摘要: |
稀疏码分多址接入(Sparse Code Multiple Access,SCMA)作为有应用前景的空口技术,在高吞吐量以及大规模连接中极具优势。针对SCMA通信系统中接收端消息传递算法(Message Passing Algorithm,MPA)计算复杂度较高的问题,提出了基于模型驱动辅助MPA法(Model driven Assisted MPA,MD MPA)的SCMA多用户检测算法。MD MPA在MPA算法迭代过程中节点更新后的信息矩阵和输出的概率矩阵之后添加权重参数,并通过神经网络训练更新参数。经训练所得权重参数可加快算法收敛速度,降低误码率,获得更佳的译码效果。仿真结果表明,MD MPA相较于MPA算法,误码率降低了20%,复杂度降低了33%。 |
关键词: 稀疏码分多址接入(SCMA) 多用户检测 消息传播算法(MPA) 模型驱动 |
DOI: |
|
基金项目: |
|
SCMA multi-user detection algorithm based on model-driven assisted MPA |
SHAO Kai,GUO Hongyao |
(School of Communication and Information Engineering,Chongqing University of Posts and
Telecommunications) |
Abstract: |
Sparse code division multiple access(SCMA),as a promising air interface technology,has great advantages in high throughput and large scale connections.For the high bit error rate(BER) and computational complexity of the message passing algorithm(MPA) at the receiving end in the SCMA communication system,a SCMA multi user detection algorithm based on model driven assisted MPA(MD MPA) is proposed.MD MPA adds weight parameters after the updated information matrix and output probability matrix of the node during the iteration of the MPA algorithm,and updates the parameters through neural network training.The weight parameters obtained through training can speed up the algorithm convergence speed,reduce the BER,and obtain better decoding results.The simulation results show that compared with the MPA algorithm,MD MPA has a 20% reduction in BER and a 33% reduction in complexity. |
Key words: sparse code multiple access(SCMA) multi user detection message passing algorithm(MPA) model driven |