CHAPTER 1

INTRODUCTION

The success of a digital communication system is the ability to

channel is an integral part of the data transmission system which may be
distracted through noises. The channel can be coaxial cable, wireless, fiber
cable, twisted cable, etc.

In the receiver, the first process is the extraction of the message

from the information-bearing waveform created by the modulation. This is
handled by the corresponding demodulator. Next, the channel decoder corrects
the error that occurred in the transmission using the incorporated EDC
algorithm and provides the original message without parity bits. Source
decoder changes the output of the channel decoder which is a binary into the
signs of source information.

Figure 1.1 Block diagram of digital communication system

1.2 TYPES OF ERROR

Channel error is an error in a signal which arrives at the receiver

of a communication system. This is due to noise in the channel. The noise is
an undesirable signal that is added or affects the message bits transmitted over
the channel. It is random in nature and unpredictable, it doesn't have single
frequencies or constant amplitude. Noise affects the recovery of the message;
this noise error can be broadly classified as random error and burst error.

Random error is normally known as white Gaussian noise which

induces normally distributed random errors in the data. This arises in the
channel where noise corrupts the individual bits of the transmitted packet in
random nature. These are usually caused by thermal noise and the radiation
collected by the receiving antenna in the contact networks. Besides, the power
spectral density of the Gaussian noise at the receiver input is white in most
cases. The white colour is a combination of all colours equally likewise white
noise affects all the frequencies of the message equally. The error control
codes such as LDPC, BCH, etc. can handle random error effectively. Burst
Errors are the bit errors that take place simultaneously in more than two bits
positioned near to one another. These are caused in a storage device by fading
in a contact channel and multipath communication. Burst errors make large
blocks of data full of faults in channels due to catastrophic faults that are
highly problematic in the digital communication networks. In general, in the
face of disastrous burst errors, all coding schemes struggle to recreate a
message, and message data may be unrecognizable at the receiver. However,
by analysing the received data as a symbol, codes like Reed-Solomon can
detect the existence of the burst error.
 Reddy and Robinson (1972) pointed out that one coding
algorithm can control burst error less than or equal to N and by using another
coding method, it can correct up to M errors. When both algorithms are added
to the coding system, it can correct up to random error n < N and burst error
m < M. To address the channel for handling both errors simultaneously
product codes can be used. Blomqvist (2020) describes the error correction
using product codes can control various errors and can correct beyond half of
the minimum distance.

1.3 ERROR CONTROL TECHNIQUES

1.3.1 Automatic Repeat Request

Automatic Repeat Request (ARQ) works at the point when the

receiver identifies a mistake in the information; it will ask the source to
retransmit if negative acknowledgment comes from the receiver as shown in
Figure 1.2. Information transmitted over the channel, from the sender to the
collector will be exposed to distraction. Subsequently, data transmitted over
the channel will get corrupted due to noise. In such cases, the collector informs
the sender to retransmit the information to bring the rehashed transmission of
the same information. Such rehashed transmissions are not just giving
postponement in data communication, also consume extra channel bandwidth.
 Figure 1.2 Process of automatic repeat request

1.3.2 Forward Error Correction

Figure 1.3 Process of forward error correction

 As a substitute to ARQ, Forward Error Correction (FEC) method
is added in the channel encoder and decoder. Here, the data to be transmitted
will be included with parity bits and transmitted over the channel. Included
repetitive bits will assist the recipient to correct errors that happen during
transmission. As shown in Figure 1.3, when the receiver identifies an error in
the received data, it will find the error location in a stream of data and will
correct it using error control codes at the receiver itself. The FEC coding is
needed because it is a far superior option to transmit information over and over
the same channel until the destination informs received data is correct.

1.4 ERROR CONTROL CODE IN DIGITAL

COMMUNICATION

FEC is emerged as a necessary element in the present digital

communication system because of its ability to sort out errors in the receiver
end. This scheme is broadly categorized as Convolutional codes and Block
codes.

Figure 1.4 Classification of error control code

1.4.1 Shannon’s Information Theory

Shannon explained that every communication should reserve bits

for forward error correction coding to receive the data without error.

