D2D Resource Allocation Optimization Algorithm Based on Deep Reinforcement Learning

In the cellular network framework, the resource allocation problem for Device-to-Device (D2D) communication technology is particularly critical. The core issue is how to efficiently allocate channel resources to D2D users while minimizing interference to cellular users (CU), ensuring the overall performance and Quality of Service (QoS) of the network. To address this challenge, This letter has proposed a resource allocation strategy based on deep reinforcement learning algorithms, aimed at dynamically adjusting channel allocation and transmission power in D2D communication. Firstly, an in-depth study and analysis of the conflicts and interference issues between D2D communication and cellular users in the cellular network environment regarding channel reuse are conducted. By quantitatively analyzing the impact of signal interference on communication quality, key factors requiring optimization are identified. Secondly, Double-Actor Critic (D-AC) algorithm is developed for dynamically allocating communication channels and adjusting transmission power. This algorithm iteratively optimizes D2D channel reuse and power control strategies, aiming to balance the efficient use of channel resources with the minimization of signal interference. Thirdly, the D-AC algorithm is implemented and evaluated for its effectiveness in enhancing system throughput, reducing collision probabilities, and maintaining quality of service standards. By comparing with existing resource management schemes, the advantages of the D-AC algorithm are demonstrated.