Machine Learning-Based Polarization Signature Analysis for Detection and Categorization of Eavesdropping and Harmful Events
Leyla Sadighi, Stefán Karlsson, Carlos Natalino, Marija Furdek
Abstract
We propose a methodology that uses polarization state changes and machine learning to detect and classify eavesdropping, harmful, and non-harmful events in the optical fiber network. Our solution achieves 92.3% accuracy over 13 experimental scenarios.
Topics & Concepts
EavesdroppingCategorizationComputer scienceMachine learningSignature (topology)Artificial intelligencePolarization (electrochemistry)Support vector machinePattern recognition (psychology)Computer securityMathematicsPhysical chemistryChemistryGeometryOptical Network TechnologiesAdvanced Photonic Communication SystemsAdvanced Fiber Laser Technologies