Définition d'un plan de contrôle pour les réseaux optiques sans filtre

Abstract

Next generation optical networks have to cope with the wide popularity of Broadband applications, including video on demand and (3D) high-definition video streaming. This success of the Internet involves a strain in the physical layer because optical architectures need to be simple and reconfigurable at the same time so as to deal with an increase of the different traffic types and of their unpredictability. Usually, all optical networks are based on active switching devices such as optical cross connects (OXC) or wavelength selective switches (WSS) in the dynamic and automatic lightpath establishment process. Agility and reconfigurability come however at the expense of a relatively high cost, and require an efficient control plane which supports the last concepts in terms of automated self-discovery, automated connection management, efficient operation and self-healing abilities. In this thesis, we recall the principles of filterless optical networks as a cost-effective solution for metropolitan and wide area networks. A filterless network exploits recent breakthroughs in tunable lasers, frequency selective coherent receivers, advanced modulation formats, and electronic dispersion compensation, in order to move agility only at the emitters and their corresponding receivers, by the use of passive optical splitters and combiners for link interconnections at intermediate nodes and local-add drop in the place of active switching devices. We then present the peculiar features of filterless networks, which make them passive broadcast and select networks, such as unfiltered channels, backward and forward ports, closed loops, wavelength reuse, and show how they can enrich the manner one can operate such a network as well as how they can harm established connections if not properly handled by a control plane. At last, we present a filterless optical control plane simulator and show how the concept of unfiltered channel is important to be mitigated so as to keep network performance at its best; as the unfiltered channels still propagate after reaching their destination, they also participate in wavelength contention and increase the blocking probability for new paths to be established. We also introduce a new metric to validate filterless solutions obtained from previous works based on the ability of these solutions to deal with unfiltered channels, given a traffic matrix, and show how, through cost and power dissipation analysis, it is possible to select the most effective solution.

Date	24 Oct 2012
Original language	French
Awarding Institution	École de technologie supérieure
Supervisor	Christine Tremblay (Supervisor) & David V. Plant (Co-supervisor)