Research On Traffic Measurement And Scheduling Method Of LEO Satellite Network

Low-earth orbit satellite networks use a group of satellites in low-earth orbit to provide seamless communication services to ground users,with advantages such as wide network coverage,low deployment costs,and fast transmission speeds,making them an important component of future communication technology.Compared to ground networks,satellite networks have highly dynamic topologies and traffic flows,which can easily lead to problems such as increased data transfer latency,packet loss,and network congestion.Therefore,it is necessary to ensure quality of service through traffic management.To achieve accurate traffic management,precise and granular traffic measurement technology is required to obtain network traffic information,and dynamically manage and optimize network traffic based on real-time traffic information.This thesis focuses on the problem of traffic measurement and traffic management in low-earth orbit satellite networks.The traffic measurement module deployed on a single satellite node is the basic unit of network traffic measurement,capable of measuring traffic information passing through that node.Chapter 2 proposes a single-satellite-node traffic measurement scheme,Multistage Sketch,which supports multiple measurement tasks.To avoid measurement errors caused by the sharing of data structure storage between large and small flows,this scheme separates traffic characteristics to better measure flow sizes using different data structures based on separation results.This thesis implements the Multi-stage Sketch traffic measurement module and evaluates its performance on various measurement tasks.The experimental results show that compared to typical traffic measurement methods,Multi-stage Sketch provides more accurate measurement capabilities at a lower cost.The measurement capability of a single satellite node is limited by satellite power consumption,storage resources,and computational resources,and accurately measuring all traffic in the network solely relying on a single node is very challenging.To address this issue,Chapter 3 proposes a network-level traffic measurement scheme,TM-GALFM,based on ground station assistance.This scheme uses the stronger processing ability of ground stations to accurately measure large flows,generates special data packets containing measurement results,and sends them to satellite nodes.This thesis evaluates the performance of TM-GALFM,and the experimental results show that under the condition of using 600 KB of memory space,the relative error rate of TM-GALFM is only 22.6% of typical traffic measurement methods.TM-GALFM achieves high accuracy in multiple measurement tasks while maintaining a low overhead.After obtaining accurate network traffic information,traffic management is required to better serve business needs.Therefore,Chapter 4 studies the problem of low-earth orbit satellite network traffic planning and scheduling.For the scenario where the satellite network has link utilization constraints,this thesis proposes a traffic planning algorithm based on hierarchical graphs and K alternate routes.The routing scheme calculated by this algorithm will be used as the initial routing for the business.However,when the business changes and the initial routing is no longer optimal,this thesis proposes two dynamic traffic scheduling algorithms based on complete distributed and semi-distributed approaches to reduce link load.Finally,this thesis uses experiments to evaluate the performance of the proposed schemes,and the experimental results show that these algorithms can alleviate network congestion while minimizing adjustment costs as much as possible.