Study of a long-range perimeter intrusion detection system using deep-learning-based optical fiber distributed acoustic sensor

In this study, we propose a deep learning algorithm that analyzes signal data collected from DAS devices to prove the concept of PIDS capable of intrusion detection and event classification. The collected signals are processed in specific time units (timeframes), and an application is developed that provides the analyzed data in the form of a web interface using deep learning based on CNN (Convolutional Neural Network). The experiment collects data by attaching optical fibers to the fence and burying them in the ground simultaneously and then introducing various events, and then using them to perform learning and inference to verify the validity of the system.