摘要
近些年,基于期望传播(expectation propagation)的EP检测算法在高阶调制的大规模MIMO(multiple input multiple output)场景下展现出了相比于传统检测算法的巨大优势,但是由于其每次迭代都需要进行矩阵求逆运算,使得该算法的计算复杂度在实际应用中是难以接受的.本文提出了一种适用于大规模MIMO系统下的逐次更新EP-SU (expectation propagation successive updating)检测算法,一方面提出了逐次更新的信息节点更新方式,加快了信息更新效率的同时规避了原有EP中的矩阵求逆运算;另一方面优化了单个节点更新算法,提升了信息更新的有效性和准确性.仿真结果表明:本文提出的EP-SU检测算法在多种大规模MIMO场景下,都具有相比于原有EP检测器更加优秀的检测性能、更快的收敛速度、更低的误码平层和更低的计算复杂度,最后这些优势都会随着天线规模的增加而更加显著.
In recent years, the detection algorithm based on expectation propagation in high-dimension highorder modulation multiple-input-multiple-output(MIMO) systems shows a huge advantage over the traditional detection algorithm;however, this MIMO-based detection algorithm suffers from the prohibited complexity caused by matrix inversion in each iteration. In this paper, an EP-SU detection algorithm for large-scale MIMO systems is proposed. On the one hand, a successive updating method of information nodes is proposed, which speeds up the efficiency of information updating and avoids the matrix inversion of original EP;on the other hand, the single node update algorithm is optimized, which improves the efficiency and accuracy of information update.Simulation results show that the proposed EP-SU detection algorithm in various large-scale MIMO scenarios exhibits a more outstanding detection performance, faster convergence rate, lower error flat layer, and lower computational complexity compared with traditional EP detection;most importantly, these advantages will be more significant at higher antenna scales.
作者
羊贵武
姚国强
胡剑浩
Guiwu YANG;Guoqiang YAO;Jianhao HU(National Key Laboratory of Communication, University of Electronic Science and Technology of China, Chengdu611731, China)
出处
《中国科学：信息科学》
CSCD
北大核心
2019年第7期853-867,共15页
Scientia Sinica(Informationis)
基金
国家自然科学基金(批准号:61571083,61501084)资助项目
关键词
大规模MIMO
期望传播
逐次更新
低复杂度
快收敛
large-scale MIMO
expectation propagation
successive updating
low complexity
fast convergence
