Joint Precoding Using Successive Over-relaxation Matrix Inversion and Newton Iteration for Massive MIMO Systems

In massive multiple-input multiple-output (MIMO) systems, Newton iteration can effectively reduce the computation complexity of regularized zero forcing (RZF) precoding algorithm in which complicated matrix inversion is required. However, the Newton iteration has slow convergence speed and it needs complicated calculation to search initial values. In this paper, by using successive over-relaxation (SOR) iteration, a matrix inversion method, namely successive over-relaxation matrix inverse (SORMI) iteration, is derived to obtain the initial values for Newton iteration. Correspondingly, the SORMI-Newton joint precoding algorithm is presented. Moreover, convergence and complexity are analyzed for the investigated joint precoding algorithm. Simulation results show that compared with the conventional Newton, the joint precoding algorithm possesses have faster convergence speed because of its appropriate initial values, at the same time, it can achieve a bit error rate (BER) performance comparable to that of the RZF.