《计算机应用》唯一官方网站 ›› 2026, Vol. 46 ›› Issue (8): 2524-2532.DOI: 10.11772/j.issn.1001-9081.2025070842

• 先进计算 • 上一篇    下一篇

用户为中心的卫星边缘计算架构优化任务卸载

陈娟1(), 陈玉杰1, 吴宗玲2, 田谛1, 钟杰1   

  1. 1.西华大学 计算机与软件工程学院,成都 610039
    2.西南交通大学 信息科学与技术学院,成都 611756
  • 收稿日期:2025-07-28 修回日期:2025-10-07 接受日期:2025-10-09 发布日期:2025-11-05 出版日期:2026-08-10
  • 通讯作者: 陈娟
  • 作者简介:陈娟(1986—),女,山西朔州人,讲师,博士,CCF会员,主要研究方向:智能边缘计算、深度强化学习
    陈玉杰(2000—),男,四川威远人,硕士研究生,主要研究方向:边缘计算、云计算
    吴宗玲(1987—),男,重庆人,工程师,博士,主要研究方向:嵌入式物联网
    田谛(2001—),男,四川德阳人,硕士研究生,主要研究方向:空天地一体化网络
    钟杰(1999—),男,四川广安人,硕士研究生,主要研究方向:卫星边缘计算、移动边缘计算。
  • 基金资助:
    教育部专项(HZKY20220578);省级高校重点实验室开放基金资助项目(KL2411);省级高校重点实验室开放基金资助项目(2024WYJ02)

User-centric satellite edge computing architecture for task offloading optimization

Juan CHEN1(), Yujie CHEN1, Zongling WU2, Di TIAN1, Jie ZHONG1   

  1. 1.School of Computer and Software Engineering,Xihua University,Chengdu Sichuan 610039,China
    2.School of Information Science and Technology,Southwest Jiaotong University,Chengdu Sichuan 611756,China
  • Received:2025-07-28 Revised:2025-10-07 Accepted:2025-10-09 Online:2025-11-05 Published:2026-08-10
  • Contact: Juan CHEN
  • About author:CHEN Yujie, born in 2000, M. S. candidate. His research interests include edge computing, cloud computing.
    WU Zongling, born in 1987, Ph. D., engineer. His research interests include embedded internet of things.
    TIAN Di, born in 2001, M. S. candidate. His research interests include air-space-ground integrated network.
    ZHONG Jie, born in 1999, M. S. candidate. His research interests include satellite edge computing, mobile edge computing.
  • Supported by:
    Special Project of Ministry of Education(HZKY20220578);Provincial University Key Laboratory Open Fund(KL2411)

摘要:

卫星边缘计算(SEC)通过将任务卸载至卫星服务器提高远程用户的计算效率;然而,由于用户设备(UE)间的信号干扰,无线传输质量可能会影响SEC系统的性能,因此,提出一种以用户为中心的SEC(UCSEC)架构,根据用户实时位置和网络状况动态分配接入点(AP)集群。针对UCSEC中的动态任务卸载问题,设计一种基于门控循环单元(GRU)与多头注意力机制的主从式多智能体近端策略优化(MSPPO_GMA)算法。该算法采用集中控制-分布执行框架,即部署在SEC服务器上的主智能体负责监控全局资源约束并协调从智能体,而部署在UE上的从智能体根据网络状态执行任务卸载和资源分配。在MSPPO_GMA算法中,GRU捕捉网络状态的时间相关性,而多头注意力机制关注任务卸载和资源分配的重要信息。实验结果表明,与MAPPO和MATD3等算法相比,MSPPO_GMA算法的任务完成延迟降低了3.7%~13.6%,UE能耗降低了6.5%~13.2%。

关键词: 卫星边缘计算, 多智能体, 深度强化学习, 任务卸载, 资源分配

Abstract:

Satellite Edge Computing (SEC) improves computational efficiency for remote users by offloading tasks to satellite servers. However, SEC system performance may be affected by wireless transmission quality, since there is signal interference among User Equipments (UEs). To this end, we proposed a User-Centric SEC (UCSEC) architecture to allocate Access Point (AP) clusters dynamically according to the real-time location of users and network conditions. To handle dynamic task offloading in UCSEC, we designed a Master-Slave multi-agent Proximal Policy Optimization with Gated Recurrent Unit (GRU) and Multi-head Attention mechanism (MSPPO_GMA) algorithm. The algorithm adopted a centralized control-distributed execution framework. The master agent, deployed on the SEC server, monitored global resource constraints and coordinated slave agents, while slave agents deployed on the UE executed task offloading and resource allocation according to network states. In MSPPO_GMA algorithm, GRU captured temporal correlations of network states, while multi-head attention focused on critical information of task offloading and resource allocation. Experimental results demonstrate that compared with MAPPO, MATD3 and other algorithms, MSPPO_GMA algorithm reduces task completion latency by 3.7% to 13.6% and UE energy consumption by 6.5% to 13.2%.

Key words: Satellite Edge Computing (SEC), multi-agent, deep reinforcement learning, task offloading, resource allocation

中图分类号: