Data Management in Participatory Sensing

In recent years there has been a proliferation of privately owned sensing devices such as GPS devices, cameras, home weather stations and, more importantly, smart-phones. Most of these devices are either intrinsically mobile, e.g., smart-phones and GPS devices, or can be easily carried by people during their daily activities. Nowadays, it is possible to embed various sensors in small devices as the result of sensor technology advancement. For example, we can consider smart-phones as sensing devices because they are equipped with several sensors such as GPS, accelerometer, gyroscope, microphone, and proximity sensors. This provides an unprecedented opportunity for a new application paradigm called participatory sensing, in which people collect and share sensing data about some phenomenon of interest in their environment. This unique opportunity is mainly due to (I) the ubiquity of smart-phones with various built-in sensors, (II) the availability of small, low-cost and pluggable sensors, and (III) the easy access to various connectivity media such as 3G, 4G, and WiFi. However, for using the full potential of participatory sensing, several challenges exist that must be addressed. These challenges include, but are not limited to, privacy protection of participants, quality assessment of collected data, efficient energy consumption of sensing devices, data unavailability due to uncontrolled mobility of the participants, and efficiently incentivizing people to participate. In this thesis we propose methods for addressing some of these issues. In particular, this thesis addresses the following topics: ** Efficient Data Acquisition in Participatory Sensing. In participatory sensing systems participant often require to make some level of effort for data collection and sharing, which includes the consumption of the limited resources on their devices. Some people might altruistically participate in such systems. However, it is not realistic to assume that all participants offer this effort altruistically. Therefore, adequate incentives must be given to people to participate. One common approach is to provide the participants with monetary incentives. Additionally, data need not be constantly collected at all places. In many applications, data collection is necessary only when there is some utility for the data. The difference between the value of the collected data to the application and the data collection cost is defined as the utility of the data. We propose a data acquisition framework in Chapter 3 for participatory sensing systems. This framework takes into account the major factors pertinent to this context and efficiently shares sensor data among queries of different types with the objective of maximizing the total utility. Queries for sensor data can come from multiple different applications with arbitrary utility considerations. ** Truthful Data Elicitation in Participatory Sensing. In participatory sensing systems, some participants might have incentives to report wrong data. For example, a participant might report higher costs for her data or wrong location tags for the data with the objective of receiving higher payments. Therefore, it is critical to prevent dishonest behavior of participants by appropriately designing the participatory system. [...]