A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering in Telecommunications, School of Engineering and Technology

Thesis (Ph.D.) - Asian Institute of Technology, 2007

On top of a power grid used to distribute electricity, a dense wavelength division multiplexing (DWDM) based optical network can be constructed to be a scalable backbone network that can reach almost all remote areas of the country. This dissertation proposes a complete optimization procedure for optical network design based on an existing power grid. We design a network as a sub graph of the power grid and divide the network topology into two layers: backbone and access networks. The design procedure includes physical topology design, routing and wavelength assignment (RWA) and optical amplifier placement. We formulate the problem of topology design into two steps: selecting the concentrator nodes with their node members and finding the connections among concentrators subject to the two-connectivity constraint on the resultant backbone topology. These two problems are formulated as integer linear programming (ILP) problems. Since the ILP problem for the concentrator connection problem grows intractable for large networks, a heuristic algorithin is developed. For RWA and optical amplifier placement problem, we solve these two problems together since they are closely related. As the ILP for solving these two problems becomes intractable with increasing network size, we propose a simulated annealing approach. We choose a neighborhood structure based on path-switching operations using k shortest paths for each source and destination pair. The optimal number of optical ilmplifiers is solved based on local search among these neighbors. We solve and present numerical results for 3D-node and IDO-node randomly generated power grid topologies and for actual EGAT's power grid topology.