Low-Complexity Data Detection Based on Combined SD and SOR Algorithms for Uplink Massive MIMO Systems

Abstract The zero forcing (ZF) and the and minimum mean square error (MMSE) conventional algorithms has been widely employed as near-optimal detectors for large-scale multiple-input multiple-output (LS-MIMO) systems, whereas their complexity is too high. In order to reduce the complexity of the MMSE algorithm while maintaining its performance in term of convergence speed, we develop in this paper a new fast detection scheme based on the joint of two low complexity iterative algorithms. The devised iterative scheme starts with an initial solution based on the inverse of diagonal matrix to calculate just one steepest decent (SD) iter-ation. So, this first solution is to be offered to the successive over-relaxation (SOR) algorithm which can use it to converge quickly with lower number of iterations. A better tradeoff between com-plexity and convergence speed compared to conventional methods is shown from theoretical investigation and simulation examples of our proposed algorithm.