New approach for massive MIMO detection using sparse error recovery

Jun Won Choi, Byonghyo Shim

Abstract: In this paper, we introduce a new symbol detection technique for large-scale multi-input multi-output (MIMO) systems. Based on the observation that detection errors produced by conventional linear detectors tend to be sparse in practical communication regime, we employ compressed sensing techniques to correct the symbol errors from the output of the linear detectors. The proposed symbol detector, referred to as post detection sparse error recovery (PDSR) technique is derived in two steps (1) sparse transform: transforming the original non-sparse system into a sparse error system and (2) sparse error recovery: applying the sparse signal recovery algorithm to estimate the error vector at the output of the transformed system. We show from the asymptotic mean square error (MSE) analysis that the proposed post detection technique based on compressed sensing can bring remarkable performance gains over the conventional detectors. The intensive simulations performed over large-scale MIMO systems also confirm the superiority of the PDSR algorithm.