Multi-layer optical network design based on clustering method

In our optical network design method, design of the logical topology is preceded by routing and wavelength assignment. The logical topology is a set of lightpaths formed to serve traffic demands. It must be flexible and robust for future changes or reconfiguration while also minimizing the cost and optimizing some objective functions. These features can be found in a multi-layer architecture. In an optical network, this architecture includes service layer, electrical layer and optical layer, where the optical layer serves as the backbone of the lower layers. Moreover, the equipment at optical nodes, such as optical cross connect (OXC), cannot serve all demands in the traffic demand matrix (e.g., the set of lightpaths is limited by the number of transmitters and receivers). Some nodes and lightpaths are preselected and form the optical network in the optical layer of the multi-layer architecture. We show an approach to multi-layer optical network design by analyzing the traffic demand matrix and physical topology in order to classify nodes into either optical layer or electrical layer using a clustering method and then design the logical topology to reach a certain limitation and objective function. Our multi-layer approach reduces the size of the optical network, resulting in lower cost of equipment and fewer constraints in the design. Besides, the clustering method provides a multivariate analysis to cluster the data by combination of network node characteristics, and can be extended to the financial cost of equipment, location data or even policy data.