Optical spectrum feature analysis and recognition for optical network security with machine learning.

ABSTRACT

Physical layer attacks threaten services transmitted through optical networks. To detect attacks, we present an investigation of optical spectrum feature analysis (OSFA) and recognition. By analyzing the spectral features of optical signals, recognition and detection of unauthorized signals can be realized. In this paper, (1) we theoretically analyzed factors influencing optical spectrum (OS) features and simulated these factors. OSs collected from the simulation are quantitatively analyzed, spectral features are extracted by principal component analysis, and the theoretical derivation is validated. (2) We proposed support vector machine (SVM) and one-dimensional convolutional neural network (1D-CNN) machine-learning OSFA methods. (3) Experimentally collected OSs from commercial small form-factor pluggable modules are used to verify the performance of the SVM and 1D-CNN methods, which achieved 98.54% and 100% recognition accuracies, respectively, demonstrating that the methods are promising solutions for optical network security.