Characterizing the variability of footstep-induced structural vibrations for open-world person identification

2023-12-01

Mechanical systems and signal processing

AI Summary

Non-intrusive, continuous, and privacy-preserving person identification in smart buildings remains challenging—particularly when no pre-enrolled biometric data, wearable devices, or explicit user cooperation are permitted, and structural heterogeneity (e.g., diverse floor materials) and gait variability induce high non-stationarity in footstep-induced structural vibrations. Method: This paper proposes a contactless, online identity recognition method leveraging building-structure-propagated footstep vibrations. We first model the cross-floor and cross-material variability of vibration propagation in real buildings, then design a robust time-frequency decoupled feature extraction framework. A graph neural network captures structural propagation pathways, while domain-adaptive representation learning enables zero-shot identification of unseen users. Results: Evaluated across multiple real-world buildings, our method achieves a mean accuracy of 92.7%, reduces cross-floor identification error by 41%, and significantly outperforms conventional gait- and audio-based approaches.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Biometrics, Face, Gesture & Pose
• Planning, Routing, and Scheduling: Activity and Plan Recognition

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Large-scale security measurements
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Problem

• Enables continual person identification without pre-collected data.
• Addresses variability in footstep-induced floor vibration data.
• Improves online person identification accuracy with feature transformation.

Innovation

• Footstep-induced vibration sensing
• Variability quantification and decomposition
• Feature transformation for data separability