Electrocardiogram Excited Fiducial Domain Predictor: An Active Method of Determining Fiducial Domains of Excited ECGs for Enhanced Authentication

Abstract

This paper proposes a novel method of extracting similar and invariant parameters from exited electrocardiogram (ECG) for enhanced authentication of non-rest state subjects. We offer a method of accurately predicting excited fiducial domains even when participants submit ECG data that is collected during different states of excitement. Rest state ECG is obtained and used as baseline. Multiple ECGs are obtained from human subject at various excitement levels. The excited ECGs are superimposed on rest state ECG baseline to obtain invariant fiducials domains which are used to model generic predictor vectors. The predictor vectors can now be used (applied) on other unexcited subject ECG to predict their exited fiducial domains. By combining traditional rest state baseline ECG authentication methods with our excited fiducial domain predictor method, we envisage enhanced True Acceptance Rate (TAR) authentication in spite of the subject’s state of rest or excitement. Results indicates that exited subjects were authenticated with TAR exceeding 50% in spite of them only initially submitting rest state ECG baseline data. Authentication rate declined with increased activity rate achieving 100% TAR authentication for subjects at rest, 84% TAR for both walking and brisk walking subjects and 58%TAR for jogging subjects.