ERRORFLOORS

Advanced coding techniques for mitigating error floors

 Partecipanti

Mappa

+-

Leaflet | Map data © OpenStreetMap contributors, CC-BY-SA, Imagery © Mapbox

 Word cloud

Esplora la "nuvola delle parole (Word Cloud) per avere un'idea di massima del progetto.

 Obiettivo del progetto (Objective)

'Iterative decoding algorithms for low-density parity-check (LDPC) codes are known to be subject to decoding failures that result in high error floors in the bit error rate performance curves. Recently, researchers became aware that the error floor was a result of a number of properties of the code's graph representation, including cycles, stopping sets, trapping sets, absorbing sets, and pseudo-codewords.

This research aims at analyzing the error floor phenomenon for these structures and developing improved code representations as well as decoder architectures to mitigate the error floor. The convolutional counterparts of LDPC codes, namely LDPC convolutional codes will also be examined in this scope. It is proposed to extend the results to the decoding of conventional codes with denser graph representations, where the error floor issue is even more severe.'

Introduzione (Teaser)

An EU team addressed performance issues affecting a common method of error control coding. They examined the circumstances of decoding failure, developed new stopping rules for decoders and devised a more realistic communication model.

Descrizione progetto (Article)

Error control coding is the branch of digital communications concerning reliable transmission of information over noisy channels, through efficient use of power and bandwidth resources. The efficiency is achieved by adding data redundancy at transmission, though the most popular such method (low-density parity-check (LDPC)) reaches a performance ceiling.

The limitations are called the error floor. The EU-funded ERRORFLOORS (Advanced coding techniques for mitigating error floors) project aimed to study the phenomenon and find ways of improving the performance of iteratively decodable codes. The undertaking ran through Turkey's Bogazici University from October 2010 to September 2014.

Consortium members began with a rigorous analysis of the code structure to understand how an error floor arises. The team analysed error-prone structures in graphical representations of LDPC codes with regard to how decoding fails. Further asymptotic work investigated how properties of code families develop as code becomes longer.

Additionally, the group designed novel schedules and stopping rules for decoders. The work maximised processing power efficiency, reducing the number of operations without significant performance loss.

The team also developed a communication model unlike the simpler versions of other groups. The project used a more realistic convoluted interleaver to break up correlations, while minimising the delay.

ERRORFLOORS contributed new understanding about the successes and failures of iterative decoding. The result was new communications standards employing error control coding technologies.