rcif thereareany voicecellsin the queue.Otherwise,the discretionary traffic impactsthe queuinganalysis.Cell lossfor thediscretionary(e.9.,besteffott) haff,rcwill be significantif a link is allocatedat closeto 100%of capacity. Traffic Shaping Traffic shapingis a termusedto denotethe controlof $ourcesof an accessnodeso thattheyconformto a panicularsetof traffic statistics(e.g.,SCRandPCR).Traffic shapingmay involveflow controlfor sometypesof fiaffic or cell discardingfor others.Becausethe statisticsof a groupof sourcesis lessvariablethanthe statisticsof with lessimpacton individual individualsources, haffic shapingcanbe implemented traffic streamsif a groupof sources(VCC$)arepackedinto a groupandshapedasa to thesourcesof theindividualVCCs(e.g., VPC.If a VPC creationprocesshasaccess 64-kbpsPCM voicechannels), additionalpossibilitiesfor traffic shapingoccur.In this processes for theindividualVCCscanbe controlledto achieve casethecompression distributedacrosseach a constafltcompositedataratewith speechqualitydegradation of thechannelsasopposedto degradation occurringto an individualconnectionasa of traffic resultof cell lossor excessive delay.For a descriptionof othertechniques shapingandan associated analysisofdelay andcell lo$s,seereferences [29 and30].

490

DATAAND ASYNCHHONOUS THANSFEFMODENETWORKS

TABLE10.1 Uncompressed Blt Ratesof DlgltlzedStandards $tandard NTSC PAL HDTV

Pixels/Line LineVFrame Frames/sec

640 s80 1920

480 575 1080

30 50 30

pixels/sec 9.216 16.675 62.208

Mbps 221 400 1493

This principle can also be applied to MpEG-z video encodingof multiple, real-time video signalst3l, 321.Maintaining a constantcompositedatarate of multiple independent channels complicates the multiplexing process becausea figure of quality most be preparedand comparedacrossthe multiple sourcesto determinewhich source is least affected by a lower bandwidth allocation. Undoubtedly, this processalso adds delay to the channelssince double-passencoding may be required.

Video Tablel0' I identifiesthreemajortelevisionstandards andthebit ratesrequiredto digitally encodethemwith no compression andusing24bitsperpixel.Thethreestandards areNTSC for North AmericanbroadcastTV, PAL for Europeanbroadcast TV, and high-definitionTV (HDTV). Two ver$ionsof digital compression encodinghavebeendefinedby the Motion PicturesExpertGroup:MPEGI andMPJG2. MPEGI is intendedfor vHS-quality videoandaudio.MPEG2addresses higherqualityrequirements of broadcast-quality videoand audioaswell asHDTV. Becauseof the encodingcomplexity,MpEG2 is primarilyusedin broadcast applications wheretheencodingcostis sharedacrossmul(MPEC2decoders tiple destinations. arerelativelyinexpensive.) Dataratesarevariabledependingon thenatureof thesourcematerial(amountof movement). Themean bit rateof MPEG-I is I.544 Mbpswhile MpEc-? hasa meanbit rateof 5 Mbps. The ATM Forumhasdefinedhow MpEG-2 TS packetsareto be packedinto a AALS-CBR frame.The useof AALI layeris advocatedby the rru oru-T J.gz). However,ITU-T hasagreedto incorporate ATM ForumdefinedAALS-cBR packing into its document.Accordingto the ATM Forum,AALS servicesare adequatefor MPEG-2streamsbecauseMPEG-2includesa prograrnclock reference(FCR) and thereforedoesnot needthe SRTSsprovidedby AALI.

10.4 INTERNETPROTOCOLTRANSPORT The Internetevolvedfrom a packet-switching networkconsistingof hostcomputers connectedwith leasedlinesto IntemetProtocol(IP) routersthat were,in turn,interconnected with leasedlines.As this evolutionoccurred,high-speed digital networks alsoevolved.FigureI 0.I 8 depictsthesituationwhereinthesetwo developments have cometogetherin variousways.Whatis shownis five differentmeansof providingIp transporJ usingvariouscombinationsof networks:ATM, soNET, framerelay,and directphysicallayerfiber.

10.4 TNTEBNETpnoTocoLTHANSpoRT491

| " , l lAtl4 I

I tr

ll

F

ll

F

I

Am,,I llF'-".t'yll f sotrEr@ lsoltEr/spHll

p

I

Fiber plant physical layer F'igure 10.18 Intemet transmissionlayer altematives.

Other than IP transport directly over fiber, IP over SONET [33] is the most etficient. The most inefficient situation is IP on top of ATM. A straightforward approach to reconciling the two protocol layers is to use a relatively static configuration for the ATM network with permanentvirtual circuits. ln this manner the IP network thinks it is using leasedlines. An AAL5 interfaceconvertsthe IP packetsinto fixed-length cells and ships them through the ATM network to an ATM destination node. At the destination node the packet is recon$tructedand passedto the IP layer. Ifthis IP node is the final destination,all is well and good. More often than not this is not a final destination becausea direct virtual connectionbetween all nodesin a large network is very difficult to manage.(Tl1enumber of paths grows arrthe sguareof the number of nodesand nodes are constantly being added and removed.) When the,first ATM node is an intermediatenode, the IP packet is passedback down through an AALS interfaceto get back into the ATM network. and so on. An altemative procedureinvolves having the ATM network set up SVCs for each packet so the ATM destinationcoincideswith the IP destination.This approachavoids the multiple trips up and down the adaptationlayersbut introducesan extremeamount of control overheadwithin the ATM network. To minimize IP over ATM problems, Cisco systemshas developeda networking procedurereferred to as "tag switching" that mergesIP and ATM protocols. A tag is essentiallyan ATM destinationaddressthat gets attachedto eachpacket (addressand all) at a tag edgerouter. The compositemessageis passedthrough the ATM network, which provides tag switching, until it gets reconstructedand delivered to the destination where the tag is removed. The simplification occurs becausea tag is easier to processthan IP adfuessingand routing. The IETF has standardizedthe basic concept of tag switching as multiprotocol label switching (MPLS). As indicated on the far right of Figure 10.18, the overheadof multiple network layers is avoided by directly connecting one IP node to anotherIP node with a dedicated fiber link. There are, however, limitations with direct connections.lP networks have been traditionally implemented with leasedcircuits that are managedand maintained by a common carrier. These leased-line services include provisioning, performance monitoring, and protection switching. Common-carrier provisioning allows rearrang-

492

DATAANDASYNcHRoNoUsTEANSFERMoDENETWoHKS

ing networkinterconnections usingSONETcross-connect equipment.performance monitoringallowsautomated identificationandisolationof faulty or marginalequipment.Physicallayerprotectionswitchingassures an applicationwith a reliableconnection despite equipmentfailures or cable cuts. when direct physical layer interconnect is used,the IP applicationmustimplementapplicationlayerprotection switchingto circumventtransmission link failures. Another,possiblymoresignificantlimitationwith Ip "overglass"is thepotential for a lackofcarrier-$upported signalregeneration. Althoughtransparent opticalhansmission(on fibersor wavelengths) is a seeminglyatmactive serviceoffering,a transparentchannel,by its definition,cannotbe supportedwith regeneration. Thecariers could offer transparent opticalchannelsfor limited distanceapplications,but longdistanceapplicationsthroughopticalmultiplexingand switchingsy$temscannotbe realizedwithoutregeneration. Regeneration is whatallowsa telephoneuserto "hear a pin drop."

HEFERENCES I R.E,Kahn,"Resource-sharing computer communications Network s,"proceedings of IEEE,Nov.1972,pp.1397-1407. 2 H. Rudin and H. Miiller, "More on Routingand Flow Control,"National Telecommunications Conference, 1979,pp.34.5.l-34.5.9. 3 J' Rinde,"RoutingandControlin a CentrallyDirectedNetwork,"l{afional Computer Conference, 1977, pp.603-608. 4 s. c. K. Youngandc. I. McGibbon,"The control systemof the DatapacNetwork," Internati onaI Conferente on Communicatio ns, lg76, pp, I 37- I 4 I . "comparison 5 T. s' Yum and M. schwartz, of Adaptive Routing Algorithms for ComputerCommunicationsNetworks,"National Telecommunications Conference1 9 7 8p, p . 4 . 1 . 1 - 4 . 1 . 5 . 6 F, E. Heart,R. E. Kahn, S. M. Omstein,W. R. Crowther,and D. C. Walden..'The Interface MessageProcessorfor the ARPA computer Network," spring Joint ComputerConference, 1970,pp. 551-556. "Principles 7 L. Kleinrock, and l,essonsin Packetcommunications,"proceedingsof IEEE,Nov, 1978,pp. 1320-1329. I Internetstandard3, "Requirementsfor InternetHosts: IETF RFC 1122," Internet EngineeringTaskForce,lg9B. 9 J' McQuillan, "Beyond 'Best-Effort' Routing," BusinesscommunitationsReview, May 1998. l0 M. Hamidi, O. Verscheure,J. p. Hubaux,L Dalgic, and p. Wang, ..VoiceService Interworking for PSTN and IP Networks," IEEE communicationsMagazine, May 1999,pp. 104-111. I I TheATM ForumTC, "Traffic Management specificationversion 4.0," April 1996. 12 N. Ghani,s. Nanukul,and s, Dixit, "ATM Traffic Managementconsiderationsfor FacilitatingBroadbandAccess,"IEEE communicationsMagazine,Nov. 199g,pp. 98-105.

REFERENcES493 P. Newman,"A FastPacketSwitchfor the IntegatedServicesBackboneNetwork," IEEE Joumal on SelectedAreasof Corununications,Dec. 1988. L4 K. Sarkies,"The BypassQueuein FastPacketSwirching,"IEEE Transactionson Communications, May 1991,pp. 766-7'74. 1 5 F. M. Chiussi,J. G. Kneuer,andV. P. Kumar,"Low-CostScalableSwitchingSolutions for Broadband Nefworking; The ATLANTA Architecture and Chipset," IEEE Communications Magaline,Dec. I 997,pp.44-53. t 6 S. C. Knauer and A. Huang, "STARLITE: A WidebandDigital Switch," IEEE IJ

Glohecom, Nov. 1984.

t 7 V. E. Benes,MathematicalTheoryof ConnectingNetworks,AcademicPress,New York, 1965.

r 8 K. E. Batcher,"SortingNetworksandTheirApplications,"Proteedingsof SpringJoint 1968,pp. 307-314. ComputerConference, l 9 Y-S. Yeh, M. G. Hluchyj, and A. S. Acampora,"The KnockoutSwitch:A Simple, Modular Architecture for High-PedonnancePacket Switching," IEEE Joumal on Selected Areasof Communications, Oct. 1987,pp. 1274-1283. "Architectures 20 E. W. Zegura, for ATM SwitchingSystems,"IEEE Communications 2l

Magazine,Feb.1993.pp. 28-37. "NonblockingArchitectures A. Pattavina, for ATM Switching,"IEEE Communicatians Magazine,Feb.1993,pp. 38-48. M. de Prycker,Asynchronous TransferModeSolutionfor BroadbandISDN,Prentice Hall, EnglewoodCliffs, NJ, 1995. R. Y. Awdeh andH. T. Mouftatr,"Surveyof ATM SwitchArchitectures,"Computer

Networksand ISDNSystems, Vol. 27, 1995,pp. 1567-1613. "Multiservice 24 Agreement,"msf-architecture-Ol.00, SwitchingForumImplementation ATMForum,Nov. 1998. 25 CircuitEmulationSeniceInteroperabiliry SpecificationVersian2.0, af-vtoa-fi)?8.000, ATM Forum.Ian. 199'7. 26 B. Steyaert,Y. Xiong, and H. Bruneel, "An Efficient Solution Techniquefor TrafFrc,"InternationalJoumal of Dirlcrete-TimeQueuesFed by Heterogeneous Communications ,9ystens, Mar./Apr.1997,pp. 73-86. 27 A. Ia Corte, A. Lombardo,and G. Schembra,'?nalysis of Packet Loss in a Continuous-Time Finite-BufferQueuewith MultimediaTraffic Stream,"Intemational fournal of Communitations, Mar./Apr.199'1,pp. 103*115, "A 28 J. ChoeandN. B. Shroff, Cental-Limit-Theorem-Based Approachfor Analyzing QueueBehavior in High-SpeedNetwork," IEEE/ACM Transactionson Networking, Oct. 1998,pp. 659-671. K, Sriram,T. G. Lyons,andY, T. Wang,'AnomaliesDue to DelayandLossin AALZ PacketVoiceSystems: Performance ModelsandMethodsof Mitigation,"IEEEfournal on SelectedAreasin Communications, Jan.1999,pp.4-17. K. Sriram and Y, T. Wang, "Voice over ATM Using AAL2 and Bit Dropping; Performanceand Call Admission Control." IEEE Joumal on SelectedAreas in Jan.1999,pp. 18-28. Communications,

494

DATAANDASYNCHRONOUS TRANSFEH MODENETWOHKS

3l

S. Gringeri, K. Shuaib, R. Egorov, A. L,ewis, B. Khasuabish, and B. Basch...Traffic shaping, Bandwidth Allocation, and Qualilty Assessmentfor MpEG video Distribution over Broadband Networks," IEEE Nen+ork,Nov./Dec. 1g98. 32 M' Krunz, "Bandwidth Allocation strategiesfor Transporting variable-Bit-Rate video Traffic," IEEE CommunicationsMagazine, Ian. 1998, pp. 40*46.

33 J. Manchester,J. Anderson,B. Doshi, and S. Dravida, ..Ip over SONET." /EEE Communications Mag,azine, May I998, pp. 136-142.

PROBLEMS l0.l

10.? 10.3

10.4 r0.5

Determinethe averagetransmission capacityrequiredto rehansmit50-kbyte mes$ages if theyaretransmitted intactacrossa singletransmission link with a bit errorrateof 10-6.what is the averageretransmission capacityrequiredif themessages arepacketized into ?-kbyrepackets? (Ignorethesizeof thepacket headers.) RepeatProblem10.1with 2 bit errorprobabilityof l0-a. Determinethe total numberof bits in an HDLC frameif an informationfield containsnothingmorethan4 bytesof all 1's. Assumetheminimumsizefor all fieldsof overhead. Includetheopeningandclosingflag in thecalculation. Determinethenumberof 2 xz switchingelementsin a 32 x 32 banyanswitch matrix. Determinethe transmission efficiencythat resultswhen a singlechannelof G.723.r compressed voice is packedinto ATM cells and transmittedwith minimal delay.G723.1utilizes30 ms processingframes.Assumethe higher of the two standmdrates(6.3 kbps).How muchdelayis addedro rhe voice channelif theATM cellsarepackedwith closeto l00zoof capacity?(Ignore speechactivityconsiderations. )

11 DIGITAL ACCESS SUBSCRIBER By theearly1990sthepublictelephone networksof theworldhadconvertedto digital technologyin virnrallyall of the internaltransmission andswitchingequipment.Towardthe latterhalf of the 1990snew digital applicationsfor traditionalanalogtechnologyfocusedon providingcustomeracces$to the digital network.The Integrated ServicesDigital Network(ISDN) digital subscriber loops,definedin themid 1980s, represent oneapproachto digitalaccess. AlthoughISDN hasbeendeployedin many penetrationfor variousreasons. markets,it hasnot achievedwidespread This chapter contrastsISDN accesswith altemateapproaches for accessing the digitalnetwork. technologies usedin providing Broadlyspeaking,therearefour basictransmission local digitalaccess:copperwire pairs,fiber,coaxialcables,andwireless.Copperaccessutilizestelephonywire pairsby replacinganalogffansmission with digitaltransmission for both voice and data (e.9., ISDN), adding digital transmissiononto (e.9.,ADSL), or carryingdigitalsignalsin the conventional analogvoicetransmission existinganalogloop usingmodems.Therapidadvancement of modemsfor realizing transmission ratesof 33.6kbpsin a V.34 voicebandmodemor roughly56 kbpsfor a V.90 modemis a mainreasonfor lessthanexpectedISDN usage. Wide-bandwidth transmission on copperpairsinstalledfor analogPOTSserviceis complicatedby severalfactors.First,loadingcoils on longerloopsmustbe removed. performance. Second,it maybe necessary to removebridgedtapsfor adequate ISDN basicrateacces$ such typicallyrequiresbridgedtap removal,but newerapproaches as ADSL utilize more sophisticated channelcharacterization andequalizationtechniquesto accommodate bridgedtaps.Crosstalkandinterference at high frequencies areotherconsiderations. A few wire pairsin a cablecancar4ithehigh-frequency signals,but if a largenumberof pairstry to be activeat thesametime,crosstalkbecomes a significantlimitation.Interference from externalsourcessuchasAM radiostations (e.g.,I MHz). Somehigherrateaccess is a consideration athigherfrequencies systems aredesignedto detectnarrowband interference andmaskit out. Coaxialcableaccessinvolvesaddingdatatransmissiononto unusedfrequency bandsof a cableTV system.Fiber accesshasthe potentialfor providingvery wide bandwidthsbut requiresinstallationof new transmissionfacilitiesdedicatedto the new services.Wirelessaccessinvolvesmicrowavedistributionsvstemslike mul4S5

496

DIGITAL SUBSCRIBER AccESs

tichannel multipointdishibution service(MMDS)andlocalmicrowave distribution (LMDS)or satellite-based system services suchasDSS. 11.1 INTEGRATED SERVICES DIGITALNETWORK In addition to the digitization of the internal portions of public telephonenetworks, a lesserknown but also significant changeinvolved the development of common-channel signaling (CCS) for network control. Both the digitization and the use of CCS srafred at internal portions of the network and migrated toward the periphery. Except for some special data service offerings and a few network-basedfeaturesderived from the signaling network, thesefacilities provided no direct benefit to the end users.As shown in Figure I I . I , ISDN i s a serviceoffering that extendsaccessto both of thesefacilities to the end user. Access to the digital transportfacilities occurs on 64-kbps bearer (B) channelswhile accessto the signaling network occurs on l6- or 64-kbps signaling (D) channels.Major featuresor benefits made availableby thesechannelsare listed in Tables 1l.l and 11.2,respectively. Two levels of digital accessto the ISDN network havebeenstandardized:basic rate accessand primary rate access.As shown in Figure I1.2, the (worldwide) basic rate interface (BRI) $tandardis also referred to as a 2B + D interface. In North America. the primary rate interface (PRI) standard is sometimes referred to as 238 + D while the ITU-T counterpartis 30B + D. The North American pRI is fundamentally a L544Mbps DSI signal with the D channel replacing one of the 24 messagechannels(usually the last one). To achieve a 64-kbps clear channel capability, a BSZS line code is used to eliminate one's density requirementsand common-channelsignaling frees up the signaling bits so the full 64-kbps bandwidth is available for user data. The ITU-T

-----

/

a

f-ffiif-

\

-*

I

ISDN AccEte Line

-----.-/ Figure 11.1 Integrated ServicesDigitalNetworkaccess to circuits,channels, leasedlines,and common-channel signaling.

11..I INTEGRATED NETWoRK DIGITAL SERVICES

497

TABLE 11.1 Features of ]SDN B Channels

1 . End-to-endfour-wiredigitalcircuits:no loss or echoesfor voice circuitsusingdigital 2. 3. 4. 5. o.

instruments $hared network access for voice, data, and leased lines Flelativelyhigh bandwidthdata channel$ (64 kbps) Lower error rates than typical voiceband modems In"seruiceperformancemonitoring Possibleexpansionof speechbandwidthbecauseeliminationof tandemencodingsallows greaterone-timequantizationBrrors

PRI is a 2.048-Mbps El digital signal with the D channel occupying the signaling channel (time slot 16). Becausea single D channel can support more than one PRI, 248 and 3lB interfacesare allowed for additional PRIs in a group of PRIs.

11.1.1 ISDNBasicRateAccessArchitecture An ISDN basic rate accessline is a standardcopper pair that has been specially conditioned to support a bidirectional 160-kbps aggregatedata rate. Transmission technology required for basic rate accessis generally refened to as the digital subscriber loop (DSL). Complications arise when using existing analogpairs. The principal considerationsare bridged taps and wire gaugechanges,both of which causereflections that impact higher speeddigital signals.To allow flexibility in the selection and deployment of the DSL, the ITU-T basicrate specification I I ] doesnot define a two-wire transmission $tandard.Instead. it establishesan interface standardthat assumesthe presenceof a network termination module that converts any chosentransmissionsystem to the standardinterface.In the interestof supportingderegulatedcustomerpremises equipment, the Exchange Carriers Standard Association in the United States establisheda basic rate transmissionstandard[2] so CPE equipmentcould connectdirectly to the transmissionlink or select network termination modules f,rom alternate vendors. Figure 11.3 depicts the architectureand associatedterminology of a North American BRI.

TABLE11.2 Featuresof ISDND Channels 1 . Signaling simultaneous withactiveconnections 2. Callingnumberidentification 3. Far-end superuision 4 . User-to-user messagetransfer 5 . Telemetry forfirealarms,$ecurity, meterreading, etc, A Accessto packet-switching network 7. $upportfor nelwork$eruices suchas multiple numbers directory sharingoneor moreB channels, trunkgroupblocking andidentification of callingnumberfor statistics, or blockedcalls abandoned

498

DIGITALSUBSCHIBEFACCESS

Prim8ry Rete Accgrs

l

2

B I B l . . . l Bl D 238+D

-rTpT

ISDN N6twgrk

BTil 2B+D Ba$ic Rate AccBBB

Figurell.2

Basicrarcandprimaryrateaccess to ISDN.

ModuleDefinitions NTI : A networkterminationmodulefor layer I functionsthatprovidesphysical and electricalterminationof the transmission link only. In essence, the NTI isolatesthe userfrom thetransmission technologybut doesnot demultiplexor processD channelmessages. NTZ: A secondlevelof networkterminationthatimplementsfunctionsassociated with layers2 and3 of theosl protocolstack.Thus,NTZequipmentextractsand processes D channelmessages. Representative NTZ equipmentincludespBXs, multiplexers,or LAN gateway$. TEl: Type I terminalequipmentsuchasa digitaltelephone thatcomplieswith the ISDN S interfacerecommendation. 714.'Terminaladapterusedto convertfrom an arbitrary(R) interfaceto the ISDN S interface. TEZ: A non-ISDN terminal that requiresa terminal adapterto interfaceto the ISDN S interface.Prevalentexamplesof a TE2 equipmentare analog telephones (RS-232)dataterminals. or asynchronous ReferencePoints U; Interfaceto thetwo-wiretransmission line.

ISDN Network

Figure 11.3 Basic rate interface architecture.

111 TNTEcRATED sEnvtcEsDrcrrAL NETWoHK 499

T: CCITT ISDN interface defined in RecommendationI.430. S; Interface to NTZ equipment identical to a T interface. R.' A non-ISDN interface such as an analog tip and ring. 11.1.2 S/T Interface The S/T interfaceis definedin ITU-T recommendation 1.430to be suppliedby network terminationequipment(NT2A{TI). It is intendedfor customerpremisesinstallationsonly. (No overvoltageprotectionis prescribed.) The mostsignificantaspects of the S/T interfaceare; 1 . Four-wirefacility (onepair for eachdirection) 2. Onekilometermaximumrequireddistance J - Alternatespace inversionline code(whichis the inverseof an AMI line code: seeFigurel l.4) 4 . Point-to-pointor point-to-multipoint configurations 5 . Datarateof 192kbpswith 48 kbpsof framing,control,andsynchronization The frame rttructureat referencepoints S and T is shown in Figure 11.5.As indicated, each 250-msecframe contains 48 bits. Thirty-eight of thesebits are cornmon to both directions of transmissionand are defined as follows: 16 bits in first B channel(Bl) 16 bits in secondB channel (82) 4 bits in the D channel I bit in the framing channel F I bit in the auxiliary framing channel F4 The remaining l0 bits are assigneddifferent functions dependingon the direction of transmission.From the TE to the NT all remaining 10 bits are defined as L bits, which are used to maintain dc balance.The definition of the 10 bits from the NT to the TE are: 2 L bits for maintainins dc balance

l r l o l o l o l t l o l r l o l

_ J L _ _ n _---Lr -l_r _ 1rFigure 11.4 Altematespaceinversionline code.

500

DIGITALSUBSCHIBERACCESS

48 Bits ln 260

NT to TE

F L Brrr tl gr Br tr Brrr E D A F^r E?tt!tt2B? |2r?t2 f, u H !t rl rl rt rr

BtB2t0LFt.

FJ._J._J

LtttlllllllllllllLoLr^LlrttBt!2828282!lL0LBlElEtltrlElBluLDL12ultit12uu|0LDttL

lrFigure 11.5 S/T framestructure,

4 E bits that echo D bits received from the TE(s) I A bit for activation I N bit, which is the complement of the Fo bit I M bit far multiframe identification 1 S bit for S channel Figure I I.5 indicatesthat the framing bit F is always a binary 0. Even though a positive voltage level is indicated, either a positive or negative voltage is allowed so the receiversare not sensitiveto wiring polarity. A transmitter always producesthe same level, however, so the receiver always receivesthe samepolarity in every framing bit. As an aid in rapid acquisition of the framing pattem, the framing bit always represenm a line code violation (it is the samepolarity as the previous 0). To maintain dc balance, an L bit with the opposite polarity of the F bir always follows the F bit. The first 0 in a data block following the L bit is encodedwith the samepolarity as the L bit, which implies another line code violation. Direct-current balancing of this violation is the purpose of the L bit at the end of each data block, which also assuresthat the next (fixed-polarity) F bit producesa line code violation. The reason for the additional L bits in the frame from the TE to the NT arises becau$emore than one TE can be connectedto the S intedace as a passivebus (Figure I 1.6). Becausethe TEs transmit independently of each other, each individual transmission (D channelbits and B channelbytes) is individually dc balanced. Passivebus operationsare also the reasonfor the existenceof the NT-to-TE E bits. Multiple station accessto the D channelis controlled by having a terminal wait for an idle code on the NT-to-TE D channelbefore hansmitting on the TE-to-NT D channel. when a terminal begins D channeltransmission,it monitors the incoming E bits. If an incoming E bit does not match the previously transmitted D bit, that terminal stops transmitting and waits to seize the channel at a later time. Two levels of priority are defined for accessingthe D channel. signaling informarion is the highest priority

11.1 INTEGHATED DIGITAL SERVICES NETWORK 501

s TEl

a

u

a I

NTt

TET

TEl

J Ftgurell.6

S-busconnections.

while userpacketmessages arethelowerpriority.All terminalson a passivebushave equalprioritieswithin eachlevel. Example 11.1. Determinethe distancelimit imposedby theneedto echoE bits in a BRI S/T inteface.Assumethe speedof transmission of a signalon a pair of wires is one-thirdthespeedof light in a vacuum.Ignoretransmitterandreceiverfilter delays andassumeno appreciable delaysin theNT logic circuitry. Solutian. From Figure I I.5 it can be seenthat the minimum delay betweena terminalhansmittinga D bit andreceivingit backin the followingE bit is sevenbit times(thisis theTE to NT D bit followingthefirst 82 byte).At a 192-kbpsdararare the durationof a bit is 5.2 psec.Thus,the total round-trippropagationtime is 7 x 5.2 = 36.4p sec.Assumingno appreciable circuitrydelaysin the NT, Maximumwire length= (36.4x 10*6)x + (3 x 108)m Becauseround-trippropagation involvesbothdirectionsof transmission = x 3.64= 1.82Lrn Maximumdistance ] Example1l.I showsthatthe BRI standardhasa fundamentaldistancelimitation thatis not far abovethe minimumtransmission distancespecification of 1 km. 11.1.3 ISDN U Interlace Prior to the e$tablishmentof a standardU interfaceby the ExchangeCarriersStandardsAssociationin NorthAmericaa numberof basicratetransmission sy$temswere developedby telephoneequipmentsuppliersaroundthe world andput into service.

504

DrcrrAL suBscHtBER AccESS

one of theseinvolvestheuseof timecompression multiplex(TCM).TcM, developed by NEC in Japan[3], providesfull-duplextransmission on a singlepair of wiresby alternatelytransmittingburstsof datain eachdirection.For thisreasonit is sometimes referredto as"ping-pong"tralsmission.one big advantage of rCM transmission is thatnear-endcrosstalkis avoidedbecause a stationis neverreceivingwhile transmitting' Thebig disadvantage is thattheburstdataratemustbe morethantwicethe desireddatathroughput. AT&T in the united statesalsodevelopeda BRI transmission systemfor the No. 5 ESSend office switchingsystemavailablewith genericreleases5E4 and 5E5 t4l. This systemhasa 160-kbpsdararateutilizinga 50zodutycycleAMI line code.Full-duplextransmission is achievedby simultaneous transmission in both directionsusinghybridsandechocancelers(ECs)to separate the two signals,as indicatedin FigureI I.7. Beginningwith generic5E6theNo. 5 ESSsupporrs both the AT&T (Lucent) u interface(referuedro as a 5E4l5E5 u interface)and the ANSI U interface. Like the 5E4l5E5U interface,the ANSI U interfaceusessimultaneous ffansmissionin bothdirectionswith echocancelers anda datarateof 160kbps.Themajordifferenceis the useof a four-levelline codereferredto as2B1e (two binarydigits in onequaternary digit).Thus,thesymbol(baud)rateon theline is 90,000symbols/sec. Becausethe line codeitself doesnot preventdc wander,dc restorationis necessary. Reference animplementation with adaptivequantizedfeedbackasbeing [5] describes thebestapproach. A 2BlQ line codewaschosenprimarilybecause thelower symbol rate minimizesthe two dominanttransmissionlimitationsin this application;intersymbolinterference andnear-endcrosstalk[6]. The frame format and superframestructureof the ANSI U interfaceareshownin Figure11.8.Eachframeconsistsof 240bits containinglg framingbits, 216payload bits (12 fieldsof I I 2B + D databits),and6 overheadbits.Becausetheframerateis 667frames/sec thedatarateis 160kbps.The6 overheadbits areorganizedasa block of 48 bits in an eight-framesuperframe. Functionsincludedin the overheadbits are 24 bits of embedded operations channel,I activationbit, I deactivation bit, I far-end block errorbit, l2 cRC bits, and9 fixed I bits. All bits exceptthe framingbits are scrambled fbr transmission.

Figure 11.7 ISDN DSL TX/RX blockdiagram.

11.2 HIGH-DATA-RATE SUBSCRIBER LOOPS DIGITAL

I

z

3 4 5 6 7 E

t8

18

18

rsw sw svv sw

28+D 28+D 28+O 28+D 28+D 28+D 28+O 28+D

28+D 28+D 28+D 28+D 28+D 2B+D 28+D 28+D

SW svlf

sw sw

28+D 28+D 2 8 +D 2E+D 28+D ?B+D 28+D 28+D

503

Mr-M" Mr-M. Mr-M" Mr-M. Mr-M. Mr-M" Mr-M" Mr-Mt

S W- S y n cW o r d= + 3 + 3 - 3 * 3 - 3 + 3 - 3 + 3 + 3 I S W : I n v w u d$ W = - 3 - 3 + 3 + 3 + 3 - 3 + 3 - 3 - 3 2 B + D * l B , l B " l D l ( l 8 l 8 f2 ) Mr-M" * Ovdrlrosdbitr Dataar€encoded as0O= -3. 01 = -1, 11 : +1, 10 = +3 Figure 11.8 ANSI U interfaceframe and superframestructure.

11.1.4 ISDN D Channel Protocol seriesof ITU-T recomrnendations; TheD channelprotocolis definedin two separate theI seriesandtheQ series.Thedatalink layer(LAPD) is definedin L441or Q.921. exceptQ.921allowsmorethan This protocolis similarto LAPB of theX.25 standard "connections" canexistfor signaling,packetnetonelogicallink. (Thereforeseparate work, or far-endterminals.)The main functionsof the datalink layer aremessagesequencing,error checkingand retransmission, and data layer link recovery.The networklayerof the D channelprotocolis definedin I.451 or Q.931. This layerprovidesconnectionsetup,alerting,routing,andreleaseof ISDN calls.Whena B channel state. accesses a packetnetwork,theX.25 protocolis usedwhile in theconnected

11.2 HIGH.DATA.RATEDIGITALSUBSCRIBERLOOPS bidirectionaldata A basicrateISDN digital subscriberloop providesan aggregate, severaltransmission rateof 160kbpson a singlepair of wires.This sectiondescribes techniques thatallow muchlargerbandwidthson copperwire pairs.Thesenewtechdigital signal niquesare enabledby the availabilityof low-cost,high-performance processing. digital subscriberlineshavebeendevelSeveralversionsof high-speed oped.Thevariousversionsarecollectivelyreferredto asxDSL (seeTableI 1.3). 11.2.1 Asymmetric Dlgital Subscriber Line ADSL allowsfor high dataratesto the subscriberandmoderateto low dataratesfrom the subscriber to thenetwork,ADSL technologywasoriginallyconceivedasa means of deliveringswitcheddigitalvideoservicesovera copperloop [7], whichobviously did not Althoughvideoapplications do notrequirehigh dataratesfrom thesubscriber.

504

DIGITAL SUBScFIBEH AccEsS

TABLE11.3 Versloneof DlgltalSubscrlberLines DSL ADSL HDSL SDSL VDSL

DigitalSubscriber Line(ISDNbasicrate) Asymmetric DSL(9 Mbpsdownstream, 640 kbpsupstream)d High-bit.rate DSL(T1lE1serviceon two pairs) Single-line DSL(T1/Elserviceon onepair) VeryhighbitrateDSL(52Mbpsdownstream, 2.8Mbpsupstream)d

'List€d data rates ere maximum posslblevalues. Downstreamls toward the subscriber.Upstr6emis toward the network.

materialize,the asymmehic datarate is also suitedfor Internet access.The philosophy behind the asymmetric data rates is that subscriberstypically need to reieive highbandwidth data (for lntemet file downloads) but normally need to rransmir (query) at a relatively low data rate. The subscriber'sreceive data rate on ADSL varies between 1.5 and 9 Mbps while the subscribertransmit rate varies between 16 and 640 kbps.* The specific data rate utilized dependson the transmissionquality of the particular wire pair. In contrastto ISDN channels,which representextensionsofthe digital facilities of the public switched telephone nerwork (psrN), ADSL channelsare separatedfrom the public nerwork at rhe line inrerface of the psrN. As shown in Figure I l.g, ADSL lines terminate at a telephonecompany central office (or remote terminal) where the data streams are forwarded to and received from a facility that is separatefrom the telephone network. Transmission between the line interface and an intemet service provider, for example,is typically provided with an ATM connection. ADSL has two major advantagesover IsDN access.First, ADSL data rates provided to the subscriberare significantly higher than the lZg-kbps ISDN basic rare. second, ADSL piggy backs digital transmissionon a standard*atog telephonewire pair. Thus, existing analog telephonesare retained on ADSL but are either replaced by digital phonesor are connectedthrough conversiondeviceswhen ISDN is utilized. Two versionsof ADSL transmissionlinks have beendeveloped:carrierlessamplitude and phase(CAP) modulation and discretemultitone (DMT) modularion. CAp is the first version deployed but DMT has been selectedas the standard [g]. Because DMT makes more intensive use of Dsp, a DMT implementation typically requires more power-a significant considerationfor remote terminal deployment. cAp, on the other hand, its not generally considered to be as flexible as DMT in achieving maximum data rateson some wire pairs or in some interferenceenvironments.For a thorough comparisonof the two alternatives,seereference[9].

DMT lmplementdtion Basicparameters of the standardADSL DMT implementation areprovidedin Table I 1.4.A block diagramof an end-to-endDMT implementationis shownin Figure I I.10. As mentionedin chapter6, a DMT implementation utilizesan inverseFFT as *A

less ambitious vetsion of ADSL rcferred to as G.Lite only attempts to achieve 1.544 Mbps clownstream and 384 kbps upstream.

11.2 HIGH.DATA.RATEDIGITALSUBSCRIBERLOOP$

P$TN Switcl'tmetrix Lineinterfaces multiplexer

Subscriberloops Splitters volce

volCc

Figure 11.9 ADSL networkconfiguration.

Param€ters TABLE11.4 AD$L DMTlmplementatlon Subchannel$eparation Maximumbits/$ubchannel Numberof subchannelsa FFT sample size Cyclic prefix Total numberof samples Sample rate FFT frame duration Pilot frequency

Numberolsubchannels FFTsamplesize Cyclicprelix Totalnumberof samples $amplerate FFTframeduration Pilotfrequency

4.3125kHz 15 Downlink 255

512samples 32 samples s44 (512+ 32) 2,208MHz(51?x 4312,5) x 106) 246.377x 10a sec (5,+4/2,208 64) 276kHz(subchannel uplink 31 64 samples 4 samples 68 (64+ 4) 276kHz(64x 4312.s) x 1Oo) ?46.977x 104 sec(68/0.276 16) 69 kHz(subchannel

the number of subchannFl8dBpendson the amount of matgin tor PoTS filterlngand on choice of echo canc€lingor FDM to isolatethe two directionsol transmission.

505

506

DIGITAL sUBscRIBER AccESs

a modulator and an FFT as the correspondingdemodulator.The following paragraphs identify the basic funcrion of eachblock in Figure I I . 10: PRSsource: A pseudorandomsequencegeneratorprovides a prescribedsequence for characterizing the channel during a training period. Characteristicsof the channel that are determined during the training periorl are attenuation and phase distortion across the band, noise/interferencelevels across the band. and the information capacity of individual subchannels. Channel Allocation: Provides assignmentof data bits to individual subchannels according to the subchannelcapacity determinationsobtained during training. FEC: A combination of Reed-solomon and convolutional coding. IFFT QAr4 Modulation: conversion of data values to quadraturechannel signal amplitudes and conversion to a time-domain waveform using an inverse FFT. D/A : Digital-to-analog conversion. TX Filter: Bandpassfilter to preventinterferenceinto the voicebandand to smooth the discreteDSP samples. POTS Splitter" Used at both ends of the line to separatethe analog voice from the data.(Simpler versionsof ADSL incorporatethe splitter function in the modems to simplify installation.) AGC.' Automatic gain control to adjust overall receive level. sktpe/Delay Equaliption' A front-end equalizerro partially flatten the frequency responseand equalizeextreme delay variations in the chaanel [10]. ,4./D.' Analog-to-digital conversion.

Source data

Wire pair

Figure ll.l0

ADSL DMT block diagram

CAFHIEB 11,3 DIGITAL LOOP SYSTEMS 507 FFT QAM Demodulation: FFT conversion of time-sampledwaveform to frequency domain where data values are related to amplitudes of quadraturecarriers. Clock Recovery: AID sample timing obtained by locking to pilot frequency. The desired sampling rate is eight times the pilot frequency so 1-of-8 phase ambiguity has to be determined by monitoring framing/synchronization bit integrity. Frequency-Domain Egualizer: Multiplication of complex (quadrature) frequency $pectrumby amplitude and phaseequalization pararnetersobtained during training period. Data Detection and Interleaving: Slicing of quadrature amplitude values to decodedata and subsequentgenerationof compositestreamidentical to original sourcedata.

11.2.2 VDSL VDSL is an expandedversionof ADSL to achieveevenhigherbandwidthson particularlyshortlinesaswouldbe availablefrom remoteterminalsof fiber to the curb systems.A primarymotivationfor VDSL is potentialdistributionof HDTV signals. Althoughseveraldifferentmodulationtechniqueshavebeenproposedfor VDSL, a DMT versionrieemsto be favored[11].

