We propose a practical system design for biometrics authentication based on electrocardiogram (ECG) signals collected from mobile or wearable devices. The ECG signals from such devices can be corrupted by noise as a result of movement, signal acquisition type, etc. This leads to a tradeoff between captured signal quality and ease of use. We propose the use of cross correlation of the templates extracted during the registration and authentication stages. The proposed approach can reduce the time required to achieve the target false acceptance rate (FAR) and false rejection rate (FRR). The proposed algorithms are implemented in a wearable watch for verification of feasibility. In the experiment results, the FAR and FRR are 5.2% and 1.9%, respectively, at approximately 3 s of authentication and 30 s of registration.
- electrocardiogram (ECG)
- mobile and wearable environments
- signal analysis