Two Simplified Multiuser Detection Algorithms for Uplink SCMA Systems via Generalized Approximate Message Passing

Abstract

Sparse code multiple access (SCMA) is a competitive non-orthogonal multiple technique for 5G, which can accommodate more communication users by exploiting the multi-dimensional codewords. However, it is inflexible to meet the various demands of detection performance, or the various demands of computational complexity in real scenario. In this paper, we propose two simplified algorithms to deal with the problem, named GAMP-MPA and SIC-GAMP-MPA. These two algorithms are developed from the mainstream SCMA detector algorithm, message passing algorithm (MPA), and inspired by generalized approximate message passing (GAMP) as well as serial interference cancellation (SIC). The modified MPA benefits from the GAMP method since GAMP converts vector problems to scalar problems which provides higher computational efficiency and easier hardware implementation. Simulation results show that the GAMP- MPA achieves better performance than MPA with an acceptable extra complexity cost, and SIC-GAMP-MPA has significantly decreased complexity compared to MPA with similar performance. Particularly, the advantages of both GAMP-MPA and SIC-GAMP-MPA are more apparent in low signal-to-noise ratio (SNR) condition.