On Exploring Sparsity in Widely Separated MIMO Radar

The scenario of widely separated multi-input multi-output (MIMO) radar is considered. For a small number of targets, the target returns are sparse in the target space. First, a decoupled Lasso approach is proposed, which by exploiting the structure of the basis matrix decomposes the large size problem into a number of smaller size problems, thus reducing computational complexity. Second, it is shown that by reordering the columns of the basis matrix, group sparsity can be introduced to the returns. This structure can be exploited by a group Lasso approach to achieve significant performance gains.