Future Internet-of-Things (IoT) applications are expected to require increasingly demanding Quality of Service (QoS) levels, jointly in terms of rate, delay, and reliability. In this context, we propose a Multi-Agent Deep Reinforcement Learning (MADRL) based framework to flexibly orchestrate the use of Sub-6 GHz and millimeter Wave bands, while fulfilling such heterogeneous QoS demands. In particular, we design two Personalized Federated MADRL (F-MADRL) methods that enable each user to adapt their learning model to the specific local mobile environment, thereby improving the overall network performance in terms of QoS outages, while reducing signaling exchanges and costs. Numerical evaluations show that our methods largely outperform benchmark MADRL schemes, while closely approaching a fully centralized solution.
Smart Band Association for Wireless IoT Networks: a Personalized Federated Multi-Agent Deep Reinforcement Learning Approach
07.10.2024
1882574 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Communication-efficient and federated multi-agent reinforcement learning
BASE | 2022
|