无线局域网中Turbo码的应用研究(英文)

UsingTurbo-CodeinWirelessLANWANGXiu-fang,WANGYang,WEIYu-heng(DaqingPetroleumInstitute,Daqing,Heilongjiang163318,China)Abstract:Turbo-codeusedinwirelessLAN(WLAN)isproposedbyanalyzingthedatatransmissionthroughWLAN,andanewcodinganddecodingschemeispresentedtoimprovetheperformanceofWLAN.TheperformanceofTurbocodewiththechangesofcorrelativeparametersisstudiedthroughcomputersimulations.PerformancecomparisonwithconvolutionalcodewhichiswidelyusedintraditionalWLANisconducted.SimulationresultsshowthatTurbo-codecangreatlyenhanceWLANperformance.Keywords:wirelessLAN;turbo-code;convolutionalcode无线局域网中Turbo码的应用研究王秀芳,王洋,魏宇恒(大庆石油学院,黑龙江大庆163318)摘要:通过分析无线局域网中数据传输的特点，对Turbo码编码和译码方案进行适合于WLAN的改进，设计出无线局域网中Turbo码编译码器的实现方案。通过计算机仿真研究相关参数的变化对Turbo码性能的影响，并与传统WLAN中应用广泛的卷积码进行性能比较，仿真结果表明，利用Turbo码编码技术能够较大地提高无线局域网传输性能。关键词:无线局域网；Turbo码；卷积码0IntroductionWLANtechnology,becauseofaseriesofadvantages,isgettingwidelyusedinglobalcommunicationsystems.Itbecomesanmoreattractivesolutionandconnectsmoreandmoreequipments,suchastheinterconnectingOfnotebookPC,cellphone,VoIP(VoiceoverInternetProtocol),PDA(PersonalDigitalAssistant)andsoon.Inordertoimprovethereliabilityofwirelesscommunicationssystemsandreducepowerconsumption,peoplehaveadoptedalotofmethodstooptimizedesigns.Amongthemethods,channelcodinganddecodingtechniqueisanimportantone.WiththeamendmentoftheprotocolsofWLAN,convolutionalcodingtechniquewithdifferentgeneratorpolynomialscan'tmeettherequirementoftransmissionspeed.Inthispaper,wepresentanewschemeforencoderanddecoderofTurbocodeinWLANbyanalyzingthecharacteristicsofdatatransmissioninordertomeettherequirementsofcommunicationqualitiesandtransmissionspeed,andprovidemorehighqualityserviceswithminimumcomplexityandlowestpowerconsumption.Firstly,weoverviewturbo-codeandselectthebestdecodingalgorithmfromfourkindsofdecodingalgorithms.Secondly,theprincipleandstructureofturbo-codeusedinWLANaredescribedindetail.Thesimulationsofbiterrorrate(BER)performancesofturbo-codeinanAdditiveWhiteGaussianNoise(AWGN)channelareperformedinSection4.InSection5,weconcludethispaper.1TurboCodeOverviewAnewschemewasintroducedbysomeFrenchresearchers,Berrou,GlavieuxandThitimajshimain1993[1].TheyusedaparallelconcatenatedschemewhichgotthenicknameTurbo-codeaftertheresemblancewiththeturboengine.Itwasoneofthemostimportantdevelopmentsinchannelcodingtheoryformanyyears.Aturbocodeisknownasalinearblockcode[2],whichadoptsasimplestructureofconcentralconvolutioncodeandMaximumaposterioriprobability(MAP)decodingalgorithm,hasattractedtheattentionofmanyresearchersforitsremarkablecodinggainsclosetoShannontheoreticallimitsinanAWGNchannel.Inthesimulation,framesizeis65535,interationnumberis18,theFrenchresearchersdecodedarater=1/3Turbo-codewithaso-callediterativedecodingalgorithm,theyclaimedthatabiterrorof10−5couldbeachievedatasingal-to-noiserate(SNR)of0.7dB.Turbo-codecanobtainhighperformance,lowconsumptionandlowcomplexity,soitisverysuitableforwirelesscommunicationssystems.2DecodingAlgorithmSelectionItiswellknownthatMAP[3]algorithmcancalculatetheaposterioriprobability(APP)oftheinformationsequencederictly,whichisthebestdecodingalgorithmforturbo-code.Butitishardtorealizeinpracticalapplication,becauseitneedsamassofcalculations.Log-MAPalgorithm[4]isthealgorithmwhichusesMAPinlogarithmfield,itsadvantageischangingmultiplyoperationtoaddoperation,soitcanreducethecalculationcomplexitygreatlyandbemorepractical,atthesame,itisapproximatelyequaltoMAPalgorithmondecodingperformance.Max-log-MAPalgorithmisasimplifiedversionoftheLog-MAPalgorithmthatusesonlyaddition,comparisonandselectionfunctions,whichisonlyanapproximationofLog-MAPalgorithm,soitcertainlywillleadtothedeclineondecodingperformance.AlthroughthecomplexityofLog-MAPalgorithmisabouttwicethatofSoftOutputViterbiAlgorithm(SOVA)[5],itisobviouslysuperiortoSOVAalgorithmonBERperformance.Throughcomparisonsabove,weselectLog-MAPalgorithmasdecodingalgorithminthispaper.Fig.1StructureofTurbo-CodeEncoderandDecoderFig.2StructureofTurbo-CodeEncoderandDecoder3PrincipleandStructureofTurbo-CodeinWLANFigure1showsthestructureofturboencoderanddecoder[6].Thetwoparallelconcatenatedconstituentencoders,interleaver,puncterandmultiplexerformtheturboencoder,asshowninFigure3(a).Theinterleaverisemployedtopermutetheinputbitssothatthetwoconstituentencodersworkonthesamebits,butdifferentsequences.Inthispaper,weadoptthatconstraintlengthKofeachrecursivesystematicconvolutional(RSC)componentcodeis4,twogeneratorpolynomialswithoctalform:g=(13,15)andrandominterleaver.Withthedelayofencodingandinterleavingnegligible,turboencoderhasthreeoutputscorrespondingtothe𝑘𝑡ℎsourcebit𝑑𝑘,oneistheinformationbit𝑥𝑘,theothersaretheparitybits𝑦1𝑘and𝑦2𝑘fromtherecursivesystematicconvolutional(RSC)encodersrespectively.Thesequencesof{𝑥𝑘,𝑦1𝑘,𝑦2𝑘}arepuncturedtoimprovetheefficiencyoftheencoder,andmultiplexedtoconstituetheoutputbits{𝑥𝑘,𝑦𝑘},fromwhichtheoutputsequences𝐶1𝑁={𝐶1,⋯,𝐶𝐾,⋯𝐶𝑁}oftheturboencoderareconstructed.Since𝐶𝑘+1isindependentof𝐶𝐾and𝑑𝐾,thesequencesaretreatedasdiscreteMarkovsourcetotheequivalentwireles