Energy-Efficient Resource Allocation in Uplink NOMA Systems with Deep Reinforcement Learning

Abstract

Non-orthogonal multiple access (NOMA) is regarded as a promising technology to satisfy the huge access demand and data rate requirements of the next generation network. In this paper, we investigate the joint subcarrier assignment and power allocation problem in an uplink multi-user NOMA system to maximize the energy efficiency (EE) while ensuring the quality-of-service (QoS) of all users. Different from conventional resource allocation methods, we propose a two-step deep reinforcement learning (DRL) based algorithm to solve this non-convex and dynamic optimization problem. In particular, with the current channel conditions as input, we design a deep q-network (DQN) to output the optimum subcarrier assignment policy, then use a deep deterministic policy gradient (DDPG) network to dynamically output the transmit power of all users, and finally adjust the entire resource allocation policy by updating the weights of neural networks according to the feedback of the system. Simulation results show that our DRL based algorithm can provide better EE under various transmit power limitations compared with other approaches.