DOA Estimation Based on Sparse Reconstruction via Acoustic Vector Sensor Array under Non-uniform Noise

This paper aims to solve the reduction in direction of arrival estimated accuracy for the acoustic vector sensor array in the existence of non-uniform noise, and proposes an iterative sparse covariance matrix reconstruction (ISCMR) method. We first define a virtual manifold matrix and establish the cost function based on the covariance matrix fitting criterion. Then, using the properties of Frobenius norm to derive the analytical expression of the cost function. Furthermore, the sparse covariance matrix is reconstructed according to the estimated power of sparse signal and noise until the iteration is terminated. Finally, a more exactly DOA estimation is completed by searching spectral peaks on the sparse signal power. The superior performance of the proposed method, comparison with the existing methods, is proved by simulation results in the existence of non-uniform noise.