Giải pháp cải thiện tỷ lệ lỗi bit (BER) trong hệ truyền dẫn số OFDM, ứng dụng trong truyền hình số DVB t thế hệ mới tt tiếng anh

MINISTRY OF EDUCATION AND TRAINING MINISTRY OF NATIONAL DEFENCE
ACADEMY OF MILITARY SCIENCE AND TECHNOLOG Y

------------------------

TRAN HUU TOAN

SOLUTION ON IMPROVEMENT BIT ERROR RATE (BER)
IN OFDM DIGITAL TRANSMISSION SYSTEM, APPLYING FOR
THE DIGITAL VIDEO BROADCASTING – NEW GENERATION
TERRESTRIAL
Major: Electronic Engineering
Code: 9 52 02 03

SUMMARY OF PhD THESIS IN TECHNIQUE

Hanoi, 2019


This thesis has been completed at:
ACADEMY OF MILITARY SCIENCE AND TECHNOLOGY

Scientific supervisors:
1.

Dr. Nguyen Van Lien

2.

Assoc. Prof. Dr. Bach Nhat Hong

Reviewer 1: Assoc. Prof. Dr. Trinh Anh Vu
University of Engineering and Technology, Vietnam National University, Hanoi
Reviewer 2: Assoc. Prof. Dr. Nguyen XuanQuyen
Hanoi University of Science and Technology
Reviewer 3: Assoc. Prof. Dr. Bui Ngoc My
Academy of Military Science and Technology

The thesis was defended at the Doctoral Evaluating Council at Academy level,
held at Academy of Military Science and Technology at …., date….. 2019

The thesis can be found at:
- The library of Academy of Military Science and Technology
- Vietnam National Library


1

INTRODUCTION
1. The urgency of thesis
In recent years, the Orthogonal Frequency Division Multiplexing (OFDM)
has been used very effectively in many fields. One of those is theDigial Video
Broadcasting - Terrestrial (DVB-T). DVB-T has more disadvantages than other
modes of transmission such as:
- Channels are degraded due to multi-path reflections
- Man - made noise
- Frequency distribution is quite thick in the spectrum for interference
television between analogues and digital.
Therefore, it has been suggested that broadcasting DVB-T is not practical.
However, the introduction of digital terrestrial television standards such as DVB-T
of Europe, ATSC of Americal and ISDB-T of Japan has overcome most of the

disadvantages mentioned above.
So far the DVB-T system has been evolved to the second generation (DVBT2); The DVB-T2 standard is the most modern. DVB-T2 mainly broadcasts high
quality digital (high definition HDTV). Therefore, the two main targets of the
transmission system are concerned: channel capacity and quality of BER. To
achieve the two above targets, many technical solutions have been adopted in
DVB-T2 [33].
However, DVB-T2 does not consider comprehensive and comprehensive
types of interference impact on the system. For example, co-channel interference
by radio stations with frequency near the digital broadcasting frequency (CCI).
The InterChannel Interference remains when the carrier mode is extended to 32K,
overlapping interference of the sub-wavelengths of the subcarrier over the original
band. At the same time, decoder quality should also be considered. All of these
factors are limited to improving the BER quality of the system
To further improve the bit error rate for the DVB-T2 standard, the additional
solution to minimize the impact of these types of above noise, while also taking
into account the quality of the decoder.
Therefore, the topic of " Solution on improvement Bit Error Rate (BER) in
OFDM digital transmission system, applying forthe Digial Video Broadcasting –
New Generation Terrestrial” is a topic of high scientific and practical.
2. Research purposes
- Study the effect of different types of noise reducing the bit error rate of the
system.
- Find out solution to improve bit error rate (BER) in OFDM digital
transmission system from which to apply in the Digial Video Broadcasting – New
Generation Terrestrial.
- Improve the quality of the decoder


