Continuous Proof of Presence based on Touchscreen Devices Interactions and Signals

点击翻译按钮获取中文摘要

Traditional access control systems that rely on static or fixed-point authentication are considered insecure andnot suitable for mission-critical systems, especially when the users share their credentials, even when thesecurity policies in place forbid those behaviours. Without continually checking that "you are who you claimyou are", a potential security breach is almost certain to happen at some point. To solve this challenges withHCI touch screen data our approach will focus on investigating the available HCI touch screen datasets andvalidate their fitness for identity recognition and verification. After studying the touchscreen interaction data,we will focus on engineering relevant features to describe the unique behaviours of the user while interactingwith touchscreen. We will investigate different feature selection and extraction techniques to identify the bestfeatures for identity verification. The next stage will focus on evaluating pattern classification algorithms thatrequire positive training only (e.g. artificial immune system) to identify the most promising algorithm fornovelty detection. Finally, we will propose an architecture for a non-intrusive continuous authentication. Theexpected outcome of the project includes the following:1. Evaluate the effectiveness of using touchscreen data for continuous authentication.2. Engineering features for pattern classifications from touchscreen interaction data3. Design a new non-intrusive continuous authentication model based on touch screen data.