Abstract

The telecommunications industry is growing very fast and frequently faces technological developments. Due to the competition between service providers and high expected reliability from their customers, they should be able first, to migrate their networks to the novel advancements in order to be able to meet their customers’ latest requirements and second, to optimally use the resources in order to maximize their profitability.
Many researchers have studied different scenarios for Network Migration Problem (NMP). In these studies, a comparison between the legacy and new technologies is investigated in terms of time frames, reduction in expenditures, revenue increases, etc. There have been no prior studies considering the operational costs of NMP e.g., technicians, engineers and travels. The first contribution of the thesis is to propose a two-phase algorithm based on the solution of column generation models that builds a migration plan with minimum overall migration time or cost.
The second contribution is an improved decomposition model for NMP by removing the symmetry between the network connections. We apply a branch-and-price algorithm in order to obtain an epsolin-optimal ILP solution.
The third contribution of the thesis is to propose a new methodology for Wavelength Defragmentation Problem to recover the capacity of WDM networks in dynamic environments and optimize resource usages. Since rerouting the lightpaths in an arbitrary order may result in a huge number of disruptions, an algorithm based on a nested column generation technique is proposed. The solution is an optimized configuration in terms of resource usage (number of links) that is reachable by no disruptions from the current provisioning.
All the algorithms presented in this thesis are based on Column Generation method, a decomposition framework to tackle large-scale optimization problems.