Abstract

Minimum mean square error (MMSE) linear detector is able to achieve the near-optimal bit error rate (BER) performance for uplink multi-user massive Multiple-Input Multiple-Output (MIMO) systems due to the increase in the number of base station (BS) antennas. However, this detector involves matrix inversion with high complexity, especially when the number of users is large. In order to reduce the complexity, in this paper, a low-complexity MMSE detector algorithm based on the Damped Jacobi (DJ) method is proposed. Further, a simple approach to determine the optimum and quasi-optimum damped parameter by exploiting the massive MIMO channel property of asymptotic orthogonality is developed. The analysis shows that the proposed algorithm can reduce the complexity of the classical MMSE detector by about one order of magnitude without performance loss. Finally, it is verified through simulation results that the proposed algorithm outperforms the recently proposed linear detector based on the conventional Jacobi method, and achieves the near-optimal performance of the classical MMSE detector with a small number of iterations in Rayleigh fading channels.