%0 Journal Article
%T 基于低轨卫星网络的遥感卫星任务计算卸载策略<br>Computation Offloading Strategy for Remote Sensing Satellite Tasks Based on Low-Orbit Satellites Network
%A 杨桂松
%A 李相霏
%A 何杏宇
%J Modeling and Simulation
%P 3130-3138
%@ 2324-870X
%D 2024
%I Hans Publishing
%R 10.12677/mos.2024.133286
%X 在低轨卫星上部署边缘服务器，可以将遥感卫星产生的遥感卫星任务在靠近边缘端进行处理，减少了任务传输时延的同时，极大的缓解了遥感卫星的计算压力。然而，由于低轨卫星自身能源有限，过度使用能源可能会缩短其寿命。此外，当大量遥感卫星任务产生时，计算资源遭遇短缺，从而增加了任务处理的延迟。因此，针对上述问题，提出了一种基于Dueling DQN的遥感卫星任务计算卸载策略，该策略可以充分利用低轨卫星的计算资源，并最大程度上减少任务处理能源消耗。最后，大量仿真结果表明，与其他卸载方法相比，所提策略能有效降低系统任务处理的平均能耗。<br />
Deploying edge servers on low-orbit satellites enables the processing of remote sensing tasks produced by remote sensing satellites at the edge end. It not only reduces task transmission delay but also significantly alleviates the computational pressure on the remote sensing satellites. However, the limited energy of low-orbit satellites presents a challenge, as excessive energy use may shorten their lifespan. Further, the occurrence of many remote sensing tasks could lead to a shortage of computational resources, consequently increasing task processing delay. To address these issues, we propose a remote sensing task computation offloading strategy based on Dueling DQN. This strategy maximizes the utilization of computational resources on low-orbit satellites and minimizes energy consumption. Ultimately, extensive simulation results indicate that compared with other offloading methods, the proposed strategy effectively reduces the system's average energy consumption for task processing.
%K 低轨卫星，遥感卫星任务，深度强化学习，计算卸载<br>Low-Orbit Satellite
%K Remote Sensing Satellite Tasks
%K Deep Reinforcement Learning
%K Computation Offloading
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=87938