Scholarlink

Abstract <p>This scholarly article elucidates a holistic approach for enhancing the operational efficacy of UAV-MEC (Unmanned Aerial Vehicle enabled Mobile Edge Computing Server) networks, focusing on the critical aspects of task offloading, trajectory control, energy utilization, and queuing stability. We introduce the Joint Offloading-Trajectory Optimization (JOTO) framework, which facilitates multi-UAV-MEC networks in offloading computational tasks from mobile devices (MDs) to UAV-MECs, and subsequently to base station (BS). The framework employs a hybrid optimization framework where Independent Proximal Policy Optimization (IPPO) handles decentralized trajectory control and maneuvering, while a Lyapunov-based convex optimization model determines adaptive task offloading ratios. This dual methodology enables each UAV-MEC to dynamically adjust its trajectory and offloading strategy to minimize energy consumption, stabilize queues, and maximize MD energy efficiency. Crucially, our model utilizes channel gain information, rather than accurate locational data, for near-optimal decision-making. Empirical simulations reveal that JOTO surpasses other deep reinforcement learning (DRL)-based frameworks in energy consumption, service duration, and decision-making efficacy, achieving performance competitive with convex optimization benchmarks. By jointly optimizing trajectory control and cooperative task offloading, our methodology significantly advances the capabilities of UAV-enabled MEC networks for real-world edge computing scenarios.</p> Graphical abstract <p></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

Empowering multi-UAV networks: a synergistic approach to task offloading and trajectory optimization

  • 期刊论文

  • 发表日期：

    2026年2月12日

查看全文
  • Pronab Kumar Adhikari,
  • Abhinav Tomar

摘要

Abstract

This scholarly article elucidates a holistic approach for enhancing the operational efficacy of UAV-MEC (Unmanned Aerial Vehicle enabled Mobile Edge Computing Server) networks, focusing on the critical aspects of task offloading, trajectory control, energy utilization, and queuing stability. We introduce the Joint Offloading-Trajectory Optimization (JOTO) framework, which facilitates multi-UAV-MEC networks in offloading computational tasks from mobile devices (MDs) to UAV-MECs, and subsequently to base station (BS). The framework employs a hybrid optimization framework where Independent Proximal Policy Optimization (IPPO) handles decentralized trajectory control and maneuvering, while a Lyapunov-based convex optimization model determines adaptive task offloading ratios. This dual methodology enables each UAV-MEC to dynamically adjust its trajectory and offloading strategy to minimize energy consumption, stabilize queues, and maximize MD energy efficiency. Crucially, our model utilizes channel gain information, rather than accurate locational data, for near-optimal decision-making. Empirical simulations reveal that JOTO surpasses other deep reinforcement learning (DRL)-based frameworks in energy consumption, service duration, and decision-making efficacy, achieving performance competitive with convex optimization benchmarks. By jointly optimizing trajectory control and cooperative task offloading, our methodology significantly advances the capabilities of UAV-enabled MEC networks for real-world edge computing scenarios.

Graphical abstract