基于DDPG的多无人机辅助视频传输资源分配算法研究  

Resource Allocation Algorithm for Multi-UAV Assisted Video Transmission Based on DDPG

作　　者：刘含 王涛[1] 彭振春 朱耀辉 王永志 

机构地区：[1]上海大学通信与信息工程学院,上海200444

出　　处：《工业控制计算机》2025年第2期55-57,共3页Industrial Control Computer

基　　金：国家自然科学基金资助项目(61671011)。

摘　　要：针对车联网中传输带宽和计算资源受限的问题,提出了基于深度强化学习的多无人机辅助边缘计算的车联网视频传输资源分配方案。以无人机集群头为中心资源决策体,联合路边单元和固定车辆作为边缘节点来优化视频传输过程,以减少向目标车辆传输视频时的时延和能耗。该资源分配方案通过深度确定性策略梯度算法来寻求最优的无人机资源分配策略。仿真结果表明,在随机动态环境中的MUECRA-DRL算法性能优于Actor-Critic、深度Q网络和随机算法,将视频传输时延缩短了27%,同时将能量损失减少了24%。For the problem of limited transmission bandwidth and computing resources in Internet of vehicles,this paper proposes a model-free deep reinforcement learning(DRL)based multi-UAV assisted edge computing for Telematics video transmission resource allocation scheme.The unmanned aerial vehicle cluster head(UCH)is used as the central resource decision maker,and the road side unit(RSU)and stationary vehicles are jointly used as the edge nodes to optimize the video transmission process,in order to reduce the latency and energy consumption when transmitting the video to the target vehicles.The resource allocation scheme seeks the optimal UAV resource allocation policy through deep deterministic policy gradient(DDPG)algorithm.

关 键 词：车联网 深度强化学习 无人机 边缘节点 深度确定性策略梯度 

分 类 号：U495[交通运输工程—交通运输规划与管理] TN929.5[交通运输工程—道路与铁道工程]