Efficient LLR Calculation for Uplink Coded Massive MIMO Systems

Abstract

In the uplink massive MIMO systems, linear minimum mean square error (MMSE) algorithm can achieve near-optimal performance in combination with a soft iterative decoder, but suffers from high computational complexity due to the complicated matrix inversion. To approximate the performance of the classical MMSE detection algorithm, a number of iterative methods were proposed with reduced complexity by eliminating the matrix inversion. However, in order to apply these methods to coded systems with soft iterative decoders the post-equalization signal-to-interference-plus-noise ratio (PE-SINR) should be calculated in each layer to produce soft output values. In this paper, we propose to approximate the PE-SINR in each layer with a universal value calculated at the base station (BS), and apply symbol mapping techniques to the estimation of soft output in each layer to further reduce the computational complexity. The simulation results demonstrate that the detection algorithm with the proposed PE-SINR calculation approach achieves approximating performance to the conventional methods.