Skip to Main content Skip to Navigation
New interface

Prédiction et estimation de très faibles taux d'erreur pour des chaînes de communication codées.

Abstract : The time taken by standard Monte Carlo (MC) simulation to calculate the Frame Error Rate (FER) increases exponentially with the increase in Signal-to-Noise Ratio (SNR). Importance Sampling (IS) is one of the most successful techniques used to reduce the simulation time. In this thesis, we investigate an advanced version of IS, called Adap- tive Importance Sampling (AIS) algorithm to efficiently evaluate the performance of Forward Error Correcting (FEC) codes at very low error rates. First we present the inspirations and motivations behind this work by analyz- ing different approaches currently in use, putting an emphasis on methods inspired by Statistical Physics. Then, based on this qualitative analysis, we present an optimized method namely Fast Flat Histogram (FFH) method, for the performance evaluation of FEC codes which is generic in nature. FFH method employs Wang Landau algorithm and is based on Markov Chain Monte Carlo (MCMC). It operates in an AIS framework and gives a good simulation gain. Sufficient statistical accuracy is ensured through dif- ferent parameters. Extention to other types of error correcting codes is straight forward. We present the results for LDPC codes and turbo codes with different code- lengths and rates showing that the FFH method is generic and is applicable for different families of FEC codes having any length, rate and structure. Moreover, we show that the FFH method is a powerful tool to tease out the pseudocodewords at high SNR region using Belief Propagation as the decoding algorithm for the LDPC codes.
Complete list of metadata

Cited literature [146 references]  Display  Hide  Download
Contributor : David Declercq Connect in order to contact the contributor
Submitted on : Wednesday, May 1, 2013 - 10:02:40 AM
Last modification on : Friday, August 5, 2022 - 2:46:00 PM
Long-term archiving on: : Friday, August 2, 2013 - 4:04:57 AM


  • HAL Id : tel-00819416, version 1


Shahkar Kakakhail. Prédiction et estimation de très faibles taux d'erreur pour des chaînes de communication codées.. Théorie de l'information [cs.IT]. Université de Cergy Pontoise, 2010. Français. ⟨NNT : ⟩. ⟨tel-00819416⟩



Record views


Files downloads