Buried Fiber-Optic Geolocalization with Distributed Acoustic Sensing

TL;DR

This paper introduces a scalable method for geolocating buried fiber-optic cables using DAS and traffic-induced seismic signals, achieving sub-meter accuracy with neural-network optimization.

Contribution

The paper presents a novel framework combining DAS measurements, vehicle trajectory data, and neural networks for accurate underground fiber geolocation.

Findings

Achieves sub-meter localization accuracy in field experiments.

Robust convergence under realistic noise and trajectory uncertainties.

Strong agreement with manual calibration methods.

Abstract

We present a scalable method for geolocalizing buried fiber-optic cables using Distributed Acoustic Sensing (DAS) and traffic-induced quasi-static seismic signals. Assuming access to one end of the fiber, the method fuses DAS measurements with vehicle trajectories obtained from either video tracking or vehicle-mounted GPS. The fiber geometry is estimated by minimizing the mismatch between the measured and physics-based synthetic strain-rate maps. The framework combines a matched-filter initialization with neural-network-based trajectory optimization, enabling robust convergence under realistic noise and trajectory-uncertainty conditions. Simulation and field experiments demonstrate sub-meter localization accuracy, often on the order of tens of centimeters, and strong agreement with manual calibration by tap-testing. This approach provides a practical tool for mapping poorly documented underground fiber infrastructure and for supporting urban sensing applications.