•  
  •  
 

Document Type

Article

Keywords

Biometric Authentication, Touch Dynamics, XGBoost, Decision-Making, Smartphone Security

Abstract

Smartphone authentication systems must balance security and user convenience, which is a persistent challenge in the digital realm. Traditional biometrics, such as fingerprints and facial recognition, face vulnerabilities to spoofing and environmental conditions, limiting reliability. This study introduces a novel approach by integrating three-dimensional (3D) touch sensors with finger location data for authentication. The goal is to develop a system that improves accuracy while minimizing false positives and negatives, leveraging touch pressure and spatial interaction as unique biometric identifiers. Data from 20 participants, including pressure levels, spatial coordinates, and timestamps, were analysed using Random Forest (RF) and Extreme Gradient Boosting (XGBoost) models. The results showed that combining pressure sensitivity with spatial data significantly improved performance, achieving an F1- score of 0.83 and an accuracy of 83%. The system demonstrated balanced precision (0.84) and recall (0.83), effectively reducing false positives and negatives. Robustness was confirmed through cross-validation tests, which validated the consistency across datasets and real-time usability scenarios. This study establishes a foundation for secure, user-friendly smartphone authentication, highlighting the potential of 3D touch technology in addressing current biometric system limitations. This approach opens avenues for further research in mobile security, integrating multimodal biometric data with advanced machine learning techniques.

Download

Share

COinS