简介：Radio-frequency（RF）tomographyisanemergingtechnologywhichderivestargetslocationinformationbyanalyzingthechangesofreceivedsignalstrength（RSS）inwirelesslinks.ThispaperpresentsandevaluatesanovelRFtomographysystemwhichiscapableofdetectingandtrackingatime-varyingnumberoftargetsinaclutteredindoorenvironment.ThesystemincorporatesanobservationmodelbasedonRSSattenuationhistogramandamulti-targettracking-by-detectionfilteringapproachbasedonprobabilityhypothesisdensity（PHD）filter.Inaddition,thesequentialMonteCarlomethodisappliedtoimplementthemulti-targetfiltering.Toevaluatethetrackingsystem,theexperimentsinvolvingupto3targetswereperformedwithinanobstructedindoorareaof70m^2.Theexperimentalresultsindicatethattheproposedtrackingsystemiscapableoftrackingatime-varyingnumberoftargets.