#DTT17: Digitalisaatio ja tulevaisuuden teollisuustyö, Eija Kaasinen VTT
Eija c It 120053
Transcript of Eija c It 120053
-
7/30/2019 Eija c It 120053
1/11
Research Article ISSN 22779140
ReceivedonMay2012,PublishedonJuly2012 405
INTERNATIONAL JOURNAL OF ADVANCES IN
COMPUTING AND INFORMATION
TECHNOLOGYAn International online open access peer reviewed journal
BERReductioninOFDMSystemsusingZCTMatrixTransform
RohitGarg1,VijayKumar
2
1Lecturer,SKIET,Kurukshetra,INDIA2Lecturer,MaharishiMarkandeshwarUniversity,Mullana(Ambala),India
doi:10.6088/ijacit.12.14009
ABSTRACT
In recent years, Orthogonal Frequency Division Multiplexing (OFDM) is preferred for transmission over a
dispersivechannel.Itisbecauseofitsseveraladvantagessuchashighspectralefficiency,lowimplementation
complexity,lessvulnerabilitytoechoesandnonlineardistortion.However,ithasfewlimitationssuchaspeak-
to-average-powerratio(PAPR)andbiterrorrate(BER),whichdeterminesthesystemspowerefficiency.HighPAPRisoneofthemajordrawbacksinOFDMsystems,whichcausesasignificantlevelofsignaldistortions,
whenthemodulatedsignalsare amplified throughhigh power amplifiers (HPAs). Inparticularly,high PAPR
increasesthecomplexityofAnaloguetoDigital(A/D)andDigitaltoAnalogue(D/A)convertersandalsoreduces
theefficiencyofHPAs.AhighPAPRandhighnoise(e.g.AWGN)maysignificantlydistortthesignal,resulting
inhighBERondemodulationofthereceivedsignal.Therefore,PAPR&BERreductionisveryessential.Forthispurpose,a Zadoff-Chumatrix Transform(ZCT) basedprecodingandpostcoding techniquesarequantitatively
evaluated in this paper in terms of its parameters q and r.From the analysis, it is inferred that anoptimumselectionofqandrcanleadtoasignificantlevelofreductioninBER.
Keywords:Zadoff-ChumatrixTransform;OFDMconventional;HPA;PeaktoAveragePowerRatioandBER.
1. IntroductionOFDM is a key multicarrier modulation technique [1], which provides high spectral efficiency, low
implementationcomplexity[2],lessvulnerabilitytoechoesandnonlineardistortion[3].Duetotheseadvantages
oftheOFDMsystem,itisvastlyusedinvariouscommunicationsystems.However,ithasapracticallimitationof
highbiterrorrateoccurringduetohighPAPRornoise[4].AlargePAPRincreasesthecomplexityoftheanalog
todigital and digitaltoanalog converter and reduces the efficiency of the radio frequency (RF) power
amplifier [5].There are a numberof techniques dealingwith the problem of BER& PAPR. Some of these
include:constellationshaping,nonlinearcompandingtransforms[6],tonereservation[7]andtoneinjection(TI),
clipping and filtering [8], partial transmit sequence [9] and precoding based techniques. However, these
techniquesdoPAPRreductionattheexpenseofincreaseintransmitsignalpower,biterrorrate[7],datarateloss
and computational complexity. The reduction in BER is very essential. Therefore, two novel reduction
techniques,namely,Zadoff-Chumatrix Transform(ZCT)precodingbasedBER reductiontechniqueandZCT
postcodingbasedBER reduction technique for OFDMsystemswere proposed in this paper. In the proposed
schemes, the reshaping oftheZCT iscarriedoutoneway for precoding and anotherway for postcoding.Forprecoding,thereshapingoftheZCTmatrixisrowwiseandisappliedbeforetheIFFT;however,forpostcoding,
thereshapingoftheZCTmatrixisdonecolumnwiseanditisappliedaftertheIFFT.
-
7/30/2019 Eija c It 120053
2/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
406
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
2.SystemModel
2.1OFDMSystemModel
TodesigntheproposedZCTprecodingandpostcodingsystem,considertheblockdiagramofanOFDMsystemshown inFigure 1. In Figure1,anOFDMsignal consistsofN subcarriersthat aremodulatedbyN complex
symbolsselectedfromaparticularQAMconstellation.
Figure1:BlockdiagramofgeneralOFDMsystem.
