Learning to Beamform in Heterogeneous Massive MIMO Networks

Finding the optimal beamformers in massive multiple-input multiple-output (MIMO) networks is challenging because of its non-convexity, and conventional optimization based algorithms suffer from high computational costs. Recently, deep learning based methods have been proposed because of their computational efficiency, but they typically can not generalize well when deployed in heterogeneous scenarios where the base stations (BSs) are equipped with different numbers of antennas and have different inter-BS distances. This paper proposes a novel deep learning based beamforming algorithm to address above challenges. Specifically, we consider the weighted sum rate (WSR) maximization problem in multi-input and single-output (MISO) interference channels, and propose a beamforming learning architecture by unfolding a parallel gradient projection algorithm. By leveraging the low-dimensional structures of the optimal beamforming solution, our constructed learning network can be made independent of the numbers of transmit antennas and BSs. Moreover, such a design can be further extended to a cooperative multicell network where users are jointly served by multiple BSs. Numerical results based on both synthetic and ray-tracing channel models show that the proposed neural network can achieve high WSRs with significantly reduced runtime, while exhibiting favorable generalization capability with respect to the antenna number, BS number and the inter-BS distance.