Decodage des codes turbo par l'algorithme MAP: Simplifications et applications aux modulations multi-niveaux (French text)

Error control coding aims to reduce noise and interference affecting the received signals in digital communication systems. Among all channel coding techniques, Turbo coding, introduced by Berrou in 1993, achieves amazing error performances close to the Shannon limit. Nevertheless, these very powerful codes suffer from high decoding complexity and long delay.; This thesis mainly concerns the analysis of the MAP algorithm, one of the most important part of the iterative Turbo decoding. Our goal is to reduce the drawbacks previously cited. We study new variants of the MAP algorithm derived from those known as Log-MAP or Max-Log-MAP. The introduction of thresholds and the use of sliding windows in the decoding process are notably proposed. These new algorithms are discussed and their performance are evaluated by simulation in conjunction with Turbo codes.; Turbo coding schemes, usually combined with BPSK modulation, don't offer the best possible spectral efficiency. A first solution to decrease the bandwidth consumption may be to use puncturing patterns to obtain high coding rates. However we prefer to introduce multilevel modulations such as M-QAM or M-PSK which offer various possibilities. We study the state of the art concerning Turbo coding schemes using high level modulation. Some schemes using 16-QAM and & PSK modulations are proposed. The resulting bit error rate performance are analyzed and the different developed schemes are compared.