Michael Franklin
Professor, Computer Science Department, UC Berkeley
Faculty Affiliate Researcher, Intel Research Berkeley

My involvement with the Intel Berkeley Lab came about during the course of the TelegraphCQ project that [lab director] Joe Hellerstein and I were running at UC-Berkeley. TelegraphCQ is a database query engine that processes data as it flows through a network. As part of that project, we were looking for sources of streaming data and were interested in getting data streams from sensor networks. We asked a Ph.D. student, Sam Madden, to spend some time at the Intel lab to learn how sensors work and to figure out how to hook them up to our system. In the process, Sam gained some interesting insights into how we could apply some of the query processing techniques we were using at the university to make sensor networks easier to program.

In fact, Sam's project ended up morphing quite a bit and became what is now known as TinyDB. TinyDB employs database queries as an interface to sensor networks in order to hide the complexity and variability of the underlying hardware and environment. The result is a dramatic reduction in the time to develop certain types of sensor applications and the ability to more quickly reconfigure a deployed network in such applications. Now that we have TinyDB in place, we are hooking it up to our host-based TelegraphCQ system and are starting to see some interesting results in that project as well. So this interaction clearly showed the benefits to both parties of collaboration between the university and the Intel lab.

At the time we started our sensor network project, there were other academic projects competing in the same space. The Intel lab gave us a tremendous advantage; by leveraging the lab's resources we were able to produce not only research results but also working systems that have been deployed in real applications.

One of the amazing things about how the Intel lab has been set up is the open and collaborative research model. The model makes it much, much easier for me to think about working with the lab and more importantly, to consider having my students collaborate with lab researchers. Under the lab's model, there is no question about ownership of intellectual property, what should be published and when, and so on. In a university environment, those kinds of restrictions make collaborative research very difficult. Having an open and collaborative agreement lowers the barrier between the lab and university and allows me and my students to flow back and forth fairly seamlessly.