A Correlation-Aware Sparse Bayesian Perspective for DOA Estimation With Off-Grid Sources

A sparse Bayesian perspective for off-grid direction-of-arrival (DOA) estimation when the correlation information about the unknown coherent sources is taken into account is formulated in this article. A key contribution is to establish a suitable statistical model which inherently exploits the nature of correlation between the nonzero components of the sparse signal. Such a model guarantees to promote the reconstruction accuracy and robustness. Using our model, we derive the adaptive updating formulas for latent variables via the variational Bayesian inference and the resulting algorithm is fully automated, i.e., no user-intervention is needed. In addition, we introduce a refined 1-D searching procedure which utilizes elucidated properties of the marginal likelihood function to refine the DOA estimates one by one based on the reconstruction result. Our experimental results on synthetic data sets show that, compared with state-of-the-art algorithms, the proposed algorithm yields superior estimation performance at the expense of a higher computational complexity.