A machine learning-based user authentication model using mobile app data

Abstract

Modern mobile phones have become ubiquitous in smart home networks for operating and controlling home appliances. As user authentication on mobile devices is mostly achieved at initial access, it is crucial to offer an authentication approach that can be continuously employed. A solution can be found by extracting information during mobile phone usage, including application access events. In this paper, we present a machine learning user authentication model for smart home networks that utilizes application access events. To validate our model, we use a real-world dataset. Considering that users may run their mobile applications differently during weekdays and weekends, the model is evaluated for continuous authentication of users who utilize shared apps at the same daily intervals during weekdays and weekends. Additionally, we assess various classifiers regarding legitimate user identification. The results show that the presented model authenticates users, with high F-measure, based on application access events.