On recovery of block sparse signals from multiple measurements

We consider the problem of recovering block sparse signals which share the same sparsity pattern given multiple measurements. We consider two different noisy measurement models. In the first model, the sensing matrix remains the same for all the measurements. In the second model, we employ different sensing matrices for different measurements. For both these models, we present greedy algorithms for block sparse signal recovery and theoretically establish the recovery guarantees of the proposed algorithms. Using numerical simulations, we study the performance of the proposed algorithms and some existing algorithms. Our results present insights on how the correlation between block sparse signals plays a role on the recovery performance.