Research Article
Improving Reliability in Hospital WLANs with RL-Based Channel Selection and Contention Window Adaptation
@INPROCEEDINGS{10.1007/978-3-032-38657-1_32, author={Kousuke Kinoshita and Kento Takabayashi}, title={Improving Reliability in Hospital WLANs with RL-Based Channel Selection and Contention Window Adaptation}, proceedings={Body Area Networks. 20th EAI International Conference, BodyNets 2025, Okayama, Japan, December 2--4, 2025, Proceedings}, proceedings_a={BODYNETS}, year={2026}, month={10}, keywords={reinforcement learning channel selection contention window WLAN}, doi={10.1007/978-3-032-38657-1_32} }- Kousuke Kinoshita
Kento Takabayashi
Year: 2026
Improving Reliability in Hospital WLANs with RL-Based Channel Selection and Contention Window Adaptation
BODYNETS
Springer
DOI: 10.1007/978-3-032-38657-1_32
Abstract
In recent years, Wireless Local Area Networks (WLANs) have been introduced into hospitals to support various medical applications such as monitoring of vital signs, telemetry, and rapid information sharing among staff. However, unsupervised deployments and the coexistence of heterogeneous devices, for example patient-owned wearables and unmanaged access points, often lead to severe interference and transmission failures. To address these issues, we investigate a scheme that combines channel selection with contention window (CW) adaptation using reinforcement learning (RL). We compare several RL algorithms, such as ε-greedy, Upper Confidence Bound (UCB), and UCB1-Tuned, against conventional CSMA/CA. Computer simulation results show that the proposed learning-based configuration reduces the transmission failure rate. A trade-off is observed in throughput, which decreases in some settings, indicating that further joint optimization of channel choice and CW parameters is required.