ABSTRACTDistributed compressed sensing exploits the correlation among multiple signals to reduce the number of measurements required for recovery. In this paper, we propose a recovery algorithm for a type of joint sparsity model, where all signals share a common sparse component and each individual signal contains a sparse innovation component. Our approach iteratively removes the information of each component from the measurements and then performs sparse recovery. We provide analytical analysis to verify the advantage of the proposed algorithm over separate recovery, which is also confirmed by simulation results. Copyright © 2012 John Wiley & Sons, Ltd.