2

3. Research subject and research scope
Research subject:
- ICI, CCI interference and solution on reduce ICI, CCI in DVB-T2 system.
- Study soft decoding for LDPC in DVB-T2 system.
Research scope:
- M-QAM transceiver system with parameters in accordance with DVB-T2
standard. Interference between subcarriers and active noise in the M-QAM system.
4. Research content
- Research application of spatial filter to minimize the effect of positive
interference (Co-Channel Interference (CCI)).
- Research and application of Kalman filter to prevent Inter-Channel
Interference (ICI).
- Study soft decoding for error correction code, thereby improving the bit
error rate
5. Research Methods
Analytical methods and computer simulations are used in the thesis .
6. Scientific and practical significance
Scientific significance:
- To proposetheapplication of spatial filter to minimize the effect of CoChannel Interference, ie, anti-interference outside the television; thereby
improving the bit error rate.
- To propose the application of spatial filter to minimize the effect of the
Inter-Channel Interference, thereby improving the bit error rate of the system.
- To propose the application Soft-Input Soft-Output decoding to improve
decoding quality for LDPC code, thereby improving the bit error rate of the
system.
Practical significance:
- The researchs in the thesis contribute the solution on improvement the
quality of the current Digial Video Broadcasting – Terrestrial.
7. The structure of the thesis
The thesis consists of 4 chapters:

Chapter 1: Overview of European terrestrial digital television (DVB-T and
DVB-T2).
Chapter 2:Research space filter, proposedpractical application diagram
Chapter 3: Research Extended Kalman Filter and proposed a detailed
algorithm to minimize the effect of frequency shifting.
Chapter 4: This chapter researchs the MAP algorithm for LDPC code and
evaluates the possibility of improving the bit error rate when using it.


3

CHAPTER 1: OVERVIEW OF THE DIGIAL VIDEO BROADCASTING –
TERRESTRIAL DVB-T
1.1. Digital video broadcasting standards
The first generation of DVB-T has three standards:
- ATSC of American
- DVB-T of Europe
- ISDB-T of Japan
Similarity of the three above standards is used MPEG-2 standard for video
signal. The main difference is the modulation method.
1.2. Digital video broadcasting with Europe standard DVB
DVB organization is separated many committees, including the following
subcommitees:
- DVB-S: Develop digital broadcasting technology via satellite
- DVB-C: Develop digital broadcasting technology via cable
- DVB-T: Develop digital broadcasting technology via terrestrial.
In Vietnam, the digital television standard is accepted and allowed for widescale deployment of DVB-T. It can be said that the development of all three
modes of transmission is very practical, because the three transmission
environment will complement for each other.
1.3. DVB-T Overview

According to [34], the general block diagram of the DVB-T system shown in
Figure 1.1.

Figure 1.1. DVB-T system scheme


4

DVB-T uses Basic technology solutions:
- Modulating and multiplexing technique
DVB-T uses OFDM modulation and multiplexing technique.
- Error correction coding solution (channel coding)
DVB-T uses channel coding as the combined coding: Internal encoder is
convulutional codes and External encoder is RS codes (Reed-Solomon codes).
Using two error correction codings as a link code is better error correction.
According to [34] the link code has the probability of bit error decreases
exponentially while the complexity increases linearly.
1.4. The Digial Video Broadcasting –second generation Terrestrial (DVB-T2)
DVB-T2 is the second generation of Digial Video Broadcasting –Terrestrial.
DVB-T2 allows for increase channel data capacity (  30% ) than DVB-T and and
increased reliability in the ground wave propagation environment. DVB-T2 is
primarily intended for high definition digital television (HDTV).
1.4.1. Basic technical solutions:
1.4.1.1. Physical Layer Pipes (PLPs)
1.4.1.2. Extended carrier modes
1.4.1.3. Channel coding (Forward Error Correction encoding FEC)
1.4.1.4. Pilot Insertion
1.4.1.5. 256-QAM modulation
1.4.1.6. Rotated Constellation
1.4.1.7. PAPR Reduction

