天地一体化信息网络2025，Vol.6Issue(2)：65-75,11.DOI:10.11959/j.issn.1000-0801.2025019
低轨卫星网络中面向时延优化的跨域边缘计算卸载策略
Cross-Domain Edge Computing Offloading Strategy for Delay-Optimized in Low Earth Orbit Satellite Network
杨桂松 1黄业超 1何杏宇2
作者信息
- 1. 上海理工大学光电信息与计算机工程学院,上海 200093
- 2. 上海理工大学光电信息与计算机工程学院,上海 200093||上海理工大学出版学院,上海 200093
- 折叠
摘要
Abstract
With the rapid development of satellite networks,edge computing on low earth orbit(LEO)satellites has been widely ad-opted to reduce latency and enhance task processing efficiency.Existing approaches primarily employ centralized or distributed man-agement,where the former suffers from high information collection latency and limited flexibility,while the latter incurs high inter-satellite collaboration costs,restricting global optimization.To address these challenges,an edge computing offloading process based on control domain partitioning is designed,and a latency-optimized cross-domain computing offloading strategy using the Actor-Critic al-gorithm is proposed.This strategy enables efficient task scheduling and dynamic optimization across computing control domains.Simu-lation results show that the proposed approach significantly reduces task processing latency and improves task completion rates,demon-strating its applicability and performance advantages in complex LEO satellite networks.关键词
卫星网络/边缘计算/计算卸载/深度强化学习Key words
satellite network/edge computing/computing offloading/deep reinforcement learning分类
信息技术与安全科学引用本文复制引用
杨桂松,黄业超,何杏宇..低轨卫星网络中面向时延优化的跨域边缘计算卸载策略[J].天地一体化信息网络,2025,6(2):65-75,11.
