Combined BP-MF algorithm for cluster sparse non-stationary channel of massive MIMO

This paper proposes a high precision iterative channel estimation algorithm for the uplink of massive multiple input multiple output (MIMO) system. The transmitted signals from a mobile single antenna, are observed as a multi-pattern sparse structure channel over antenna array on account of multi-path. As a result, the whole discrete channel tap vectors received by the array can be clustered into a number of different sets by a non-stationary prior – Dirichlet process (DP) – under the Bayesian framework. Mean Field (MF) – one of the message passing rules – has been used popularly in channel estimation problem, but still suffers from low accuracy. In order to leverage the high accuracy of Belief Propagation (BP), another message passing rule, a combined message passing channel estimation algorithm are developed by combining BP and MF rule based on factor graph. Simulation results show that the proposed algorithm results in a significant performance improvement compared to the state-of-the-art channel estimation algorithms with a similar complexity.