基于两阶段的毫米波大规模MIMO低复杂度混合预编码算法

A Two⁃Stage Based Low Complexity Hybrid Precoding Algorithm for Millimeter⁃Wave Massive MIMO

针对现有基于矩阵分解的混合预编码算法信道容量有损和算法复杂度高的问题,本文提出了一种基于两阶段的低复杂度混合预编码算法.该算法分为获取最优全数字预编码器和求解混合预编码器两部分.首先,本文联合奇异值分解(Singular Value Decomposition,SVD)与注水算法以容量无损的要求设计最优全数字预编码矩阵.其次,为了降低搜索超完备矩阵列的复杂度,提出两阶段混合预编码(Two⁃Stage Hybrid Precoding,TS⁃HP)算法求解混合预编码矩阵.第一阶段,根据天线阵列响应矩阵的相关性获取模拟预编码矩阵备选集;第二阶段,利用贪婪搜索对备选集进行搜索构建混合预编码矩阵.仿真结果表明,所提算法能够有效改善系统性能,降低复杂度.

Aiming at the problems of loss channel capacity and high algorithm complexity of the existing matrix decomposition⁃based hybrid precoding algorithms,a two⁃stage based low complexity hybrid precoding algorithm is pro⁃posed.The algorithm is divided into two parts:obtaining the optimal full⁃digital precoder and the hybrid precoder.Firstly,this paper combines singular value decomposition(SVD)and water filling algorithm to design an optimal full⁃digital pre⁃coder with the requirement of lossless capacity.Then,to reduce the complexity of searching columns of the overcomplete matrix,the two⁃stage hybrid precoding(TS⁃HP)algorithm is proposed to obtain the hybrid precoding matrix.In the first stage,the candidate set of the analog precoding matrix is obtained according to the antenna array response matrix correla⁃tion.In the second stage,greedy search is used to search the candidate set to construct a hybrid precoding matrix.Simula⁃tion results show the proposed algorithm can effectively inprove the system performance and reduce the algorithm com⁃plexity.