Recherche et analyse de codes convolutionnels doublement orthogonaux simplifies au sens large.

摘要

This thesis presents a new class of error correcting codes called Simplified-Convolutional Self-Doubly Orthogonal codes (S-CSO2C). These codes have the property that they can be easily decoded with an iterative threshold decoder without the need of interleavers.; Based on the algebraic properties of Convolutional Self-Doubly Orthogonal codes (CSO2C), S-CSO2C are codes for which we have reduced some conditions on the double orthogonality of the CSO 2C codes. This action allows a important reduction of the span of the codes, which in return leads to a reduction of the total decoding delay. A Punctured version of S-CSO2C's codes (S-PCSO2C) is also investigated. We observed that the error performances was not decreasing dramatically, just a few tenths of a decibel at low signal to noise ratio, when we use simplified codes instead of CSO2C codes. At high SNR, the error performances of simplified codes are the same as CSO 2C codes.; Finally, the use of S-CSO2C's codes offers a good tradeoff between the error performances degradation and the time delay reduction at the decoder. For that, this study was fruitful and allows an easy implementation of this error control coding technique into real time communications systems.