基于SAMP的低复杂度大规模MIMO信号检测算法

Low-complexity signal detection based on SAMP method for massive MIMO

在大规模MIMO系统上行链路中,最小均方误差(MMSE)算法能达到近似最优的线性信号检测性能。但是,MMSE算法引入了高计算复杂度的矩阵求逆运算。文中提出了一种基于简化近似信息传递(Simplified Approximate Message Passing,SAMP)的低复杂度迭代阈值算法以替代矩阵求逆,并通过重构准确的有效噪声方差及设置合适的初始值,进一步提高了检测性能。仿真结果表明,在设定的有效噪声方差和合适的初始值条件下,仅通过少数几次迭代,SAMP算法就能够以较低的计算复杂度快速接近MMSE算法的检测性能。

Minimum mean square error （MMSE） linear detection algorithm can achieve nearly optimal performance in large-scale MIMO system uplink. However, the MMSE detection algorithm involves a complicated matrix inversion. This paper proposes a low-complexity iterative thresholding algorithm, referred to as simplified approximate message passing （SAMP） method to substitute the matrix inversion. In addition, the exact effective noise variance and the initial solution are presented to further improve the detection performance. Simulation results show that the SAMP algorithm can quickly approach the detection performance of MMSE algorithm with low computational complexity by only a few iterations under the conditions of the effective noise variance and the suitable initial value.