Deep Learning-Based Beam Management and Interference Coordination in Dense mmWave Networks

Abstract

Due to severe signal pathloss in millimeter wave (mmWave) band, beamforming enabled directional transmission is critical to overcome the attenuation challenge in future mmWave communication systems. Furthermore, in order to improve signal coverage of mmWave networks, network densification has to be used at the same time. However, the concurrent use of directional transmission and network densification will make the radio resource management (RRM) of dense mmWave network dramatically more complicated than that of microwave network. In order to maximize the sum-rate of the entire network, tedious and complex RRM algorithms are usually needed to obtain good results, which require high complexity of computation. To address this challenge, we proposed a deep learning-based beam management and interference coordination (BM-IC) method in dense mmWave network, through which the conventional complex BM-IC algorithm is transformed into a deep neural network (DNN)-based approximation. Because DNN only requires a series of simple calculations (e.g., some additions and multiplications), the complexity of computation is greatly reduced. Simulation results show that the proposed deep learning-based BM-IC approach can obtain comparable sum-rate to conventional BM-IC algorithm, but with much less computation time. Thus, deep learning could be a powerful tool to mitigate the complexity of RRM problems in dense mmWave networks.