WORLDCOMP'07 Tutorial
![]() |
Factor Graphs for Advanced Algorithm Design in Wireless Communications
Dr. Henk Wymeersch , Massachusetts Institute of Technology, Cambridge, Massachusetts,USA Date: June 27, 2007 Time: 6:00 - 9:30 PM Location: TBA |
With the introduction of turbo codes in the early 1990s, a major leap forward was achieved in communications research. Not only were we able to finally develop practical capacity-approaching error-correcting codes, but we also managed to apply the general idea behind turbo codes (the so-called turbo principle) to a long list of applications. These include multi-user detection, channel estimation and MIMO systems. This has led to a myriad of algorithms, many of which are ad-hoc, using different terminology, different notations, different implementations, and so forth. Any kind of coherent underlying mechanism was unclear.
Only with the development of a type of graphical models, known as factor graphs, it has been recognized that there is in fact an underlying unifying theory for most, if not all, of these turbo algorithms. The factor graph framework is simple and elegant, and allows for a deep understanding of wide variety of algorithms, based on a set of simple rules and a common notation. These insights are developed in detail in the upcoming book Iterative Receiver Design, to be published by Cambridge University Press in the summer of 2007.
Objectives
The goal of this tutorial is to present the factor graph concept in a rigorous yet understandable manner, and show how advanced algorithms can be understood and designed in a systematic fashion using the sum-product algorithms. After this course the audience should be able to cast inference problems in their factor graph form, and recognize existing algorithms (e.g., decoding of LDPC codes, multi-user detection) as particular instances of the sum-product algorithm.
Intended Audience
This tutorial is aimed at graduate students, academic researchers and practitioners in the field. A basic knowledge of digital communications and probability theory is assumed.
Biography of Instructor
Henk Wymeersch is a postdoctoral associate with the Laboratory for Information and Decision Systems (LIDS) at the Massachusetts Institute of Technology (MIT). He obtained the Ph.D. degree in electrical engineering in 2005 from Ghent University, Belgium. In 2005-2006, Henk Wymeersch was a postdoctoral fellow of the Belgian American Educational Foundation at MIT, and in 2006 he won the Alcatel Bell Scientific Award for his Ph.D. thesis. He is a member of the IEEE, and author of the forthcoming book from Cambridge University Press Iterative Receiver Design. His research interests include algorithm design for wireless transmission, statistical inference and iterative processing. A list of his publications can be found at http://www.mit.edu/~hwymeers/.
