Enhancing Vibration Detection in ϕ-OTDR Through Image Coding and Deep Learning-Driven Feature Recognition

The phase-sensitive optical time domain reflectometer (<inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Phi $ </tex-math></inline-formula>-OTDR), a distributed fiber optic sensing technology, excels in precise vibration detection, making it optimal for security monitoring. Traditional <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Phi $ </tex-math></inline-formula>-OTDR for vibration detection typically involves laborious and inefficient analysis based on manual extraction of vibrational features from 1-D signal. This study introduces an innovative technique for recognizing vibration events based on 2-D image coding and a deep learning neural network (NAM-HorNet) to simplify and enhance the vibration recognition process. Converting 1-D vibration signals into 2-D images and using HorNet for feature recognition, our approach eliminates the necessity of manual feature extraction. Testing our method in discerning six distinct vibration events, including common noises and intrusion activities, such as stone knocking, scratching actions, and climbing attempts, we show that our approach offers an impressive vibration detection accuracy greater than 94.25% when combined with NAM-HorNet. Our method significantly outperforms conventional vibration detection techniques by enhancing recognition accuracy and minimizing false positives. Furthermore, the proposed method shows great promise not only in augmenting <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\Phi $ </tex-math></inline-formula>-OTDR-based vibration detection but also for a broad spectrum of sensor-based recognition applications.