Algorithm Design and Optimization for Sensor Systems: (Invited Talk)

Distributed collaborative computation will play a key role in next generation of smart sensor systems. Till now, most research has focused upon development of networking protocols and simple data aggregation algorithms to facilitate robust communication and localized processing in sensor systems. One of the key challenges involved in realizing future sensor systems is design of efficient and scalable algorithms to facilitate large scale collaborative computation in such networks. In the state-of-the-art, partitioning of computation tasks among nodes, nodelevel optimization, and data routing is done in an ad-hoc, largely empirical manner. While this does not necessarily result in inferior designs, it does require the application developer to be aware of the details of the underlying node hardware, networking paradigms, and their performance. We believe that next logical step in the evolution of networked wireless sensor systems is to develop computation models and programming abstractions to complement the existing body research. A layer of abstraction from an algorithm design perspective is needed to provide the end user with a modular, layered, composable paradigm to design and optimize networked sensor systems. In this talk, we discuss various issues in addressing the following questions: