Designing Efficient Context-Aware Mobile Systems

24 Jan
Friday, 01/24/2014 4:00am to 6:00am
Ph.D. Dissertation Proposal Defense

Abhinav Parate

Computer Science Building, Room 151

The proliferation of sensors on smartphones, tablets and wearables has led to a plethora of context classifiers designed to sense the individual's context on mobile devices. These contexts provide a rich source of information about an individual user and its surroundings such as individual's daily habits, activity, physiology, locations, emotional and social contexts. In this thesis, we investigate ways in which these insights into individual's behavior can be enhanced and utilized by answering the following three questions: 1) How can we improve the performance of context sensing using the insights into individual user's behavior? 2) How can we utilize user contexts to improve the user experience with mobile apps? 3) How can we leverage wearable accessories such as a wristband in conjunction with a smartphone to gain additional behavioral insights about the user?

In this thesis, we first design a context-query engine that performs information fusion across contexts for an individual user to enable optimizations like i) energy-efficient sensing, and ii) accurate context inference, while minimizing the privacy risks associated with the sensitive contexts. Next,  we explore the utility of user contexts in improving the freshness of the mobile apps such as Facebook that present users with the latest content fetched from the remote servers.  We present an algorithm that utilizes user contexts to decide when to automatically refresh an app so as to present a user with the freshest content. Finally, we demonstrate the utility of a wristband equipped with inertial sensors in detecting health-related behaviors such as smoking and eating in real-time. Unlike existing works, our approach detects these behaviors in a passive manner without any explicit user interaction and does not require use of any cumbersome device.

Advisor: Deepak Ganesan