11.3 DIGITALLOOP CARRIER$YSTEMS in Chapter1,theprimarypurposeof a digitalloopcarrier(DLC) system As discussed is to reduceor eliminatecopperpairsfiom a centraloffice to the vicinity of a group link from the centraloff,rceto theDLC of subscribers. Eventhoughthetransmission links from theremoteterminalto thesubremoteterminalis digital,thetransmission loops. Thus,themainpurposeof a DLC is analog scribersaretypicallyconventional provide someDLC systems(particudigital Nevertheless, subscriberaccess. not to provide T1, for ISDN, or xDSL digitalinterlarly fiber-based systems) options DLC faces. Moreover, the fact that fhe DLC remote terminal is relatively close to the subscriber locationsfacilitatestheuseof theseinterfaces.(A shortcopperdropfrom andcrosstalkandis lesslikely to haveloading theremoteterminalhaslow attenuation Althougha DLC taps, multiple of or sections wire with varyinggauges.) coils,bridged servicesthroughISDN or xDSL digital is a naturalmechanismto provideenhanced of theseinterfacesrepreinterfaces, mechanicalpackagingandpowerconsiderations with to strictly POTS applications. sentuniquerequirements respect 11.3.1 Universal Digltal Loop Carrier Systems As shownin FigureI1.11,a universaldigitalloop carier (UDLC) system[12] canbe interfacedto anyswitchingsystem:analogor digital.Theinterfacebetweenthelocal

508

DtetrAL suBscRlBER AccEss

Customer loops COT

RT

Figure 11.11 Universal digital loop carrier system.

switching system (end office) of the public network and the central office terminal (coT) involves individual circuits (e.g., individual analog rip and ring connections). The multiplexed digital transmissionlinks between the COT and the remote terminal (RT) can be wire pairs or fiber. Each interface of a COT is paired with a corresponcling subscriberintetface at the RT so the use of a UDLC is transparentto both the switch and the subscriber.In its simplest mode of operation,the uDLC usespure multiplexing betweenthe COT and the RT so that there i$ a one-to-onecorrespondencebetween a particular TDM channel and the coT/RT interface pair. some systemscan also be configured with concentration wherein the Cor/RT pairs are dynamically assigned transmissionchannels.If the number of requestedconversationsexceedsthe number of channels,blocking occurs. The possibility ofblocking introducesnontransparency and implies that some means of returning a reorder tone (fast busy) is neededin the RT. UDLC installations are configured to match eachparticular interface of the central office switch with a complimentary interface in the RT. A fully capablesystemmust provide a wide variety of interfaces such as loop-start line, ground-start pBX trunk, foreign exchangelines, and coin telephoneinterfaces.In someearly systemsthe confrguration processinvolved nothing more than physically installing matching interfacesin respectiveequipment slots of the CoT and RT. More recent.systemstypicatly utilize line units with multiple-service capabilities. These systemscan be configured electronically (i.e., no straps)with either a local or a remore manasementinterlace.

11.3.2 IntegratedDigitatLoop CarrierSystems Whenevera universaldigitalloopcarriersystemis interfacedwith a digitalswitch,obviousinefficienciesoccurin termsof back-to-back demultiplexing-multiplexing and D/A-A/D conversion.As shownin Figure 11.12,an integrateddigital loop carier 0DLC) systemeliminarestheinefficiencyby directlyconnecring theDLC TDM link to the digital matrix. Typically rhe direct digital connectionsare eitherDSI or El crorts-connect signals.Thus,a fiber-basedDLC systemwill typically interfacewith theswitchthroughmultiplexing/demultiplexing equipmentassomenumberof digital crostt-connect signals.

LOOPCARRIEH SYSTEMS 509 11.3 DIGITAL

Customer loops

Figure11.12 Integrated digitalloopcariersystem.

From a functionalpoint of view an earlyIDLC systemswasnothingmorethana distributedswitchingsystemwhereinsomeline interfacesof the switcharemovedto providesucha capabilityremotelocations.Most cenfialoffice switchmanufacturers thatcanbecolosometimes assimplyasusingchannelbanksfor analogline interfaces catedwith the matrix or remoted.Central office switch vendorsalso offer remote switchingmoduleswhereinsomeportion of the matrix itself is remotelylocated. for efficientuseof theconnecting provideremoteconcentration Theseconfigurations transmission link and,in somecases,pmvidelocal switchingin theremotemoduleso link. connections betweentwo por"tsof a remotemoduledo not usethetransmission (If remotelocalswitchingis notprovided,a connection betweentwo portsof a remote modulerequirestwo channelsof thehansmissionlink.) Remotemodulesof a particular switching systemvendorare often implemented processes thatprecludeuseof DLC equip' with proprietarysignalingandmanagement mentfrom othervendors(exceptwhena DLC vendordevelopsthe ability to emulate a particularswitchvendor'sremotemodules).In the interestof openingIDLC applicationsto competingvendors,Bellcore establishedan IDLC standardreferredto as GR-303[13, 14] that compliantswitchingsystemvendorsmustsupport(possiblyin additionto a proprietaryIDLC capability).Includedin the GR-303specificationare andperformance definitionsfor signaling,provisioning,testing,alarmsurveillance, monitoring. Due,in part,to a moveto unbundleLEC local loop servicesGR-303hasassumed a muchbroaderscopethanjust anIDLC application.Becausea GR-303capabilityincludesbeingableto defineandadministera myriad of switch interfacetypes,theGR303 $tandardcan be usedfor interfacingother typesof equipmentsuchas xDSL equipment.ETSI has establisheda similar IDLC standardfor intemational(ITU) switchingequipmentrefenedto asa V5 interface[5, 16]. 11.3.3 Next-GenerationDigital Loop Carrler Syetems The term next-generationdigital loop carrier (NGDLC) hasbeenadoptedwithin the industry to refer to DLc sy$temsfhat adhereto GR-303andprovide additionalconan Thereis no precisedefinitionof whatconstitutes figurationoptionsandinterfaces. to GR-303,opticalfiber transNGDLC system.Four basicattributesareadherence missioncapabilities(e.g.,SONET/SDH),generallylargerline sizes,andan ability to interfacewith a cenhaloffice operationalsupportsystemfor diagnostics,alarms,and

510

DIGITAL SUB$CRIBER AccESS

Figure ll.l3

Next-generationdigital loop carrier.

remoteprovisioning[l4]. other aspectscommonlyavailablein NGDLC systemsare depicted in FigureI 1.13. The mostimportantaspectof an NGDLC $ystemfrom a servicespoint of view is theavailabilityof new servicessuchascopperor fiber digitalsubscriber interfacesin additionto theconventional POTSinterfaces. Themostdesireddigitalinterfacesare Tl/El, primaryrateandbasicrateISDN, ADSL, andvDsL. other digitalinrerfaces mayalsobe providedfor serviceslike broadband dataanddigitalvideo.Althoughthe diversityof interfacesbeliesuseof a singlemultiple serviceline intedacefor total electronicprovisioning,the systemmust report inconsistencies betweeninstalled hardwareandthe elechonicdatabase. A particularlydesirablefeatureof the system shownin Figure I l. I 3 is drop-and-insert capability,which allowsdistributedaccess to a singlebackbonedigitalroute.GR-303identifiesstarconfigurations, linearADM distribution,andADM ringsasdesirable topologies. NGDLC sysrems tharsupportadvancedinterfacesor topologiesnecessarily usea COTto separate outthespecialservicesfrom thePors. In essence, thecor performscross-connect operations to groom anddistributevarioustypesof traffic.

11.4 FIBERIN THE LOOP Fiber in the loop (FITL) is a genericterm that refersto oneof threemore specificdescriptionsof the useof fiber for local distribution.The flrst categoryis fiber to the cabinetor fiber to the neighborhood. Thesesystemsareofteninstalledby local telephonecompanies aspartof thefeederportionof theirlocaldistributionfacilities.Traditional copper pairs for voice service extend from the cabinet to subscriber residences. Becausea relativelylong copperpair is usedfor ,.thelastmile,,'they do not providemuchopportunityfor enhanced serviceofferings.Theiruseandjustification arestrictlybasedon reducingrhecosrof pOTSdisFibution. The secondcaregoryof FrrL is a fiber-to-the-curb (FTTC) system.As the name implies,thesesystems aredesigned to reachwithin 1000feetof a subscriber residence. An FTTC systemis generallyintendedto provideenhanced servicessuchasvideoor high-speed datausingADSL or VDSL. Distributionof theenhanced servicesfrom rhe "curb" location is caniedover wire pairsor coaxialcable.Thesesystemsareessentially identicalto advanced DLC sysremswith opticalrransport. Thethird categoryof FITL is f,rberro thehome(FTTH).Thesesysrems obviously off'er oppornrnitiesfor extremelylarge bandwidthsto the homebut havesignifircant

11.s HYBRTDFTBERcoAXSYSTEMS 511

is expensive neighborhoods First,installationin established bedeployment obstacles. causeundergroundinstallations(understreetsanddriveways)arenormally required. Second,providingnetworkpowerto subscriberequipmentis a majorproblem.The lack of power for enhancedserviceapplicationsis not much of a considerationbut power to telephonesis. Local exchangecarriersgo to great effotts to ensureindependence from commercialpowersystemsfor boththeiroffice switchingsystemsand the connectedsubscribers.(A centraloffice typically maintainsenoughbatteriesto keepa systemup for ?4 hoursif commercialpoweris lost.If poweris out for longer than this, dieselgeneratorsare available.)A further complicationof providing telephoneserviceoverFTIH is the needto converta digitalvoicechannelto analogfor interfacingto conventionaltelephones-anadditionalexpenseandpowerproblem. Batteriesareoften offered asan alternativeto line poweredtelephonesbut logistical problemsremain;who maintainsthe batteryand what happensif the batteryis dead whencommercialpoweris lost andsomeonewantsto diat 911?

11.5 HYBRIDFIBERCOAX SY$TEMS The cableTV systemsinstalledaroundthe countrycan be augmentedwith downto subscribers by merelyadding"cablemodems"utilizing strearndatatransmission from the subunusedor displacedTV channelbandwidth.Upstreamtransmission scriberto a cableTV headendis muchmoredifficult. Although manycableTV systems were designedand installedwith upstreamtransmissionas an option, the bandwidthavailableto theupstream channelsis generallylimitedandoftensubjectto very high noiseand interferencelevels.Upstreamtransmissionlimitationscan be telephone connectionwith voiceband somewhat alleviatedby utilizinga conventional modemswith dataratesup to 28.8kbps.Thetelephonemodemconnectionis usedin the samebasicmanneras the upstreamchannelof an ADSL while relativehighof this solubandwidthdownstrearn dataarecarriedon thecable.The disadvantages cost large telephone line, of numbersof the tion includethe needfor a subscriber provider, and need to into a service the coordinatedialchannelconnection$ telephone particular cablechannelusers. up telephonecomectionswith Thebasicconfigurationof a hybridfiber coax(tFC) $ystemis depictedin Figure areconfiguredastree-and-branch I 1.14.Thecoaxialcableportionsof CATV system$ signal.Ambroadcast receivingthesame,multichannel topologieswith all customers low from attenuation plifiersarein$ertedwhereverthe signallevelgetsunacceptably andbranchingloses.Bidirectionalamplifiersareshownin Figure11.14undertheassumptionthat this sy$temis an applicationwith a returnpath from the residences.In thereturnpath(if thereis one)is usedfor premiumchannelseTV-only applications lection.In expanded$erviceapplicationsthereturnpathcarriesvoice or datawith frequencydivisionmultiplexedcablemodemsin a bandfrom 5 to 42 MHz. The optical fiber transmissionportion of an HFC sy$temrepre$entsa replacement of relativelylong haulcoaxialcablesectionswith numerousamplifiers.For this reason,theopticaltransmittersandreceiversaredesignedto carrya widebandanalogsignal. Noticethat the opticallinks aresharedby a largenumberof cu$tomers(anywhere

512

DIGITAL SUBS0RIBER AccEsS

Fiber node

Figure 11.14 Hybrid fiber coaxsystemconfiguration. from 100 to 1500). upgrading an IIFC system for new $ervices typically requires Sreaterpenetrationof the optical frber portions so that fewer householdsare connected to a common coaxial cable segment.In the limiting situation, wherein eachhouseholdis connectedthrough a dedicatedcoaxial cable, an FIFC systembecomesan FTTC system. Downstream digital servicescan utilize cable modems that typically pack 30-40 Mbps into a 6-MHz analog TV channel. 64-eAM modulation is commonly used.In newer FIFC system.s,new downsfieam digital servicescan be canied at fuquencies above 450 MIJZ while the band from 54 ro 450 MHz is reservedfor haditional analog Tv. A major impediment to upgrading an IIFC systemfor return channelservicesis the shareduse of a coaxial cable segmentcommon to some number of households.The network termination within eachhome is passiveand bidirectional, which meansthat all noise and interferencewithin a home is passedonto the cornmon cable to all other homes.Thus, a single sourceof interferencecan disrupt the signal to all other homes served by the common coaxial cable. Furtherrnore,the noise and interferenceof all householdsare additive, indicating the needto limit the number of householdsserved by a single coax segment.An additional drawback of the sharedcable is the need for some form of encryption for content security. TDMA return channelshelp minimize the interferenceproblem by blocking all output energy from a residenceexcept when an allotted time slot occurs. using an HFC systemfor POTS hasthe samebasic drawbacksas an FTTH sy$tem in that there is no inherent facility for line powering rhe telephones.Thus, FIFC might not be used for primary ("lifeline") POTS distribution but could be quite effective in providing secondarytelephoneapplications.The main attribute of HFC for enhanced servicesis the ability to provide dynamic assignmentof high-rate digital downstream channelsand relatively low rate full-period upstreamchannels.

11.6 VOICEBAND MODEMS voiceband modem technology improved dramatically in the early 1990swith the culmination of 33.6 kbps becoming standardwith ITU recommendationv.34 [17]. The

11,6 VOICEBAND MODEM$ 513

rapidadvanceof voicebandmodemperformance wasdueto two primaryfactors;the cancelingandthe availabilityof economicalDSP technologyfor equalization/echo improvedquatityof thenetworkin termsof lowernoiseanddistortionresultingfrom the near-all-digitalimplementation,In an all-digital network the only significant conversion. sourceofnoise is thequantizationnoiseofthe analog-to-digital Example 11.2. Determinethe theoreticalmaximum data rate of a perfectly equalizedvoicebandmodemundergoinga singleA/D andD/A conversion. ratio (SQR)is derivedin Chapter3 to be Solution. The signal-to-quantizing-noise to a powerratio of 3981.Usingthis value on the orderof 36 dB, which conesponds in Shannon'stheoremfor thetheoreticalcapacityof a channelyields C= lVlogz(l + SNR)

bps

= 3100logr(l + 39Bl) = 37 kbps wherethe channelis limited by the transformercoupling,60 Hz eliminationfrlters, andD/A smoothingfilters to extendfrom 300to 3400Hz.

11.6.1 PGMModems As the previous problem indicates,a V.34 voicebandmodem provides data rates that arenearthe theoreticallimit imposedby quantizationnoise alone.Recognition that the principal sourceof noise in the end-to-endconnectionis the quantizationnoise of the A./D convertersleads to alternative modem implementations that directly utilize the digital 64-kbps channel and eliminate the quantization noise [18]. Thesemodems are commonly referred to as PCM modems. As shown in Figure 11.15, a V.90 PCM modem relies on the digital network to carry an unaltered digital signal from a digital source to a digital-to-analog conversion device (codec)at an analog subscriberinterface.The codecconvertsthe PCM codewordsto PAM samplesthat are detectedby the receiving customerpremisesmodem and convertedback to the original PCM data. Successfuldata detection requires the receiving circuitry to adequately equalize the combined distottion of the D/A smoothing filter and the transmission link, to know the quantization levels of the codec,and to becomesynchronizedto the D/A conversionclock. The equalizationand quantizationrequirementsare determinedduring an initialization proce$$while clock synchronization requires processing of data transitions in the multilevel received waveform. It might seemthat a PCM modem could provide a data rate of 64 kbps. However, severalfactors restrict the data rate to something less than 64 kbps. The first of fhese is the bandpassfilter in the D/A codec (for 60 Hz elimination and samplesmoothing). A secondconskaint is the possibility that the digital path through the network might

514

DIGITAL SUBSCHIBER AccEsS

Tl or ISDN digitatlnterface Dlgital network

Subscrlberloop

m1orrm 10mfi0010t10011

F

+'*.-.H* mlor roo 1fiI}l roo 1oltmii

Figure 11.15 Downstream V.90 modemconcept.

include a digital pad for changing the signal level of rhe assumedanalog signal.* A third constraint is rhe possibility (in North America) that robbed bit signaling might be in use on one or more of the digital links. The fact that the overall bandwidth of the channelis slightly over 3kHz implies that the maximum, intersymbol, interference-freepulse rate is just over 6 kHz. Thus, the actual sample rate of 8 kHz implies that some amount of intersymbol interferenceis inevitable (assumingthe PCM samplesare independent). The lack of a low-frequency transmissionresponseis accommodatedby V.90 modems by utilizing every eighth pcM sample soreryfbr dc restoration.Thus, the maximum data rate is 56 kbps. If robbed bit signaling is present,its effects are minimized by determining, during initialization, which frames in the six frame sequencesare signaling frames and then using only 7 bits per samplein thoseframesund g bits p*r ru*_ ple in the nonsignaling frames. Digital pads can be accommodatedby detectins their presenceand modifying the digital codewords according to the particular amolunt of attenuationbeing insertedby the pad. The v'90 uplink direction is implemented as a convenrional (v.34) modem. thereby implying asymmetry in the data rates. It is conceivable that pAM signaling could be usedin both directions,but the uplink is more complicatedto implement and is often unnecessarybecausemost applications (e.g., Intemet access)are inherently asymmehic in the dataratesrequired. It is also possibleto utilize analog PAM on both -Dgital Ievels.

pads are often included in the codecsfor digital setting of gain levels to desired voiceban d sigflal

DISTFIBUTION SEHVICE 515 11,7 LOCALMICHOWAVE

are endsof the connection(asopposedto just one end),but theseimplementations morecomplicat€d[19, 20]. considerably Themain attractionof PCM modemsis that theyprovidealmostthe samedatarate in theline interfacesor specialtreatasanISDN B channelbut do notrequirechanges mentof thecustomerloop (e.g.,theeliminationof bridgedtaps).Whenthecustomer loop is a shortdrop from a remoteterminalof a digital loop carrier,themaximumdata rateof 53-56 kbpsis assured.

11.7 LOCAL MICROWAVEDISTRIBUTIONSERVICE microwavefrequencybandsfor digitalradioacTheFCC hasallocatedtwo separate Thefirst of theseis MultichannelMultipointDistribuservices. cesscommunications at 2 GHz.MMDS is essentially a wirelesscable tion Service(MMDS), whichoperates Reversechannelcommusy$temand,as such,providesonly one-waytransmission. throughthe telephonenetwork. nicationsrcquiresdial-upmodemconnections Local MicrowaveDistributionService(LMDS) operatesat 28 GHz andprovides Bandwidthsavailablewith LMDS aredependtwo-waycommunications. cell-based, provider's frequencyplan.Typically,theseplansprovide ent on a particularservice bidirectionaldatarate$ontheorderof 1.5-50Mbps,althoughthetotalallocatedbandwidth of almost1 GHz canbe partitionedto providehigherrate$or asymmetricrates if desired.InternationalLMDS allocationshavebeenmadein a rangeof 25-31 GHz. Systems Thesesystemsare also refbrredto as Local Multipoint Communications (LMCS) in CanadaandBrazil. by LMDS arevoice,video,andhigh-bandwidth Theprincipalapplicationssupported data.The immenseamountof bandwidthavailableis the main attraction.The major Thus,after radioequipment. with thesubscriber costof anLMDS systemis associated begin additionalcostsareincurredonly assubscribers the cell sitesareestablished, payingfor service.This situationis in contrastwith newfiber-based facilitiesthatrequire major,up-frontinvestmentbeforerevenuesare realized.The mostimmediate digitalleasedlinesfor opportunityfor LMDS is to offer an alternativeto high-speed areseconHigh-speed InternetaccessandHDTV to individualresidences businesses. daryopportunities. The major disadvantages of LMDS are the needfor FCC licensing,line-of-sight in heavyrainfalls,potransmission, distancelimits of 2-3 miles,extremeattenuation as satellites in the other such samebands,andthe services tentialinterferencefrom (telephone) or dynamic power sites. margins Fade in all subscriber needfor local rainfall rates. powercontrolon the orderof 40 dB overcomeall but themo$tsevere meansthata singlebasestationcanTherequirement tbr line-of'-sight transmission not communicate with all locationsin a cell (exceptmaybein WestTexasor theAusbut somelocationsare tralianoutback).Overlappingcellsprovidegreatercoverage, sureto be in theshadowofall basestationsunlessreflectorscanbeinstalled.Line-ofof buildingsor trees canbefurthercompromised by theappearance sighttransmission initial deployment. after

516

DtctrALsuBscRtBER AccESS

11.8 DIGITALSATELLITESERVICES Direct SatelliteService(DSS)hasrecentlybecomea viablealternativeto cableTV services.BecausetheDSS systemutilizesdigitaltransmission for its TV channels, it is straightforward for a DSSproviderto offer datacommunications servicesto theTV subscribers. However,becauseDSS is a one-waycommunications service,dial-up modemconnectionsthroughthe telephonenetworkare requiredfor two-way data communications. DSSis alsolimitedro thetoralbandwidth(400MFIz)of thesatellite transponders, which mustbe sharedby the Tv ffansmissions and any allocateddata channels. In contrastto DSS,whichusesgeostationary satelliteorbits,Low Earthorbit satellitesoflridium andTeledesicprovideopportunities for lowercost,bidirectionaluser terminals.As mentionedin Chapter9, Iridium is primarilyorientedto relativelynarrow bandapplicationssuchasvoice,messaging, andpaging.Teledesic,on the other hand,is orientedto wide-bandwidth dataapplications.

REFERENCES

r0 n

Recommendation I.430 ISDN user-NetworkInterface;Layer I Recommendations, Fascicle IlL9,CCITTBlueBook,1988,pp. lll- 240. AmericanNationalstandardsfor Telecommunications,,.Integrated servicesDigital Network-Basic AccessInterfacefor use on Metallic Loopsfor Applicationon the NetworkSideof theNT-LayerI Specification," ANSI TI.60l-1988. R. Komiya, K. Yoshida,and N. Tamaki,"The Loop coveragecomparisonbetween TCM andEchocancellerundervariousNoiseconsiderations,"rEEE Transactions on Communications, Nov. 1986,pp. 1058-1067. [email protected] ISDN BasicRateInrerfacespecification,5E5 GenericprogramIssue I.00,AT&T, Dec.1987,pp.II-B-t/15. s. sugimoto' K. Hayashi,and F. Mano, "Design of 2Ble rransceiverfor ISDN subscriberLoops,"IEEEIntemationalconferenteon communication,Iune19g9,pp, 8 . 1r.- 8 .1 . 5 . J. w. Lechleider,"Line codes for Digital subscriberLines,"IEEE communications Magazine,Sept.1989,pp.25-32. w' chen andD. waring, "Applicability of ADSL to supportvideo Dial rone in the CopperLttop,"IEEE Communications Magazine,May 19g4,pp, 102_109, "Networkand customerInstallationInterfaces-AsymmetricDigital subscriberLine (ADSL)MetallicInrerface," ANSI Tl.4t3-1995,Aug. 1995. B' saltzberg,"comparisonof single-carrierand Multitone Digital Modulationfor ADSL Applications,"IEEE Communications Magasine,Nov. 199g,pp. ll4_lLL. w. Y. chen, DSL simulation Techniquesand standardsDevelopment Digital for subscriberLine,sysferns, MacmillanTechnicalpublishing,Indianapolis,l9gg. J. M. Cioffi, V. Oksman,J. J. Werner,T. polletr,p. M. p. Spruyt,J. S. Chow,andK. S. Jacobsen,'-very-High-speedDigital subscriber Lines," IEEE communications Magazine,Apr. 1999,pp.72-79.

PROBLEMS 517 "Functional

Criteria for Digital Loop Carrier Systems," Technical ft.eference, TR-NWT-000057, Bellcore, Morristown, NJ, Jan. 1993. "Integrated Digital Loop Carrier System Generic Requirements, Objectives and t3 Interface," Generic Requirement, GR-NWT-000303, Bellcore, Morristown, NJ' Dec' I 998. "Generic 14 Operations Interfaces Using OSI Tools; Information Model for Integrated Digital Loop Carrier and Fiber in the l,oop Systems," Generic Requirements, GR-NWT-002833, Bellcore, Morristown, NJ, Dec. 1996' 12

r5 l6

t7 l8 19

"V-Interfaces at the Digital Local Exchange RecommendationG.964 (06194), (LE)-VS.l-Interface (Basedon 2048kbiVs)for theSupportof AccessNetwork(AN)' ITU, GenevaSwitzerland, RecommendationG.965 (03/95)-V-Interlaces at the Digital Local Exchange (LE)-VS.2Interface(Basedon 2048kbit/s)for the Supportof AccessNetwork(AN). ITU, Geneva,Switzerland. "The V'34 High-Speed G. D. Forney,L. Brown, M. V. Eyuboglu,and J. L. Moran, Maga4ine,Dec. 1996,pp' 28-33. ModemStandard,"IEEE Communications B. Townsend, "High Speed CommunicationsSystem for Analog Subscriber Sept.l, 1998. Connections," U.S.Patent5,801,695, "TheCapacityof PCM VoicebandChannels," I. Kalet,J. E. Mazo,andB. R, Saltzberg, 1993,pp. 507*51l. Conference. IntemationalCommunications "High Speed E. Ayanoglu, G. Golden, R. Jones, J, Mazo, and D. Shaw, Quantization-Level-Sampling Modem with Equalization Arrangement," U.S. Patent 5,528,625,June 18, 1986.

PROBLEMS systemutil11.1 Determinethemaximumloop lengthof a ping-pongtransmission izing 8-kHzdataburstsof 50 psecdurationin eachdirection.Assumethe veis one-thirdthe speedof light. locity of propagation ll,2 Determinethemaximumtheoreticaldataratethatcanbe achievedby a voicebandmodempassingthroughthepublicnetworkanda singleuniversaldigital loop carriersystem.Assumethereare no signalimperfectionsin the digital poftionsof thefacilities. Determine the range of data rates achievable with a V.90 modem passing "robbed bit" signaling. through six Tl facilities with 11.4 Assume a multipair cable systemis used to carry bidirectional high-bandwidth AMI signals and that near-endcrosstalk coupling from one pair to another is IVo (-2O dB). If the systemis crosstalklimited, determinethe maximum num ber ofactive pairs for a bit error rate of 10-6.

r1-3

marginfor noiseandotherim11.5 RepeatProblemI1.4 with a 3-dB interference perfections. t',t#Fi

It#t",*" x h't

\t'l\\

12 TRAFFICANALYSIS of networkis composed loops,a telephone Exceptfor stationsetsandtheirassociated interstage a variety of commonequipmentsuchas digit receivers,call processors, switchinglinks, andinterofficetrunks.The amountof commonequipmentdesigned into a network is determinedunder an assumptionthat not all usersof the network needserviceat onetime.Theexactamountof commonequipmentrequiredis unpreNetworksconceivably dictablebecauseof therandomnatureof the servicerequests. couldbe designedwifh enoughcommonequipmentto instantlyserviceall requests peaks.However,this solutionis of very rareor unanticipated exceptfor occurrences muchof thecommonequipmentis unusedduringnormalnetuneconomical because work loads.The basicgoalof traffic analysisis to providea methodfor determining of networks. of varioussizesandconfigurations thecost-effectiveness networkrefersto theaggregateof all userrequestsbeTraffrc in a communicatrons the servicereque$ts ing servicedby the network,As far asthe networki$ concerned, arriverandomlyandusuallyrequireunpredictableservicetimes.The fluststepof trafof traffic arrivalsandservicetimesin a probabilistic fic analysisis thecharacterization framework.Then the effectivenessof a network can be evaluatedin terms of how muchtraff,rcit carriesundernormalor averageloadsandhow often the haffic volume exceedsthe capacityof the network. loss Thetechniques oftraffic analysiscanbe dividedinto two generalcategories: fiystemsanddelay systems.The appropriateanalysiscategoryfor a particularsystem on thesystem'streatmentof overloadtraffic.In a losssystemoverloadtraff,tc depends is rejectedwithoutbeingserviced.In a delaysystemoverloadtrafficis heldin a queue until the facilitiesbecomeavailableto serviceit. Conventionalcircuit switchingoperatesasa losssystemsinceexcesstraffic is blockedandnot servicedwithouta reffy "lost" callsactuallyrepresent a lossofreveon ttrepartofthe user.In someinstances nueto thecarriersby virtueof theirnot beingcompleted. the basic messageor packetswitchingobviouslypossesses Store-and-forward operation however,a packet-switching characteristics of a delaysystem.Sometimes, of a loss$ystem.Limited queuesieesandvirnralcircanalsocontaincertaina$pects duringtraffic overloads.Circuit-switchingnetworks imply operations loss both cuits operations ofa delay naturein additionto the loss operation incorporate certain also 519

520

THAFF|c ANALysts

of the circuits themselves.For example,accessto a digit receiver,an operator, or a call proce$soris normally controlled by a queuing proces$. The basic measureof performancefor a loss systemis the probability of rejection (blocking probability). A delay system, on the orher hand, iJ measured in terms of servicedelays' Sometimesthe averagedelay is desired,while at other times the probability of the delay exceedingsome specified value is of more interest. some of the analysespresentedin this chapter are similar to those presented . in chapter 5 for the blocking probabilities of a switch. chapter 5 is concerned mostly with matching loss-the probability of not being able to set up a connection througir a switch under normal or averagetraffic volumes. This chapter, however, is mostly concernedwith the probability that the number of active sourcesexceedssome specified value' Typically, the specified value is the number of trunk circuits in a route.

12.1 TRAFFICCHARACTERIZATION Becauseof theratrdomnatureof networktraffic,thefollowing analysesinvolve certain fundamentals of probabilitytheoryand srochasticpro""r*r*. L tni, heatment only the mostbasicassumptions andresultsof traffic analysisarepresented. The intentis to provideanindicationofhow to applyresultsofhaffic analysis,not to delve deeplyinto analyticalformulations.However,a few basicderivationsarepresented to acquainttheuserwith assumptions in themodelssotheycanbeappropriately applied. In the realm of applied mathematics,where thesesubjectsare nlatea more formally, blockingprobabilityanalysesarereferredto as congestiontheoryanddelay analysesarerefenedto asqueuingtheory.Thesetopicsarealsocommonlyreferred to astraffic flow analysis.In a circuit-switched network,the..flow" of messages is not so much of a concernas are the holding times of cornmonequipment.A circuitswitchednetworkestablishes an end-to-endcircuitinvolvingvariousnetworkfacilities (transmission links andswitching$tages)that areheldfor the durationof a ca1l. From a networkpoint of view, it is the holdingof thesere$ources that is important, not theflow of informationwithin individualcircuits. on theotherhand,message-switching andpacket-switching networksaredirectly concernedwith the actualflow of information,sincein thesesystemstraff,rcon the transmission links is directlyrelatedto the activityof the sources. As mentionedin Chapter7, circuitswitchingdoesinvolvecertainaspects of traffic flow in theprocessof settingup a connection. connectrequests flow from thesources to the destinations acquiring,holding,andreleasingcertainresources in theprocess. As wasdiscussed, controllingtheflow of connectrequestsduringnetworkoverloads is a vital functionof networkmanagement. Theunpredictablenatureof communicationstraffic arisesasa resultof two underlying randomprocesses: call arrivalsandholdingtimes.An arrivalfrom anyparticular useris generallyassumed to occurpurelyby chanceandbetotallyindependent ofar_ rivals from otherusers.Thusthenumberof arrivalsduringanyparticulartime interval is indeterminate. In mostcasesholdingtimesarealsodistributedrandomly.In some applications this elementof randomness canbe removedby assumingconstanthold-

521 tz.t rBAFFtc cHAHAcrEFlzATloN to a neting times(e.g.,fixedJengthpackets).In eithercasethetraffic loadpresented on both the frequencyof arrivalsandthe average work is fundamentallydependent situationin which holdingtime for eacharrival.Figure12.1depictsa representative The are 20 sources unpredictable. different of the times holding the arivals and both while top displays source the individual of each bottomof the figuredepictsactivity total of all activity.[f we assumethatthe20 sourcesareto be contheinstantaneous nectedto a hunk group,theactivitycurvedisplaysthenumberof circuitsin useat any particulartime.Noticethatthemaximumnumberof circuitsin useat anyonetimeis 16 andthe averageutilizationis a little underI I circuits.In generalterms,thetrunks arereferredto asservers,anda trunk group is a serYergroup' Traffic Measurementg of networkcapacityis thevolumeof traffrccarriedovera periodof time. Onemeasure Traffic volumeis essentiallythe sumof all holdingtimescarriedduringtheinterval. in Figure12.1is theareaundertheactivitycurve(apThetraffic volumerepresented proximately84 call minutes). A moreusefulmeasureof traffic is the traffic intensity(alsocalledtraffic flow). Trafficintensityis obtainedby dividingthetrafficvolumeby theIengthof timeduring theaverageactivityduringapeThustraffic intensityrepresents whichit is measured. dimenriod of time (10.5in Figure12.1).Althoughtraffic intensityis fundamentally in units of erlangs,afterthe sionless(time dividedby time), it is usuallyexpressed Danishpioneertlaffic theoristA. K. Erlang,or in termsof hundred(century)call secondsperhour(CCS).TherelationshipbetweenerlangsandCCSunitscanbe derived by observingthatthereare3600secin anhour:

Es I

E.E !s E E t

Tims lfiinu$rl

Flgure 12.1 Activity profile of networktraffic (all callscanied).

522

TRAFFICANALYSIS

I erlang= 36 CCS Themaximumcapacityof a singleserver(channel)is l erlang,whichis to saythat theserveris alwaysbusy.Thusthemaximumcapacityin erlangsof a groupof servers is merelyequalto thenumberof seryers.Because traffic in a losssysLmexperiences infiniteblockingprobabilitieswhenthetrafficintensityis equalto thenumberof servers,theaverageactivityis necessarily lessthal thenumberof servers.similarly,delay rrystems operateat lessthanfull capacity,on average,becauseinfinite delaysoccur whentheaverageloadapproaches thenumberof servers. Two importantpalametersusedto characterizemafficarethe averagealival rate L andtheaverageholdingtime t*. If thetrafficintensityl, is expressed in erlangs,then

a =fu*

(12.1)

whereI' andf* areexpressed in like unitsof time (e.g.,callsper secondandseconds per call,respectively). Noticethat traffic intensityis only a measureof averageutilizationduringa time periodanddoesnot reflectthe relationshipbetweenarrivalsand holdingtimes.That is, manyshortcallscanpraducethesametraffic intensityasa few longones.In many of the analysesthat follow the resultsaredependent only on the trathc intensity.In someca|tes, however,theresultsarealsodependent on theindividualarrivalpattems andholdingtime distributions. Publictelephonenetworksaretypicallyanalyzedin termsof the average activity duringthebusiesthourof a day.Theuseof busy-hourtrafficmeasurements to design andaralyzetelephonenetworksrepresentrr a compromisebetweendesigningfor tle overallaverageutilization(whichincludesvirtually unusednighnimehours) anddesigningfor short-duration peaksthat may occurby chanceo, *, u resurtof rv commercialbreaks,radiocall-inconte$ts, andsoon. Busy-hourtrafficmeasurements indicatethatanindividualresidentialtelephone is typicallyin usebetween5 and l0zo ofthe busyhour.Thuseachtelephone represenm a traffic loadof between0.05and0.l0 errangs.Theaverageholdingtime is between 3 and4 min, indicatingthara typicartelephoneis involvedin oneoi two phone calls duringthe busyhour. Businesstelephonesusuallyproduceloadingpatternsdifferentfrom residential phones'First, a businessphoneis generallyutilizedmoreheavily. second,the busy hourofbusinesstraffic is oftendifferentfrom thebusyhourofresidential traffic.Figute 12.2showsa typicalhourly variationfor both sourcesof traffic. The trunks of a telephonenetworkaresometimes designedto takeadvantage of variationsin calling patternsfrom differentoffices.Toll connectingtrunksfrom residential areasareoften busiestduring eveninghours,and trunksfrom businessareasareobviously busiest duringmidmorningor midafternoon. Traffic engineering depends not only on overall traffic volumebut alsoon time-volumetraffic patternswithin the network. A certainafirountof caremustbe exercisedwhendeterminingthetotal traffic load of a systemfrom theloadingof individuallinesor trunks.For example,sincetwo tele-

12.1 TRAFFICCHARACTEFIEAT|0N523

: I

F

t

o

rillAr'r2

3

.

t

6

t

t

t

l0 ll

l2H I

t

3

4

I

6

t

I

ftuq ol dof I

oti(|;uomrdi*ia

t

ofi*

. *.ll.,ttot

or'i"

on time of day. Figure 12.2 Traffic volumedependence phonesare involved in eachconnection,the total load on a switching systemis exactly one-half the total of all traffic on the lines connectedto the switch. ln addition, it may be important to include certain setupand rcleasetimes into the averageholding times of some common equipment. A lQ-sec setup time is not particularly significant for a 4-min voice call but can actually dominate the holding time of equipment used for short data messages.Common equipment setup times also become more significant in the presenceofvoice traffic overloads.A greaterpercentageofthe overall load is representedby call attempt$since they increaseat a faster rate than completions. An important distinction to be made when discussingtraffic in a communications network is the difference between the offered haffic and the carried traffic. The offered trffic is the total traffic that would be carried by a network capable of servicing all requestsas they arise.Since economicsgenerally precludesdesigning a network to immediately cany the maximum offered traffic, a small percentage of offered traffic typically experiencesnetwork blocking or delay. When the blocked calls are rejected by the network, the mode of operation is referred to as blocked calls cleared or lost calls cleared.In essence,blocked calls are assumedto disappearand never return. This assumption is most appropriatefor trunk Broup$with alternateroutert.In this case a blocked call is normally serviced by another tnrnk group and does not, in fact, retum. The carried traffic ofa loss system is always less than the offered traffic. A delay system,on the other hand, does not reject blocked calls but holds them until the necessaryfacitities are available. With the assumptionthat the long-term averageof offered traffic is less than the capacity of the network, a delay system caries all offered

524

TRAFFIc ANALYSIS

ffaffic. If the number of request$that can be waiting for serviceis limited, however, a delay system also takes on properties of a ross system.For exampre,if the queue for holding blocked arrivals is finite, requestsarriving when the queueis full are cleared.

12.1.1 Arrival Distrlbutione The mostfundamental assumption of classicaltraffic analysisis thatcall arivals are independent.That is, an arrivar from one sourceis unrelatedto an arrival from any othersource.Eventhoughthis assumptionmay be invalid in someinstances, it has generalusefulness for mostapplications. In thosecaseswherecall arrivalstendto be correlated, usefulresultscanstill be obtainedby modifyinga randomarrivalanalysis. ln this mannerthe randomarrival assumptionprovidesJmathematicalformulation that.catrbe adjustedto produceapproximate solutionsto problemsthatareotherwise mathematically intractable. Negatlva Exponenilal lnterarrlval Timeg Designatetheaveragecall arrivalratefrom a rargegroupof independent sources(subscriberlines)as1,,Usethefollowing assumptions: l. Only oneanival canoccurin anysufficientlysmallinterval. 2. The probability of an arrival in any sufficiently small iriterval is directly proportionalto the length of the interval. (The probability of an arrival is l" lr, whereAr is theintervallength.) 3. The probability of an arrival in any particularinterval is independent of what hasoccurredin otherintervals. It is straightforward [1] to showthatrheprobabilitydisnibutionof interarrivaltimes is P()(I"t) = e-M

(r2.2)

Equationl?'2 definestheprobabilirythatno arrivalsoccurin a randomly selected intervalt' This is identicalto theprobability that f secondselapsefrom one arrival to the next. Example12.1. Assumingeachof 10,000subscriberlines originate one cail per hour,how oftendo two callsarrivewith lessthan0.01secbetweenthemt Solutinn.

