Ordered Iterative Methods for Low-Complexity Massive MIMO Detection

In this paper, two ordered iterative detection methods are proposed for better signal detection performance in massive multiple-input multiple-output (MIMO) systems. First of all, in order to reduce error propagation in the traditional iterative detection schemes with sequential order, the ordered iterative detection (OID) algorithm is proposed, which achieves a better detection performance with low complexity. Then, we show that the convergence performance chiefly depends on the residual component during the iterations. Therefore, a dynamic ordering strategy is given for further performance improvement, which leads to the modified ordered iterative detection (MOID) algorithm. After that, we extend the proposed MOID algorithm via deep learning network (DNN), and parameters like relaxation factor are trained to optimal for further performance gain.