Efficient Information Access in Data-Intensive Sensor Networks

Recent advances in wireless communications and microelectronics have enabled wide deployment of smart sensor networks. Such networks naturally apply to a broad range of applications that involve system monitoring and information tracking (e.g., fine-grained weather/environmental monitoring, structural health monitoring, urban-scale traffic or parking monitoring, gunshot detection, monitoring volcanic eruptions, measuring rate of melting glaciers, forest fire detection, emergency medical care, disaster response, airport security infrastructure, monitoring of children in metropolitan areas, product transition in warehouse networks etc.).Meanwhile, existing wireless sensor networks (WSNs) perform poorly when the applications have high bandwidth needs for data transmission and stringent delay constraints against the network communication. Such requirements are common for Data Intensive Sensor Networks (DISNs) implementing Mission-Critical Monitoring applications (MCM applications).We propose to enhance existing wireless network standards with flexible query optimization strategies that take into account network constraints and application-specific data delivery patterns in order to meet high performance requirements of MCM applications.In this respect, this dissertation has two major contributions: First, we have developed an algebraic framework called Data Transmission Algebra (DTA) for collision-aware concurrent data transmissions. Here, we have merged the serialization concept from the databases with the knowledge of wireless network characteristics. We have developed an optimizer that uses the DTA framework, and generates an optimal data transmission schedule with respect to latency, throughput, and energy usage. We have extended the DTA framework to handle location-based trust and sensor mobility. We improved DTA scalability with Whirlpool data delivery mechanism, which takes advantage of partitioning of the network. Second, we propose relaxed optimization strategy and develop an adaptive approach to deliver data in data-intensive wireless sensor networks. In particular, we have shown that local actions at nodes help network to adapt in worse network conditions and perform better. We show that local decisions at the nodes can converge towards desirable global network properties e.g.,high packet success ratio for the network. We have also developed a network monitoring tool to assess the state and dynamic convergence of the WSN, and force it towards better performance