Scholarlink

With high flexibility, easy deployment and controllability, UAVs have be-come an important part of future wireless networks, which can be flexibly deployed according to communication needs. UAVs hover over each sensor node to communicate with it. For the UAV path planning problem, combined with the UAV energy, flight speed and other constraints, to minimize the UAV data collection task completion time to establish a mathematical optimization problem for the objective function. In this paper, the problem is modeled as Markov decision process and solved by DQN optimization. The simulation results show that the reward value of DQN is higher than that of Q learning, which proves the superiority of the algorithm.

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

UAV-Assisted Data Collection for WSN: A Deep Reinforcement Learning Method

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shengchun Wang,
  • Kunfu Wang,
  • Weifeng Wang,
  • Pengyi Zhang,
  • Qijia Gu,
  • Baiqiao Huang

摘要

With high flexibility, easy deployment and controllability, UAVs have be-come an important part of future wireless networks, which can be flexibly deployed according to communication needs. UAVs hover over each sensor node to communicate with it. For the UAV path planning problem, combined with the UAV energy, flight speed and other constraints, to minimize the UAV data collection task completion time to establish a mathematical optimization problem for the objective function. In this paper, the problem is modeled as Markov decision process and solved by DQN optimization. The simulation results show that the reward value of DQN is higher than that of Q learning, which proves the superiority of the algorithm.