Blended Biometric Authentication System Using Machine Learning
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric security systems face challenges in constructing effective fuzzy extractors for various biometric data, particularly due to high entropy biometric sensors being intrusive and requiring conscious re-authentication, which reduces overall security and convenience.
Innovation Solution
The integration of multiple biometric sensors, combining high-entropy active sensors with low-entropy passive sensors, using machine-learning to process and convert biometric data into a common form, allowing for continuous authentication with reduced intrusiveness and improved security by leveraging variations in entropy, intrusiveness, and availability.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If high-entropy biometric sensors are used for authentication, then security is improved, but intrusiveness increases and continuous authentication becomes difficult
Solution Approach 1:
The patent segments the authentication system into two distinct components: high-entropy biometric sensors (fingerprint, face, iris) for initial authentication to establish security, and low-entropy passive sensors (accelerometer, gyroscope, microphone) for continuous authentication to maintain security state. This segmentation allows each sensor type to operate in its optimal mode without compromise.
Solution Approach 2:
The patent merges multiple sensor types with different entropy characteristics into a unified authentication system. The passive sensors continuously monitor device usage patterns and combine their low-entropy data with periodic high-entropy biometric verification, creating a composite authentication mechanism that achieves both continuous monitoring and high security.
2Reliability
If active biometric sensors are used for authentication, then entropy and security are improved, but re-authentication requires conscious user action which reduces convenience
Solution Approach 1:
The system performs preliminary authentication using high-entropy biometric sensors to establish the user's identity and security state. Once authenticated, the system enters a sustained unlocked state where passive sensors continuously verify usage patterns, eliminating the need for repeated conscious biometric authentication while maintaining security.
Solution Approach 2:
The passive sensors automatically and continuously monitor device usage patterns without requiring conscious user participation. The system self-verify whether the authenticated user is still in possession and control of the device by analyzing accelerometer, gyroscope, and microphone data, providing continuous authentication without user burden.
3Ease of operation
If passive biometric sensors are used for continuous authentication, then intrusiveness is reduced, but entropy and security are compromised
Solution Approach 1:
The high-entropy biometric authentication acts as an intermediary that periodically reinforces the security state established by passive sensors. The passive sensors provide continuous low-entropy verification of device possession, while periodic high-entropy biometric checks prevent accumulation of uncertainty and maintain overall system entropy at secure levels.
Data Source
AI summary
A biometric processing system for authentication combines multiple biometric signals using machine learning to map the different signals into a common argument space that may be processed by a similar fuzzy extractor. The different biometric signals may be given weight values related to their entropy allowing them to be blended to increase security and availability while minimizing intrusiveness.