大规模MIMO系统低复杂度混合迭代信号检测

摘要

在大规模MIMO系统上行链路信号检测算法中,最小均方误差(MMSE)算法能获得接近最优的线性检测性能.但是,传统的MMSE检测算法涉及高维矩阵求逆运算,由于复杂度过高而使其在实际应用中难以快速有效地实现.基于最速下降(steepest descent,SD)算法和高斯一赛德尔(Gauss-Seidel,GS)迭代的方法提出了一种低复杂度的混合迭代算法,利用SD算法为复杂度相对较低的GS迭代算法提供有效的搜索方向,以加快算法收敛的速度.同时,给出了一种用于信道译码的比特似然比(LLR)近似计算方法.仿真结果表明,通过几次迭代,给出的算法能够快速收敛并接近MMSE检测性能,并将算法复杂度降低一个数量级,保持在O(K2).%Among the uplink signal detection algorithms for massive MIMO systems,the minimum mean square error (MMSE) algorithm can achieve the near-optimal linear detection performance.However,conventional MMSE usually involves high complexity due to the required matrix inversion of large-size matrix,which makes it hard to implement in realistic applications.Based on joint steepest descent (SD) algorithm and Gauss-Seidel iteration,a low complexity hybrid iterative detection algorithm was proposed.The SD algorithm was employed to obtain an efficient searching direction for the following Gauss-Seidel to speed up convergence.Meanwhile,an approximated method was also proposed to compute the bit log-likelihood ratio (LLR) for soft channel decoding.Simulation results verify that the proposed algorithm can converge rapidly and achieve its performance quite dose to that of the MMSE algorithm with only a small number of iterations.Meanwhile,the complexity is reduced by an order of magnitude,which is kept consistently of O(K2).