Distributed Computation in the Physical World

Summary form only given. Networks of intelligent sensors that are distributed through the physical world will revolutionize practices in the life sciences, civil engineering, manufacturing, security, agriculture, ubiquitous computing, and many other areas. They also present an opportunity and a need to explore distributed algorithms that are wedded to the noisy, localized, time varying physical world. Bandwidth, storage, and energy limitations make in-network processing essential - within the node and among collections of nodes. The algorithms should be resource efficient, but also deal with noise, uncertainty and dynamically changing connectivity. A broad research community has been exploring these issues in the context of TinyOS and the Berkeley motes. This talk will highlight novel distributed algorithms coming out of these efforts and discuss issues in making such networks robust and programmable