Locality Sensitive Hashing for ECG-based Subject Identification

There are more than 3.5 billion people around the world has utilized Internet-based services for various purposes such as news, businesses, communication and office applications; and accessing to the Internet by various modern devices for instance smartphone, tablet and laptop. Indeed, the way of doing businesses across the world also have been changed toward relying upon providing services through the internet network; at the same time the threats usually exist alongside within these technologies. Different technologies have been utilized to secure access to these services. Biometrics approaches, likewise, fingerprint has a popular use for user identification but they have some serious limitations such as cost, reliability, and etc. In this paper, we propose electrocardiogram (ECG)-based subject identification system using Locality Sensitive Hashing (LSH) with fiducial and non-fiducial feature extraction approaches. The proposed identification system was evaluated on 42 subjects of Capnobase dataset. The results show that the system scored 99.07% identification accuracy for LSH-fiducial based approach and 99.16% for the LSH-non-fiducial based approach.