Model-Driven Distributed WMMSE for Downlink Massive MIMO Systems

Distributed signal processing based on the decentralized architecture plays an important role in the next generation of wireless communications. In this paper, we propose a model-driven distributed WMMSE algorithm (MDD-WMMSE) for downlink massive MIMO systems. The proposed MDD-WMMSE is model-driven, which is designed based on a distributed version of the traditional WMMSE algorithm. Specifically, each distributed unit (DU) in it has a local private network and calculates in parallel, while the related distributed training is achieved by simply exchanging the local information. Simulation results show that the proposed MDD-WMMSE achieves competitive performance compared to the centralized WMMSE algorithm but with much lower complexity.