Bidirectional Polarization-Based Integrated Sensing and Communication for In-Service Metropolitan Fibres: Train-Induced Vibration Localization and Train Flow Monitoring

Le résumé fourni par la source

In-service metropolitan fibre sensing can extend the value of deployed communication links, but practical sensing schemes must remain compatible with live traffic and modest hardware budgets. Here, we present a low-overhead state-ofpolarization (SOP)-based integrated sensing and communication (ISAC) framework that combines simplified SOP monitoring with synchronized bidirectional transmission on live metropolitan fibres. We also introduce TAG-Former, a target-aware anchorguided Transformer for localizing vibration events from synchronized bidirectional polarization dynamics. In controlled experiments over five reference positions along a 73-km fibre span, TAG-Former reduces the localization mean absolute error from 4.15 km to 1.71 km relative to the Cross-CorrelationAssisted Convolutional Neural Network (CC-CNN) baseline. We then deploy and validate the complete framework in a twoweek field trial on live metropolitan fibres, demonstrating stable concurrent communication and sensing while achieving a testset mean distance error of 0.47 km on the mixed-event fieldlink benchmark and approximately 0.56 km for the trainpassage events. Together, these results establish a communicationcompatible bidirectional SOP-based ISAC architecture, validate TAG-Former as an effective localization engine, and, to the best of our knowledge, provide the first field demonstration of traininduced vibration localization under the tested live metropolitanfibre conditions. Long-term observations also reveal environmentand traffic-sensitive polarization behaviour, highlighting broader sensing potential for polarization-based ISAC under the tested conditions.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.