TABLE OF CONTENTS The turbocode principle was introduced by C. Berrou during the ICC Geneva congress in 1993 [BER 93] where, for the first time, an error correcting code operating within less than 0,5 dB of the Shannon limit [SHA 48] was announced. These results have initially surprised all the specialists in the field who were persuaded that it was not possible to reach this level of performance with a reasonable complexity. Very quickly many researchers, such as Hagenauer, Benedetto, Divsalar [HAG 96, BEN 96, DIV 95, ROB 94, WIB 95] and a number of others confirmed the results of Berrou and within a few years the turbocode became essential in the field of the error corrector coding as the 21 st century solution.
The principle described by Berrou consists of carrying out iterative decoding of two CRS (convolutional recursive systematic) codes concatenated in parallel through a random or non-uniform interleaver. This iterative processing is based on SISO (soft input soft output) decoding and on the optimal transfer of the decoding information from one decoder to the next. To that end he has introduced the concept of extrinsic information, which plays a fundamental part in the operation of the convolutional turbocode (CTC).
In view of the first results of Berrou it was obvious that it became possible to obtain performances comparable to block codes. To get to that point several problems had to be solved:
which type of concatenation should be adopted?
how should a SISO...
