Cancelable Biometric Authentication via Client-Side Inversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing biometric authentication systems face challenges in protecting user privacy and security, particularly when biometric templates are leaked or shared across different systems, as they are not easily changeable and can be vulnerable to forgery and unauthorized access.
Innovation Solution
A method for cancelable biometric authentication is implemented, where biometric images are randomized using cross-correlation and random filters, keeping the original images confidential from the server, allowing for secure transformation and regeneration of templates, thus maintaining privacy and security even if templates are leaked.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If biometric templates are stored on the server for authentication, then authentication functionality is achieved, but privacy security deteriorates due to potential leakage and inability to change templates
Solution Approach 1:
The patent segments the biometric authentication process into two parts: the server stores only encrypted template data without decryption capability, while the client holds the decryption key. This segmentation ensures the server cannot access raw biometric information even if templates are leaked, resolving the contradiction between maintaining authentication functionality and protecting privacy security.
Solution Approach 2:
The patent introduces encryption as an intermediary mechanism between the biometric template and the server storage. The template is encrypted using a key held by the client, creating a protective layer that prevents the server from accessing or leaking raw biometric information, thus protecting privacy while maintaining authentication capability.
2Reliability
If biometric information is encrypted and transmitted to the server, then transmission security is improved, but privacy protection deteriorates because the server can still decode and potentially leak the information
Solution Approach 1:
The patent inverts the traditional encryption model by making the client the key holder rather than the server. Instead of the server encrypting and managing keys, the client encrypts the template and retains the decryption key, reversing the power dynamic to ensure the server cannot access raw biometric information even after decryption occurs on the client side.
Solution Approach 2:
The patent uses client-side decryption as an intermediary step that prevents raw biometric information from ever residing on the server. The encryption/decryption process acts as a mediator that ensures transmission security while simultaneously protecting privacy by eliminating the server's ability to store or access plaintext biometric data.
3Adaptability or versatility
If the same biometric information is registered to different systems, then versatility is improved, but security deteriorates because a leak in one system compromises all systems
Solution Approach 1:
The patent applies local quality by allowing each system to have its own unique encryption key while maintaining the same underlying biometric template. Each system's encrypted template has local security properties specific to that system, so a leak in one system does not compromise other systems that use different encryption keys.
Solution Approach 2:
The patent uses parameter changes by varying the encryption key parameter for each system while keeping the biometric template constant. This allows the same biometric information to be registered across multiple systems with different security parameters, ensuring that a security breach in one system (with its specific key) does not affect other systems with different keys.
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
A personal authentication method is provided for authenticating a user by cross-relation between an enrolled image and a verification image of biometric information of the user. Upon registration of the biometric information, the method generates a filter for scrambling the image and an inverse filter thereof, and applies the filter to the enrolled image generated from the biometric information to generate a registration template which is then stored to a memory. Upon authentication of the user, the method applies the inverse filter to the verification image generated from the biometric information collected from the user, and then verifies the identity of the user based on cross-relation between the verification image after application of the inverse filter and the registration template.