脉冲噪声环境下的韧性匹配滤波检测方法

256(332005)BPSKAlphaRobustDetectionMethodofMatchedFilteringinImpulsiveNoiseEnvironmentJIANGJin-long,ZHADai-feng,LIANGNing-li(CollegeofElectronicEngineering,JiujiangUniversity,Jiujiang332005)AbstractArobustdetectionmethodofMatchedFiltering(MF)isproposedtoimproveperformanceofconventionallocaloptimizationMFdetectioninimpulsivenoiseenvironment.Anonlineartransformofobservedsignalisfoundbyprobabilitydensityfunctionofimpulsivenoisedistribution,andMFdetectionisusedtotransformedsignal.Themethodisdulltocharacteristicexponentofimpulsivenoise,itcanrestraineffectivelyimpulsivenoise.SimulationresultsonMFdetectionofBPSKmodulationshowthattheproposedmethodhasbetterdetectionperformancethanconventionallocaloptimizationMFinimpulsivenoiseandisrobusttonoisecharacteristicexponent.Keywordsrobustdetection;MatchedFiltering(MF);nonlineartransform;AlphastabledistributionComputerEngineering3614Vol.36No.1420107July2010··10003428(2010)14025603ATN911.71(MatchedFilter,MF)(MMSE)α[1]α[2][3]2Alphaα4α(02α)βγa2α=αα[1]()(SSα)0,0aβ==αγ2α()exp{j||[1jsgn()(,)]}uauuuuαφγβωα=−+(1)tan(π)12(,)2()ln||]1πuuααωαα⎧≠⎪⎪=⎨⎪=⎪⎩(2)0γ2σ2/2γσ=02αa−∞∞[1,1]β∈−sgn()uuαXα(02α)αp[||]0[||]ppEXpEXpαα⎧∞⎪⎨=∞⎪⎩(3)X()(60772037)α(GJJ10266)2009GES006820092076(09YB472)(1969)[e-mail]257[4]11[ln||](1)lneEXCγαα=−+(4)2221111[ln||]()(1)ln62eEXCπγααα=++−+(5)(SSα)(1),()exp{||}uuααγφγ=−(6)(6)SSαj,,1()()ed2πufxuuωαγαγφ∞=∫−∞(7)(2α=)(1,0αβ==)(Pearson,1/2,1αβ==−)αSSα,()fxαγ[2]3α0011()()(),1,2,,()()(),1,2,,HxnsnwnnNHxnsnwnnN=+==+=(8)()isn1,2i=2()wnαγSSα()isn0H1H(9)Λ0111([()()])ln([()()])NnNnfxnsnfxnsnαα==⎧⎫−∏⎪⎪Λ=⎨⎬⎪⎪−∏⎩⎭011([()()])ln([()()])Nnfxnsnfxnsnαα=⎧⎫−⎪⎪=∑⎨⎬−⎪⎪⎩⎭(9)ηηΛ0()sn1()sn0η=BPSK(0()sn)01MF01()()HNnHsnxnη=Λ=∑(10)1()sn(LocalOptimization,LO)(9)01LO01'(())()()(())HNnHfxnXsnfxnη=Λ=−∑(11)'()()()fxgxfx=−(12)(12)()xn(11)()xn0()sn1()sn(7)(2α=)2()/(2)/gxxxγσ==(13)(1α=)22()/()gxxxγ=+(14)(02,1αα≠)()gx1α()gx[2]-4-3-2-101234-2-1.5-1-0.500.511.52'()/()fxfx−1.5α=2α=1.2α=0.8α=0.5α=1.0α=1α4ααM()mst1,2,,mM=T()ntSSααγ()rt()()()mrtstnt=+,0tT(15)α[5](12)(12)α01α(14)()()22'()/()sgxxxqPγ=++(16)sPq1~3sqP(14)5α(GeneralizedSNR,GSNR)[1]258210lg(/)dBsGSNRσγ=(17)2sσγα2/2γσ=GSNR3dB()mstBPSK6.25df=Hz12.5cf=Hz100sf=Hz10000N=1α=SSα02(14)(BER)(14)(16)0246810121416182010-410-310-210-1100BERofBPSKGSNR(dB)BERnononlinearnonlinearmodifiednonlinear21,1αγ==α(16)30123410-410-310-210-1BERofBPSKGSNR(dB)BERalpha=2.0,notransformalpha=2.0alpha=1.5alpha=1.031γ=α1α=(16)(2α=)α12GSNR0dB~3dB3dBα6αBPSK[1]ShaoM,NikiasCL.SignalProcessingwithFractionalLowerOrderMoments:StableProcessesandTheirApplications[J].ProceedingsoftheIEEE,1993,81(7):986-1010.[2]ShaoM,NikasCL.SignalDetectiononImpulsiveNoiseBasedonStableDistributions[C]//Proc.ofthe27thAsilomarConferenceonSignal,SystemsandComputers.PacificGrove,Canada:[s.n.],1993:218-222.[3]ChuahTC,SharifBS,HintonOR.RobustAdaptiveSpread-spectrumReceiverwithNeural-netPreprocessinginNon-GaussianNoise[J].IEEETransactionsonNeuralNetworks,2001,12(3):546-558.[4]ZhaDaifeng,QiuTianshuang.NewAnalysisModelandFrameworkofImpulsiveNoisewithHeavyTail[J].JournalofCommunicationandComputer,2008,38(1):1-8.[5]JiangJinlong,ZhaDaifeng.GeneralizedFractionalLower-orderSpectrumofAlphaStableDistributionProcess[C]//Proc.ofthe4thIEEEInternationalConferenceonWirelessCommunications,NetworkingandMobileComputing.Dalian,China:IEEEPress,2008.~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~(255)4.4ColdFireFlashColdFireFlash44FlashMCUPCFlash5ColdFireMCF52233Flash3ColdFireMCUColdFireFlash[1],.MC68HC908GP32Flash[J].,2002,19(7):15-19.[2],.UF32FlashU-ISP[J].,2010,36(2):56-59.[3],,.[J].,2009,26(29):76-79.[4],.[J].,2006,33(11):259-262.[5].[J].,2002,28(12):22-24.[6],,.JECM[J].,2007,24(7):194-197.