Professional Documents
Culture Documents
COMM1208 Unit6 PCM Sampling
COMM1208 Unit6 PCM Sampling
1. Introduction.......................................................................................................................... 2
2. TimeDivisionMultiplexing(TDM)Principle..............................................................3
3. SamplingTheorem............................................................................................................... 4
3.1 SamplingMethods............................................................................................................ 4
3.2 AliasingError................................................................................................................... 5
4. PulseAmplitudeModulation.............................................................................................6
5. PulseCodeModulation.......................................................................................................7
5.1 Quantization..................................................................................................................... 7
5.2 Companding...................................................................................................................... 8
5.3 PCMEncodingProcess(HDB3).......................................................................................8
5.4 PCMTimingandSynchronisation...................................................................................9
6. Differentialpulsecodingschemes.................................................................................11
6.1 DeltaModulation............................................................................................................11
6.2 DifferentialPCM(DPCM)andADPCM........................................................................11
7. TDMandCodecs.................................................................................................................12
8. DigitalTransmissionHierarchies..................................................................................13
8.1 PrimaryRateFrame.......................................................................................................13
9. PlesiochronousDigitalHierarchy(PDH).....................................................................14
1 03/07/2017
1. Introduction
Inthesimplestmodelofatelephonespeechcommunicationthereisadirect,dedicated,
physicalconnectionbetweenthetwoparticipantsintheconversation,andthislinkisheldfor
thedurationoftheconversation.Theanalogueelectricalsignalproducedbythetelephoneat
eitherendissentontoconnectionwithoutmodification.
InPulseAmplitudeModulation(PAM),theunmodifiedelectricalsignalisnotsentontothe
connection.Instead,shortsamplesofthesignalaretakenatregularintervals,andthese
samplesaresentontotheconnection.Theamplitudeofeachsampleisidenticaltothesignal
voltageatthetimewhenthesamplewastaken.Typically,8,000samplesaretakenper
second,sothattheintervalbetweensamplesis125s,andthedurationofeachsampleis
approximately4s.
Becauseeachsampleisveryshort(~4s)thereisalotoftimebetweensamples(~121s).
Samplesfromotherconversationsareputintothissparetime.Usuallythesamplesfrom32
separateconversationsareputontoasingleline.ThisprocessiscalledTimeDivision
Multiplexing(TDM).
Eachsampleisveryshort,andwillbedistortedasittravelsacrossacommunications
network.Inordertoreconstructtheoriginalanaloguesignaltheonlyinformationthereceiver
needstohaveaboutasampleisitsamplitude,butifthisisdistortedthenallinformation
aboutthesamplehasbeenlost.Toovercomethisproblem,thepulseisnottransmitted
directly,insteaditsamplitudeismeasuredandconvertedintoan8binarynumbera
sequenceof1sand0s.Atthereceiverend,thereceivermerelyneedstodetectifa1ora0has
beenreceivedsothatitcanstillrecovertheamplitudeofaPAMpulseevenifthe1sand0s
usedtodescribeithavebeendistorted.
Theprocessofconvertingtheamplitudeofeachpulseintoastreamof1sand0siscalled
PulseCodeModulation(PCM)
NotethattheprocessofPAMandPCM(butwithouttheuseofTDM)isessentiallyusedto
storemusicandspeechonCDs,butwithahighersamplerate,morebitspersampleand
complexerrorcorrectionmechanisms.
Sometermsare:
Sampling Theprocessofmeasuringtheamplitudeofacontinuoustimesignalatdiscrete
instants.Itconvertsacontinuoustimesignaltoadiscretetimesignal.
Quantizing Representingthesampledvaluesoftheamplitudebyafinitesetoflevels.It
convertsacontinuousamplitudesampletoadiscreteamplitudesample.
Encoding Designatingeachquantizedlevelbya(binary)code.
Samplingandquantizingoperationstransformananaloguesignaltoadigitalsignal.
