Objectives: The use of mobile devices such as smartphones and tablets have fundamentally changed the way that users interact with ICT applications and information, moving away from a keyboard as a primary input device to a swipe and touch methodology. Devices also contain a plethora of sensors such as GPS that may be used to provide additional information about the device usage. These ‘modality’ data provide a novel opportunity to explore behavioural biometrics in both continuous and instantaneous scenarios – the former examining interaction over a prolonged period of use, whilst the latter involves obtaining a behavioural sample as part of a donation ceremony. This work will build upon extensive expertise in the field of signature biometrics wherein both the static (outcome) and dynamic (temporal) aspects of signature production are used to authenticate a signer. The work will explore two central themes. Firstly, we shall explore feature stability of individual sensor data channels allowing us to assess the biometric power of each of separate modalities for verification purposes in an instantaneous context. Secondly, we shall explore how modalities are affected by donation scenarios related to application task, environment and other usage variables. In this way we shall be able to ascertain which features remain stable for users and therefore provide candidates for a novel assessment of continuous authentication on a mobile platform.

Expected Results: There are three expected outcomes from this project: a) a novel exploration of the instantaneous biometric performance of a range of mobile platform sensors and data, b) an assessment of the stability of modalities across usage scenarios with application to continuous authentication and c) the design of an implementation framework for continuous biometrics authentication in collaboration with our industrial partner.

Planned secondment(s): Two secondments would take place within this project. The first is a five month visit to the team at WUT wherein a range of novel feature extraction algorithms will be explored. The expected results of this secondment are the rigorous design and testing of algorithms based on expertise at WUT, particularly with respect and temporal assessment and quality. The second five-month secondment would be with IBM wherein the ESR would assess authentication performance and the practical commercial aspects of implementing a continuous and instantaneous authentication method. The results of this secondment are based on the extensive use of IBM’s data analysis and exploration tools, alongside their experience with devices and platforms, and insight into novel emerging technologies in this field.

Start Date: October 2017