E = p log2(1/p) + (1 − p) log2(1/(1 − p)) (1.1)

Where p - The probability of a bit transmitted without error

(1 – p)- The probability of error
E - Reserved bits for error corrections

Shannon’s information theory helps to estimate the amount of coding

data that can be added to the message which helps data destination to handle
independently rectifying errors that happen in a channel. Figure 1.5 shows that
the Shannon limit on Eb/No consists of Channel’s spectral efficiency (Cp) as
one of the functions. When Cp =2, the Shannon limit on Eb/No is equal to
1.76 dB. If Cp =1, the limit comes to 0 dB. When Cp =0, it goes to -1.59 dB.

Cp =
R

BW
 Figure 1.5 Eb/No vs Spectral efficiency

This value is called the Shannon power efficiency limit. This limit
informs the minimum requirement of energy per bit at the transmitter for
reliable communication. It is one of the important measures in designing a
coding scheme.

1.4.2 Block Codes

Elias (1954) introduces the block error-correcting codes. The

block codes divide the incoming message stream as a block that has 'k' number
of bits and then applied the coding procedure which produces 'q' parity bits.
After encoding, error control bits (q) are appended to the message block. The
resultant code words will have n number of bits.

Code word (n)=Block message bits(k)+Parity bits(q) (1.3)

There will be 2k code words and each codeword is a linear

combination of other code words. This is classified into linear code and
nonlinear code. Linear codes are widely used because it has efficient
properties, concise description, and easy coding than nonlinear coding. Figure
1.6 shows the general structure of the block encoding System. Here, the blocks
from the input stream (i) are encoded with parity bits (p) to produce code set
(v). The block codes do not require any buffers and can correct a specified
number of errors. It is used in a system where BER is relatively low and has
very high code rates. The code rate is a ratio between codeword (n) to the
message (k). LDPC code, Hamming code, BCH code, RS code are some of the
leading block codes.

Figure 1.6 General structure of block encoding System

1.4.2.1 Low-Density Parity Check (LDPC) code

Gallager (1962) had given this code as one of the linear block
codes for error correction. This encoding will have more numbers of 0's than
1's in the parity check matrix; hence it is called a low density parity check
code. The LDPC codes are having a code rate closer to the Shannon channel
limit. It has nearly 0.04dB of Shannon's limit. This code uses an iterative
decoding method that is easy to implement and has enhanced performance in
the error rectifying process by providing a low error floor in the BER curve.
 10

This code can be constructed for any block length and code rate. Figure 1.7
shows the function of the LDPC with variable node (v) and parity check node
(c).

Figure 1.7 Depiction of LDPC code

1.4.2.2 Bose, Chaudhuri, and Hocquenghem (BCH) Code

Bose & Ray-Chaudhuri (1960) proposed the BCH codes. Due to

its cyclic and finite derived algorithmic approach, the encoding and decoding
process becomes simple. Kashani & Shiva (2006) explained that the BCH
code is characterized through the Galois Field for the wireless digital network.
This code can address different arbitrary errors and based on the number of
error correction capabilities, the message size could likewise be chosen. It’s a
cyclic code yet has direct properties of Hamming code. BCH codes are
characterized by polynomials; a generator polynomial is picked after which
encoding is executed. The advantage of this code over other codes is, it has
precise control on error correction ability and can be designed using low
intricate architecture. The BCH code is the generality of hamming codes,
contains the following parameters.

• Block length: n = 2 m – 1
• Number of parity-check digits: n – k ≤ m * t
• Minimum Distance: d min ≥ (2*t) + 1
 11

1.4.2.3 Reed and Solomon (RS) code

In this modern communication, RS code is used in numerous

applications like telecommunications, deep space communication, and
particularly the application requires to handle burst error correction. It works
well for the channel that has memory. While message size is expanded, this
code works as more prominent coding. RS code is a linear systematic code. If
more burst errors happen, this code will default due to its bounded distance
decoding property. It can correct up to half of its parity bits. The encoding and
interpreting of the RS code could likewise be designed as a BCH code; but,
variation is, here product and sum are performed in Galois Field. The same
encoding, decoding hardware of binary BCH code could likewise be utilized in
RS code also. Tan et al. (2014) used Reed - Solomon for a digital sensor-based
system to overcome the burst errors.

