| 摘要: |
| 卫星算力网络(Satellite Computing Network,SCN)在支撑实时服务时,面临着卫星高速运动引发的拓扑时变、链路资源随机性波动,以及时延、带宽等多目标需求冲突等路由保障难题。然而,现有的路由方法多基于确定性拓扑快照与单目标优化,难以应对网络状态的随机波动,导致路径频繁失效与服务失衡。为解决上述静态模型与动态环境失配、单目标与多目标失衡的问题,提出了一种基于最大概率的确定性需求保障路由算法(Max-probability Deterministic Demand-guaranteed Routing Algorithm,MP-DDGRA)。首先,构建统计时变图模型,将链路时延、带宽占用及节点缓存建模为高斯随机变量,实现动态网络资源的时空关联表征。其次,设计动态权重调整机制,根据业务剩余时间比例与链路负载状态,自适应平衡时延与带宽的差异化需求。最后,结合贪心策略Dijkstra框架,在多项式时间内求解满足业务需求的最大概率路径。仿真结果显示,较于传统的OSPF(Open Shortest Path First)算法以及TOSPF(Transient OSPF)算法,该算法平均往返时延分别降低29.35%、13.05%,算力任务传输总时延分别减少18%、11.3%。 |
| 关键词: 卫星算力网络 确定性路由 统计时变图 动态权重调整 最大概率保障 |
| DOI:10.20079/j.issn.1001-893x.250623001 |
|
| 基金项目: |
|
| Max-probability Routing Algorithm for Satellite Computing Networks |
| LI Huan,HU Xinqian,SHANG Kefeng,LIANG Huicui,SUN Xun |
| (1.Southwest China Institute of Electronic Technology,Chengdu 610036,China;2.School of Electronic Science and Technology,Xiamen University,Xiamen 361100,China;3.Intelligent Measurement and Control and Space-based Information Application Laboratory,Chengdu 610036,China;4.School of Electronic Information and Electrical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China?Symbol`@@) |
| Abstract: |
| The satellite computing network(SCN) faces routing challenges in supporting real-time services,including topology variations caused by high-speed satellite motion,random fluctuations in link resources,and conflicting demands for multiple objectives,such as latency and bandwidth.However,existing routing methods,mostly based on deterministic topology snapshots and single-objective optimization,struggle to cope with random fluctuations in network state,leading to frequent path failures and service imbalances.Therefore,to address the mismatch between static models and dynamic environments and the imbalance between single and multiple objectives,a maximum-probability deterministic demand-guaranteed routing algorithm(MP-DDGRA) is proposed.First,a statistical time-varying graph model is constructed,modeling link latency,bandwidth utilization,and node cache as Gaussian random variables to represent the spatiotemporal correlations of dynamic network resources.Second,a dynamic weight adjustment mechanism is designed to adaptively balance differentiated latency and bandwidth demands based on the proportion of service remaining time and link load status.Finally,by combining the greedy Dijkstra framework,the maximum probability path that meets service requirements is solved in polynomial time.Simulation results show that compared with the traditional Open Shortest Path First(OSPF) and Transient OSPF(TOSPF) algorithms,the proposed algorithm reduces average round-trip latency by 29.35 |
| Key words: satellite computing network deterministic routing statistical time-varying graph dynamic weight adjustment maximum probability guarantee |