Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Computer Science

First Advisor

Zhipeng Cai

Second Advisor

Yingshu Li


Mobile devices are becoming more and more indispensable in people’s daily life. They bring variety of conveniences. However, many privacy issues also arise along with the ubiquitous usage of smart devices. Nowadays, people rely on smart devices for business and work, thus much sensitive information is released. Although smart device manufactures spend much effort to provide system level strategies for privacy preservation, lots of studies have shown that these strategies are far from perfect. In this dissertation, many privacy risks are explored. Smart devices are becoming more and more powerful as more and more sensors are embedded into smart devices. In this thesis, the relationship between sensory data and a user’s location information is analyzed first. A novel inference model and a corresponding algorithm are proposed to infer a user’s location information solely based on sensory data. The proposed approach is validated towards real-world sensory data. Another privacy issue investigated in this thesis is the inference of user behaviors based on sensory data. From extensive experiment results, it is observed that there is a strong correlation between sensory data and the tap position on a smart device’s screen. A sensory data collection app is developed to collect sensory data from more than 100 volunteers. A conventional neural network

based method is proposed to infer a user’s input on a smart phone. The proposed inference model and algorithm are compared with several previous methods through extensive experiments. The results show that our method has much better accuracy. Furthermore, based on this inference model, several possible ways to steal private information are illustrated.