Abstract:
Sensor networks hold promise to allow researchers to gather data more efficiently and thoroughly than previously possible. These networks consist of small battery-powered nodes equipped with sensors and radios, the latter used to return sensed data through the network. Communication is typically achieved by having nodes organize themselves into an ad hoc network, where they transmit among each other over limited distances.

This necessitates a measure of cooperation among the nodes, since some are invariably located between pairs of sources and destinations, and must relay messages on their behalf, even while transmitting their own readings. These varied workloads lead to unbalanced consumption and are of particular concern. If nodes acting as forwarders fail and cause a network partition, remaining nodes may continue sensing, but have no way to transmit data to their destinations.

In this talk, I will discuss the SEESAW MAC layer protocol. The goal of this work is to alleviate the burden on forwarders and balance energy consumption across all nodes, such that the useful network lifetime is extended. SEESAW uses a number of innovative techniques that afford it great flexibility in shifting communication overhead costs among nodes. I will discuss how to automatically tune these techniques to achieve balance in a variety of changing application workloads. I will present experimental results, both simulated and from a Mica2 sensor node implementation, evaluating SEESAW's flexibility and distributed tuning algorithm.

Bio:
Rebecca Braynard received her B.S. degree in Computer Science from Illinois State University in Normal, Illinois in 2000. She immediately entered the graduate program at Duke University in Durham, North Carolina where she graduated with her PhD in May 2006 under the advisement of Dr. Carla Schlatter Ellis. Her research interests include distributed systems, wireless networking, overlay networks and energy-efficient sensor network applications.