DATE: Thursday, February 24, 2005
TIME: 12:00 pm - 1:00 pm
PLACE: Hamerschlag Hall D-210

Suman Nath

Exploiting Redundancy for Robust Sensing

Over the last few years, we have seen a number of real-world deployments of applications that use live sensor data. One of the challenges of these applications is to reliably collect the desired data, since they must operate within various constraints and failures are norms in their deployment environments.

In this talk, I will show how several domain-specific properties can be efficiently exploited to make these sensing applications extremely robust against failures. In particular, for the wireless sensors (e.g., motes), I will show how the underlying broadcast medium can be exploited to robustly compute aggregates over sensor data and to dynamically adapt the topology. For the sensors (e.g., webcams) directly connected to the Internet, I will show how the weak data consistency requirements of the applications can be exploited to tolerate many large failures common to today's Internet. I will describe the challenges and my solutions in the context of IrisNet, a sensing infrastructure that I have designed and implemented as part of my thesis work.

Suman Nath is a PhD candidate of the Computer Science Department. He earned a B.S. at Bangladesh University of Engineering and Technology in 1998, and a M.S. at CMU in 2003. His research interests lie in the intersection of sensor networks, distributed systems, and databases.

For Further Seminar Info Contact:

or visit