| Motivated by diurnal variation of traffic patterns in the large backbone networks, the first part of this dissertation addresses the fundamental question of under which conditions the deployment of reconfigurable optical layer can be justified. A comprehensive empirical study is conducted to answer the question. The benchmarking is carried by comparing a traffic engineering method capable of rerouting the LSP's with fixed optical layer to mitigate the adverse effect of traffic fluctuation.;Depending on the traffic fluctuation scenarios, and the network resources, this study provides an answer for IP/MPLS over WDM networks. In this dissertation, for small networks the results are presented using a flow based ILP model. However, for larger networks the problem is tackled by taking advantage of a novel heuristic (BinCluster).;Once the reconfigurable optical layer is deployed, the next question is how to migrate the virtual topology of the network. The second part of this dissertation addresses the virtual topology migration problem under traffic fluctuation scenarios. Algorithmic solutions are presented to conduct a smooth migration of the virtual topology from the perspective of upper layer traffic. Two optimization problems are considered, one with a single objective function and the second a multi-objective optimization with primary and secondary objectives.;Finally, a numerical method is presented to obtain a lower bound on the lightpath capacity of the network which guarantees the non-disruptive reconfiguration of the network. |
