Adaptive Channel Assignment for Wireless Mesh Networks Based on Multi-Intelligent Body Reinforcement Learning

Hang Su,Fang Ye,Chunxia Su,Xue Sun
DOI: https://doi.org/10.1109/iceict61637.2024.10671181
2024-01-01
Abstract:With the popularity of emerging applications and the diversification of IoT device needs, wireless mesh networks (WMN) have been recognized as an effective way to solve "last mile" coverage. In the scenario where the network environment status is unknown, an adaptive channel allocation algorithm based on multi-agent reinforcement learning is proposed to solve the problem that static channel allocation in wireless mesh network cannot adapt well to traffic changes in the network environment, resulting in the degradation of link communication quality. Map the channel allocation problem as a multi-agent reinforcement learning process. The state space based on multiple factors such as interference and load is established, and the relationship between exploration and utilization in the algorithm is balanced by using the improved greedy action selection mechanism, and the return function based on link satisfaction is designed to guide the channel selection to improve the quality of link communication. The simulation results show that the proposed algorithm can effectively adapt to the dynamic changes of traffic and source nodes in the network to ensure the quality of user service.
What problem does this paper attempt to address?