Programming wireless sensor networks is notoriously difficult because of severe resource constraints for sensors, unstable and low-bandwidth communication links, unreliable nodes, inaccessible and hostile deployment environments for certain applications, and potentially a large number of nodes. To ease the building of sensor network applications, the objective of this research is to develop a high-level programming abstraction that facilitates resource-efficient, data-centric, and trustworthy computing for large-scale sensor networks, and is applicable to a wide range of sensor network applications.
This approach involves designing a sensor coordination model called active dataspace (ADS), an active data repository that provides associative operations for data access, and developing techniques to implement the ADS model in a resource-efficient, robust, and trustworthy fashion. This research extends previous work on the tuple space coordination model to address the challenges for sensor networks. Specifically, the ADS model presents a novel construct called virtual tuple that supports a data-on-demand strategy to conserve the resources of data producers when their services are not needed, and presents constructs that facilitate in-network aggregation and exploiting locality.
This research is expected to provide an effective means to develop sensor network applications for a wide range of problem domains and to impact education through student participation. The results will be disseminated through papers and the Web.
