Research On Traffic Engineering Based On MPLS

With the rapid growth of the Internet scale and the emergence of a lot of new services, how to optimize the utilization of the network resources, balance the load on the links and improve the Quality of Service (QoS) of the Internet becomes the problem which urgently need for solution. IP is a connectionless protocol and tries its best to meet the requirement of users. It can not ensure enough throughputs and appropriate transmission delay. If there isn't any new method, we will not be able to develop new services on a large scale. In addition, because of the inhere defects of the IP network, a lot of devices and bandwidth of the network can not be used sufficiently. That is one of the main reasons for network congestion. So, how to optimize the utilization of the network resources, balance the load on the links and improve the performance of the network is the urgent problem for solution, and that is also the very problem for Traffic Engineering to solve.In the real applications, Traffic Engineering can be achieved by modulating the IP metrics. And it can be achieved by deploying the overlay model, too. But both of these methods have their localizations. MPLS (Multiprotocol Label Switching) is supposed to be the backbone network technology of the next generation Internet. It takes good use of the switching technology of the core of network and the IP routing technology of the edge of network. With the reasonable control of network traffic, MPLS is able to optimize the utilization of network resources and improve the performance of network. So MPLS is considered as a good tool for Traffic Engineering.MPLS uses explicit routing LSP(ER-LSP) to implement Traffic Engineering. ER-LSP can be created artificially by administrators. And it also can be created automatically by explicit routing algorithms. In this thesis, we bring forward a new online explicit routing algorithm for Traffic Engineering which is named as Load Balance routing algorithm based on Adjacent Region (LBAR). The LBAR computes Adjacent Region of each link based on the network topology and distributes the flows on the link to its Adjacent Region if the resource utilization rate of this link reaches a critical value. We validate the feasibility and validity of this algorithm by NS2 (Network Simulator 2). Simulation results show that it can balance the load on each link more equably, reduce the transmission delay, increase the throughputs of network and reduce the packet loss rate. So it optimizes the utilization of network resources and improves the performance of network.