Efficient data retrieval in wireless sensor networks

Wireless sensor networks that are deeply embedded in the environment are increasingly deployed in reality, providing an interface between the Internet and the physical world. By means of the widespread distribution of sensor nodes equipped with sensing, communication, and actuation capabilities, it is now possible to observe and monitor physical phenomena at unprecedented high resolutions. These tasks are implemented by a process of data retrieval to extract relevant observations from the sensor nodes, including the identification, sampling and collection of sensor data in response to application requests. Data retrieval in wireless sensor networks faces many challenges because of disparate application requirements and strict resource constraints. In this dissertation, we identify and solve various research problems related to data retrieval. We propose a general system structure and elaborate on the implementation and optimization of principal components of the system. First, we present the design of a query layer that translates application requests into queries in a declarative language, generates efficient query execution plans, and processes the queries with in-network aggregation. For applications where the query layer receives simultaneous requests, we develop a multi-query optimization algorithm that minimizes the overall communication cost. Next, we demonstrate modeling and compression algorithms to reduce the size of all data returned to the server by exploring sensor data coherence. To deliver the data to the server reliably and energy efficiently, we extend the interface of the routing layer and modify an existing ad-hoc routing protocol, by considering the special communication pattern and particular requirements of sensor network applications. Finally, we present several scalable network scheduling protocols to further improve the communication performance. Our network scheduling protocols set up a transmission schedule through standard data aggregation and dissemination processes, and significantly reduce the probability of collisions and communication interference.