Towards Optimal and Scalable Solution for Routing and Spectrum Allocation

Abstract

The advent of the flexible spectrum grid (flexgrid) technology for the slicing of the optical spectrum has brought new opportunities in terms of efficient utilization of the spectrum, but also new challenges in terms of scalable algorithms for efficiently provisioning the user requests in a flex optical network, i.e., solving the Routing and Spectrum Allocation (RSA) problem. Despite the research that has been done to solve the RSA problem, there is still a considerable gap between the literature and industry sizes of the RSA data sets.In this study, we go one step further towards solving the RSA problem for large traffic and network instances. In doing so, we propose a new column generation model with lightpaths (i.e., paths provisioned with a given number of spectrum slots in the context of RSA), in which the lightpath generators (pricing problems) can be efficiently solved using a shortest path algorithm. The resulting solution process is much faster than the previously proposed algorithms (up to 2 orders of magnitude faster), with a significantly increased proven solution accuracy.