ITR: (NHS+ASE) - (int+dmc) - Activity-Driven Computing and Communication for Cooperative Distributed Networks

Cornell University, Ithaca NY

Investigators

Abstract

Natural and man-made disasters are, by their very nature, unpredictable and rare and thus any monitoring infrastructure for detection of such events will operate with an extremely low duty cycle. Coverage, scalability, and longevity are the key metrics to judge the effectiveness of such networks rather than traditional metrics like latency and bandwidth. This investigation examines monitoring networks from the perspective of minimizing energy usage at all levels of abstraction: circuit, node architecture, signal processing, communication, and networking. The overall goal is to develop technologies that enable sensor networks that have unprecedented longevity---on the order of years rather than days or weeks. Research on node design focuses on the development of an ultra-low-power asynchronous microprocessor optimized for sensornetwork applications. Asynchronous circuits enable extremely fine-grained power management, and lead to a significant reduction in energy per operation as well as transition times between active and idle states. Research on communication and networking focuses on collaborative signal processing techniques. Nodes in the system cooperate at the physical layer, resulting in a simultaneous reduction in the energy and latency required for information extraction. Both these research vectors are to be evaluated in the context of an experimental hardware testbed developed over the course of this project.