Novel turbo-equalization techniques for coded digital transmission

by Dejonghe, Antoine

Abstract (Summary)
Turbo-codes have attracted an explosion of interest since their discovery in 1993: for the first time, the gap with the limits predicted by information and coding theory was on the way to be bridged. The astonishing performance of turbo-codes relies on two major concepts: code concatenation so as to build a powerful global code, and iterative decoding in order to efficiently approximate the optimal decoding process. As a matter of fact, the techniques involved in turbo coding and in the associated iterative decoding strategy can be generalized to other problems frequently encountered in digital communications. This results in a so-called turbo principle. A famous application of the latter principle is the communication scheme referred to as turbo-equalization: when considering coded transmission over a frequency-selective channel, it enables to jointly and efficiently perform the equalization and decoding tasks required at the receiver. This leads by the way to significant performance improvement with regard to conventional disjoint approaches. In this context, the purpose of the present thesis is the derivation and the performance study of novel digital communication receivers, which perform iterative joint detection and decoding by means of the turbo principle. The binary turbo-equalization scheme is considered as a starting point, and improved in several ways, which are detailed throughout this work. Emphasis is always put on the performance analysis of the proposed communication systems, so as to reach insight about their behavior. Practical considerations are also taken into account, in order to provide realistic, tractable, and efficient solutions.
Bibliographical Information:


School:Université catholique de Louvain

School Location:Belgium

Source Type:Master's Thesis

Keywords:turbo principle iterative detection and decoding techniques digital communications


Date of Publication:12/10/2004

© 2009 All Rights Reserved.