Touchscreen biometrics for continuous authentication
Year: 2014
Type: dissertation
Abstract: Nowadays, touchscreen devices are part of people’s daily tasks and there is, progressively, less need to use traditional computers, whether to work or entertainment activities. Consequently, these devices contain an increasingly higher quantity of sensitive information that must be protected from unauthorized access. Generally, these devices have only a single layer of authentication and, once passed this phase, it is assumed that the device is used only by its owner or authorized user. However, no matter how strong is the authentication method, the device is still vulnerable to attacks from malicious users who gain access to it in an unlocked state. This way, is pertinent to adopt techniques that provide continuous authentication of users. In this dissertation is proposed a system for continuous authentication in devices with touchscreen, aiming to serve as a proof of concept about the possibility of using the interactions of users with the touchscreen as a behavioral biometric characteristic. Was developed a mobile application to collect the touch data of some controlled gestures, performed by different users, and to extract a selected set of features, used to build their biometric template. The classification module was simulated through machine learning software. Lastly, are presented the system evaluation results. Globally, the results show that the system, using the selected features and classifier, is not capable of perform a strong continuous authentication. In fact, was verified that the system behaved quite distinctly with different users. Besides the results, it is believed that, with the right set of features and classification techniques, the system has a promising future.
Author: Sérgio Emanuel da Cruz Lopes
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2233097732
