Thèse soutenue

Auteur / Autrice : Gurvan Huiban
Direction : Afonso FerreiraGeraldo Robson Mateus
Type : Thèse de doctorat
Discipline(s) : Informatique
Date : Soutenance en 2006
Etablissement(s) : Nice
Ecole(s) doctorale(s) : École doctorale Sciences et technologies de l'information et de la communication (Sophia Antipolis, Alpes-Maritimes)


FR  |  

An optical telecommunication network is configured to transmit a given traffic, to meet a given objective. However, the demand changes with the time and the development of infrastructures. The reconfiguration problem lies in this context : being able to change the configuration of the network to adjust it to the new demands. In general, to reconfigure a network, it is required to interrupt totally or partially the network traffic. Considering the amount to data flowing on it, it may not be possible to regularly stop the network, even if it is only for a few minutes. Many parameters have to be taken into account to find out what will be a good solution. We propose a mathematical model allowing to represent the reconfiguration problem. However, solve the proposed model exactly may require a high computation time. Then, we also propose a greedy and a simulated annealing heuristics. Depending on the metric optimized, the solutions have different characteristics. The greedy algorithm provides solutions competing with the optimal ones. In a second part, we focus on the multiobjective aspect of the reconfiguration problem. It considers at the same time the different metrics and searches for a seat of solutions representing different interesting trade-offs instead of a unique solution. We proposed an algorithm based on our mathematical formulation. We also adapt an evolutionary algorithm. The proposed methods succeed in finding different interesting trade-offs. Giving a little flexibility with respect to a metric generally allows to significantly improve the solutions with respect to the other metrics.