The averageanival rateis l, = 3600/10,000 = 2.78arrivals/sec

FromEquationlZ.z,theprobabilityof no arrivalin a O.0r-secintervalis

525 12.1 TRAFFICCHARACTERIZATION = 0.973 PoQ'0215): e4'o?78 'lhus

2.7Voof the arrivats occur within 0.01 secof the previous arrival. Since the arrival rate is ?.78 arrivatsper second,the rate ofoccurrence ofinteranival times lessthan 0.01 secis

2.78x0.027 = 0.075times/sec madein derivingthenegativeexponentialarrivaldistriTheflrst two assumptions howThe third assumptron, butioncanbe intuitivelyjustifiedfor mostapplications. ever,impliescertainaspectsof the sourcesthat cannotalwaysbe supported.First, certainevents,Suchas televiSioncommercialbreaks,might stimulatethe sourcesto placetheir callsat nearlythe sametime.In this casethe negativeexponentialdistributionmay still hold but for a muchhighercallingrateduringthecommercial. involvesthenumarrivalassumption A moresubtleimplicationof theindependent notjust theircallingpattern$.Whentheprobabilityof ananival in any berof sources, small time intervalis independentof other arrivals,it implies that the numberof sourcesavailableto generaterequestsis constant.If a numberof arrivalsoccurimmediatelybeforeanysubintervalin question,someof thesourcesbecomebusyandcanrequests. Theeffectof busysourcesis to reducethe averagearrivalrate. not generate Thusthe interarriva'ltimesarealwayssomewhatlargerthan whatEquation12.2predicts themto be. The only time the arrival rateis truly independentof sourceactivity is whenan infinite numberof sourcesexist. If the numberof sourcesis largeandtheir averageactivityis relativelylow, busy reducethearrivalrate.Forexample,consideranendoffice $ources do notappreciably with 0.1 erlangof activityeach.Normally,thereare thatservices10,000subscribers newartivals.Ifthe numavailableto generate 1000activelinks and9000subscribers by anunlikely 507oto 1500activelines,thenumber increa$es berof activesubscribers Thusthearrivalrateis relareducesto 85fi), achangeof only 5.6Vo. of idle subscribers tively constantovera widerangeof sourceactivity.Wheneverthearrivalrateis fairly constantfor the entirerangeof normal sourceactivity, an infinite $ourceassumptron is justified. in Chapter5 Actually,someeffectsof finite $ourceshavealreadybeendiscussed Lee graph pointed out that It is switch. probabilities of a blocking when analyzing if somenumberof interstage analyses overestimate theblockingprobabilitybecause, links in a groupareknownto bebusy,theremaininglinks in thegrouparelesslikely analysisproducesa morerigorousandaccuratesolutionto to be busy.A Jacobaeus theblockingprobability,particularlywhenspaceexpansionis used.Accurateanalysesofinterarrivaltimesfor finite sourcesarealsopossible.Theseareincludedin the to follow. blockinganalyses Poisson Arrlval Dlstributlon providesa meansof determiningthedistributionof interarrival Equation1.2.2merely times.It doesnot, by itself, providethe generallymoredesirableinformationof how

526

TRAFFICANALYSIS

manyarrivalscanbeexpectedto occurin somearbitrarytimeinterval.using thesame assumptions presented, however,theprobabilityofj arrivalsin aninterval, canbede_ termined[] as

r,{M)=p

r-u

(12.3)

Equation12.3is thewell-knownpoissonprobabilitylaw. Noticethatwhenj = 0, the probabilityofno arrivalsin an intervalr is ps(r),asobtainedin Equation 12.2. Again,Equation12.3assumes arrivalsareindependent andoccurat a givenaveragerate1,,irrespectiveof the numberof arrivalsoicuning just prior to an interval in question'Tfius the Poissonprobabilitydistributionstrouldonty be usedfor arrivals from a largenumberof independent sources. Equationl?.3 definestheprobabilityof experiencing exactlyjardvalsin, seconds. Usuallythereis moreinterestin determiningtheprobabilityofj or morearrivals in r seconds;

p=Itu) =Er,fUt ej

./:l

: r -Er,flrt r'{

= I _p*lftf)

(r2.4)

whereP;(l,r)is definedin Equation12.3. Example12.2. Given a message-swirching nodethat normally experiencesfour arrivalsper minure,what is the probabilitythat eight or more arrivals occur in an arbitrarilychosen30-secinterval? solution,

Theaveragenumberof arrivalsin a 30-secintervalis

L r = xa # = ? Theprobabilityof eightor morearrivals(whentheaverageis 2) is

=i ",,r, P>8(2) r=8

I2.1 TRAFFICCHARACTERIZATION

527

7

=t-Lr,(z) r{

=r-a{r

) " *' {'r!

) = 0.0011 Exampte12.3. What is the probability that a 1000-bitdata block experiences exactlyfour errorswhile beingtransmittedovera transmissionlink with alit enor rate (BER) of l0-5? Solation, Assuming independenterrors (a questionableassumptionon many transmissionlinks), we canobtainthe probability of exactlyfour errorsdirectly from the Poisson distribution. The average number of errors (arrivals) lut= 101x tO-s= 0.01.Thus

= P4(0.01) = prob(4errors) ry

- 4.125x 16-10 e-{'or

An alternativesolutioncanbe obtainedfrom the binomial probability law:

: oo*Jtt t - p)ee6 prob(4errors) [t = 4.I01x 10-10

wherep= lfl-s

As canbe seen,thetwo solutionsof Example12.3arenearlyidentical.Theclosenessof the two answersreflectsthe fact that the Poissonprobabilitydistributionis often derivedasa limiting caseof a binomialprobabilitydistribution.Becauseit is to a binoeasierto calculate,a Poissondistributionis oftenusedasanapproximation mial distribution' 12.1-z Holdlng Time Distrlbutions Thesecondfactorofffaffic intensityasspecifiedin Equation12.1is theaverageholding time tr-.In somecasestheaverageof theholdingtimesis all thatneedsto beknown aboutholdingtimesto determineblockingprobabilitiesin a losssystemor delaysin to know the probabilitydisribution of a delaysystem.In othercasesit is necessary the holdingtimesto obtainthe desiredresults.This sectiondescribesthe two most holdingtime disnibutions:constantholdingtimesandexponencommonlyassumed tial holdingtimes.

528

rRAFFtcANALysts

ConstantHoldlng Times Althoughconstantholdingtimescannotbeassumed for conventional voiceconver$ations,it is a reasonable assumption for suchactivitiesasper-callcall processing requirements, interofficeaddresssignaling,operatorassistance, andrecorded-*ssag" playback'Furthermore, constantholdingtimesare obviouslyvalid for transmission timesin fixed-lengthpacketnetworks. when constantholding time messages are in effect,it is straightforwardto use Equation12.3to determinethe probabilitydistributionof activechannels.As$ume, for thetimebeing,thatall requests areserviced.Thentheprobabilityofj channelsbe_ ing busyat anyparticulartime is merelytheprobabilitythat; arrivalsoccurredin the time intervalof lengthr* immediatelyprecedingtheinstantin question.sincetheaveragenumberof activecircuits over all time is the traffic intensityA = l,f*, the probability ofj circuitsbeingbusyis dependent only on the traffic intlnsity: P;(i.r-) = Pj(A)

=N7- "a where

(12.5)

l, = arrival rate fm = constflrltholding time A = traffic inten$ity (erlangs)

ExponentlalHolding Tlmes The most commonly assumedholding time distribution for conventional telephone conversationsis the exponential holding time distribution: P(>t) = s-t/t^

(12.6)

yhere r* is the averageholdingtime. Equation12.6specifiesthe probabilitythat a holdingtime exceedsthevaluer. This relationshipcan-bederivedfro* u few simple assumptions concerningthenatureof thecall terminationprocess.Its basicjustification, however,lies in thefact thatobservations ofactual uoice"onuersations exhibit a remarkablycloseconespondence to an exponentialdistribution, Theexponentialdistributionpossesses the curiouspropertythattheprobabilityof a terminationis independent of how longa call hasteen in progress. Thaiis, no matter how long a call hasbeenin existence, theprobabiliryofit Lsting anotherr seconds is definedby Equation12'6.In this senseexponentialholdingtimesrepresentthemost randomprocesspossible.Not evenknowledgeofhow longa call hasieenin progress providesanyinformationasto whenthecall will terminaie. combininga Poissonarrivalprocesswith an exponentialholdingtime process to obtainthe probabilitydisribution of activecircuitJis morecompilated thanit was for constantholdingtimesbecausecallscanlast indefinitely.rhe final result,however'provesto be dependent on only theaverageholdingtime.ThusEquation12.5is

12.I TRAFFIC CHARACTERIZATION529

holdingtimesaswell asfor constantholdingtimes(or anyholdvalid for exponential ing time distribution).Equation12.5is thereforerepeatedfor emphasis;The probability of j circuitsbeingbusy at any palticularinstant,assuminga Poissonarrival processandthatall requestsareservicedimmediately,is

PIA)=fte-A

(r2.1)

whereA is ttretraffic inten$ityin erlangs.This resultis true for any distributionof holdingtimes.

Example12.4. Assumethat a tnrnk group has enoughchannelsto immediately carry all of thetraffic offeredto it by a Poissonprocesswith an arrival rateof onecall per minute.Assumethat the averageholding time is 2 min. What percentageof the total traffic is carriedby the first five circuits, andhow much traffic is carriedby all remainingcircuits? (Assumethat the haffic is always packedinto the lowest numberedcircuits.) thesystemisA=I x2=2erlangs. Solutinn. Thetrafficintensity(offeredload)of Thekaffic intensitycarriedby I activecircuitsis exactlyi erlangs. Hence the traffic carried by the fust five circuits can be determined as follows;

+ 5Pr(?) As = 1Pr(2)+2Pr(2)+ 3Pr(2)+ 4Pa(2)

=e

^z + . Z x z ?*. -33xi z_3*.-4Tx-z a, s x z s)

zl

st )

= l.g9 erlangs All of the remainingcircuits carry 2-1.89=0.llerlang theprincipleof diminishingreturnsasthe Theresultof Example12.4demonstrates greater andgreaterpercentage$ ofthe offered capacityofa systemis increasedto carry traffic. The first five circuitsin Example12.4carry94.57oof the traffic while all remainingcircuitscarryonly 5.5Voof theftaffic. If thereare 100sources,95 exka circuitsareneededto carrythe5.5Vo.

530

TRAFFICANALYSIS

12.2 LOSS SYSTEMS Examplel2'4 providesanindicationof theblockingprobabilitiesthatarisewhenthe numberof servers(circuits)is lessthanthemaximumpossibletraffic load(numberof sources). The exampledemonstrates thatg4.5%o of the traffic is carriedby only five circuits'Theimplicationis thattheblockingprobability,if only five circuitsareavailableto carrythetraffic,is 5.svo.Actually,Example12.4is carefullywordedto indi_ catethatall ofthe offeredtraffic is carriedbut thatonly thetraffic carriedby thefirst five circuitsis of interest.Thereis a subtlebutimportantdistinctionbetweentheprobability that six or morecircuitsarebusy(ascanbe obtainedfrom Equation12.7)and theblockingprobabilirythatariseswhenonly five circuirsexisr. Thebasicreasonfor thediscrepancy is indicatecl in Figure12.3,whichdepictsthe sametrafficpatternarisingfrom z0 sources asis shownpreviouslyin Figuretz. t. nlgure l2'3, however,assumes thatonly l3 circuitsareavailableto carryttretraffic.thus thethreearrivalsat t = 2.2,2.3,and2.4 min areblockedandassumed to haveleft the system.Thetotalamountof traffic volumelost is indicatedby theshadedarea,which is thedifferencebetweenall trafficbeingservicedasit arriverandtrafficbeingcarried by a blockedcallsclearedsystemwith l3 circuits.Themostimportantfeatureto notice in Figure I2.3 is that the call arrivingat r = z.g is not blocied, eventhoughthe originalprofile indicatesthatit arriveswhenall 13circuitsarebusy.Thereasonit is not blockedis thatthepreviouslyblockedcallsleft thesystemandthereforereduced thecongestionfor subsequent arrivals.Hencethepercentage of time thattheoriginal traffic prof,rleis at or above13is not the $ameastheblockingprobabilitywhenonly 13circuitsareavailable.

gs E.E Efi .tE P E

= E G

Tim6 (rninutEl Figure

12.3

Activity

profile of blocked calls cleared ( I 3 channels).

12.2 LOSSSYSTEMS 531

12.2.1 Loet CallsCleared The first personto accountfully andaccuratelyfor theeffectof clearedcalls in the calculationof blockingprobabilitieswasA. K. Erlangin 1917.In this sectionwe discussErlang'smost often usedresult;his formulationof the blocking probability for a lostcallsclearedsystemwith Poissonarrivals.RecallthatthePoisson arrival assumptionimplies infinite sources.This resultis variouslyreferredto as Erlang'sformulaof the first kind, El,y(A);the Erlang-Bformula;or Erlang'sloss formula. A fundamentalaspectof Erlang'sformulation,anda key contributionto modern processtheory,is theconceptof statisticalequilibrium.Basically,statistical stochastic equilibriunrimpliesthattheprobabilityof a system'sbeingin a particularstate(numof the time at whichthe system ber of busycircuitsin a trunk group)is independent is examined.For a systemto be in statisticalequilibrium,a longtime mustpass(severalaverageholdingtimes)from whenthe $ystemis in a knownstateuntil it is again examined. For example,whenatrunkgroupfirst beginsto accepttraffic,it hasnobusy the systemis mostlikely to haveonly a few busy circuits.For a shorttime thereafter, circuits.As time passes,however,the systemreachesequilibrium.At this point the mostlikely stateof the systemis to haveA = I'rr busycircuits' Whenin equilibrium,a $ystemis aslikely to havean arrivalasit is to havea terA, de' to increaseabovetheaverage mination.If thenumberof activecircuitshappens parturesbecomemorelikely thanarrivals.Similarly,if the numberof activecircuits happensto dropbelowA, an arival is morelikely thana depadure.Thusif a system is perturbedby chancefrom its averagestate,it tendsto return. it is notpreAlthoughErlang'selegantformulationis notparticularlycomplicated, The interresults. application of the we aremostlyinterestedin sentedherebecause estedreaderis invitedto seereference[2] or [3] for a derivationofthe result:

B=Erl,A)= Mffi

(12.8)

whereN = numberof servers(channels) ,4 = offeredtraffic intensity,\,t^ (erlangs) Equation12.8specifiestheprobabilityof blockingfor a systemwith randomarrivals from an infinite sourceand arbitraryholding time distributions.The blocking probabilityof Equation12.8is plottedin Figure12.4asa functionof offeredtraffic of Erintensityfor variousnumbersof channels.An oftenmoreusefUlpresentation theoutputchannelutilization lang'sresultsis providedin Figure12.5,whichpresent$ Theoutpututilizationp repfor variousblockingprobabilitiesandnumbersof servers. resentsthe traffic carriedbv eachcircuiu

j

h

g sT - Ug F E

@ =

t

;

6 t

H b

J E

X H

= ! t

E . E

? E

= €

o t r F o . E b o L ' t r .it X q cl P HO tl

g ^rilHrqod 0sHoolg

\

I

\ \

l \\

\

\

\

\\ \

i q

\

\'

\

\

\

\

\

\

8 d

\

I

h

r

i

H u E

I

\

\ \

\ \

\

I

\

\ \

U o C

t r H r E ; =

\

\

t

(

rr

L

aj

\

a '

_Q

i\

\

r

J

d

le

\ \

\

\

O i

\

\

E

H

t

EJ

(

€ *

o

E

ol Fi

\ \

E

U?

I

\ \

t

E v

14.'

\

\

!

f t r t

$ \

\

o

E

a , H rlr .5 . a =

(\ Jt

\

t

$

JT

t

q = i !

' i

P

\

\

b0

\

\ \

\

l-,

\ \

\

,r\ \ \ \

oj (€ ' tr ec h t q ou lo r ot t o ' ? o t o ! -o 1

(uotterilrrn tauueqclndtno)d

o

TRAFFICANALYSIS

(1-B)4

P=-F whereA N B (l -B)A

(12.e)

offered traffic number of channels blocking probability carried traffic

Blockingprobabilitiesarealsoprovidedin tabularform in AppendixD. Example 12.5. A Tl line is to be usedasa tieline trunkgroupberweentwo pBXs. How muchhaffic canthe trunk groupcany if the blockingptouuuitityis to be 0.1? What is the offeredtraffic intensitv? Soluti'on- From Figure 12.5it can be seenthat the outputcircuit utilization for B = 0.I andN = 24 is 0.8.Thusthecarriedtraffic intensityis 0.g x24 = 19.2erlangs. sincetheblockingprobabilityis 0.l, themaximumlevelof offeredtraffic is lq?

A=ffi=2l.3erlangs Example 12.6. Fourclustersof dataterminalsareto beconnected to a computerby way of leasedcircuits,as shownin Figure 12.6.In Figure lz.6athetraffic from the clustersusesseparate group$of sharedcircuits.In Figure lz.6b thetraffic from all clustersis concentrated onto one commongroup of circuits.Determinethe total numberof circuits requiredin both caseswhen the maximum desiredblocking probabilityis 57o,Assumethat 22 terminalsarein eachclusterandeach terminalis activellVo of thetime. (Usea blockedcallsclearedanalysis.)

Figure 12.6 Dataterminalnetworkof Example10.6:(a) four separate groups;(b) all traffic concenfratedinto onegroup.

1a.aLosssYsrEMS 535 Solution. The offeredtraffic from eachclusteris22x0.l =2.2 erlangs.Sincethe averagenumberof active circuits is much smaller than the numberof sources,an infinite sourceanalysiscanbe used.UsingTableD.l, thenumberof circuitsrequired for B = SVoat a loadingof 2.2 erlangsis 5. Thusthe configurationof Figure 12.54 requiresa totalof 20 circuits. Thetotalofferedtraffic to theconcentrator of theconfigurationof Figure12.6bis 4 x2.2 = 8.8 erlangs.From TableD.1, l3 circuitsarerequiredto supportthe given traffic load. of smalltrafficgroupsinto onelarge Example12.6demonstrates thatconsolidation provide group in traffic can significant savings total circuit requirements.Large groupsare more efficient than multiple small groupsbecauseit is unlikely that the arrismall groupswill becomeoverloadedat the sametime (assumingindependent group. Thus group use idle in another can circuits vals).In effect,excesstrafficin one thosecircuits that areneededto accommodafetraffic peaksbut arenormally idle are utilizedmoreefficientlywhenthe,trafficis combinedinto onegroup.This featureis oneof themotivationsmentionedin Chapter10for integratingvoiceanddatatraffic into a commonnetwork.The total savingsin ffansmissionco$tsis most significant whenthe individual traffic intensitiesarelow. Henceit is the peripheralareaof a network that benefitsthe mostby concentratingthe traffic. The greatercircuit efficiency obtainedby combiningtxaffic into large groupsis of largegroupsizes.This efficiencyof circuitutilioftenreferredto asthe advantage Insteadof interconzationis thebasicmotivationforhierarchicalswitchingstructures. groups eachpair, it between nectinga large numberof nodeswith rathersmall trunk one individual nodes into large trunk is moreeconomicalto combineall traffic from groupandroutethetraffic througha tandemswitchingnode.Figure12.7contrastsa switchingnodeat thecenter.Obviously, meshversusa starnetworkwith a centralized justified whenthesavingsin totalcircuitmiles thecostof thetandemswitchbecomes is largeenough.

(a) Flgure

12.7

Use of tandem switching

ft) to concentrate ffaffic:

(a) mesh; (b) star.

536

THAFFIC ANALYSIS

Example 12.7. what happensro rhe blocking probabiliries in Figure rz.6a andb discussedin Example 12.6 when the traffic intensity increase$by 50va? soluti.on. If the traffrc intensity of eachgroup increasesfrom z.zto3.3erlangs, the blocking probability of the configuration of Figure 12.6aincreaserrfrom 5Zoto almost l4Va. In the configuration of Figure 12.6b a 507oincreasein the traffic intensitv causes a 400Voincreasein the blocking probability (from 5 to ZOVo). Example 12.7 demonstrate$some important consideration$in network design. As indicated, blocking probabilities are very sensitive to increasesin traffic intensities, particularly when the channelsare heavily utilized. Becauselarge trunk groups utilize their channelsmore efficiently, they are more vulnerable to traffic increasesthan are a number of smaller groups designedto provide the same grade of service. Furthermore, failures of equal percentagesof transmissioncapacity affect the performanceof a large group more than the performarce of severalsmall groups.In both casesthe vulnerability of the large group$ arisesbecauselarge groups operatewith less sparecapacity than do multiple small groups. A secondaspectof blocking analysesdemonstratedin Example l Z.7 is that the calculatedresultsare highly dependenton the accuracyofthe traffic intensities.Accurate valuesof traffic intensitiesare not always available.Furthermore,even when accurate traffic measurementsare obtainable,they do not provide an absoluteindication of how much growth to expect.Thus only limited confidence can be attachedto calculations of blocking probabilities in an absolutesense.The main value of theseanalysesis that they provide an objective meansof comparing various network sizes and configurations' The most cost-effectivedesignfor a given gradeof serviceis the one that should be chosen,even if the traffic statisticsare hypothetical. If a network is liable to experience wildly varying traffic patternsorrapid growth, thesefactors mustbe considered when comparing design alternatives.A network with a somewhat larger initial cost may be more desirableif it can absorbor grow to accommodateunanticipatedtraffic volumes more easily.

12.2.2 Lost Catts Returning In the lost callsclearedanalyses just presented, it is assumedthat unserviceable requestsleavethe systemandneverretum.As mentioned,this assumption is mostappropriatefor trunk groupswhoseblockedrequestsoverflowto anotherrouteandare usuallyservicedelsewhere. However,Iostcallsclearedanalysesarealsousedin in$tanceswhereblockedcalls do not get servicedelsewhere. In manyof thesecases, blockedcallstendto returnto the systemin the form of retries.someexamplesare subscriber concenhator systems, corporatetie linesandpBX trunks,callsto busytelephonenumbers,andaccessto WATS lines (if DDD altemativesarenot used).This

1z.z LosssYsrEMS 537 sectionderivesblockingprobabilityrelationships for lost callsclearedsystemswith retries. random assumptions regardingthenaThefollowing analysisinvolvesthreefirndamental tureof theretumingcalls: l. All blockedcalls return to the systemand eventuallyget serviced,even if multiple retriesarerequired. The elapsed times between call blocking occurrencesand the generation of retries are random and statistically independentofeach other. (This assumption allows the analysisto avoid complications arising when retries are correlatedto each other and tend to cause recurring traffic peaks at a particular waiting time interval.) The typical waiting time before retries occur is somewhat Ionger than the averageholding time of a connection.This assumptionessentiallystatesthat the $ystem is allowed to reach statistical equilibrium before a retry occurs. Obviously, ifretries occur too soon, they are very likely to encountercongestion "relax." In the limit, if all retries are since the system has not had a chanceto immediate and continuous, the network operation becomes similar to a delay system discussedin later sections of thi$ chapter. In this case, however, the system does not queuetequests-the sourcerldo so by continually "redialing." When consideredin their entirety, theseassumptionscharacterizeretries as being statistically indistinguishable from first-attempt traffic.* Hence blocked calls merely add to the first-attempt call arrival rate. Consider a system with a first-attempt call arrival rate of 1,.If a percentageB of the calls is blocked, B times L retries will occur in the future. Of theseretries, however, a percentageI will be blocked again. Continuing in this manner, the total arival rate l,i after the systemhas reachedstatisticalequilibrium can be determinedas the infinite series

l,' = l, +Bl,+ R7"+B37r-. . . L

(12.10)

I-B whereB is the blockingprobabilityfrom a lost callsclearedanalysiswith traffic inten'tityA' = l,'fm. Equation12.10relatesthe averagearrival ratel,/, including theretries,to the firstattemptarrival rate andthe blocking probability in termsof l,'. Thusthis relationship in terms doesnotprovidea directmeansof determiningl,' or d sinceeachis expressed of the other.However,the desiredresultcan be obtainedby iteratingthe lost calls *First-attempt

traffic is also referred to as demand traffic: the service demands assuming all arrivals are serviced immediately. The offered haffrc is the demand traffic plus the refries,

538

THAFFIc ANALysts

clearedanalysis of Equation 12.8. First, determine an estimateof B using L and then calculatel,'. Next, use l,' to obtain a new value of B anclan updatedvalui of 1,,.continue in this mamer until values of l,' and B are obtainecl. Example 12.8. what is the blocking probability of a pBX to a central office trunk group with I 0 circuits servicing a first-attempt offered traffic load of 7 erlangs?What is the blocking probability if the number of circuits is increasedto 13?Assumerandom retries for all blocked calls. solution. It can be assumedthat the 7 erlangs of haffic arise from a large number of PBX stations.Thus an infinite sourceanalysisis justified. The blocking probability forA = 7 erlangsand N= l0 serversis about 87a.Thus the total offered load, including retries, is approximately 7.6 erlangs. with N = l0 and,4'= 7.6, the blocking probability is llvo. Two more iterations effectively produce convergenceatA, = g erlangs and B = lTvo.rf the number of circuits in the trunk group is increasedto 13, the blocking probability of a lost calls clearedsystemis 1.57o.Thus a first approximation to the retuming traffic intensity is 7/0.985 = 7.1 erlangs. Hence the blocking probability including all retuming traffic increasesonly slightly above the l.5Zo. Example l2'8 demonstratesthat the effect of retuming traffic is insignificant when operating at low blocking probabilities. At high blocking probabilities, however, ir is necessaryto incorporate the efTectsofthe returning traffic into the analysis.This relationship between lost calls cleare{ and lost calls retuming is shown in Figure 12.g. when measurementsare made to determine the blocking probability of an outgoing trunk group, the measurementscanxot distinguish betweenfirst-attempt calls (demand traffic) and retries. Thus if a significant number of retries are contained in the measurements,this fact should be incorporatedinto an analysisof how many circuits must be addedto reduce the blocking of an overloadedtrunk group. The apparentoffered load will decreaseas the number of serversincreasesbecausl the number ofret.0 .5

E .100 E .080 { E .010

.F .om J

t E .oot

u. I

tt,z

u.ir 0J

u,+ 0.,0

u.[ 0.[

o.o 0.8

0,7 0.7

0,9

(ritfigll Of|rttd tilffh Of|rttd rilftTo Inrfirlry Inrfirlry Frr ctunrut (ritfigll Frrctunrd Figure 12.8 Blocking probability of lost calls reruming.

12,2 LoSSSYSTEMS 539

tries decreases. Thus feweradditionalcircuitsareneededthanif no retriesarecontainedin the measurements. 12.2.3 Lost Galls Held In a lostcallsheldsystem,blockedcallsareheldby thesystemandservicedwhenthe facilitiesbecomeavailable.Lost callsheldsystemsaredistinctlydifferent necesrrary from thedelaysystemsdiscussed laterin oneimportantrespect;Thetotalelapsedtime of a call in the system,includingwaitingtime andservicetime,is independent of the waitingtime.In essence, eacharrivalrequiresservicefor a continuousperiodof time andterminates its requestindependently of its beingservicedor not.Figure12.9demthebasicoperationof a lostcallsheldsystem.Noticethatmostblockedcalls onstrates eventuallyget someservice,but only for a portion of the time that the respective sourcesarebusy. Althougha switchedtelephonenetworkdoesnot operatein a lost callsheld manner,somesystem$ do.Lost callsheldsystemsgenerallyarisein real-timeapplications in whichthe sourcesarecontinuouslyin needof service,whetheror not thefacilities are available.Whenoperatingunderconditionsof heavyftaffic, a lostcallsheldsystem typicallyprovidesseryicefor only a portionof thetime a particularsourceis active. Eventhoughconventional circuitswitchingdoesnotoperateaccordingto thetheoreticalmodelof lost callsheld,Bell Systemtraffic engineers haveusedit to calculate blockingprobabilitiesfor trunkgroups[4]. A lostcallsheldanalysisalwaysproduces a largervaluefor blockingthandoesErlang'slossformula.Thusthe lost callsheld designthat helpsaccountfor retriesandday-to-day analysisproducesa conservative

E,:

Es € E -E gG

Figure 12,9 Activity profile sf lost callsheld,

540

THAFFIC ANALYSIS

variations in the busy-hour calling intensities. In contrast, CCITT recommendations [5] stipulateErlang-B formulas should be used in determining blocking probabilities. One example of a system that closely fits the lost calls held model is time assignment speechinterpolation (TASD. A TASI sy$temconcentratessomenumber of voice $ourcesonto a smaller number of transmissionchannels.A sourcereceivesservice(is connectedto a channel) only when it is active. If a source becomesactive when all channels are busy, it i$ blocked ard speechclipping occurs. Each speechsegment starts and stops independenfly of whether it is serviced or not. TASI systemswere originally used on analog long-distancetransmissionlinks such as underseacables. More modern counterpartsof TASI are referred to as digital circuit multiplication (DCM) systems.[n contrast to the original rASI systems,DCM systemscan delay speechfor a small amount of time, when necessary,to minimize the clipping. In this case,a lost calls held analysisis not rigorously justified becausethe total time a speech segmentis "in the sy$tem" increase$as the delay for service increases.However, if the averagedelay is a small percentageof the holding time, or if the coding rate of delayed speechis reduced to allow the transmissionchannel time to "catch up," a lost calls held analysisis stilljustified. Recall that controlling the coding rate is one technique of traff,rcshaping used for transporting voice in an ATM network, Lost calls held systemsare easily analyzedto determinethe probability of the total number of calls in the systemat any one time. Since the duration of a source'sactivity is independentof whether it is being serviced,the number in the systemat any time is identical to the number of active sourcesin a systemcapableof carrying all traffic as it arises.Thus the distribution of the number in the systemis the Poissondistribution provided earlier in Equationl2.3. The probability that i sourcesrequestingserviceare being blocked is simply the probability that i + N sourcesare active when N is the number of servers.Recall that the Poissondistribution essentiallydeterminesthe desired probability as the probability that i + N arrivals occurred in the preceding f. seconds. The distribution is dependentonly on the product of the averagearrival rate l" and the averageholding time tm. Example 12.9. what is the probability that a talk$purt experiencesclipping in a TASI system with l0 sourceriand 5 channels?with 100 sourcesand 50 channels? Assume that the activity factor of each talker is 0.4. (Ignore finite sources..l solution, For the first case, the clipping probability can be determined as the probability that five or more sourcesare busy in a poisson processwith an averageof A = O.4x l0 = 4 busy servers.Using Equation 12.7, ,

, : f, r,r+) prob(crippingl ="*l/ ,#* # * +.#.fr l: o.ru Fi

l,"'

)

With 100sources, theaveragenumberof busycircuitsis,4 = 0.4 x 100: 40.A speech segmentis clippedif 50 or moretalkersareactiveat once.Thusthe clippingprobability canbe determinedas

1z.a LosssysTEMs 541 99

prob(clippingl= f, rr{+O)= 0,04 'Fso Example12.9demonshates that TASI system$aremuchmoreeffectivefor large groupsizesthanfor smallones.The 36Voclippingfactoroccuningwith 5 channels producesunacceptable voicequality.On the otherhand,the4Voclippingprobability for 50 channelscanbe toleratedwhenthe line costsarehigh enough. In reality,thevaluesfor blockingprobabilitiesobtainedin Example12.9areoverly pessimisticbecause an infinite sourceassumption wasused.Thesummations did not includethecaseof all sourcesbeingactivebecause thereneedsto be at leastoneidle sourceto createanarrivalduringthetimecongestion. A moreaccuratesolutionto this problemis obtainedin a latersectionusinga finite sourceanalysis. 12.2.4 Lost Calls Cleared-Finite Sources As mentionedpreviously,a fundamental assumption in the derivationof thePoisson arrivaldistribution,andconsequently loss Erlang's formula,is thatcall arrivalsoccur independently of thenumberof activecallers.Obviously,this assumption canbejustified only whenthenumberof sources is muchlargerthanthenumberof servers. This sectionpresentssome fundamentalrelationshipsfor determiningblocking probabilitiesof lost callsclearedsystemswhenthe numberof sourcesis not muchlarger thanthe numberof servers.Theblockingprobabilitiesin thesecase$arealwaysless thanthosefor infinite rtourcesystem$sincethearrivalratedecreases asthenumberof busysources increases. Whenconsidering finite $ourcesystems, traffictheoristsinffoduceanotherparameter of interestcalledtime congestion. Time congestionis the percentage of time that all serversin a grouparebusy.It is iderticalto theprobabilitythatall server$arebusy at randomlyselectedtimes.However,time congestionis not necessarily identicalto blockingprobability(whichis sometimes refenedto ascall congestion). Time congestionmerelyspecifiestheprobabilitythatall serversarebusy.Beforeblockingcan occur,theremustbe an arrival. In an infinite lrourcesystem,time congestionandcall congestionareidenticalbecausethepercentage ofarrivals encountering all serversbusyis exactlyequalto the (Thefact that all serversarebusyhasno bearingon whetheror not time congestion. an arrivaloccurs.)In a finite sourcesystem,however,thepercentage of arrivalsenis smallerbecause counteringcongestion fewerarrivalsoccurduringperiodswhenall (blockingprobability) $erversarebusy.Thusin a finite riourcesystem,call congestion is alwayslessthanthetime congestion. As anextremeexample,considerequalnumbersof sourcesandservers.Thetime congestionis theprobabilitythatall senersare busy.Theblockingprobabilityis obviouslyzero. The samebasictechniques introducedby Erlangwhenhe determined thelossforfor infinite mula sources canbeusedto derivelossformulasfor finite sources[3]. Us-

542

TRAFF|cANALysts

ing thesetechniques, we find theprobabilityof n serversbeingbusyin a systemwith M sourcesandN serversis

( 1 2 . 1)

wheret is the calling rateper lzle sourceandf* is theaverageholding time. Equation 12.1I is knownasthetruncatedBernoulliandistributionandalsoastheEngsetdishibution. settingn = N in Equation12.11 producesan expression for thetime congestion:

rL[Y)rrr-r'

(r2.r2)

using the fact that the arrival ratewhenN serversarebusyis (M -M)/r4 timesthe arrival ratewhenno serversarebusy,we candeterminetheblockingprobabilityfor lost callsclearedwith a finite sourceasfollows;

t)t^"-r u= [t" ELF,I-tlrrrJ'

(12.13)

t'')

which is identical to P1,,(the time congestion) for M - I sources.

Equationsl2.ll, 12-12,and 12.13areeasilyevaluatedin termsof theparameters ir,'andt-. However,Ll andr. do not, by themselves,speci$,the averageactivity of a source'In a lost callsclearedsystemwith finite sourcestheeffectiveofferedloaddecrease$ asthe blockingprobabilityincreases because blockedcallsleaveanddo not retum' When a call is blocked,the averageactivity of the offering sourcedecreases, which increasesthe averageamountof idle time for that source.The net resultis that U decreases becausethe amountof idle time increases. If the averageactivity of a sourceassumingno traffic is clearedis designateda$p = It*, the valueof l/t canbe determinedas

} j r"*': = * - j I - P(l-B)

(r2.14)

12.?LosssysTHMs543 whereB is theblockingprobabilitydefinedby Equation1?.13. The difficulty with usingthe unblocked$ourceactivity factorp to characterize a $ource'sofferedloadis now apparent. Thevalueof li t* depends on B, whichin furn dependson l"'I*. Thus someform of iterationis neededto determineB whenthe sourcesarecharacterized by p (aneasilymeasuredparameter)insteadof il. If thetotal offeredload is consideredto be Mp, the carriedtraffic is A"*ie*l =MPI

- B)

( I 2.1s)

A tableof traffic capacities for finite sourcesis providedin AppendixD.2, where the offeredIoadA =Mp is listedfor variouscombinationsof M, N, and8. Someof theresultsareplottedin Figure12.10,wheretheycanbe comparedto blockingprobabilitiesof infinite sourcesystems.As expected,infinite $ourceanalyses(Erlang-B) areacceptable whenthenumberof sourcesM is large. Example12,10, A groupof callersgenerate request$ at a rateof f,rvecallsperhour pertelephone (includingincomingandoutgoingcalls).Assumingtheaverageholding time is 4 min, whatis theaveragecallingrateof eachidle source?How manycallers can be supportedby a l2-channel concentrator/multiplexer if the maximum acceptable blockingprobabilityis lVo'! Solution, Sinceeachcalleris typicallyactivefor ?0 min ofevery hour andplaces anaverageof five callsduringthe40 min of inactivity,thecallingratefor idle $ources M = 5l4O= 0.l?5 callsperminute.Theofferedloadfor M sources, assumingall traffic is carried,is 0.33M. TableD.2 mustbe searched to find thelargestM suchthat0.33 M is lessthanor equalto the maximumofferedloadfor B = lVo andN = 12.Using interpolationfor M = 2l revealsthat 12 serverscansupport7.11erlangsat B = LVo. Since2I x 0.33= 6.93is theofferedload,2l sourcesis an acceptable solution.If 22

! 5

c o C

E

t

d

ffi

0.6 0.8 0,7 0.4 Offeredtraffic p€rrErvsr(orlangBl

0.9

0.9

Figure 12.10 Blockingprobabilityof lost callsclearedwith finite sources.

544

THAFFIoANALysts

sourcesare used,the offeredload of 7.26 erlangsis higherthan the 7.04 erlangs obtainablefrom interpolationin TableD.2 asthemaximumofferedloadfor B = lvo. It is worthwhilecomparingthe resultof Example12.10to a resultobtainedfrom aninfinite sourceanalysis(Erlang-B).For a blockingprobabilityof lvo,TableD.l revealsthatthemaximumofferedloadfor 12serversis 5.88erlangs.Thusthemaximum numberof sourcescanbe determined as5.88/0.333= t/.94. Hencein this casean infinite sourceanalysisproducesa resultthatis conservative by l5Vo. 12.2.5 Lost Calls Hetd-Flnlte Sources A lost callsheld systemwith finite sourcesis analyzedin thesamebasicmannerasa lost callsheld $ystemswith infinite sources.At all timesthe numberof calls"in the system"is definedto be identicalto the numberof callsthat wouldbe servicedby a strictlynonblockingservergroup.ThusEquation12.11is usedto derermine theprobability thatexactlyn callsarein the system:

(1,'t-)"

r%[Y)rr,t-r'

[y),",-, ql + l"'r*)tr

(r2.16)

Becauseno calls are cleared,the offered load per idle sourceis not dependenton B:

r,,*=*h=f;

(12.r7)

combining Equationsl?.16 and 12.17producesa more usefulexpressionfor the probabilitythatn callsarein the system:

+=[f;)ot'-p)M-n

(12.18)

If thereareN servers,the time congestionis merelythe probability that N or more serversarebusy:

Pr* EP, n4'l

(12.19)

12.2 LOSSSYSTEMS 545

The blockingprobability,in a lost callsheld sense,i$ theprobabilityof an arrival encounterins try'ormorecallsin the svstem:

ff^ Br,=

f ,prob(anival lrusourcesarebusy) average arrival rate

=iU

t)o'ct - p)Mr-'r

(12.20)

wherep = offeredloadper source M = numberof sources N = numberof servers Exampte12.11. Determinethe probability of clipping for the TASI systems describedin Example12.9.In this case,however,usea lost callsheld analysisfor finite sources. Solution, In this exarnplewe areconcemedonly with theprobabilitythata speech utteranceis clippedfor someperiodof time until a channelbecomesavailable.Thus Equation12.20providesthe desiredanswerusing p = 0.4 for the offeredload per source.In the first case,with 10 sourcesand5 channels, 9

/rt\

=o.z7 Bn= E | ] lto.+fto.o)%n ,=s(-/

In thesecondcasewith 100sourcesand50 seryers.

6)ee-'= o'023 . =,1f;)(o'4)n(0 Theresultsof Example12.11showagainthata TASI systemrequireslargegroup sizesto providelow clippingprobabilities.Whencomparedto theresultsof Example 12.9,theseresultsindicatethat an infinite sourceanalysisoverestimates theclipping probabilityin bothcases(0.36versus0.27and0.04versus0.023).Noticethatthepercentageerrorin the infinite $ourceanalysisis almostidenticalfor the IO-source systemand the l0O-sourcesystem.Hencethe validity of choosingan infinite sourcemodelis more dependenton the ratio of sourcesto serversthanit is on the numberof sources.

546

TRAFFTc ANALYSIs

F E

o

.s o .s g

E 5

E I

o

ll

s

.E 4

.to

$

E

s E o g

E

L

$ourceactivlty(erlangsl F'igure l2.Il

Clipping probability and clipping duration of TASI.

Example I2. I I merely determines the probability

that a speech segment encoun-

ters congestionand is subsequentlyclipped. A complereanalysisof a TASI (or DCM) system must consider the time duration of clips in addition to their frequency of occuffence. In essence,the desired information is representedby the amount of traffic volume in the clipped $egments.weinstein [6] refers to the clippert $egmentsas "fractional speechloss" or simply "cutout fraction." This is not the silne as the lost traffic, since a conventional lost calls held analysisconsidersany arrival that encounterscongestion as being completely "lost"-even if it eventually receives some service.The cutout fraction is determined as the ratio of untransmittedtraffic intensitv to offered traffic intensitv:

iz.g NErwoFrKBLocKtNc pFroBABtLtlES

547

M

n",:i.E(,-1v)P"

(r2.21)

n=N+l

whereM = numberof sources A = ofTeredload,= Mp N = numberof servers Pn= probabilityof n callsin thesystem(Equation12.18) Example12.12, Determinetheaveragedurationof a clip in thetwo TASI systems of Example12.11.Assumethe average durationof a speech$egment is 300 msec. (Theaveragelengthof a speechsegmentis dependent on theactivitythreshold,which alsoinfluencestheactivityfactorp.) Solution. In thefirst case,

:0.05e tn- s)f1,0.)(0.4r(0.6)10-n

a,,=II

n+t

\-/

Thus,on average,5.9Va,ot 17.7msec,of every300-msecspeechsegmentis clipped. Since?7%o of thesegmentsexperienceclipping(Example 12.l l),theaverageduration of a clip for clippedsegments is 0.059/0.27= 22Vo,or 66 msec-an obviouslyintolerableamount.In the secondcasefor 100sourcesand50 channels. 100

- 50) =0.00r u",:+,*E,,n fifl),r.-t*.6)rtx]Thusin thiscaseonly 0.1%of all speechis clipped,whichimpliesthatwhenclipping occurs,300x 0.001/0.023=13msecof thesegment is lost. Example12.12showsthat largegroupsizesnot only greatlyreducethe clipping probabilityof TASI systemsbut alsoreducethedurationof theclips.Therelationship of clippingprobabilitiesandclippingduration(fractionalspeechloss)to groupsize andsourceactivityis providedin FigureI ?.I l. As shown,theclippingprobabilityis extrcmelysensitiveto s(turceactivity(offeredload).For a discussion of theeffectsof clippingon speechquality,seereferences [7] and[8].

