Research On SDN Segment Routing Algorithms For Traffic Engineering

With the rapid development of communication network infrastructure,the scale of Internet is expanding.The rapid growth of network resource demand puts forward strict requirements for traffic scheduling ability,and traffic engineering problems are becoming increasingly important.The further innovation of network architecture can make full use of traffic engineering technology to improve network throughput,increase data flow transmission performance,and ensure the availability and reliability of the network.Combining Segment Routing(SR)and Software-Defined Networking(SDN),this thesis studies segment list management,load balancing,quality of service(QoS)routing and other issues involved in actual traffic engineering.The research results are as follows:(1)In view of the large number of labels in the source routing,the high performance requirements for network devices,and the physical limitation of maximum stack depth(MSD),reversed longest match relay push(RLMRP)of segment list management algorithm is proposed,and the number of network resources are analyzed through mathematical model.Experiments comparing with multiple algorithms prove that the RLMRP algorithm can make multiple links share the same label according to the initial forwarding table,which can reduce the number of labels,the packet header overhead and the number of flow entries.RLMRP can decouple routing calculation and segment list calculation.It is also possible to push stack at relay nodes of any long path for forwarding.(2)In view of the current emerging business requirements represented by cloud computing and big data,the existing Multi-Protocol Label Switching(MPLS)network has some problems such as complex protocol,poor scalability,operation and maintenance difficulties,etc.The segment routing load balancing algorithm based on Constrained K-Shortest Paths(CKSP)is proposed,which can support rerouting by binding group table to port.The experimental results show that the CKSP can nonuniform weight link utilization and hop number,increase the network throughput,smooth the traffic distribution,reduce the average delay of data flow and the total network loss rate.(3)In view of the QoS performance requirements of multi service characteristic data flow in current network,this thesis introduces the differentiated service model,realize the measuring method of QoS index in SDN,and proposes the ant colony optimization based traffic control(ACOTC)algorithm combined with the differentiated services code point of data flow.The algorithm uses meter table and queue to control the real-time transmission rate of data stream,and finds the optimal path with the minimum QoS metric through ant colony iterative routing.Experimental results show that the ACOTC algorithm utilizes global network view and traffic control of SDN,can increase the residual bandwidth,reduce packet loss rate,delay and jitter of data flow,and improve the QoS routing effect.