InFigure1, the basebandmodulated symbolsarepassed through serial toparallelconverter which generates
complexvectorofsizeN.ThiscomplexvectorofsizeNcanbeexpressedas
(1)
XisthenpassedthroughtheIFFTblockofsizeNN IFFTmatrix.TheresultedcomplexbasebandOFDMsignalwithNsubcarrierscanthenbewrittenas
n=0,1,...,N1.
(2)
Afterparallel-to-serialconversion,acyclicprefix(CP)witha lengthofNgsamplesisappendedbeforetheIFFT
outputtoformthetime-domainOFDMsymbol,s=[s0,.. ,sN+Ng1].TheusefulpartofOFDMsymboldoesnot
include theNg prefix samples and has duration ofTu seconds. The samples (s) are then amplified, with the
amplifiercharacteristicsisgivenbyfunctionFandtheoutputofamplifierproducesasetofsamplesdenotedbyy.
Atthereceiverfrontend,thereceivedsignalisappliedtoamatchedfilterandthensampledatarateTs=Tu/N.After dropping the CP samples (Ng), the received sequencez, assuming an additive white Gaussian noise
(AWGN)channel,canbeexpressedas
z=F(Wd)+ (3)
Where,thenoisevectorconsistsofNindependentandnormallydistributedcomplexrandomvariableswithzero
mean.Subsequently, thesequencez isfedtothefastFouriertransform(FFT),whichproducesthefrequency-
domainsequenceras
r=WHz (4)
Where,kthelementofrisgivenby
-
7/30/2019 Eija c It 120053
3/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
407
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
k=0,1,2,,N-1
(5)
Finally,theestimatedsymbolsvector canbeobtainedfromr.Itistobenotedthatideally,thedemodulationisperformed based on the assumption of perfect symbol timing, carrier frequency, and phase synchronization.
Howeveritisnotpossibletoachieveinpractice;therefore,theBERisintroducedandwhichcanbedefinedasthe
differenceoftheobtaineddemodulatedsignalrkfromtheinputsignal.
2.2ZADOFF-CHUSEQUENCES
Ingeneral,aZadoffChusequenceisacomplex-valuedsequencewhich,whenappliedtoradiosignals,givesrisetoanelectromagneticsignalofconstantamplitude,wherebycyclicallyshiftedversionsofthesequenceimposed
on a signal result in zero cross-correlation with one another at the receiver. Mathematically, ZadoffChu
sequencescanbedefinedforasequenceoflengthLas:
(6)
Where,k=0,1,2L-1,qisanyintegerandrisanyintegerrelativelyprimetoL.Fromtheaboveequation,q
and rcomesinthenumeratoroftheexponentialpowerfraction.Thismeans,thattheirvariationcanleadtoa
significantchangeintheoutput.Thisisthekeyfocusoftheresearchreportedinthispaper.ThisZCTsequence
canbeappliedintwoforms:pre-codingandpost-coding,asdiscussedbelow.
2.3ZCTPRE-CODINGBASEDOFDMSYSTEM
TheblockdiagramofZCTpre-codingbasedOFDMsystemisshowninFigure2.IntheZCTpre-codingbased
OFDMsystem,thebasebandmodulateddataispassedthroughS/Pconvertorwhichgeneratesacomplexvector
ofsizeNthatcanbewrittenasX=[X0,X1,X2XN-1]T.ZCTpre-codingisthenappliedtothiscomplexvector
whichtransformsthiscomplexvectorintonewvectoroflengthNthatcanbewrittenasY=[Y0,Y1,Y2YN-1]T;
where,RisaZCTbasedrow-wiseprecodingmatrixofsizeL=N*N.Byreordering,
(7)
AndthematrixRwithrowwisereshapingcanbewrittenas
(8)
Where,RisaN*N,ZCTcomplexorthogonalmatrixwithlength L2=N*N.Accordingly,pre-codingXgivesrise
toYasfollows:
Y=RX (9)
Or,
(10)
Where, means themth rowand l
th column ofpre-codermatrix. Therefore, the complexbasebandOFDM
signalwithNsubcarrierswithZCTpre-codingisgivenby
-
7/30/2019 Eija c It 120053
4/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
408
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
(11)
Theaboveequationcanbeexpressedas theIFFTofconstellationdataXlpre-multipliedwithZCTmatrix.The
BERofthissignalissameasbefore.
Figure2:BlockdiagramofanOFDMsystemwithZCTpre-codingapproach.