12.3 NETWORKBLOCKINGPROBABILITIES In theprecedingsectionsbasictechniques ofcongestiontheoryarepresented to determineblockingprobabilitiesof individualtrunk groups.In this sectiontechniques of calculatingend-to-endblockingprobabilitiesof a networkwith morethanoneroute

548

TBAFFIcANALysts

betweenendpointsis considered.In conjunctionwith calculatingthe end-to-end blockingprobabilities,it is necessary to considerthe interactionof raffic on various routesof a network.Foremostamongtheseconsiderations is the effectof overflow ffaffic from onerouteontoanother.Thefollowingsectionsdiscusssimplifiedanalyses only. More sophisticated techniques for morecomplexnetworkscanbe obtainedin references t9l, t101,andI l]. 12.3.1 End-to-End Blocking Probabilities Generally,a connectionthrougha largenetworkinvolvesa seriesof transmission links, eachone of which is selectedfrom a set of alternatives. Thus an end-to-end blockingprobabilityanalysisusuallyinvolvesa compositeof seriesandparallelprobabilities.The simplestprocedureis identicalro the blockingprobability(marching loss)analysespresentedin chapter5 for switchingnetworks.For example,Figure I 2.I 2 depictsa representative setof alternativeconnectionsthrougha networkandthe resultingcompositeblockingprobability. Theblockingprobabilityequarionin Figure12.12containsseverarsimplifyingassumptions.First, the blockingprobability(matchingloss)of the switchesis not included.In a digital time divisionswitch,matchinglosscanbe low enoughthat it is easilyeliminatedfrom theanalysis.In otherswitches,however,thematchinglossmay not be insignificant.when necessary, switchblockingis includedin the analysisby consideringit a sourceofblocking in serieswith theassociated ffunk groups. when morethanoneroutepassesthroughthe sameswitch,asin nodec of Figure 12'12,propertreatmentof correlationbetweenmatchinglossesis an additionalcomplication.A conservative approachconsidersthematchinglossto be completelycorrelated.In this casethematchinglossis in serieswith thecommonlink. On theother hand,an optimisticanalysisassumes thatthematchinglossesareindependent, which impliesthattheyarein serieswith theindividuallinks.Figure12.13depictsrhesetwo approaches for includingthe matchinglossof switchc into theend-to-end blocking probabilityequationof Figurel2.r2.In rhiscase,thelink from c to D is thecommon link.

1 - qz{t

B=pr11-qpqsl Figure 12.12 hobability graph for end-ro-endblocking analysis.

12.3 NETWoRK BLoGKING PRoBABILITIES 549

{

l E/

-wfl"-

1 -q+

fr%-

@-q\ \0,

ffito'

d

; bo"o ,o,

Qr*= 1- (1 -0r9")Q

B= 1-(1-slSsxl-Q2fte)

g=1-{rofig

Figure 12.13 Incorporatingswirch*matching loss into end-to-endblocking analysis:(a) independent switchblocking;(b) correlatedswitchblocking.

A secondsimplifyingassumption usedin derivingtheblockingprobabilityequation in Figure12.I? involvesassumingindependence for theblockingprobabilitiesof the trunk groups.Thus the compositeblockingof two parallelroutesis merelythe productof the respectiveprobabilities(Equation5.4). Similarly,independence impliesthattheblockingprobabilityof two paths-in series-is I minustheproductof the respectiveavailabilities(Equation5.5). In actualpracticeindividual blocking probabilitiesarenevercompletelyindependent. This is particularlytruewhena large amountof traffic on oneroute resultsasoverflow from anotherroute.Wheneverthe first routeis busy,it is likely thatmorethanthe averageamountof overflowis being divefiedto thesecondroute.Thusan alternaterouteis morelikely to bebusywhena primaryrouteis busy. In a largeprrblic network,trunksto tandemor toll switchesnormally carry traffrc to manydestinations. Thusno onedirectroutecontributesan overwhelmingamount of overflowtraffic to a paniculartrunkgroup.In thiscaseindependent blockingprobabilitieson alternateroutesarejustified.In someinstances public of the network,and oftenin privatenetworks,overflow traffic from oneroutedominatesthetraffic on tandem routes.In thesecasesfailure to accountfor the conelationin blockingprobabilitiescanleadto overlyoptimisticresults. Example 12.13. Two trunk groupsare to be usedas direct routesbetweentws switchingsystems.The first group has 12 channelsand the secondgroup has 6 channels.AssumeI0.8 erlangsof haffic is offleredto the l2-channelgroup and overflowsareofferedto the6-channelgroupwhenthefirst groupis busy.Whatis the blockingprobabilityof thefirst group,andhow muchtraffic overflowsto the second group?Using the overflow traffic volumeasan offeredload, determinethe blocking probabilityof the secondkunk group.What is theprobabilitythatbothtrunk groups arebusy?Comparethis answerto the blockingprobabilityof one t8-channeltunk group.

550

THAFFI0 ANALysts

solution. using a lost calls cleared analysis with an infinite source, we determinethatblockingofthefirsrgroupis l|vo(A= l0.B,N= 12).Thereforethe overflow traffic is 10.8x 0.ltsI.62 erlangs. The blocking probability (assuming random arrivals?) of the secondgroup is o.Svo(A = 1.62, N = 6). The probability that both trunk groups are busy simultaneously can be determined (assuming independence?)as

B=0.15x0.005=0.00075 In contrast,thecorrectblockingprobabilityof an l8-channeltrunk groupwith an offeredloadof 10.8erlangsis B:0.013 Thequestionmarksin the solutionof Example12.13pointto two sourcesof error in the determinationof the first blocking probability value.one error is the assumption ofindependence ofblockingin thetwo trunkgroups.A seconderrorresultsfrom the useof an analysispredicatedon purely random(poisson)arrivalsfor over{low trafficinto thesecondtrunkgroup.Resolutionofthis erroris discussed in thenextsec* tion. separatingthe 18channelsof Example12.13into two groupsis anobviousartifice. This exampleis usefulin thatit demonshates anextremecaseof correlationbetween blockingprobabilitiesof two trunk groups.when correlationexists,the composite blocking probability of a direct route and an alternateroute shouldbe determinedas follows: B = (8,) (8211)

(12.22)

where 81 = blocking probability of group I B2ll = blocking probability ofgroup 2 given that group I is busy In the artificial caseof dividing a trunk group into two subgroups,the conditional blocking probability can be determined as

Brll =B(MN1) ; prob (N serversarebusywhenN, areknown to be busy) P*

-_! 1 V p -n=Nr'n

AN/NI

-_\ a N r un=71;An/nl)

(r2.23)

wherePnis theprobabilityrhatexactlyn of N serversarebusy(Equationl?.3).

pRoBABtLrrES 551 la.s NETWoRK BLoCKTNG

EvaluatingEquation12.23for A = 10.8,Nr = 12,andN = 18 revealsthat the appropriateconditionalprobabilityB2ll for Example12.13 is 0.033.Thusthecomposite blockingprobabilityper Equation12.22isB = 0.15x 0.033= 0.005.The remaining inaccuracy(0.005versus0.013actually)is dueto nonrandomcharacteristics ofoverflow trffic. Equation 12.23is valid only for the contrivedcaseof an altemateroute carrying overflowtraffic from only oneprimaryroute.It canbeused,however,asa worst-case solutionto situationswhereoverflow from oneroutetend$to dominatethe traffic on an alternateroute. The correlationsbetweentheblockingprobabilitiesof individualroutesarisebecausecongestionon onerouteproducesoverflowsthat tendto causecongestionon otherroutes.Extemaleventsstimulatingnetworkwideoverloadsalsocausetheblockblocking Thusa third assumption in theend-to-end ing probabilitiesto be correlated. probabilityequationof Figure 12.12is that traffic throughoutthe networkis independent.If fluctuationsin the traffic volumeon individual links tendto be conelated (presumably of extemaleventssuchastelevisioncommercials, etc.),signifibecause cantdegradationin overall performanceresults. 12.3.2 OverflowTratfic The secondsourceoferror in Example12.13occurredbecausean Erlang-Banalysis usedthe averagevolume of overflow traffic from the first group to determinethe traffic blockingprobabilityof the secondtrunk group.An Erlang-Fanalysisassumes arrivalsarepurelyrandom,thatis, theyaremodeledby a Poissondistribution.However, a Poissonanival distributionis an erroneousassumptionfor the traffic offered to the secondtrunk group.Eventhougharrivalsto thefirst groupmay be random,the overflowprocesstendsto selectgroupsof thesearrivalsandpassthemon to the secondtrunk group.Thusinsteadofbeing randomthearrivalsto thesecondgroupoccur in bursts.This overfloweffectis illustratedin Figure12.14,whichportray$a typical randomardval patternto onehunk groupandthe overflow patternto a secondgroup, If a significantamountof the traffic flowing onto a trunk group resultsas overflow from othertrunk groups,overly optimistic valuesof blocking probability arisewhen all of thetraffic is assumed to be purelyrandom. The mostcornmontechniqueof dealingwith overflow traffic is to relatethe overflow traffic volume to an equivalentamountof randomtraffic in a blocking probability sense.For example,if the 1.62erlangsof overflowtraffic in Example12.12is equatedto 2.04erlangsof randomtoaffic,a blockingprobabilityof LSVIis obtained for the secondtrunk group.(This is the correctprobability of blocking for the second groupsincebothgroupsarebusyif andonly if the secondgroupis busy.) This methodof treatingover{lowtraffic is refenedto astheequivalentrandomtheory [12]. Tablesof traffrccapacityareavailable[3] thatincorporatetheoverfloweffectsdirectlyinto the maximumofferedloads.The Neal-Wilkinsontablesusedby compriseonesuchsetof tables.TheNeal-WilkinsontaBell Systemtraffic engineers bles,however,alsoincorporatethe effectsof day-to-dayvariationsin the traffic load.

552

THAFFICANALYSIS AfflYrh to flil

trunk gru.rp

Drprturd frotn llnt tunk $uup frprclty of fim mrnk ero{F

Olnrflofr rtylnh to rmnd trunk Foup

l i l l

Figure 12.14 Bursty characteristicof overJlowtraffic. (Forty erlangson one day and 30 erlangs on anotheris not the $ameas 35 erlangson both days.) These tables are also used for trunk groups that neither generatenor receive overflow traffic. The fact that cleared traffic doe$not get serviced by an alternate route implies that reffies are likely. The effect of the rehies, however, is effectivelv incorporatedinto the value of B by equivatentrandomness.

12.4 DELAYSYSTEMS The secondcategoryoftelefaffic analysisconcern$systemsthat delay nonserviceable requestsuntil the necessaryfacilities become available. These systemsare variously referred to as delay systems,waiting-call systems,and queuing system$.call arrivals occurring when all serversare busy are placed in a queueand held until servicecommences. The queue might consist of storage facilities in a physical sense,such as blocks of memory in a message-switchingnode, or the queuemight consist only of a list of sourceswaiting for service.In the latter ca$e,storageof the messagesis the responsibility of the sourcesthemselves. using the more general tnrm queueingtheory, we can apply the following analyses to a wide variety of applications outside of telecommunications.Some of the more coiltmon applications are data processing, supermarket check-out counters. aircraft landings, inventory control, and various forms of service bureaus.These and many other applicatronsare consideredin the field ofoperations research.The foundations of queuing theory, however, rest on fundamentaltechniquesdevelopedby early telecommunications traffic researchers.In fact, Erlang is credited with the first solution to the most basic type of delay system.Examplesof delay systemanalysisapplications in telecommunicationsare messageswitching, packet switching, statisticaltime division multiplexing, multipoint data communications,automatic call distribution, digit receiver access,signaling equipment usage,and call processing.Furthermore, many

12.4 DELAY$YSTEMS 553

to corporate PBXshavefeaturesallowingqueuedacce$$ tie linesor WATS lines.Thus somesystemsformerlyoperatingaslosssystemsnow operateasdelaysystems. ln general,a delayoperationallows for greaterutilization of servers(fransmission facilities)thandoesa losssystem.Basically,theimprovedutilizationis achievedbecausepeaksin thearrivalprocessare"smoothed"by thequeue.Eventhougharrivals to thesy$temarerandom,the$erversssea somewhatregulararrivalpattern.Theeffect of the queuingprocesson overloadtraffic is illustratedin Figure 12.15.This figure displaysthe sametraffic patternspresented earlierin FiguresI2.1,12.3,and I2.9. In produceavailable thiscase,however,overloadtrafficis delayeduntil callterminations chamels. thatall traffic offeredto thesystem it is assumed In mostof thefollowinganalyses is thattheofferedhaffic eventuallygetsserviced.Oneimplicationof this assumption intensityA is lessthanthenumberof serversN. EvenwhenA is lessthanN, thereare two casesin whichthecaniedtraffic mightbelessthantheofferedtraffic.First,some sourcesmighttire of waitingin a longqueueandabandontherequest.Second,thecapacityfor storingrequests be rejected mayoccasionally maybefinite.Hencerequests by thesystem. in the following analysesis that infinite $ourcesexist.In a A secondassumption delaysystem,theremay be a finite numberof sourcesin a physicalsensebut an infinitenumberof sourcesin anoperationalsensebecause eachsourcemayhaveanarbitrary numberof requestsoutstanding(e.9., a packet-switchingnode).There are conbut notin theapplications instances in whicha finite sourceanalysisis necessary, sideredhere. An additionalimplicationof servicingall offered traffic ariseswhen infinite source$exist. This implicationis the needfor inJinitequeuingcapabilities.Even

Ee

F G - + - - r - + - - - '

E E.E gs E C

6

Figure 12.15 Activity profile of blockedcallsdelayed(13 servers).

554

THAFFIC ANALYSIS

though the offered traffic intensity is less than the number of servers,no statistical limit exists on the number of arrivals occurring in a short period of time. Thus the queue of a purely losslesssystem must be arbitrarily long. In a practical sense, only finite queuescan be realized, so either a statisticalchanceof blocking is always pr+ sent or all sourcescan be busy and not offer additional traffic. When analyzing delay systems,it is convenientto separatethe total time that a request is in the sysrem into the waiting time and rhe holding time. In delay systems analysisthe holding time is more corlmonly referredto as the servicetime. In contrast to loss systems,delay systemperformanceis ggnerally dependenton the distribution of servicetimes and notjust the mean value Im.Two servicetime distributions are consideredhere;constantservicetimes and exponentialservicetimes. Respectively,these distributions representthe most deterministic and the most random servicetimes possible. Thus a system that operateswith some other distribution of service times performs somewherebetweenthe performanceproducedby thesetwo distributions. The basic purpose of the following analysesis to determine the probability distribution of waiting times. From the distribution, the averagewaiting time is easily determined. sometimes only the average waiting time is of interest. More generally, however, the probability that the waiting time exceedssome specified value is of interest.In either case,the waiting times are dependenton the following factors: I. Intensity and probabilistic nature ofthe offered traffic 2. Distribution of service times 3. Number of servers 4. Number of sources 5. Service discipline of the queue The service discipline of the queue can involve a number of factors. The first of these concerns the manner in which waiting calls are selected.commonly, waiting calls are selectedon a first-come, first-served (FCFS) basis,which is also referred to as first-in, first-out (FrFo) seryice.sometimes,however, the serversystemitself does not maintain a queuebut merely polls its $ourcesin a round-robin fashion to determine which onesare waiting for service.Thus the queuemay be servicedin sequentialorder of the waiting sources.In some applications waiting requestsmay even be selectedat random. Furthermore, additional service variations arise if any of these schemesare augmentedwith a priority discipline that allows some calls to move aheadof others in the queue. A secondaspectof the service discipline that must be consideredis the length of the queue.If the maximum queuesize is smaller than the effective number of sources, blocking can occur in a lost calls sense.The result is that two characteristicsof the grade of service must be considered: the delay probability and the blocking probability' A common example of a system with both delay and loss characteristicsis an automatic call distributor with more accesscircuits than attendants(operatorsor reservationists).Normally, incoming calls are queuedfor service. under heavy loads,

12.4 DELAYSYSTEMS 555

however,blockingoccursbeforethe ACD is evenreached.Reference[4] contains an analysisof a delaysystemwith finite queuesandfinite $ervers. To simplify the characterization of pafiicular systems,queuingtheoristshave adopteda concisenotationfor classifyingvarioustypesof delaysystems.This notato identifyaltion, which wasintroducedby D. G. Kendall,usesletterabbreviations listed.Althoughthediscussions in thisbookdo not ternativesin eachof thecategories sothereadercanrelatethe rely on thisnotation,it is introducedandusedoccasionally of each following discussions to classicalqueuingtheorymodels.The interpretation letteris specifiedin Figure12.16. formatpresented anextension Thespecification in Figure12,16 actuallyrepresents of the formatcommonlyusedby mostqueuingtheorists.Thusthis formatis sometimes abbreviatedby eliminating the last one or two entries.When theseentriesare eliminated,infinite casespecifications sys* areassumed. For example,a single-server tem with randominput andnegativeexponentialseryicetimesis usuallyspecifiedas Mll,lUl. Both thenumberof sourcesandthepermissiblequeuelengthareassumed infinite. 12.4.1 Exponential Servlce Times The simplestdelaysystemto analyzeis a systemwith randomarrivalsandnegative exponentialservicetimes:M/IVI/I.{.Recallthat a randomarival distributionis one with negativeexponentialinterarrivaltimes.Thusin the shorthandnotationof queuing theorists,theletterM alwaysrefersto negativeexponentialdistributions(anM is a purelyrandomdistributionis memoryless). usedbecause

Gonffrl {nqs$urhpt;ofi;

Input rfclflcttlon

Purelyrrndom

{; Gl

ServlcetimE distrihrtiorl

Gcnlfd {no ffiumptiofil

Mr ;{egltiw rxponintirl D: Corltrnt

Numbtr of *rrrot

N:

rinittnumuer

Numbffof rourcdJur

Inrinitc

L*

I I

or*"

|

|

L'

r"r$fi { L*'

Flnlte number

Flnitelensrttr Inflnitclefisth

t t

l2l 3/4/E Figure 12.16 Queueing $ystemnotation.

556

THAFFIo ANALysts

In the IM/lvVl systemand all other systemsconsideredhere,it is assumedthat calls are serviced in the order of their arival. The following analysesal$o assumethat the probability of an arrival is independentof the number of requestsalreadyin the queue (infinite sources).From theseassumptions,the probability that a call experiencescongestion and is thereforedelayed was derived by Erlang:

(r2.24) whereN = numberof servers A = offeredload(erlangs) B = blockingprobabilityfor a lost callsclearedsystem(Equationl2.g) Theprobabilityof delayp(>0) is variouslyrefenedro asErlang'ssecondformula, Ezn(A);Erlang'sdelayformula;or the Erlang-c formula.For single-server $ystems (N= l) theprobabilityof delayreducesto p, whichis simplyrheoutpururilizationor traffic carriedby the server.Thusthe probabilityof delayfor a single-server sysrem is alsoequalto theofferedloadl,t- (assuming fu. < l). Thedistributionof waitingtimesfor randomarrivals,randomservicetimes.anda FIFO servicedisciplineis p(>t) : p(>0) "-{N-A)t/to,

(r2.?5)

wherep(>0) = probability of delay given in Equation 12.24 fm = flY€rfl8oservice time of negative exponential service time distribution Equation 12.25definesthe probability that a call aniving at a randomly choseninstant is delayed for more than t/t^ service times. Figure 12.17 presentsthe relationship of Equation 12.25by displaying the traffic capacitiesof various numbersof serversas a function of acceptabledelay times. Given a delay time objective r,/f., Figure l2.l7a displaysthe maximum traffic intensity if the delay objective is to be exceededfor only lovo of the arrivals. similarly, Figure lz,.r7b displays the maximum traffic intensity if the delay objective is exceededfor only lzo of the arrivals. Notice that atp(>r) = 0.01, the serversystemsdo not approachtheir maximum capacity (number of servers) unlessthe acceptabledelay is severaltimes larger than f*. By integrating Equation 12.25 over all time, the averagewaiting time for all arrivals can be determinedas

;f = - P(>O)r,n N_A

(12.26)

NoticethatTistheexpecteddelayfor all arrivals.Theaveragedelayofonly thosear.


E I

I

B .B E

F

OF

1.6

1,0

2.0

aE

It/t-l

2.0

e.s

(yr-l

(t)

$ o !

u o E

r.0 ,rrr.u

Figure 12.17 Traffic capacityof multiple-serverdelay systemswith exponentialservice times;(a)probabilityof t,p(>t)=lVo. exceeding LpFt)=10To;(b) probabilityofexceeding rivals that get delayedis commonly denotedas * - t *

(r2.27)

v-N-a

Example12.14. A message-switching network is to be designedfor 95Vo utilization of its hansmissionlinks. Assumingexponentiallydistributedmessage per minute, what is the averagewaiting lengthsand an arrival rate of 10 messages time,andwhatis theprobabilitythatthewaitingtime exceeds5 min? network uses a single channel Salutinn. Assumethat the message-switching betweeneachpair of nodes.Thusthereis a singleserveranda singlequeuefor each transmissionlink. Sincep is given to be 0.95 and l, = 10 arrivalsper minute,the

558

TRAFFIcANALYSIS

averageservicetime can be determinedas f* = 0.g5l10= 0.095min. The averase waitingtime (notincludingtheservicetime) is easilydetermined as ;=

o'q5x-q:095 = l.Bo5min I - 0.95

UsingEquation12.25,we candeterminetheprobabilityof thewaitingtimeexceeding 5 min as - 0.068 p(>5)= (0.95)e-tt-o'sr)5/0.0es Thus6.87oof themessages experience queuingdelaysof morethan5 min. Example12.15. Dererminerhenumberof digit receiversrequiredto support1000 telephones with anaveragecallingrateof two callsperhour.Assumethediatingtime is exponentially distibutedwith anaverageservicetimeof 6 sec.Thegradeof service objectiveis to returndial tonewithin I secofthe off-hooksignalfor 99Zoofthe call attempts.comparethe answerobtainedfrom a delay systemanalysisto an answer obtainedfrom a losssystemanalysisatB = lVo. If theblockingprobabilityis lessrhan l%o,fewerthanlVoof thecallsaredelayed. Solution. The calling ratel, andthe offeredhaffic intensiry,4areeasilydetermined as0.555callsper secondand3.33erlangs,respectively. sincethenumberof servers N cannotbe solvedfor directlyfrom theequations, Figure lz.ljb is usedto obtaina valueof eightserversfor t/t^=t. Examinationof TableD.1 revealsthatgg.Svoof the call attemptscanbe serviced immediatelyif thereateninedigit receivers. Thusin thiscasetheabilitv to delavservice providesa savingsofonly oneserver. Example12.15demonstrates rhat a blockingprobabilityanalysisproducesapproximatelythe sameresultsasa delaysystemanalysiswhenthe maximumacceptabledelayis a smallpercentage of theaverageservicetime.Thetwo resultsarealmost identicalbecause, if a digit receiveris not imrnediatelyavailable,thereis only a small probabilitythatonewill becomeavailablewithin a shorttime period.(With an averageservicetime of 6 sec,theexpectedtime for oneof eightdigit receiversto be releasedis 6/8 =0.75 sec.Hencethedelayoperationin thiscaseallowsa savingsof one digit receiver.) Because a digit receivermustbeavailablewithin a relativelyshorttimeperiodafter a requestis generated, digit receivergroupsizingis oftendeterminedstrictlyfrom a blockingprobabilityanalysis.Thefact thatdigit receiveraccessis actuallyoperated asa delaysystemimpliesthe gradeof serviceis alwaysbetterthanthatcalculated. 12.4.2 Constant Service Timee Thissectionconsiders delaysystemswith randomarrivals,constantservicetimes,and a singleserver(IWD/I). Again,FIFO servicedisciplinesandinfinite $ourcesareas-


sumed.Thecasefor multipleservershasbeensolved[3] but is tooinvolvedto include systemswith constantservicetimesareavailablein here.Graphsof multiple-server reference[5]. The averagewaiting time for a singleserverwith constantservicetimes is determinedas Pf'= ze*p)

-

(r2.28)

l2,28producesan average wherep =,4istheserverutilization.NoticethatEquation asingle-server systemwithexponential waitingtimethatisexactlyone-halfofthatfor Exponentialservicetimes averagedelays becausethere causegreater servicetimes. increatingthedelay.Inbothtypesofsystems,dearetworandomprocessesinvolved laysoccurwhena largeburstof arivals exceedsthecapacityof theservers.With exservice ponentialservicetimes,however,long delaysalsoarisebecause of excessive message-switching timesofjust a few arrivals.(Recallthatthisaspectof conventional up into packetsin a packetsystemsis one of the motivationsfor breakingmessages switchingnetwork.) If the activityprofile of a constantservicetime system(M/D/l) is comparedwith theactivityprofile of anexponentialservicetime system(lWIWl), theM/D/l $ystem is seento be activefor shorterandmorefrequentperiodsof time. That is, the Ivl/ii{,/l systemhasa highervanancein the durationof its busyperiods.The averageactivity of both system$is, of course,equalto the serverutilizationp. Hencethe probability systemwith constantservicetimesis identicalto that for of delayfor a single-server exponentialservicetimes:p(>0) = l"t . Theprobability of congestionfor largerN is relatively closeto that for exponential forp(>0) for servicetimes.ThusEquation1?.25canbe usedasa closeapproximation time arbitrary service distributions. multiple-server with systems For single-serversystemswith constantholdingtimes,the probabilityof delay greaterthanan arbitraryvalue f is p(>r)=p[>(ft+ r)t*l k

pili - t/t^1itP$-t/t^)

=t_(1_p)E I4

k

=1-(l-p)er'ieff whereft = largestintegralquotientof t/t* r = remainderof t/t^ P = serverutilization, = fu-

(r2.2e)

560

THAFFIcANALysts

Comparisons of the waitingtime distributionsfor single-server systemswith exponentialand constantservicetimes are shownin Figure l2.lg. For eachpair of curves'theupperoneis for exponentialservicetimesandtheloweroneis for constant servicetimes.Sinceall otherservicetimedistributionsproducedelayprobabilitiesbetweentheseextremes, Figure12.18providesa directindicationof therangeof possibledelays. Example 12.16. A packet-swirching node operateswith fixed-lengthpacketsof 300 bits on 9600-bpslines.If the link utilizationis to be 90%,whatls the averase

q E o o

x o g o

E .E

o &

DElaytime. r/r-

Figure 12.18 Delayptobabilitiesofsingle-server$ystems(exponential andconstantservice times).


of packetsencountermorethan0.35 secof delaythrougha node?What percentage by 107o? delay?Whatis theaveragedelayif theofferedloadincreases Soluti,on. Messagelengthsof 300 bits anda datarateof 9600bps imply that the fixedJengthservicetime is 300/9600= 0.031sec.FromEquation12.28,the average waitingtime is

0.9x 0.031 f = F = 0-. 1 4 2(1 0.e)

sec

processing, is obtained by adding delaythrough Thetotalaverage thenode,excluding the average waiting time to the service time:

Averagedelay= 0.140+ 0.031= 0.171sec Sincethe servicetime is 0.031sec,0.35secof delayoccurswhenthewaitingtime is = 10service to 0.319/0.031 times.FromFigure 0.35* 0.031= 0.319.Thiscorresponds 0.12.Thus lZVoof the 12.18,theprobabilityof delayfor t/t^- 10 is approximately packetsexperience delaysof greaterthan0.35 sec.An increaseof 10Voin the traffic intensityimpliesthat the new offeredloadis 0.99erlang.From Equation1?.30,the averagewaitingtime becomes

-

0.99x 0.031

f =-:=

2(t-o.ee)

1 . 5 3s e C

Thuswhentheofferedloadincreases by only l0%, theaveragedelaythroughthenode + increases ninefoldto a valueof 1.53 0.03I = I .56sec! for heavilyutilizeddelaysysthesamecharacteristic ExampleI 2.16demonstmtes performance is very sensitiveto inThe for losssy$temsr temsthatwasdemonstrated flow controlis a critical in Chapter10, creases in traffic inten$ity.Thus,asdiscussed particularly packet-switching there real-timedelivery operation, when are aspectof a objectives. 12.4.3 Finlte Queues presented sofar haveassumed thatanarbitrarilylarge All of thedelaysystemanalyses this ascouldbe placedin a queue.In manyapplications numberof delayedrequest$ havesignificantlylimited sumptionis invalid. Examplesof systemsthat sometimes queue$izesare store-and-forward swirchingnodes(e.g.,packetswitchesand ATM switches),automaticcall distributors,andvarioustypesof computerinput/outPutdevices.Thesesystemstreatarrivalsin threedifferentways,dependingon the number "in the $vstem"at the time of an arrival:

562

TRAFFIo ANALYSI$

I . Immediate service if one or more of N serversare idle 2. Delayed service if all serversare busy and less than 1-requestsare waiting 3. Blocked or no service if the queueof length Z is full In finite-queue systemsthe arrivals getting blocked are those that would otherwise experiencelong delays in a pure delay system. Thus an indication of the blocking probability of a combined delay and loss sy$tem can be determined from the prob-_ ability that arrivals in pure delay systemsexperiencedelays in excessof some specified value. However, there are two basic inaccuraciesin such an analysis.First, the effect of blocked or lost calls clearedis to reduce congestionfor a period of time and thereby to reduce the delay probabilities for subsequentanivals. Second,delay times do not necessarily indicate how many calls are "in the system." Normally, queue lengths and blocking probabilities are determinedin terms of the number of waiting requests,not the amount of work or total service time representedby the requesd. with constantservicetimes, there is no ambiguity betweenthe size of a queueand its implied delay. with exponential service times, however, a given size can represent a wide range of delay times. A packet-switching node is an example of a system in which the queue length is mo$t appropriatelydeterminedby implied servicetime and not by the number of pending requests.That is, the maximum queue length may be determined by the amount of store-and-forwardmemory availablefar variablelength messagesand not by some fixed number of messages. For a system with random input, exponential service times, N servers.an infinite source,and a maximum queuelength of z (wMlNl*lL), the probability ofj calls in the svstemis

PrA)=hni 0
INIri-

(12.30)

N
where 4 = offered load (erlangs), = f/* N = number of servers I = maximum number in the queue Here, Pq(A) is chosento make the sum of all p,(A) = l;

"',o)=[=i#.,-1,,---tl

=[-i#.#]#J

(12.30a)


Thetime congestion, or probabilityof an arrivalbeingdelayedor blocked,canbe foomEquation12.30as determined N+L

P(>o)=EPIA)=Pru(A)++ l-P

(r2.3r)

j=w

wherep =A/N is the offeredloadper server. Theloss,or blockingprobability,is determined as P^(A)AN*L "'

B=P-,(A)=

' N!N"

(r2.32)

It is worthnotingthatif thereis no queue(I = 0), theseequationsreduceto thoseof theErlanglossequation(12.8).If t is infinite,Equation12.31reducesto Erlang'sdelay formula,Equation12.24.Thustheseequations represent a generalformulationthat producesthepurelossandpuredelayformulasasspecialcases. The waitingtime distribution[3] is GI

-n,I

t

p(>r):PN(A) I \ ) xte-*dx i+"'u'/'*

(12.33)

from which the average delay can be determined as lp(>O) -Pff+r(A)lt*

(r2.34)

N_A

to Equation12.26for aninfinite{ueue(L = m). Again,Equation12.34is identical Single-ServerEquations involvesingle-server theprevious mostqueuing applications configurations, Because equationsare listed explicitly for N = l: , Prob(7 calls in system)(12.30):

ntG):rn(P)d

P.$) =[,

' p+p*"I'=[E,'l'

(r2.3s)

(12"36)

564

TRAFFIG ANALysts

Probabilityof delay(12.31): p(>0):

Po(p)p(l- p*t)

p(l

I-p

(r2.37)

Probabilityof loss(12.32): o_{t-p)pal

u----=-:-

l-p*"

(12.38)

Averagedelay(12.34):

;

tp(>o)- Pr*,(p)Jr* p(l - pz)rl-p ( t _ p X l_ p L r z )

(12.3e)

The blockingprobabilityof a single-seryer system(N = 1) is plottedin Figure 12.19.when usingFigure12.19,keepin mindthattheblockingprobability(Equation 12.38)is determined by thenumberof waitingcallsandnot by theassociated service time' Furthermore, $incethecurvesof Figure12.19arebasedon exponentialservice times,they overestimate the blockingprobabilitiesof constantholdingtime system$ (e.g.,fixed-lengthpacketnetworks).However,if fixed-lengthpacketsariseprimarily from longer,exponentiallydishibutedmessage$, the arrivalsare no longer independent,andtheuseof Figure12.19(or Equationt 2.38)asa conservative analvsisis moreappropriate. ATM Cell Queues Analysisof queuingdelaysandcell lossin an ATM switchingnodeis complicated. Thecellshavea fixed lengthof 53 bytessoit wouldseemthata constantservicetime analysiswouldbe appropriate. This assumption is valid for voicetraffic insertedonto wide-bandwidthsignalssuchas 155-MbpssTS-ls. In this caserhe servicetime is muchshorterthatthedurationof a speechburst(e.g.,2.7 psecversusseveraltensof milliseconds). Eventhoughcorrelatedarrivalsoccurfrom individualsources. the arrival times are separatedby many thousandsof servicetimes so they appearindependent. WhenATM voiceis carriedin CBR trunk groups,a differentsituationresults.In this casetheservicetimesof thevoicecellsmaybe only slightlysmallerthantheintervalbetweenvoicecell generation, andtheaveragedelaywouldindicatethattwo or morecellsfrom the samesourcecould be presentin the queueat onetime. Thus,a queuinganalysisthatassumes exponentiallydistributedservicetimesis moreappropriateeventhoughthevariable-length talk spurtsarebrokenup into fixed{engttrciffs. Example12.17. A 64-kbpscBR virtual path in an ATM networkis to carry lg voicesignalsthatarecompressed to 7.25kbpsduringvoicespurts.Assuminga speech activityfactorof 407o,determineeachof thefollowing;

I2.4 DELAYSYSTEMS 565

1.000 f, I 2 3 5

tt o' 1 0 0 -o (s

10

-o

e s. (D

.g .s IJ

g -0.

