QoSn-Based Management for Participatory Sensing

Open Access
Zhao, Yu
Graduate Program:
Computer Science and Engineering
Master of Science
Document Type:
Master Thesis
Date of Defense:
March 22, 2013
Committee Members:
  • Sencun Zhu, Thesis Advisor
  • QoSn-Based
  • Participatory Sensing
To enhance user experience, vendors have equipped mobile phones with various kinds of sensors in recent years. Though small, these embedded sensors have greatly improved the communication devices by making them “smarter”. Things are becoming more interesting as people are starting to share their local knowledge by providing sensing data from their phones. This is so-called “participatory sensing”. People have already seen the power of “participatory sensing” in traffic monitoring, health analysis and many other research areas. However, while sensing data are helping people understand the world, they may also leak personal private information. Though existing permission-based sensor management systems have claimed to be able to prevent privacy leakage, recently reported attacks on smartphones have shown their limitations and ineffectiveness in protecting sensitive user information. In this thesis, we propose a Quality-of-Sensing (QoSn) based management framework for participatory sensing. This framework would help people manage their sensing tasks and control the quality of sensing through a unified interface on the phone, which provides both convenience and security to users in participatory sensing. To make participatory sensing a win-win game, the framework also allows privacy bargain to let the users trade off their privacy for better rewards. This thesis presents the implementation of the framework’s phone client on the Android OS. The phone client provides unified interface to efficiently manage the sensing tasks and control the sensing quality of all the onboard sensors. We have successfully restricted the quality of sensing by creating policies. We have also made 9 sensors on Nexus 10, which is a tablet with Android system, running together at their highest rates in SENSOR_DELAY_NORMAL mode with very low network overhead and little impact to the phone, which shows the scalability of the sensing framework.