Group Bayesian Sparse Channel Estimation for Massive MIMO Systems

In massive MIMO systems, different wireless channels starting from the same user antenna to different base station (BS) antennas usually share a common channel support. In this paper, Bayesian channel estimation is studied to jointly recover uplink channels sharing common support in massive MIMO systems. Since wireless channel is usually modeled as a mixed Bernoulli-Gaussian random process, it is employed as a prior knowledge for Bayesian estimation. A group Bayesian sparse channel estimation algorithm is proposed by enhancing the posterior probability of common support after jointly processing the received signal at BS. To reduce the computational complexity of Bayesian estimation, the genuine support set of uplink channels is approximated by a main support set. Moreover, the estimation of channel sparsity is also derived. Simulation results show that the proposed group Bayesian algorithm performs much better than the existing sparse recovery algorithms.