简介：Inanactiveradar-trackingsystem,thetarget-motionmodelisusuallymodeledintheCartesiancoordinates,whiletheradarmeasurementusuallyisobtainedinpolar/sphericalcoordinates.Thereforethetarget-trackingproblemintheCartesiancoordinatesbecomesanonlinearstateestimationproblem.Anumberofmeasurement-conversiontechniques,whicharebasedonpositionmeasurements,arewidelyusedsuchthattheKalmanfiltercanbeusedintheCartesiancoordinates.However,theyhavefundamentallimitationstoresultinfilteringperformancedegradation.Infact,inadditiontopositionmeasurements,theDopplermeasurementorrangerate,containinginformationoftargetvelocity,hasthepotentialcapabilitytoimprovethetrackingperformance.AfilterisproposedthatcanuseconvertedDopplermeasurements(i.e.theproductoftherangemeasurementsandDopplermeasurements)intheCartesiancoordinates.ThenovelfilteristheoreticallyoptimalintheruleofthebestlinearunbiasedestimationamongalllinearunbiasedfiltersintheCartesiancoordinates,andisfreeofthefundamentallimitationsofthemeasurement-conversionapproach.Basedonsimulationexperiments,anapproximate,recursiveimplementationofthenovelfilteriscomparedwiththoseobtainedbyfourstate-of-the-artconversiontechniquesrecently.Simulationresultsdemonstratetheeffectivenessoftheproposedfilter.