Robust Sparse Bayesian Learning for off-Grid DOA Estimation With Non-Uniform Noise

The performance of traditional sparse representation-based direction-of-arrival (DOA) estimation algorithm is substantially degraded in the presence of non-uniform noise and off-grid gap caused by the discretization processes. In this paper, a robust sparse Bayesian learning method is proposed for off-grid DOA estimation with non-uniform noise. In the proposed method, the covariance matrix of non-uniform noise is reconstructed by a modified inverse iteration method. Then, the discrete sampling grid points in the spatial domain are treated as dynamic parameters, and the expectation-maximization algorithm is used to iteratively refine the position of the discretization grid points. This refinement procedure is implemented by solving a polynomial. The simulation results indicate that the proposed method can maintain excellent DOA estimation performance with uniform or non-uniform noise. Furthermore, it can also achieve satisfactory performance under a coarse grid condition.