Scholarlink

<p>Local mutual exclusion (LME) extends traditional mutual exclusion by preventing two neighboring nodes from accessing the Critical Section (CS) simultaneously while allowing concurrent access for non-neighboring nodes. In Flying Ad-Hoc Networks (FANETs), shared resources are hosted on Unmanned Aerial Vehicles (UAVs), and user nodes within a UAV’s transmission range can request access to the resources. However, resource allocation and token management in FANETs remain unexplored. This paper addresses the LME problem for FANETs through Q-learning-based Local Mutual Exclusion (Q-LME). Q-LME employs a token-based LME algorithm with a Q-learning-based leader selection mechanism&#xa0;(QLS). The leader is responsible for coordinating access to shared resources and managing tokens in the distributed system.&#xa0;The proposed QLS mechanism dynamically calculates Q-parameter values by considering environmental factors. Rewards are derived from the number of hops, direction of node movement, link quality, and distance to the resource. The learning rate is calculated based on the number of role changes of a node, while the discount factor reflects the node’s speed. The Q-LME algorithm ensures safety, prevents starvation, and enhances system performance by reducing the leader selection frequency, the average waiting time for CS access, the average number of messages per CS and the average hop count. Additionally, Q-LME improves efficiency and fault tolerance in dynamic FANET environments. </p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Q-LME: Q-learning-based local mutual exclusion algorithm for flying ad hoc networks

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年7月8日

查看全文
  • Sindhuja S,
  • Mary Anita Rajam V

摘要

Local mutual exclusion (LME) extends traditional mutual exclusion by preventing two neighboring nodes from accessing the Critical Section (CS) simultaneously while allowing concurrent access for non-neighboring nodes. In Flying Ad-Hoc Networks (FANETs), shared resources are hosted on Unmanned Aerial Vehicles (UAVs), and user nodes within a UAV’s transmission range can request access to the resources. However, resource allocation and token management in FANETs remain unexplored. This paper addresses the LME problem for FANETs through Q-learning-based Local Mutual Exclusion (Q-LME). Q-LME employs a token-based LME algorithm with a Q-learning-based leader selection mechanism (QLS). The leader is responsible for coordinating access to shared resources and managing tokens in the distributed system. The proposed QLS mechanism dynamically calculates Q-parameter values by considering environmental factors. Rewards are derived from the number of hops, direction of node movement, link quality, and distance to the resource. The learning rate is calculated based on the number of role changes of a node, while the discount factor reflects the node’s speed. The Q-LME algorithm ensures safety, prevents starvation, and enhances system performance by reducing the leader selection frequency, the average waiting time for CS access, the average number of messages per CS and the average hop count. Additionally, Q-LME improves efficiency and fault tolerance in dynamic FANET environments.