A column generation approach for large-scale RSA-based network planning

In flexgrid-based optical networks, the problem of finding optimal route and spectrum allocation (RSA) for the demands in a traffic matrix becomes harder to solve than similar problems in fixed DWDM networks, mainly due to the finer spectrum granularity and the spectrum contiguity and continuity constraints. Additionally, the increasing traffic volumes and the size of real networks lead to network planning problem instances consisting of hundreds of thousands, or even millions, of (binary) variables. In this work, we present a column generation decomposition method to obtain feasible solutions for RSA-based network planning problems. Numerical results show the utility of this method for solving intractable instances.