1.4.1.8. Mapping bits onto constellations
1.4.1.9. Cell Interleaver, Time Interleaver
1.4.1.10. DVB-T2 frame structure
The above are the technical solutions used in DVB-T2. The breakthrough
solution is: while DVB-T uses inner and outer correction code are Convulutional
Codes and RS, DVB-T2 uses LDPC and BCH. These codes enable better BER
performance and transfer better data over the same channel.
1.4.2. Comment on DVB-T2
First of all it should be affirmed that: DVB-T2 is the most modern terrestrial
digital television standard. The system has inherited the DVB-T technology
solutions and added new technology solutions to improve transmission capacity
and dramatically improve the bit error rate.
Today, demand for audiovisual is increasing. TV programs are required not
only for content but also for quality requirements. On the other hand, the trend of
HD technology is increasingly being developed, HD production equipments is
replacing the existing SD production system. All of these issues require further
refinement of DVB-T2 standards on a number of issues such as: the MPEG-4
Compressed Application Research, continue to improve the bit error probability...


5

1.5. Summary of published works
So far, there has been a number of studies to overcome and improve the
quality of DVB-T2 such as [15], [16], [19], [44], [60], [61] và [79]. However, it
can be seen that:
- There is no technical solution to reduce noise in front of the TV receiver.
- There is no effective technical solution to minimize the impact of ICI
interference on high frequency offset and Doppler frequency.
- The DVB-T2 decoding problem is still in hard mode.

So, with the aim of thesis “The solutions to improve Bit Error Rate (BER) in
OFDM digital transmission system, apply for the Digial Video Broadcasting –
New Generation Terrestrial”, the researching content of this thesis will be studied
in three aspects as follows:
Firstly, research application of spatial filter to improve SNR at the TV
receiver input, this means improving the bit error probability of the system. The
results of this study are presented in chapter 2.
Secondly, research application of Kalman filter to minimize the effect of
Inter-Channel Interference in case the frequency offset changes random.The
results of this study are presented in chapter 3.
Thirdly, research application of soft encoder to substitute for hard encoder to
to improve decoder’s quality, this means improving the bit error probability of the
system. The results of this study are presented in chapter 4.
CHAPTER 2: RESEARCH APPLICATION OF SPATIAL FILTER TO
MINIMIZE INTERFERENCE OCCURRING IN TELEVISION RECEIVER
2.1. General problems about space filter and applications
2.1.1. Space and time signal
2.1.2. Spatial Filter
Perform spatial signal processing with a spatial filter circuit which has
impulse response h( x ,t ) is a multi-dimensional linear system. Then system’s
input is spatial signal S ( x ,t ) and system’s output is f ( x ,t ) .
Filtering process is to separate the desired signal components with a certain
frequency and propagation direction.


6

2.1.3. Basic beam structure
2.1.4. Applicability of spatial filter
To minimize Co-Channel Interference (CCI), according to the theory of

spatial filter circuits, M elements can create M-1 negative directions and so can
only eliminate the M-1 of CCI noise source. The efficiency of interference
filtering of spatial filter circuits is usually evaluated by SINR ratio. When the
larger the M, the larger the SINR. However, at that time, a great deal of
calculations were required (especially multiplication). Therefore, to improve SINR
of the spatial filter, attention is paid to the number of calculations (relative to the
speed of the signal processing) and this is huge limitation on the possibility of
applying the model to reality. To reduce the signal level in the direction of the
positive interference source, thesis proposed using self-compensation method
which is a type of spatial filter that is made entirely of hardware and possible
application in practice.
2.3. Quadrature positive self-suppressor
2.3.1 Principle diagram
The principle diagram of Quadrature positive self-suppressor is shown in
Figure 2.6.

