中国邮电高校学报(英文) ›› 2018, Vol. 25 ›› Issue (5): 67-74.doi: 10.19682/j.cnki.1005-8885.2018.0023

• Networks • 上一篇    下一篇

On-line learning algorithm for dynamic sensitivity control in IEEE 802.11ax network

李齐山,胡智群,温向明,路兆铭,亓航   

  1. 北京邮电大学
  • 收稿日期:2018-03-22 修回日期:2018-05-21 出版日期:2018-10-18 发布日期:2018-10-18
  • 通讯作者: 李齐山 E-mail:515600789@qq.com
  • 基金资助:
    国家自然基金;中央在京高校重大成果转化项目

On-line learning algorithm for dynamic sensitivity control in IEEE 802.11ax network

  • Received:2018-03-22 Revised:2018-05-21 Online:2018-10-18 Published:2018-10-18
  • Contact: Qi-Shan LI E-mail:515600789@qq.com
  • Supported by:
    National Natural Science Foundation of China

摘要: The popularity of IEEE 802.11 based Wireless Local Area Network (WLAN) increased significantly in recent years and resulted in the dense deployment of WLANs. While densification can contribute to increasing coverage, it could also lead to increasing interference and cannot insure high spatial reuse due to the current physical carrier sensing of IEEE 802.11. To tackle these challenges, the dynamic sensitivity control (DSC) is considered in IEEE 802.11ax, which dynamically selects the appropriate carrier sensing threshold (CST) to improve spectrum efficiency and enhance spatial reuse in densely deployed network. A dynamic Q-learning based CST selection method is proposed to enable a network to select the optimal CST according to the channel condition. Simulation results show that the propsoed scheme provides 40% aggregate throughput gain of a dense network when compared with legacy IEEE 802.11.

关键词:

 IEEE 802.11ax, reinforcement learning, Dynamic Sensitivity Control, dense network

Abstract: The popularity of IEEE 802.11 based Wireless Local Area Network (WLAN) increased significantly in recent years and resulted in the dense deployment of WLANs. While densification can contribute to increasing coverage, it could also lead to increasing interference and cannot insure high spatial reuse due to the current physical carrier sensing of IEEE 802.11. To tackle these challenges, the dynamic sensitivity control (DSC) is considered in IEEE 802.11ax, which dynamically selects the appropriate carrier sensing threshold (CST) to improve spectrum efficiency and enhance spatial reuse in densely deployed network. A dynamic Q-learning based CST selection method is proposed to enable a network to select the optimal CST according to the channel condition. Simulation results show that the propsoed scheme provides 40% aggregate throughput gain of a dense network when compared with legacy IEEE 802.11.

Key words:

IEEE 802.11ax, reinforcement learning, Dynamic Sensitivity Control, dense network