010

0.001 Offeredbaffic(erlangs) Figure 12.19 Blocking probabilityof singleserverloss/delaysystem(exponentialservice times).

(a) Outputchannelloading (b) Servicetime (c) Probabilityof delaywith an infinite queue (d) Probabilityof delaywith a queueof length20 ATM cells (e) Averagetime in queuefor exponentialservicetimeswith aninfinite queue (f; Averagetime in queuefor exponentialservicetimeswith a finite queueof 20 (g) Averagetime in queuefor constantservicetimes(infinitequeue) (h) Probabilityof cell loss(assuming exponentialservicetimes)

566

THAFFIC ANALYSIS

Solution (a)An ATM cell consistsof48 bytesofpayloadand5 bytesofoverhead.Thus,rhe offeredloadto thechannel* (18 x 7.25kbpsx 0.4) (53/4g)/64kbps= 0.9 erlangs. (b) The servicetime of a cell is 53 x g/64kbps= 6.625msec. (c) Theprobabilityof delayfrom Equation12.24is 0.9. (d) Theprobabilityof delayfrom Equarion12.37is 0.g9. (e) The averagequeuingdelayfrom Equation12.26is 59.6msec. (0 Theaveragequeuingdelayfrom Equation12.34is 5g.2msec. (S)The averagequeuingdelayfrom Equationl?.29 is 29.gmsec. (h) Theprobabilityof cell lossfrom Equation12.39is 0.012. Theresultsof Example12.17illustrareseveralimporrflntpoints.First,(c) and(d) indicatethattheprobabilityof delayin a finite-queuesystemis smallerthanthatof an infinite-queuesystem-becau$esometraffic is rejected.with a reasonablysized queuetheeffecton the delayprobabilityis small.similarly, a comparison (e) of and (f) illustratesthatthe averagequeuingdelayin a reasonably sizedqueueis not much differentthanthatof an infinite queue.As discussed earlier,thereis a significantdifferencebetweentheaveragedelayof a systemwith exponentiallydistributedservice timesasopposedto constantservicetimes.Beforeassumingthat a consrantservice time analysisshouldbe used,the averagedelaymustbe comparedto the delaybe_ tweenardvalsof voicecells.The durationof a speechsegmentcarriedin the ATM cellsof Example12.17canbederermined as4g xgl72s0= 53 msec.Theaveragedelay of a constantservicetimeanalysis(29.8msec)indicatesthatcellsfrom a particular sourceareusuallyservicedbeforea subsequent arrivalfrom the$amesource.but certainly not always.If Equation12.34is usedto determinetheaveragedelay,the5g.2_ msecresultindicatesthat, on average,an arriving cell from a sourceencounters a previouscell from the same$ource.Thusthe assumptionof independent arrivalsis marginal.To be safe,a cell losscalculationassumingexponentialservicetimesaccountsforcorrelatedarrivals.The cell lossprobabilityof l.zvo is marginalfrom a voicequalitypointof view.Noticethatthisvalueof cell lossrelatesto theCBR gateway (AALI adapration layer),whichis presumably theonly significantsourceof cell loss. The solutionto exampleproblem12.17necessarily useda simplemodelfor the traffic andthequeue.A thoroughanalysisof anATM or packet-switching networkis muchmorecomplicated. Factorsthat mustbe considered arethe switchingnodear_ chitectures andqueueimprementation (e.g.,sharedqueuesversusdedicatedqueues;, serverdisciplines(e.g.,priorities),cell discardalgorithms,call admissionconholal_ gorithms,andtraffic statistics.Someof theseissuesarecoveredin references [16-lg]. 12.4.4 Tandem eueues All ofthe equations providedin previoussections for delaysystemanalysishavedealt with theperformance of a singlequeue.In manyapplications a servicerequestundergoesseveralstagesofprocessing, eachoneofwhich involvesqueuing.Thusit is often

REFEHENcES 5S7 Locrl rrrlvrlr

Arrinh lrom pr|Yiodt qu6lC

Output quctre

queues. Figure12.20 Tandem of a systemwith a numberof queuesin series. desirableto analyzethe performance requests Figure 12.20depictsa seriesof queuesthatreceive,asinputs,locallygenerated and outputsfrom other queues.Two principalexamplesof applicationswith tandem switchingnetworks. systemsandstorc-and-forward queue$aredataprocessing in derivingforin queuingtheoryhavenot beengenerallysuccessful Researchers mulasfor the performanceof tandemqueues.Often, simulationiS usedtOanalyzea queuesarisingin systemslike store-and-forof interdependent complexarrangement that specialaspectsof a network'sopward networhs.Simulationhasthe advantage eration-like routingandflow control-can beincludedin thesimulationmodel.The of simulationareexpen$eand,often,lessvisibility into the demain disadvantages on variousdesignparameters' pendence of systemperforrnance one tandemqueuingproblemthat hasbeensolved[19] is for randominputsand holdingtimesfor all queues.The solutionof this sysrandom(negativeexponential) temis basedon thefollowingtheorem:In a delaysystemwith purelyrandomarrivals andnegativeexponentialholdingtimes,the instantsat whichcallsterminateis alsoa negativeexponentialdistribution. The significanceof this theoremis that outputsfrom an IvI/IvIA.{systemhavestatisticalpropertiesthat areidenticalto its inputs.Thusa queuingprocessin onestage doesnot affect the arrival processin a subsequentStage,and all gueuescan be anaSpecifically,if a delay$ystemwith N servershasexponentially lyzedindependently. distributedinterarrival times with average1/1,,and if the averageservicetime is f., callsleaveeachofthe serversaccordingto exponentiallydistributedintercompletion timeswith averagel/LN. analysisof tandemqueuescanbe rigorouslyjustified only Althoughindependent is often assumedin other for purely randomarrivalsand servicetimes,independence the in questionshouldbeexhowever, systems assumptions, using Before such cases. queue the operationof influence of can if the one state to closely determine amined queue in the system. another

REFERENCES Wiley'New to ProbabilityTheoryand ltsApplications, I W. Feller,An Introduction York.1968. 2

A. A. Collins and R. D. Pederson,Telecommunicatians-a Timefor Innovation,Metle Collins Foundation, Dallas, TX, 1973.

568

rRAFFtcANALysts

3 R. syski, Introductionto congesfionTheoryin Telephone,sysfems, oliver andBoyd, London.1965. 4 TechnicalStaff,Bell TelephoneLaboratories, Engineeringand Operationsinthe Bell System, WestemElectric,Indianapolis,lg7?. 5 'Determination of the Number of Circuits in Automatic and Semi-Automatic operation,"ccITT Recommendarion E. 520,orangeBook,vor. II.2, r976,p.zrr. 6 c. J. weinstein, "Fractional speech Loss and ralker Activity Model for packet switchedspeech,"IEEE Transactions on communications Technology. Aug. I 97g,pp. t253-1256. 7 H. Miedema and M. schachtman,"TASI eualiry-Effect of speech Detection and Interpolation,"Bell SystemTechnical Joumal,July 1962,pp. l4i5_1473. I G. Szarvasand H. Suyderhoud,'-Voice-Activated-Swirch PerformanceCriteria," ComsatTechnicalReview,SpringI g80,pp. I Sl *177. 9 S. S. Katz, "ImprovedNetworkAdministationhocessUtilizing End-to-EndService Considerations," IntemationalTeletrafficConference,lg7g, 10 R. Dayem,"AlternateRoutingin High Blockingcommunications Networks.,, National Telecommunications Conference, lg7g,pp. 2g.4.I _2g.4.6, l l P. R. BoorstynandH. Frank,"Large*scaleNetworkTopologicaloptimization,"rEEE Transactions on Communicaflrrzs, Jan.lg77, pp, Zg_47. 12 R. I. wilkinson, 'Theories for Ton rraffic Engineeringin u.s.A.,- Beil system TechnicalJournal,Mar. 1956. 13 "calculation of theNumberof circuits in a Groupcarrying overflow Traff,rc,"ccITT Recommendation E.52r, orangeBoak,vol. 2, No. 2, Geneva,switzerrand,p. 2rg. 14 J. A. Morrison,"Analysisof some overflow hoblems with eueueing,,'Bell system Technical Joumal,Oct.1980,pp. 1427-1462. 15 TelephoneTragic Theory,Tables,and charts, siemensAlctiengesellschaft, Munich, t970. 16 Y.-s. Yeh' M. G. Hruchyj,and A. s. Acampora,"The Knockoutswitch: A simple, Modular Architecture for High-perfonnancepacket switching,,, IEEE Journal on SelectedAreasin Communications, Oct.l9g7, pp. ll74_llg3. 17 K. sriram,T. G. Lyons,andr.-T. wang, "AnomariesDueto DelayandLossin AAL2 Packetvoice systems:Performance ModelsandMethodsof Mitigati on,,'IEEEroumal on Selected Areasin Cornmunitations, Jan,I g99,W, 4_17. 18 K. Sriram and y. T. yang, ..Voice over ATM Using AALZ and Bit Dropping: Perfbrmanceand call Admission control," IEEE Joumal on selected Areas in Communications, Jan.1999,pp. 18-28. 19 L. Kleiruock and R. Gail, eueueing sysferns,problems and solutions,wiley, New York. 1996,

PROBLEMS l2.l

A central-office-to-PBX funk groupcontains fourcircuits.Ifthe average call duration is 3 minandthebusy-hour offeredhafficintensity is 2 erlangs, deter_ mine each of the followine:

PHOBLEMS 569

(a) Busy-hourcallingrate (b) Probabilitythattwo arrivalsoccurlessthan I secapart (c) Blockingprobabilityassuminga lost callsclearedoperation (d) Amountof losttraffic (e) Proportionof time the fourthcircuit is in use(assumingfixed-orderselection) to a centraloffice, 12.2 A Tl line is usedto cany traffic from a remoteconcentrator

12.3 12.4

12.5

12.6

12.7

systemsuppofrat0.SVo canthe concentrator How many l0 CCS subscribers finite to sourceanalysis'Asanalysis a blocking.Comparean infinite source sumeblockedcallscleared. 20 erlangsof averagebusy-hourtraffic load Two switchingofficesexperience betweenthem.Assumea singleTI line provides24 directtrunksbetweenthe offices.How muchbusy-hourtraffic over{lowsto a tandemswitch? A PBX with 200 stationshasfive trunksto the public network'What is the blockingprobabilityif eachstationis involvedin threeexternalcallsper 8-hr workingday with an averagedurationof 2 min percall?Assumethe average callingrateis constantduringthe day (no busyhour)andblockedcallsreturn with randomretries.Whatis tlreofferedload?Whatis thedemandtraffic? How manydial-upinput-output(VO) portsareneededfor a computercenter to support40 userswith a blockingprobabilitylimit of 57o?Assumeeachu$er durationof 30 min. If three four callsperdaywith anaverage$ession averages grade of servicefor theremaining all day,whatis the u$srsremainconnected 37 users? A 24-channel trunk groupis dividedinto two Sroupsof 12 one-waytrunksin eachdirection.(A one-waytrunk is onethat canonly be seizedat one end.) How manyerlangsof traffic canthis systemsupportat O-SVIblocking?How manyerlangscanbe supportedif all 24 trunksaretwo-waytrunks?(Thatis, everytrunk canbe seizedat eitherend.) The following 10 A.M. to 11 A.M. busy-hourErlang(E) statisticshavebeen observedon a 3z-channelinterofficetrunk group.what is theoverallblocking probability?Whatis theblockingprobabilityfor thesamebusyhourif day-totogether? dayflucfuationsareaveraged Monday Tuesday Wednesday Thursday Friday 30E 19E 22E' 19E' 20E

on a PBX-to-central-officetrunk groupindicatethatdur12.8 Traffic measurements ing the busiesthour of the day the trunk$are807outilized.If thereareeight trunksin the group,what is the blockingprobability,assumingblockedcalls do not retum?How manytrunksmustbe addedto achievea maximumblocking probabilityof SVo? tz.9 RepeatProblem12.8assumingblockedcallsreturnwith randomretries. 12,10A smallcommunitywith 400 subscribersis to be servicedwith a community originates0.1 erlangof dial office switch.Assumethatthe averagesubscriber traffic. Also assumethat20Voof the originationsare local (intracommunity)

570

TRAFFIC ANALYSIS

callsandthat 807oaretransitcallsto tlreservingcentraloffice.How manyerlangsof trafficareofferedto thecommunity-dial-office-to-central-office g1nk group?How manyfrunksareneededfor 0.5voblockingof thehansittraffic? 12'11 For the comrnunityof Problem12.10determinethe numberof concentrator channelsrequirediflocal callsarenot switchedlocallybut aremerelyconcentratedinto pair-gain$ystems andswitchedat thecentraloffice. 12.12 Repeathoblems 12.10and 12.l I if g0%of rheoriginationsareinrracommu_ nity calls andZOVa aretransitcalls. u.13 A groupof eightremotefarm housesareservicedby four lines.If eachof the eightfamiliesutilizestheirtelephones for I0zo of thebusyhour,comparethe blockingprobabilitiesof thefollowingconfigurations: (a) Fourpartylineswith two srationsper line (b) An 8-to-4concentration sy$tem 12.14 A PBX providesqueuingandautomaticcalrbackfor access to outgoingwATS lines.If therearez0 requestsper hourfor thewATs lines,andif theaverage call is 3 min in length,how manyWATS linesareneededto providederaysof lessthan I hr for 90Zoof therequests? f2.15 A call processor has507oof its timeavailablefor servicingrequests. If eachrequestrequires50 msecof processing time, whatarrivalratecanbe supported if only I 7oof theservicerequests aredelayedby morethanI sec?Assumethat processor time is slicedinto iOO-msec time slots.(Thatis, 500 msecareallo_ catedto call processing andthen500 msecto overheadfunctions..l 12'16 A groupof 100sourcesoffersmessages with exponentiallydiskibutedlengths to a 1200-bps line.Theaveragemessage lengthis 200bits,includingoverhead, andeachsourcegenerates onemesrlage every20 sec.Accessto theline is con_ trolledby message-switching concentration with an infinite queue.Determine thefollowing: (a) Probabilityof enteringrhequeue (b) Averagequeuingdelayfor all arrivals (c) Probabilityof beingin thequeuefor morethan I sec (d) Utilizationof thetransmission link 12.17 An airlinecompanyusesan automaticcall distributorto servicere$ervations andticketpurchases. Assumethatthe processing time of eachinquiry is randomtydistributedwith a 40-secaverage. Also assume thatif customers arepur on hold for more than2 min, they hangup and cail anotherairrine.If eachof 200inquiriesperhourproducesg30worthof sales,on average, whatis theop_ timum numberof reservationists? Assumeeachreservationist coststhe company$20/hr(includingoverhead). 12'18 A radio stationtalk showsolicitsthelisteningpublicfor cornmentson the ineptne$$ of government (I assume thiswill bea topicalsubjectfor thelife of this book).Assumethateachcallertalksfor a randomlengthof time with an averagedurationof I min. (Eithertheshowis unpopularor thepublichasgivenup

PRoBLEMS 571 on the govemment.) How many incoming lines must the radio station have to keep the idle time below 5% if the call arrival rate is 3 calls/min? 12.19 RepeatExample I 2' 17 for a queuelength L = 40' (Although a rigorous solution requires calculation of a ZQ-termsummation, only fhe first few terms are significant.)

A APPENDX

OF EQUATIONS DERIVATION 3.2 NOISEPOWER:EOUATION A.1 QUANTIZING to beuniform; is assumed densityfunctionof a noisev sampled Theprobability

p(n)=iq lO

otherwise

The averageor expectedvalueof noisePoweris determinedas

noir*po*"rJf euantization

* [n1jrt

=[#)n 4.1 A.2 NRz LINEGODE:EQUATION ,tt) =

[t {O

t rst + 7 other'wise

rUo))=j 71t1e-i'ntdt

574

AppENDtxA

=[i'Jt"'-"' - e-i$T/z)

=671'in(a44) (aT/2) Note: FQw)is thespecfumof a singlepurse;rF(7co)z gnl isthepowerspectraldensity of a randompursetrain assumingpositiveandnegativepulses areequaltylikery andoccurindependently.

4.3

DlclTAL BIpHASE: FtcURE 4.19

^r=l:, -lT
o< t < l r

otherwise

(t/z)r

0 |

F(7'to)= !

"-*dr-

_(t/?\T

=[*)"

....

I

J **i*dt 0

_ rl( /2laT_ s-iTrz)ar * 11

=[.,,')'*'[T) A.4 FBAME ACQUISITIONTIME OF SINGLE.BITFRAME CODE: EOUATION4.10 Framingis established by successively examiningonebit positionafteranotheruntil a sufficiently long framingpatternis detected.In this derivationit is assumed that the framingpatternalternates I's and0's. Furthermore, it is assumed thatwhenbeginning to testa particularbit positionfor ftaming,the valueof the first appearance is saved andcomparedto the secondappearance. Thusthe minimum time aorejectan invalid framingpositionis oneframetime. If we denotebyp the probability of a I andby s I r _ p the probability that a 0 is receivedfirst, the averagenumberof ftamesrequiredto receivea mismatchis

4,10 CODE:EQUATION FRAME ACQUISITION TIMEOFSINGLE"BIT FRAME

575

Ao = (l) (probability of mismatch at end of first frame) + (2) (probability of mismatch at end of second frame) + (3) (probability of mismatch at end of third frame) . . .

= (1)4+ (zxl * q)p+ (3Xl - q)(r- p)q + ( a X -l i l 2 0 - i l p + . . . = (1)4+ Q)pz+ (3)pqz+ {4)p3q + (5)p'q3+ . . =(q+2p?+ps\[I +(Z)ps+(3)pzqz + (4)p3q3 +. . .] = [7 - p + Zpz+ p(l-p)1(1 + pq + pzqz+ psqt+ . . .)'

= l+p3 tr*P 1+p ='=-

r_pq

Similarly,if a I is receivedftrst, the expectednumberof framesbeforereceivinga mismatchis l+a At=T_fi The overall averagenumberof framesrequiredto detecta mismatchis A = q A o *p A 1 _I +Zpg 1- pq If we assumea randomstartingpoint in a framewith N bits, the averagenumberof bits that mustbe testedbeforethe true framing bit is encounteredis Frametime = (l /Ztl)(A. M + 1/2N = 1 / z N ( A . N + 1 )b i t t i m e s

576

AppENDtx A

If I's and0'sareequallylikely (p =q=l),A = 2 soframetime=Ap + t/2Nbittimes (Equation 4.10).

A.5 FRAMEACQUISITIONTIME OF SINGLE.BITFRAN,IING CODE USINGA PARALLEL$EARGH:EQUAT|ON4.11 This framingalgorithmas$umes that all bit positionsin a frameare simultaneously scanned for theframingpattern.If weassume thatanalternating-bit framecodeis used andthat I 's and0's in theinformationbitsareequallylikely, th-eprobabiliryrhata par_ ticularinformationbit doesnot producea framingviolationin n framesis

/rY o^=lil The probabilitythat a framing violatidn iuJ u**n receivedin n or lessframes is I -p,' The probabilitythat all N - I informationbits in a frameproducea framing violationin rzor lessframesis

o.=f,-fri]-'

(4.lr)

A.6 FRAMEACQUISITION TIMEOF MULTIBIT FRAMEGODE: EQUATION 4.13 N= Iength of frame including framing code ,L = length of framing code p= G)L: probability of matching frame code with random data The expected number of frames examined before a particular frame position mismatchesthe frame code follows:

A = (OXl -p) + (l)p(t - p) + (Dpz(l - P ) + " ' : (l - p)p(I + Zp -t 3p?+ 4p3+ . . . ) = (1 -p)p(l +p + pz +p3 + . . .)?

= ( r - p )l/p+' r \ |2 ['-pJ _

P l-p

s.11 A.7 PATHFINDING TIME:EOUATION

577

The averagenumberof bits that passbeforethe frameposition is detected(assuming a randomstartingpoint andthetestfield is movedonebit positionwhena mismatch occurs)is givenas

r=ftruJraxnn*it r.d/otzt'*t. t ,, =T:769-1''

(4.13)

4.13 Equation 4.l0 because toEquation Equation 4.13withZ = I isnotidentical Iy'afe.. assumes an alternating code. 4.10 a fixedframecodewhileEquation assumes

5.11 TIME:EQUATION A.7 PATHFINDING Assume that all paths through a switch are independently busy with probability p. Let the probability that a path is not busy be denotedby q= | -p. The probabilityp; that exactly i paths are tested before an idle one is found is the probability that the first i - | arebusv and the ith is notr

P'= P(i-t)n The expected number of paths tested before an idle path is found is

+ (k)pk ruo* (1)a+ (Z)pq+ (3)pzq+. . . + (k)pk*Lq

wherethelasttermrepresentstheexpectationthat all possiblepathsk areunavailable. as A closedform for A is determined A t = ( 1 * p X l + 2 p + 3 p 2+ . . . + k p k -+rk p k 1 = I + p + p z+ p j + , . . + p k - ' I

t-p ' =l: Po l-p

k( r \

[t

-oJ (s.ll)

578

APpENDtx A

Figure A.l.

probabiliqrgraphof No, 4 ESSfour_stage spaceswitch.

A,8 LESS SPACE STAGE BLOCKING PROBABILITY5.21 The moststraightforward way to calculatethe blockingprobabilityof thefour-stage spaceshuctureshownin theprobabilitygraphof FigureA.I is to list all elementary, mutuallyexclusiveprobabilityterms,determinetheirprobabilitiesof occurrence, and generate thesumof thosethatrepresent blocking.Because thereareeightlinks,which are.assumed to be independently busyor idle, therearezs = 256elementaryterms. Ratherthanlaboriouslylist themall, theanalysiscanbegreatlysimplified,with a risk oJmiscountingblockingterms,by groupingthetermsaccordingto thenumberof busy links-Thefollowingtableliststhegroupsandthecorrespondinfnumbersof termsthat blockanddo not block. Numberof BusyLinksI 0 1

2 3 4 5 6 7 I

Combinations (:)

NumberThat Block

NumberthatDo Not Block

1 I 28 56 70 56 28 B 1

0 0 2 16 50 52 2B I 1

1 I 26 40 20 4 0 0 0

256

157

99

Entriesin thelasttwo columnsaredeterminedby analyzingthetopologyof thenet_ work. For example,whentwo links arebusy,onry2 of the2g combinations produce blocking( I and2 and7 and8).when threelinksarebusy,I 6 of thecombinations produceblocking.To determinethe remainingenhiesin column3, it is easierto determine the numberof combinationsthat do not block and subtractfrom the total. For example,whenonly threelinks areidle, thereare4 of 56 combinationsthat do not block.using theentriesin column3, theblockingprobabilityis determinedas B = Zpzq6+ l6p3qs + S}paqa+ Sbps qt t Zgp6 qz+ gp7q + pg wherep is theprobabilirythata link is busyandq = | _ p is theprobabilitythatit is idle' All ofthe interstage links areequallyloadedbecause thereis no concentration or expansionin the stages.

B APPENDX

ENCODING/DECODING FOR ALGORITHMS PCM SEGMENTED 8.1 EIGHT-BITp255 CODE format of p255 PCM codewordsusea sign-magnitude The encodedrepresentations whereinI bit identifiesthe samplepolarityandtheremainingbits specifythemagnitudeof thesample.The7 magnitudebitsareconvenientlypartitionedinto a 3-bit segment identifier S and a 4-bit quantizingstepidentifier Q. Thus thebasic structureof an 8-bit p255PCM codewordis shownin FigureB.l. of encodinganddecodingalgorithms,it is assumed, In thefollowing descriptions that analoginput signalsare scaled for conveniencein using integerrepresentations, all amplitudesandsegmentidentifiers to amaximumamplitude of 8159.Furthermore, binaryrepresentations. The actualento be encodedusingconventional areassumed systems,however,complementthe codewordsto incodersusedin T1 transmissiOn creasethedensityof I's in a transmittedbit stream. 8.1.1 Algorithm 1: Direct Encodlng (Table 8.1) polar.tty Dttr = [o for uositivesamplevalues ll for negativesampleroalues Given a samplevaluewith a magnitude.x,the first stepin the magnitudeencoding areidentifiedby processis to determinethesegmentidentifierS.Themajorsegments and8159'Thus'Scanbe endpoints: 31, 95, 223,479,991,2015,4063, the segment I v\-rJ\-1/--/

P Figure B.l.

s

I

I

Q

Eight-bitp225PCM codeformat.

580

APFENDIXB

TABLE 8.1

Piecewlee Linear Approxlmation to FZSSCompoundlngd QuantizationEndpointsby SegmentCode S

000 0 1

3 5 7 I 11 13 15 17 19 21 23 25 27 29 31

001

010

011

100

101

31 35 39 43 47 51 55 59 63 67 71 75 79 83 87 91 95

95 103 111 119 127 135 143 151 159 167 175 183 191 199 2Q7 215 223

223 239 255 271 287 303 319 335 351 367 383 399 415 431 447 463 479

479 5 11 543

991 1055 1119 11 8 3 1347 1 3 11 1375 1439 1503 1567 1631 1695 1759 1823 1887 1951 2015

s7s 607 639i 671' 703 735 767 799 831 863 89s 927 959 991

110

111

Quantization Code Q

0 1 2 3 4 E

6 7 I I 10 11 12 13 14 15

asamplevaluesarerelerenced to a yaly6oj 8159.Nogative samplesareencodedin sign-magnitud6 Iujl--sc€le formetwitha polarltybitof 1. In actuel transmission thecod6salreinvertedto in"reaietn" o"nsrtyof 1,swhen low signalamplitude$ are €ncoded.Anelogoutputsamplesere docodedas th6 centerof the encod€d quantizationinterval. Quantizationerror is i'he difference between the teconstructooout[ut vatu6 ancl th. origir-lalinput sample value. ,!

determined asthe smallestendpointthatis greaterthanthe samplevaluex. Here,s is equalto the smallesta suchthat x<64-T-33

a=0,I,...,7

After themajorsegmentcontainingthesamplevaluehasbeendetermined, theparticularquantizationintervalwithin the major segmentmustbe identified.As a first stepa residuer is determinedasthe differencebetweenthe input amplitudeandthe lowerendpointof the segment:

S=0 '=fi- (32.zs33) S = I , 2 , , . . . 7 The valueof Q cannow be determined asthequantizationintervalcontainingthe residuer. Here,fl is equalto the smallestb suchthat

lzu*t r<[1zs+rxb+ l)

,5=0 S=1,2,...,"1

whereb = 0, l, . . . , 15.Noticethat this processidentifiesquantizationintervalsin segment^s= 0 ashavingupperendpoint$at l, 3, 5, . . . , 3t while theothersegments

8.1 EIGHT-EITp255CODE581

havequantizationendpointsfhataremultiplesof 4, 8, 16,32,64, 128,256 for S = 1, 2, 3, 4, 5, 6, 7, respectively. After,SandO havebeendetermined,they arebinary encodedinto 3 and4 bits, reof S andO producesa 7-bit word thatcanbe convenspectively.The concatenation this integeridentifies integer represented as an between0 and 127.In essence, iently amplitude. compressed signal quantization intervalsof a oneof the 128 polarityto an analogoutThedecodingproces$involvesassigningthedesignated put sampleat the midpointof the nth quantizationint€rvaln = 0, l, ' ' . , 127'Using thevaluesof S andQ directly,we candeterminea discreteoutputsamplevalueas yn=(2Q+ 33X2r)- 33 of ,5and wheren is the integerobtainedby concatenatingthe binary representations

a.

Example8.1. thefollowingcodeword: of +242produces An inputsample

0 , 3 ,1 = f f i The decoderoutputbecomes t$=(2. 1+33X23)-33 :247

quantization intervalfrom?39to 255. whichis themidpointof theforty-ninth A.1.2 Algorithm2: LinearCodeConversion with p255 is The fundamentalreasonfor usinga p-law compoundingcharacteristic to and can converted be digitally approximation the with which segmented the ease the implement algorithms that the section describes basic from a uniform code.This enimplementing PCM 4p255 themeansof Thefirst algorithmprovides conversions. coderusinga 13-bituniformencoderfollowedby digitallogic to providethecompressionfunction.Thesecondalgorithmindicateshow to implementthedecoderfunction codeinto a 13-bitlinearcodeto beusedin generating offirst expandinga compressed theouQutsamples. Justasin algorithm1, thepolaritybit P is determinedas for positivesamplevalues 'o -_ I0 11 for negativesamplevalues

s82

APPENDIX B

The simplicity of converting from a linear code to a compressedcode is most evident if the linear code is biasedby adding the value 33 ro the magnitudeof all samples.Notice that this bias shifts the encoding range from 0*gl5g ro 33-gl9z. The addition processcan be performed directly on the analog samplesbefare encoding or with digi_ tal logic after encoding. In either case,the generalform of all biased linear code patterns and the correspondingcompressedcodesare as follows: p255EncodingTable BiasedLinearInputCode 0 0 0 0 0 0 0 0 0 0 0 0 0 1 l w

0 0 0 0 0 0 0 0 0 1 1 w w x x y

0 0 0 l w x y z

0 0 l w x y z a

0 l w x y z a b

1 w x y z a b c

w x y x y z y z a z a b a h c b c d c d e d e f

z a a b b c c d d e e f f s g h

0 0 0 0

Compressed Code w x y w x y w x y w x y w x y w x y w x y w x y

0 0 0 1 1 O l 1 0 0 0 1 1 O l 1

z z z z z z z z

From the foregoing table it can be seenthat all biased linear codeshave a leading I that indicatesthe value of the segmentnumber ,s.specifically, rhe value of s is equal to 7 minus the number of leading 0's before the I. The value of is directly available e as the 4 bits (w, x, y, z) immediately following the leading 1. All trailing bits (a-h) are merely ignored when generatinga compressedcode. In reversefashion the following table indicateshow to generateabiased linearcode from a compressedcode. An unbiasedoutput can be obtainedby subtracting 33 from the biasedcode: p255DecodingTabte 0 0 0 0

0 0 1 1 0 0 1 l

Compressed Code 0 w x y z 1 w x y z 0 w x y z 1 w x y z 0 w x y z 1 w x y z 0 w x y z 1 w x y z

0 0 0 0 0 0 0 1

0 0 0 0 0 0 1 w

0 0 0 0 0 1 w x

BiasedLinearQutputCode 0 0 0 0 l w x 0 0 0 1 w x y 0 0 1 wx y z 0 1 w x y 2 1 1 w x y 2 1 0 w x y z 1 0 0 x y z 1 0 0 0 y z 1 0 0 0 0

y z 1 z 1 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Both of thesetablesindicatethat 13bitsof thelinearcodeareusedto represent the magnitudeof the signal.In chapter3 it is mentionedthat a p255 pcM coderhasan amplituderangeequivalentto l2 bits.Thediscrepancy occursbecause thefirst quantizationintervalhaslengthI while all othersin thefirst segmentareof lengthz. Thus theexhabit is neededonly to specifythefirst quantization interval.Noticefurrherthat theleastsignificantbit in thetablescarriesno informationbut is includedonly to facilitatetheintegerrelationships. kr particular,theleastsignificantbit in theencoding tableis completelyignoredwhendetermininga compressed code(assumingthatthe

B.r EIGHT BrrA-LAW coDe

583

biasis addedto the analogsample).Furthermore,the leastsignificantbit in the output codesis completelyspecifiedby thesegmentnumberS.It is a 0 for segmentzeroand a I for all othersegments. Example 8.2. An input codeword of +242is biasedto producea valueof 275.The binaryrepresentationof 275is 0000 I 000 I 00 1 1 (biasedlinearcode) codeis Hencefrom theencodingtableS = 011andwxlz = fi)01, andthecompressed (compressedcode)

ffi

codeproducesthefollowingbiasedlinUsingthedecodingtable,thiscompressed earoutputcode: 0000 1000 I 1000

(biasedlinearoutput)

The decimal repre$entation of the foregoing code is 280, which corresponds to an unbiasedoutput equal to +247.

8.2

EIGHT BIT A-LAW CODE

algorithmsfor segmentedAlawcodesusethe samebaThefollowingcompounding for thep255codes.Onedifferencethatdoesoccur, sic procedures a$thosepresented however,involvesthe eliminationof a biasin the linearcodefor conversionto and code.Anotherdifferenceoccursin the useof theinteger4096as from a compressed If desired,the scale the maximumamplitudeof a samplein anAlaw representation. by doublingtheAJaw factorsfor thetwo systemscanbebroughtinto closeagreement scaleto 8192. Alaw codeis usually referredto as a As mentionedin Chapter3, the segmented l3-segmentcodeowing to the exisknceof sevenpositiveand sevennegativesegmentswith the two segment$nearthe origin beingcolinear.In the following descriptions,however,the first segmentfor eachpolarity is dividedinto two pal'tsto produce This point of view permitsa codeformat eightpositiveandeightnegativesegments. codewordconsistsof a that is identicalto the p255 codeformat. Thus,a compressed signbit P followedby 3 bits of a segmentidentifierS and4 bits of quantizerlevel Q. 8.2.1 Algorlthm 1r Direct Encodlng The segmentendpointsof anA-law codeare32,64,128, 256, 5I2, l0Z.,2048, and 4096.Thus for a samplewith magnitudex the major segmentidentifier ,5can be determinedasthe smalle$ta suchthat

584

APPENDIX B

TABLE8.2 SegmentedA-[aw EncodlngTable Ouantization Endpointsby SegmentCode 001 0 2 4 6 B 10 12 14 16 18 20 22 24 26 28 30 32

32 34 36 38 40 42 44 46 48 50 52 5,4 56 58 60 62 84

010 64 68 72. 76 80 84 BB 92 96 100 104 108 112 116 120 1?4 128

011 128 136 144 152 160 168 176 184

19e 200 208 216 224 232 240 248 256

101 256 272 288 304 320 336 3s2 368 384 400 416 432 M8 464 480 496 512

110

111

512 544 576 608 640 672 704 736 768 800 832 864 896 928 960 992 1024

Quantization Code

0 1 2 3 4 5 6 7 B I 10 11 12 13 14 15

Not6i Every oth6r bit is invertedfor tran8mi8slon.

x<32'2"

a = 0 ,l , . . . , 7

After S hasbeendetermined,the residuer canbe obtainedas , =[* lx- 16.2s

S=0 S=1,2,...,7

The valueof B canthenbe determinedasthe smallestb suchthat

'.{#[..'i,

S=0 S=1,2,...,7

Justasin the casefor p255 coding,anAlaw magnitudecanbe represented asan integern= 0, l, .. ',121 derivedfromtheconcatenation of 3 s-bitsand4 p-bits.An outputmagnitudeI canthenbeexpressed as

S=0 [zB+r ,,=ltt,n+ 16f)

S=I,2,...,7

Exampla8.8. An inputsample of +121produces thefollowingcodeword:

8.2 EIGHTBITA.LAWCODE

585

=+46indecimal

ffi The decoderoutputbecomes

+ te |) )+e= 2s(14 _ | tr,,

whichis themidpointof thequantizationintervalfrom 120to 7M. 8.2.2 Algorithm 2: Linear Code Converslon Thefollowingtablesindicatehow to converta lZ-bit linearcodedirectlyinto a compressed.A-law code.The algorithmis basicallythe sameasfor the p255 conversion anda first segmentcodedoesnot exceptthatbiasingthe linearcodeis unnece$sary as7 minusthenumber havea leadingL ThusthesegmentnumberScanbedetermined of leadingzerosasbefore.The p field dataareobtainedasthe 4 bits (wxyz) immediatelyfollowingtheleadingl, exceptwhen,S= 0, in whichcasetheQ field is contained in the4 bits following thesevenleading0's. A.Law Encodlng Tabl€

Compressed Code

Linear Code

0 0 0 0 O 0 O l

0 0 0 0 0 O l w

0 0 0 0 0 l w x

0 0 0 0 l w x y

0 0 0 1 w x y z

0 0 1 w x y z a

0 w 1 w w x x y y z z a a b b c

x x y z a b c d

y y z a b c d e

z z a b c d e f

a a b c d e f g

0

0

0 w x y z 0 0 1 w x y z 0 1 O w x y z 0 1 1 w x y z 1 O 0 w x y z 1 0 1 w x y z l l 0 w x y z l l l w x y z

The following table providesthe meansof generatinga linear codeworddirectly from a compressed codeword.Theoutputvalueconespondsto themiddleof thequantization interval designatedby S and p. Table A-LawDecoding Compressed Qode 0 0 0 0 1 1 ' 1

0 0 1 1 0 0

0 1 O 1 0 1

LinearOutputCode

w x y z 0 0 0 0 0 0 0 w x y z 0 0 0 0 0 0 1 w x y z A 0 0 0 0 1 w w x y z 0 0 0 0 1 w x w x y z 0 0 0 1 w x y w x y z 0 0 1 w x y z f 1 O w x y z O 1 w x y z 1 1 1 w x y z 1 w x y z 1 0

w w x y z 1 0 0

x x y z 1 0 0 0

y z 1 y z 1 z l O 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0

586

AppENDtx B

Eachof thesetablesrelates12bits of magnitudein a linearcodeto a compressed codewith 7 bitsof magnitude. Notice,however,thattheleastsignificantbit of theencoderis alwaysignored.Thustheencoderneedsonly 1l bitsof resolutionif all of its outputsareimmediatelycompressed. If anysignalprocessing (suchasaddingtwo signalstogether)is to takeplacebeforecompre$$ion, however,theextrabit is usefulin reducingthecompositequantizationerror. Example The previously used sample value Lzr is representedin binary form as 000001I 11001.Fromtheencodingrable,S= 010ande = I I 10.Thusthecompressed codeword is

ffi using the decodingtable,the linearourputcanbe determinedas000001I l l0l0, whichis 122decimal.

APPENDX C

ANALYTIC FUNDAMENTALS OF DIGITALTRANSMISSION C.l

PULSESPECTHA

This sectionpresents thef'requency $pectra of commonpulsewaveformsusedfor digital transmission. Thesearesquarepulsesasgenerated at a source.Sincethe spectraof squarepulseshaveinfinite frequencycontent,the spectrapresented heredo not correspondto pulseresponses at theoutputofa channelwherethepulseshapes havebeen alteredby bandlimitingfilters.In thenextsectionchanneloutputpulseresponses are described.Thenthe necessary combinationsof input pulseshapesandfilter designs to produceparticularoutputpulsesareconsidered. Thevariouspulseshapes frequencyspectraarepresented andcorresponding in FigureC.1.In derivingthespectra,thefollowing conditionsandassumptions made: were 1. 2. 3. 4. 5.

All pulseshaveequalenergy. All systemssignalat rutfrIlT. Thewaveform$shownareusedto encodea 1. Oppositepolaritiesareusedfor a 0. It is equallylikely for l's and0's to occurandto occurat random(complete independence).

C.2 CHANNELOUTPUTPUL$E RESPONSES Althougha digitaltransmission systemcanbe designed to producea varietyof output pulseresponses, themostcommonis definedas v--\0 | '/rc\r/

sin(nr,zl) cos(ant/7) IEI/T

(c.l)

I-(Zat/T)z

where1/Tis the signalingrateandcr,i$ an excessbandwidthfactorbetween0 and 1. EquationC.l represents the response of a raised-cosine chamel-so-called because thefrequencyspectrumconesponding to yrc(t)in EquationC.l is

588

APPENDTx c

NRZ llnc code

-|r

*3

tr

:2

fltl * llrl{+ r

-l F (lr^rl -

Dlelul biphrre

-

,frt- r -tr
2

-

r

t

t

r r pl*S*n' r Sr

| + D Oorrahtivr ficodhg

-3

-2

-1

I

?

3

?

3

rqr.,r=ffipII

l"r;-r

_ t

| -Dcornhth'.

'ncodrns

I

l'

,

TE]_

t't:t:*'-r:1,**'o

-3

-2

-l

r

F(r.l - tf;t .rnrtrtt

| * pr Cotolrtiw Encodlng _l_ I

fld "; __ I ?

-F*'*-f

-3

-z

..r

r l*ot -t$rn t'n *n rfl

E*r*T

Figure C.1. Spectraof commonpulseshapes.

C.2 GHANNELOUTPUTFULSEHE$PONSES 589

F,"(f) = I

,' {o' -* I - o 2T

=il'.*'[#-r*r]]

1-o*rfl<1*o - '''2T 2T

:"",'[#q#) =Q

(c.2)

otherwise

Theoriginofthe apellation"raisedcosine"is apparent in thethird line ofEquation C,2, The parameteru in EquationsC.l andC.2 is referredto as an excess-bandwidth parameter. If o = 0, thespectrumdefinedin EquationC.2is exactlyequalto thetheoreticalrninimumbandwidthll2T for signalingrate llT. As fl increases from 0 to l, theexcessspectrumwidthincreases to 1(X)7o. Raised-cosine channelspecFaareillustratedin FigureC.2 for severalvaluesof cr. Practicalsystemsaretypicallydesigned for excessbandwidthsof 307oor morefor severalreasons.First, "brick wall" filters neededto producethe infinite attenuation slopesimpliedby fl = 0 arephysicallyunrealizable. Second,asshownin FigureC.3, the time-domainpulseresponse for smallvaluesof u,exhibitslargeamountsof ringing. Slight errorsin the sampletimescausesignificantdegradations in performance dueto intersymbolinterference. Third a slightdeviationin thesignalingratefrom the designratealsoproducessignificantintersymbolinterference. It mustbeemphasized thatEquationC.2definesthedesiredspectrumat theoutput ofthe channel(theinputto thedecisioncircuit).Thusthedesiredresponse resultsfrom a combinationof the input pulsespectrumandthe channelfilter responses. Often, the inputpulse$pectrumarisesfrom a squarepulseof durationT:

Y,"lfl

-1 lo=0

I

I

i

t I I

-l

t +

I I

0.

.5

** "-.-+

tr'igureC.2. Raised-cosine spectrums fbr variousvaluesof c.

590

APPENDTx c !*ltl .l tnputpura" Ouput puber

Normrlirrdtime I r/Tl Figure C.3.

Raised-cosinepulse responsesfor various values of s.

[t r,(r)= lo

td<+r otherwise

(c.3)

The frequencyspectf,umcorrespondingto x"(r) in Equationc.3 is the ,.sin(.r)/x" spectrumalsoreferredto asa "sine"function:

*"cn=I4ffi = f sinc(n/T)

(c.4)

when thechannelinput spectrrrm is asdefinedin Equationc.4, thefilter function of a channelto producea raised-cosine ouryutis determinedas

H(n=ffi

(c.5)

The channelfilter functionsimpliedin Equationc.5 areshownin Figurec.4 for thesamevaluesof fl, shownin Figurec.2. Figurec.2 displayschanneloutputspectra while FigureC.4displayscorresponding frequency-domain transferfunctionsfor the channel.

Normrlirrdf lugurrtyl.f TI figure C.4. Channelfilterresponses neededto produceraised-cosine outputswhenexcited by sin(4172)/(rn72) pulses.

C,2 CHANNELOUTPUTPULSEHESPoNSES 591

I

-'5 no-"Lr*u.rlv urr

FtgureC.5. "Optimum" ffansmit andreceive filterfunctions for raised-cosine response with s = 0,3andsin(ro72)/(coZl?) excitation. As mentionedin Chapter3, the designof a smoothingfilter in a digital voicedecodersometimes requiresa modificationlike thatdefinedin EquationC.5.In fhecase mentioned,the ideal "flat" filter for reconstructing speechfrom narrowimpulselike samplesshouldbe modifiedby dividing the flat re$ponse by the $pectrumof the finite-widthsamples(EquationC.4).Whenthesamplesarenarrowerthanthe signaling interval,the sin(x)/xresponseis essentiallyflat acrossthe bandwidthof the filter. However,whenthe samplesaremadeto last for the entiredurationof the sampleinterval,thefilter response shouldbe "peaked"to compensate for sin(x)/xroll-off. As discussed in Chapter6, a channelfilter functionis usuallypartitionedbetween the transmitterand the receiver.The so-called"optimum" filter partitioningoccurs when the receivefilter responseis the squareroot of the desiredoutputresponse Y*(fl, andthetransmitfilter responseis whateveris necessary to han$formthechannel inputspectrumintothecomplexconjugateof thereceivefilter response. (Theoutput'ryecrilhen the trum of thetransmitteris alsothesquareroot of thedesiredchannelresponse.) 'bptimum' pulse channelinputis a of duration7, the filter functionsaredefinedas H**(.f) = {I'"(,f)}r/2

(c.6)

HwU)=#

(c.7)

Noticethat the transmitspectrumresultingfrom HnE(f) in EquationC.7 is equal to thesquareroot of thedesiredoutputresponse, no matterwhattheinputpulseshape is. Hence,whenoptimumpartitioningis used,thetransmitspectrumandthereceiver designareindependent of thechannelexcitation. The transmitand receivefilter functionsfor square-wave excitations(Equations C.6 and C.7) arc shownin FigureC.5.* Noticethat the transmitfilter functionhas peakingat frequenciesotherthandirect current.The implication for midbandattenuationis oneof thereasonswhy "optimum"partitioningmaynot be optimumin a system with sourcepowerlimitations. *This

discussionassumesbasebandtransmission.The principles are easily extendedto carrier-based systemsby translstingthe filter functionsto the carrierfrequency.

