基于EXIT图的SCMA通信系统多用户检测算法收敛性分析

Convergence Property Analysis for Multiuser Detection Schemes in SCMA Systems Based on EXIT Chart

消息传递算法(message passing algorithm,MPA)是基于稀疏码多址接入(sparse code multiple access,SCMA)技术中码本的稀疏性,提出的一类接近联合最优的多用户检测算法。针对现有的基于并行策略和串行策略的MPA算法,存在算法收敛性论述缺乏理论分析的问题,该文通过外信息转移图(extrinsic information transfer,EXIT)技术分析MPA算法的收敛性。首先将SCMA接收机检测器分为函数节点译码器和变量节点译码器,然后给出译码器的EXIT图的理论分析,最后通过比较两类算法的EXIT曲线估算出达到收敛状态所需的迭代次数。仿真结果验证了EXIT图分析的正确性。

Thanks to the sparse structure of sparse code multiple access （ SCMA）, a message passing algorithm （MPA） was proposed to efficiently approximate the optimum maximum a posteriori scheme. Based on a lack of theoretical analysis on the convergence property of parallel and serial strategies, in this paper, extrinsic information transfer （EXIT） is proposed to analyze the convergence property. First, the SCMA detector is divided into two parts： one function node decoder and one variable node decoder, and then, analyzing the decoders＇ EXIT convergence property. Finally, by comparing the EXIT curve of the two kinds of algorithms, the number of iterations required to achieve convergence state is estimated. Simulation results validate the EXIT chart analysis.