Improving the Resilience of Transport Networks to Large-scale Failures

Abstract

Telecommunication networks have to deal with fiber cuts, hardware malfunctioning and other failures on a daily basis, events which are usually treated as isolated and unrelated. Efficient methods have been developed for coping with such common failures and hence users rarely notice them. Although less frequently, there also arise cases of multiple failures with catastrophic consequences. Multiple failures can occur for many reasons, for example, natural disasters, epidemic outbreaks affecting software components, or intentional attacks. This article investigates new methods for lessening the impact of such failures in terms of the number of connections affected. Two heuristic-based link prioritization strategies for improving network resilience are proposed. One strategy is built upon the concept of betweenness centrality, while the second is based on what we call the observed link criticality. Both strategies are evaluated through simulation on a large synthetic topology that represents a GMPLS-based transport network. The provisioning of connections in a dynamic traffic scenario as well as the occurrence of large-scale failures are simulated for the evaluation.