简介:AspecializedHungarianalgorithmwasdevelopedhereforthemaximumlikelihooddataassociationproblemwithtwoimplementationversionsduetopresenceoffalsealarmsandmisseddetections.Themaximumlikelihooddataassociationproblemisformulatedasabipartiteweightedmatchingproblem.Itsdualityandtheoptimalityconditionsaregiven.TheHungarianalgorithmwithitscomputationalsteps,datastructureandcomputationalcomplexityispresented.Thetwoimplementationversions,Hungarianforest(HF)algorithmandHungariantree(HT)algorithm,andtheircombinationwiththenaveauctioninitializationarediscussed.ThecomputationalresultsshowthatHTalgorithmisslightlyfasterthanHFalgorithmandtheyarebothsuperiortotheclassicMunkresalgorithm.
简介:TheperformanceofthetraditionalVoiceActivityDetection(VAD)algorithmsdeclinessharplyinlowerSignal-to-NoiseRatio(SNR)environments.Inthispaper,afeatureweightinglikeli-hoodmethodisproposedfornoise-robustVAD.Thecontributionofdynamicfeaturestolikelihoodscorecanbeincreasedviathemethod,whichimprovesconsequentlythenoiserobustnessofVAD.Divergencebaseddimensionreductionmethodisproposedforsavingcomputation,whichreducesthesefeaturedimensionswithsmallerdivergencevalueatthecostofdegradingtheperformancealittle.ExperimentalresultsonAuroraIIdatabaseshowthatthedetectionperformanceinnoiseenvironmentscanremarkablybeimprovedbytheproposedmethodwhenthemodeltrainedincleandataisusedtodetectspeechendpoints.Usingweightinglikelihoodonthedimension-reducedfeaturesobtainscom-parable,evenbetter,performancecomparedtooriginalfull-dimensionalfeature.
简介:Duetotheopennessofthecognitiveradionetwork,spectrumsensingdatafalsification(SSDF)canattackthespectrumsensingeasily,whilethereisnoeffectivealgorithmproposedincurrentresearchwork,sothispaperintroducesthemalicioususersremovingtotheweightsequentialprobabilityradiotest(WSPRT).Theterminals'weightisweightedbytheaccuracyoftheirspectrumsensinginformation,whichcanalsobeusedtodetectthemalicioususer.Ifoneterminalownsalowweight,itcanbetreatedasmalicioususer,andshouldberemovedfromtheaggregationcenter.SimulationresultsshowthattheimprovedWSPRTcanachievehigherperformancecomparedwiththeothertwoconventionalsequentialdetectionmethodsunderdifferentnumberofmalicioususers.
简介:Toestimatethespreadingsequenceofthedirectsequencespreadspectrum(DSSS)signal,afastalgorithmbasedonmaximumlikelihoodfunctionisproposed,andthetheoreticalderivationofthealgorithmisprovided.Bysimplifyingtheobjectivefunctionofmaximumlikelihoodestimation,thealgorithmcanrealizesequencesynchronizationandsequenceestimationviaadaptiveiterationandslidingwindow.Sinceitavoidsthecorrelationmatrixcomputation,thealgorithmsignificantlyreducesthestoragerequirementandthecomputationcomplexity.Simulationsshowthatitisafastconvergentalgorithm,andcanperformwellinlowsignaltonoiseratio(SNR).
简介:Blindidentification-blindequalizationforFiniteImpulseResponse(FIR)MultipleInput-MultipleOutput(MIMO)channelscanbereformulatedastheproblemofblindsourcesseparation.Ithasbeenshownthatblindidentificationviadecorrelatingsub-channelsmethodcouldrecovertheinputsources.TheBlindIdentificationviaDecorrelatingSub-channels(BIDS)algorithmfirstconstructsasetofdecorrelators,whichdecorrelatetheoutputsignalsofsubchannels,andthenestimatesthechannelmatrixusingthetransferfunctionsofthedecorrelatorsandfinallyrecoverstheinputsignalusingtheestimatedchannelmatrix.Inthispaper,anewapproximationoftheinputsourceforFIR-MIMOchannelsbasedonthemaximumlikelihoodsourceseparationmethodisproposed.TheproposedmethodoutperformsBIDSinthepresenceofadditivewhiteGaussiannoise.