UseofquantizingandencodingdistinguishesPCMfromanaloguepulsemodulationmethods.
Thequantizingandencodingoperationsareusuallyperformedinthesamecircuitatthe
transmitter,whichiscalledanAnaloguetoDigitalConverter(ADC).Atthereceiverendthe
decodingoperationconvertsthe(8bit)binaryrepresentationofthepulsebackintoan
analoguevoltageinaDigitaltoAnalogueConverter(DAC)
2 03/07/2017
2. Time Division Multiplexing (TDM) - Principle
Whensendingsamplesofasignalinsteadofthesignalitselfthereistimeavailablebetween
eachofthesamples.Samplesfromotheranaloguesignalscanbeputintothisspace.The
Receiver processofsplittingupthe
Transmitter
Timing Timing timeintoslotsandputting
differentsignalsintothe
Ch1
Buffer LPF1
Ch1 timeslotsisknownasTime
i/p o/p
Ch1 TransmissionLine Ch1 DivisionMultiplexing
Buffer LPF2
i/p
SW1 SW2
o/p (TDM).AbasicrealTDM
Ch1 Ch1
i/p Buffer LPF3 o/p systeminterleaves32
signalsanduseselectronic
switches.Thisisadiagramofa3channelPAMTDMsystem.
ThisdiagramshowsthewaveformsproducedduringtheoperationofthePAMTDMsystem
Theswitchesconnectthetransmitterandthereceivertoeachofthechannelsinturnfora
Ch1
Ch2
Ch3
TDM
Signal
GuardSlots
Time
Slots 1 2 3 1 2 3 1 2 3
specificintervaloftime.Ineffecteachchannelissampledandthesampleistransmitted
Whentheswitchesareinthechannel1position,channel1formsaPAMchannelwithanLPF
forreconstruction,andsoonforchannels2and3.Theresultisthattheamplitudessamples
fromeachchannelsharethelinesequentially,becominginterleavedtoformacomplexPAM
wave,asshownabove.
AmajorproblemsinanyTDMsystemisthesynchronisationofthetransmitterandreceiver
timingcircuits.Thetransmitterandreceivermustswitchatthesametimeandfrequency.
AlsoSW1mustbeinthechannel1positionwhenSW2isinthechannel1position,sothatthe
switchesmustbesynchronisedinpositionalso.
Inasystemthatusesanaloguemodulation(PAM)thetimeslotsareseparatedbyguardslots
topreventcrosstalkbetweenchannels.
3 03/07/2017
3. Sampling Theorem
Considerabandlimitedsignalwithnofrequencycomponentsaboveacertainfrequencyf m.
Thesamplingtheoremstatesthatthissignalcanberecoveredcompletelyfromasetof
samplesofitsamplitude,ifthesamplesaretakenattherateoffs>2fmsamplespersecond.
Thisisoftencalledtheuniformsamplingtheoremforbasebandorlowpasssignals.
Theminimumsamplingrate,2fmsamplespersecond,iscalledtheNyquistsamplingrate(or
Nyquistfrequency);itsreciprocall/(2fm)(measuredinseconds)iscalledtheNyquistinterval.
fs=2*fmiscalledtheNyquistsamplingrate.
Fortelephonespeechthestandardsamplingrateis8kHz(oronesampleevery125s).
Time
Spectrum of Sampled Signal
Ineffectthesignalm(t)ismultipliedbyatrainofimpulsesgivingrisetoatrainofpulsesas
inthelowerlineofthediagram.Thetrainofsamplingimpulseshasafrequencyspectrum
consistingofallharmonicsormultiplesoffsandallareatthesameamplitude.
ThissampledsignalhasaspectrumasshownwhereM(f)isrepeatedunattenuated
periodicallyandappearsaroundallmultiplesofthesamplingfrequency(f s=1/ts).
Torecoverm(t)fromthesampledsignalweneedonlypassthesampledsignalthroughalow
passfilterwithastopfrequencyoffs/2.Allofthehigherfrequencycomponentswillbe
dropped.Inthediagram,iffsisgreaterthantwicethehighestfrequencyinm(t)the
repetitionsofthesampledspectraaroundtheharmonicsofthef sdonotoverlap.
4 03/07/2017
3.2 Aliasing Error
Ifasignalisundersampled(sampledataratebelowthe
Nyquistrate),thespectrumconsistsofoverlapping
repetitionsofthesampledspectrum.Becauseofthe
overlappingtailsasinglerepetitionofthespectrumno
Spectrum of Sampled Signal longerhasthecompleteinformationabouttheunsampled
signal,anditisnolongerpossibletorecoveritfromthe
sampledsignal.Torecovertheoriginalsignalatthereceivingendthesampledsignalis
passedthroughalowpassfilterwithacutoffoffs/2,wegetaspectrumthatisnotthesampled
signalbutisadifferentversiondueto:
Lossofthetailofthesampledsignalspectrumbeyondf s/2
Thissametailappearsinverted,orfolded,ontothespectrumatthecutofffrequency.
Thistailinversionisknownasaliasing,(orspectralfoldingorfoldoverdistortion).
Thealiasingdistortioncanbeeliminatedbycuttingthetail(i.e.filtering)ofthesampled
signalbeyondf>fs/2beforethesignalissampled.Bysodoing,theoverlapofsuccessive
cyclesinthesampledsignalisavoided.Theonlyerrorintherecoveryoftheunsampledsignal
isthatcausedbythemissingtailabovefs/2.
fm fs 2fs 3fs Itissimplertoconsideraliasingbyconsidering
asinglefrequencycomponentofm(t).Wewill
lookatthefrequencyfmanditissampledata
fs-fm fs+fm 2fs-fm 2fs+fm 3fs-fm 3fs+fm
ratefs.Thediagramsshowthefrequencies
whichwillbepresentinthesampledsignal.
Therewillbefrequencycomponentsatfm,fsfm,
fs+fm,2fsfm,2fs+fm,3fsfm,3fs+fm,etc.etc.
fm
fm
Frequency
Inthefirstcasefmisverymuchlessthanfs,sothatfsfmismuchhigherthanthecutoffof
thefilter(fs/2).
Inthesecondcasefmisbelow,butclosetofs/2,sothatasharpcutofffilterisrequiredto
ensurethatfmispassedbutfsfmisstopped.
Inthethirdcasefmishigherthanfs/2,sothatfsfmislessthanfs/2.Thelowpassfilterwitha
cutoffoffs/2willthereforeblockfm(theactualsignalfrequency)butwillpassasignalwith
frequencyfsfm.
Thisisaliasing
Strictlyspeaking,abandlimitedsignaldoesnotexistinreality.Itcanbeshownthatifa
signalistimelimiteditcannotbebandlimited.Allphysicalsignalsarenecessarilytime
limitedbecausetheybeginatsomefiniteinstantandmustterminateatsomeotherfinite
instant.Hence,allpracticalsignalsaretheoreticallynonbandlimited.
5 03/07/2017
Arealsignalcontainsafiniteamountofenergy,thereforeitsfrequencyspectrummustdecay
athigherfrequencies.Mostofthesignalenergyresidesinafiniteband,andthespectrumat
higherfrequenciescontributeslittle.Theerrorintroducedbycuttingoffthetailbeyonda
certainfrequencyBcanbemadenegligiblebymakingBsufficientlylarge.
Thus,forallpracticalpurposesasignalcanbeconsideredtobeessentiallybandlimitedat
somevalueB,thechoiceofwhichdependsupontheaccuracydesired.Apracticalexampleof
thisisaspeechsignal.Theoretically,aspeechsignal,beingafinitetimesignal,hasan
infinitebandwidth.Butfrequencycomponentsbeyond3400Hzcontributeasmallfractionof
thetotalenergy.WhenspeechsignalsaretransmittedbyPCM theyarefirstpassedthrough
alowpassfilterofbandwidthof3500Hz.(Thisfilteriscalledanantialiasingfilter).Higher
samplingrates(i.e.8000samples/sec)permitsrecoveryofthesignalfromitssamplesusing
relativelysimplefiltersi.e.itallowsforguardbandsbetweentherepetitionsofthespectrum
(otherwiserecoveringsignalssampledattheNyquistratewouldrequireverysharpcutoff
(ideal)filters).
Insummary,aliasingdistortionproducesfrequencycomponentsinthedesiredfrequency
bandthatdidnotexistintheoriginalwaveform.Aliasingproblemsarenotconfinedtospeech
digitisationprocesses.Thepotentialforaliasingispresentinanysampledatasystem.
Motionpicturetaking,forexample,isanothersamplingsystemthatcanproducealiasing.A
commonexampleoccurswhenfilmingarotatingwheel.Oftenthesamplingprocess(the
picturerefreshrate)istooslowtokeepupwiththewheelmovementsandspuriousrotational
ratesareproduced.Ifthewheelrotates355 0betweenframes,itlookstotheeyeasifithas
movedbackwards50.
6 03/07/2017
5. Pulse Code Modulation
PulseCodeModulation(PCM)isanextensionofPAMwhereineachanaloguesamplevalueis
quantizedintoadiscretevalueforrepresentationasadigitalcodeword.
Thus,asshownbelow,aPAMsystemcanbeconvertedintoaPCMsystembyaddinga
suitableanaloguetodigital(A/D)converteratthesourceandadigitaltoanalogue(D/A)
converteratthedestination.
Modulator PCMisatruedigitalprocess
Analogue PCM
ascomparedtoPAM.In
Input
A to D Binary
Parallel Digital Output PCMthespeechsignalis
Sampler to Serial Pulse
Converter Coder
Converter Generator
convertedfromanalogueto
digitalform.
Demodulator
PCM Analogue
Serial to
Input D to A Output
Parallel LPF
Converter
Converter
PCMisstandardisedfortelephonybytheITUT(InternationalTelecommunicationsUnion
Telecoms,abranchoftheUN),inaseriesofrecommendationscalledtheGseries.For
exampletheITUTrecommendationsforoutofbandsignalrejectioninPCMvoicecoders
requirethat14dBofattenuationisprovidedat4kHz.Also,theITUTtransmissionquality
specificationfortelephonyterminalsrequirethatthefrequencyresponseofthehandset
microphonehasasharprollofffrom3.4kHz.
Inquantizationthelevelsareassignedabinarycodeword.Allsamplevaluesfallingbetween
twoquantizationlevelsareconsideredtobelocatedatthecentreofthequantizationinterval.
Inthismannerthequantizationprocessintroducesacertainamountoferrorordistortion
intothesignalsamples.Thiserrorknownasquantizationnoise,isminimisedbyestablishing
alargenumberofsmallquantizationintervals.Ofcourse,asthenumberofquantization
intervalsincrease,somustthenumberorbitsincreasetouniquelyidentifythequantization
intervals.Forexample,ifananaloguevoltagelevelistobeconvertedtoadigitalsystemwith
8discretelevelsorquantizationstepsthreebitsarerequired.IntheITUTversionthereare
256quantizationsteps,128positiveand128negative,requiring8bits.Apositivelevelis
representedbyhavingbit8(MSB)at0,andforanegativeleveltheMSBis1.
5.1 Quantization
Thisistheprocessofsettingthesampleamplitude,whichcanbecontinuouslyvariabletoa
discretevalue.LookatUniformQuantizationfirst,wherethediscretevaluesareevenly
spaced.
levels,eachofstepsize,givenby
-mp +mp
Input
=2mp/L
7 03/07/2017
Asampleamplitudevalueisapproximatedbythemidpointoftheintervalinwhichitlies.
Theinput/outputcharacteristicsofauniformquantizerisshown.
5.2 Companding
InauniformorlinearPCMsystemthesizeofeveryquantizationintervalisdeterminedby
theSQRrequirementofthelowestsignaltobeencoded.Thisintervalisalsoforthelargest
signalwhichthereforehasamuchbetterSQR.
Example: A26dBSQRforsmallsignalsanda30dBdynamicrangeproducesa56
dBSQRforthemaximumamplitudesignal.
InthiswayauniformPCMsystemprovidesunneededqualityforlargesignals.Inspeechthe
maxamplitudesignalsaretheleastlikelytooccur.ThecodespaceinauniformPCMsystem
isveryinefficientlyutilised.
Amoreefficientcodingisachievedifthequantizationintervalsincreasewiththesample
value.Whenthequantizationintervalisdirectlyproportionaltothesamplevalue(assign
smallquantizationintervalstosmallsignalsandlargeintervalstolargesignals)theSQRis
constantforallsignallevels.Withthistechniquefewerbitspersamplearerequiredto
provideaspecifiedSQRforsmallsignalsandanadequatedynamicrangeforlargesignals
(butstillwiththeSQRasforthesmallsignals).Thequantizationintervalsarenotconstant
andtherewillbeanonlinearrelationshipbetweenthecodewordsandthevaluesthey
represent.
Originallytoproducethenonlinear
quantizationthebasebandsignalwas
compressed signal level
passedthroughanonlinearamplifierwith
input signal level
input/outputcharacteristicsasshown
beforethesamplesweretaken.Lowlevel
signalswereamplifiedandhighlevel
signalswereattenuated.Thelargerthe
samplevaluethemoreitiscompressed
beforeencoding.ThePCMdecoderexpandsthecompressedvalueusinganinverse
compressioncharacteristictorecovertheoriginalsamplevalue.Thetwoprocessesarecalled
companding.
Thereare2compandingschemestodescribethecurveabove:
1. LawCompanding(alsocalledlogPCM)
ThisisusedinNorthAmericaandJapan.Itusesalogarithmiccompressioncurvewhichis
idealinthesensethatquantizationintervalsandhencequantizationnoiseisdirectly
proportionaltosignallevel(andsoaconstantSQR).
2. ALawCompanding
ThisistheITUTstandard.ItisusedinEuropeandmostoftherestoftheworld.Itisvery
similartotheLawcoding.Itisrepresentedbystraightlinesegmentstofacilitatedigital
companding.
Originallythenonlinearfunctionwasobtainedusingnonlineardevicessuchasspecial
diodes.ThesedaysinaPCMsystemtheAtoDandDtoAconverters(ADCandDAC)include
acompandingfunction.
8 03/07/2017
5.3 PCM Encoding Process (HDB3)
Theoutputfromtheanaloguetodigitalconverter(ADC)hasnparallelbits.Inthecaseof
telephonyn=8.Themostsignificantbitisthesignedbit.Ifthemeasuredsampleispositive
thenthesignedbitis0.Ifthemeasuredsampleisnegativethenthesignedbitis1.The
remaining7bitsareusedtocodethesamplevalue.TheITUTdefinealookuptablewhich
allocatesaparticularbinarycodetoeachquantifiedAlawvalue.
Thelinecodingwhichisusedassignsoppositepolaritiestosuccessive1s.Thiseliminates
anyDCvoltageontheline,andreducestheintersymbolinterferenceifadjacentbitsare1.
IfthereissilenceonthePCMchannelthenthemeasuredsampleswillbe0V rmsandthe
outputoftheDACwillbe10000000.Astreamofallzerosisnotdesirableonanactive
channelbecause
allzeroscouldalsobeafaultconditionand
itisdifficulttorecovertheclocksignalfromtheincomingsignal.
ThecodingsystemHDB3isusedandwasdevelopedtoeliminateallzeros,andtoassign
oppositepolaritiestosuccessive1s.
Thisisabipolarsignallingtechnique(i.e.reliesonthetransmissionofbothpositiveand
negativepulses).
InAMIpositiveandnegativepulses(ofequalamplitude)areusedforalternativesymbols1.
Nopulseisusedforsymbol0.Ineithercasethepulsereturnsto0beforetheendofthebit
interval.ThiseliminatesanyDContheline.
HDB3encodingrulesfollowthoseforAMI,exceptthatasequenceoffourconsecutive0'sare
encodedusingaspecial"violation"bit.The4th0bitisgiventhesamepolarityasthelast1bit
whichwassentusingtheAMIencodingrule.Thispreventslongrunsof0'sinthedatastream
whichmayotherwisepreventareceiverfromtrackingthecentreofeachbit.Byintroducing
violations,extra"edges"areintroduced,enablingaDigitalPLLtoreliablyreconstructthe
clocksignalatthereceiver.TheHDB3istransparenttothesequenceofbitsbeing
transmitted(i.e.whateverdataissent,theDigitalPLLcanreconstructthedataandextract
thebitsatthereceiver).
TopreventaDCbeingintroducedbyexcessiverunsofzerosanyrunofmorethanfourzeros
encodesasB00V.ThevalueofBisassigned+oralternatelythroughoutthebitstream.
Example 11111111 = ++++
BBBBBBBB
10101010 = +00+00
B0B0B0B0
10000001 +000+00
= B000V00B
10000110 = +000++0
= B000VBB0
9 03/07/2017
APCMchannelissampledat8,000Hzoronceevery125s.Ifthereisonechannelor30
data clock 64 kbit/s TDMchannelsthesamplingperiodisfixedat125
15.625 s
sandthisperiodisknownasaframe.Therefore
theframeclockmusthaveaperiodof125s.The
frame clock 125 s risingedgeoftheframeclockinformsthereceiver
thatthenextbitwillbeBit1ofanewsample.The
fallingedgeofthedataclockinformsthereceiver
thatitmustreadthedatabit.
B1 B2 B3 B4 B5 B6 B7 B8 B1
1 0 1 0 0 1 1 1 ?
Whenthebitstreamistransmittedalongalinethepulsesbecomedistortedandtheriseand
falltimesbecomesignificant.Ideally,a1willbehighfor15.625s.Inpracticethepulse
mayonlybeabovethehighthresholdforafewssoitisveryimportantthatthebitisread
withinacertaintimelimitoftheclockpulse.
ThesimplestwaytosynchroniseaPCMsendertoaPCMreceiveristosendtheclocksignals
ondifferentcircuitstothedataThiswouldbedoneinaselfcontainedsystemsuchasprivate
branchexchange(PBX).Telephonyisfullduplexsothatthereisacoderandadecoderateach
port,buteachwouldusethesameclock.
Tominimisethenumberofcircuitsitispossibletousealinecodingschemewhichallowsthe
receivertoextracttheclocksfromthePCMsignal.Inthiscasethereceiverwillhavefree
runningclocksthatlock(usingaPLL)tothephaseandfrequencyofthetransitionsinthe
datastream.Thelinecodingschemeensuresthatthereisatransitionforeverydatabit.
10 03/07/2017
6. Differential pulse coding schemes
PCMtransmitstheabsolutevalueofthesignalforeachframe.Insteadwecantransmit
informationaboutthedifferencebetweeneachsample.Thetwomaindifferentialcoding
schemesare:
DeltaModulation
DifferentialPCMandAdaptiveDifferentialPCM(ADPCM)
11 03/07/2017
individualamplitudesample.ThesamplingrateisoftenthesameasforacomparablePCM
system,unlikeDeltaModulation.
