2010 Fritz Kutter Prize is awarded to René Müller

René, my sincere congratulations on the Fritz Kutter Award. You presented in your doctoral thesis the data processing on two different platforms: the wireless sensor networks and the field-programmable gate arrays. Could you please explain to the non-scientists among us what the technological solution is all about?
Thank you. I am honored by this recognition of my work. The work of my thesis addresses a problem we face in today's information processing. Data volumes we have to deal with are humongous and steadily increasing. A significant part of the data arrives in form of continuous streams. It turns out that we are currently generating data much faster that we can provide storage capacity for it. This is not necessarily a problem, as not all data are equally important to retain. The challenge is just how to find out what is the important fraction. This selective data processing on continuous data streams is the topic of my thesis.

So, which objective did you pursue with your research?
Well, I was very keen to find out how data processing can be offloaded from traditional information systems so that the user does not need to worry about it. I looked at two different scenarios. In the first one, data processing is offloaded to the data source, to the sensors of a wireless sensor network. The idea here is to prevent the generation of irrelevant data right at the source. The challenge is to come up with a way to push almost arbitrarily processing onto the resource constrained senor nodes in an energy-efficient manner.

The second scenario investigates a way of how to offload processing into the data path to a traditional information system to prevent overloading. The example I use here is a bank that it is no longer able to keep up with the processing of the data volume from a stock exchange. I worked on a hardware solution that is based on field-programmable gate arrays, called FPGAs. They are re-programmable logic chips that can implement any digital circuit. The goal was to see to which extent it is possible to build an information processing system without a traditional processor. The thesis then presents a way to translate queries into digital circuits on a chip.

With a certain distance now and retrospectively, where do see the "thrill" about the research for your thesis?
The exciting aspect was that I could work at different levels in the system stack. It started at a high abstraction level during design of algorithms for the query optimization problem. The work also touched the operating system and network stack. Finally, for the FPGA backend, it went all the way down to hardware gates on a chip. Of course, each level has its own complexity. However, once one takes control on different levels in a system, interesting cross-layer optimization becomes possible. As an example, it gives you far more control on energy and resource aspects in sensor networks. Also, during the research for the thesis I had the chance to broaden my knowledge in various disciplines and to deal with different areas like databases, wireless networks and VLSI hardware design. At the end of the day, I think, the most exciting side of doing a systems PhD is simply, if all goes well, to have a working system.

Ten years ago, Andy Grove, the former chairman of Intel, told Business Week magazine: "Brains don't speed up. You can reach people around the clock, but they won't think any better or any faster just because you have reached them faster." Now, ten years later, does your particular sort of data processing solution help empowering the knowledge workers?
Not necessarily directly. It will not get us thinking faster, but it might get us answering our emails in less time. I believe that the point Andy Grove also made was that IT has to help in reducing overhead. In daily business we need to have the relevant information available immediately to make decisions. We need to know the current state of the market or our supply chain. That's why we add more and more sensors or build information systems that closely observe the behavior of the financial market. Again, the difficulty then lies in quickly finding the relevant pieces of information. For that the offloading or selective data acquisition approaches discussed in my thesis could help. It can cut overhead by getting the data faster, or more importantly, spare us from dealing with irrelevant information.

Currently, the City of Zurich runs the eZurich campaign. The initiative is modeled on Silicon Valley in the United States. René, with which visionary ideas would you personally contribute to transform Zurich into one of the world's great technology centers?
That is a tough question. But I think the results of the FPGA work could be used to build a low-latency solution for automated financial trading in the stock market. There is a significant interest from the financial sector in accelerating certain operations. My current employer is engaged in different projects along this line. Putting the risks of such an endeavor aside the creation of an ETH startup and a close collaboration with a banking institute may be an approach. ETH can further contribute to Zurich as a technology center by a more direct technology transfer. An example of such a knowledge transfer to the industry already happens inside the computer science department. The professors of Systems Group founded the Enterprise Computing Center. That center tightly collaborates with a small number of companies. The current interactions include help in architectural decisions for IT systems up to proof-of-concept prototypes. The gap that might be filled by Zurich IT companies is to turn these ideas into solutions that are suitable for the market.

René, thank you very much indeed, and don't forget to celebrate your success.