MPLS-based Network Fault Recovery Research

With the rapid development of the network scale and a large number of new business, how to improve the network quality of service has become the urgent need to address one of the issues. MPLS uses label to execute fast packet switching in network, and integrates switching technique and IP routing technique for purpose of constructing a new kind of network structure with higher stability and higher agility. MPLS, a next generation backbone architecture, it can speed up packet forwarding to destination by label switching. However, if there exists no pre-established backup path when the network is failed, MPLS frames cannot be forwarded to destination. Therefore, fault recovery has become an important research area currently.Research for fault recovery based on MPLS is not only one of important research fields but also the base of development of next generation network. Research for fault recovery of MPLS mostly focuses on two aspects, Protection Switching and Rerouting. This thesis focuses mainly on the Protection Switching. At present, Protection Switching has two famous methods, Makam and Haskin, other methods basically come into being on base of them. When the place of failure is away from ingress node, a problem will come out. For Makam, more packets will be loosed. For Haskin, path of Packet retrace is longer and delay time of packet is longer and disorder of packet is prone to emerging.Based on research for present methods, the thesis presents an improved method of MPLS-based recovery. It combines with the advantages of Makam, Haskin and Hundessa, using the Reverse Backup Path to solve the wrapping of data back to the path is too long. As long as a reverse path is established, the data flow will switch on it directly. At the same time, the improved method combined with the Hundessa of Buffer and Tag to resolve the issue out of order packets. Improvement program to the next during the switching of a packet marked with a marker, indicate that it is the need to reverse the last packet wrap. When once again receive this packet, we will know prior to sending the data packets sent have been all around the back, and then transmit the data Buffer out, so that around the back to back packets marked with a special label, to resolve the problem of packet loss. Later, add the Ping protocol to detect whether the path is open, whether the fault has been resolved, and after recovery it will switch back to the original data flow path of the work, in this way, it saves time for recovery and reduce delays. In the end of the thesis, we use NS-2(Network Simulator-version 2), a tool of network simulation, to implement simulation for the improved method and to compare with present methods. The result of simulation experiment shows that improved method of MPLS-based recovery has less packet disorder and much lower delay and packet losses.