Biometric Verification Using Fully Homomorphic Encryption
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional biometric verification systems face challenges with intra-user variability, security concerns due to computations performed in unencrypted domains, and loss of information through quantization of feature vectors, leading to low matching performance and compromised template security.
Innovation Solution
The proposed system employs fully homomorphic encryption for secure biometric verification, performing computations and dimensionality reduction in the encrypted domain using an encrypted random projection matrix, eliminating the need for quantization and ensuring secure, accurate matching without loss of information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If computations are performed in unencrypted domains to improve processing speed, then computational efficiency is improved, but security and privacy of biometric data are compromised
Solution Approach 1:
The patent introduces a third-party trusted server that performs encrypted biometric verification computations. The server encrypts biometric templates and probe features before transmission, performs matching computations on encrypted data, and returns encrypted results. This intermediary approach allows secure verification without exposing raw biometric data to the server, resolving the contradiction between computational efficiency and security by enabling encrypted processing on powerful servers while maintaining data privacy through proper encryption protocols.
Solution Approach 2:
The patent replaces traditional unencrypted mechanical computation with encrypted computation systems. Instead of computing on raw biometric data and then protecting it, the system performs all matching operations on encrypted representations of the data. This substitution of the computational paradigm allows high-speed processing on encrypted data, simultaneously achieving both computational efficiency and security through cryptographic operations.
2Productivity
If feature vectors are quantized to reduce data complexity, then processing efficiency is improved, but information loss occurs and matching performance deteriorates
Solution Approach 1:
The patent changes the parameter representation from quantized discrete values to real-valued continuous parameters. Instead of quantizing feature vectors to reduce complexity, the system maintains high-precision real-valued feature vectors throughout the encryption and processing pipeline. This parameter change allows the system to achieve both processing efficiency through encrypted computation and information preservation through real-valued representations, eliminating the trade-off between efficiency and information loss.
3Ease of operation
If biometric templates are stored in unencrypted form to simplify system architecture, then ease of operation is improved, but template security is compromised
Solution Approach 1:
The patent applies preliminary encryption action to biometric templates before storage. Templates are encrypted using asymmetric encryption schemes before being stored in the database, ensuring that even if the storage system is compromised, the raw biometric data remains protected. The encryption keys are managed separately through a key management system, allowing simple storage operations while maintaining high security through pre-encryption of all sensitive data.
Data Source
Figure 1~2A
Figure 2B
Figure 3
AI summary
This disclosure relates generally to a method and system for biometric verification. Conventional biometric verification method and system performs one or more computations in non-encrypted domain, thereby leading to security threats. The disclosed method includes performing computations such as enrollment and verification feature vector computation, dimensionality reduction of said feature vectors, and comparison of dimensionally reduced encrypted feature vectors to obtain matching scores indicating the extent of match therebetween between in encrypted domain using fully homomorphic encryption, thereby leading to secure biometric verification.