Skip to Main content Skip to Navigation
Theses

Des codes correcteurs pour sécuriser l'information numérique

Abstract : Error-correcting codes are used to reconstitute digital data, which are proned to alterations during their storage and their transport. They can also be employed in cryptography as a tool to encrypt data and authenticate people. These different aspects are treated in this document. First of all, we study the class of cyclic codes defined by the zero set {1, 2^i+1, 2^j+1}, where i and j are distinct positive integers. We focus on the characterization of three-error correcting codes in this class as well as the weight distribution of these codes. We improve the Schaub algorithm, which gives a lower bound on the minimum distance of cyclic codes. We implement this algorithm to compute the spectral immunity of Boolean functions. This quantity is related with the minimum distance of cyclic codes and is important to guarantee the security of certain stream ciphers. Subsequently, we propose a solution to speed up the polynomial roots computation over finite fields of characterisitic two. This computation is the slowest step during the decoding of McEliece-type cryptosystems based on classical binary Goppa codes. We provide a complexity analysis of the underlying algorithm named BTZ. We complete our works by a study of low-cost authentication solutions based on the protocol HB, adopting a syndrome decoding approach, instead of the standard approach, based on the LPN problem.
Complete list of metadatas

https://tel.archives-ouvertes.fr/tel-00657110
Contributor : Vincent Herbert <>
Submitted on : Thursday, January 5, 2012 - 7:29:56 PM
Last modification on : Wednesday, December 9, 2020 - 3:12:28 PM
Long-term archiving on: : Monday, November 19, 2012 - 12:31:02 PM

Identifiers

  • HAL Id : tel-00657110, version 1

Citation

Vincent Herbert. Des codes correcteurs pour sécuriser l'information numérique. Théorie de l'information [cs.IT]. Université Pierre et Marie Curie - Paris VI, 2011. Français. ⟨tel-00657110⟩

Share

Metrics

Record views

823

Files downloads

8817