2.4ZCTPOST-CODINGBASEDOFDMSYSTEMTheblockdiagramofZCTpost-codingbasedOFDMsystemisshowninFigure3.IntheZCTpost-codingbased
OFDMsystem,thebasebandmodulateddataispassedthroughS/Pconvertorwhichgeneratesacomplexvector
ofsizeNthatcanbewrittenasX=[X0,X1,X2XN-1]T.Then,IFFTisperformedtothiscomplexvectorwhich
transformsthiscomplexvectorintonewvectoroflength Nthatcanbewrittenasy=IFFT{X}=[y0,y1,y2yN-
1]T.Similarly,re-orderingasinpre-codingtohave:
(12)
And,ZCTmatrixCwithcolumnwisereshapingiswrittenas
(13)
Inotherwords,theN2pointlongZadoff-Chusequencefillsthepre-codingmatrixcolumn-wise.Thematrix Cis
N*N, ZCT complex orthogonal matrix with length L2 =N*N. Accordingly, postcodingy gives rise to w as
follows:
w=C.y (14)
Where,CisaZCTcolumnwisepost-codingmatrixofsizeL2=N*N.Forq=1andr=1,thecomplexbaseband
ZCTpostcodingbasedOFDMsignalwithNsubcarrierscanbewrittenas:
-
7/30/2019 Eija c It 120053
5/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
409
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
(15)
Where,wmistheIFFTofconstellationdataXlpremultipliedwithquadraticphaseandIFFTpostcoded,andthenalternatedwith1.TheBERofthisZCTpostcodingbasedOFDMsignalissameasbefore.
Figure3:BlockdiagramofanOFDMsystemwithZCTpost-codingapproach.
3.ImplementationTodotheBERanalysisandtoevaluatetheperformanceofZCTprecodingandpostcodingbasedOFDMsystems,
thestepsundertakenareshowninFigure4andaredetailedbelow.
1. TodotheBERanalysisofZCTprecodingandpostcodingbasedOFDM,firstlybinarydataisgeneratedrandomly.
2. ThegeneratedbinarydataisthenconvertedintosymbolsandisthenmodulatedbyM-QAM(whereM=4,
16,64,256).TheZCTmatrixiscalculatedforthelengthofthedata.
3. TheserialdataisthenconvertedintoparalleldataandIFFTisperformed.TheZCTmatrix(Transform)isappliedbeforeandaftertheIFFT,respectivelyforprecodingandpostcoding.
4. TheparallelsignalisthenconvertedtoserialdataandispassedthroughamultipathchannelwithAWGNnoiseadded.
5. Thereceivedsignalisagainconvertedfromserialtoparalleldata.ZCTinversematrixisappliedbeforeandaftertheFFT.Thefinaldemodulatedsignalisparalleldata,whichisthenconvertedintoserialdata.
The serial data is then demodulated to retrieveback the signal. The BER iscalculated by taking the
differenceofthedemodulateddataandtheinputdata.
-
7/30/2019 Eija c It 120053
6/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
410
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
Figure4:Implementationprocedure
4.ResultsandDiscussionsToevaluatetheperformanceofZCTprecodingandpostcodingbasedOFDMsystems,extensivesimulationshave
been performed in Matlab. The simulationsare performed for different values ofq andr. To show theBERanalysisofZCTprecodingandpostcodingbasedOFDMsystems,dataisgeneratedrandomlyandthenmodulated
byM-QAM(whereM=4,16,64,256).TheorderofFFTandIFFTistakenas64.Theresultsobtainedare
showninFigures5to11andTable1,forvariouscombinationsofqand r,andEBN0from1to7.Thekey
conclusionsdrawnfromtheresultsarediscussedbelow:
1. FromFigures5to 11, itcanbeinterpreted that the resultsobtainedforZCTprecodingandpostcodingaredifferent.TheBERismorereducedincaseofZCTpostcodingforthevaluesofEBN0(1to7),andq=1,10andr=11and111andM=4,16,64and256.
2. FromFigures5to11,itcanbeinterpretedthattheBERdecreaseswithincreaseinEBN0,forallthreecasesi.e.nocoding,ZCTprecodingandpostcoding.Itisbecause,withincreaseinEBN0,thenoisewilldecrease
andthustherewillbelessnumberofbitsinerror,whichmeanslowBER.
3. FromFigures5to11,forZCTpostcoding,theBERismorereducedwhenvalueofrisincreasedfrom11to111,foranyvalueofq.ThiscanalsobeseenfromdatainTable1.ItisbecauseinZCTpostcoding,thematrix
isappliedcolumn-wisewhichwillleadtohigherreductioninPAPRandthusmorereductioninBER.