592

AppENDtx c

C.2.1 OptlmumFilterlngfor MinimumShiftKeylng As discussed in chapter6, minimumshiftkeyed(MSK)modulation canberepresented as quadrature channel modulation with basebandexcitation defined as cosine pulse shapes:

.,,=f-[*J ttts+T

(c.8)

otherwise

Thetransformofx"(r) is v , A =[zrJ f r') cosn/r x.ul | -ffi

(c'e)

Whena raised-cosine ouQutresponseandoptimumfilter partitioningis desired, theMSK filter functionsare

H*"(f) =lYn(frlt/z

(c.10)

ro(f)=w

( c.ll)

Thetransmitspectrum andthereceiverof anoptimallypartitioned MSK systemare identical to counterparts in anoptimallypartitioned, offset-keyed 4-psKsystem! C.2.2 Partlal-Response Systems As anotherdigital hansmissionsystemdesignexample,considera I + D partial-responsesystem.Thedesiredtime response of thechannelis definedin EquationC.12 andthe associated frequencyspectnrmin EquationC.13.Thepulseresponse of a I + D partial-response channelis shownin FigureC.6.Noticethata singlepulsecontributes equallyto the responseat two successive sampletimesbut crosseszeroat all other sampletimes: 4 cos(nt/T\

Y,\I)= n|__eiffi fcosnlr Y.ffi=i

Io

nt+i otherwise

(c.12)

(c.13)

PULSEBESPoNSES 593 C.? cHANNELoUTPUT

rI

I I

1

-1 I F---tnprt I

putrr

puls6Jrrlrl

-2.5

-r.6

t.5 2.5 Noffirlirid tiffi6 {tr?}

3.S

Figure C.6. Outputpulseof I + D partial-response channel. Optimum filter pafiitioning is again achievedwith a receive filter having an amplitude responseequal to the squareroot ofthe desiredoutput respon$eY"(/). Ifthe channel is excited by squarepulsesof duration ?, as defined in Equation C.3, the optimum filter functions are

Hnx(fl=[Y"ff)]L/z

vls+

(c.r4)

H*(f) - Lv,(f)1"' X'fl

vts+

(c.r5)

where Xr(f) is defined in Equation C.4. The optimum filter functions defined in EquationsC.14 and C.l5 are shown in Figure C.7 along with the desiredoutput responseof the channel.Notice that, unlike the full-response(raised-cosine)systems,optimum partitioning of a partial-responsesystem does not require peaking of the transmit filter.

-t

o.5 1 Normelitrdf ruqurncyl.f Tl

Figure C.7. Spectra of 1 + D PRS channel and "optimum" filter responsesfor sin(aT| 2)| (aT | 2) excitation.

594

c.3

APPENDIXC

ERRORRATEANALYSES:BASEBANDSYSTEMS

C.3.1 BinaryTransmission considerthe receivermodelof thedigital transmission systemshownin Figurec.g. Thereceiverconsistsof two parts:signalprocessing circuitryanda datadetector(decisioncircuit).For thetimebeing,assumethattheoutputof thesignalprocessing circuitproducesapulseof amplitude+vwhenaI istransmittedandapulseof amplitude *v whena 0 is transmifted.obviously,the detectormerelyexaminesthepolarityof its input at the sampletimesdefinedby the sampleclock.A decisionerroroccursif noiseat the sampletimeshasan amplitudegreaterthanv anda polarityoppositeto thetransmittedpulse. Themostcommonlyanalyzedtypeof noiseis assumed to havea Gaussian or normal probabilitydistribution.Thustheprobabilityof enor ps canbe determined as

po = ^-l- l r-'tzo' dt - \txr o "v

(c.16)

whereoz is therms noisepowerat thedetector.Usingtheenor function, , erf a=|l

l

t

z

.vr, e-r-dt ;

(c.r7)

Equation C.16is sometimes rewrittenas P"=|(1-erfz)

(c.18)

where.e= v/"12o. TheerrorprobabilityPs canalsobe expressed in termsof thecomplimentaryerror function:

;t4# Noire Figure C.8.

Digital receiver model,

BASEBAND ERROBBATEANALY$ES: SYSTEMS 595

(C.19)

P"=ferfcz where erfc4-1-erfz

-

{2o

In lieu of evaluatingthe integralin EquationC.17or C.19 (which hasno closed form solution),theerrorfunctionscanbe approximated as

p-z

srfs g -.:F(e'vn;

(e>> l;

C.3.2 Multllevel Transmlssion The error rateof a multilevel basebandsystemis easilydeterminedby an appropriate reductionin theerrordistance.If themaximumamplitudeis V, the errordistanced betweenequallyspacedlevelsat thedetectoris d=*

(c.21)

whereL is thenumberof levels.Adjustingthe errordistanceV of a binarysystemto thatdefinedin EquationC.21providestheerrorrateof a multilevelsystemas

""=['=;)F}.'{.t#)

(c.22)

wherethe factor (L * 1)/Z,reflectsfhe fact that interior signallevels arevulnerableto bothpositiveandnegativenoiseandthefactor Illog2L arisesbecause themultilevel systemis assumed to becodedsosymbolerrorsproducesingle-biterrors(log2.Lis the numberof bits per symbol). EquationC.22revealsthat,with respectto peaksignal-to-noise ratiosat thedetector, L-leveltransmission incursa penaltyof 20log10(/-- l) decibels.If Vis increased by a factorL - I, theerrorrateof thellevel systemis identicalto theenor rateof the binarysystem[exceptfor thefactorslftogyL and(f - 1/L, whichtypicallyrepresent only a few tenthsof a decibell. EquationC.22relateserrorrateto thepeaksignalpowerIE. To determinetheerror ratewith respectto averagepower,the averagepowerof an L-level systemis determinedby averagingthepowerassociated with thevariouspulseamplitudelevels:

596

APFENDTx c

.[#J 3L[*l 7

(Il)"",

?

2

. . F ]

'

\

/

\

=ffiT^Qi-t)?

(c.23)

wherethelevels

v (r- 1)} r_, {+t,+3,...,* areassumed to be equallylikely. C.3.3 Energy.per-Blt-to-Noiee-Den$ityRailos The foregoingerror rate equationsrelatePBto the signalenergyat the sampletimes andthermsnoisepowerat thedetector.Whencomparingvariousdigital modulation formats,multilevel systemsin particular,it is morerelevantto relateerror performanceto signalpowerandnoisepowerat theinputto thereceiver(in front of thesignal processing circuitry).As a first stepin developinganerrorrateequationbasedon signal-to-noise ratiosat thereceiverinput,thenoisepowerat thedetectoris determined. Thevarianceofthe noiseo2 usedin thepreviousequationsis exactlyequalto the rmspowerthatwouldbemeasured atthedetectorin theabsence of a signal.Thisnoise powercanbe determined analyticallyas

o": I lHffi(|No)t'df

= NDJ IHU)I' df

(c.24)

0

=No'Brv

(c.2s)

where-.rNs is the one-sidednoise power spectraldensity in watts per hertz and Brv= Jo=lH(fllz df is thenoise-equivalent bandwidthor simplythenoisebandwidth of thereceiverfilter functionHll). In Equationc.?5, thenoisesourceis assumed to be white.Thatis, a uniformspectral densityexistsacrosstheentirebandof interest.This noisemay existin thetans-

C.3 EHHORRATEANALYSES: BASEBAND SYSTEMS 597

missionmediumitself,or it mayoccurin the "front-end"amplifierof thereceiver.If the rms noisepowerpassingthrougha $quare(ideal)frlter of bandwidthB is measured,a readingof (IV$(B)wattswould beobtained.ThusB1yrepresentsthe bandwidth of a perfectly squarefilter that passesthe sameamountof noiseasthe receiverfilter in its amplituderesponsel. H(l) lH(f) may be decidedlynonsquare providesa compromisebetweentwo conThe receiverfunctionH(fl necessarily flicting objectives.First,it mustminimizethe amountof noisepassedto thedetector (i.e.,minimizeEry).Second,the differencebetweensamplevalues(+Vand *I4 must be maximized.Obviously,the signal-to-noise ratio at the detector(Vld) shouldbe maximizedto minimizetheerrorprobability.A classicalresultof digitalcommunication theorystatesthat V/o is maximizedwhenHIJ) is'?natched"to thereceivedsignal. Whenviewedin thetimedomain,a "matchedfilter" is implemented by correlating (multiplying)thereceivedsignalwith eachof thereceivable(noise-free) pulseshapes. The outputsof the correlatorsareintegratedacrossa signalinterval to determinethe overall averagecorrelationduringthe interval.The integratorwith the maximumoutput indicatesthemostlikely symbolto havebeentr-ansmitted. In mostsystemsall signalsor pulseshavethesameshapebut differ only in amplitudeandpolarity.Thusa singlematchedfilter canbe used.Detectionmerelyinvolves comparingthematchedfrlteroutputto appropriate decisionlevels.Theoutputof a single correlatorft(r) andits integratoris expressed as T

V=)s(t)h(t)dt 0

= J ls(r)Fdr

(c.26)

0

wheres(r)is the signalor pulseshapebeingmeasured. Noticethat Vis, in essence, a of theenergyin the signaloverthe signalinginterval7. measure Whenanalyzedin thefrequencydomain,amatchedfilterresponse f(/) is thecomplex conjugateof the channelpulsespectrumS(f). Thus the matchedfilter output Y(/) canbe expressed in thefrequencydomainas

YU)=H(f)'SU) = S.(fl . ,S(fl

(C.zt1

Frequency-domain representations aremost convenientwhen the transmitsignalis stictly bandlimited,implyingthat the durationof thepulsercsponseis theoretically unlimited.In thiscasetheenergyin a pulse(e.g.,raisedcosine)is directlyproportional

598

APPENDTx c

to the detectorvoltageat the optimumsampletime. Henceoptimumdetectionis achievedby merelysamplingthe outputof the receivefilter at the propertime. Usingtherelationshipof EquationC.25andrheparameterE5 to represent symbol energy,we expressthebinaryerrorrateEquationC.I 9 as P" = j erf(z)

(c.28)

wherez2= Es/(Nfiil. Notice that for a given system(fixed 81,,),the error rate is dependentonly on the ratio of thesymbolenergyE5andthenoisedensityNs.This ratio is commonlyreferred to asa signal-to-noise ratio,althoughit is not a signal-power-to-noise-power measurement.Equationc.?8 is thepreferredform of theenor rateequationfor comparingdifferentmodulationschemes. In a binaryscheme,the symbolenergyEs is equalto the energyperbit Es. As an exampleof a specificsy$tem,considera baseband raised-cosine channel with optimumpartitioning.Theoutputspectrumof the hansmitteris the squareroot of theraised-cosine spectrumY*(fl definedin Equationc.2. Thematchedreceiverfilter alsohasa squareroot of a raised-cosine (EquationC.6).Hencethenoise response bandwidthof the receivercanbe determinedas

; BN=J tHU)t' df 0 (l+fl)/27

= ! tY*u)tdf 0

I

2T

(independent of u)

(c.?9)

As defined in Equation c.l and shown in Figure c.3, the (normalized)peak sample value at the detector of a raised-cosinepulse is L Using unnormalizedpulses of amplifude Es, we can determinethe error rate of a binary (+Es, -Es) raised-cosinechannel as

Pu = | erfc(z)

(c.30)

wheree2= (EslNg)f andZis thedurationof a signalinrerval. AlthoughEquationC.30wasderivedfor a raised-cosine channel,it is moregeneral in thatit is applicableto anybinarysystemusingantipodalsignaling.ThusEquation c.30 is plottedin Figure4.23asthebestperfonnance achievable by *y digitalhans-

c.3 EHFORRATEANALYSES:BASEBANDSYSTEMS 5gg

missionsystemdetectingonepulseat a time.(Lowererrorratesarepossiblewhenredundantsignalsor enror-coffectingcodesare used.) presented The enor rateperformance in EquationC.22 for multilevelsystemsis basedon noisepower at the detector.As long asthe signalbandwidthis identicalfor all systems, EquationC.22is valid sincethe noisebandwidthof thereceiveris independentof the numberof levels.Whenthe signalingrateis held constant,however, thebit rateincreaseswith thenumberof levels.To comparemultilevel systemson the basisof a given datarate,the signalinginterval7 and,hence,the noisebandwidths mustbe adjustedaccordingly. If 7 is thesignalingintervalfor a two-levelsystem,thesignalingintervalI for an /,-levelsystemprovidingthe samedatarateis determinedas

Tr=TlogrL

(c.31)

Usingthenoisebandwidthof a raised-cosine filter in EquationC.29,we extendEquation C.22to multilevelsystemsas

"=[,#)[#)-'*o .

v/(L * L)

EquationC.3?canbe simplifiedandpresented in a morecustomaryform by usingthe relationshipthatenergyper symbolEr= Ealog2L = IPTI, whereE6is theenergyper bit:

"=[,#J[tt)*'r.r r=

(c.33)

(logrL)r/z(Eo\"

1-1

l1r I \'0,/

EquationC.33is plottedin Figure4.26for 2, 4, 8, and 16levels.Thesecurvesrepresentthe ideal relativeperformances of multilevelbasebandsystemsat a constant in propordatarate.Thebandwidthrequirementsof thehigher-levelsystemsdecrease to log2L. tion

600

APPENDTx c

Equationc.33 and Figure4.26 representthe performanceof multilevelsystems with respectto E6lNp(commondataratebut differentbandwidths).The following relationshipcanbe usedto determineerrorrateswith respectto signal-to-noise ratiosat thedecisioncircuit (differentdatararesbut commonbandwidth): siTal Powet nolsepower

Slgp:

_(E)(rogzL)(r/T) (No)0/2r) / F \

=(2)(log2z)l# | l-"0i

(c.34)

wherell?Tpis theminimum"Nyquist"bandwidthof the signal. ThesNR obtainedin Equationc.34 repre$ents theratioof signalpowerat thesample time to noiseat thedetector.This ratiois sometimes refenedto asa postdetection sNR because it is the signal-to-noise ratio at the outputofthe signalprocessing circuiffry. Somecommunications theoristsusepredetectionsignal-to-noise ratiosin determining elror rates.SincepredetectionSNRsare measuredprior to bandlimiting the noise,a noisebandwidthmustbehypothesized to establisha finite noisepower.commonly,a bit ratebandwidth( l/7) or a Nyquistbandwidth(l/zT) is specified.Thelatrer specification producesSNRsidenticalto thatin EquationC.34.Exceptionsoccurwith double-sideband modulationusing coherentdemodulation(e.g.,2-psK) wherethe predetection sNR is 3 dB higherthanthepostdetection sNR. (All signalpoweris coherentto thedemodulator carrierreference, but only halfofthe noiseis .,coherent.") C.3.4 Partlal-ResponseSystems Theenor rateequationfor a I + D partial-response systemis obtainedby incorporating thefollowingmodificationsto a full-response system: l. The error distanceis exacflyone-halfthe error distance of a corresponding full-response system(seeFigureC.6). 2. The noise bandwidthof the receiver(assuming$quare-rootpartitioning) is obtainedby integratingIIsx of F4uationC.14: | /27

B"=J cosMTdf 0

_ l nT

(c.35)

TRANSMISSION C.4 CARRIEF SYSTEMS601 Thusthenoisebandwidthof a 1 + D patial-respon$e systemis 2 dB lowerthanthe (full-response) noisebandwidthof a raised-cosine systemobtainedin EquationC.29. Sincetheerrordistanceis reducedby 6 dB, partial-response filteringincursa netpenalty of 4 db-with respectto unfilteredsignalpowerat the source.In termsof channel powers,the panial-response systemincursa smallerpenaltyowing to greaterspectrum truncationin the hansmitfilter. Thedifferencein channelpowersof thetwo sy$tems canbeobtainedby integrating partitioningis used,thecharurel the respectivechannelpowerspectra.If square-root powerof the partial-response systemis 2 dB belowthe channelpowerof the correspondingfull-response systems.In this case,the differencein channelpowersi$ exactlyequalto thedifferencein noisebandwidthsbecause thereceiverfilter responses arematchedto therespectivechannelspectra.That is, integrationofchannelspectra hasessentiallybeenaccomplished in EquationsC.29andC.35. pafiiWhenspectrumtruncationlossin a transmitteris considered, a square-root tionedpartial-re$ponse systemis only 2 dB worsethanthat of a corresponding fullresponse system.Of course,thepartial-response systemrequireslessbandwidththan the full-response$ystem.For completeness, the erTorrateequationof a square-root partitionedI + D partial-response systemis

"=['#Jf-r)-'r-t.r

(c.36)

/2(logzL)trzrrt - 1)andE6is theenergyperbit on thechanwhere{ = (rE/4)(Eb/No)t nel. EquationC.36is identicalto EquationC.33exceptfor thefactorrd4resultingfrom thelower errordistance,thelower noisebandwidth,andthe lower channelpowerof thepartial-re$ponse system.

C.4 CARRIERTRAN$MISSION$YSTEMS C.4.1 Filter Design Exceptfor a few relativelyuncommonfrequency-modulated systems, digitallymodulatedcarriersystemscanbe designedand analyzedwith baseband-equivalent channels.Carrier-based filters to bandpass filters arederivedby kanslatingthe baseband filters centeredat the carrierfrequency.The outputpulseresponseof the channelis determinedby the compositeof the baseband-equivalent filters. Thuspulseshaping be by the signals or the can achieved filtering baseband modulatedsignals.Partitionof the application. In all casesthecoming channelfilter functionis dependent on the positechannelresponse identical that in Equation C.2 is to defined for raised-cosine channelsor EquationC.13for partial-response systems.

602

APPENDIX c

Q.4.2 Error BateAnatysis Error rate analysesof basebandsystemscan be direcfly applied to carrier $ysremsunder one important condition: Coherentmodulation and demodulationmust be used. For example,coherentdemodulation of a 2-pSK signal y(r) = cos[fitf + Q(I)] involves implementing the following equations;

+ Q(r)]2 cos(rrlr)] S(t)= lowpass{cos[rol = lowpass{[(cosQ(r)cos(Dr* sin S(r)sin rryl2cosrof ] = lowpass{cos $(rxl + cosrrlr)- sinS(r)sin 2cor} = cosS(r)

_ J r for Q(t)= 6 [-r

(c.37)

for S(t)= 7s

Notice that coherentdemodulationinvolves multiplying the received signal by a local carrier that is exactly in phasewith the respectiveincoming signal.Hence coherentdemodulation is closely related to matched filter detection as presentedin Equations c.26 and c.27. To complere optimum detection of a digitally modulated signal, the basebandequivalent of the receiver filter function must also be matchedto the carrier envelope: cos Q(r). When coherent demodulation (also called coherent detection) is used,the error rate performanceis identical to the analogousbasebandsystem.Hence the error rate of a coherently detectedz-psK system is provided in Equation c.19 or

c.28.

The correspondenceof coherentdemodulation to matchedfilter detection is illustrated in Figure c.9. For convenience,digital biphase (diphase) is compared to an NRZ basebandsignal. The basic principle also applies to sine wave carriersat any frequency-The important point to notice in Figure C.9 is that the output of the coherent demodulator (or equivalently, the matched filter) is identical to the NRZ sisnal. Fur-

I t l t l o l r l o l r l r-l t-l Dffi I-l

T,H|[lf'-"-lJ+ f f i hodrm

l-r

Figure C.9.

t-t

Mlxcr (defitodulfiorl

r

|

_-r

Coherent demodulation (detection) of digital biphase signal.

I

THANSMtsstoN c.4 cARRIEB sysrEMS 603 thermore,thenoisepowerdensitycomingoutof thecoherentdemodulator is identical system.(Positive-weighted to thenoisepowerdensityof thebaseband white noiseis statisticallyno differentthannegative-weighted whitenoise.) FigureC.9 demonstrates that coherentlydemodulated carrier-based signalsproratiosat the detectorasbaseband systems,despitethe ducethe samesignal-to-noise fact that (double-sideband) carriersystemsrequiretwiceasmuchbandwidth.Coherent demodulationleadsto a receivernoisebandwidttrequalto the baseband-equivalent noisebandwidthbecause only one-halfof the noisepowerin thecarriersignalband(Noisein the carriersignalbandwidth width is passedby the coherentdemodulator. that is out of phasewith respectto thecoherentreferenceis ffanslatedto twice the carrier frequencyand thereforeeliminatedby the low-passfilter, sin ot 'cos o)f = I sin 2rot.) C.4.3 QAM Error Rates The error rateequationof a coherentlydetectedQAM systemis identicalto the error rate of the conespondingmultilevelbasebandsystemappliedindividually to each quadrature of a 16-QAM systemis providedin channel.Thusthe eror performance EquationC.33for I = 4 levels.In a QAM systemcoherentdemodulation causesonehalf of thenoisepowerin the carriersignalbandwidthto showup at the/-channeldetectorand one-halfshowsup at the Q-channeldetector.Of course,the total signal poweris dividedin two sothesignal-to-noise ratio at theindividualdetectorsis identicalto thecarier signal-to-noise ratio(i.e.,predetection SNRis equalto postdetection SNR).Theerrorrateperformance of QAM systemsis plottedin Figure6.20in terms ratios,useEquationC.34usingL of E/Ns. To relatethoseresultsto signal-to-noise asthe numberof levelson eachquadrafurechannel. C.4.4 PSK Error Rates The errorrateof a multilevelPSK systemis derivedmosteasilyby usingquadrature channelrepresentations for thesignals.For example,FigureC.10displaystheregions of decisionerrorsfor a repre$entative signalphasein an 8-PSKsignal.Thereceived signals signalis processed by two orthogonalphasedetectorsto producequadrature Y{t) and Yp(t)in Equations6.10and6.11,respectively. If thetransmittedphaseis rd8 (corresponding to datavalues011),a decisionerror resultsif noisecausesIg(0 to go positiveat the sampletime [yo(t) is positivedownwardto represent thep (sine)channelaslaggingthe1(cosine)channell.Thenormalizederrordistanceis sin (n/8).A decisionerroralsoresultsif noisecauses-Ia(t) to exceedY{t), indicatingthephaseis greaterthan-nl4. This latterconditioncanberepresentedas a negative value for the hansformed signal Y3(t)=O.lOiYBft) + A.7O7Y6[I). fHere,Y6(f)is aprojectionof thereceivedsignalontoa -nl4 basisvector (seeFigure6.l3)1.Examinationof FigureC.l0 revealsthattheerrordistancefor this secondtypeof erroris alsosin (d8). Sincethe noisevarianceyB(f)is identicalto the

604

APPENDIX c Feglon 2 whero 0lO ir dstsctod

Rcgion I nlrcre lll hdstsrsd

FigureC.10. Regions of decision errorfor 8-pSKsignalatldg (01I ). noisevarianceof r{f) (by virtue of the 0.707multipliers),both typesof errorsare equallylikely. In general terms, the error distanceof a psK system with N phasesis v' sir{n/M, wherev is the signalamplitudeat a derector(i.e.,the radiusof a psK signalconstellation). A detectionerroroccursif noiseof theproperpolarityis present at theoutputof eitherof two phasedetectors. A detectionerror,however,is assumed to produceonly a single-biterror.Thegeneralexpression for thetheoreticalerrorrate for PSKmodulationis now determined by modifyingEquationC.22as

"":['chJ*'oo where

^

sin(n/MV r/Zo

(c.38)

The signal amplitude V can be expressedas

v=frorros, r [+)"'

(c.3e)

and the rms noise voltage o as

"=[",[,+)'"

(c.40)

c.4 CARRIER TRANSMISSIoN SYSTEMS 605

for noisein a Nyquist bandwidth. CombiningEquationsC.38,C.39,andC.40relatesPSK errorrate$to energyper bit to noisedensityon thechannel:

""=['=lo)w'r

(c.41)

/rr\ ..^ (a ,1 " lOt/2| + | z= sin| ; '/lttosz

\

l.toj

EquationC.4l is plottedin Figure6.16 for PSK systemswith variousnumbersof phases.To determineerrorrateswith respectto signal-power-to-noise-power ratios, usethefollowing: (N> 2) sNR=t"-,"h1J

(c.42)

For 2-P$K systems,the errorratesas specifiedin EquationC,38 or C.41 shouldbe divided by 2 because only onephasedetectoris neededandit produceserrorsfor onepolarity of noiseonly.

APPENDIX D

TRAFFICTABLES TableD.l is a tableof maximumofferedloadsA for variousblockingprobabilitiesB with lost andnumberof serversN. Theblockingprobabilitiesarefor infinite $ources callscleared(Erlang-B,Equation10.8). TableD.2 is a tableof maximumofferedloadsA for variousblockingprobabilities B, numberof serversN, andfinite numberof sourcesM. TheofferedloadA is deterno callsarecleared.The minedasMp, wherep is theaveragesourceactivityassuming from Equation12.13. blockingprobabilityfor finite sourcesis determined Note: The following tableswereobtainedfrom TelephoneTraffic Theory,TaMunich, 1970. hles,and Charts,SiemensAktiengesellschaft, TABLED.l MaxlmumOfieredLoadVersusBand M 0.01 0.0s .0001 .0005 ,014 .032 .087 .1s2 .23s .362 ,452 .649

0.1 .001 .046 .194 .439 .762

0.5

1.0

10

15

20

,00s .105 .340 .701 1,13

.010 .1s3 ,45s ,869 1.36

.020 .223 .602 1.09 1.66

.053 .381 .899 1.62 2.22

.11'l .595 1.27 2.05 2.BB

.176 ,796 1.60 2.s0 3.45

.250 1.00 1.93 2.9S 4.01

30 .429 1.45 2.63 3.89 5.10

.667 2.00 3.48 $.42 6,60

10

.728 1.05 1.42 1.83 2.26

.9S6 1.39 1.83 2.30 2.80

1.15 1.58 ?.0S 2.56 3.09

1.62 2.16 2.73 3.33 3.96

1.91 2.50 3.13 3,78 4.46

2.28 2.94 3.60 4.34 5.08

2.96 3.74 4.54 5.37 6.22

3.76 4.67 5.60 6.55 7.51

4.44 s.46 6.50 7.s5 8.62

5.11 6.23 7.37 8.52 9.68

6.51 7,86 9,21 10,6 12.O

8.19 9.80 11.4 13.0 14.7

11 12 13 14 15

2.72 3.21 3"71 4.24 4.78

3.33 3.65 4.61 5,16 5.84 7.08 s.88 4.23 5.28 s.88 6.61 7,95

8.49 9.47 4.45 4.83 5.96 6.61 7.40 8.83 1 0 . 5 5.03 5.45 6.66 7.35 8.20 9.73 1 1 . 5 s.63 6.08 7.38 8.11 9.01 1 0 , 6 1 2 . 5

9.69 10.8 11,9 13.0 14.1

10.9 12.0 13.2 14.4 15,6

13.3 14.7 16.1 17.5 1B.g

16.3 18.0 19.6 21.2 22.9

16 17 18 19 20

5.34 s.91 6.50 7,09 7.70

6.25 6.88 7.52 8.17 8.S3

15.2 16.3 17.4 18,5 19.6

16.S 18,0 19.2 20.4 2 1. 6

20,3 21.7 23.1 ?4.5 25.9

24.5 26.2 27.8 29,5 31:2

MB 1 2 3 4 5 6 7 I o

6.7? 7.38 8.05 8.72 9,41

8.10 8.83 9.58 10,3 11. 1

8.88 9.65 10.4 11.2 12.0

9.83 10.7 11.5 1e.3 13.2

11.5 12.5 13.4 14.3 15.2

13,5 14.5 15.5 16.6 17.6

40

608

APPENDIx D

TABLED.1 (Continued) NIB

2

10.1 10.8 11.5 12.2 13.0

0.5 11,9 12.6 13.4 14.2 15.0

1.0

21 22 23 24 25

0.01 0.05 8.32 9.50 8.95 10,2 9.58 10.9 10.2 11.6 10.9 1?,3

12.8 13.7 14.5 1s.3 16.1

14.0 14.9 15.8 16,6 17.5

5 16.2 17.1 18.1 19.0 20.0

187 19.7 aQJ 21.8 22.8

15 20.8 ?1.9 23.0 24.2 ?5,3

22.8 24j 25,3 26.5 277

30 27.9 ?a.7 30.1 31,6 33.0

32.8 34.5 36.1 37.8 39.4

26 27 28 29 30

11.5 1?.2 12.9 13.6 14.2

13,0 13.7 14.4 15.1 15.9

13.7 14.4 15.2 15.9 16.7

15.8 16.6 17.4 18.2 19.0

17.0 17.8 18.6 19.5 20.3

18.4 19.3 20.2 21.0 21.9

20.9 21.9 22.9 ?3,8 24.8

23.9 e4,9 26.0 27.1 28.1

26.4 27.6 28.7 29.9 31.0

28.9 30.2 31.4 32.6 33,8

34.4 35.8 97.2 38.6 40.0

41.1 42.8 44.4 46.1 47.7

31 32 33 34 35

14.9 15.6 16.3 17.0 17.8

16,6 17.3 18.1 18,8 19.6

17.4 18.2 19.0 19.7 20,5

19,9 20.7 21,5 22.3 23.2

21.2 22.0 22.9 23.8 24.6

22.8 23J 24.6 25.5 26.4

2S.8 26.7 277 28.7 29.7

2s.2 30.2 31.3 s2.4 33.4

32.1 33.3 U.4 35.6 s6J

3s.1 36.3 s7.5 38.8 40.0

41.s 42.9 44.9 45.7 47.1

49.4 51.1 52.7 54.4 56.0

36 37 38 39 40

18.5 19.2 19,9 20.6 21.4

?0,3 21.1 21.9 22.6 23.4

21.3 2e.1 22.s 23.7 24.4

24.0 24.8 25.7 26.s 27.4

25.5 26.4 27.3 28.1 29.0

27.3 28.3 29.2 30.1 31.0

30,7 31.6 32.6 33,6 34.6

34.s 35.6 36.6 s7.7 38.8

37.9 39.0 4Q.2 41.3 42.5

41.2 42.4 4s7 44.9 46.1

48.6 50.0 51.4 52.8 54.2

57.7 5e.4 61.0 62.7 64.4

41 42 43 44 45

22.1 22.8 23.6 24.3 25.1

24.2 25.0 25.7 26.5 27.3

25.2 26.0 26.8 27.6 28.4

28.2 29.1 29.9 30.8 31.7

29.9 30.8 31.7 32.s 33.4

31.9 32.S 33.8 U.7 3s.6

3s.6 36.6 37.6 38.6 39.6

39.9 40.9 42.0 43.1 44.2

43.6 44.8 4s.9 47.1 48.2

47.4 48.6 49.9 51.1 s2.3

55,7 57j s8.5 59.S 61.3

66.0 67.7 6e.3 71.0 727

46 47 48 49 50

?5.8 26.6 27.3 28.1 28.9

28.1 28.9 29.7 30.5 31.3

29.3 30.1 30.9 31.7 32.s

32.5 33.4 34.2 35.1 36.0

34.3 35.2 36,1 37.O 37.9

36.s 37.5 38.4 39.3 40.3

40.5 41.5 42.5 43.s M.5

45.2 46.3 47.4 48.s 49.6

49.4 50,6 51.7 52.9 s4.O

53.6 s4.8 56.0 57.3 s8.s

62.8 64.2 65.6 67.0 68.5

74.3 76.0 77.7 7s.3 81.o

51 52 53 54 55

29,6 30.4 31.2 31,9 32.7

32.1 3?,9 33.7 34.5 35.3

33.3 34.2 35.0 3s.8 36.6

36.9 37.7 38.6 39.5 40.4

38.8 39,7 40.6 41.5 42.4

41.2 42j 43.1 44.0 44.s

45.5 46,s 47.5 48.5 49.s

50.6 51.7 52.8 53.9 s5.0

55.2 56.3 57.5 58.7 59.8

59.7 61.0 62.2 63.5 64.7

69.9 71.3 72J 74.2 7s.6

82J 84.3 86.0 87,6 89.3

56 57 58 59 60

33.5 34.3 3s.1 35.8 36.6

36.1 36.9 37.8 38.6 39.4

97.5 38.3 39,1 40.0 40.8

41.2 42.1 43.0 43.9 44.8

43.3 M.2 45.1 46.0 46.9

45.9 46.8 47.8 48.7 49.6

50.5 51.5 52.6 53.6 54,6

56.1 57.1 58,2 59.3 60.4

61.0 62,1 63.3 64.5 65,6

65.9 67J 68.4 69.7 70.9

77.O 78.4 79.8 81.3 827

91.0 92.6 94.3 96.0 97,6

61 62 63 64 65

37.4 38.2 39.0 39.8 40.6

40.2 41,0 41.9 42.7 43.5

41.6 42.5 43.3 44.2 45,0

45.6 46.s 47.4 48.3 49.2

47.9 48.8 49.7 50.6 51.5

50.6 51.s 52.5 53,4 54.4

55.6 56.6 57.6 58.6 59.6

61,5 62.6 63.7 64,8 65.8

66.8 68.0 69,1 70.3 71.4

72j 7s.4 74.6 75.s 77.1

84.1 8s.s 87.0 88.4 89.8

99.3 101. 103. 104. 106.

0.1

10

20

40

APPENDIXD

609

TAELED.l (Gontlnuedl ME 66 67 68 69 70

1.0

10

15

20

30

40

44.4 48.2 46.0 46,8 47.7

0.1 45.8 46.7 47.6 48.4 49.2

0.5

41.4 42.2 43.0 43.8 44.6

50.1 51.0 51.9 52.8 53.7

52.4 53.4 54.3 s5,2 s6.1

55.3 56.3 57.2 58.2 59.1

60.6 61.6 62.6 63.7 64.7

66.9 68.0 69.1 7Q.2 71.3

72.6 73.8 74.9 76.1 77.3

78.3 79.6 80.8 82.1 83.3

91.2 92.7 94.1 95.5 96.9

108, 109. 111. 113. 114,

71 72 73 74 75

45.4 46,e 47.0 47.8 48,6

48.5 49.4 50.2 51.0 51.9

50.1 50.9 s1.8 52.7 53.6

54.6 55.5 56.4 57.3, 58.2

57.0 58.0 58.9 59.8 60.7

60.1 61.0 62.0 62.9 69.9

65.7 66.7 ts7.7 68"7 69.7

7?.4 73.5 74.6 75.6 76.7

78.4 79.6 80.S 81.9 83,1

84,6 8s.B 87.0 88.3 89.5

98.4 99.8 101, 103. 104.

116. 118. 119. 121. 123.

76 77 78 79 80

49.4 50.2 s1.1 51.9 52.7

52,7 53.6 54.4 55,3 56.1

s4.4 55.2 s6,1 56.9 57.8

59.1 60.0 60.9 61.8 62,7

61,7 62.6 63.5 64,4 65.4

64.9 65.8 66,8 67.7 68.7

70.8 71.8 72.8 73.8 74.8

77,8 78.9 80.0 81,1 82.2

84.2 85.4 86.6 87.7 88.9

90.8 92.0 93.3 94.5 95.7

10s. 107. 108. 110. 111.

124. 126, 128. 129. 131.

81 82 89 84 8s

s3.5 54.3 55.1 56.0 s6.8

5S.9 s7,8 58.6 59,5 60.4

58.7 59.5 60.4 61.3 62.1

63.6 64.5 65.4 66.3 67.2

66.3 67.2 68.2 69.1 70.0

69.6 70.6 71.6 72.5 73.5

75.8 76.9 77.9 78.9 79,9

83.3 S4.4 85.5 86.6 87.7

90.1 91.2 92.4 93.6 94.7

97.O S8,2 99.5 101. 102.

113. 114. 115. 117, 118.

133" 134. 136, 138. 139.

86 87 88 89 90

57.6 58.4 s9,3 60.'1 60.9

61.2 6e.1 62.9 63.8 fr.6

63,0 63.9 M.7 65,6 66.5

68.1 69.0 69.9 70.8 71.8

70.9 71.9 72.8 73.7 74.7

74,5 75.4 76.4 77.3 78,3

80.9 82.0 83,0 84.0 8s.0

88.8 89.9 91.0 92.1 93.1

9s.9 97.1 S8.2 99,4 101.

103. 104. 106, 107. 108.

120. 121., 123. 124. 126.

141, 143. 144. 146, 148,

91 92 93 94 95

61.8 62,6 69.4 64,2 65.1

65.s 66.3 67.2 68.1 68.9

67.4 68,2 69.1 70.0 70,9

72.7 73.6 74.5 75.4 76.3

75.6 76.6 77.5 78.4 79.4

79.3 80.2 81.2 82.2 83.1

86.0 87.1 88.1 89,1 90.1

94.2 95.3 96,4 97.5 98.6

102. 103. 104. 105. 106.

109, 111, 112. 11 3 . 114.

127. 128. 130. 131. 133.

149. 151. 1s3. 154. 156,

96 s7 98 99 100

65.9 66.8 67.6 68.4 69.3

69.8 70.7 71.5 72.4 73.2

71.7 72.6 73.5 74.4 75.2

77.2 78.2 79.1 80.0 S0.9

S0.3 81.2 82.2 83.1 84.1

84.1 85.1 86.0 87.0 88.0

91.1 92.2 93.2 94.2 95.2

99.7 101. 102. 103. 104.

108. 116. 134. 1 0 9 . 1 1 7 . 135. 110. 118. 137. 1 1 1 . 1 1 9 " 138. '11?. 121. 140.

158. 1$9. 161. 163. 164.

0.01 0.0s

a/Vis the nufibef of servers. Th6 numericalcolumn h€edingsindicateblocklngprobability(%).

