Achieving application quality of service in resource-constrained wireless sensor networks.

Wireless sensor networks (WSNs) consist of autonomous devices that can sense the environment, process information and communicate wirelessly. They provide distributed monitoring capabilities and hence their applications include habitat monitoring, object tracking, military surveillance and search-and-rescue. These applications necessitate sensor nodes to be low cost, battery powered and have a small form factor, which results in their having very limited resources like bandwidth, computational power, memory and energy. Due to the limited resources, the diverse deployments and the non-deterministic communication medium it becomes highly challenging to provide the desired level of performance and Quality of Service (QoS) to the applications. This dissertation presents the design, development and evaluation of novel protocols and services that provide the required application QoS despite these challenges. The contributions of this dissertation are twofold. The first set of contributions include network protocols that provide application-specific QoS to a subset of WSN applications that involve mobility. Specifically, they include: (1) two wake-up and topology maintenance protocols that provide spatiotemporal performance guarantees to a class of applications involving mobility, (2) a data collection protocol called Roadmap Query which supports safe mobile entity navigation in dynamic environments, and (3) a multi-resolution location directory service that supports a rich set of multi-granular spatial queries about mobile entities, over wide areas covered by multiple sensor and IP networks. The second set of contributions involve a first look at achieving application QoS in multi-application shared WSNs. Since the concept of shared WSNs is relatively new, this dissertation aims to show the feasibility of using such systems to realize integrated sensing systems like automated building management. Significant contributions in this direction include the design, development and evaluation of a utility-based multi-application allocation and deployment environment (UMADE) for shared WSNs. UMADE achieves a near optimal distribution of sensor nodes among contending applications based on application QoS requirements, at deployment time. It supports the end-to-end process of (1) application QoS specification, (2) dynamic node allocation that maximizes overall system utility under sensor node memory constraints, (3) automatic application deployment, and (4) adaptation to network and application dynamics.