Block compressive channel estimation and feedback for FDD massive MIMO

Channel state information (CSI) at the base station (BS) is required to fully exploit the advantages of massive MIMO. However, due to massive BS antennas, the estimation and feedback of downlink CSI are very challenging in frequency division duplexing (FDD) massive MIMO. This paper proposes a block compressive channel estimation and feedback scheme for FDD massive MIMO, which can reduce the overhead for CSI acquisition substantially. Specifically, we first propose the non-orthogonal pilots, which is essentially different from conventional orthogonal pilots. Then, a block orthogonal matching pursuit (BOMP) algorithm is proposed to estimate CSI according to the feedback signal from users, where the analog channel feedback is adopted, and the spatial common sparsity of time-domain massive MIMO channels is exploited to reduce the overhead for CSI acquisition. Moreover, we exploit the temporal common sparsity of channels to estimate channels with reduced complexity. Simulation results demonstrate that the proposed scheme with significantly reduced CSI acquisition overhead can approach the performance bound.