Figure 2.6. Quadrature positive self-suppressor


7

- The equalizer performs inverse the noise vector components and is the
amplitude control element of one of the two orthogonal component interference
vectors.
- The multiplier is the phase difference signal splitter of the two orthogonal
components
- The electronic locks on duty: In the absence of interference, the system
will interrupt, inthe opposite, the system will be connected by the control voltage.
The multiplicator’s input has two signals:
The first one is the interference signal in the secondary antenna are analyzed

into two orthogonal components (due to the 00 and 900 phase reversal).
The second is the distorted signal between the interference vector from the
secondary antenna and the interference vector from the lateral wing of the main

antenna from the adder’s output S .
The task of the multiplier is to separate the false signal for two orthogonal
channels. The output signal of the multiplier:


 

(2.29)
U ra  K TS U n 00 ,900 U S  cos n  sin n 
2
2 

Sensitivity of the multiplier (considered as phase separation), with the
variation of  n p , is characterized by sensitivity of phase separation  :


  K TS U ra  cos

n

 sin

n 

(2.30)


2
2 

The gain factor of the control element is K đc . Then the amount of leftover
error of the system will be:
U S
(2.31)
U Sdu 
1  K đc
The noise control factor is:
U
K CA  S  1  K đc
(2.32)
U Sdu
Thus, the noise control factor of self-compensator depends on K đc and
sensitivety of the multiplier  , with K đc  K cb K sbc (where K cb is gain factor of the
equilibrium modulator, K sbc is gain factor after adder).
2.3.2 Operational principle
The operational principle of quadrature positive self-suppressor is shown in
Figure 2.7.


8

Figure 2.7. The operational principle of achannelof quadrature positive self-suppressor

Where: S0 is signal vector obtained from the lateral wing of the main
antenna.
S n is signal vector obtained from secondary antenna.

In the secondary channel, the vector S n is separated by two orthogonal
vectors S Y and S X n , then two vectors are reversed in the equilibrium modulator,
n

get two vectors  SYn and  S X n .

The distorted signal S is applied to the two multiplicator of the two
branches, generates distorted signals for two branches, passes through electronic
locks and integral circuits lead to amplitude control of  SY and  S X . When the
n

n

amplitude of two these components changes, this leads to phase of  S n changes.

When phase of  S n is opposite to phase of S 0 then S  0 , the self-suppressor
stops working.
Thus, the self-suppressor has created concave slot in the direction of
thepositiveinterference source, so the positive interference source will not entrance
the receiver. This will improve BER quality of the system.
To be able to create four concave slots towards four positive source
interferences, by using four auxiliary antennas placed perpendicular to each other
– one auxiliary antenna avoids creating a concave slot in a right angle.
2.4. Simulation and results
As known, bit error probability is a function of signal to noise ratio (SNR) at
S
the receiver input Pb  f   . In digital information, SNR is evaluated (or
N

9

E 
expressed) through Eb/N0 ratio.So can rewrite Pb  f  b  . If using the
 N0 

E 
quadrature positive self-suppressor then Pb  f  K CA b  .
N0 

According to [17], the error bit probability of the M-QAM signal is
determined by the following formula:

Pb 





4 M  1  3 log 2 M Eb 

Q
M log 2 M  M  1 N 0 

(2.35)

Where: M is the signal constellation level
Q is a negative cumulative distribution function of normalized
random variables
u2

1  2
Q x  
 e du
2 x

(2.36)

If there is a quadrature positive self-suppressor at the receiver input, the bit
error probability at the receiver input will be:
Pb 









E  4 M  1  3log 2 M
4 M  1  3log 2 M
1  K đc  Eb
Q
K CA b  
Q
N0 
N0
M log 2 M  M  1
M log 2 M  M  1


 (2.37)



To evaluate the performance of the quadrature positive self-suppressor, using
the Matlab software to simulate, setting for parameters according to Table 2.1.
Bảng 2.1. Setting BER evaluation’ paremeters of quadrature positive self-suppressor
No

Parameters

Settings

1

Eb/N0

0:30

2

Channel type

AWGN

3

Modulation type

QAM

4

Demodulation type

Coherent

5

Modulation order

4, 16, 64, 256

6

Voltage control factor K CA

1, 5, 10

 Simulation results:


