DOA-estimation-for-monostatic-MIMO-radar-based-on-

Thisarticlewasdownloadedby:[UniversityTownLibraryofShenzhen]On:04May2013,At:19:38Publisher:Taylor&FrancisInformaLtdRegisteredinEnglandandWalesRegisteredNumber:1072954Registeredoffice:MortimerHouse,37-41MortimerStreet,LondonW1T3JH,UKInternationalJournalofElectronicsPublicationdetails,includinginstructionsforauthorsandsubscriptioninformation::08Jan2013.Tocitethisarticle:WeiWang,XianpengWang,XinLi&HongruSong(2013):DOAestimationformonostaticMIMOradarbasedonunitaryroot-MUSIC,InternationalJournalofElectronics,DOI:10.1080/00207217.2012.751319Tolinktothisarticle::://dx.doi.org/10.1080/00207217.2012.751319DOAestimationformonostaticMIMOradarbasedonunitaryroot-MUSICWeiWang,XianpengWang*,XinLiandHongruSongCollegeofAutomation,HarbinEngineeringUniversity,Harbin150001,China(Received15January2012;finalversionreceived13October2012)Directionofarrival(DOA)estimationisanimportantissueformonostaticMIMOradar.ADOAestimationmethodformonostaticMIMOradarbasedonunitaryroot-MUSICispresentedinthisarticle.Inthepresentedmethod,areduced-dimensionmatrixisfirstutilisedtotransformthehighdimensionofreceivedsignaldataintolowdimensionone.Then,alow-dimensionreal-valuecovariancematrixisobtainedbyforward–backward(FB)averagingandunitarytransformation.TheDOAoftargetscanbeachievedbyunitaryroot-MUSIC.DuetotheFBaveragingofreceivedsignaldataandtheeigendecompositionofthereal-valuedmatrixcovariance,theproposedmethodownsbetterangleestimationperformanceandlowercomputationalcomplexity.Thesimulationresultsoftheproposedmethodarepresentedandtheperformancesareinvestigatedanddiscussed.Keywords:MIMOradar;DOAestimation;reduced-dimensionmatrix;unitarytransformation;unitaryroot-MUSIC1.IntroductionMultiple-inputmultiple-output(MIMO)radar(Fishleretal.2004),whichusesmultipleantennastotransmitorthogonalwaveformsratherthancoherentwaveformsintraditionalphased-arrayradarsandmultipleantennastoreceivethereflectedsignal,hasbeenwidelyinvestigatedinrecentyears,owingtoitspotentialadvantageovertheconventionalphased-arrayradars,suchasincreaseddegreesoffreedom(DOF)andangularresolution(BekkermanandTabrikian2006),betterparameteridentifiability(JianandStoica2007),andmoresensitivityfordetectingslowlymovingtargets(JianandStoica2007).Accordingtotheconfigurationofthetransmitterandreceiverantennas,MIMOradarcanbedividedintotwogroups.OneisthecloselyspacedMIMOradar(BekkermanandTabrikian2006;JianandStoica2007),includingbistaticMIMOradarandmonostaticMIMOradar,whosetransmitterantennasandreceiverantennasareclosedspaced.TheotheristhewidelyspacedMIMOradar(Fishleretal.2006;Haimovichetal.2008),inwhichatleastoneofthetransmitterantennasandreceiverantennasiswildlyspaced.Theformeraimsatobtainingvirtualaperturewhichislargerthanrealaperture,soitbringsnarrowerbeamwidthandlowersidelobesandprovideshigherresolutionspatialspectralestimationsandmoreDOFofMIMOradar.Thelattercanobtaincoherentprocessinggainforsolvingtargetscintillationproblem.InMIMOradar,asweknow,angleestimationisanimportantissueandalotofmethodsaredeveloped.Forexample,thedirectionofdeparture(DOD)anddirectionof*Correspondingauthor.Email:[e-mail]2013Taylor&FrancisDownloadedby[UniversityTownLibraryofShenzhen]at19:3804May2013arrival(DOA)estimationbasedonspatialspectrumpeaksearching(Yan,Li,andLiao2008;ZhangandXu.2010;Zhangetal.2010)arepresentedinbistaticMIMOradar,buttheperformanceofangleestimationdependsonsearchingspaceandsearchingresolution.Toavoidthespatialspectrumpeaksearching,theESPRITtechnique(DuoFang,Baixiao,andGuodong2008;Chen,Gu,andWeimin2008)isappliedtobothtransmitterarrayandreceiverarray,andtheDOAandDOAcanbeobtained.Additionally,aneffectivebeamspaceESPRITisdevelopedtoestimatetheDODandDOAoftargets(Guo,Zhang,andTong2011),buttheangleestimationperformancesignificantlydegradeswhenthebeamspacetransformationisinverylowdimension.Toachievedhigherangleestimationprecision,theDODandDOAcanbeobtainedusingthepolynomialrootfindingtechnique(Bencheikn,Wang,andHe2010),andtheDODandDOAoftargetisalsoautomaticpairing.Allthemethodsmen