A Low-Complexity Precoding Algorithm Based on Improved SOR Method for Massive MIMO Systems

For massive multiple-input multiple-output (MIMO) systems, linear precoding can achieve near-optimal performance due to the asymptotically orthogonal channel property, but it suffers from high computational complexity of matrix inversion. To avoid exact matrix inversion in massive MIMO systems, we propose a low-complexity successive over relaxation (SOR) based precoding algorithm. However, due to the complex calculation of the optimal relaxation parameter, the SOR-based algorithm is not applicable in practical systems. To deal with this problem, we obtain the optimal relaxation parameter based on random matrix theory. Furthermore, to guarantee the performance of practical massive MIMO systems, we propose a linear fitting method to obtain the optimal relaxation parameter. In addition, the convergence rate and computation complexity of different iterative methods are compared, from which we can see that the improved SOR-based method has a faster convergence rate than other methods but with almost the same computational complexity. Finally, numerical simulations further verify the advantages of our proposed method in bit error rate (BER) and user average rate performance.