10

Figure 2.8. BER performance of quadrature
self-suppressor with 4-QAM modulator

Figure 2.10. BER performance of quadrature

self-suppressor with 64-QAM modulator

Figure 2.9. BER performance of quadrature
self-suppressor with 16-QAM modulator

Figure 2.11. BER performance of quadrature
self-suppressor with 256-QAM modulator

Based on the simulated results, the author makes Table 2.2 to calculate the
gain at BER = 10-4.
Table 2.2. The gain of the quadrature self-suppressor at BER = 10-4
Modulator

K CA  1
(Do not use
self-suppressor)

K CA  5

Gain

K CA  10

Gain

4-QAM

8.5dB

7dB

1.5dB

5.5dB

3dB

16-QAM

13dB

10.5dB

2.5dB

8dB

5dB

64-QAM

17dB

14dB

3dB

11.5dB

5.5dB

256-QAM

22.5dB
18.5dB
3.5dB
15.5dB
7dB
Based on the simulated results and the result of gain in Table 2.2 we can see:
With using the quadrature positive self-suppressor in front of the TV receiver has


11

effectively eliminated noise, thereby improving the quality of the system. When
the more modulation level increases (the noise increases), the more effective the
self-suppressor, such as: with 4-QAM modulator, K CA  10 the gain is 3dB; while
with 256-QAM modulator, K CA  10 , the gain is up to 7dB. This means that we
can apply the quadrature positive self-suppressor to the Digial Video Broadcasting
– Second Generation Terrestrial (DVB-T2) using the modulator up to 256-QAM.
CHAPTER 3: RESEARCH APPLICATION OFKALMAN FILTER TO
PREVENT INTER-CHANNEL INTERFERENCE (ICI)
3.1. The effect of OFFSET and DOPPLER
Digial Video Broadcasting –Terrestrial system uses OFDM transmission
system. According to [52], [67], one of the fundamental disadvantages of OFDM
is very sensitive to the frequency shifting between transmitter and receiver signals.
Frequency offsetmay due to differences between the internal oscillator frequencies
at transmitter and receiver sides or Doppler frequency shift ing in the channel.
Frequency offset is the causes of lossing of orthogonality between the subcarriers
creating ICI interference.

3.2. ICI interference cancelling solutions
Synthetize references materials on preventing ICI interference can be
presented in the classification scheme of the ICI interference preventing methods
shown in Figure 3.2.

Figure 3.2. Classification of ICI interference preventing methods
3.2.1. Self – Cancellation Scheme method (SC)
Self – Cancellation scheme method proposed by Yuping Zhao and SvenGustas Haggman in 2001 [82]. The main idea of this method is to modulate the
input data on a group of subcarriers with predetermined coefficients for the ICI
interference generated in the group will self-cancel each other. Thus, the total of
ICI interference components number is reduced by half, because only the even


12

subcarriers are added. Therefore, if using the Self - Cancellation scheme, BER
performance will improve, but bandwidth efficiency will be reduced by half.
3.2.2. Maximum likelihood method (ML)
This method was proposed by Moose [68]. The main idea of this method is to
estimate the frequency offset  by the closest algorithm and then eliminate the
ICI noise at the receiver. To implement this method, it is necessary to create an
OFDM version before the transmission and then compare each subcarrier between
the next symbols.
According to [63], it was found that when N and  are small, the Self –
Cancellation method and the closest method have good BER performance.
However, both methods reduce bandwidth efficiency, because each subcarrier has
a backup. When N and  are high, the two methods are no longer effective. In this
case, according to [63] should apply the Extended Kalman Filter method, because
the EKF which allows us to estimate accurately is a powerful recursive estimation
method. Therefore, ICI interference will be eliminated, thereby improving the

system's BER quality.
3.3. Kalman Filter
3.4. Extended Kalman Filter (EKF) [49]
A Kalman filter has expectant and linearized covariance matrix called the
Extended Kalman Filter. The Kalman filtering method consists of two processes:
status estimation process and calibration process (measured value updated
process) of the Extended Kalman Filter.
3.4.1. Status estimation
The updated equations over time ofExtendedKalmanfilterare as follows:
xˆ k  f (xˆ k 1 , u k 1 ,0)
(3.54)

