Topology discovery of Synchronous Optical NETworks

Telecommunication networks are comprised of interconnected network elements which provide communication services to end users. The port-to-port connectivity of these network elements is referred to as the network topology. Due to the ever growing demand, these networks undergo frequent changes in their topology as new fiber optic links, nodes and circuits are regularly provisioned and upgraded. Topology discovery of legacy optical networks, though currently a challenge for network operators, has not been studied much in the literature. It is a difficult problem, that resembles a puzzle, as very often, some network information is missing or erroneous. We propose a new algorithm for the physical (link) layer network topology. We define the concept of port signature with the combination of four different network parameters, and then a weighted matching model in order to identify the pairs of connected ports. Confidence coefficients are assigned to the discovered port pairs to express their connectivity likeliness. Numerical results are presented on a data set coming from a customer of Ciena Corporation. We discuss the validation of the results, and the performance of the proposed tools for topology discovery.