An Improved Newton-Schulz Iterative Algorithm for Massive MIMO Detection.

The exact matrix inversion required by the conventional minimum mean square error (MMSE) detector for massive multiple-input multiple-output (MIMO) brings unaffordable computational burden that hinders efficient implementation. In this paper, a linear iterative detector based on Newton-Schulz algorithm is proposed to avoid the inversion of MMSE. An initialization scheme and a successive-update schedule is proposed to enhance the convergence rate and accuracy of the Newton-Schulz detector. Simulation results are illustrated to show that the proposed detector can approach the bit-error-rate (BER) performance of MMSE with reduced complexity. Moreover, it can outperform state-of-the-art (SOA) linear detection methods under various MIMO scenarios.