Two-Factor Biometric Verification with ECG: Two Cancelable Approaches

TL;DR

This paper introduces two novel cancelable ECG-based biometric verification schemes that enhance privacy and security, achieving low error rates by addressing inter-beat variability.

Contribution

It presents two new cancelable biometric verification methods using ECGs with bioconvolving and correlation filters, improving privacy and robustness.

Findings

Achieved equal error rate as low as 1%

Effective handling of inter-beat variation

Enhanced privacy with revocable templates

Abstract

Biometric authentication relies on an individual's physiological or behavioral traits to verify their identity before granting access permission to a system or device without remembering anything. Although electrocardiograms (ECGs) have been considered a biometric trait, an ECG biometric recognition system that operates in verification mode is rarely considered. This study proposes two two-factor cancelable biometric verification schemes that enable identity recognition using ECGs. Using bioconvolving and minimum average correlation energy biometric filters, revocable and irreversible templates can be constructed to avoid privacy invasion and security concerns associated with ECG biometric recognition. An interquartile range-based method is adopted to determine if an identity match exists, enabling identity verification under the influence of inter-beat variation. The experimental results indicated that the proposed schemes could achieve an equal error rate as low as 1% when the inter-beat variation was properly addressed.