Deployed sensor networks represent infrastructure that can and in fact must be used by many applications. This raises the question of where in the stack to place certain functionality. A particular issue our group has been wrestling with lately in this regard is that of data cleaning. Our original thought was to draw a line in the architecture beneath which we would place cleaning functionality. This allows all applications to built without having to worry about dirty data. As I've looked into this further, however, it's become apparent that a) data cleaning for sensors is hard, and b) the notion of "clean" can be application-dependent. Thus, I believe that we need to figure out how to build more flexible architectures that address the potentially conflicting goals of protecting applications from complexity and allowing cleaning functionality to be adjusted on a per-application basis.

Michael Franklin is a Professor of Computer Science at the University of California, Berkeley. He is also Co-Founder and CTO of Amalgamated Insight, Inc., a technology start up in Foster City, CA. At Berkeley his research focuses on the architecture and performance of distributed data management and information systems. His recent projects cover the areas of wireless sensor networks, XML message brokers, data stream processing, scientific grid computing, and data management for the digital home. He spent several years as a database systems developer in industry prior to receiving his Ph.D. from the University of Wisconsin, Madison in 1993. He recently served as program committee co-chair of the 2005 ICDE conference, and is on the editorial board of the ACM Transactions on Database Systems and the VLDB Journal. In 2003 he was an Executive-in-Residence at the Mayfield Fund, where he focused on emerging opportunities in sensor networks, RFID, and distributed data management. He is a Fellow of the Association for Computing Machinery, a recipient of the National Science Foundation Career Award, and the ACM SIGMOD "Test of Time" award.