ORCID ID

0000-0002-9500-5554

Date Awarded

2018

Document Type

Dissertation

Degree Name

Doctor of Philosophy (Ph.D.)

Department

Computer Science

Advisor

Qun Li

Committee Member

Gang Zhou

Committee Member

Pieter Peers

Committee Member

Adwait Jog

Committee Member

Jennifer A Stevens

Abstract

New personal computing technologies such as smartphones and personal fitness trackers are widely integrated into user lifestyles. Users possess a wide range of skills, attributes and backgrounds. It is important to understand user technology practices to ensure that new designs are usable and productive. Conversely, it is important to leverage our understanding of user characteristics to optimize new technology efficiency and effectiveness. Our work initially focused on studying older users, and personal fitness tracker users. We applied the insights from these investigations to develop new techniques improving user security protections, computational efficiency, and also enhancing the user experience. We offer that by increasing the usability, efficiency and security of personal computing technology, users will enjoy greater privacy protections along with experiencing greater enjoyment of their personal computing devices. Our first project resulted in an improved authentication system for older users based on familiar facial images. Our investigation revealed that older users are often challenged by traditional text passwords, resulting in decreased technology use or less than optimal password practices. Our graphical password-based system relies on memorable images from the user's personal past history. Our usability study demonstrated that this system was easy to use, enjoyable, and fast. We show that this technique is extendable to smartphones. Personal fitness trackers are very popular devices, often worn by users all day. Our personal fitness tracker investigation provides the first quantitative baseline of usage patterns with this device. By exploring public data, real-world user motivations, reliability concerns, activity levels, and fitness-related socialization patterns were discerned. This knowledge lends insight to active user practices. Personal user movement data is captured by sensors, then analyzed to provide benefits to the user. The dynamic time warping technique enables comparison of unequal data sequences, and sequences containing events at offset times. Existing techniques target short data sequences. Our Phase-aware Dynamic Time Warping algorithm focuses on a class of sinusoidal user movement patterns, resulting in improved efficiency over existing methods. Lastly, we address user data privacy concerns in an environment where user data is increasingly flowing to manufacturer remote cloud servers for analysis. Our secure computation technique protects the user's privacy while data is in transit and while resident on cloud computing resources. Our technique also protects important data on cloud servers from exposure to individual users.

DOI

http://dx.doi.org/10.21220/s2-rtcz-4157

Rights

© The Author

Download

Share Feedback

Share

COinS