Â鶹´«Ã½Ó³»­

In Memoriam: Ralf Kötter
This winter, the information theory community lost one of its giants: Ralf Kötter passed away in Munich, Germany, in the early morning hours of February 2, 2009. During the past 15 years, his seminal contributions have transformed our field, and his work will surely remain a source of inspiration for years to come. Those of us who have had the privilege to know Ralf personally will cherish the memory of a towering intellect, with a big heart and a gentle soul.
Apr 30, 2009
koetterPortrait

Ralf Kötter was born in Königstein im Taunus, Germany, on OctoberÌý10, 1963. He received his Diploma in Electrical Engineering from the Technische Universität Darmstadt, and earned a Ph.D. in Electrical Engineering from Linköping University, Sweden, under the supervision of Thomas Ericson. After a brief sojourn at the IBM Almaden Research Center, Ralf joined the University of Illinois Urbana-Champaign, where he remained for the major part of his career, from July 1997 until December 2006. During the last two years of his life, Ralf was at the helm of research in information theory and communications in Germany, as Head of the Institute for Communications Engineering at the Technische Universität München (TUM).

Ralf received numerous awards for his work, including the Information Theory Society Paper Award (2004), the Vodafone Innovations Award (2008), the Best Paper Award from the Signal Processing Society (2008), and the ComSoc & IT Joint Paper Award (2009). He served as Associate Editor for the Â鶹´«Ã½Ó³»­ Transactions on Inform- ation Theory and the Â鶹´«Ã½Ó³»­ Transactions on Communications. He was Technical Program Co-Chair for the 2008 International Symposium on Information Theory, and twice Co-Editor-in-Chief for special issues of the Â鶹´«Ã½Ó³»­ Transactions on Information Theory (one on factor graphs and iterative decoding, and the other on networking and information theory). During 2003-2008, he served on the Board of Governors of the Â鶹´«Ã½Ó³»­ Information Theory Society, taking active part in several key committees and initiatives.

Goethe once said: what is demanded of genius, first and foremost, is love of truth. Ralf's love of truth was limitless. He had the ardent will and the innate ability to address every research problem he encountered at its most fundamental level. He always strove to understand the very heart of the problem, wherein he often found the truth and the beauty he was seeking. It is no coincidence that so many of Ralf's results are now counted among the most "beautiful" theorems in our field.

A good example is Ralf's work in network coding. He started on this track in earnest at the Information Theory Workshop in Metsovo, Greece, in June 1999. At the workshop, Raymond Yeung presented his paper on "Linear codes for network information flow," which later won the Information Theory Society Paper Award. It was a great result, and everyone in the audience was happy to accept it at face value. Everyone, that is, except Ralf. He went on musing, thinking, seeking the beauty and the truth.... These musings eventually led to his paper with Muriel Médard which not only established the algebraic foundations of the field but also brought the ideas of network coding much closer to practice. Indeed, the principle of randomized network coding , first developed in a series of papers by Ralf and his colleagues, critically relies on the underlying algebraic structure. The end result is an ingenious, completely decentralized, network-coding protocol, which bridges a fundamental gap between the theory of network coding and the practice of communication networks. The impact of network coding in general, and of Ralf's contributions to this field in particular, on the design, optimization, and operation of future high-performance networks is one of the most exciting areas of research in information theory today.

Ralf's early work, dating back to his student days at Linköping, was already a treasure-trove of wonders. One of the algorithms sketched out in his Ph.D. thesis is now known as the Kötter algorithm and universally recognized as the best approach to multivariate interpolation over finite fields. Much later, Ralf and I, in collaboration with several others, developed VLSI for this algorithm running at over 3.0 Gbps. According to Dick Blahut, Ralf's thesis also "made the well-known Forney formula obsolete," by showing how to compute error values on the fly during Berlekamp-Massey decoding. His 1995 paper with Wiberg and Loeliger developed the foundations for the theory of factor graphs . This paper still underlies much of the current research into the behavior of message-passing algorithms, and is one of the most cited references in this area. Ralf's short stint at the IBM Almaden Research Center led to key patents in the area of magnetic recording: the results of his work during that period are now implemented in IBM products. During the year and a half we spent working together at the University of Illinois, we produced half a dozen papers on a broad array of subjects, ranging from the theory of tail-biting trellises to signal-space characterization of iterative decoding. The papers I wrote with Ralf are among the best papers I ever wrote.

It would be futile to attempt recounting all of Ralf's contributions in the confines of this brief eulogy; a comprehensive tribute to his work is planned for a later date. Thus let me now fast-forward to the last few years of his life, a life that was cut so woefully short by cancer. Many have described Ralf's fight with cancer as brave and heroic. It was. It was also unyielding. While his body was giving up on him, the disease never got the upper hand in the battle with his spirit. The last major conference that Ralf attended was the Oberwolfach Kodierungstheorie Workshop in December 2007. He arrived there after a heavy dose of chemotherapy and confided to me, in private, that he felt really sick. While most people knew, nobody could tell. Ralf was his usual self: happily soaking up every bit of new research, going for long walks to discuss his latest ideas, eagerly taking part in lively conversations that often lasted into the morning hours. In fact, those few attendees that were blissfully unaware of his condition left Oberwolfach in the same state of blissful ignorance (I recall one of them asking Ralf on the last day why he decided to shave his head, a question that Ralf simply shrugged off).

I also recall visiting Ralf at the Freising hospital near Munich in January 2009. He could no longer walk. He had just undergone a terrible operation. His blood counts were dangerously low. Yet, what concerned Ralf the most that wintry day was that, for the first time ever, someone else was teaching his lectures at TUM. He spent most of the day on the phone with his assistant, making sure that the latter was well-prepared for the lecture. During that week in January 2009, we talked about so much.... Ralf was eager to catch up on the latest news in research. Which papers at the ISIT in Toronto turned out to be the most interesting? What was I working on? What was everyone else doing? There, at the hospital in Freising, Ralf and I started a new research project going: list-decoding of the Kötter-Kschischang codes on the operator channel. It was fun again!

I believe that information theory has inherent intellectual beauty that has always attracted the best and the brightest people. We have been particularly fortunate to have Ralf Kötter join our research community. His passion for information theory knew no bounds, his mighty intellect and wondrous spirit etched an everlasting stamp on our field. He has been a shining light, a light that was extinguished much too soon. We will miss you, Ralf.

Ìý

by Alexander Vardy — last modified Apr 30, 2009 02:06 AM