Resource Management Based on Reinforcement Learning for D2D Communication in Cellular Networks

Recently, the integration of Device-to-Device (D2D) communication to cellular networks became a vitality task with the growth of mobile devices, as well as requirements of enhanced network performance in terms of spectral efficiency, energy efficiency, and latency. In this paper, we propose a spectrum allocation framework based on Reinforcement Learning (RL) for joint mode selection, channel assignment, and power control in D2D communication. The objective is to maximize the overall throughput of the network while ensuring the quality of transmission and guaranteeing low latency requirements of D2D communications. The proposed algorithm uses reinforcement learning (RL) based on Markov Decision Process (MDP) with a proposed new reward function to learn the policy by interacting with the D2D environment. An Actor-Critic Reinforcement Learning (AC-RL) approach is then used to solve the resource management problem. The simulation results show that our learning method performs well, can greatly improve the sum rate of D2D links, and converges quickly, compared with the algorithms in the literature.