Federated Reinforcement Learning-Based Resource Allocation in D2D-Enabled 6G

The current 5G and conceived 6G era with ultra -high density, ultra -high frequency bandwidth, and ultra -low latency can support emerging applications like Extended Reality (XR), Vehicle to Everything (V2X), and massive Internet of Things (IoT). With the rapid growth of transmission rate requirements and link numbers in the wireless communication network, how to allocate resources reasonably and further improve spectrum utilization challenges the traditional approaches. To address these problems, technologies such as device -to -device (D2D) communication and machine learning (ML) are introduced to the traditional cellular communication network to improve network performance. However, due to the interference caused by spectrum reusing, efficient resource allocation for both cellular users and D2D users is necessary. In this article, we consider underlay mode D2D-enabled wireless network to improve the spectrum utilization, and deep reinforcement learning (DRL)-based federated learning (FL) -aided decentralized resource allocation approach to maximize the sum capacity and minimize the overall power consumption while guaranteeing the quality of service (QoS) requirement of both cellular users and D2D users. The performance of the proposed schemes is evaluated through simulations under 5G millimeter -wave (mm -wave) and 6G terahertz (THz) scenarios separately. The simulation results show that the proposal achieves significant network performance compared with the baseline algorithms.