-
7/30/2019 Eija c It 120053
7/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
411
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
4. FromFigures5to11,forZCTprecoding,theBERisalsosignificantlyreducedwithincreaseinconstellationsizeforanygivenSNR.
5.ItcanalsobeseenthatZCTpostcodinghasquickandsharprolloffthanZCTpre-coding.Also,thereislessirregularityinZCTpostcodingcomparedtoZCTprecoding.
6. As the constellation size increases, i.e., from M = 4, 16, 64, 256; there is significant reduction in BER;however,notwithoutcoding.Therefore,ZCTprecodingandpostcodinghaspotentialforsignificantreduction
inBER.
7. Itcanbealsoseenthatforsomecases,theBERforZCTpostcodingisnotobserved.ItisbecauseitisveryefficientandhasquicklyreducedtheBERtozero.
From the above discussion, the value of q and r significantly changes the BER for any given EBN0 and
constellationsizeM.Also,forsamevaluesofqandr,thevariationinconstellationsizealsosignificantlychanges
theBER. It isalso found that ZCT postcoding is more effective than ZCT precoding in reducing theBER.
However, ZCT precoding and postcoding and its parameters q and r significantly reduces the PAPR also.
Therefore,theselectionofvalueofq,r,MandZCTprecoding/postcodingdependsuponthelevelofPAPRand
BERreductionrequired.Hereinthispaper,itispreferredtochooseasmallvalueofqandlargevalueofrfor
ZCTpostcodingandbothsmallvaluesofqandrforZCTpre-coding.ItispreferredtochooseZCTpostcoding
becauseofitssharproll-offinreducingBER;however,italsodependsuponthedesiredlevelofreductionin
PAPR.
ConstellationSize(M=4)
Figure5:Thesimulatedbinarydataanditsconversationtosymbolsforconstellationsize(M=4)
Figure6:ThesimulateddatamodulatedusingQAM(M=4)andAWGNnoise(EBN0=1,2and3).
-
7/30/2019 Eija c It 120053
8/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
412
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
Figure7:ThesimulateddatamodulatedusingQAM(M=4)andAWGNnoise(EBN0=4,5and6)
ConstellationSize(M=4)
Figure8:ThevariationforBERwithEBN0forQAM
(M=4)andwithZCTprecodingandpostcodingandq=1
andr=11.
Figure9:ThevariationforBERwithEBN0for
QAM(M=4)andwithZCTprecodingand
postcodingandq=10andr=111.
ConstellationSize(M=16)
Figure10:ThevariationforBERwithEN0forQAM(M=16)andwithZCTprecodingandpostcodingand
q=1andr=11.
Figure11:ThevariationforBERwithEN0forQAM(M=16)andwithZCTprecodingand
postcodingandq=10andr=111.
-
7/30/2019 Eija c It 120053
9/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
413
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
ConstellationSize(M=64)
Figure12:ThevariationforBERwithEN0forQAM(M=64)andwithZCTprecodingandpostcodingand
q=1andr=11.
Figure13:ThevariationforBERwithEN0forQAM(M=64)andwithZCTprecodingand
postcodingandq=10andr=111.
ConstellationSize(M=256)
Figure14:ThevariationforBERwithEBN0forQAM
(M=256)andwithZCTprecodingandpostcodingand
q=1andr=11.
Figure15:ThevariationforBERwithEBN0for
QAM(M=256)andwithZCTprecodingand
postcodingandq=10andr=111.
5.Conclusion
ThispaperpresentedtheresultsobtainedonapplyingaZCTpre-codingandpost-codingforreducingBERinOFDMsystemswithM-QAMmodulation,whereasM=4,16,64,256andtheparametersq=1,10andr=11and
111.ItisobservedfromthegraphsthattheresponseofbothZCTpre-codingandpost-codingisdifferentforrand
q.TheBERisfoundtosignificantlyvarywiththevaluesofM,randqforagivenEBN0.Foranyparticularvalue
ofr,qandM,theBERreductionismoreforZCTpost-codingthanforZCTpre-coding.Fromtheresults,itis
preferredtochooseZCTpostcodingbecauseofitssharproll-offandquickreductioninBERforagivenSNR
andM.
