Mesh, supports and along-fence structure of a metal fence
Metal-fence photograph (perimeter context). Maxx@night from Austin, Texas. USA (resized/WebP) · CC BY 2.0

Monitoring task and sensing route

Nearby fiber distances do not always represent nearby fence locations, particularly with return loops or different installation heights. Channel-to-site mapping should precede analysis.

Original conceptual layout: Fence and channel map and the optical-fiber route
Original OriginSense application concept showing observation relationships, not a customer installation.

Published research observation

Jeong and colleagues combined channel-graph structure and temporal modeling for anomaly detection using controlled fence-activity data. [1]

Research evidence and proposed validation
ItemDescription
Study settingControlled fence-activity experiments
Research observationSpatial channel relationships support anomaly analysis
Proposed comparisonDocument upper, lower and return fiber paths, attachments and slack sections.

Validation and integration priorities

Retain weather, traffic and normal-work backgrounds with independent activity logs.

Split tests by independent event and date; report false alarms, missed events and processing latency separately.

Reference

[1] Jeong S et al. Spatiotemporal Anomaly Detection in Distributed Acoustic Sensing Using a GraphDiffusion Model. Sensors (Basel, Switzerland) (2025). DOI: 10.3390/s25165157.