AdaptiveDifferentialPulseCodeModulationADPCMisstandardisedbyITUT
recommendationsG.721andG.726.Themethoduses32,000bits/spervoicechannel,as
comparedtostandardPCMs64,000bits/s.Fourbitsareusedtodescribeeachsample,which
representsthedifferencebetweentwoadjacentsamples.Samplingis8,000timesasecond.It
makesitpossibletoreducethebitflowbyhalfwhilemaintaininganacceptablequality.While
theuseofADPCM(ratherthanPCM)isimperceptibletohumans,itcansignificantlyreduce
thethroughputofhighspeedmodemsandfaxtransmissions.
TheprincipleofADPCMistouseourknowledgeofthesignalinthepasttimetopredictthe
signalonesampleperiodlater,inthefuture.Thepredictedsignalisthencomparedwiththe
actualsignal.Thedifferencebetweentheseisthesignalwhichissenttolineitistheerrorin
theprediction.Howeverthisisnotdonebymakingcomparisonsontheincomingaudiosignal
thecomparisonsaredoneafterPCMcoding.
ToimplementADPCMtheoriginal(audio)signalissampledasforPCMtoproduceacode
word.Thiscodewordismanipulatedtoproducethepredictedcodewordforthenextsample.
Thenewpredictedcodewordiscomparedwiththecodewordofthesecondsample.Theresult
ofthiscomparisonissenttoline.ThereforeweneedtoperformPCMbeforeADPCM.
TheADPCMwordrepresentsthepredictionerrorofthesignal,andhasnosignificanceitself.
Insteadthedecodermustbeabletopredictthevoltageoftherecoveredsignalfromthe
previoussamplesreceived,andthendeterminetheactualvalueoftherecoveredsignalfrom
thispredictionandtheerrorsignal,andthentoreconstructtheoriginalwaveform.
ADPCMissometimesusedbytelecomoperatorstofittwospeechchannelsontoasingle
64kbit/slink.Thiswasverycommonfortransatlanticphonecallsviasatelliteupuntilafew
yearsago.Now,nearlyallcallsusefibreopticchannelsat64kbit/s.
12 03/07/2017
8. Digital Transmission Hierarchies
Overanygivenroute,themostcosteffectivetransmissionwill,ingeneral,beachievedby
multiplexingasmanyaspossibleofthePCMchannelstogether.Astechnologyhasadvanced,
ithasbecomefeasibletocombineanincreasingnumberofchannelsbyoperatingatever
highersignallingrates.Someroutes,ofcourse,willonlyneedafewchannels,sowillnot
requireveryhighsignallingrates.
Transmissionnetworksinthepublicswitchedtelephonenetwork(PSTN)aredesignedaround
hierarchiesoftransmissionrates,correspondingtoincreasingnumbersofchannelsconveyed
onasinglemultiplexedlink.Thesehierarchiesaredefinedinnationalandinternational
standards.InEuropethehierarchyisbaseduponthe30channel2,048kbit/sprimaryrate,
whileintheUSAandJapanitisbaseduponthe24channel1,544kbit/sprimaryrate.Recall
thateachPCMchanneloperatesat64kbit/s.Inaddition,ITUTdefinedtheworldstandard
knownasSynchronousDigitalHierarchy(SDH).Thisisdesignedtointerfacewithboththe
2048kbit/sand1544kbit/shierarchies.
13 03/07/2017
9. Plesiochronous Digital Hierarchy (PDH)
Theprimaryrateinformationcanbefurthermultiplexedontoevenhigherdataratesas
showninthetablebelow.
Digital US European Frequency
hierarchy hierarchy hierarchy Tolerance
level (kbit/s) (ppm)
(kbit/s)
1 1,554 2,048 50
2 6,312 8,448 30
3 44,736 34,364 20
4 139,264 15
Hierarchy1(primaryrate)carries30speechchannels+2overheadchannels.
Hierarchy2carriesfourprimaryrates+overheadbits =120speechchannels.
Hierarchy3carriesfourhierarchy2s+overheadbits =480speechchannels.
Hierarchy4carriesfourhierarchy3s+overheadbits =1920speechchannels.
14 03/07/2017