Wide-band DOA Estimation Method Based on Fast Sparse Bayesian Learning

Bayesian methods are promising for sparse representation based problems, which could appropriately avoid the parameter tuning procedure and desirably provide statistical information. However, the sparse Bayesian methods often suffer from high computational complexity, where practical applications to wide-hand DOA estimation problem is largely constrained. In this paper, a fast sparse Bayesian algorithm is developed for wide-hand DOA estimation problem, where the proposed method can be applied to estimate the DOA with substantially reduced computational costs. In the derived methods, the algorithm operates in a constructive manner, which can only update one basis in each iteration. On the one hand, the proposed algorithm could avoid parameter-tuning procedure, compared with the convex optimization based methods. On the other hand, the algorithm can efficiently obtain estimation of the sources compared to the conventional variational Bayesian implemented sparse Bayesian approaches. Results from numerical experiments have demonstrated that the proposed algorithm can achieve desirable performance with substantially reduced computational complexities.