1.4.3 Convolutional Codes

Convolutional codes are alternatives to block coding in which

encoding and decoding can take place on a continuous data bitstream instead
of the static block code process. In this code, a set of bits are taken to the shift
register. The codeword is produced using a designed XOR system, when the
next bit of message comes in, the data in the register shifted and the codeword
is produced. Likewise, the process continues up to the last bit of the message.
The output is a combination of the previous bit and the present bit, so it
requires memory for buffering. Figure 1.8 shows the general structure of the
convolutional encoding system. A continuous bit stream is converted into a
single codeword. The widely used decoding algorithm is Viterbi because it
gives an efficient decoding system. Mahdavifar et al. (2014) implemented the
 12

convolutional code as a product code for the digital video broadcasting and
wireless sensor network. Trellis coded modulation and Turbo codes are
convolutional coding systems.

Figure 1.8 General structure of Convolutional encoding system

1.4.3.1 Trellis Coded Modulation (TCM) code

1.4.3.2 Turbo Code

Turbo code is having two Recursive Systematic Convolutional

codes (RSC) that process the incoming message as parallel and combines at
the output of the encoder. Turbo code is having three outputs, one encodes the
incoming message directly using the convolutional algorithm like RSC,
another convolutional encoder gets the same incoming message after
interleaving, and the third one is the un-coded original message. Turbo codes
having high error capability where information rate can be passed close to
theoretical limit and error rate close to zero. So that Condo and Masera (2014)
implemented turbo for deep-space communication. These codes are also useful
in UMTS, CDMA2000, mobile WiMAX systems, etc.

1.4.4 Product Codes

Figure 1.9 Product code structure

 Product code is a technique to encode a block of data with one
EDC code in a row and another EDC code in a column. First, the input data
stream is divided into k1 bits and q1 parity bits added by one algorithm,

n1=k1+q1 (1.4)
then, The n1 data set is exchanged from row to column using interleaver, and
the second algorithm is added. It adds q2 parity bits. Consider n1 as k2. Hence,
n2=k2+q2 (1.5)

Minimum distance of the product code = d1*d2 (1.6)

Where, d1= Minimum distance of codeword C1

d1= Minimum distance of codeword C2

So, the product code increases the minimum distance which is the
important parameter of error correction ability. The communication which
vulnerable to errors can be dealt with by incorporating the error control code
capable to correct random error in row and code capable to correct burst error
in the column or vice versa. Based on the application four code has been
chosen. The first product code is designed for DVB which is using LDPC and
BCH codes. The second product code is designed for mobile communication
which has more burst errors. So LDPC and RS code has been chosen. The
third product code is designed for deep-space communication. Here, turbo
code has been chosen along with LDPC to control burst errors.

1.5 CHANNEL ERROR MODELS IN COMMUNICATION

The channel error models like Additive White Gaussian Noise

(AWGN) Channels and Rayleigh fading channels are used to create the
random errors and burst errors respectively in the communication design
 15

model. This gives error estimation like transmitting as in real-time wireless

digital communication scenario and helps to analyze the performance of
coding algorithms Chowdhury et al. (2017) described that the Rician fading
model is quite inefficient for real-time application. So AWGN and Rayleigh
models are considered to add the noises in the channel that occurs during the
movement of signal from transmitter to receiver.

1.5.1 Additive White Gaussian Noise (AWGN) channel

The performance of the coding system is analyzed for the

correction ability of random error in the presence of noise, especially thermal
noise. These noises are additive to the signal while transferring from
transmitter to receiver, the autocorrelation of AWGN is zero since the power
spectral density is uniform and the noise samples follow the Gaussian random
process. The AWGN channel is a linear and time-invariant system. The
AWGN is used to create random errors in the message bitstream.

Figure 1.10 Power spectral density of the AWGN channel model

(McClaning & Vito 1959)

1.5.2 Rayleigh fading channel

Rayleigh fading is used to analyze the performance coding

system in the event of burst error occurrence at the digital communication
 16

system. This model creates a burst error in the signal which will arise in the
actual system due to multipath reception. Here the deep fading will arise in
every half of the wavelength due to the random process. The Rayleigh fading
channel model having its amplitude like the Rayleigh model and the derivative
of amplitude follows Gaussian which is amplitude independent. The phase is
uniform and the derivative of the phase is Gaussian which is amplitude-
dependent. It is useful in calculating the Probability Density Function (PDF) of
the incoming signal, the rate of fluctuation happening in a channel, effect of
multiple interferences occurring in the signal while using satellite and mobile
communication. The depth of burst error can be changed by modifying
variance (σ) as shown in Figure 1.11.
Probability

Random variable

Figure 1.11 Probability density function of the Rayleigh channel

model (Siddiqui 1964)
1.6 AN OVERVIEW OF PRODUCT CODES IN DIGITAL
COMMUNICATION SYSTEM

Digital communication is a self-sustained transmission system

which can correct errors while recognizing message at its receiver through
error control codes. In high-speed technological development, enhancing these
error control codes according to the present scenario becomes necessary.

