Joint Power Allocation and Hybrid Beamforming of mmWave NOMA Systems Using DDPG

Joint Hybrid Beamforming (HBF) and Power Allocation (PA) in mmWave nonorthogonal multiple access (NOMA) systems are investigated in this letter. We consider downlink mmWave NOMA systems, in which HBF is done at the Base Station (BS), and all users have only one antenna. At the BS, user grouping uses the K-means-based algorithm according to the user’s channel correlation. Subsequently, inspired by the benefits of Deep Reinforcement Learning (DRL), a novel algorithm based on DRL is proposed to output the HBF matrix and PA of all the users. Simulations represent the superiority of the proposed algorithm in comparison with the state-of-the-art schemes in both perfect Channel State Information (CSI) and imperfect CSI.