Behavioural Biometrics for Authentication

点击翻译按钮获取中文摘要

The research project aims at investigating the feasibility of using various behavioral biometrics in authentication scenarios. This field of study focuses on establishing uniquely identifying patterns in human activity such as keystroke, mouse and touchscreen dynamics as well as voice, gait and cognitive behaviour. This is typically a non-invasive method for authentication as it does not require users to learn how to operate a particular system or remember unique passcodes and phrases. Furthermore, there are no active steps required for authentication but rather it is a seamless integration with the regular operation of the system. Often authentication systems based on behavioural biometrics can be used as a multifactor safeguard in conjunction with other more traditional cybersecurity measures. Despite being a somewhat well research area with some apparently successful projects currently there are few successful commercial systems employing the technology. The goal of the research is to design, develop and test novel systems for authentication based on behavioural biometrics and close the gap between promising research and practical applications. For instance, developing a continuous authentication model based on phone usage patterns such as touchscreen gestures and gyroscope micro movements in space. Another aspect of the project focuses on identifying problems in past research in the area and some of the reasons it tends to be unsuitable for practical use. For example, the reported sample sizes in some of the studies might not be large enough to accurately represent the population using such systems. Finally, there are unique privacy challenges stemming from the use of highly accurate authentication systems based on behavioural biometrics. One way to maliciously employ this technology is to create unique fingerprints for users which can then be exploited for tracking behaviour and identity throughout multiple non-connected systems. It also might be possible to reveal personal information about users through their behaviour patterns. Gender, age and cultural groups could exert specific traits which might be detectable by the technology described above. This project falls within the ESPRC Cybersecurity research area.