LDPC performance: A rigorous evaluation

Hadjer ZAAF and Merouane

Bouzid
Electrical Engineering Faculty, P.O. Box 32, El-Alia, Bab-Ezzouar,
Speech Communication and University of Sciences and Technology Algiers, Algeria
Signal Processing Laboratory, Houari Boumediene (USTHB), [email protected],
[email protected],

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Abstract— method of selecting the flipped bits, achieving different
degrees of bit error rate (BER) and convergence rate
Linear error-correcting codes, particularly Low-Density performance enhancements at the cost of higher complexity.
Parity-Check (LDPC) codes, are crucial for guaranteeing
Figure 1 illustrates the general classification of hard
reliable communication over noisy channels. LDPC codes
achieve this by employing a sparse parity-check matrix, where decision decoding approaches available for LDPC codes.
each row and column contains only a limited number of non-
zero elements. This sparsity contributes to their exceptional
performance, allowing them to theoretically approach the
Shannon limit – the fundamental capacity for error-free II. PERFORMANCE ANALYSIS OF LDPC
communication in the presence of noise. Notably, a key
advantage of LDPC codes lies in their ability to deliver A. LDPC encoding
superior bit-error-rate (BER) performance across diverse The encoding of LDPC is similar to that of any linear
channel conditions.This article provides a comprehensive block codes for which a generator matrix (G) and parity
overview of well-established LDPC hard-decision decoding check matrix (H) are defined. In order to achieve a
algorithms. Additionally, it presents simulations comparing the systematic LDPC code, G must be in the following form G [I
performance of various LDPC decoding algorithms. P]k Where I k is an identity matrix and P defines the parity
bits. In some cases, a code may be specified by only the H
Keywords— Codes LDPC, systèmes de communication sans fil,
matrix and it becomes necessary to solve for the G matrix.
codage canal, encodage, décodage, transmission, sans et avec
The H matrix is often in an arbitrary format, it must be
codage.
converted into canonical form [ ] n k .H =[PT I ] ,where I
n−k is an identity matrix and defines the parity bits .
I. INTRODUCTION (HEADING 1) Typically, encoding consists of using the G matrix to
This template, modified in MS Word 2007 and saved as a compute the parity bits and decoding consists of using the H
“Word 97-2003 Document” for the PC, provides authors matrix and soft-decision decoding. In the encoding stage, the
with most of the formatting specifications needed for main task is identifying the fixed bits position. As in the
preparing electronic versions of their papers. All standard systematic LDPC codes, the value of the transmission
paper components have been specified for three reasons: (1) codeword is the same with the value of the H matrix‟s
ease of use when formatting individual papers, (2) automatic message word.
compliance to electronic requirements that facilitate the
concurrent or later production of electronic products, and (3) B. LDPC decoding
conformity of style throughout a conference proceedings. Soft-decision, hard-decision and hybrid decoding
Margins, column widths, line spacing, and type styles are algorithms have been proposed for decoding LDPC codes.
built-in; examples of the type styles are provided throughout The decoding algorithms proposed by Gallager had two
this document and are identified in italic type, within common features. Firstly, based on the observed channel
parentheses, following the example. Some components, such output, these algorithms tried to iteratively find the codeword
as multi-leveled equations, graphics, and tables are not that was sent over the channel. Secondly, these algorithms
prescribed, although the various table text styles are operated locally in the sense that they combined partial
provided. The formatter will need to create these information that could then be used in other partial-
components, incorporating the applicable criteria that follow. information combining.
The LDPC codes were first introduced by Gallager in his 1) Bit flipping algorithm
PhD thesis in 1960[15]. An LDPC code is a linear error- Le Bit-Flipping, ou basculement de bits, est une technique
correcting code that has a parity check matrix H with small de décodage employée pour les codes LDPC à entrée et
number of nonzero elements in each row and column. sortie ferme ou dure (hard). Cette méthode repose sur le
Among various coding schemes, Low-Density Parity-Check principe de décisions binaires fermes pour chaque bit reçu,
(LDPC) codes are a special class of linear block codes effectuées par le détecteur et transmises au décodeur. Dans
which has become popular in recent years due to its versatile le cadre de l'algorithme BF, les messages transmis le long
error correcting characteristics [7] and near Shannon limit des branches du graphe de Tanner sont également binaires.
performance. Basically, according to their error correcting Un nœud de variable envoie un message indiquant s'il s'agit
mechanisms, decoding algorithms for LDPC codes can be d'un "1" ou d'un "0", tandis que chaque nœud de parité
categorized into two main classes, i.e., soft-decision [12] envoie un message à chaque nœud de variable connecté,
and hard-decision [13] based approaches.Among the hard précisant sa décision sur la valeur de ce bit (en fonction des
decision algorithms, bit-flipping (BF) algorithms flip one or informations disponibles au niveau du nœud de parité). Le
a group of bits based on the values of the flipping functions nœud de parité peut également transmettre un message
(FFs) computed in each iteration. The FF associated with a indiquant si son équation de contrôle de parité est satisfaite.

reçu 𝑈 en examinant les résultats des équations de parité.

variable node (VN) is a reliability metric of the L'algorithme de décodage BF corrige les bits d'un vecteur

