@inproceedings{8fa6a6afd3814c4ca5462a66c9ce71bb,
title = "Adaptive Non-Primary Channel Access in IEEE 802.11Bn with Deep Reinforcement Learning",
abstract = "Efficient spectrum utilization is critical in modern Wi-Fi networks as traditional systems require primary channel occupancy for transmission, limiting efficiency in overlapping BSS (OBSS) environments. IEEE 802.11bn introduces nonprimary channel access (NPCA) capability, yet optimal decision strategies remain challenging. This paper presents a deep reinforcement learning approach for adaptive NPCA decision-making using Semi-Markov Decision Process formulation with Deep QNetwork. Trainings across varying network density scenarios demonstrate that our approach achieves stable convergence and significant throughput improvements over baseline strategies are observed. The learning algorithm exhibits conservative strategies favoring long-term stability, providing insights for nextgeneration Wi-Fi channel access mechanisms.",
keywords = "Channel Access, Deep Reinforcement Learning, DQN, Non-Primary Channel Access, OBSS, Semi-MDP, Wi-Fi Networks",
author = "Taewon Song",
note = "Publisher Copyright: {\textcopyright} 2026 IEEE.; 40th International Conference on Information Networking, ICOIN 2026 ; Conference date: 14-01-2026 Through 16-01-2026",
year = "2026",
doi = "10.1109/ICOIN68469.2026.11480475",
language = "English",
series = "International Conference on Information Networking",
publisher = "IEEE Computer Society",
pages = "870--873",
booktitle = "40th International Conference on Information Networking, ICOIN 2026",
address = "United States",
}