610

APPENDIXD

TABLED.2 MaxlmumOfferedLoadVersusB, A|,and Flnite $ourceellla N

M

.01

0.1

5

0.5 1 . 0 10 0002 ,0010 .0020 ,0100 .0200 .0400 .100 .202 .0002 .0008 ,001s .0075 ,0151 .0303 .077 .1s9 .0001 ,0007 .0013 ,0067 .0134 .0270 ,069 .143 . 0 0 0 1. 0 0 0 6 . 0 0 1 3 . 0 0 6 3 . 0 1 2 6 . 0 2 5 4 . 0 6. 51 3 6

3 4 5 6 7

.030 .023 .021 .019 .018

2 '' r t ' 43

4 5 6 7 8 9 10 15 5 6 7 8 9 10 r5 6 7 I 10 12 15 20 7 I I 10 15 20 30 8 9 10 15 2Q 30

.067 .0s2 .046 .043 .041

.095 .074 .065 .061 .058

.212 .167 ,149 .139 .133

.300 .238 .213 .200 .191

.425 .342 .308 .289 .277

.678 .s60 .510 .482 .464

15

20

30

.307 .246 .224 .213

.417 .341 .312 .298

.6ss .559 .s19 .498

.980 1.23 1.47 1.s7 .832 1.07 1.30 1.78 .787 .gs7 1.22 1.70 .731 .955 1.18 1,65 .707 .928 1.1s 1.6? .186 .317 .400 .685 .864 1.09 1.s0 1.95 2.31 2,65 3.35 .148 .254 .322 ,561 .715 .918 1.30 ,t.72 2.OB 2.42 3.13 .131 .227 .288 .505 .648 .837 1.20 1.62 1.97 2,g1 3.02 j22 .211 .268 .473 .609 .790 1.14, 1.5s 1.90 2.24 2.e5 .116 .201 .255 .452 .583 .759 1.10 1.51 1.85 2.19 2.90 ,111 .194 .246 .437 .s65 .737 1.Q7 1.47 1.82 2.16 2.86 .108 .1S8 .240 .426 ,551 .720 1.05 1.45 1.79 2.13 2.84 .100 j74 .222 .396 .514 .67s ,994 1.38 1.72 2,06 2.76 .500 .748 .889 1.33 1.59 1.89 2.24 2.98 3.43 3,86 4.76 .408 .617 .737 1.12 1.36 1.64 2.15 2.71 3.16 s.6o 4.51 .365 .554 .665 1.02 1.24 1.52 2.01 2.56 3.02 3.46 4.38 .340 .517 .621 .963 1.17 1.44 1.9? 2.47 2.93 3.37 4.30 .323 .492 .592 .922 1.13 1.39 1.86 2.4't 2.87 3.31 4.?4 .310 .4't4 .571 .892 1.09 1,35 1.82 2.36 2.82 g.27 4.20 .280 .429 .518 .816 1.00 1.25 1.71 2.24 2.7Q 3.14 4.08 .9s1 1.31 1.51 2.08 2.40 2]7 3,39 4.Q4 4.58 5.09,6.17 .794 1.11 1.28 1.80 2.10 2.45 s.o7 s.7s 4.28 4.80 5.91 .7161,00 1.16 1.66 1.94 2.29 2.90 3,56 4j2 4.65 5.77 .668 .940 1.04 1.56 1.84 2.18 2.78 3.45 4.01 4.55 5.68 .695 .896 1,09 1.50 1.77 2.11 2.71 3.s7 3.94 4.48 s.6.1 .592 .,839 .979 1.42 1.68 2.01 2,60 9.27 3.84 4.38 5.52 ,556 .791 ,924 1.35 1.60 1.92 2.51 3.18 3.74 4.2g 5.44 .525 ,748 .876 1.28 1.53 1.84 2.42 3.09 3.66 4.21 5.37 1.51 1.97 2.21 2.90 3.26 3.69 4.38 5.12 5.73 6.32 7.59 1.28 1.70 1.91 2.55 2.90 3.32 4.o2 4.78 5.41 6,02 7.g2 1.17 1.55 1.76 2.37 2.71 3.12 3.82 4.59 5.24 5.S5 7.17 1.09 1.46 1.6s 2.25 2.58 2.98 3.69 4.47 s.12 5.74 7.O7 .926 1.2s 1,43 1.97 2.29 2.68 3.38 4.17 4.84 s.48 6,84 .865 1.17 1.34 1.87 2j7 2.56 g.A6 4.OS 4.72 S.i7 E:74 .8131.10 1.27 1.77 2.O7 2.45 3.15 3.94 4.62 s.28 6.66 2,15 2.70 2.98 3,76 4j7 4.63 5.39 6.20 6.89 7.56 9.01 1.8s 2.36 2.62 3.36 3.7s 4.22 s.00 5.8s 6.56 7.25 8.74 1.70 2j7 2.42 3.13 3.52 3.99 4.78 5.64 6.97 7.07 8.s8 1.39 1.80 2.02 2.68 3.05 3.51 4.31 5.21 5.97 6.70 8.2s 1.28 1,67 1.88 2.52 2.88 3.33 4J4 s.Os s.81 6.s6 8.13 1.19 1.56 1.76 2.38 2.74 3,18 3.99 4.90 5,68 8.44 8.03

APPENDTx D

611

TABLED.2 (Continuedl

,v

10

15

7.30 6.92 6.71 6.56 6.29 6.07 5.89

8,05 7.71 7.51 7.37 7.13 6.93 6.77

M

.01

.05

0.1

0.5

t.0

I 10 11 12 15 20 30

2.85 2.49 2.25 2.16 1,93 1.76 1.63

3.48 3,08 2.85 2.70 2.43 2.24 2.08

3.80 3.37 3.14 2.97 2.70 2A9 2.32

4.65 4.20 3.94 3.77 3.46 3.23 3.03

5.09 4.64 4.38 4.20 3.89 3.65 3.45

5.59 5.14 4.89 4.17 4.40 4.16 3.95

6,41 6.00 5.76 5,59 5.29 5,06 4.86

10 11 12 13 14 16 18 20 30

3.59 3.18 2.94 2.78 2.66 2.50 2.39 2.31 2j2

4.30 3.84 3.57 3,39 3.26 3.08 2.95 2.87 2.64

4.84 4.17 3.89 3.71 3.57 3.38 3.25 3.15 2.91

5.57 5.O7 4.75 4.s9 4.44 4.33 4.09 3.99 3,73

6.03 5.54 5.26 5.06 4.91 4.70 4,56 4.46 4.19

6.56 6,09 5.81 s.61 5.41 5.27 5.13 5.02 4.76

7.44 8.39 7.OO 8.01 6,75 7.78 6.57 7.62 6.43 7,50 6.34 7.33 6.11 7.21 6.02 7.13 5,77 6.90

10

11 12 13 14 15 16 1B 2a 25 30

4.38 3,91 3.63 3.45 3.31 3.20 3.04 2.93 2.75 2.65

5.15 4.64 4.34 4.13 3.98 3.86 3.68 3,56 3.36 3,25

5.52 5.00 4.6S 4.47 4.32 4.19 4.01 3.88 3,68 3.56

6.49 s.97 5.65 5,43 5.27 s.14 4,95 4.81 4.s9 4.47

6,98 8.47 6.16 5.94 5,78 5.6s s.46 5.32 5.10 4.98

7.54 7.04 6.74 6.54 6.38 6.2s 6.07 5.93 5,72 s.59

8.47 8.02 7.75 7,56 7.41 7.30 7.13 7.O1 6.81 6.69

11

12 13 14 1s 16 17 18 20 2s 30

s.19 4,68 4.37 4.15 3.99 3.86 3.76 3.60 3.36 3.32

6.01 5.46 5.13 4.90 4.72 4.s9 4.48 4.31 4.04 3.90

6.41 5.85 5.51 5.27 5.09 4.95 4.84 4.66 4.39 4.24

7.44 6,88 6.54 6.30 8.12 5.98 5.86 5.68 5.40 s.24

7,95 7.40 7.O7 6.84 6.66 6.52 6,41 6.23 5.95 5.79

8.53 8.01 7.69 7.47 7.30 7.17 7.06 6.88 6.62 6.46

9.s0 10.6 11.6 9,04 10.2 11.2 8.76 9,94 11.0 8.56 9.77 10.8 8.40 9.63 10.7 8.28 9.s3 10.6 8.18 9,44 10.5 8.03 9.31 10,4 7.79 9.10 10.2 7.64 8.98 10,1

12

13 14 15 16 17 18 20 25 30

6.03 5.47 5.13 4.88 4.7Q 4.56 4.34 4.03 3.85

6.90 6,31 5.95 5.69 s.50 5.35 5,'t2 4.77 4.s8

7.31 6.72 6.35 6.09 5.90 s.74 5.51 5.16 4.96

8.39 7.80 7.44 7.18 6.9S 6.83 6.60 6.24 6,04

8.92 8.35 8.00 7.75 7.56 7.41 7.18 6,83 6.63

9,52 10.s 11.7 B.9B10.1 11.3 8.65 9.77 11.O 8.42 9.56 10,8 8.24 9.40 10.7 8.09 9.27 10.6 7.87 9.08 10.4 7.54 8.78 10,2 7.35 8.61 10.0

the

numericalcolumn headingsindlcateblockingprobability(%).

20

30

8.80 10.4 8,48 10.2 8.2S 9.99 8.17 9.88 7.94 9.69 7.76 9.s4 7.61 9.41

9.22 10.0 11.9 8.87 5.72 1' 1.6 8.66 9.52 11.4 8.51 9.39 11.3 8.41 9.2S' |1.2 8.25 9.15 11.1 8.15 9.06 11.0 8.07 8.99 11.0 7.87 8.81 10.8

9.49 10.4 11.3 9.09 10,0 11.0 8"86 9.81 10.8 8.69 9:66 10'6 8.56 9.55 10.5 8.46 9.46 10.4 8.31 9.33 10,3 8,21 9.23 10.2 8.04 9,08 10.1 7.93 8.99 10.0 12.s 12,2 12.0 11.9 11.7 11.7 11.6 11.5 11.3 11.2

13.3 13.0 12.8 12.7 12.6 12.6 12.s 12.4 12.3 12.2 14.7 14.4 14.2 14.1 14.0 14.0 13.9 13.8 13.7 13.6

12.7 13.8 16.1 12.4 13.4 1s,8 '12.1 13,2 15.7 12.0 13.1 15.5 11.8 13.0 15,5 11.8 12.9 15.4 11.6 1e.8 1s,3 11.4 12.6 15.1 11,3 12.5 1s.0

GLOSSARY 2BlQ. Four-levelline codeusedin ISDN basicrateaccesslines. 997oPowerbandwidth.Bandwidthcontaining99Voof theenergyof a signal' Abbreviateddiating. ,SeeSpeeddialing. Accesstandem. "switching systemwithin a LATA used as an accesspoint by long-distance carriers." only a portionof a Add-drop multiplexer (ADM). Networkelementthataccesses intermediatepoint local at insert traffic an extract and higherlevel digital signalto of a route. Added-channelframing. TDM framing format utilizing an additionalchannelwith the samerate as the messagechannelsfor the expresspurposeof defining frame boundaries. Added-digit framing. TDM framingformatutilizing anadditionalbit in everyftame for the expresspurposeof definingframe with a prescribedrepetitivesequence boundaries. Advanced mobile phone service (AMPS). Analog cellular mobile telephone standardof NorJhAmerica. Alternate mark inversion(AMI) signal.Three-levelsignalwith values+V, 0' -y. A space(logic 0) is encodedwith 0 V. A mark (logic 1) is encodedwith altemating valuesof +Vand-t/, whichale nonnallyonly 507odutycyclepulses.Also refered to asa bipolarsignal. Alternate mark inversion violation. In AMI coding, the occunenceof two successive markswith the samepolarity.Also refenedto asa bipolarviolation. Amplitude distortion. Distortionof a transmissionsignalcausedby nonuniform passband asa functionof frequency. attenuation Amplitude modulation(AM). Modulationof the amplitudeof a carrierwavewith signal. the amplitudeof a baseband Antipodal signaling. Techniqueof encodingbinarysignalssothatthe symbolfor a I is theexactnegativeof the symbolfor a 0. Antipodal signalingprovidesoptimum ratio. in termsof the signal-to-noise errorperformance 613

614

GLoSSAFY

Asynchronousnetwork. A networkin which the clocksof the transmissionlinks andswitchingsystemsarenot synchronized to eachother. Asynchronoustransfer mode (ATM). packet-switched methodof time division multiplexing(labeledmultiplexing)utilizing fixed packet(cell) sizesof 53 bytes. Asynchronoustransmission.Mode of communication characterized by start-stop transmissionswith undefined time intervals between transmissions.Each transmission burstgenerallycontainsa singleword or byteof information. Automatic call distributor (ACD). switching sy$temusedto evenly distribute incoming calls to a number of stationson a first-come,first-servedbasis. Applicationsincludeoperatorassistance andairlinereservations. Automatic number identification (ANr). processof identifyingandforwardinga callingnumberto networkcall conffolservices. Automatic repeat request (ARo. Error correctionprocessinvolving the use of redundantcheckbits to detectcomrptedblocksof dataand triggerrequestsfor retransmission of the same. Availability. (l) With respectto switchingsystems:the numberof outletsavailable from a particularinlet.(2) with respectto equipmentin general:thepercentage of time theequipmentis providingacceptable service. Balancedcode. Line codewith equailyoccurringpositiveand negativeenergyro precludea dc componentin thefrequencyspectrum. Baseband.Literally, the frequencyband of an unmodulatedsignal. A baseband signalis an information-bearing signalthat is eithertransmitteaas is or usedto modulatea carier. Baud rate. unit of signalingspeed(symbolsper second).For binary signalingthe datarateis the sameasthe baudrate.For multileveltransmission the datarateis equalto thebaudratetimeslog2(L),whereL is thenumberof levels. Bipolar coding. ,SeeAltematemarkinversion. Bis. secondversion.Ter meansthree.For example,v.27bis and v.27ter are the secondandthird versionsof theV.27 standard, respectively. Bit leaking. Processof. convertingbyte-sizedtiming adjustmentsinto multiple bit-sizedor fractionalbit-sizedtiming adjustments. Blockedcallscleared(BCC). Servicedisciplinein whichunserviceable requests are rejectedby thesystemwithoutservice.Also calledlosrcallscleared(LCC). Blockedcalls held (BcH). servicedisciplinein which unserviceable requesmsray "in thesystem" withoutbeingservicedbut havinga portionof theirdesiredservice time elapseuntil servicebegins.Also calledlosrcallsheld (LCH). Blocking- Inabilityof a callingpartyto be connected to a calledparrybecause either all circuitsarebusyor thereis internal(matchingtoss)blockingwithin a switch matrix. Bridged tap. Extrapair of wiresconnectedin shuntto a maincablepair.The extra pairis normallyopencircuitedbut maybeusedat a futuretimeto connectthemain

GLOSSAFY 615

tapsdo not affectvoicefrequencysignalsbut pair to a newcustomer.Shor.t-bridged canbe extremelydetrimentalto higherftequencydigital signals. Building integratedtiming supply (BITS)- Mastertiming supplyfor all equipment in a building. Busy hour. The 60-minuteperiod of a day (and sometimesof a weet) in which the averageofferedtraffic load is highest. Call admissioncontrol (CAC). ATM controlprocessresponsiblefor negotiating quality-of-serviceparametersfor new connectionsdependenton the traffic loads of the networkat the time. Call congestion.Blockingprobabilityof a trunk group. of signalinga switchingsystemto reroutecallsintendedfor Call forwarding. Process a particularnumberto someothernumberon a temporarybasis. Call waiting. Processof insertinga shorttoneinto the voicepathof an activeuser informing the userthat anothercall is waiting to be connected. Catling number identiticafion (CM). Service whereby a telephonenetwork providestheidentityof a callingnumber(or person)with incomingringing' Carrier sensemultiple accesdcollisiondetection(CSMA/CD). Accessprotocol for a commonbus or radio link in which sourcesflustny to detectthe presenceof a carrier and then begin transmittingwhen the facility is idle. If anothersource begins transmitting during a period of time correspondingto the maximum transmissiondelay,acollisionis detectedandall actiyenodesceasetransmission for a randomarrount of time. technique Carrierlessamplituddphase(CAP) modutation. QAM implementation followed by waveforms quadrature signal (DSP) filtered generation of usingdirect signal. DSPsummationfor thecomposite as so manyhundredcall secondsper CCS. Measureof traffic intensityexpressed = I edang. CCS hour;36 that specifiesthe maximum Cell delay variation (CDV). An ATM QoSparameter delaythrougha network' rangeofcell propagation cell transfer delay (cTD). An ATM QoS parameterspecifying the average propagationtime for a cell to be transferredfrom its sourceto its destination. Central office. Public network switchingoffice (and particularlythe switching machinewithin it) to which subscriberlines areconnected.Also referredto as an end office or class5 office. In a more generalsen$e,any switch in the public networkis sometimesreferredto asa centraloffice. Centrex service. Privateswitchingserviceprovidedby a local exchangecarrierto an organizationutilizing hardwareand softwarepartitionsof a centraloffice switch. CEPT-|. Term usedto designatethe 2.048-Mbpsfirst-level digital signalof the fTU-T digital hierarchy.(More often referredto a$an EI signal') signalingchannelsfor each signaling(CAS). Use of $eparate Channel-associated may be separatefrom the channels individual signaling messagechannel.The

616

GLoSSARY

message channels(asin El) or derivedwithin themessage channels themselves (as in Tl). channel bank. Equipmentthat convertsmultiple analoginterfacesto and from a time-division-multiplexed digitalbit stream(24 analogio oneDSI or 30 analogto oneEl). Channelbanksarealsousedfor FDM applications. channel service unit (csu)- Equipmentinstalledon customerpremisesat the interfaceto phonecompanylines to terminatea DDS or Tl ciicuit to provide networkprotectionanddiagnosticcapabilities. Also calleda customerserviceunit. circuit Emulation service (cEs). An ATM servicein which constant-bit-rare virtualcircuitsuseAALI adaptation to emulateanend-to-endphysicalcircuit. Circuit switching. The principleof establishing an end-to-endconnecrionbetween usersof a network. The associatedfacilities are dedicatedto the particular connectionandheldfor thedurationof thecall. clear-channelcapability.Ability to providea transparenr 64-kbpschannelthrough a North Americannetwork(usingBSZSandcommon-channel signaling). clock. Time baseusedto controlthe transferof digitalinformation. C'messageweighing. Selectiveatlenuationof voicebandnoisein accordance wifh the subjectiveeffects as a function of frequency (noise weighting filter characteristic usedin NorthAmerica). codec(coder-decoder). Integrated circuit providing analog-to-digital and digital-to-analog conversionof voicesignals. coded mark lnversion (cMr). Two-levelline codein which a binary 0 is coded with a positivelevel and an equal-magnitude negativelevel for half a unit time intervaleach.A binaryI is codedwith a full-periodpositiveor negativelevelin an alternatingmannerto maintaindc balance. coherent demodulation. Demodulation using a carier reference that is synchronized in frequencyandphaseto thecarrierusedin themodulationprocess. common-channelsignaling. use of a separatechanneldedicatedto tran$portof signalinginformationpertainingto a multiplenumberof message channels. community dial office (cDo). small, normallyunanended switchingsystemthatis usedin smallcommunitiesandis controlledfrom a rargercentraro}fice. companding. Processof compressing a signalat the sourceandexpandingit at the destinationto maintain a given end-to-enddynamic range whiie reducing the dynamicrangebetweenthe compressor-expander. concentration. Processof switching somenumber of lighfly used channels or sourcesontoa smallernumberof moreheavilyusedchannels. Conferencecall. Call in which threeor morestationsareinterconnected in a manner that allows all parriesto talk andbe heardby all otherparties. connection admissioncontrol (cAc). An ATM functionthatdetermines whether a vifiual circuit (vc) connectionrequestshouldbe accepted or rejected. constant (or continuous)bit rate (cBR). An ATM classof servicethat supports transmission of a continuousdatarate.

GLOSSARY 617

Constantenvelope.A (frequencyor phase)modulatedcarrierwith no the carrieramplitude. Controlled stip. Slip involving the repetitionor deletionof an integralnumberof TDM frames. Crossbar switch. Rectangulararray of crosspointsarrangedso that an input connectedto a row of crosspointinput$ can be switchedto any output to a columnof crosspointoutputs. connected Crosstalk. Unwantedsignaltransferfrom onecircuitinto another' Custom Local Area Signaling Services(CLASS). Signalingservicefor analog call' linesprovidedbetweenthefirst two ringsof anincomingtelephone telephone Principalfeaturesenabledby CLASS servicesare Caller ID, Call Return,Call Call Forwarding,PriorityRinging,andRepeatDialing' Preferued Screening, (from thepoint of view of a packetnetwork). Datagram. Singlepacketmessage telephone DataphoneDigital Service(DDS). Serviceoffering of common-carder 1544 kbps. 4.8, or at 9.6, 56, 2.4, providing channels digital companies dBm. Powerlevel in decibelsrelativeto I mW' dBrnC. Power level of noise with C-messageweighting expressedin decibels noisepoweris -90 dBm = 10-l?W. relativeto referencenoise.Reference dBrnC0. Noise power measuredin dBmC but referencedto the eero-level level point. transmission Decisionfeedbackequalization(DF'E).Equalizationtechniqueinvolvingthe useof in a receiverthat are deftcteddatavaluesto producesimulatedchannelresponses to cancel channel signal Samples received from subsequent then subtracted distortion. Deltamodulation.Techniquefor digitallyencodingananalogsignalby continuously measuringandtransmittingonly thepolarityof thesignalslope.Theanalogsignal of theinput' asa staircase approximation is reconstructed (offered traffic not includingretries.) Demandtraffic. First attempttraffic. Desynchronizer. Circuitry that extractsa tributary digital signalfrom a composite, higherlevel signalandderivesthe sourceclock frequencyof thetributary. Digitat advanced mobile phone service (D-AMPS), Digital mobile telephone standardof North America,alsoknownastS-I36 TDMA. Digital circuit muttiplication {DCM). Processof increasingthe numberof voice compression link throughtheuseof speech signalscarriedon a digitaltransmission or silenceremovalor both. multipledigitalsignals Digitsl loop carrier (DLC). A TDM systemfor transporting lines. to andfrom multiplesubscriber changes PCM samplevaluesfor the putposeof pad. that Digitat circuit Digitat theencodedanalogsignal. (attenuating) power level of the adjusting by satellite(DBS)systemdeveloped Digital satetlitesystem(DSS). Directbroadcast HughesElectronicsCorporation.

618

GLOSSAHY

Digital speechinterpolation (DSI). Dieital equivalenrof rASI whereindigital speechsignalsaremonitoredandconnected to a hansmissionchannelonly when voiceactivityis present(a form of DCM). Digital video broadcasting(DvB) group. Europeandigitat rv standardfor TV, audio,anddata.DvB canbe broadcast via satelliteor terrestrialsystems. Direct broadcastsatellite(DBS)- One-wayTV broadcast servicefrom a sarelliteto a small l8-in. dishantenna. Direct inward dialing (D.rD). processof a publictelephone networkprovidingpBX extensionnumberswith incomingcalls. Double-sidebandmodulation. Modulationtechniquein which a basebandsignal with no dc energydirectly moduratesa carrier to produceboth upper and lower sidebands but no carrierenergy. Dual'tone multifrequency (DTMF) signaling. Generic name for pu$h-button telephonesignalingequivalenrto theBell Sysrem'sTOUCH_TONE. Dynamic rflnge. Rangeof power levels(minimumto maximum)achievabre by a signalor specifiedfor equipmentoperation. E&M lead signaling. Interfacebetweena switchingsystemand a transmission systemutilizingpairsofwires for signalingthatareseparate from thevoicepairs. outgoing controlsignalsfrom eachfacility arecardedby respectiveM-leads to E-leadsof theotherfacility. Echo canceler.Devicethatremovestalkerechoin the returnbranchof a four-wire circuit by subtractinga delayedversionof the signalhansmittedin the forward path. Echosuppressor.Devicethatis activatedby voiceenergyin onepathof a four-wire circuit to inserta high amountof signallossin the otherpathior the purposeof blockinganecho. Echo. Reflectedand delayedsignal.common sourcesof echo in the telephone networkareelectricalreflectionsat four-wireto two-wireconversions andacoustic reflectionswith speakerphones. Edge switch. An ATM switch providing gareway interfaces to non-ATM communication links. Elastic store. First-in,first-out databuffer that acceptsdataunderconftol of one clockandoutputsdataundercontrolof anotherclock. Electronic automatic exchange(EAx). Designationof stored-program control switchingmachinesdevelopedby cenerarrelephoneandErectronic*s. Electronic switching system (ESs). Designation of stored-program conhol switchingmachinesdeveloped by AT&T. End office.class5 switchingoffice.Arsoreferredto asa centraloffice. Envelopedelay. Derivativeof channelphaseresponsewith respectto frequency. Ideally, the phaseresponseshould be linear, indicating ttrat att frequencies are delayedequally.

GLOSSAHY 619

Envelopedetector. Methodof detectingamplitudemodulationby trackingonly the peakvaluesof a carrierwave. for transmissiondistortionswith fixed or Equalization. Practiceof compensating adaptivecircuitry. Erlang. Measureof traffic intensity.Basically,a measureof the utilization of a resource(e.g.,the averagenumberof busycircuitsin a kunk group,or theratio of time an individualcircuitis busy). Error second.A l-sec intervalcontainingat leastI bit error. purposesthat Exchangearea. Contiguousareaof servicedefinedfor administrative typicatlycomprisesanenriretownor city andincludestheimmediatecountryside and suburbs.An exchangeareamay have one end office or many end offices interconnected by trunksandtandemoffices. Expansion. Switchingof a numberof inputchannelsontoa largernumberof output channels. Extended superframe format (ESF). A Tl framing format that embedsdiagnostic capabilitiesanda facility datalink into the8-kbpsDSI framingchannel. link failuredefinedas 10 consecutive Failed seconds(FS). An (ESF)transmission severelyerroredseconds. Far-end crosstalk(FEXT). Unwantedenergycoupledfrom onechannelor circuit into anothercircuit andappearingat the far endof the transmissionlink. Fiber distributed data interface (FDDI). High-bandwidthfiber transmission local areanetworks. systemfor interconnecting FIFO. First-in,first-outservicedisciplinefor a queue. Flow control (in data network). Procedurefor controlling the rate of transferof packetsfrom onenodeto another. Foreignexchangecircuit (FX). Extensionof servicefrom oneswitchingofficeto a subscriber normallyservicedby a differentswitchingoffice, Forward error correction (FEC). Error correction techniquewith sufficient into thesignalfor correctingerrorswithoutretransmission. redundancy embedded Four-wire circuit- Circuit using two separatechannelsfor each direction of transmission.When wireline transmissionis involved, each direction of pair of wires. is providedby a separate transmission Fractionalspeechloss. Fractionof speechthatSetsclippedin a TASI or DSI system all channelsarein usewhentalk spurtsbegin. because Frame aliSnment. Identificationof thebeginningandendof a TDM frameusinga framing daiapattern, Frame. Set of consecutivetime slots in a TDM format in which the positionof individualtime slotscanbe identifiedby referenceto a framealignmentsignal' Frequencydiversity. In radio systemsthe useOfoneor moreback'uptransmitters, channels,andreceiversto protectagainstatmospheric(multipath)fading-

620

cLossAny

Frequencydivisionmuttiplex (FDM). Partitioningthebandwidthof a rransmission link into separatechannelsof lesserbandwidttrinto which separatesignals are insertedandextracted. Frequencymodulation (FM). Modurationprocessthat variesthe frequency of a carriersignalin accordance with a baseband informationsignal. n'requencyshift keying ff'sK). Digital modutationtechniquein which data are represented by discretefrequencies. n'ricative. speechsoundproducedwith continuousair flow through one or more vocaltractrestrictionsto causeair turbulence(friction).Examplefricativesounds ares,f, t, x, orj, Full'duplex. Transmissionin two directionssimultaneously(also referred to as simplya duplexoperation). Gateway network element (GNE). A soNET node with an interface to unsynchronized tributarysignals. Gaussiannoise. Backgroundnoisewith a uniformfrequencyspectrumacross a band of interestand havingamplitudesamplevaluesthit foliow a normal(gaussian) probabilitydistribution. Glare. simultaneousseizureof both endsof a two-way trunk by two switching machinestrying to setup two separate connections. Global positioning system(Gps). Global satellitenavigationsystem commonry usedby telecommunications sy$temsas a time and friquency synchronization source. Half'duplex transmission.Transmission in bothdirectionsbut only in onedirection at a time. HDB3. Modified AMI (bipolar)line codein which stringsof four 0,s areencoded with an AMI violationin thelastbit. Head'of-line hlocking. Blocking of cells or packetsin a FIFO queue becausethe nextenffrycannotbe servicedwhile otherentriesin thequeuecouldbe serviced. Highway. A TDM pathinsidea digital (TDM) switchingmachine. Hook flash. Momentarydepressionof the switchhookto alert afl operatoror equipment,but not solong asto be interpretedasa disconnect. Hot standby. Redundantequipmentkept in an operationalmode as backupfor primaryequipment. usually,automaticswitchingiothestandbyequipmentoccurs whentheprimaryequipmentfails. Huffman coding. processof assigningvariabreJengthdigitar codewords to a messageset whereinthe length of a codewordis inverselyproportional to the probabilityof occurrenceof respectivemessages. Thus a freiuentty hansmitted message requiresfewerbits thanthe averagerengthof all messages.' Hybrid fiber coax(HFc). cable terevisiondistributionsysteminvolving rheuseof fiber in the feederportion of the networkfrom a head-endto fiber nodes followed by coaxdrop$to residences in neighborhoods.

GLossAHY 621 Hybrid. Device used to connecta two-wire, bidirectionalcircuit to a four-wire circuit. Idle channelnoise. Noiseoccurringduringsilenceintervalsof speech' link when transmission insertedinto a synchronous Idle character. Controlcharacter thereis no informationto be sent' Impulse noise. Short-durationspike of noise that is much larger than randomly noise. disnibuted(Gaussian) Inband signaling. Signalingtran$mittedwithin the samechanneland band of toaffic. frequencies usedfor message (IDN). Network in which digital TDM transmission Integrated digital network links aredirectlyinterfacedinto digitalTDM switchingmatrices. Integrated servicesdigitat network (ISDN). Integrateddigital network in which ofa signalingchannelandone interfacesareprovidedwith directaccess subscriber channels. or moredigital64-kbps Intercept. Processof diverting a caII from an intendedstation to an operatoror recordedannouncement. International Telecommunications Union (ITU). International standards theCCITT in Marchof 1993(renamedITU-T)that subsumed organization Internet. Connectedsetof networksusingTCP/IP suite. Intersymbol interterence (ISD. Interferencein a transmissionsystemcausedby a symbolin onesignalinginternalbeingspreadout andoverlappingthe sampletime of a symbolin anothersignalinterval. Jitter. Short-termvariationsof the significantinstant$of a digital signalfrom their ideal positionsin time. Short-termvariationsare often defined to correspondto above10He. frequencies Joint photographic experts group (JPEG). ISO standardsgroup that defineshow to compressstill pictures. Justifrcation (pulsestuffing). hocess of adaptingtherateof digital signalby adding non-information-carryingbits at a sourceinterfaceandextractingthe samebits at a destinationinterface. rateusedto synchronize Justificationratio. Ratioof thejustifrcation(pulse-stuffing) a hibutarysignalto the maximumpossiblejustificationrateallowedby a higher level multiplex format. (A justification ratio of { implies one-third of the rate adjustrnentopportunitiesare null bits and two-thirds of the opportunitiesare fributarydatabits.) Key system. Small,customerpremisestelephoneswitchingsystemthatallowsusers to directly selecttelephonelines (andreceivedial tonefrom a cenffaloffice). LAN Emulation. An ATM serviceoffering that emulatesEthernetor token ring LANs. Line administration. SeeLoad balancing.

622

GLossARy

Line code. Setof electrical(or optical)waveformschosento represent datafor the purposeof transmission. Load balancing. Adjustingthe assignment of very activecustomerlinesso that all groupsof customerlines in a multistageswitchreceiveapproximately the same amountof traffic. Loading coils. Lumped-element inductorsinsertedat periodicpointsin cablepairs to flattenouttheirvoicefrequencyresponse. Althoughloadingcoilsimprovevoice frequencyffansmission, theyseverelyattenuate higherfrequJncies asrequiredfor digitaltransmission. Local accessand transport area (LATA). serviceareaof a local exchange carrier (localoperatingcompany). Longitudinal current. Electricalcurrentpassingalonga pair of wiresin a common direction.ContrastMetalliccurrent. Loop start. Off-hook signalingprotocolinvolving theflow of dc current betweenthe tip andthering of a subscriber loop. Loop timing. Synchronizingthe hansmit timing of a bidirectional link to the receivedtiming of the samelink. Lost calls cleared(alsolosssystem).Modeof operationin whichblocked callsare rejectedby the networkandmay or may not return. Lost catls held. Mode of operationthat holdsblockedcall requesrs until a channel becomesavailable.Theportionof a call thatgetsblockedis lost. Mfil networkmultiprexingequipmentusingbit stuffing {ltiplexer. Asynchronous for 28 individual1.544-Mbps Dsl signalsto createa1/..ii6-lvrbpsns3 signal. Main distrlbuting frame (MDF). Frameworkusedto crossconnect oursideplant cablepairs to centraloffice switching equipment.The MDF providesprotection andtestaccessto theoutsideplantcablepairs. Managementlnformation base(MrB), coilectionof softwaredefined objectsthat canbe accessed via a networkrnanagement protocol(e.g.,sNMp or TMN). Master clock. Particularlyaccurateand stablefrequencysourcefrom which various nodesin a networkextracttheir operatingfrequency(clock). Master frame. set of consecutiveTDM framesthat areidentifiedby a masterframe alignmentsignal(MAS). Matching loss. Blockingwithin a mulristageswitchmahix resulting from at least onelink in all suirablepathsthroughthematrixbeingbusy. Maximum time interval error (MTrE). Largesttime interval error occurring in all possiblemeasurement intervalsof lengths within a measurement periodT. Messageswitching. practiceof transportingcompletemessages from a $ourceto a destination in non-real time and without interaction between source and destination, usua.llyin a store-and-forward fashion. Messageunit accounting. Activity-basedbilling as opposedto flat-rate billing, whichis independent ofusage.

GLOSSARY 623

Metallic current. Currentflowing in oppositedirectionsin a wire pair asa resultof a differencesignalpropagatingalongthe pair. ContrastLongitudinalcurrent' Minimum shift keying (MSK). Digital frequencyshift modulation wherein the magnitudeof a frequencyshift is theminimum amountrequiredto producea 180" of phaseshift in a symbolinterval. Modem. Contractionof the termsmodulationand demodulation.A deviceusedto overtelephonelines.A modem generate"voicelike"datasignalsfor transmission "data Bell terminology. in older System set" is referredto asa Multifrequency (MF) signaling, Signalingmethodusedfor interofficeapplications' MF signalingusestwo of six possibletonesto encode10 digits andfive special auxiliarysignals. phenomenon in whicha singletransmitsignalfollowstwo Multipath. Transmission or moreseparatepathsto a receiverwith differing delays. Muttiplexing. Processof combiningmultiple signalsinto a compositesignalfor overcoilrmonfacilities. transmission a networkin which Mutually synchronizednetwork. Techniquefor synchronizing all nodes derive their operating frequency as the average of their own of somenumberof othernodes free-rUnningclock frequencyandthefrequencies in thenetwork. Near-endcrosstalk(NEXT). Unwantedenergycoupledfrom a ffansmitterin one circuitinto a receiverof anothercircuit at the samelocation(nearend). Network element (NE). Internal node of a SONET network whoseintetfacesare signalsof theSONEThierarchy. Network management.Network managementis the function of supervisinga communicationsnetworkto ensuremaximumutilization of the networkunderall conditions.Supervisionrequiresmonitoring,measuring,and, when necessary, actionto control the flow of traffic. Next generationdigital loop carrier (NGDLC). A DLC systemcapableof using links andprovidingbasiccomplianceto theBellcoreGR 303 SONETtransmission standard. Nonblocking. Switchingnetworkthat alwayshasa freepathfrom anyidle incoming trunk or line to any idle outgoingkunk or line. Nonreturn to zero (NRZ). Line code that switches directly from one level to another.Eachlevel is heldfor the durationof a signalinterval. Nyqulst rate. (1) Minimum samplingraterequiredto extractall informationin an analogwaveform.The Nyquist rate is equalto twice the bandwidthof the signal channelwith a bandwidthB, the maximumrate beingsampled.(2) In a baseband (28) that pulsescanbe transmittedwithout intersymbolinterference. Off hook. Stateor conditionthat a telephonereceiveris requestingseryiceor in use. Also a supervisorysignalto indicateacfivestatusof a telephoneor line. Offered traffic. Amount of traffic carried by a systemassumingthe systemhas inJinitecapacityandthereforedoesnot block or delayanycalls.

624

clossAny

On hook. Inactivestatusof a telephoneor line. One-waytrunk. Trunk circuitthatcanbe seizedat only oneend. Open systemsinterconnection(OSI). Communications protocolreferencemodel introducedby ISo consistingof sevenlayers(physical,link, network,transport, presentation, session, application). out'of-band signaling. signaling techniquethar u$esthe samepath as message traffic but a portion of the channelbandwidthaboveor below that usedfor voice. Pair-gain system. subscriber transmissionsystem that serves a number of subscribers with a smallernumberof wire pairsusingconcentration, multiplexing, or both. Parity. Processof addinga redundantbit to a group of informationbits to maintain eitherodd or evennumbersof l's in the compositegroupof bits. A parity error resultsif an oddnumberof l's is detectedwhenevenparityis transmittedor vice versa. Partial-responsesignallng (pRS). use of controlledintersymbolinterferenceto increasethesignalingratein a givenbandwidth. Permanentvirtual circuit (pvc). virtual circuit(x.25), virtualconnection(Frame Relay), or vinual channel connection(ATM) that emulatesa leased-line connection, Personalcommunicstions system(pcs). Digiral mobile cellular telephonesysrem operatingin l9fi)-MHz frequencybandsin North America(1g50-1910MHz for mobileto basestarion1930-lgg0 MHz for basesrarionto mobile). Per-trunk signaling. Method of signaling in which rhe conrrol information pertainingto a pafricularcall is transmittedover the samecircuit (channel) that carriesthe call. Phasedistortion. Signaldistortionresultingfrom nonuniformdelayoffrequencies within thepassband. Phasereversalkeying(PRK). specialcaseof phaseshiftkeyinginvolvingonly two phasesl80o apart. Phaseshift keying (PSK). Formof digitalmodulationthatusesZndistinctphasesto represent n bits ofinformationin eachsignalinterval. Plesiochronous digital hierarchy (pDH), Designation of the original digital multiplexing hierarchy utilizing free-running clocks at all leiels of the multiplexinghierarchy. Pleslochronous.Method of network synchronizationinvolving the use of independent (unsynchronized but highly accurate) clocksat thesamenominalrate. Pointer burst. In soNET, the occurrenceof multiple pointerjustificationevents (PJEs)of onepolaritywithin the time constantof a desynchronizer circuit. Pointer justiflcation event (pJE). changein a pointervalueto accommodate phase drift.

GLOSSARY

625

Polar signaling. Two-level line coding for binary datausing balanced(symmetric) positiveandnegativelevels. Powerspectraldensity. Distributionof signalpowerasa functionof frequency. Primary referencesource (PRS). Top level (Stratum1) clock sourcewith an of+l v lfl-tt. accuracy Priority ringing. Featurefor incomingcall discriminationin which a telephonerings with a specialringingpatternfor callsfrom a selectsetof phonenumbers, Private branch exchange(PBX). Switchingequipmentused by a companyor organization to providein-houseswitchingandaccessto thepublicnetwork. taffic from onefransmission Protection switching. Practiceof transferringmessage link whenthe formeris degradedin somefashionor link to a spafetransmission for maintenancete$tsor equipmentupgrades. Provisioning. Allocatingor deallocatingfacilitiesin suppotlof a particularservice offering(tuminga seryiceon or off;. by CCIfi. filter recommended Psophometricweighting. Noise-weighting a continuousanalog Pulseamplitude modulation (PAM). hocess of representing of thesamplesare Theamplitudes samples. waveformwitha seriesof discrete-time analog in nature. and therefore continuous Pulsecodemodulation (PCM). Processof convertingPAM samplesinto discrete by digitalcodewords. levelsrepresented Pulsestuffing. SeeJustification. signalingon each QPRS. Quadraturechannelmodulafionusing partial-response channel. Quadrature amplitude modulation (QAM). Independentamplitudemodulationof two orthogonalchannelsusingthe samecarrierfrequency. Quantlzation noise. Differencebetweenthe discretesamplevaluerepresentedby a digital codeandthe original analogsamplevalue. channelwith a particulartypeof pulse channel. Digital transmission Raised-cosine responsethat producesno intersymbolinterferenceat the sampletimesof adjacent "raisedcosine"is derivedfrom theform ofthe signalingintervals.Thedesignation analyticalfrequencyspectrum(l + cosineor cosinesquared). Recirculation. Passinga cell backto an input stageof a multiple-stageATM switch blockingin an it encountered whenthecell wouldotherwisebe discardedbecause outputstage. Reframing time. Amount of time it takes to detect and synchronizeto a frame alignmentsignal. Regeneration. Process of recognizing and restoring a digital signal before perturbafionsof the signalaccumulateto the point that bit elrors occur' Regenerativerepeater. Device usedto detect,amplify, reshaPe,and retransmita digital bit stream.

626

GLossAFty

Return loss. Differencein decibelsbetweenreflectedandincidentenergyat a signal reflectionpoint. Ringback. Signalingtonereturnedby switchingequipmentto a callerindicatingthat a calledpartytelephoneis beingalerted(ringing). Ringing voltage. Low-frequencyac signalusedto activatetheringerof a telephone (typically,a20-Hzsignalat 90 V rms). Robbedbit signaling.NorthAmericanpracticeof usingtheleastsignifrcantpCM bit of everysixthframefor channel-associated signaling. Router. Store-and-forward packet-switching elementthat processes packetheaders anddetermines an appropriate outputrink for eachpacketusingroutingtabresand routingprotocols. severelyerrored seconds(sES). An (ESF)transmission link performance measure definedastheoccurrence of 320or moreerroredseconds. Sidetone.Portionof a talker'ssignalthatis purposelyfedbackto theearpiecesothat thetalkerhearshis or her own speech. signaling. Exchangeof electricalinformation(otherrhanby speech)specifically concemedwith the establishment andcontrolof connections andmanasement in a communication network. simple network managementprotocol (sNMp). standardprotocolfor perforrning networkmanagement functions(createdby theIntemetcommunity). Simplextransmission.Mode of operationinvolving transmission in onedirection only. singing. Audibleoscillationof a telephonecircuitcausedby a netamountof gainin a four-wire$egment of thecircuit. Singlemodefiber. Fiberthatis narrowenoughto precludeffansmission of anoptical signalalongmorethanonepath. single-frequency(sF') signating. Methodof conveyingdial-pulseandsupervision signalsfrom oneendof a trunk or line to the otherusingthepresence or absence of a singlespecifiedfrequency.A 2600-Hztoneis commonlyused. slip. Deletionor repetitionof datain a datasffeamcausedby an over{lowor an underflow of an elastic store due to variations in the write and read rates. A controlledslip is a deletionor repetitionof an entireframeof datasotheposition of theframingbits is undisturbed. Slopeoverload.Saturation oftherangeofdifference signalsinadifferential encoder causedby a signalwith a large,rapidchangein amplirude, soNET island. collection of one or more soNET GNEs and NEs that createa continuousSONETpathfor a digital signal. spacediversity. In radiosystems, theuseof two receivingantennas andpossiblytwo separatereceiversto provide protection against atmosphere-induced signal attenuation (fading).

