Matrix Inversion-Less Signal Detection Using SOR Method for Uplink Large-Scale MIMO Systems

摘要

For uplink large-scale MIMO systems, linear minimum mean square error (MMSE)signal detection algorithm is near-optimal but involves matrix inversion withhigh complexity. In this paper, we propose a low-complexity signal detectionalgorithm based on the successive overrelaxation (SOR) method to avoid thecomplicated matrix inversion. We first prove a special property that the MMSEfiltering matrix is symmetric positive definite for uplink large-scale MIMOsystems, which is the premise for the SOR method. Then a low-complexityiterative signal detection algorithm based on the SOR method as well as theconvergence proof is proposed. The analysis shows that the proposed scheme canreduce the computational complexity from O(K3) to O(K2), where K is the numberof users. Finally, we verify through simulation results that the proposedalgorithm outperforms the recently proposed Neumann series approximationalgorithm, and achieves the near-optimal performance of the classical MMSEalgorithm with a small number of iterations.