Fast sparse channel estimation for many access MIMO system: On the benefit of block sparsity

We consider efficient methods for active users detection in Multiple-Input Multiple-Output (MIMO) system. To support massive connectivities, all active users are allowed to transmit information simultaneously. We use greedy algorithm to get the set of active users and their channel state information(CSI) in limited time. The Block Orthogonal Matching Pursuit (BOMP) algorithm needs many iterations and the knowledge of sparsity level. To make explicit use of block sparsity, adaptive Compressive Sampling Matching Pursuit(ACoSaMP) algorithm is proposed to decrease the iterations we need. And new stop conditions has been found to get accurate sparsity level which is the number of active users. Combining those two methods, we can have better performance and lower complexity than BOMP at the same time.