Blind Multiuser Detection in Massive MIMO Channels with Clustered Sparsity

In this paper, we propose a blind multiuser detection algorithm for massive MIMO systems equipped with a uniform planar antenna array. We model the channel in the angular domain as a sparse matrix with the non-zero coefficients forming clusters. The channel sparsity is described by a binary Markov random field (MRF). Then, we develop a turbo message passing framework by incorporating the MRF model into the bilinear generalized approximate message passing (BiGAMP) algorithm. We show that the proposed scheme considerably outperforms the schemes without exploiting the clustered sparsity of the channel.