DOA Estimation of Coherent Signals Based on Improved SVD Algorithm

Abstract

In the light of that the general super-resolution subspace algorithms are invalid to coherent signals and the resolution of singular value decomposition (SVD) algorithms is reduced in the presence of low-SNR, an improved SVD algorithm to direction-of-arrival (DOA) estimation is proposed. Firstly, a new matrix is constructed from the maximum eigenvector of the signal covariance matrix according to certain rules. Secondly, the reconstruction matrix is corrected by using the idea of matrix decomposition to enhance the decorrelation ability. Finally, ESPRIT method is utilized to DOA estimation. The simulation results show that the proposed algorithm has high estimation success probability, small estimation bias, and low estimation standard error for DOA estimation of the coherent signals in low-SNR condition.