ED Sciences Physiques et de l'Ingénieur
Distance properties of polar codes: theory and applications
by Malek ELLOUZE (Laboratoire de l'Intégration du Matériau au Système)
The defense will take place at 10h15 - Amphi Jean Paul Dom Laboratoire IMS 351 cours de la libération 33405 Talence CEDEX
in front of the jury composed of
- Christophe JEGO - Professeur des universités - Université de Bordeaux - Directeur de these
- Charly POULLIAT - Professeur des universités - Université de Toulouse - CoDirecteur de these
- Philippe MARY - Professeur des universités - INSA RENNES - Rapporteur
- Charbel ABDEL NOUR - Professeur - IMT ATLANTIQUE - Rapporteur
- Iryna ANDRIYANOVA - Professeure des universités - Université Paris Cergy - Examinateur
- Stephan TEN BRINK - Full professor - University of Stuttgart - Examinateur
Error-correcting codes are essential for ensuring reliable data transmission, especially in contexts where various interferences may compromise data integrity. Polar codes are one of the most competitive families of error-correcting codes. They can achieve Shannon channel capacity through efficient encoding and decoding for very large code lengths. For these reasons, polar codes have been included in the 5G standard. Additionally, they are the subject of several research efforts for the future 6G standard. However, polar codes, as originally designed for successive cancellation (SC) decoding, exhibit limited performance for moderate code lengths. This is in part due to their weak distance properties and partly to the nature of hard decision decoding. However, with the use of mainly list decoding and several other enhancements, including pre-transformation, polar codes are now competitive with LDPC and turbo codes. In this context, this thesis aims to study and analyze polar codes focusing on two fundamental aspects that influence their performance: their distance properties and their performance for list decoding. After a comprehensive review of polar code definition, various variants, decoding algorithms, and concepts related to their distance spectrum, a first contribution characterizes some distance properties of classical and pre-transformed polar codes. This method has the advantage of being entirely independent of code construction, making it applicable to different configurations. Moreover, the proposed approach distinguishes itself by having lower computational complexity than methods in the existing literature. Polar code puncturing and shortening techniques are introduced as variants to obtain polar codes whose sizes are not necessarily powers of two. A second contribution involves generalizing the developed approach within the thesis to punctured and shortened polar codes. It is noteworthy that this approach can be applied regardless of the puncturing and/or shortening technique used. Finally, the question of the list size necessary for list decoding (SCL) to achieve maximum likelihood performance is addressed. Since this depends on code construction, an algorithm is proposed to estimate the average list size required to achieve the best decoding performance. This constitutes a very useful contribution for constructing codes that offer a compromise between distance properties and list decoding with controlled computational complexity.
study of the physics of shock ignition
by Diego VIALA (Centre Lasers Intenses et Applications)
The defense will take place at 16h00 - Bâtiment A29 amphi B Université de Bordeaux 351 cours de la Libération 33405 Talence
in front of the jury composed of
- Dimitri BATANI - Professeur des universités - Université de Bordeaux - Directeur de these
- Sébastien LE PAPE - Directeur d'études - Laboratoire pour l'Utilisation des Lasers Intenses - Rapporteur
- Valeri GONCHAROV - Directeur de recherche - Laboratory for Laser Energetics - Rapporteur
- Benoît CANAUD - Directeur de recherche - CEA/DIF - Examinateur
- Javier HONRUBIA - Professeur - Universidad Politécnica de Madrid - Examinateur
- João Jorge SANTOS - Professeur des universités - Université de Bordeaux - Examinateur
A decade of experiments at the National Ignition Facility has proven that inertial confinement fusion is a credible approach to energy production, with experiments having exceeded the ignition regime. However, the indirect-drive approach is not suited for high gain implosions and reliable energy production. The direct-drive ignition approach is favoured for energy production as it features simpler target designs and couples more energy to them. There are currently no ignition-scale laser facilities configured for the standard direct-drive approach. Integrated direct-drive experiments have mostly been focused on understanding the physics at reduced scales, with the ultimate goal of demonstration of necessity and feasibility of construction of an international direct-drive laser facility. This thesis manuscript presents a study on the validation of state-of-the-art 3D radiative hydrodynamics codes and the understanding of low modes and laser coupling which play crucial roles in the study of inertial fusion energy. Careful examination of CBET models is of paramount importance in this context, ensuring the accuracy of simulations and contributing to the design of future direct-drive facilities. In addition, the investigation of laser homogeneity on target is imperative to understand its overall impact on the system.