-
7/30/2019 Eija c It 120053
10/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
414
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
Table1:VariationofBitErrorRatewithEBN0
Constellation
Size
q r EBN0
1 2 3 4 5 6 7
None
BitsinError 239 135 89 35 23 6 3
BER 0.0478 0.0270 0.0178 0.0070 0.0046 0.0012 0.0006
ZCTPostcoding
BitsinError 66 22 16 0 0 0 0
BER 0.0132 0.0044 0.0032 0 0 0 0
ZCTPrecoding
BitsinError 177 108 46 17 8 0 0
4
1
11
BER 0.0354 0.0216 0.0092 0.0034 0.0016 0 0
None
BitsinError 240 145 102 38 21 5 0
BER 0.0480 0.0290 0.0204 0.0076 0.0042 0.0010 0
ZCTPostcoding
BitsinError 16 0 0 0 0 0 0
BER 0.0032 0 0 0 0 0 0
ZCTPrecoding
BitsinError 157 92 39 4 0 0 0
4
10
111
BER 0.0314 0.0184 0.0078 0.0008 0 0 0
None
BitsinError 1411 1355 1274 1179 1110 1008 945
BER 0.2822 0.2710 0.2548 0.2358 0.2220 0.2016 0.1890
ZCTPostcoding
BitsinError 0 0 0 0 0 0 0
BER 0 0 0 0 0 0 0
ZCTPrecoding
BitsinError 26 7 0 0 0 0 0
256
1
11
BER 0.0052 0.0014 0 0 0 0 0
NoneBitsinError 1388 1340 1307 1190 1090 951 943
BER 0.2776 0.2680 0.2614 0.2380 0.2180 0.1902 0.1886
ZCTPostcoding
BitsinError 0 0 0 0 0 0 0
BER 0 0 0 0 0 0 0
ZCTPrecoding
BitsinError 37 15 7 0 0 0 0
256
10
111
BER 0.0074 0.0030 0.0014 0 0 0 0
-
7/30/2019 Eija c It 120053
11/11
BERReductioninOFDMSystemsusingZCTMatrixTransform
415
RohitGarg,VijayKumarInternationalJournalofAdvancesinComputingandInformationTechnology
References
1. E.A.-Dalakta,A.A.-Dweik&C.Tsimenidis(2012).EfficientBERReductionTechniqueforNonlinear
OFDMTransmissionUsingDistortionPrediction.IEEETransactionsOnVehicularTechnology.Vol.61,NO.5,pp.230-36.
2. H.-B. Jeon, J. S. and D. J. Shin (2011).A Low-Complexity SLM SchemeUsingAdditiveMappingSequencesforPAPRReductionofOFDMSignals.IEEETransactionsOnBroadcasting.Vol.57,NO.4,
pp.866-75.
3. GhassemiandTA.Gulliver(2007).SLM-BASEDRADIXFFTFORPAPRREDUCTIONINOFDMSYSTEMS.IEEEInternationalConferenceonSignalProcessingandCommunications(ICSPC2007).pp
1031-34.
4. C.P.Li,S.H.WangandC.L.Wang(2010).NovelLow-ComplexitySLMSchemesforPAPRReductioninOFDMSystems.IEEETransactionsOnSignalProcessing.Vol.58,NO.5,pp2916-21.
5. I.BaigandV.Jeoti(2010).PAPRReductioninOFDMSystems:Zadoff-ChuMatrixTransformBasedPre/Post-Coding Techniques. Second International Conference on Computational Intelligence,
CommunicationSystemsandNetworks.
6. H.Liang,Y.R.Chen,Y.F.Huang,C.H.Cheng(2009).AModifiedGeneticAlgorithmPTSTechniquefor PAPR Reduction in OFDM Systems. Proceedings of the 15th Asia-Pacific Conference on
Communications(APCC2009).pp182-185.
7. H.Li,T.JiangandY.Zhou(2011).AnImprovedToneReservationSchemeWithFastConvergenceforPAPRReductioninOFDMSystems.IEEETransactionsOnBroadcasting.Vol.57,No.4,pp.902-06.
8. L.WangandC.Tellambura(2005).ASimplifiedClippingandFilteringTechniqueforPARReductioninOFDMSystems.SignalProcessingLetters,IEEE.Vol.12,No.6,pp.453-456.
9. P.VarahramandB. M.Ali (2011). PartialTransmit SequenceSchemewithNew Phase SequenceorPAPRReductioninOFDMSystems.IEEETransactionsonConsumerElectronics.Vol.57,No.2,pp.
366-71.