Entropy Measurement for Biometric Verification Systems.

Area of Science:

• Computer Science
• Information Security
• Biometrics

Background:

• Biometric verification systems permit multiple similar measurements per user to account for natural data variations.
• Accepting multiple measurements enhances genuine user acceptance but compromises system security by lowering imposter difficulty.
• Quantifying this security degradation is crucial for user assurance.

Purpose of the Study:

• To develop a novel entropy-measuring model for biometric systems that accommodate multiple similar measurements per user.
• To quantify the security level of biometric systems in the presence of intrauser variations.
• To provide a truthful security assurance to users of biometric verification systems.

Main Methods:

• Developed an entropy-measuring model based on the concept of guessing entropy.
• Adapted the model to handle biometric systems accepting multiple similar measurements.
• Quantified security by measuring adversarial guessing effort against two practical attack scenarios.

Main Results:

• The proposed model accurately quantifies biometric system security, even with multiple acceptable measurements.
• Analytic results showed excellent agreement with experimental simulations on synthetic and benchmark face datasets.
• Validated the correctness and feasibility of the entropy-measuring approach for biometric security.

Conclusions:

• The developed entropy model effectively measures security in biometric systems with multiple accepted measurements.
• This approach provides a reliable method for assessing and assuring the security of biometric verification.
• The findings support the practical application of entropy-based security measurement in biometrics.