Intelligent Fiber Optic Perimeter Intrusion Detection System Using Mach-Zehnder Interferometer and Machine Learning

This paper presents development of an intelligent Perimeter Intrusion Detection and Classification System (PIDS) based on the integration of a Mach Zehnder interferometer and machine learning techniques. The main objective is to enhance the perimeter security against the intrusion activities, including shaking, climbing, and knocking, by providing accurate intrusion detection and classification. Experimental design included positioning a pressure sensitive fiber optic interferometer on a fence in order to capture the disturbances generated by different intrusions. Pre-recorded signals were processed with signal processing, background subtraction and de-noising. Statistical features including variance, root mean square (RMS) and standard deviation were extracted. A Decision Tree classifier is trained using these statistical features and an accuracy of 83% is obtained to classify the intrusion types. The findings support the effectiveness of integrating fiber optic sensing capability and machine learning to provide accurate intrusion detection and classification, with profound applications for enhancing security in strategic infrastructure.