Traceable and comparable evaluation methodology for biometric system usability
Barbara Corsetti · e-Archivo (Carlos III University of Madrid) · 2020
Biometrics is currently replacing traditional authentication mechanisms in more and more contexts. Biometric recognition allows us to identify ourselves quickly and easily without requiring any effort as when remembering secret codes or passwords. For this reason, more and more users choose to unlock their smartphones through the face or finger recognition. Hence, the release of mobile services supported by biometric recognition is increasing. This growing use of biometrics has raised new questions about the user interaction. Is it easy to interact with the new biometric applications? Can anyone use them? Are people aware of how this new technology works? Can mobile biometrics be applied to improve accessibility in daily contexts? This thesis aims to bring improvements to the study of user interaction and usability in biometrics. The study focuses on the user and his or her characteristics (e.g., experience, age, health status). For this reason, two scenario evaluations were carried out proposing biometric systems to a heterogeneous group of data subjects. The experiments were developed and designed following the directives of the ISO/IEC 21472. While the evaluation of the user interaction was conducted according to a novel methodology presented for the first time in this Thesis. The novel methodology sets specific metrics to report how the user’s characteristics impact the outcome of the biometric process. Its key point is the accessibility and all its aspects: the accessibility of the scenario, of the biometric system, and how the user’s accessibility concerns influence the interaction with biometric applications. Results show a strong correlation between the age, experience, health status of the user, and the outcome of the recognition process (in terms of performance, usability, accessibility, and biometric sample quality). This demonstrated that the accessibility brings to a lot of information regarding the human-biometric system interaction. Thus, all its aspects must be considered to improve the traceability and comparability of future assessments. The main contributions of this Thesis include: • First application of the ISO/IEC 21472 to design user interaction evaluations in Biometrics • A proposal of a novel methodology to analyse the accessibility of biometric recognition systems • Validation of this novel methodology through two user interaction evaluations.