Abstract

Fiber Bragg Grating (FBG) sensors detect strain over long distances with very low loss (around 0.2 dB/km) and are immune to electromagnetic interference. We used these properties to build an intruder detection system that monitors a real fence in real time using FBG sensors, signal processing, and a photonics setup. The work proceeded in three stages. First, we collected intrusion datasets from two experiments: an indoor lab setup, and an outdoor setup on a real fence installed around a garden at the College of Engineering. These experiments characterized the relationship between the strain on the fence and the wavelength reflected from the FBG sensor under different environmental conditions. Second, we processed the captured signals using the short-term-average/long-term-average (STA/LTA) algorithm together with machine learning classifiers. STA/LTA on its own detected intruder presence at both high and low optical signal-to-noise ratios (OSNR), but could not classify the intrusion type. Machine learning classified the reflected waveform well, and combining STA/LTA with ML improved classification accuracy by 6% at low OSNR. Third, we built a LabVIEW graphical interface that flags detection whenever the reflected waveform changes. The system distinguished intruders from wind disturbance on the fence, the main source of false alarms in conventional perimeter sensors. The FBG approach also avoids the size, weight, and EMI limits of electronic vibration sensors, so it scales to perimeters several kilometers long.

Biography

Mohammed Falah Alqahtani is a PhD candidate whose research covers photonic sensing, machine learning, and robotics. He earned his Master’s in Robotics and AI from King Fahd University of Petroleum and Minerals in 2024 (GPA 3.57/4.00, Second Honor) and his B.Sc. in Electrical Engineering from King Saud University in 2022 (GPA 4.35/5.00, Second Honor). He worked as a Senior Network Engineer at SABIC, with earlier roles at Saudi Aramco and STC. His research interests include fiber-optic sensing, deep learning, UAV-based detection, and applied AI for industrial security.