A novel continuous authentication method using biometrics for IOT devices

In this paper, we examine continuous authentication for IoT devices using real-time biometrics of a person's electrocardiogram (ECG) and electromyography (EMG). ECG is mainly used as a biometric identifier because it has specific features such as mathematical, morphological, and wavelet characteristics. EMG is a bio-signal defining a hand gesture of a person. Our authentication system would require no human interaction as it will have a continuous authentication schema. As soon as the user leaves a specific perimeter, the session will be killed by the system. In this paper, we propose a challenging and integrated methodology for developing, prototyping, and evaluating a continuous authentication scheme to ensure that currently insecure IoT networks are improved to have a high level of security with two layers of biometric security with continuous authentication to perform authentication in an automated manner. We used the dataset from PhysioNet for ECG, which contains samples of around 12 K for 298 people. We also used the EMG dataset available on the geostatic python library containing 150 K samples. In this experiment, we concluded that it is viable to use our continuous authentication for IoT devices with the lowest performance consumption and power consumption. The experimentation also demonstrates that the training model on two bio-signals helps obtain higher accuracy on continuous authentication within an average of 99.6%-99.99%. Our authentication schema can be implemented and integrated on any IoT device with having at least one wireless frequency that can receive and send a signal to the sender/authenticator.