Research On Coarse-grained Differentiated Routing In Traffic Engineering Based On MPLS

Internet has brought the revolutionary change for the world, the emerging voice, video and other multimedia services on the network has put forward higher requirements. It's hard to meet the demand just by adding more infrastructures to extend the capacity of the network. Hence, it's more important to improve the current transmission technology while adding the network resource. MPLS technology is proposed based on the needs of network development, as well as the application experience of many mature technologies. It makes full use of the advantage of the switch function and traffic management of the second layer and the flexible routing function of the third layer.Traffic Engineering has attracted much attention of researchers since the date it was proposed; its implementation can be directly used to relieve network congestion and to distribute the network resource reasonably. In addition, it can achieve the QoS guarantee indirectly. The thesis deploys around Traffic Engineering, explores the advantage of MPLS to realize TE combined with existing routing algorithms and the contribution that MPLS TE made to improve the network performance and to satisfy the QoS.A new routing algorithm CDLI (Coarse-grained Differentiated Least Interference routing algorithm) with DiffServ-Aware is presented via analyzing DORA (Dynamic Online Routing Algorithm), PBR (Profile Based Routing) and LIP (Least Interference Path algorithm) in depth. This algorithm is composed of off-line and on-line stages, taking into account both real-time traffic and best-effort traffic. Off-line stage is to determine the shortest path set DP (Disjointed Path) database for real-time traffic, and to identify Link Critical Value by traffic profile information of real-time traffic and DP database; on-line stage is at first to select route in the DP database for real-time traffic, if there is no path to meet their needs, the dynamic routing will be operated. On-line routing algorithm chooses the relative short path for real-time traffic to meet its bandwidth requirements, and for best-effort traffic chooses a lighter load. The simulation results show that compared with the DORA algorithm and CSPF algorithm, the new algorithm can significantly improve network throughput and reduce the average path length of real-time traffic. It guarantees QoS of real-time traffic while improving the utilization of network resources.