Wireless sensor networks are a new class of computing platform that integrate low-power computation, sensing, and wireless communication to form distributed networks capable of measuring physical systems at scale. Sensor network "macroscopes" have the potential to deeply impact many scientific disciplines by enabling continuous, in situ collection of sensor data at scales that have been previously unattainable. We have developed and deployed sensor networks for a range of scientific studies, including volcano seismology, neuromotor disease rehabilitation, and urban-scale environmental monitoring. Through these experiences we have identified key research challenges that must be addressed to enable high data fidelity and long network lifetimes in challenging field settings. These include managing the limited resources of individual sensor nodes, and reducing the complexity of programming network-wide behaviors. In this talk, I will describe several of our efforts to tackle these challenges. I will describe resource aware programming, a new approach to capturing resource management and adaptivity at the node and network level, and a new operating system design based on this model. I will also describe macroprogramming, a new approach to distributed programming in which the global behavior of the network is specified at a high level, and automatically compiled down to the node-level program. I will also present results from several field deployments of sensor networks.Bio:
Matt Welsh is an Associate Professor of Computer Science at Harvard University. His research interests focus on operating system, network, and programming language support for complex distributed systems. He received his Ph.D. and M.S. from UC Berkeley and his B.S. from Cornell University.