基于动态因子图缩减的SCMA低复杂度检测算法

针对现有稀疏码分多址接入(sparse code multiple access, SCMA)上行链路系统的接收端采用基于遍历的消息传递算法(message passing algorithm, MPA)进行多用户检测，存在复杂度过高的问题，通过在原始MPA算法基础上提出一种基于动态因子图缩减的消息传递算法(dynamic factor graph reduction based MPA, DFGR-MPA)。该算法根据码字概率的收敛情况对变量节点进行分类，并将码字概率未达到目标收敛的变量节点以及与其相关的功能节点重构成新的因子图用于后续迭代。DFGR-MPA算法不仅能够减少参与消息迭代的节点个数还能降低消息迭代的平均迭代次数，从而降低迭代检测的复杂度。仿真结果表明，该算法在误比特率(bit error ratio, BER)性能与检测复杂度之间可以达到较好地平衡。

A low complexity SCMA detector based on dynamic factor graph reduction

Sparse code multiple access (SCMA), a competitive air-interface technology with the message passing algorithm (MPA) employed at the receiver, has a broad application prospect. However, the computational complexity of MPA is still an extreme challenge for SCMA detection. A low complexity multiuser detection algorithm, dynamic factor graph reduction based MPA (DFGR-MPA), is proposed for the SCMA uplink system in this paper. The algorithm classifies the variable nodes according to the convergence of the codeword probability, and reconstructs a new factor graph with the variable nodes whose codeword probability has not reached the target convergence and the related function nodes for subsequent iteration. The DFGR-MPA algorithm can reduce the detection complexity by reducing the number of nodes involved in message iterations and the average number of iterations. Numerical results demonstrate that the proposed scheme can achieve an affordable trade-off between bit error ratio (BER) performance and detection complexity for SCMA system.