AirSignatureDB: Exploring In-Air Signature Biometrics in the Wild and its Privacy Concerns

2025-08-11

AI Summary

This study exposes a long-overlooked privacy threat in air-signature biometrics: inertial measurement unit (IMU) data enables high-fidelity reconstruction of three-dimensional signing trajectories, undermining the prevailing assumption that air gestures leave no trace. To address this, we introduce the first large-scale, publicly available air-signature dataset—comprising genuine and spoofed samples from 108 participants, collected across 83 smartphone models in realistic settings. We propose the first end-to-end 3D trajectory reconstruction method using IMU data alone, and systematically evaluate diverse traditional and deep learning-based verification models, analyzing the impact of sensor modalities and enrollment strategies. Experimental results demonstrate trajectory reconstruction errors as low as 2.3 cm, confirming the strong behavioral identifiability embedded in inertial signals. Our work establishes a new paradigm and empirical foundation for privacy risk assessment and mitigation in behavioral authentication systems.

Technology Category

• Computer Vision: Biometrics, Face, Gesture & Pose
• Machine Learning: Privacy
• Humans and AI: Interaction Techniques and Devices

Application Category

• Security and Privacy: Large-scale security measurements
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems
• Web Mining and Content Analysis: Web data provenance, reliability, and authenticity

Abstract

Behavioral biometrics based on smartphone motion sensors are growing in popularity for authentication purposes. In this study, AirSignatureDB is presented: a new publicly accessible dataset of in-air signatures collected from 108 participants under real-world conditions, using 83 different smartphone models across four sessions. This dataset includes genuine samples and skilled forgeries, enabling a comprehensive evaluation of system robustness against realistic attack scenarios. Traditional and deep learning-based methods for in-air signature verification are benchmarked, while analyzing the influence of sensor modality and enrollment strategies. Beyond verification, a first approach to reconstructing the three-dimensional trajectory of in-air signatures from inertial sensor data alone is introduced. Using on-line handwritten signatures as a reference, we demonstrate that the recovery of accurate trajectories is feasible, challenging the long-held assumption that in-air gestures are inherently traceless. Although this approach enables forensic traceability, it also raises critical questions about the privacy boundaries of behavioral biometrics. Our findings underscore the need for a reevaluation of the privacy assumptions surrounding inertial sensor data, as they can reveal user-specific information that had not previously been considered in the design of in-air signature systems.

Problem

• Evaluating robustness of in-air signature biometrics against attacks
• Reconstructing 3D trajectories from inertial sensor data
• Assessing privacy risks in behavioral biometrics data

Innovation

• Public dataset of in-air signatures from 108 participants
• Benchmarking traditional and deep learning verification methods
• Reconstructing 3D trajectories from inertial sensor data