Apply machine learning and optimization in optical networks

Abstract

In the EONs, that is, Elastic Optical Networks, a sophisticated management and planning approach is employed to address the physical impairments and resource limitations inherent in high-capacity data transmission over optical networks. In this project, we conduct a comprehensive assessment of simulation tools and the theoretical framework surrounding EONs through an extensive literature review and development of an elastic optical network. The literature review focuses on understanding the NLI-aware algorithm and the NLI-unaware algorithm, and replaces the fixed 50 percent OSNR-margin threshold with a better reasonable parameter to classify risky and safe lightpath movements. This literature on RMSA, Qot-aware resource allocation and physical layer modelling, including ASE, NLI, ISRS, and spectral-grid simulation, provides the foundation for this study. By evaluating for different 30 percent, 50 percen,t and 70 percent, demonstrating how Qot aware defragmentation performance varies with risk tolerance and offers insights for optimizing elastic optical network operations.