Biometric Template Secret Sharing Across Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Previous biometric systems face vulnerabilities in protecting biometric templates, including software-based encryption approaches that introduce additional security risks and hardware-based solutions that are limited in availability, expensive, and require re-enrollment on new devices, leading to inconvenience and security breaches.
Innovation Solution
A distributed biometric system splits the biometric template into multiple shares using a polynomial-based secret sharing scheme, distributing them across multiple components, including user devices and a remote server, allowing reconstruction with a threshold number of shares, ensuring that no single location has the complete template and enabling flexible matching on either devices or servers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If biometric templates are stored in a central database or user devices with software-based encryption, then the templates can be accessed for comparison, but the system introduces additional security vulnerabilities requiring cryptographic key storage and complex access controls
Solution Approach 1:
The biometric template is divided into multiple shares using secret sharing schemes, where each share is stored in different locations (central database and user devices). No single location contains the complete template, eliminating the need for cryptographic keys while maintaining security. The template is segmented such that a threshold number of shares are required for reconstruction.
2Reliability
If hardware-based security modules are used to protect biometric templates, then security is improved, but the modules are limited in availability, expensive, and require re-enrollment on new devices
Solution Approach 1:
The secret sharing system enables the biometric template to be universally accessible across multiple different devices without requiring hardware security modules. The template shares can be reconstructed on any device that has the necessary shares, making the system device-agnostic and eliminating re-enrollment requirements when users switch devices.
3Productivity
If biometric templates are stored in a central database, then matching can be performed centrally, but a security breach would result in significant damages
Solution Approach 1:
The complete biometric template is segmented into multiple shares distributed between the central database and user devices. Even if the central database is breached, attackers cannot reconstruct the template without the additional shares stored on user devices. This segmentation reduces the impact of security breaches while maintaining centralized matching capability.
4Reliability
If BioHashing is used to transform biometric data, then anonymous templates can be generated, but the technique is vulnerable to intrusion and linkage attacks and requires re-enrollment on each new device
Solution Approach 1:
Instead of directly hashing biometric data as in BioHashing, the system uses secret sharing as an intermediary mechanism. The biometric template is first divided into shares, then these shares are distributed and protected. This intermediary approach provides stronger security against intrusion and linkage attacks compared to direct hashing, while enabling cross-device functionality.
Data Source
AI summary
Biometric information from an initial sample is used to generate a biometric template for a user. The biometric template is split into multiple template shares using a polynomial secret sharing scheme, such that at least some threshold number of the resulting template shares must be combined to reconstruct the biometric template. After the biometric template is split, the resulting template shares are distributed to multiple components in the system, such as a server, and/or one more user devices, and the original copy of the biometric template is destroyed. To subsequently verify the identity of the user, the threshold number of template shares are obtained and combined to reconstruct the user's biometric template, and the reconstructed template is compared with biometric information extracted from one or more subsequently collected biometric samples. If there is a match between the reconstructed biometric template and the extracted biometric information, the user's identity is verified.