1.6.1 Low Density Parity Check Code

Jian-bin et al. (2018) constructed a matrix of QC-LDPC which is

applied in Wireless Sensor Networks (WSN). The basic matrix shortens the
RS codes and has good minimum distances, BER performance, and less
complexity in hardware design and implementation. The simulation output
shows the WSN system with good energy efficiency. Along with fully
structured features, good minimum distance codes are used to ensure the girth
of the designed parity check matrix. The error correction is similar to the PEG
algorithm. It can be used in the practical applications of ECC for WSN.

Peng et al. (2018) explain a deterministic structure to resolve the

high coding complexity and inflexible code length selection in quasi-cyclic
low-density parity check. The perfect cyclic shifts ensure less storage space as
well as reduced hardware complexity. The simulation part shows the
improvement by 0.13 dB, 0.32 dB of perfect cycle different sets, and BER of
10-6 with AWGN. A special structure is designed and simulated with
combinational mathematical models. The sum-product algorithm is used to
give excellent SNR without error levelling.
 Sang & Joo (2006) analyse the effect of block interleaving in
LDPC codes. Here decoding is soft compared to other RS codes, so the LDPC-
Turbo codes generate higher performance and this can be improved further.
The average number of iterations will be minimized in the decoding part of
LDPC-Turbo codes. The results obtained in the simulation ensure the better
performance of the RS turbo codes. The 2D block interleaving is used with a
low BER region. It is suitable for high-quality communication systems.

Xiumin et al. (2019) propose a system with density evolution

analysis with Gaussian approximation. Here the Belief Propagation (BP)
algorithm and the Turbo Decoding Message Passing (TDMP) algorithm are
used. The symmetry condition obtained a convergent result. The density
evolution method provides a simple theoretical bays algorithm. The simulation
results in the convergence of the speedy decoding TDMP algorithm with the
normalized factor. It simulates six code rates of LDPC codes which are used
under IEEE 802.16e standards.

Moataz & Ashraf (2015) propose a system with many micro-

sensors that constitute the wireless sensor networks. Battery lifetime and
power consumption are considered as the most important impact, which
ensures a reliable and sustainable network operation. LDPC and ECC ensure
the network. It gives low SNR with less complex circuits. The energy
efficiency is with concatenated LDPC codes and simulated outputs give the
improved BER. The energy-efficient system proposed here is suitable for
wireless sensor networks. The control coding parameters include frame size,
code rate, and BER.

Alireza et al. (2019) designed a new binary message-passing

decoding algorithm. The algorithm is based on Bounded Distance Decoding
(BDD), which is a soft coding algorithm. However, the messages are hard it
 19

reduces the data flow. It ensures the reliability of the decoder output. The
density evolution analysis is used for generalized low density parity check
codes. This code achieves the performance gain up to 0.29 dB and 0.31 dB.
This method is expensive and has a minor increase in complexity. It is mainly
applied in fiber-optic communication systems.

Lin et al. (2013) propose an early termination technique used for

low density parity check codes. The decoding part is focused much on iterative
operations, with high reliability along with the soft log-likelihood ratio. The
threshold value is maintained throughout all the layers. It gives up to 60%
negligible loss of coding gain at Eb/No of 3.0 dB. This criterion gives the
highest reliability under simulation on IEEE 802.16e standards with high
efficiency provided the highest data rate.

Jin Xie et al. (2009) propose an improved layered min-sum

algorithm for usual low density parity check codes decoding. It is ensuring the
high speed of the decoding algorithm with regular updates. The simulation part
is compared with the analytical outputs and similar results were ensured. The
number of iterations is reduced by 8.4% - 62.6%. But increases the hardware
complexity. Here all the layers were enhanced with improved efficiency and
amplification.

