简介：Inthispaper,alowcomplexityESPRITalgorithmbasedonpowermethodandOrthogo-nal-triangular(QR)decompositionispresentedfordirectionfinding,whichdoesnotrequireaprioriknowledgeofsourcenumberandthepredeterminedthreshold(separatesthesignalandnoiseei-gen-values).Firstly,accordingtotheestimationofnoisesubspaceobtainedbythepowermethod,anovelsourcenumberdetectionmethodwithouteigen-decompositionisproposedbasedonQRde-composition.Furthermore,theeigenvectorsofsignalsubspacecanbedeterminedaccordingtoQmatrixandthenthedirectionsofsignalscouldbecomputedbytheESPRITalgorithm.Todeterminethesourcenumberandsubspace,thecomputationcomplexityoftheproposedalgorithmisapproximatedas(2log_2n+2.67)M~3,wherenisthepowerofcovariancematrixandMisthenumberofarrayele-ments.ComparedwiththeSingleVectorDecomposition(SVD)basedalgorithm,ithasasubstantialcomputationalsavingwiththeapproximationperformance.Thesimulationresultsdemonstrateitseffectivenessandrobustness.