493: Undecipherable

Gesture-based User Authentication on Mobile Devices using Accelerometer and Gyroscope

Abstract: Mobile devices offer their users lots of possibilities and a feeling of freedom. However, this freedom comes along with new security threats. Sensitive data might be stolen and abused, if an unauthorized person gets unrestricted access to such devices. Therefore, user authentication mechanisms are required. So far authentication mechanisms like PINs and passwords do not take into account the limited capabilities of user interfaces of mobile devices. So, it is necessary to create and develop specially adapted mechanisms, which are designed to be usable under these restrictions. Prior work has shown that gestural interfaces have potential as gestural interaction is natural to humans. In this thesis a behavioral biometric user authentication mechanism for mobile devices based upon gestures is presented using an embedded 3-dimensional accelerometer and a 3-dimensional gyroscope. As recognition algorithms Dynamic Time Warping and Hidden Markov Models were studied. The designed mechanism of this thesis was evaluated in a user study including a realistic attack. The conducted user study confirmed that gesture-based authentication is feasible, usable and promising for mobile devices.

Paper