简介:Forthedirectionofarrival(DOA)estimation,traditionalsparsereconstructionmethodsforwidebandsignalsusuallyneedmanyiterationtimes.Forthisproblem,anewmethodfortwo-dimensionalwidebandsignalsbasedonblocksparsereconstructionisproposed.First,aprolatespheroidalwavefunction(PSWF)isusedtofitthewidebandsignals,thentheblocksparsereconstructiontechnologyisemployedforDOAestimation.Theproposedmethodusesorthogonalizationtochoosethematchingatoms,ensuringthattheresidualcomponentscorrespondtotheminimumabsolutevalue.Meanwhile,thevectorsobtainedbyiterationareback-disposedaccordingtothecorrespondingatomicmatchingrules,sotheextraatomsareabandonedinthecourseofiteration,andtheresidualcomponentsofcurrentiterationarereduced.Thustheoriginalsparsesignalsarereconstructed.Theproposedmethodreducesiterationtimescomparingwiththetraditionalreconstructionmethods,andtheestimationprecisionisbetterthantheclassicaltwo-sidedcorrelationtransformation(TCT)algorithmwhenthesnapshotissmallorthesignal-to-noiseratio(SNR)islow.
简介:Thetraditionalsuper-resolutiondirectionfindingmethodsbasedonsparserecoveryneedtodividetheestimationspaceintoseveraldiscreteanglegrids,whichwillbringthefinalresultsomeerror.Tothisproblem,anovelmethodforwidebandsignalsbysparserecoveryinthefrequencydomainisproposed.Theoptimizationfunctionsarefoundandsolvedbythereceiveddataateveryfrequency,onthisbasis,thesparsesupportsetisobtained,thenthedirectionofarrival(DOA)isacquiredbyintegratingtheinformationofallfrequencybins,andtheinitialsignalcanalsoberecovered.Thismethodavoidstheerrorcausedbysparserecoverymethodsbasedongriddivision,andthedegreeoffreedomisalsoexpandedbyarraytransformation,especiallyithasapreferableperformanceunderthecircumstancesofasmallnumberofsnapshotsandalowsignaltonoiseratio(SNR).