资源描述:
《[2002]Particle Filters for Tracking an Unknown Number of Sources.pdf》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、2926IEEETRANSACTIONSONSIGNALPROCESSING,VOL.50,NO.12,DECEMBER2002ParticleFiltersforTrackinganUnknownNumberofSourcesJean-RenéLarocque,Member,IEEE,JamesP.Reilly,Member,IEEE,andWilliamNg,StudentMember,IEEEAbstract—Thispaperaddressestheapplicationofsequentialim-Inrecent
2、years,therehavebeenseveralmethodsdevelopedportancesampling(SIS)schemestotrackingdirectionsofarrivalforestimatingortrackingtheDOAsofmovingtargetsusing(DOAs)ofanunknownnumberofsources,usingapassivearraypassivesensorsorarraysofsensors,e.g.,[9]–[11],etc.Liketheofsensor
3、s.Thisproposedtechniquehassignificantadvantagesinhigh-resolutionmethods,theseapproachesalsoassumethetar-thisapplication,includingtheabilitytodetectachangingnumberofsignalsatarbitrarytimesthroughouttheobservationperiodandgetsarestationaryoveralimitedtimeinterval.The
4、approachthattherequirementforquasistationarityoveralimitedintervalin[10]isbasedonadaptivelyestimatinganoisesubspacebasismayberelaxed.fromthereceivedsignalcovariancematrix.ThesemethodsthenWeproposetheuseofareversiblejumpMonteCarloMarkovrelyonahigh-resolutiontechniqu
5、esuchasMUSIC[5]toesti-chain(RJMCMC)steptoenhancethestatisticaldiversityofthematedthedesiredDOAs.In[9],amethodbasedonmaximumparticles.Thisstepalsoenablesustointroducetwonovelmovesthatsignificantlyenhancetheperformanceofthealgorithmwhenlikelihoodestimationofanovelsta
6、te-spacerepresentationfortheDOAtrackscross.Thesuperiorperformanceofthemethodistrackingispresented.demonstratedbyexamplesofapplicationoftheparticlefiltertoAnimportantconsiderationintargettrackingproblemsisthesequentialtrackingoftheDOAsofanunknownandnonstationarydata
7、associationproblem,i.e.,theassociationoftrackswithnumberofsourcesandtoascenariowherethetargetscross.Ourmeasurements.InthecasewherepassivearraysofsensorsareresultsarecomparedwiththePASTdmethod.used,thedataassociationproblemreducestotheassociationIndexTerms—Arraysign
8、alprocessing,Bayesianestimation,oftargetsbeforeandaftertheirDOAtrackscrosseachother.modelorderdetection,particlefilters,tracking.In[11],amethodfo