Joint channel estimation algorithm based on structured compressed sensing for FDD multi-user massive MIMO

Accurate channel state information (CSI) at transmitter is of importance to sufficiently exploit the merits of massive multiple input multiple output (MIMO). Because of the large amount of antennas at base station (BS), the pilot overhead becomes unaffordable, especially in frequency-division duplexing (FDD) massive MIMO systems. To alleviate the overwhelming pilot overhead, a novel channel estimation algorithm for multi-user massive MIMO system employing structured compressed sensing (CS) theory is proposed. Firstly, the angular domain channel representation of massive MIMO is analyzed. Then, due to the practical scattering environment, the common sparsity and private sparsity structure of channel matrix exist in multi-user massive MIMO system. Finally, basing on the statistical information of multi-user channel matrix, a structured joint subspace matching pursuit (SJSMP) algorithm is proposed, which is to estimate channels with limited pilot jointly at the BS. Particularly, the common support and private support of multi-user channel matrix are separately estimated to reduce the pilot overhead with improved CSI estimation quality in terms of MSE.