Traditional design approaches that treat the processor node, network, and sensor/actuator subsystems as distinctly independent components do not apply well to the domain of sensor webs. This is a result of sensor web systems being comprised of hundreds of thousands of small, autonomous devices that dynamically engage or disengage in the generation of data. Nodes within these systems are deployed in an ad-hoc fashion, with no a-priori knowledge of network and sensor/actuator connectivity.
These new requirements call for a new machine model that more closely supports the sensor webs unique operational and behavioral requirements. This research focuses on a holistic system-centric approach that defines such a new integrated communications/computation machine model for sensor web nodes. The new machine model is expressed in terms of a Meta instruction set architecture (ISA). This Meta ISA defines primitive system operations, such as domain selections, data reduction, and broadcast across ad hoc networks of nodes, much like the ISA of a CPU defining the CPU's primitive operations. The Meta ISA enables consistent and reliable system level operation for dynamically changing numbers and locations of sensors, and supports multimode data fusion. The Meta ISA explicitly supports attribute based naming, which enables the dynamic creation of unique knowledge domains across unknown numbers of heterogeneous sensor types. A new routing protocol has is studied as an integral part of the integrated computation/communication model. The protocol is active in issuing and decoding Meta instructions throughout the sensor web and has been designed specifically to support attribute-based naming and formation of knowledge domains across large ad hoc webs.
The broader impact of this project includes the development of an educational focus area in the engineering of computer-based systems. Courses include embedded and real-time systems, sensor webs, run time software for embedded systems, and distributed systems. The project is achieving impact for both graduate and undergraduate students by integrating the research agenda and results into the classroom, and focusing collaborative activities. Software tools and hardware platforms developed in this project will enhance student laboratories for experimentation and projects.
