Improving the Decoding of LDPC Codes for the Packet Erasure Channel with a Hybrid Zyablov Iterative Decoding/Gaussian Elimination Scheme

نویسنده

  • Mathieu CUNCHE
چکیده

This work focuses on the decoding algorithm of the LDPC large block FEC codes for the packet erasure channel, also called AL-FEC (Application-Level Forward Error Correction). More specifically this work details the design and the performance of a hybrid decoding scheme, that starts with the Zyablov iterative decoding algorithm, a rapid but suboptimal algorithm in terms of erasure recovery capabilities, and, when required, continues with a Gaussian elimination algorithm. For practical reasons this work focuses on two LDPC codes for the erasure channel, namely LDPC-staircase and LDPC-triangle codes. Nevertheless the decoding scheme proposed can be used with other LDPC codes without any problem. The performance experiments carried out show that the erasure recovery capabilities of LDPC-triangle codes are now extremely close to that of an ideal code, even with small block sizes. This is all the more true with small code rates: whereas the Zyablov iterative decoding scheme becomes unusable as the code rate decreases, the Gaussian elimination makes the LDPC-triangle codes almost ideal. In all the tests, when carefully implemented, the LDPC-triangle codec featuring the proposed decoding scheme is fast, and in particular always significantly faster than the reference Reed-Solomon on GF(2) codec. The erasure recovery capabilities of LDPC-staircase codes are also significantly improved, even if they remain a little bit farther from an ideal code. Nevertheless, a great advantage is the fact that LDPC-staircase codes remain significantly faster than LDPC-triangle codes, which, for instance, enables their use with larger blocks. All these results make these codes extremely attractive for many situations and contradict the common belief that using Gaussian elimination is not usable because of a prohibitive processing load. Moreover the proposed approach offers an important flexibility in practice, and depending on the situation, one can either choose to favor erasure recovery capabilities or the processing time. Key-words: FEC codes for the packet erasure channel, AL-FEC codes, LDPC codes, LDPC-staircase codes, LDPC-triangle codes, Gaussian elimination, Zyablov iterative decoding This work is supported by the ANR/RNRT 2006 contract number 06TCOM01901. in ria -0 02 63 68 2, v er si on 2 14 M ar 2 00 8 Amélioration du Décodage de Codes LDPC pour le Canal à Effacement de Paquets avec une Approche Hybride Décodage Itératif de Zyablov/Pivot de Gauss Résumé : Ce travail aborde les algorithmes de décodage des codes FEC de type grand bloc pour le canal à effacement de paquets, aussi appelés AL-FEC (Application-Level Forward Error Correction). Plus précisement, ce travail détaille la conception et les performances d’une approche de décodage hybride, qui débute avec l’algorithme de décodage itératif de Zyablov, un algorithme rapide mais sous-optimal en terme de capacités de correction d’erreurs, et se poursuit, lorsque ceci est nécessaire, avec un pivot de Gauss. Pour des raisons pratiques ce travail considère deux codes AL-FEC pour le canal à effacement de paquets, les codes LDPC-staircase et LDPC-triangle. Cependant le schéma de décodage proposé peut s’appliquer à d’autres codes LDPC sans aucun problème. Les tests éffectués montrent que les codes LDPC-triangle atteignent des performances en terme de capacité de correction d’effacements extrêmement proches de celles d’un code idéal, ceci même avec de petites tailles de blocs. Ceci est encore plus vrai avec de petits code rates: alors que le décodage itératif de Zyablov devient inutilisable au fur et à mesure que le code rate diminue, le pivot de Gauss rend les codes LDPC-triangle quasiment optimaux. Dans tous les tests, lorsqu’il est soigneusement implémenté, le codec LDPC-triangle doté du schéma de décodage proposé est rapide, et en particulier toujours significativement plus rapide que le codec Reed-Solomon sur GF(2) de référence. Les capacités des correction d’effacements des codes LDPC-staircase sont également significativement améliorées, même si elles restent un peu plus éloignées d’un code idéal. Cependant un gros avantage est le fait que les codes LDPC-staircase sont significativement plus rapides que les codes LDPC-triangle, ce qui permet, par exemple, de les utiliser avec des tailles de blocs plus importantes. Tous ces résultats rendent ces codes extrêmement attractifs pour de nombreuses situations et contredisent une idée largement répendue selon laquelle le pivot de Gauss ne serait pas utilisable du fait de coûts de calcul prohibitifs. De plus l’approche proposée offre une très grande flexibilité à l’usage, et suivant la situation, on pourra choisir de privilégier soit les capacités de correction, soit le temps de calcul. Mots-clés : Codes FEC pour canaux à effacement de paquets, codes AL-FEC, codes LDPC, codes LDPCstaircase, codes LDPC-triangle, Pivot de Gauss, décodage itératif de Zyablov in ria -0 02 63 68 2, v er si on 2 14 M ar 2 00 8 Improving the Decoding of LDPC Codes with a Zyablov Iterative Decoding/Gaussian Elimination Scheme 3

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Search Based Weighted Multi-Bit Flipping Algorithm for High-Performance Low-Complexity Decoding of LDPC Codes

In this paper, two new hybrid algorithms are proposed for decoding Low Density Parity Check (LDPC) codes. Original version of the proposed algorithms named Search Based Weighted Multi Bit Flipping (SWMBF). The main idea of these algorithms is flipping variable multi bits in each iteration, change in which leads to the syndrome vector with least hamming weight. To achieve this, the proposed algo...

متن کامل

Search Based Weighted Multi-Bit Flipping Algorithm for High-Performance Low-Complexity Decoding of LDPC Codes

In this paper, two new hybrid algorithms are proposed for decoding Low Density Parity Check (LDPC) codes. Original version of the proposed algorithms named Search Based Weighted Multi Bit Flipping (SWMBF). The main idea of these algorithms is flipping variable multi bits in each iteration, change in which leads to the syndrome vector with least hamming weight. To achieve this, the proposed algo...

متن کامل

On the erasure-correcting capabilities of low-complexity decoded LDPC codes with constituent Hamming codes

Low-density parity-check (LDPC) codes can be constructed using constituent block codes other than single parity-check (SPC) codes. This paper considers random LDPC codes with constituent Hamming codes and investigates their asymptotic performance over the binary erasure channel. It is shown that there exist Hamming code-based LDPC codes which, when decoded with a low-complexity iterative algori...

متن کامل

Achieving a near-optimum erasure correction performance with low-complexity LDPC codes

SUMMARY Low-density parity-check (LDPC) codes are shown to tightly approach the performance of idealized maximum distance separable (MDS) codes over memoryless erasure channels, under maximum likelihood (ML) decoding. This is possible down to low error rates and even for small and moderate block sizes. The decoding complexity of ML decoding is kept low thanks to a class of decoding algorithms, ...

متن کامل

Adaptive message-passing decoding: ways to implement and limits of applicability

I n this report we study the decoding algorithms for a family of codes called LDPC codes used over the binary erasure channel (BEC). LDPC codes are able to work at rates near Shannon capacity with linear time decoding algorithms. Those linear time decoding algorithms sometimes fail even when Gaussian elimination would still be able to restore the codeword. In this report we try to experimentall...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2008