Abstract :
A unified view of iterative receivers and decoders can be obtained by using "factor graphs". These provide an exceedingly versatile tool: they allow one to model and design iterative receivers, to categorize in a simple and natural way the approximations simplifying them, and to investigate their convergence properties. Originally developed for decoding turbo codes, it was recognized that factor graphs are a natural setting for the description of a number of general iterative techniques for detecting coded signals transmitted on a variety of channels. In addition, they provide a unified framework allowing one to understand the connections among seemingly different detection problems. This talk summarizes the application of normal factor graphs to a number of these problems, such as turbo and LDPC decoding, equalization of coded signals, multiuser detection, decoding of multilevel coded modulation, and reception of space-time coded signals in MIMO systems.
