| Satellite networks are widely used in military,disaster relief and other fields due to their global coverage,uninterrupted communication and other advantages.Especially,Low Earth Orbit(LEO)satellite networks have received extensive attention from various countries and research institutions in recent years due to their lower launch and network construction costs,shorter communication delays and other advantages.In the future 6G network,industry is also researching how to carry Time Sensitive Network(TSN)business through LEO satellite networks for full coverage and support of various differentiated services.However,due to the frequent network topology switching and delay changes of LEO satellite networks,the traditional per-frame scheduling method for time-sensitive flows is difficult to achieve.Therefore,this thesis focuses on static and dynamic routing optimization problems of time-sensitive flows in LEO satellite networks from the perspective of routing,combined with the ATS shaper and dwell time mechanism of the data plane,aiming to meet the Qo S requirements of time-sensitive business.To cope with the high-speed topology change of LEO satellite networks and meet the delay requirements of time-sensitive business,a static routing algorithm FG-DVTR based on fine-grained virtual topology and K-alternate routes is proposed in Chapter 2.FG-DVTR comprehensively considers switching costs,path delays and link utilization rates,and can effectively solve problems such as inter-satellite routing,ground-to-satellite routing,and reliability routing guarantees.In addition,by combining the priority queue of the ATS shaper,FG-DVTR can guarantee the transmission priority and delay determinacy of time-sensitive flows.Experimental results show that compared with other algorithms,FG-DVTR can significantly reduce the number of path switches and network congestion,and reduce the delay and packet loss rate of time-sensitive business.For the case where the static routing scheme may not be able to cope with the dynamic change of network status,causing local congestion and network delay increase,a dynamic routing optimization system DFG-DVTR based on fine-grained virtual topology is proposed in Chapter 3.DFG-DVTR adopts a congestion flow location mechanism based on delay measurement to obtain a set of business on the congested links.In addition,by setting joint optimization objectives of minimizing link utilization rates and adjusting the minimum number of businesses,the number of businesses that need to be re-routed is reduced.Experimental results show that DFG-DVTR can accurately locate congested links in LEO satellite networks,and only by adjusting 22.29% of the traffic on the congested links can alleviate local network congestion and reduce transmission delay.LEO satellite TSN switches using dwell time mechanism determine the waiting time of data packets by calculating the difference between the arrival time and sending time of the data packets,thereby ensuring the determinacy of delay for time-sensitive business.However,the changes in business transmission paths caused by dynamic routing optimization and topology switching will cause large delays in jitter,resulting in frequent calculation and configuration of new dwell times by satellite nodes and generating significant overhead.Therefore,in Chapter 4,we studied how to reduce the problem of delay jitter,established a mathematical model for minimizing delay jitter,decoupled complex constraints among multiple businesses using the Lagrange relaxation method,and proposed two approximate algorithms S-Greedy and K-Greedy.Experimental results show that compared with other algorithms,S-Greedy and K-Greedy can reduce the average value of delay jitter by about 0.7ms,and the maximum value and median of delay jitter are also lower in low load and high load scenarios.To simulate a more realistic LEO satellite network environment,a verification system based on OVS-DPDK for LEO satellite networks is designed and implemented in Chapter5.The system adopts an SDN architecture,modular design,and has functions such as LEO satellite link delay and on/off state simulation,network performance measurement,ATS shaping,routing algorithm Restful services,information caching,traffic injection and analysis,and human-computer interaction.It can effectively perform route algorithm performance testing in a real and practical manner. |