Abstract: For the uplinkchain of massive multiple-input multiple output (MIMO) system, conventionalminimum mean square error (MMSE) linear detection algorithm achieves nearlyoptimal performance when the number of antennas at the base station (BS) ismuch larger than the number of single-antenna users. However, the MMSEdetection algorithm involves complicated matrix inversion, thus making itdifficult to be implemented rapidly and effectively. In order to make atradeoff between detection accuracy and computational complexity, this paper comparesand analyzes polynomial expansion based matrix inversion approximation method andlinear equation iterative solving based equivalent approach of matrix inversion.Finally, we further propose to apply them in soft decision to make full use ofthe soft information of channel coding and decoding. In this way, it is feasibleto reduce complexity while achieving near optimal detection performance.

Key words: massive multiple-input multiple-output, matrix inversion, low complexity, soft decision