Selection of regularization parameter in sparse inverse problems for DOA estimation

This article deals with the selection of the regularization parameter of a sparse L0 regularized criterion for DOA (Direction-of-Arrival) estimation. This parameter is generally empirically obtained. In this paper, we establish a theoretical range for this parameter allowing a correct estimation for a single observation. Considering that an observation is the eigenvector associated with the maximum eigenvalue of the covariance matrix, we extend the analysis to multiple observations. Numerical results support our theoretical investigations.