Seminar: SNR Awareness in Elastic Optimal Networks using Column Generation

Abstract

    Telecommunication companies use software with heuristic algorithms to plan their routing. However, with network demands increasing at such a rapid pace, the effectiveness of these heuristics become a critical issue. Therefore, our research focuses on developing a large-scale optimization model for solving the Routing and Spectrum Assignment (RSA) problem. The goal of this model is to serve as a benchmark that can be used to accurately measure the performance of the aforementioned heuristics.

    In the first part of our research, we employ the Column Generation (CG) algorithm, a powerful decomposition technique, to create an exact mathematical model, namely the link-based decomposition model, capable of solving large datasets with near-optimal solution.

    The second part of our study involves the integration of the Optical Signal-to-Noise Ratio (OSNR) using the Gaussian Noise (GN) model into the link-based decomposition model created in the first part. To handle the complexity of non-linear elements, we utilize a combination of Column Generation (CG) and Tabu Search (TS) techniques.

    In the third part of our research, we address the shortcomings of the second part, which arose from using TS. We have developed an exact method to solve the mathematical model developed in the second study.