Optical Communication-Sensing (O-CommSense): Fiber Sensing Using Ultra-Short Pulse Locking and Machine Learning for Physical Event Detection

This paper introduces a transformative forward-sensing paradigm for existing telecommunication optical fiber networks that leverages ultra-short pulse laser locking to extract high-precision arrival time of a pulse train, coherent detection, and multi-dimensional signal analysis to detect and characterize physical disturbances, and propose a framework for localization of physical disturbances of interest. Unlike conventional backscatter-based distributed acoustic sensing (DAS) systems, which suffer from signal-to-noise ratio limitations and spatial resolution trade-offs, our approach employs a primary-secondary mode-locked laser (MLL) configuration with femtosecond-level pulse synchronization to capture phase and polarization perturbations along the entire fiber path. The core innovation lies in formulating the error signal from the phase-locked loop as a multi-dimensional data cube comprising fast-time (intra-pulse dynamics), slow-time (inter-pulse evolution), and polarization dimensions. We hypothesize that physical events such as acoustic vibrations, mechanical intrusions, or environmental changes imprint unique space-time signatures within this cube, which can be extracted via machine learning techniques. Preliminary experiments using controlled acoustic excitations demonstrate clear separability between disturbed and undisturbed states in principal component analysis (PCA) and kernel PCA reduced spaces. The proposed methodology bridges high-precision photonics and data-driven sensing at no extra hardware costs, with promising applications in critical infrastructure security, environmental monitoring, and smart sensing networks.