Sparse Representation-Based DOA Estimation with Concentration Ratio Criteria

A sparse representation-based direction-of-arrival (DOA) estimation method is proposed which defines a concentration ratio (CR) criterion for selecting the regularization parameter, shorten as the SRCR method. The proposed SRCR method performs regardless of the statistics of noise and thus it is applicable in the case of noise with unknown statistics. In particular, the SRCR method defines the CR of the recovered sparse vector as a criterion for selecting the regularization parameter. In addition, it optimizes the regularization parameter to ensure the CR is near to 1. By this way, the optimized regularization parameter recovers the sparsest signal vector, which results in correct DOA estimation. Simulation results demonstrate that the SRCR method is independent of the statistics of noise, and it performs significantly better than the SR-based DOA estimation method with the discrepancy principle (DP) for the regularization parameter selection.