Au début, il calcule le syndrome 𝑆 du vecteur reçu. Si le

corresponding bit decision and depends on the binary-
valued checksums of the VN’s connected check nodes
(CNs). Although BF algorithms are much simpler than the syndrome est nul, l’algorithme s’arrête en considérant que le
SPA, their performance is far from optimal. To reduce the vecteur reçu est correct et par conséquent, il le décode par le
performance gap, many variants of Gallager’s BF message correspondant . Dans le cas contraire (𝑆 ≠ 0),
algorithms have been proposed. Most of them tried to l'algorithme bascule les bits reçus erronés qui causent le plus
improve the VN’s reliability metric (the FF) and/or the grand nombre d'erreurs de parité. Ensuite, le processus se
répète jusqu'à ce qu'un syndrome nul soit obtenu (indiquant Save As command, and use the naming convention
un mot de code valide pouvant être décodé en un message) prescribed by your conference for the name of your paper. In
ou un nombre d'itérations prédéterminé est atteint sans this newly created file, highlight all of the contents and
obtenir un syndrome nul, l'algorithme s’arrête alors en import your prepared text file. You are now ready to style
signalant une erreur de décodage. your paper; use the scroll down window on the left of the MS

Bit-flipping (BF) decoding algorithm is a hard-decision [1] G. Eason, B. Noble, and I. N. Sneddon, “On certain integrals of
decoding algorithm which is much simpler than the soft Lipschitz-Hankel type involving products of Bessel functions,” Phil.
Trans. Roy. Soc. London, vol. A247, pp. 529–551, April 1955.
decision decoding techniques like the Sum-Product (references)
Algorithm (SPA) or its modifications but does not perform [2] J. Clerk Maxwell, A Treatise on Electricity and Magnetism, 3rd ed.,
as well. To reduce the performance gap between SPA and vol. 2. Oxford: Clarendon, 1892, pp.68–73.
BF based decoders, variants of the latter such as weighted [3] I. S. Jacobs and C. P. Bean, “Fine particles, thin films and exchange
bit-flipping (WBF) and improved modified bit-flipping anisotropy,” in Magnetism, vol. III, G. T. Rado and H. Suhl, Eds.
(IMWBF) algorithms have been proposed. They provide New York: Academic, 1963, pp. 271–350.
tradeoffs between computational complexity and error [4] K. Elissa, “Title of paper if known,” unpublished.
performance. We denote a regular binary LDPC [5] R. Nicole, “Title of paper with only first word capitalized,” J. Name
Stand. Abbrev., in press.

Check Node (CN) degree dc by (𝑁, 𝐾)(𝑑𝑣, 𝑑𝑐).

code with Variable Node (VN) degree dv and
[6] Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron

The code is the null space of 𝑀 × 𝑁 parity

spectroscopy studies on magneto-optical media and plastic substrate

check matrix 𝑯 = [𝐻𝑚𝑛]. Let 𝒖 be a codeword

interface,” IEEE Transl. J. Magn. Japan, vol. 2, pp. 740–741, August
1987 [Digests 9th Annual Conf. Magnetics Japan, p. 301, 1982].

(BPSK) modulation is used so that codeword 𝒖

and assume that Binary Phase Shift Keying [7] M. Young, The Technical Writer’s Handbook. Mill Valley, CA:
University Science, 1989.

= [𝑢1, 𝑢2, ⋯ , 𝑢𝑁] is mapped into a bipolar

sequence 𝒙 = [𝑥1, 𝑥2, ⋯ , 𝑥𝑁 ]. This is sent
through an Additive White Gaussian Noise

density of 𝑁0⁄2 W/Hz. Let 𝒚 = [𝑦1, 𝑦2,⋯, 𝑦𝑁]

(AWGN) channel with two-sided power spectral

be the soft channel values obtained at the

sequence 𝒛 = [𝑧1, 𝑧2, ⋯ , 𝑧𝑁] is obtained by

receiver’s coherent matched filer output. The

taking hard decision on the components of 𝒚.

Let 𝒖̂ = [𝑢̂1, 𝑢̂2, ⋯ , 𝑢̂𝑁] be the decoded binary

𝒔=[𝑠1, 𝑠2, ⋯ , 𝑠𝑀] is computed by 𝒔 =

sequence. The syndrome vector

𝒖̂𝑯𝑇(𝑚𝑜𝑑2). Also, we denote the nth VN by 𝑣𝑛

and mth CN by 𝑐𝑚. Let ℳ(𝑛) denote the set of
indices of the neighboring CNs of 𝑣𝑛 and 𝒩(𝑚)

of 𝑐𝑚
denote the set of indices of the neighboring VNs

III. SIMULATION RESULTS

La Figure 3.1 présente le schéma bloc de notre modèle de
simulation utilisé pour évaluer les performances des codes

génère 100 blocs de données binaires de taille 𝑘. Ces blocs

LDPC. Il est composé d’un générateur pseudo aléatoire qui

sont ensuite codés par un codeur LDPC de rendement R=1/2,

qu'il soit régulier ou irrégulier, puis modulés et transmis via
un canal AWGN. Après démodulation et décodage LDPC
par l’algorithme BF, une évaluation de la performance de
transmission en termes du taux d'erreur binaire TEB est
effectuée en comparant la séquence des bits décodés à la
séquence des bits transmis initialement. Notons que dans les
figures des évaluations de performance, nous ajoutons
toujours la courbe non-codée pour les comparaisons, c-à-d, la
courbe de performance sans codage de canal.

IV. CONCLUSION
After the text edit has been completed, the paper is ready
for the template. Duplicate the template file by using the