| In static and semi-dynamic routing, the traffic requirements in the Wavelength Division Multiplexing (WDM) network are given as a set of point-to-point connections of different granularities between pairs of access stations. Finding the exact optima for most of the network optimization problems is generally impossible with today's computation facility even for the medium size networks. Therefore, finding a sub-optimal solution within a reasonable computation time is the only choice, while knowing the proximity of the sub-optimal solution to the exact-optimum will be an additional advantage.; We provide in this thesis an efficient optimization framework to static and semi-dynamic connection routing problems, which contains new integer linear programming (ILP) formulations and a solution methodology employing Lagrangean Relaxation and Subgradient Method (LRSM). We are able to provide a good bound at the same time as obtaining a sub-optimal solution, with reasonable time complexity. We prove the efficiency and the easy adaptation of our framework by studying the optimization problems for various WDM networks with different models, different constraints and different objective functions in generalized network topologies. We use it to investigate the static Routing and Wavelength Assignment (RWA) problem, semi-dynamic RWA problem, as well as the traffic grooming problem. We also study a wide range of influences, such as the number of resources and the network design parameters, on the network behaviours under different network structures and assumptions. On the other hand, we provide new formulations for the network optimization considering various aspects of the network operation and design concerns. Compared with many other methodologies proposed in some previous research, our framework demonstrates a high efficiency in terms of both computation results and time complexity. |
