Research On Low-Complexity Algorithms Of DOA Estimation And DOA Tracking In Array Signal Processing

Direction of arrival(DOA)estimation and DOA tracking are the key technologies in array signal processing.Many current DOA estimation algorithms are proposed with high computational complexity,thus it is important to study low complexity algorithms for DOA estimation in different scenarios.At the same time,the research on DOA tracking is also significantly practical because the angle of a moving source usually changes with time in real scenarios.Based on previous analysis,this dissertation focused on low complexity DOA estimation and DOA tracking algorithms for multiple applications.The research topic is highly valued in both theory and application.Main contributions of this dissertation are as follows:(1)Based on reduced dimension(RD)and rank reduced(RARE)theories,two low complexity algorithms were proposed for near-field sources localization with uniform linear array.In the first RARE-Capon algorithm,the RARE characteristic of the angle associated matrix are utilized to change the two-dimensional spectral peak searching within the traditional Capon algorithm to one-dimensional peak searching,which remarkably reduces the computational complexity and can automatically obtain paired 2D-DOA estimation.The parameter estimation performance of this RARE-Capon algorithm is close to the traditional algorithm.In the second reduced dimension multiple signal classification(RD-MUSIC)algorithm,two-dimensional spectral peak searching is avoided by partitioning the angle information and range information of the direction matrix.Compared with the traditional MUSIC algorithm and spectral peak searching RARE algorithm,this RD-MUSIC algorithm has low computational complexity and almost equal parameter estimation performance.(2)Based on successive theory,three low complexity successive algorithms for 2D-DOA estimation were proposed.In the first successive propagator method(PM),based on uniform rectangular array,the rotational invariance of the propagator matrix is utilized first to obtain the initial angle estimation,then the accurate angle is estimated via one-dimensional local spectral peak searching,resulting in lower computational complexity compared with the traditional 2D-PM.In addition,the angle estimation performance of this algorithm outperforms the estimation of signal parameters via rotational invariance techniques(ESPRIT)and PM algorithm,close to 2D-PM.The second successive DSPE algorithm is proposed based on the research distributed signal parameters estimation(DSPE).In this algorithm,parameters of coherently distributed(CD)source are estimated via three steps local spectral peak searching thus gain high performance.The third one is an improved generalized ESPRIT(G-ESPRIT)algorithm proposed to the direction of departure(DOD)and DOA estimation for bistatic Multiple-Input Multiple-Output(MIMO)radar.Compared with the traditional G-ESPRIT algorithm which needs global peak searching and extra matching process,in this algorithm paired DOD and DOA estimation are obtained automatically via only one-dimensional local peak searching,thus to improve the performance.(3)Based on the RD transformation,two algorithms were proposed for the low complexity DOA estimation of the MIMO radar.In the first RD-PM algorithm for uniform linear MIMO radar,the DOA estimation can be done with no spectral peak searching requirements,thus low computational complexity and better angle estimation performance can be achieved.The second RD-ESPRIT algorithm is proposed for double parallel MIMO array with low computational load.Compared with ESPRIT algorithm,this method has better angle estimation performance,especially under low signal to noise ratio(SNR)condition.(4)Based on fast convergence FARAFAC(FC-PARAFAC)decomposition,two FC-PARAFAC DOA estimation algorithms were proposed.The first FC-PARAFAC algorithm was proposed for arbitrary acoustic vector sensor array with better 2D-DOA estimation performance and lower computational complexity than the traditional PARAFAC algorithm.The second FC-PARAFAC method was proposed for coprime array to eliminate the phase ambiguity.By applying initial estimations of the direction matrix via PM,lower computational complexity and closer angle estimation performance can be achieved compared to the traditional PARAFAC method.(5)Three low complexity DOA tracking algorithms were proposed for arbitrary acoustic vector sensor array.The first algorithm is based on projection approximation subspace tracking with deflation(PASTd)and has lower complexity with better angle tracking performance than traditional PAST method.The second proposed algorithm combines the Kalman filtering and traditional Orthogonal PASTd(OPASTd)to implement angle tracking.It has better DOA tracking performance than the traditional PASTd algorithm.The third proposed PARAFAC-RLST algorithm is based on the PARAFAC model and has lower complexity and closer DOA tracking performance compared with traditional PARAFAC algorithm.