AI-Powered Strategies for Addressing Fiber Nonlinearity in High-Capacity Elastic Optical Networks

Abstract

To fulfill the rising need for high-capacity and flexible communication systems, Elastic Optical Networks (EONs) have emerged as a possible alternative. However, the growing transmission rates and intricate modulation formats in EONs present substantial difficulties, such as fiber nonlinearity, which may deteriorate signal quality and restrict the network's performance. The fundamental components of EONs are fiber Nonlinear Interference (NLI) modeling and monitoring. Traditionally, they were created and studied independently. Furthermore, for heterogeneous dynamic optical networks, the previously suggested approaches' accuracy must still be increased. In this study, we demonstrate how Artificial Intelligence (AI) is used in NLI monitoring and modeling. We specifically propose to measure the drawbacks of the most current fiber nonlinearity estimates using AI approaches. The Gaussian Noise (GN) framework is used as an instance and Binary Differential-Support Vector Machine (BD-SVM) is used to demonstrate an important enhancement. In addition, we suggest combining modeling and monitoring strategies with AI for a more accurate prediction of NLI variation. Extensive simulations with 2411 connections are done to compare and assess the efficacy of various systems. The results of these simulations demonstrate that the AI-aided modeling and monitoring combo works better than other possible solutions.