Tian et al. (2020) describe a decoder for a non-binary LDPC code

with less complexity. Even though non-binary LDPC is used widely, its high
usage of hardware is a concern in its implementation. To solve this issue, a
trellis based min-max decoder with some optimized check node processing
unit architecture is proposed. This results in 1/3 hardware reduction and
increases clock frequency.
 20

Eran et al. (2017) propose the power-efficient LDPC decoder by

adding a trellis dependency in any quasi cycle LDPC. This system is encoded
by a parallel concatenated QC-RSC encoder based on the trellis. The results of
this system having 2 times lesser hardware complexity than conventional
QC-LDPC code for the same data rate SNR and BER. This parallel
concatenated LDPC code along with the QC-MAP decoder behaves well than
conventional QC-LDPC code by 0.5 dB. So, this system is useful in a high-
speed mobile system

Bocharova et al. (2019) present a decoding method for LDPC

code by using the Maximum Likelihood (ML) method. This reduces the
complexity of a decoder. In this method, when ML decoding fails to decode
and correct the errors, the system combines ML decoding with the list coding
method in the Binary Erasure Channel (BEC). The difference between BP
decoding and BP list coding method is presented in the Wi-Max standard.

Sunita & Bhaaskaran (2013) explains that regular error control

codes not able to correct right various mistakes in recollections, but a
significant number of those are fit for distinguishing numerous errors. This
system presents an exceptional translating calculation to recognize and
appropriate various mistakes. The calculation utilized is such it can address a
limit of 11 errors during 32-bit information and a limit of nine mistakes during
16-bit information. The proposed strategy is frequently used to improve the
correcting ability in the nearness of numerous bit upsets. The perpetual
arrangement of information bits is influenced when highly energetic particles
from outer radiation strike, and cause soft errors. It performs superior to the
recently known procedure of error recognition and rectification utilizing
matrix codes.
 21

The system of Qin et al. (2017) enumerates that all the reliability
vector elements are supported into separate sets. It is noted that trellis-based
extended min-sum decoding methods having a path. So it reduces the search in
all trellis of check nodes. The simulation results show that the described
methods decode the low-density parity-check code in a better manner in terms
of complexity also.

The system of Eran et al. (2015) gives an idea about trellis-based

quasi-cyclic Low-Density Parity-Check codes. This is built by introducing
trellis-based convolution dependency on any type of quasi-cyclic LDPC. It
also describes the QC Viterbi decoding algorithm. These two methods are
simulated and compared with conventional LDPC code and for the decoder;
the belief propagation method is taken. The results show that the proposed
method having 1 dB more for the same BER and complexity.

Aragon et al. (2019) present a substitution group of rank

measurement codes: Low Rank Parity Check (LRPC). This system proposed a
productive probabilistic translating calculation. This group of codes is
frequently observed because it might be compared to old-style LDPC codes
for the rank measurement. This is an all-inclusive rendition of the structure
presenting LRPC codes, with a significantly new function.

1.6.2 LDPC and BCH product code

Coşkun et al. (2020) enumerate a system of Successive

Cancellation List (SCL) decoding for product codes. Generally, SCL decoding
depends on a 2×2 Hadamard kernel which is used as a description for the
product system. This analysis focused on the product codes used in wireless
communication systems based on extended hamming codes. The results show
that SCL algorithms performed similarly to belief propagation algorithms.
 22

Also highlighted, the product code performs well when a high rate algorithm is
used as an outer code.

Li & Zhang (2010) present a scheme for product code. Here, the
TCM codes and BCH codes are used in the concatenation form for the error
control coding scheme. The TCM encoder and decoder are embedded on-chip
and BCH hardware is kept as off-chip. The improved TCM is also proposed
which relieves the burden of BCH code when correcting errors. The result
shows the improvement in error correction compared to the BCH code by
using the same number of extra redundant bits.

Kihoon et al. (2012) present a two-iteration concatenated Bose-

Chaudhri-Hocquenghem code. It is high speed with low-complexity suitable
for 100Gb/s optical communications with high data processing rates. A low
complexity syndrome is used in the architecture that ensures the high-speed
processing and 10-5 decoder output BER. It is used for potential applications,
the design and implementation of the two-iteration concatenated BCH code
give better performance. The block interleaving methods allow the same. It is
also applied for a very high-data processing rate as well as good error
correction.

