Telecommunication networks are pervading every human activity. The increasing number of activities relying on Internet and cloud services call for high quality and reliable services. Networks must therefore be resilient to failures, that is to say, it must be possible to reroute the traffic demands, or a subset of them, in case of failure. The design of resilient networks is a crucial problem in telecommunications and has gained much attention from the optimization community. Resiliency may be guaranteed applying different strategies, which may protect demands or part of the network. We focus on the end-to-end shared protection strategy. Two paths are defined for each demand: a primary and a backup one. The capacity devoted to backup paths may be shared among different demands, provided that their primary paths are fault disjoint. We consider two optimisation problems related to shared protection: the network loading problem with shared risk groups and shared protection and the energy-aware survivable network design with shared protection. In the first one beside single link/node fault, multiple failures are considered which are represented by the so-called shared risk group. In the second, the network energy consumption is minimized by switching off links when they are not used. In both problems, shared protection related constraints prove to be computationally challenging. We propose formulations and approaches to tackle the two problems. The idea of relaxing shared protection related constraints and dinamically re-add them is promising.
Groupe d’études et de recherche en analyse des décisions