Reliability, efficiency, and timeliness as selectable services in wireless sensor networks

Data dissemination protocols for sensor networks provide resource discovery and data delivery to applications in unadministered, energy-limited, volatile environments. In order to be useful to a wide variety of applications these protocols must balance effectiveness with energy conservation. Data dissemination protocols, therefore, typically provide constrained levels of reliability, efficiency, and timeliness to the applications that use them. The levels of service actually delivered to an application are the result of design trade-offs at lower layers as well as environmental factors such as radio interference and topology. While the resultant levels of service may be appropriate for some applications, other applications require higher levels in order to function as intended. We therefore propose and demonstrate a well-engineered approach to the provision of service levels that fits nicely into existing sensor network architectures.;This dissertation defines a service space for data dissemination and demonstrates how to provide regions in this space in a manner based on the proven architectural principles of abstraction, layering, and dynamic protocol stacks. We furnish a wide range of service levels related to resource discovery and data delivery in sensor networks with emphasis on those that are difficult to achieve in real-world deployments. We eschew the construction of monolithic data dissemination protocols that are specific to the service needs of a particular class of application. Rather we combine our service components with existing general purpose network modules to produce desired levels of service.;We introduce three new protocols in this dissertation that provide non-default levels of service. The Reliable Broadcast Propagation (RBP) protocol provides probabilistic guarantees that a broadcast will propagate to all nodes in a network. Bayesian Assisted Resource Discovery (BARD) provides efficient resource discovery by modeling search and discovery as a stochastic process. Reliable Multi-Segment Transport (RMST) is a transport layer for sensor networks that provides data delivery guarantees. Our protocols expand the service space made available to applications, which is a sure way to increase the viability and applicability of sensor networks in solving real-world problems.