# Enhancing WiFi Multiple Access Performance with Federated Deep Reinforcement Learning ###### tags: `Paper` ## Abstract - (CSMA/CA) 是 IEEE 802.11 的主幹 MAC 協議網絡 ## Introduction ## System Model 1. Markov Decision Process and Monte Carlo Reward Estimation Method ![](https://i.imgur.com/lAYtVIq.png) ![](https://i.imgur.com/RZgEFra.png) ![](https://i.imgur.com/43UeVkI.png) 2. Reinforcement Learning 3. Semi-gradient Algorithm in Deep Learning 4. Federated Learning and FRMA ![](https://i.imgur.com/jYxGoce.png) ## Performance Evaluation : Simulations CPU : (Intel Core i9-9700k) GPU : (NVIDIA GeForce GTX 2080Ti) 1. Performance review of IEEE 802.11 CSMA/CA 2. Simulation Setup ![](https://i.imgur.com/puOkRfA.png) 3. Results ## Conclusion 1. Distributed DRL 2. Each station is installed with a QNN and trained by its local information