Biometric personal identification system using biomedical sensors

Abstract

This paper proposes a biometric system to identify people using common biomedical sensors. A dataset is built by groups of signals acquired from 25 people, and is used in the system. The present proposal applies a combination of principal components analysis and support vector machines to identify people by a group of biometric signals: electrocardiogram, airflow, temperature, pulse oximetry, electromyogram and galvanic skin response. The testing results have achieved a 92% of correct identification rate.