Scholarlink

<p>Flying Ad Hoc Networks (FANETs), composed of unmanned aerial vehicles (UAVs), provide a flexible and rapidly deployable platform for remote data collection and mission-critical operations. However, their performance is severely challenged by high node mobility, frequent topology variations, and stringent energy constraints. These characteristics make efficient network organization particularly difficult, especially for clustering mechanisms. In this context, the cluster head (CH) election problem can be formulated as an NP-hard combinatorial optimization task, requiring adaptive and computationally efficient solutions capable of handling dynamic and uncertain environments. This work introduces a distributed Fuzzy Q-Learning (FQL) clustering framework that integrates the short-term responsiveness of Fuzzy Logic (FL) with the long-term adaptivity of Q-Learning (QL). While the FL controller evaluates CH suitability using multi-criteria metrics, the QL optimizer refines these decisions over time to learn energy-efficient policies. Extensive simulations demonstrate that the FQL framework significantly outperforms LEACH, ESOFCluster (FL-only), and QL-Clustering across all stability, energy, and reliability metrics. Quantitatively, FQL maintains a Packet Delivery Ratio (PDR) of 85–90% even under extreme mobility (30 m/s) and high density (<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(N=90\)</EquationSource> </InlineEquation>), where LEACH drops below 70%. Furthermore, FQL achieves the longest cluster lifetimes and the latest First Node Die (FND) and Last Node Die (LND) milestones, indicating superior energy balancing and network longevity. By effectively mitigating the "cold-start" limitations of pure Q-Learning and the static constraints of pure Fuzzy Logic, the proposed FQL emerges as a robust, scalable, and energy-aware solution for next-generation FANETs.</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

FQL: a fuzzy Q-learning framework for reinforcement-driven clustering in flying Ad Hoc networks

  • 期刊论文

  • 发表日期：

    2026年7月16日

查看全文
  • Mohamed Aissa,
  • Badia Bouhdid,
  • Adel Ben Mnaouer,
  • Bassem Ouni,
  • Maroua Abdelhafidh,
  • Mahmood Al-Bahri,
  • Khalfan ALmayah

摘要

Flying Ad Hoc Networks (FANETs), composed of unmanned aerial vehicles (UAVs), provide a flexible and rapidly deployable platform for remote data collection and mission-critical operations. However, their performance is severely challenged by high node mobility, frequent topology variations, and stringent energy constraints. These characteristics make efficient network organization particularly difficult, especially for clustering mechanisms. In this context, the cluster head (CH) election problem can be formulated as an NP-hard combinatorial optimization task, requiring adaptive and computationally efficient solutions capable of handling dynamic and uncertain environments. This work introduces a distributed Fuzzy Q-Learning (FQL) clustering framework that integrates the short-term responsiveness of Fuzzy Logic (FL) with the long-term adaptivity of Q-Learning (QL). While the FL controller evaluates CH suitability using multi-criteria metrics, the QL optimizer refines these decisions over time to learn energy-efficient policies. Extensive simulations demonstrate that the FQL framework significantly outperforms LEACH, ESOFCluster (FL-only), and QL-Clustering across all stability, energy, and reliability metrics. Quantitatively, FQL maintains a Packet Delivery Ratio (PDR) of 85–90% even under extreme mobility (30 m/s) and high density ( \(N=90\) ), where LEACH drops below 70%. Furthermore, FQL achieves the longest cluster lifetimes and the latest First Node Die (FND) and Last Node Die (LND) milestones, indicating superior energy balancing and network longevity. By effectively mitigating the "cold-start" limitations of pure Q-Learning and the static constraints of pure Fuzzy Logic, the proposed FQL emerges as a robust, scalable, and energy-aware solution for next-generation FANETs.