-------------------------------------------------------------------------
Speaker:	Prof. Oyvind Ytrehus
		Department of Informatics
		University of Bergen, Norway

Topic:		"On the error floors and weight distributions of
		 turbo codes"

Date:		Friday, 4 October 2002

Time:		2:30pm - 4:00pm

Venue:		Room 2404
		(Phase I, via lift nos. 17/18)
		HKUST

ABSTRACT:

Turbo codes were introduced in 1993 by Berrou et. al. These codes provide
error rates low enough to be of practical interest at signal-to-noise
ratios close to the Shannon bound. However,  upper bounds show that the
minimum distance of these codes is in general rather poor. Hence, in the
performance curves (often obtained by simulation) of simple turbo codes,
an error floor dominates the high signal-to-noise ratio range. In this
region the error rate after decoding is disappointingly high, compared to
the impressive performance at lower signal-to-noise ratios. For more
powerful turbo codes the error floor is still known to be present, but it
may be extremely time consuming to determine the exact position and
steepness of it. Under an assumption of maximum likelihood decoding, the
exact position and steepness of the error floor depends on the exact
minimum distance and the lower terms of the code's weight distribution.

We discuss algorithms for determining the lower terms of the weight
distribution of a turbo code. Several improvements of the recently
introduced algorithm by Garello et al. are outlined. The techniques
presented in this paper improve the observed asymptotic complexity by a
factor proportional to the information length. As an example, the improved
algorithm is applied to the determination of the minimum distance of the
turbo codes for the UMTS/3rd generation mobile/cellular phone. Further, we
apply the improved algorithm to high rate turbo codes using high rate
non-punctured constituent codes. To reduce complexity, the constituent
codes are represented by an information bit oriented trellis.



***********************
Biography:

Oyvind Ytrehus is a professor at  the Department of Informatics,
University  of Bergen, where he also received the Dr. Scient. Degree in
1989. His research interests include error correcting properties and
decoding complexity of algebraic codes, convolutional codes, turbo codes,
and codes based on graphs; applications of coding theory in communication
and storage; and the interaction between coding theory and cryptology.


		For enquiry, please call 2358 7008
		  **** ALL are Welcome ****
-------------------------------------------------------------------------