大规模MIMO检测的理论性能分析和算法设计

项目名称： 大规模MIMO检测的理论性能分析和算法设计

项目编号： No.61501043

项目类型： 青年科学基金项目

立项/批准年度： 2016

项目学科： 无线电电子学、电信技术

项目作者： 钱荣荣

作者单位： 北京邮电大学

项目金额： 19万元

中文摘要： 大规模天线（Large MIMO）是5G系统等未来移动通信系统实现超高速率通信的关键技术之一。面向大规模天线系统的MIMO检测（简称大规模MIMO检测）需要综合考虑算法复杂度、错误概率、复用增益和分集增益等多方面因素，因此建立一个适合的理论性能分析体系对大规模MIMO检测算法设计与实现有重要的理论和现实意义。当前MIMO检测性能分析体系尚不能综合化地考虑多方面性能因素，尤其是引入大规模天线后，很难有效区分多种大规模MIMO检测算法的性能优劣，影响了大规模天线在5G系统中的实现。本课题拟利用随机矩阵理论、自由概率理论和凸体几何理论开展渐进分析，试图从理论层面建立适合于大规模MIMO检测的性能分析体系，提出具有良好综合度和区分度的性能分析方法，以综合分析并有效区分不同大规模MIMO检测算法的性能优劣，并在此基础上设计综合性能最优/次优的大规模MIMO检测算法并研究算法在5G系统中的适用性。

中文关键词： 大规模天线；大规模MIMO检测；性能分析；算法设计；渐近分析

英文摘要： Large MIMO is one of the key technologies which enable future mobile communication systems (e.g., 5G) to provide super high data rate services. MIMO detection for Large-MIMO systems (Large-MIMO detection for short) shall take a number of performance factors (including algorithmic complexity, error probability, multiplexing gain and diversity gain) into consideration. Therefore, a proper framework of performance analysis is indispensable to design and implement Large-MIMO detection. However, the current performance analysis framework for MIMO detection cannot comprehensively consider many factors. Especially, for Large-MIMO systems it is still hard to discriminate the performance of different Large-MIMO detection algorithms. This makes Large MIMO difficult to be realized in 5G systems. Our project will perform asymptotic analysis via making use of the theories of random matrix, free probability and convex geometry so as to establish a proper performance analysis framework for Large-MIMO detection, overcome the defects of the current framework, and obtain a method of performance analysis with highly comprehensive and discriminable capability for Large-MIMO detection. Besides, our project aims to obtain the optimal/near-optimal algorithm of Large-MIMO detection and investigate the applicability of the algorithm to 5G systems.

英文关键词： Large MIMO;Large-MIMO Detection;Performance Analysis;Algorithm Design;Asymptotic Analysis