الفهرس | Only 14 pages are availabe for public view |
Abstract The network traffic characteristics has changed over the last decade; higher bit-rates are required up to 1 Tb/s which consumes a lot of the available optical spectrum using fixed frequency slot techniques. Regular optical networks are not utilizing the available spectrum in an optimal way. The conversion of conventional optical network to elastic optical network became inevitable to use the available bandwidth efficiently. However, this conversion is facing many challenges. This thesis investigates different techniques and algorithms to solve some of those challenges, which are related to the routing and spectrum assignment. The main challenge related to the routing and spectrum assignment is how to find a route and allocate enough bandwidth on that route to satisfy each traffic request in an optimal way. The allocated bandwidth must not conflict with any other traffic requests’ bandwidth on any of the optical links. The routing and spectrum assignment problem is NP-Hard. Solving the problem for exact solution is not feasible for large network. Most of the work done in the last decade utilized heuristics to solve the problem. This thesis introduces new heuristics algorithm inspired from task scheduling algorithms and also introduces a new efficient exhaustive search algorithm that can search the complete search space of the problem in a scalable manner. The proposed algorithm is scalable and has been applied to NSFNet and GEANT2 networks. The result shows that the proposed algorithm has succeeded to identify the best candidate routing tables which consequently can result in optimum spectrum assignment. |