GLOSSARY 62f

Spaceexpansion. Use of more centerstagesin a multistageswitch than there are array. inputsto a first-stagearrayor outputsfrom a last-stage T1 line from endto endbut not includingchannelbanks. Spanline. Repeatered Speeddiating. Use of a short addresscode to representan often*calledtelephone number.The commoncontrolcomputerin a PBX or endoffice providingsucha seryicehanslatesthe shortcodeinto the desirednumber. Spread spectrum. Processof distributing an information-bearingsignal acrossa bandwidththatis muchwiderthantheinherentbandwidthof thesignal. Star network. Networkwith a singlenodeto which all othernodesareconnected' Statisticalmultiplexor ($tat-mux). Multiplexerthat assignskansmissioncapacity to a tributary only when the source of the tributary is actively transmitting information. switchingsystemusingstep-by-step Step-by-step(SXS)switch. Elecffomechanical up aseachdigit is dialed. progressively a connection set switchelementsto switching. Stored-programcontrol (SPC). Computer-controlled Stratum clock. Definition of free-runaccuracyof a clock dependingon stratumlevel network. in synchronization Stuff Rstio. SeeJustificationratio. or to control Supervisory signal. Signal used to indicate the statusof a equipmenton theline. wide-area Switchedmultimegabit data service(SMDS). Public,packet-switched serviceofferedby commoncarriers. datacommunications Switchhook. Hook or buttonsuponwhich a telephonehandsetrestswhenit is not beingused. Symbolrate. SeeBaudrate. Synchronizer.Circuitry that insertsa nibutary digital signal into a higher level multiplexedsignalandperformsjustification(pulsestuffing)in theproce$$. Synchronousnetwork. Network in which the clocks of separatenodesoperateat identicalrates. in which discretesignal Synchronoustransmission.Mode of digital transmission elements(symbols)aretransmittedat a fixed andcontinuousrate. Synchronous. Mode of operationin which two or more pieces of equipmentof networkperformtheir operation$at preci$elythesameratefor anindefiniteamount of time. The rateof operationis derivedby distributionof a timing reference. processthatintegrates networklayer(layer3) Tag switching. Fastpacket-switching (layer 2) switching to simplify addressand protocol routing and data link layer processing. Talker echo. Ponion of a talker's voicethat is retumedto thetalker andheardby the talker. The amountof annoyanceto a talker is dependenton both the amountof delayandthe amplitudeof thereturningsignal.

628

cLos$ARy

Tandemoflice. In general,any intermediateswitchusedto establisha connection. In specificterminology,a tandemofficeis a switchusedto interconnect endoffices in an exchangearea. TASI. Time assignment speechinterpolation,the practiceof concentrating a group of voicesignalsontoa sma]lergroupof channelsby dynamicallyswitchingactivl voicesignalsto idle channels. Ternary coding. use of all statesof a three-levelcodeto sendmorethan I bit of informationin a singlesymbol.This is in contrastto bipolarcoding,which uses three levels, but only one of two in any particularinterval. one methodof interfacingbinary datato a ternaryline codeis to encode4 bits with threeternarv symbols(483T). Tie line. Dedicatedcircuitconnectingtwo privatebranchexchanges. Time compression multiplex. Transmitting in two directions on a single transmissionlink using altematingdirections of transmission("ping-pong" transmission). Time congestion.Ratio of time that all facilitiesof a sy$temarebusy(congesred). Time congestionrefersto the statusof the systemanddoesnot necessarily imply thatblockingoccurs. Time deviation (TDEV). Squareroot of thetime variance(TVAR). Time division multiplexing (TDM). sharinga rransmission link amongmultiple usersby assigningtime intervalsto individualusersduringwhich they havethe entirebandwidthof a system. Time expansion.Useof moretime slotson internallinks of a switchthanexist on externallinks. Time interval error (TrE). variation in time delay of a given signalrelative to an idealtiming signalovera particulartime period. Time variance (TVAR). Measureof the expectedtime variationof a signalas a functionof time separation. Traffic carried. Amount of raffic canied (by a group of circuits or a group of switches)duringany periodis the sum of the holding timesexpressed in hours

(cclrr).

Trafric engineering.Networkmanagement activitythat analyzesaverageandpeak traffic loadsto determinehow to designor reconfigurea networkto maximizethe traffic capacity of the network and to ensurean acceptablevalue of blocking probability. Traflic flow. Amount of traffic (in erlangs or ccs) carried or offered to a transmission link or switchingsy$tem. Transhybrid loss. Amountof isolation(in decibels)betweengo andrerurnparhson thefour-wiresideof a four-wireto two-wirehybrid. Transmissionlevelpoint (TLP). specification,in decibels,of thesignal tr)owerara point in a transmissionsystemrelative to the power of the samesignal at a (hypothetical) zerotransmission levelpoint (0_TLp).

GLOSSARY 629

Transmultiplexer(transmux). EquipmentthatconveflsFDM voicesignalsto TDM voicesignalsandvice versa. Transversal equalieer. Time-domainequalizerutilizing a tappeddelay line and weightingcoefficientsat eachof the tapsto removeintersymbolinterference. indicating connectionsetupmessages Traveling classmark. Codethataccompanies may provisions that be desired' thenatureof the servicerequestandany special Trunk. Circuitor channelbetweentwo switchingsystems. Two-wny trunk. Trunk circuitthatcanbe seizedat eitherendof thecircuit. Two-wire circuit. Circuit consistingof a single pair of wires and capableof carryingtwo signalsin oppositedirections. simultaneously Binary line codeusingsinglepolaritypulsesandzerovoltagefor thetwo Unipolar. codinglevels. Unit interval (UI). Nominaldifferencein timebetweenpulsepositionsat a specif,ied datarate. Unspecifiedbit rate (UBR). Classof serviceof ATM without specificquality "besteffort" service. assurances sometimesreferredto as User datagram protocol (UDP). An Internet protocol without network delivery (I-ost or erroredpacketsarenot retransmitted.) guarantees. aslong as Variablebit rate(VBR). Classof serviceof ATM with qualityassurances averageandpeakdatarates' the sourceadheresto preestablished Vector quantization. Choosing one codeword representativeof one discrete waveformof a set of discretewaveformssuchthat the selectedwaveformhasthe bestmatchto a segmentof thesignalbeingencoded. Vestigiat sideband transmission.Form of single-sidebandtransmissionthat includesa vestigeof the deletedsidebandanda small amountof carrierenergynetworkthroughwhich all packets Virtual circuit. Paththrougha packet-switched associated with a particular"connection"flow. Waiting time jitter. Timing jitter causedby waiting for a timing adjustment opportunitybeyondthe time whenthe needfor an adjustmentarises' Wander. Long-termvariationsof thesignif,rcantinstantsof a digital signalfrom their idealpositionsin time,wherelongtermimpliesphaseoscillationsof frequencyless point (typically l0 Hz) that is specifiedfor each thanor equalto a demarcation interfacerate. to Wavelengthdivisionmultiplexing (WDM). Useof multipleopticalwavelengths signals. optical multiple carry White noise. Noisewith a flat frequencyspectrumacrossa bandof interestin which samplestakenat twicethebandwidthor lower havezerocorrelation. services(WATS). Servicethatpermitscustomers Wide area telecommunications calls to make(OUTWATS)or receive(INWATS or 800 Service)long-distance andto havethembilled on a bulk basisratherthanindividually. packet-switched networks. X.25. Internationalprotocolstandardfor accessing

PROBLEMS TOSELECTED ANSWERS 1.1 2000pWC, 2(10)-6mWC 1.3 500pwC 1.5 250pW at -3 dB TLP 3.r l kHz,2 kHz, 3 kHz 3.2 2.43dB degradation 3.4 49.4Mbps 3.6 38.5dB 3.8 0/110/0110 3.10 23.5dB 3.14 (a) 13;(b) 33 (c) Quantization Sample Noise 30.2 .2

123.2 -2336.4 8080.9

-.8 .4 .9

NoisePower .04 .64 .16 .81

SQR(dB\ 43.6 43.7 75.3 79.1

(b) 1.76dB 3.17 (a)3 dB by 20.6dB 3.19 SQRincreases 3,21 0/l I 0/0001,0/I 10/000l, 1/110/0001, and I /110/000I (noninverted) 4.1 37.5bits 4.3 1.76dB 4.5 (a) Averageframetime = .097seconds. (b) Maximumaveragereframetime = 0.193seconds. 4.7 141bits 4.9 94 ms (b) 3.78dB 4.ll (a) 1.99dB (c) 18 (b) 19 4.15 (a) 81 631

632

ANSwERSToSELEcTEDpHoBLEMS

4.17 +l (a) (b) (c)

l+D l-D l-D2

-3 |, -4

+l

-1

a

0

+4 0

a

+2

+ 2 + 4 +2

5.1 5.2 s.3 5.5 s.7 5.9

250 (a)621 (a).11 (b) .2s (a)55,296 (a)4l (b) 10,828 (a) Total numberof memorybits = 48,0fi). (b) Complexity= 1504. 5.11 900bits (a) .5 dB degradation NewBERis l(10)-8 1600Hz (a) lTVoof peaksignalpower. (b) 15.3dB reducederrordistance. (a) 6.7 7.14dB advantage for 32-QAM. (b) 4.78dB advantage for 32-QAM. 6.9 1l.2 dB degradation in eruordistance 6.11 1.5(161*r 6.13 3 d B 6.1 6.2 6.4 6.5

7.1 195Bits if thereceiverknowsthenominaldatarate. 370Bits to accommodate peak+o-peak shifts. 7.3 +1.3UI with 99%probability 7.5 5.37(10)-5misframeVsec (onceevery5.2hoursT 7.7 ?5.6dB relativeto I UP 7.9 l.ZVo 7.11 (a) TIE = +4.736g1 (b) MTIE = 44.7?i61JI. 8.1 8-3 8.5 8.1 8.9

.188dB/km l6 km Al" at 1300nm is 5 timeslargerthanAl, ar 1550nm. (a)2m for mBlP (b) m + I for mBlC (a) MinimumDSl rate= 1.542Mbps. (b) MaximumDSI rate- 1.546Mbps. 8.11 (a)Minimumrate= 44.712Mbps. (b) Maximumrate= 44.784Mbps. 8.13 68.7psec

+ 0 +4

6 -

0 6 -6

ANSWERSTOSELEoTEDPHOBLEMS 633

9.1 9.3 9.5 9,7

6500kbps. I1.4kbps ChannelI output= +7; channel2 output= +9 SIR= 12dB

10.1 (a) 2.5kbits/message (b) 100bits/message 10.3 Totalnumberof bits is 86 bits. f0-5 (a) 0.aa5 by 30 milliseconds ft) delayis increased 11,f 1250meters 11.3 56 to 62.67kbps (e) .453 l2.l (a)40 calls/hour (b) .989 (c)9.5Vo (d) .19E 12.3 t.4E 12,5 (a) 15ports(14 portswith finite sourceanalysis) (b) B = 97o 12.7 (a)2.2Vo (b) rEo 12.8 (a) B = 307a (b) add6 channels 12.10 (a) At *.it = 32 E. (b) 46 circuitsrequiredfor B < .57o. 12.13 (a)B = lOVo (b) B = .787o 12,14 2 WATS lines 12.16(a)83.3Vo (b).833sec (c)30.6Vo (d).833

INDEX Abandoned calls,48 Abbreviateddialing, 16, 17 Accesstandem,12 Adaptivetransmitterpowerconhol(ATFC),323 Advancedmobilephoneservice(AMPS),53,70, 437-454 AdvancedResearch ProjectsAgency(ARPA/ ARPANET),456,468,473 Aliasing,95,96 pagiflg,444 Alphanumeric AmericanNationalStandards Institute{ANSI),4, 22t,f,81 Analogbridge,86 Analoginterfaces,47,62,63,87 ,132,2'10 Analogradio,6,63-65,85, 189,284,303,320, 383 Asynchronous transfermode(ATM), 3, 154,204, 208,310,331,,159*504,540,564-568 adaptionlayers(AALs),484-493,566,568 availablebit rate,475,476 calladmissioncontrol,479,489 cell discarding,489,566 cellloss,480,489 cell packing,487 cell transferdelay,476 circuitemulation,485,487 constantbit rste,415,484 qualityof service(QoS),476,484 switching,477,484,564 synchronous rcsidualtime stamps,485 taffrc shaping,489 unspecified bit rate,475, 416 variablebitrate, 475,476,484,487 virtualpathconnections, 477 Automaticcalldistributor(ACD),76,226,555,570 Automaticgain conbol (AGC), 120,506 Automaticnumberidentification(AM), 12

Automaticprotectionswitching,210,322,411 Automaticrepeatrequest(ARQ), 170, 459 Backwardestimation,120,136,146 Badframemasking,440 Bandwidthdistanceprnduct(BDP),388,390, 393,406,435 Banyannetwork,482,483,494 Baselinerestoration, 172 Batcher,K. E.,483,493 Battery,35,47,272, 511 Baud,54,184,185,309,326,502 Bell, Alexandercraham,6, 383 Benes,V. E.,483,493 Bemoullian,542 Binomialprobabilitydisribution,527 Bipolarviolation,175-178,193-199,214,215 Bit insertion.168 Bit leaking,428 Blocking blockedcallsclemed,523,530,569 Clos,232 Jacobaeus, 238,239.273,274,525 t,eegraph,?0, 155,221,234-239,2ffi,262, 271,274,525 probability,16,17,25,217,234-275,376, 481-570 BORSCHT,47, 272 Bragggratings,403 Bridgedtap,88, 185,316,495,497,507,515 Bridgingclips,267 Bursterroredsecond,204 Busyhour,234,241,265,522,540,569,57O Busytone,42, 43,374,380 Byte stuffing,409,415,416 Cablemodems,5l l, 512

635

636

INDEX

CableTV, 65,387,5l I, 516 Call congestion, 541 Call distribution,226 Call forwarding,16, 18 CalItracing,16 Call trarsfer,18 Callwaiting,16,l8 Catrtierrecovery,28, 282,297,335,350 CCIR (InternationalRadioConsultativeCommittee),4,320 CCS(hundredcall seconds), 521,522,569 Cellulardigitalpacketdata(CDPD),288,453,454 Cellulargeographic servicearea(CGSA),53 Centrallimit theo,rem.447 Centraloffice terminal(COT),508,510 Centralizedattendant,I I Centrex,17,18,50 Channelbanks analog A5.26 LsE,29 LMX.28 digital Dl, 107-114,2t0, 211,222,4tr DlA,59 D2, 109,110,156,187_189.223 D3, 103,109,tl4, 156,160,212,220,222. 4ll D,l, 109,212.4r3 D5, 109.178.215 Circuitswitching,3, 269,4#, 475, 480,5I 9, 520, s39 Classof service.473,476 Clearchannelcapability,178,496 Clipping,25, 40, 540,541,545,541 Clockrecovery,60, 297 Clos,Charles,232,273,482 C-message weighting,35,41,99, 114 Coaxialcable,18,26-30,16t*176 Codeblocking,379 Codedivisionmultipleaccess(CDMA), 151,365. M5-454 Coherentmodulation,283,291.297,321 Coin telephone, 508 Compact disc,9l,92 Companding, 37,91,106-128,317 AJaw, 115,116,130,132,154,155,272 instantaneous, I l6 nearlyinstantaneous, l2l F255,109-114,159,160,174,272 syllabic,ll6 Competitivelocalexchange canier(CLEC),12 Compressed voice,I 51, 154,478,484,487,494 ConcenEation,24

Conferencing, n2,214 bridge,87, l40' 2'72,273 digiral,87 Congestion theory,265, 52Q,547 Constantenvelope,288,291,301,309,314,Ml Constantholdingtimes,528,529 Constantservicetimes,489,554-566 Constraintlength,2O5,2W, 223 Cordlessphones

crz, 133 DECT. I33 Crossconnect switches,46, 52, 218, 226,U5, 265-26e,362.492.510 Crosstalk,34,293,495 CSMA/CD,453 Customerserviceunit (CSU),215 Cutoutfraction, 546 Datagrams, 465, 467,468 DATAPAC,465 Dataphone digitalservice(DDS),63,168,268. 456 Dataundervoice(DW), 64 DATRAN,456 dBrnC/dBmC0,36, 42,72, 101 dc balance,l8l, 395,399.500 dc restoration,172,ll3,396, 502,514 dc wander,172-194,396,398,502 Decisionfeedback,173,317,328 Decorrelation,262, 482 Degradedminute,2O4 Delaysystems,519,522,539,552^562 Delayvariation,413,474 Demandtraffrc, 537, 569 Desynchronization,426, 427 Differential detection,298 Differential encoding,I 83 Diffractive grating,40I DiffServ.473 Digitalcellular,3, 54,81,84,91,93,133,141, 151, 153, 277-288, 329,437488 codedi vision multiple access,l5l , 365, 444452 global systemfor mobile communications, 148,157,288,441454 Noflh Americandigitalcellular,l5l, 437 Digital circuitmultiplication(DCM), tU, l3.L, 133,141,540,546 Digital loop carrier(DLC),67, 121,26i,269. 507r509 integrateddigitalloop carrier(IDLC),63,267, 508.509. 517 next generationdigital loop carrier (NGDIf), 509,510

607 subscriberloop carier (SLC), 62 subscriber loopmultiplex(SLM), 62, l2l universaldigitalloopcarier system(UDLC), 267.507.508 Digitalpad,514 Digital phones,504 Digitalradio,3, 53,64-85,169,204,278-333, 382,406,456,515 Digital sensemultipleaccesswith collisiondetection (DSMA/CD),453 (DSPJ,40,82-88, 130Digital signalprocessing 138,r49,165,267,271,1W,315, 330,4?9, 504.506. sr3 Digital speechinterpolation(DSI), 124 (DVB), 33, 69,205,316, Digital videobroadcast JJI

Directbruadcast satellite(DBS),33 Directinwarddial (DID),48 Directprogressive control,I 3, 45,483 Ditectsatellite service(DSS),33,496,516 Distortion 23,37,60, 132,165,328,506 amplitude, delay,389 foldover,95,105,134 harmonic,55, 132 phase, 37,55, 164,328,506 quantization,57 Diversity aflgle,327 ftequency,3 l, 32, 326 space, 32,328 Doppler,337,119 (DTMF),43-49,83, 153 Dualtonemultifrequency Dynamicpowerconfrol,5I5 Dynamicrange,I03, I 14,304,349 Eb/No,192 Echo,23,34,39-55,78-87,120,154,270-273, 437,441,418, 500-502,513 cancellation, 39,40,46,51,82,154,271, 273, 437,44t,502 listenet.39 suppression, 39,40,46,55,82 Elasticstorcs,339 Electxomechaflical switching,13 ElecnonicIndustriesAssociation(ElA), 4, 438, 445,450 Elechonic serialnumbers,451 Emergency calling(91l), 12,511 Encryption, 73,74, 81, l4l, 169,350,451,5I2 Bngsetdistribution,542 specialized Enhanced mobileradio(ESMR),453 Envelopedelay,38,54 Envelopedetection,282,297

55,82, 164,173,3M, 315,317, Equalization, 328,437 ,495,507,5 13 amplitudeequaliz-ation, 315,507 frequencydomain,507 phase,165,507 quantieed feedback,173 Erlang,A. K., 454,521-563 Erlang'sdelayfonnula,556,563 Erlang'slossformula,531,54I Erlang'ssecondformula,556 Enor contol, 45, 80,204,376, 440-474 convolutionalcoding,205,207,221, 440,443 correction,119, 138,154,175, 195,204,354, 423453 cyclic redundancy check(CRC),79,198,204, 213-216,223,331,440,443,460,470 41 detection,203,1tr;0, forward error corrcction,2O4,506 Reed-Solomon(RS),205,207,223, 453,506 Erroredseconds. 20/'.216 Error free second,204 EuropeanTelecommunicationStandardsInstitute (Ersl), 69,316,331,44I, 509 Exchangearea,9-77,1OB ExchangeCarriersStandardsAssociation (ECSA),4,406, s01 Exponentialdistribution,525,528,567 Exponentialholding times, 528.,529, 567 Exponentialservicetimes,489,554,555,559, 560,562,564-566 Extendedsuperframe(ESF), 199,212-216, 223 Fa*imile, 46, 54, 80,92,132,r41, 152,350 Fademargin,31,323,325,326,328,448,515 Fa
638 Flow conhol,43,376-379,4fu-470, 476,489, 561.567 Foreignexchange (FX), 17,46,267,508 Formantfrequencies,127 Forwardestimation,120,121,136,146 Fractionalspeechloss,546,547 Framealignmentsignal(FAS), 213 Framerelay,466,472 Ftamerelayaccessdevice(mAD),472 Frarning addeddigit,210 bitenors, m3,216 byte,215,364 lossoi 203,209,210.485 statistical.209 uniquelinecode,214 Frequencyagility,444 Frequency hoppinE,444,445,M8 justification,414,415 Frcquency Fricatives.123 Full width half magnitude(FWHM), 3,gZ,Ns Gaussian, I 90, I 9l, 202,2O3,222,286,333,344, 47 Geostationary satellites, 338,451,516 Glare,47,48,379 Global positioningsystem(CPS),365, i66, 372, 373 GR-303,509, 5t0 Gradeof service,234,241,252,256,258,377, 457,466,536, 554,558.569 &ading,227 Grooming,267,510 Guardtime,85,86,438,442 Hierarchicalnetwork,6, 8, 5l, 52, 535 High definitionTV (HDTV),490,507,515 High leveldatalink conffol(IIDLC), 169,214, 449,470,472,494 Hookflash,43 Hybrid,22, 23, 38, 40,4'1,54, 93, 148,151, 2702t2 Hybridfiber coax(HFC),5l I , 512 Impedance matching,23,,10,271 Incumbentlocal exchangecarrier, 12 Lrfinitesource,525,531-569 Infbrmationcapacity,506 Informationdensity,85,278-288,3 10-317,441 INMARSAT,33,I48 Instituteof ElectricalandElectronicEngineers (rEEE),4 Integratedsewicesdigital networkflSDN), l, 456,474-516

(BRI),69,173,185,214, basicrateinterface 316.495-510 broadbandintegratedservicesdigital network (BrsDN),474 D channel.189.496-503 L430.499.516 I.,l4l,503 I.451,503 NTI/I.IT2.498.499 primaryrateinterface(PRI),46,69,496 Q.9?1,so3 Q.922,472 Q,931,503 S interface.498.500 S/T basicrate interface,2 I 4 S/T interface.499.501 TElrrEz,498 U interface.50l. 502 Intelligent network,49, 70 INTEL$AT,32 Interexchange carrier(IXC), 10, 12,406 Interference adjacentchannel,288,317-320,448 cochannel.329 electromagnetic(EMI), 280, 393 intersymbol, 54,60,74,98,163-188,288 multipath,322 mutual,43,,14, 80,84,85 narrowband.317,495 krtemationalmobile equipmentidentity (IMEI), 444 Internationalmobile subscriberidentity (MSI), 444 IntemationalStandardsOrganization(ISO), 5 InternationalTelecommunicationUnion (IIU), 4, 4l Intemet,l,61, 141,U\453-515 access, 1,453,504,515 EngineeringTaskForce(IETF),473 protocol(IP),204,473, 474,49D-494 serviceprovider(ISP),61, U,l,5M telephony(IP telephony),473 TCP/IF.473 userdatagramprotocol (UDP) 473 Iridium.45l.516 Jitter 38, 176,210,324.337-358 rnapping,359 measurements, 342 phase,343,344,358,417 removing,359 systematic, 338 waitingtime,339,358-360,3M, 416 Justification, 343,351,360,419,510,528

INDEX Kendall,D. C.,555

639

tokenpassingring, 218,485 802.3,183,485 Laserdisc,341 802.5,218 Li ghtningprotectron,272 842.6,119 Limitedavailability,5?, 76, 227 local exchangecanier (LEC), 10,406, Line administration, 16,241 487,509 Line codeviolation,214,5m Local microwavedistributionservice(LMDS), Linecoding,80,81, l6l-174, 190,282 496,515 2BlQ,220,502,516 Local multipoint communicationssystems altematemarkinversion(AMI), 1,74-178,337, (LMCS),515 499.502.517 Longitudinalcurrcnt,2 I alternatespaceinversion,499 Loop timing,340,346 antipodal,193,283,291,298,333 519,553, 554,561 Losssystems, bipolar,172*198,223 Lost callscleared,523,531-569 binaryN-zerosubstitution(BNZS),176-181 Lost callsheld,539,54,0 8325, 176,177,222 Loudesttalker, 273 B6Z3, 177, 178,1.80,222 Low earthorbit satellite,451,516 B8ZS,168,178 codedmarkinversion(CMI), 183,184,395Ml2 multiplexer,61, 354,356-359,361,429 398,407,433 Manual swirchbosxds,12 conelativelevelencoding,185-188 Matchingloss,265 digitalbiphase(diphase),181-1 84,337,396, Maximumaverageframetine,2l1, 214,222 4m Maximumlikelihood,I 75, 184,207,312 duobinary,185,188 Maximumtime intervalenor (MTIE), 367-373, (HDB3), high densitybipolar3 177 382 Manchester coding,183,337,395-398,400, 535 Mesh,6-8,52, 494 Messagesequencing, 376,503 mBlC, 400,433,435 Messageswitching,455462, 552,559 mBlP.399,435 Messageunit accounting,16 mBnB,396,399 Metalliccurent, 2l 3848.399 Mobileassisted handoff (MAHO),440 4858.398 Mobile swirchingoffrce (MTSO), 52, telephone 5868, 398,399,405 53 6888.398 Modems 8Bl0B.399 v.32,40 (NM), 172,l8l-193, 223, non-retum-to-zero v.33,315 219 -288, 298,305,392-396,425,435 v.34,88,279,303,315, 331,495,513,5r4, panialresponse, 165,185-187,198,288,3055L'7 3il.320 v.90, 55, 153,279,495,5 13,514,5 17 (RZ), 174,188,389-396,405, return-to-uero Moduletion 425 analog temary,lEO amplitude,?6,94, 98, 279-283,297,301, (PST),179-181,195 pairselected 305 483T, tg0,Z22 doublesideband,94 unbalanced. 171.399 frequency(FI\4),30,221,n7,284, 331, unipolar,l7l, 193,195,282 387,453 Link accessprocedure balanced(LAPB), 469,503 index.280.286 l,oadingcoils,23,28,17,54,60, 88, 190,.195, linear,280 507 phase,284,286 Lccal accessandtransportarea(LATA), 10, 12 pulseamplitude(PAM),94-98, 107,158, Localaxeanetworla(LANs),471-473, 485,498 M6,274,513,514 distributedqueueddualbus(DQDB),219 singlesideband,85 emulation(LANE),485 ethemet, 183,453,485 suppressed carrier, 297, 322

640 Modulation (Cantinued) digital carrierless amplitude and phase (CAP), 280,

309.504 continuousphasefrequencyshift keying (CPFSK),285 discretemulti-tone(DMT), 316,317,504, 507 frequencyshift keying (FSK), 284-286 gaussian minimumshiftkeying(GMSK), 286,330,44r, 453 minimumshift keying(MSK), 285,286, 308,309,315,330,331 multilevelcoded(MLCM), 315 ott/off keying, 282 orthogonalfrequencydivision multiplexing (OFDM),316 phasereversalkeying(PRK),283-290 phaseshiftkeying(PSK),283-302,438,441 quadrature amplitude(eAM), 88,288,301328 quadraturepartial responsesignaling (QPRS),310,3il trellis,207,312,314 MPEGIA{PEGz,490 Multichannelmultipoint distributionservice (MMDS),496,515 Multiframe alignmentsignal,2 I 3 Multipath,30-32, 53, I2l, 322-329,437,444449 Multiplexing frequencydivision multiplexing (FDM), 26* 86, 207,216, 303,317, 402,437453, 5t7 . guardbands,44S hierarchy,27,28 mastergroup,28, 63 supergroup,28 time division,58,444,461 analog,74 asynchronous, 5S,208,351,461 hierarchy,74,351 loop,216-218 ring,216 statistical(statmux),208,461-463,552 synchronous, 3, 58,2O8,352,420,46I Multiprotocol label switching(MPLS), 491 Netloss,34,39,41,42,82,271,384 Network cofltrol point (NCP), 50 Networkcontol protocol(NCP),473 Networkmanagement, 43, 376,377,407,411, 413.477- 520

Noise background,147-153, 445 Gaussian, 190,191, 202,222,333,3M idle channel,72, 81, 102-120,273 impulse,35,55,60,80, 190,203 modepartition, 392 quantization noise,35,99-l19,128,135,I38. 1 5 91 , 60,317.513 thennalnoise,35, 3U' 333 whitenoise,35.99.190 Noisebandwidth,l9l, 192,298-320 Noisefigure,29, 324,3?5 Noisepower,36-42 Nordic mobile telephone(NMT), 54, 437 Normal distribution. I 9l Nyquistsamplingrate,94,99, 149,159*165,305, 326 Offeredload, 244, 265, 377, 529-569 Offset keying, 305 Opensystemsinterconnection (OSI),5, 498,5 17 Openwire, 18 Overflowtraffic,519.548-553 Overvoltageprotection,272 Packetswitching,208,456472, 490,519, 552566 Paging,49, 4a+,451,516 Pairisolation,185 Parity,199,409 Passivephotonicloop, 401 Pathfinding,U2-24 Peakcell rate,475-490 Performance monitoring,79, l7 l-182, 195,I 99, 2rs,2r6,376.400.509 Permanent virhralcircuits,4#,466, 491 Petsonalaccesscommunications system(PACS), 133 Personalcommunication system(PCS),444,450 Personalhandyphone system(pHS),133 Phaselockedloop(PLL),336-343,381,428,485 Phonemes,92 Pickup, 18,385 Pilot,28,79,315,5M Pitch,123,l4l 139,144-149 Plainold telephoneservice(POTS),54,495,506, 507.510.512 splithr,506 Pointof presence (POP),10, 12 Poisson, 203,526,527,529,531, 540,541,550. 551 Poweramplifiers,30,?84,291,318,319 Precoding,I 86 Primaryreference clock (PRC),485

641 himary rcferencesource(PRS),186,187,311, 3't0,372,427, 506 Privatebranchexchange(PBX), 16,17,4'l-49 Privatemobile radio,452 Prcb(delay),556 delay,33,451 Propagation hotection switching,404 Provisioning, 267 (PRS),506 Pseudorandomsequence Psophomeric,35, 99 Pulsecodemodulation(PCM)modem,153,513, 515 Pulse*haping,96, 307,308 Pulsestuffing,78, 351-364,461 Quadraturemultiplexing, 293 290 Quadraturesignalrepresentations, 458,488, 520,552,555,567 Queueing, delays,458,465,558,564 finite,561 tandem,566 M/D/l,559 MA{/I, 555,556,559 MA,I/N.555.562,567 MA.4INI*|L.562 Radiocommoncarrier(RCC),52 Radiosystemavailability,322 Rainattenuation,30, 32,321-323,515 Raisedcosine,319 Randomrefies, 537,538,569 Rayleighfading,389 Reedtelays,I 6, 35 Refraction,329,387,388,401 Reframetime,209,222,355 59,60,74-Bl, 98, l l 1, 167*176 Regeneration, Reis,PhiIIip,l, 69 Remoteconcentrators, 227 Remoteprovisioning,5 10 Remoteswitching,24,62, 509 Remoteterminal,508 Repeater 30,65,190,195,323,388-406 spacing, Residualerror,136,138 Returningtraffic,538 Ringback,42,43,374 Ringingvoltage,43, 47,48 Routing,376,463,4& alternate, 13,379 besteffort,4?3 crankback,51 dynamic,463,4M fixed path,463 packels,463

satellite,26,32,54,75,86, 148-154 Molniya,32 Westar,33 81, 170,322,396,410,425 Scrambling, Seaplow,404 Secondgeneration cordlesstelephones, 133 Segregation,26T numbers,470 Sequence Servicecontrolpoint(SCP),50 Severelyerroredseconds, 216 Shannon's theorcm,513 Shortmessage service(SMS),444 Sidetone,39 13,14,42-88, I 84-331,500 Signaling, A andB bits,212,216 channelassociated, 213 comrnonchannel, 18,43-53,75,109,153, [l 8, 214,377-380,421,496 E&M.47.49 groundstart(GS),48 loop start(LS),47, 48, 508 pertuk,43 robbedbit, 109,178,216,514 wink,48 Signalingsystem(SS7),50,52 Signalingtones,153,314 (MF),43,49,83,88,153 multifrequency singlefrequency(SF),43, 49,83, 153,212, 216,222,336,391 sin(.r)/x,162,ll2, 28Q,298 Singing,34,38,39,4I, 120,154,27o,nl Skyphone, 33,69, 148,157 Slips,216,341-350, 362 buffer,420 349,485 controlled, rate,349,350 Slopeoverload, 135,136,l4l Soft handoff.450 Sourcecoderestriction, I 68 Spaceexpansion ,235-262, 525 Spaceswitching,67,257 Spanline,59,60, 181,215 Spanswitching,431 Specialized mobileradio(SMR),453 127,14 Spectogram, 284,291,386,445,446, Spectrumspreading, 448,451 Speechcompres sion,124,437,441,49 pauses, Speech 34,80,8l, 102,120,122,124,170 Speechrecognition,46,86,93, 147 Speechstorage,9l Splices, 88,387,390,404,405 Spreadspectrum, 445, 448,450 Starnetwork,217,5lO

642 Statisticalequilibrium,531,537 Storedprogramcontrol($PC),16,,14,49,51, 225,24 Stuffingratios,360,429 Submarine cables,384,403,540 NLl. 398.434 TransPacific.400 Subscriber identitymodule(SIM),M, 451 Subscriberloops analog,line interfaces, 272,5M digitar,75,87, 185,204,27o,456,495,49'1, 503,516 asymmetricdigital subscriberline (ADSL), 3 10,316,317,331,495,503-516 interfaces,5I0 vDsL.507.510 XDSL,503,507,509 Superframe, 212,211, 215,419,502 Supervision, 47, 272 Supervisory audiotone($AT),439 Swirchedmultimegabitdataservice(SMDS),4&4 Switchedviilual circuit(SVC), 46/.,470,487 Swirches, network class3, 56 class4, 6, 12 class5 (endoffice),9-87,121,225,226,Mt, . 262,270,379,380,502,508,5?5 crossbar, I3-16,49,66 DMS,16,66,272,274 electronicswitchingsystem(ESS),16,49,52, 65,7 L, 225,233,259_274, 502 No. 1, 16,65,225,233 No. 101.273 No. lA. l6 N o . 2 ,1 6 N o . 3 .1 6 No. 4, I 6, 49-7l, 225,259,262,273,366, 381 No.5, 16,272,274.502 No. 5 EAX, 16 EWSD,260,273 stcp-by*step,13-55, 225, 244, 483 toll,6,9,10,67,236 Switching,time division,246 analog,246,252 digital,66,67,226,238,U7 ,253,340,548 rnemoryswitch,248,478-480

sTs.255 TSSSST,262 TSSST.260.261 TST,255*260,275, 348,48Q,482 Synchronization, 85, 89,221,331,3#-381.427 clock,I I l, 209-211,335, 351,374.513 frame,59,209,214,347 hieraxchical,374

network, 85,210,335, 351,362,485 isochronous, I 67 mutual,36?.364 networkmaster,362 plesiochronous, 362,363,370,381 stratum. 370*373.382 Synchronous datalink protocol(SDLC),469 Synchronous digitalhierarchy(SDH),5, 352, 4Q6,407,434 Synchronous opticalnetwork(SONET),3,5,52, 78, lM, 219,M9, 269,352,373,396,405_ 4ll asynchronous mapping,419, 420 concatenation, 267,269,408 gatewaynetworkelement(GNE), 426-428 line overhead,41Q414, 425 pathoverhead, 410425, 478 pointer adjustments,4 I 5, 426, 427 pointerburst,428 rings,52, 219,405,429 bidirectionalline swirched(BLSR),431,432 unidirectional pathswitched(lJpSR),429, 435 sectionoverhead,410,411,425 payloadenvelope(SPE),413, 414 synchronous fransportoverhead, 4O8-410,415 virtual tributary (VT), 417,429 Synchronoustransmission,I 67 Systemavailability,322,326,403 Systemgain,323,405 T-carrier,59, 63, 132,167-178, 2lO, 215,222, 268,335,337,340 El, 59, 177,213,359,382,40'_41 t t, 421,435, 484.497.510 82, 330,354,359,407.413 H, t84, 407,423.436.478 repeaters, 60, 62, I I l, 358 TI48. 181 Ttc,56,60, 185 T l D . 5 6 .1 8 8 Tlc, 185 T2,56,60,6t, 67,69, 169,177,220 T3.56 T4,56 T4M, 169,t70,2l9 Tl committee,4 TI,4, 5, 56-67,84,91,t08, I I I, 132,156,t67_ 174 Tagswitching,49I Talker echo.39 Tandemclockrecovery,338' Tandemswitching, 8, I, 12,66,67,226,236,237, 370.535.569 Telephoto,54

643 Televisiondistribution,29, 33,54 Tic lines,L7,46, U0,268, 553 Time assignment speechinterpolation(TASI), 25, 124,540-547,568 Time compression multiplex(TCM),207,280, 312-314,502,516 Timeconge$tion, 541,542,563 Time divisionmultipleaccess(TDMA), 86, 89, 154,438-452,512 Time expansion, 258,259,261,26?,482 Timeintervalerror(TIE), 359,367,369,382 Time slotintetchmge(TSI), l'18, 249-254, 269, 275,348 Time variance(TVAR), 367, 369,370 Tip andring,48,499,508 Toll network,6, 9-34, 49,60-69,108,ll3, 271 Traffic intensity, 241,521-561 Transmission levelpoint (TLP), 41,42,7Z 33,286 Transpondcr, filters,328 Transvemal Trunkdirectionalization, 379 Trunkedradio,452 TYMNET.465.468 U.S.Independent Telephone Association (usITA),4, 366,370 U.S.TelephoneAssociation(USTA),4 Unit interval(UI), 343,344,361,382 Universalcoordinated time,372 v5,419,422,509,5r7 Vectorquantization, 150 Via netloss(VNL),39,41,82 Video,77,2Q4,2'17 ,387,474-515 Virtual circuit,464-469,477,487,491 Viterbidetectors,175,184,195,I 98, 201,220, 3l 2. 33I. 454 data,38,40,55,83,98, 124,132,140, Voiceband r4tt,161,167,265,277,303,350,485,51I Voicecoding adrlptivedeltamodulation(ADM), 136,I40, 156.429.510 adaptivedifTercntial PCM (ADPCM),1321 4 01, 5 3 - 1 5 6 , 2 1 5 adaptivepredictivecoding(APC), 136,138, r49,151,157 analysisandsynthesis, 92 channelvocoder,141-14 deltamodulation, 62,69,9l, 93, 133-136, r4t.153.2M diff'erentialpulsecodemodulation(DPCM), 9 3 ,I 2 7 - 1 3 6 , 1 4 1 , 1 5 3 easilydigitallylinearizable coding(EDL), 108, 110,27?. formuntvocoder.141.144

Iinearpredictive coding,92, 138,141*158,487 algebraiccodeexcited(ACELP),154,155, 487 codeexcited(CELP),150-158 multipulse(MPLPC),147-149,151 pulseexcited(pelp),I 47 Qualcommcodeexcited(QCELP),l5l,449 residualexcited(RELP),147,149,151.158 vectorsumexcited(VSELP),151,I 58,439, 453 pulsecodemodulation(PCM),71, 88-l14, 128,136,140,159 predictiongain,I 36 138,139,142 subband, vocodem,93, 102,123,I27,141-153 c.704.213 G.7ll.154 G.12t,t32,133, 154 G.722,140,154,157,? 15 G.723.154,155,494 G.723.r,154, 155,494 c.726. 154.155 G.727,155 G.728.154.155 G;729,r54,155,158,487 G . 8 tl . 3 6 3 3 . 81 G,821.2M.216 Voice messaging, 91,92, I 32 Voicequality,80,99, 102,113,133,146,151, 443,547 broadcast,15l communications, l5l, 152 measure(DAM), I47, djagnosticacceptability 151 diagnosticrhymetest(DRT). 147,l5l meanopinionscore(MOS),133,l5l signal-to-quantizing ratio (SQR),I02-1 15, 1 5 91 . 60.513 synthetic, l5l,154 toll.132.151-155 Voltagecontrolledoscillator(VCO),285, 336343,485 WATS, 17,380,553,-570 Wavelengthdivisionmultiplexing(WDM), 26, 385-405.434.415 C. J.,546,568 Weinstein, Wire gaugechanges,497 x.25.469-472.503 Zero bit insertion, 169,469 Zerobytetime slot interchange, 178 Zerolosstransmission, 270

ghrltntrnf sftrtfi;r Btr

-- e r r t l h y

r TTTTT

[T.T.E.

Digital Telephony

Recommend Documents