WMMSE-based multiuser MIMO beamforming: A practice-oriented design and LTE system performance evaluation

Abstract

This paper deals with the multiuser MIMO downlink beamforming scenario. In this scenario, a popular approach for designing the beamformer is the weighted minimum mean square error (WMMSE) approach for sum rate maximization. This work considers the WMMSE method from an implementation perspective, where we take into account practical factors - such as linear receivers (rather than successive interference cancellation receivers, which are implicitly assumed in some existing works but are more complex to implement) and imperfect channel knowledge at the transmitter - for deployment in systems such as long term evolution (LTE). A modified WMMSE algorithm that incorporates such factors is proposed. We evaluate the performance of the proposed WMMSE algorithm using a time division duplexing (TDD) LTE simulation platform, and show that our practice-oriented design can improve the system throughput by 3 ~ 30% in comparison with the conventional WMMSE algorithm.