A Continuous Identity Authentication Scheme Based on Physiological and Behavioral Characteristics.

Area of Science:

• Biometric Authentication
• Wearable Technology
• Human-Computer Interaction

Background:

• Current wearable authentication methods are often one-time, lacking continuous user certification.
• Wearable devices offer significant advantages and are increasingly used for identity verification.

Purpose of the Study:

• To develop and evaluate a continuous, two-step authentication method using a novel fingertip sensor device.
• To address the limitations of single-time authentication in wearable systems.

Main Methods:

• Simultaneously captured motion (acceleration, angular velocity) and physiological (photoplethysmography) data from a fingertip sensor.
• Feature extraction from multisensor data, followed by human activity recognition as the first authentication step.
• Application of a one-class machine learning algorithm for identity authentication as the second step, operating automatically at set intervals.

Main Results:

• The proposed two-step authentication method achieved an average accuracy rate of 98.5%.
• An F1-score of 86.67% was obtained, demonstrating robust performance across various operational scenarios.
• The system demonstrated effective continuous authentication for legitimate users.

Conclusions:

• The developed fingertip sensor-based authentication scheme offers a feasible and practical solution for continuous identity verification.
• The integration of human activity recognition and machine learning enhances the security and usability of wearable authentication systems.
• The study validates the effectiveness of multisensor data fusion for reliable biometric authentication.