| 李亚,张原,王卫岗,郭一枫.车联网边缘计算中的多跳任务卸载决策[J].高技术通讯(中文),2026,36(4):364~373 |
| 车联网边缘计算中的多跳任务卸载决策 |
| Multi-hop task offloading decision in edge computing for Internet of Vehicles |
| |
| DOI：10. 3772 / j. issn. 1002 - 0470. 2026. 04. 004 |
| 中文关键词: 车联网； 移动边缘计算； 集中式动态多跳卸载； 近端策略优化算法 |
| 英文关键词: Internet of Vehicle, mobile edge computing, centralized dynamic multi-hop offloading, proximal policy optimization algorithm |
| 基金项目: |
| 作者 | 单位 | | 李亚 | （河南理工大学物理与电子信息学院焦作 454000） | | 张原 | | | 王卫岗 | | | 郭一枫 | |
|
| 摘要点击次数: 34 |
| 全文下载次数: 27 |
| 中文摘要: |
| 在车联网（Internet of Vehicles，IoV）移动边缘计算卸载系统中，多跳卸载可将任务卸载到路侧单元（road site unit，RSU）范围外的车辆上，可以有效地利用空闲车辆的计算资源，然而由于车辆的高速移动，无法保证节点之间的稳定性。本文提出了一种基于A*算法的集中式动态多跳卸载策略，该策略以最小化任务时延为目标，将问题建模为马尔可夫决策过程（Markov decision process，MDP），采用A*算法来确定车辆任务卸载的最优队列，并利用近端策略优化算法（proximal policy optimization，PPO）进行求解。仿真结果表明，该方法与深度Q学习和贪婪策略相比，任务完成率明显提高，并且平均时延降低了30%左右。 |
| 英文摘要: |
| In Internet of Vehicle (IoV) edge computing offloading system, multi-hop offloading enables tasks to be offloaded to vehicles outside the coverage area of road-side unit (RSU), effectively utilizing the computational resources of idle vehicles. However, the high-speed mobility of vehicles poses challenges to maintaining stability between nodes. This paper proposes a centralized dynamic multi-hop offloading strategy based on the A* algorithm. The strategy aims to minimize task delay by modeling the problem as a Markov decision process (MDP). The A* algorithm is employed to determine the optimal task offloading queue for vehicles, and the proximal policy optimization (PPO) algorithm is utilized for problem-solving. Simulation results show that, compared with deep Q-learning and greedy strategies, the proposed method significantly improves task completion rates and reduces average delay by approximately 30%. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
