A fast multiple orthogonal matching pursuit algorithm for jointly sparse recovery

To recover the jointly sparse signal efficiently, a fast multiple orthogonal matching pursuit algorithm (FMOMP) is proposed in the paper. By choosing multiple indices per iteration, the FMOMP converges much faster and improves the computational efficiency over the existing OMPMMV algorithm. We also prove that FMOMP performs the exact recovery of any K row jointly sparse signal from the aspect of sensing matrix's restricted isometry property (RIP). Empirical experiments show that FMOMP is very efficient in recovering jointly sparse signal compared to the state of the art recovery algorithms.