Êtes-vous un étudiant de l'EPFL à la recherche d'un projet de semestre?
Travaillez avec nous sur des projets en science des données et en visualisation, et déployez votre projet sous forme d'application sur GraphSearch.
Modelling, dimensioning and performance analysis of optical backbone networks is the main concern of this thesis. Modelling consists in representing a network under a abstract form, simpler and easier to analyse, the model. Dimensioning is achieved when network characteristics required to fulfil given requirements are fixed a priori. Performance analysis consists in identifying the criteria characterising the network performance and in verifying whether dimensioned networks do fulfil the requirements. It also aims at estimating the deployment cost of a dimensioned network. Dimensioning a network while deploying it is unrealistic. Optical networking equipment are expensive. It is thus absurd to buy large amounts of equipment and use a part of it only. Moreover, backbone optical network spans over very large distances and is thus costly to set up. Dimensioning must thus be done based on a model, and modelling and dimensioning are therefore achieved in a integrated way. Network performances and optimal network are generally impossible to predict by means of mathematical formulations and transformations. Numerical models are thus involved to reproduce network properties. Iterative dimensioning procedures are similarly used to draw conclusions about optimal network dimensions. As both numerical models and iterative procedures involve many operations, these are most often automated. In this way, they are implemented in software programs and executed on computers. Several numerical models are presented in this thesis. Associated with corresponding dimensioning procedures, they aim at analysing the properties of various optical backbone network architectures, at devicing rationalised network layouts, and eventually at exploiting these architectures in the optimal way. Besides, the development, verification and usage of these models and procedures are time consuming activities. To speed-up them, functionalities and operations can be shared among models and procedures. They must however not only be implemented, but also organised in a structured manner within a framework. This thesis thus also presents such a software framework.