通信学报2026，Vol.47Issue(3)：137-155,19.DOI:10.11959/j.issn.1000-436x.2026058
车辆聚类及UAV轨迹优化辅助下基于MADDPG的边缘卸载算法
MADDPG-based edge offloading algorithm assisted by vehicle clustering and UAV trajectory optimization
摘要
Abstract
In order to solve the problems of overload of roadside units and blind area coverage of ground nodes in existing edge computing,a mulit-agent deep deter ministic policy gradient(MADDPG)-based edge offloading optimization algo-rithm assisted by vehicle clustering and unmanned aerial vehicle(UAV)trajectory optimization(VCTOEM)was pro-posed.Firstly,a"vehicle-road-air"joint edge offloading system was constructed,and a multi-objective optimization model was established.Secondly,a vehicle clustering mechanism was designed to aggregate vehicles with remaining re-sources into groups.In addition,an improved particle swarm optimization(PSO)algorithm was introduced to optimize the trajectory of UAV,ensuring efficient coverage of ground computing power blind spots while reducing flight energy consumption and transmission delays during task offloading.Finally,a collaborative offloading algorithm based on multi-agent learning was designed for final decision-making.Simulation results showed that the proposed algorithm could effec-tively adapt to the dynamic changes in the network caused by high-speed vehicle movement,maintaining excellent con-vergence stability in high load scenarios.The average rewards were about 10.5%,29.7%,22.8%,9.1%,6.7%and 9.9%higher than the other six benchmark algorithms,respectively.关键词
边缘卸载/聚类机制/轨迹优化/多智能体学习Key words
edge offloading/clustering mechanism/trajectory optimization/multi-agent learning分类
信息技术与安全科学引用本文复制引用
陈赓,夏聪慧,孔令志,曾庆田..车辆聚类及UAV轨迹优化辅助下基于MADDPG的边缘卸载算法[J].通信学报,2026,47(3):137-155,19.基金项目
国家自然科学基金资助项目(No.61701284) （No.61701284）
山东省自然科学基金资助项目(No.ZR2022MF226) （No.ZR2022MF226）
山东科技大学青年教师人才培养计划(No.BJ20221101) （No.BJ20221101）
青岛市应用基础研究计划基金资助项目(No.19-6-2-1-cg) （No.19-6-2-1-cg）
山东科技大学菁英计划基金资助项目(No.skr21-3-B-048) （No.skr21-3-B-048）
山东省泰山学者计划基金资助项目(No.tstp20250506) The National Natural Science Foundation of China(No.61701284),The Natural Science Foundation of Shan-dong Province(No.ZR2022MF226),The Talented Young Teachers Training Program of Shandong University of Science and Technol-ogy(No.BJ20221101),The Innovative Research Foundation of Qingdao(No.19-6-2-1-cg),The Elite Plan Project of Shandong Uni-versity of Science and Technology(No.skr21-3-B-048),The Taishan Scholar Program of Shandong Province(No.tstp20250506) （No.tstp20250506）
