Planning and dimensioning of multilayer optical transport networks

• Main Author: Morais, Rui Manuel Dias
• Publication Date: 2015
• Language: eng
• Source: Repositórios Científicos de Acesso Aberto de Portugal (RCAAP)
• Download full: http://hdl.handle.net/10773/14094
• topic: Engenharia electrotécnica; Redes ópticas; Arquitectura de redes; Algoritmos genéticos

This thesis presents a study on the planning of multilayer optical transport networks considering the various hardware implementation constraints. Optimization methods for the various stages of the network planning are proposed, namely physical topologies design, and green eld and multi-period nodes dimensioning. The developed methods rely on integer linear programming models (ILP), heuristic algorithms and statistical methods. Initially, the enabling technologies and the di erent node architectures usually employed in multilayer optical transport networks are presented. Additionally, the main grooming schemes and transport modes are also discussed. Assuming that the node localization is known, the rst stage of the overall network planning process is the deployment of the network links. Thus, a genetic algorithm for the design of survivable physical topologies with minimum cost is proposed, as well as a dimensioning model for links. Within the heuristic algorithm various genetic operators are evaluated, compared and benchmarked using an ILP model. After, the nodes are planned. Therefore, optimization models for the nodes dimensioning to use in green eld scenarios are proposed. The models are based on ILPs and calculate the number and type of modules required to implement xed and exible architectures, for both the electrical and the optical layer, taking into consideration the hardware implementation constraints. Using the developed models, comparative techno-economic analysis are performed focusing on the CapEx, power consumption, and footprint requirements. As a result, and based on the outcome of a large set of simulations, an optimization method based on simple rules for node architecture selection is proposed. Finally, optimization methods based on ILPs to use in multi-period planning are presented. The models consider the various electrical layer architectures, and enable a planning considering hitless re-grooming. Then, a comparative techno-economic analysis is performed and the savings attained by exploiting hitless re-grooming evaluated, highlighting the conditions where such savings are more signi cant.