Title:Federated Deep Reinforcement Learning-Based Intelligent Channel Access in Dense Wi-Fi Deployments

Abstract:The IEEE 802.11 MAC layer utilizes the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) mechanism for channel contention and access. However, in densely deployed Wi-Fi scenarios, intense competition may lead to packet collisions among users. Although many studies have used machine learning methods to optimize channel contention and access mechanisms, most of them are based on AP-centric single-agent models or distributed models, which still suffer poor generalization and insensitivity to dynamic environments. To address these challenges, this paper proposes an intelligent channel contention access mechanism that combines Federated Learning (FL) and Deep Deterministic Policy Gradient (DDPG) algorithms. Additionally, an FL model training pruning strategy and weight aggregation algorithm are designed to enhance the effectiveness of training samples and reduce the average MAC delay. We evaluate and validate the proposed solution using NS3-AI framework. Simulation results show that in static scenarios, our proposed scheme reduces the average MAC delay by 25.24% compared to traditional FL algorithms. In dynamic scenarios, it outperforms Average Federated Reinforcement Learning (A-FRL) and distributed Deep Reinforcement Learning (DRL) algorithms by 25.72% and 45.9%, respectively.