Agu, Emmanuel O.
Smartphones are increasingly essential to users’ everyday lives. Security concerns of data compromises are growing, and explicit authentication methods are proving to be inconvenient and insufficient. Meanwhile, users demand quicker and more secure authentication. To address this, a user can be authenticated continuously and implicitly, through understanding consistency in their behavior. This research project develops a Behavioral Model System (BMS) that records users’ behavioral metrics on an Android device and sends them to a server to develop a behavioral model for the user. Once a strong model is generated with TensorFlow, a user’s most recent behavior is queried against the model to authenticate them. The model is tested across its metrics to evaluate the reliability of BMS.
Worcester Polytechnic Institute
Major Qualifying Project
All authors have granted to WPI a nonexclusive royalty-free license to distribute copies of the work, subject to other agreements. Copyright is held by the author or authors, with all rights reserved, unless otherwise noted.