Concatenate and Boost for Multiple Measurement Vector Problems (0906.2609v2)

Abstract: Multiple measurement vector (MMV) problem addresses the recovery of a set of sparse signal vectors that share common non-zero support, and has emerged an important topics in compressed sensing. Even though the fundamental performance limit of recoverable sparsity level has been formally derived, conventional algorithms still exhibit significant performance gaps from the theoretical bound. The main contribution of this paper is a novel concatenate MMV and boost (CoMBo) algorithm that achieves the theoretical bound. More specifically, the algorithm concatenates MMV to a larger dimensional SMV problem and boosts it by multiplying random orthonormal matrices. Extensive simulation results demonstrate that CoMBo outperforms all existing methods and achieves the theoretical bound as the number of measurement vector increases.