Uryvsky & Osypchuk (2014) give a procedure for the complexity

of the LDPC check matrix (H) searching with good error-correcting ability
grows exponentially with increasing a codeword length. There is no need for
BCH codes to perform the search matrix procedure because of the nature of
encoding/decoding processes. This is an advantage of BCH codes. Research
showed that LDPC codes can be characterized as anti-noise codes with better
error-correction properties. The relative number of corrected errors per
codeword is almost the same for LDPC (n=1000) and BCH (n=1023) codes.
LDPC codes have a little bit better error-correcting abilities than BCH codes
 23

have if code rate R<0.7. It is implied that other parameters like code length,
signal to noise ratio, manipulation method, and required reliability are the
same for LDPC and BCH. The coded rates are obtained for LDPC (n=1000)
and BCH codes (n=1023) for the Gaussian channel when the signal to noise
value is known. According to numerical LDPC and BCH code rates values,
LDPC code can be recommended as more effective if the signal to noise value
is larger than 7 dB.

The system described in Spinner et al. (2016) gives the snapshot

for the soft input algorithm. It describes the structure for concatenated codes. It
is working on the concept of super code trellis which is more efficient than
ordinary trellis in terms of complexity. Super code trellis reduces the memory
required for BCH codes. This is achieved by using a sequential decoding
algorithm rather than using an ordinary decoding algorithm. This results in
better hardware architecture. The complexity of the decoder increases to 82%
compared to the system of Spinner & Freudenberger (2015). So this system is
useful where a low residual error rate is required.

The system in Xinmiao et al. (2011) prescribes the soft decision

algorithm for BCH decoder instead of hard decision decoding for satellite
systems. It is achieved by using the cyclic property with non-uniform
quantization to reduce the hardware. The simulation results show the described
system has a low hardware architecture and better BER performance than a
conventional hard decision algorithm.

Swaminathan et al. (2020) propose a blind estimation method for

product codes for the scenario at the receiver is non-cooperative, needs blind
reconstruction of the parameter. Generally, product codes use many
components like different coders, interleaves, etc. These components and their
parameters affect the performance. The efficiency of the algorithm in terms of
 24

the probability of correct estimation is analyzed using the product system of

BCH and RS. The results show that when the code dimension and modulation
order increase, the performance increases.

1.6.3 LDPC and RS product code

Qiu et al. (2020) describe the coding system that uses special case
LDPC, Spatially Coupled (SC) low-density parity-check as inner codes, and
reed Solomon as outer codes. A decoding algorithm based on belief
propagation is proposed as sliding window decoding. For RS coding,
Berlekamp-Massey is used. This system improves the Bit error rate up to 10-8
error floor.

Chengen et al. (2012) propose a product code system that uses

hamming code along the column and RS codes along the rows in the matrix of
input message bits. This system is compared with plain BCH code and RS
code. The result in comparison shows that the proposed system having 18%
lower latency and 40% lower space constraints. This work also explains a
flexible production system that moves to stronger ECC code by compensating
with latency when the error in the message is high. This work helps to design a
product system with less area and latency.

Jianguo et al. (2012) explain the construction of the product code

of RS with a low density parity check code. It compares the other construction
codes and produces 5.42% redundancy, which is a lower redundancy and is
less complex too. This method is highly suitable for optical transmission
systems. The simulated analysis shows the comparison of classical RS code
and it is observed that the novel LDPC code has better advantages including
error correction, lower decoding complexity, etc. It can be widely used for
high-speed ultra-long-haul optical transmission systems with FEC codes.
 25

Bingrui et al. (2019) generate a decoding model for Ship-based

Satellite Communications On-The-Move (SSCOTM) between ships and
satellites. When the ship antenna deviates there will be a burst error that leads
to data loss. A method to model the direction of wind resembling the reduction
of SNR is analysed. An ECC system is designed using the RS code with
LDPC and applied for error correction. The simulation output shows the better
performance from this system and it is reliable. This system is with simple
hardware implementation and a shorter code length.

