Efficient EP Detectors Based on Channel Sparsification for Massive MIMO Systems

Multiuser detection based on the expectation propagation (EP) algorithm for the massive multiple-input multiple-output (M-MIMO) system has received considerable attention in recent years due to its good performance-complexity tradeoff. However, its performance degrades when the system is highly overloaded. Furthermore, the computational complexity increases fast as the number of receive antennas increases. In this work, we propose an improved EP detector via channel sparsification based approach, which is shortened as S-EP. The key idea is to sparsify the channel first, such that the average degree of the variable node (VN) and function node (FN) in the associated factor graph (FG) is significantly decreased, thus rendering a reduced effective interference. Therefore, the inference accuracy can be enhanced. The associated computational complexity of the message computation can also be eliminated. Analytical and simulation results illustrate that the proposed S-EP based detection scheme achieves noticeable performance gains compared with the conventional EP based one while requiring even a lower computational complexity.