Joint Subcarrier and Power Allocation in Mobile Scenario of the OFDM Systems Based on Deep Reinforcement Learning

The increasing number of base station (BS) service users has made spectrum resources more valuable, leading to the need for efficient resource allocation. To address this issue, this paper proposes a novel deep reinforcement learning (DRL) architecture that solves the subcarrier-user matching and subcarrier power allocation problem in orthogonal frequency division multiple (OFDM) systems. This approach improves the system spectral efficiency (SE) by reducing inter-subcarrier interference (ICI) through subcarrier matching and power allocation, while ensuring that the minimum rate requirements for users (mobile or stationary) are met. We approach this problem by dividing it into two interconnected components: the subcarrier matching part (SMP) and the subcarrier power allocation part (SPAP). To address these two parts, the paper proposes a DRL method based on joint allocation of resources by three agents (JAoR-TA). In SMP, the REINFORCE algorithm is used to match subcarriers, and the results are then given to SPAP. In SPAP, two Actor-Critic network frameworks are proposed to address the subcarrier power allocation problem. Based on the simulation results, it has been observed that the JAoR-TA algorithm outperforms the REINFORCE algorithm in mobile scenarios, as it achieves a higher system SE.