Smartwatch Multifactor Authentication for Keyboard based Methods
Joseph G. Maes, Khandaker Abir Rahman · 2022 International Telecommunications Conference (ITC-Egypt) · 2022
A method for leveraging a smartwatch as an additional form of user authentication has been proposed in this work. Both inherent and subtle intentional wrist gestures are examined as means of appropriately identifying a user. This system leverages a custom-developed Android Wear OS application to record sensor data while typing the username-password on the keyboard. A total of 49 accelerometer samples were collected from three users and performed 3,614 different genuine and impostor comparison tests. Experimental setups include several hostile scenarios where username-password is considered compromised. To thwart such impostor attacks, the proposed method, as a standalone multifactor authentication, achieves Equal Error Rate (EER) values as low as 0% and an average of 18.7% for the observed three axes of accelerometer data.