Research On Survivability Techniques Of MPLS Networks

The economic importance of the Internet, the increasing complexity of the network technologies and the huge amount of traffic transported over a single network element require sophisticated survivability mechanisms against failures like fiber cuts or node breakdowns. Network survivability has become a key research issue for IP-based transport networks.Survivability is the capability of a network to maintain service continuity in the presence of faults within the network. Network survivability techniques include two components, survivable network design and restoration schemes. These two components are complementary to each other and cooperate to achieve seamless services upon failures. The focus of this thesis is in MPLS network survivability.In the research of survivable network design, we present a novel matrix-based model and its approximation algorithm of network partition. By using this method, we can divide large network into several sub-networks for pre-planning the topology and protection area for potential failures.In the research of restoration scheme, by using the information of network partition we propose a relevant distributed restoration scheme. Based on this scheme, most faults can be restored in a partitioned sub-network. Then we do research on the spare capacity allocation (SCA) problem. The SCA problem is to decide how much spare capacity should be reserved on links and where to route backup paths to protect given working paths from a set of failure scenarios. Based on the theory of chaos optimization method, we propose an algorithm which is called DSR to solve the SCA problem. We evaluate the effectiveness of the proposed schemes through simulation studies. Our studies show that the DSR algorithm can converges quickly and tranquilly on a stable network.