Deep-Reinforcement-Learning-Based Resource Allocation for Energy Harvesting D2D Communication

Abstract

In this paper, we studied the spectrum resource allocation for the D2D communication under cellular networks. Based on the current situation that energy harvesting is rarely considered in reinforcement learning optimization schemes, this paper combined with energy harvesting to solve the energy consumption problem of communication devices. Specifically, this paper proposes an intelligent allocation scheme for spectrum resources based on the Double DQN algorithm that can realize energy harvesting. The simulation results show that the scheme of this paper obtains superior performance than the traditional DQN scheme and the random allocation scheme. Finally, based on the scheme proposed in this paper, the energy harvesting power that conforms to the energy efficiency performance is presented, and then the overall scheme design is completed.