低轨卫星网络中业务图驱动的星间协作计算方案  

Inter-Satellite Cooperative Computing Scheme Driven by Business Graph in LEO Satellite Network

作　　者：国晓博[1] 任智源[2] 程文驰[2] 纪哲 GUO Xiaobo;REN Zhiyuan;CHENG Wenchi;JI Zhe(The 54th Research Institute of China Electronics Technology Group Corporation,Shijiazhuang 050081,China;State Key Laboratory of Integrated Services Networks,XiDian University,Xi’an 710071,China;School of Information and Communication Engineering,Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]中国电子科技集团第五十四研究所,河北石家庄050081 [2]西安电子科技大学综合业务网理论及关键技术国家重点实验室,陕西西安710071 [3]北京邮电大学信息与通信工程学院,北京100876

出　　处：《天地一体化信息网络》2021年第2期35-44,共10页Space-Integrated-Ground Information Networks

基　　金：科技部重点研发计划资助项目(No.2019YFB1803305)。

摘　　要：低轨卫星网络具备覆盖范围广、星地链路短等优点,成为当前的热点研究领域。针对将业务卸载到云计算中心产生过高传输时延的问题,提出在低轨卫星网络中实现业务图驱动的星间协作计算方法。该方法首先依托加权时间扩展图模型,屏蔽低轨卫星网络的动态性;之后联合多颗卫星的计算资源和星间链路资源,将有向无环图形式的业务调度到传输路径内的卫星上,实现业务在低轨卫星网络内的“边传输边计算”。仿真结果表明,当数据量取5 MB时,星间协同计算的时延性能相比地面云计算提高了58.9%。Low earth orbit-satellite network(LEO-SN)has the advantages of wide coverage and short satellite-to-earth link,and has become a current hot research fi eld.A business graph-driven inter-satellite collaborative computing method in LEO-SN was proposed to solve the problem of high transmission delay when the task was offl oaded to the cloud computing center.This method shielded the dynamics of LEO-SN by relying on the weighted time expansion graph model,combined multiple satellite computing resources,and dispatches tasks in the form of a directed acyclic graph to the satellites in the transmission path,thereby realizing the“computing while transmitting”of user services in the satellite cluster.The simulation results showed that when the data volume was 5 MB,the latency performance of on-orbit collaborative computing was improved by 58.9%compared to ground cloud computing.

关 键 词：低轨卫星网络 时间扩展图 星间协作计算 

分 类 号：TN927.23[电子电信—通信与信息系统]