| 摘要: |
| 针对低地球轨道遥感卫星资源受限与遥感数据高时效性需求之间的矛盾,研究了移动边缘计算与卫星网络融合架构中的星间协同计算问题。为实现系统能耗与时延的联合优化,提出了一种基于两阶段优化的分布式协同计算卸载方法。首先,将原非凸问题分解为资源分配、传输调度与卸载比例3个可求解的凸子问题,并推导出其闭式最优解。然后,将卫星间的协同决策过程建模为马尔可夫决策过程,利用多智能体近端策略优化算法实现分布式卸载策略的高效学习,以最小化系统总代价。仿真结果表明,与现有基线方法相比,所提算法能够将任务处理时延降低34%,系统能耗减少12%。 |
| 关键词: 低轨遥感卫星 星间协同 计算卸载 资源优化 多智能体强化学习 |
| DOI:10.20079/j.issn.1001-893x.250914002 |
|
| 基金项目: |
|
| Collaborative Computation Offloading and Resource Optimization for Delay-sensitive Remote Sensing Satellites |
| LI Yuxuan,DAI Jianmei |
| (School of Space Information,Space Engineering University,Beijing 101426,China) |
| Abstract: |
| For the conflict between resource constraints on low Earth orbit(LEO) satellites and the demand for high-timeliness remote sensing data,inte-satellite collaborative computing in integrated mobile edge computing(MEC) and satellite networks is investigated.A distributed computation offloading method based on a two-stage optimization strategy is proposed to jointly optimize system energy consumption and latency.The original non-convex problem is decomposed into three tractable convex subproblems:resource allocation,transmission scheduling,and offloading ratio,for which closed-form optimal solutions are derived.Furthermore,satellite collaboration is modeled as a Markov decision process,and the Multi-agent Proximal Policy Optimization(MAPPO) algorithm is employed to efficiently learn distributed offloading strategies for total cost minimization.Simulation results demonstrate that the proposed algorithm reduces task processing latency by 34 |
| Key words: LEO remote sensing satellite inte-satellite cooperation computation offloading resource optimization multi-agent deep reinforcement learning |