Study On DOA Estimation Algorithm Based On Compressive Sensing

Direction-of-Arrival (DOA) estimation, which is one of the most important research fields in smart antenna technology, is widely used in radar, communication and earthquake, etc. Especially in modern military war, accurate localization for signals is one of important processing for war. So, research on DOA estimation algorithm has very vital significance. Spatial spectrum estimation based on the array is the main method of DOA estimation due to its high resolution. However, this kind of methods requires many samples, which is too hard to achieve in real-time and it has bad estimation performance for coherent signals and in low signal-to-noise ratio (SNR). It is necessary to develop new DOA estimation method to satisfy the actual needs. As a new analytical idea of signal, compressive sensing (CS) theory has broken through the limitation of traditional Nyquist sampling theorem, has been widely applied in many fields and has many excellent qualities. In this paper, compressive sensing theory is applied to DOA estimation, a model around space narrowband signal is given, and DOA estimation algorithm of space narrowband signal is presented based on compressive sensing. The main contributions in this paper are illustrated as follows:(1) The basic principle of compressive sensing is elaborated. Three core problems of compressive sensing have been analyzed, including sparse representation of signal, compressive measurement of signal and recovery algorithm of signal.(2) The existing DOA estimation approach based on compressive sensing can not obtain sparser solution, which degrades DOA estimation performance. Aiming at this problem, a weighted-L1-SRACV algorithm based on sparse representation of array covariance vectors is presented in this paper. By making use of the orthogonality between noise subspace and signal subspace spanned by the array manifold matrix, a weighted matrix is constructed to constrain the sparse vector which is going to be recovered. This operation can ensure that the recovered sparse vector has sparser solution. The proposed method can effectively suppress spurious peaks and obtain better DOA estimation performance.(3) The existing DOA estimation method based on compressive sensing has high computational complexity and strong sensitivity to the assumed number of signals. Aiming at this issue, a weighted-L1-SVD approach based on sparse representation of array receiving data vector is developed by utilizing the sparsity of signal in space domain in this thesis. Signal subspace and noise subspace can be acquired by taking singular value decomposition (SVD) on array receiving data matrix. By utilizing the orthogonality between the two subspaces, a weighted matrix is constructed to constrain the sparse vector to guarantee effective sparse solution. The presented algorithm has low computational complexity due to avoiding the calculation and inversion of array covariance matrix, can effectively suppress spurious peaks, has low sensitivity to the assumed number of signals and possesses better DOA estimation performance.(4) The processing of generating signals, controlling parameters and collecting data on Lyrtech platform has been researched. Based on above processing, W-L1-SRACV and W-L1-SVD algorithms have been simulated on Lyrtech platform. Simulation results are shown to verify the efficiency of the two algorithms.