A Simplified Message Passing Detection Algorithm for Massive MIMO Systems

Abstract

The original message passing detection (MPD) algorithm based on channel hardening theory has shown excellent performance in large-scale multiple-input multiple-output (MIMO) systems. However, with the increase of the numbers of users and modulation order, the high computation complexity of the original MPD algorithm is very unfriendly to hardware implementation. Therefore, this paper proposes a simplified MPD algorithm based on probability approximation, which is called MS-PA-MPD for short. This algorithm simplifies the calculation of log-likelihood ratio in original MPD algorithm, and eliminates many exponential and division operations. It also sorts and selects the symbol probabilities in each iteration. Only the constellation points with high symbol probabilities can be selected to update the probabilities, which greatly reduces the computation complexity. Moreover, an improved algorithm HMS-PA-MPD is proposed to solve the problem of MS-PA-MPD's performance degradation under higher-order modulation. Simulation results show that, for 16QAM, the performance loss of MS-PA-MPD can be almost ignored compared with the original MPD. Besides, the HMS-PA-MPD also shows good detection performance under higher-order modulation. Both of them greatly reduce the computation complexity and hardware overhead, and are especially suitable for large-scale MIMO systems.