Iteratively-Decoded Erasure-Correcting Coding for Distributed Storage

Vous pouvez (re-)voir l'exposé "Iteratively-Decoded Erasure-Correcting Coding for Distributed Storage", présenté par Iryna Andryianova le 16 mars 2016 dans le cadre du séminaire "Nexus entre les théories de l’information et du calcul" de l'Institut Henri-Poincaré Paris.


This talk will present some constructions of iteratively-decoded sparse-graph codes over various erasure channel models, coming from distributed storage systems. Although the state-of-the-art of coding for distributed storage is built on short algebraic block codes, there have been several attempts to use sparse-graph codes, with the aim to improve the decoding complexity and the scalability of the storage system in whole. This talk will introduce the existing code constructions and will discuss the use of graph-based codes in the framework of distributed storage.