Polynomial-Based Cancellable Biometrics for Feature Data Secrecy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric authentication systems face challenges in maintaining the secrecy of feature data, as malicious server administrators can potentially decode feature data images, especially when multiple authentication attempts are made, leading to a risk of information leakage and security breaches.
Innovation Solution
The proposed method employs a polynomial-based cancellable biometrics system where feature polynomials are transformed using helper and mask polynomials, stored on an IC card, and used to generate authentication polynomials, ensuring that only a correlation polynomial is shared with the authentication server, maintaining the secrecy of the original feature data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If feature data is transmitted to authentication server for verification, then authentication function is achieved, but feature data secrecy is compromised
Solution Approach 1:
The feature data is segmented into multiple components: authentication polynomial (transmitted to server), helper polynomial (stored on IC card), and mask polynomial (stored on IC card). Only the authentication polynomial is transmitted, while the secret components remain on the client side, enabling authentication while preserving secrecy of the original feature data.
Solution Approach 2:
Polynomial transformation acts as an intermediary mechanism between the original feature data and the authentication server. The feature data is transformed into authentication polynomials through mathematical operations involving helper and mask polynomials, allowing verification without direct exposure of the original biometric information.
2Loss of information
If feature data is transformed using secret parameters for privacy protection, then information secrecy is improved, but system complexity increases
Solution Approach 1:
The system changes the parameter representation from raw feature data to polynomial form. By transforming biometric feature data into polynomial coefficients and applying mathematical transformations with secret parameters (helper and mask polynomials), the system achieves enhanced secrecy while maintaining manageable complexity through structured polynomial operations.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
At a time of enrollment, a client terminal (200): generates a feature polynomial from biometric information for enrollment; multiplies the feature polynomial by a prescribed integer; stores a helper polynomial (h) obtained by multiplying an inverse polynomial of a template polynomial (ϕ), a polynomial having a small norm; and transmits the template polynomial (ϕ) to an authentication server (100). The authentication server (100) stores the template polynomial (ϕ) in a storage unit (105). At a time of authentication, the client terminal (200) : generates a feature polynomial from biometric information for authentication; multiplies the helper polynomial (h) by the feature polynomial; adds a random polynomial having a small norm to the authentication polynomial; and transmits the authentication polynomial to the authentication server (100). The authentication server (100) determines whether or not the biometric information for enrollment and for authentication can be authenticated, based on the result obtained by multiplying the authentication polynomial by the template polynomial (ϕ).