Date of Award

9-12-2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Computer Science

First Advisor

Mengjun Xie

Abstract

Designed as personal smart assistant, smartphones and smartwatches have dramatically changed people's lives in every aspect from social network communication, navigation to both online and offline shopping. A large amount of personal data such as messages, emails and payment information is stored in such a device. Such data, if lost, can lead to privacy leakage and financial loss. Therefore, protecting such devices from unauthorized access is crucial. Despite the existence of personal identification numbers (PIN) and unlock patterns for user authentication, they have well known drawbacks. Although physiological biometrics such as fingerprints and facial recognition has become popular, they suffer from privacy and usability issues. Hence, new and user-friendly authentication schemes are needed. In this direction, behavioral biometrics authentication has attracted a lot of attentions. We have conducted a series of studies towards authenticating smart device users using behavioral biometrics. Beginning with signature, we first propose a signature dynamics based electronic consent model to enhance the verification of a remote signer's identity. Secondly, we conduct a comparative study on user behaviors in PIN typing and pattern drawing, and find that the same level of accuracy can be achieved using behaviors only. Thirdly, we propose a 3D in-air gesture based behavioral biometric authentication system for wrist worn smart devices to overcome the PIN typing challenge on a small-screen or no-screen device. Fourthly, we conduct an in-depth analysis of signature, PIN, pattern and 3D gesture behaviors, and then propose a behavior segment based approach for improving the accuracy of behavioral biometric authentication. Lastly, we systematically study the performance of user-created 3D motion gestures and then propose a guideline for creating a secure and usable 3D motion gesture. Through those studies, we demonstrate the viability and effectiveness of the behavioral biometric authentication schemes that we have developed.

Download

Share

COinS