Photonic Reservoir Computing to Detect Anomalies and Intrusions in Optical Networks

Abstract

Optical networks are a critical infrastructure subjected to attacks at its physical layer. Different methods have been presented to prevent, detect and solve these attacks. The current systems in charge of detecting intrusions in optical networks rely on the metrics obtained from OPM equipment. This equipment has a high cost, making the massive deployment through the optical network unfeasible. Photonic reservoir computing is a technology that has been able to solve tasks as header recognition, channel equalization, etc. with a good performance and a cost that allows it to be deployed over all the network. In this thesis it is designed and assessed by simulation a photonic reservoir computing system able to detect and classify attacks in optical networks.