Poster: An Improved Approach to Gesture-Based Authentication for Mobile Devices
Niklas Kirschnick, Sven Kratz, Sebastian Möller · 2010
We present a gesture-based authentication method for mobile devices using acceleration and angular rate data. Using data gathered from 17 test subjects we analyzed how the choice of gesture, classifier, extracted features and dimensionality reduction affects the authentication accuracy. The results enabled us to determine an optimal configuration of these variables for authentication with an accuracy of 99.94%. We show that gesturebased authentication is feasible on mobile devices, and can be improved by using a gyroscope in addition to the accelerometer presently found on many devices.