Recent advances in ubiquitous computing have led to the emergence of wireless sensor networks. Wireless sensor networks enable continuous data collection on unprecedented scales and give us new opportunities for observing and interacting with the physical world. However, collecting data from wireless sensor networks is challenging because of limited resources of sensor nodes. Since the energy constraint is the most critical issue in wireless sensor networks, it is crucial to reduce the energy consumption for query processing. This dissertation deals with energy-efficient query processing problems in wireless sensor networks.
An wireless sensor network is usually deployed with static sensor nodes to collect sensor data in the region of interest. Nowadays, mobile wireless sensor networks have emerged with the advances in distributed robotics and low power embedded computing technologies. Mobile wireless sensor networks have similar characteristics to their static counterparts but they have more challenges because of the mobility of sensor nodes. In this dissertation, we focus on both static and mobile wireless sensor network environments.
First, we study the iceberg query processing problem in static wireless sensor networks. The iceberg query finds data whose aggregate values exceed a pre-specified threshold. This iceberg query can be used in various wireless sensor network applications such as environmental monitoring, industrial maintenance and battlefield surveillance to extract meaningful information from wireless sensor networks. To process an iceberg query in wireless sensor networks, all sensor data have to be aggregated and then sensor data whose aggregate values are smaller than the threshold are eliminated. Whether a certain sensor data is in the query result depends on the other sensor data values. Since sensor nodes are distributed, communications between sensor nodes are required to know the sensor data from the other sensor nodes. However, sensor nodes have limited energy resources and communication is a primary source of the energy consumption. Thus, reducing the communication overhead is the most critical issue in wireless sensor networks. In this dissertation, we propose an energy-efficient iceberg query processing technique in static wireless sensor networks. To compactly represent the data transmitted, a lossless sensor data compression method based on an established mathematical property is devised. To reduce the energy consumption caused by the number of data transmitted, a filtering based query processing method is devised. Using the temporal correlation of sensor data and the semantics of an iceberg query, a prediction model for the future query result is proposed. Based on the predicted future query result, sensor nodes effectively filter out unnecessary transmissions.
Second, we tackle the top-k query processing problem in mobile wireless sensor networks. A mobile wireless sensor network is a wireless sensor network in which sensor nodes are mobile. The mobile sensor nodes move around and explore their surrounding areas. Top-k queries are useful in many mobile wireless sensor network applications. However, the mobility of sensor nodes incurs new challenges in addition to the problem of static wireless sensor networks. Since mobile sensor nodes tend to move continuously, the network condition changes frequently and they consume considerably more energy than static sensor nodes. In this dissertation, we propose an efficient top-k query processing framework in a mobile wireless sensor network environment. To construct an efficient routing topology, we devise a mobility-aware routing method. Using the semantics of the top-k query, we develop a filter-based data collection method which can save the energy consumption and provide more accurate query results. We also devise a data compression method for disconnected sensor nodes to deal with the problem of limited memory space of sensor nodes.
Finally, we perform extensive experiments to evaluate the performance of the proposed approaches using both real and synthetic datasets. The experimental results confirm the energy-efficiency and the effectiveness of the proposed approaches. For the iceberg query processing problem in static wireless sensor networks, on the average, our approach consumes approximately 90% less energy than an existing approach while providing nearly 100% accurate query results. For the top-k query processing problem in mobile wireless sensor networks, our approach reduces energy consumption up to about 68% than other top-k query processing approaches and provides near 100% accuracy of the query results. In addition, on the average, our approach uses approximately 47.5% less memory than a compression method for low powered wireless sensor nodes.