Heart of IoT: ECG as biometric sign for authentication and identification.

As IoT (Internet of Things) has expanded year over year enabling the presence of sensing in almost everywhere. This leads to increase the concerns about authentication and security. In this scenario, the academy is engaged with alternatives to automated recognition of individuals and provide proof of liveness. In this context, the application of physiological features such as ECG (electrocardiography), PPG (photoplethysmography), and EMG (electromyogram) is a promising approach for continuous authentication. Specifically, ECG has been used by many researchers as biometric identification, since it has features that are unique to an individual, such as, statistical, morphological, and wavelet features. In this paper, we proposed a particular feature selection, using only fiducial points related to amplitude and time that can be found directly from the signal acquired, without any kind of complex processing. We also investigate some of the most used machine learning algorithms for user identification. Evaluation results show the potential of the proposed solution, which has reached accuracy higher than 98.2% in the continuous authentication and identification scenario, which seems to be a feasible approach to increase security in many critical applications and services.