Blaum (2020) generates the alternative of the Integrated

Interleaved and Extended Integrated Interleaved codes for locally recoverable
codes. The comprehensive definition is given to such specially generated new
codes. Improvements in the special case of coding techniques are observed and
the minimum distance with upper bound is also ensured. The systematic
coding techniques use the parity symbols in an iterative decoding algorithm on
rows and columns. It also involves the Reed-Solomon type of codes. The
generalization of the code is allowed with encoding and decoding
optimization.

Liu et al. (2014) enumerates the performance of the LDPC-RS

product code system for the use of the nextgen broadcasting system and also
provides a novel iterative scheme for the product coding system. The proposed
hybrid product scheme outperforms the regular hard decision of RS and SPA
of the LDPC algorithm by providing higher error correction ability and
robustness to the decoding system. The results show a decrease in decoding
threshold and complexity. This work helps to build a hybrid decoding scheme.

Salah (2018) describes the various forward error detection and

correction codes under various noises. Providing a better channel is difficult
nowadays because of interference, noises, and topology changes, etc. So it
 26

suggests emerging channel coding algorithms like LDPC, RS codes also

described the difference between Binary LDPC and Non-binary LDPC codes.

The described method in Jieun et al. (2016) uses the nonbinary

LDPC code as an outer code and Reed Solomon code as an inner code for the
application of modern tape storage. The result shows that the mentioned
method gives a better gain compared with RS-RS codes. Also, it handles
decoding complexity quite effectively.

Greaves & Syatriadi (2019) discuss the developments of some

extraordinary cases like RS codes [n, k] in limited fields (n to n+1) where
parity generator matrix gets bind output. Moreover, this system is added with
estimation for constrained generator matrix. Lovett's estimate is explained in
detail which indicates the estimation usability.

Park & Kim (2020) provides an estimation method using a

generator polynomial model for the error correction code. This model gives
high reliable generator polynomial for RS and BCH code. The probability
density function of the codeword is also highly accurate. At the system level, it
provides high system iuntegrity. This estimation gives an eye diagram similar
to transient simulation for RS and BCH code

1.6.4 LDPC and Turbo product code

Condo & Masera (2014) proposes a product code scheme for

deep space communication with error correction codes and low BER. The use
of different coding techniques leads to hardware complexity. To avoid this,
non-custom LDPC and turbo codes are used with the BCJR algorithm. The
serial concatenation of Turbo code and LDPC codes is used for simulation.
The results have been compared with state of art competitive results used in
 27

deep space. The early analysis of the area makes the system occupy less space
and also with low cost.

Yuan et al. (2017) propose two product codes based on LDPC

code along with used the constrained turbo product codes and constrained
turbo block conventional codes. The gain obtained is nearer to uniform
interleaving that increases the minimum hamming distance. Here new
improved code is used to give high performance. This is suitable for wireless
application that ensures better performance in the wireless communication
systems compared to WiMax and LTE standards respectively.

Gooru & Rajaram (2014) provides an overview of the novel turbo

codes capable of performing the operations nearer to Shannon's limit. The
turbo coding techniques with many iterative decoders are most commonly
used. The MAP algorithm is made use to reduce the number of iterations
applied in the decoding part. The FPGA implementation is used for the
simulation that encodes and decodes the data. It gives improved channel
coding techniques and the simulation outputs are verified with similar coding
techniques. The SISO decoder will reduce the number of iterations also.

Blaum & Hetzler (2018) generate the extended product codes that
process the encoding part more efficiently compared to other coding
techniques. This code along with extended integrated interleaved code form a
special case of codes that naturally unifies the upper bound. Here minimum
distance is also improved and enhancement is done at the decoding part with
uniform distribution of parity symbols. The upper bound on the minimum
distance was presented. It requires a small finite field in practical cases. The
simulation outputs ensure the encoding and decoding algorithms with
minimum distance.
 28

Winstead et al. (2005) propose Turbo codes and other iteratively

decoded codes have been joined into a few advanced correspondences systems
like 3GPP, DVB, and DVB-RCS. Because of the iterative idea of the
disentangling calculation, turbo decoders are deserted to interpret inactivity
and enormous energy utilization.

The survey by Mukhtar et al. (2016) is useful to analyse the code

rate and the complexity of various Turbo Product Codes (TPC). It gives
significance to multidimensional, nonbinary, modified row-column
interleaving TPC along with, irregular, extended and shortened, and nonlinear
TPCs. It also gives knowledge about product codes.