Pk  A k Pk 1ATk  Wk Qk 1WkT
(3.55)
Like the discrete Kalman filter, state vector and covariance matrix are
calculated from the previous step k  1. A k and Wk is the Jacobian matrix of the
kth step, Q k is the covariance matrix of the process.
3.4.2. Measured value update process
Measured value updated equations of Extended Kalman filter are as follows:
1
(3.56)
K k  Pk HTk H k Pk HTk  Vk R k VkT 
xˆ k  xˆ k  K k z k  h( xˆk ,0)
(3.57)
Pk  (I  K k H k )Pk
(3.58)
Measured value updated equations of Extended Kalman filter will correct the
state vector and the error covariance matrix based on the actual measured value z k .
H k and Vk is the Jacobian matrix of the observation process at the kth step,
R k is the interference covariance matrix of the measurement process. Synthetize

updated process over time and measured value update process, it is possible to


13

build up the general operation diagram of the extended Kalman filter shown in
Figure 3.5.

Figure 3.5. Operation of the Extended Kalman filter
3.5. Application of Kalman filter to prevent inter-channel interference (ICI)
To cancel ICI interference, basic problem is to accurately estimate the
frequency offset  . Using the above analysis, it is possible to use the extended
Kalman filter as the most feasible method for exactly estimating  without
reducing the system bandwidth. In order to be able to apply the Kalman filter, we
need to represent the mathematical equations for the updated process over time
and the measured value update process.
To construct operation algorithm of the extended Kalman filter and to
estimate  (n) , we assume the following:
- Frequency offset  (n) is considered constant in an OFDM frame
- Status does not change according to the steps A  1
- There is no control entrance u  0
- The measurement process interference is the state matrix H  1
- The fluctuation of the process is very small, choose Q  0
So we have the mathematical equations for the updated process over time and
the measured value update process are as follows:
The equations update over time:
ˆ n  ˆ n1
(3.67)

Pn  Pn1

(3.68)
The equations update the measurement process:
K n  Pn ( Pn  R )1
(3.69)


ˆ n  ˆ n  K n ( zn  ˆ n )
(3.70)
Pn  1  K n Pn
(3.71)
The algorithm schema of the extended Kalman filter to estimate  (n) is
described in Figure 3.6.


14

Figure 3.6. The algorithm schema of the extended Kalman filter to estimate ˆ n
Simulation and results:
To compare ICI interference cancellation performance of the above three
interference preventing methods, the author simulates by using Matlab software.
The simulation diagram shown in Figure 3.9.

Figure 3.9. The simulation diagram evaluates ICI interference cancellation performance


15

The program is run with the following parameters:
- Frequency offset   0.05 ,   0.15 and   0.3 .
- M-QAM modulator with M= 4, 16, 64 and 256.

Simulation results:
With the 4-QAM modulator, the results of BER performance simulations of
the three methods are shown in Figure 3.10, Figure 3.11 and Figure 3.12:

Figure 3.10. BER performance of interference
cancellation diagrams of 4-QAM modulator
with   0.05

Figure 3.11. BER performance of interference
cancellation diagrams of 4-QAM modulator
with   0.15

Figure 3.12. BER performance of interference cancellation diagrams of 4-QAM
modulator with   0.3

With the 16-QAM modulator, the results of BER performance simulations of
the three methods are shown in Figure 3.13, Figure 3.14 and Figure 3.15:


16

Figure 3.13. BER performance of interference
cancellation diagrams of 16-QAM modulator
with   0.05

Figure 3.14. BER performance of interference
cancellation diagrams of 16-QAM modulator
with   0.15

Figure 3.15. BER performance of interference cancellation diagrams of 16-QAM modulator with   0.3

With the 64-QAM modulator, the results of BER performance simulations of
the three methods are shown in Figure 3.16, Figure 3.17 and Figure 3.18:

Figure 3.16. BER performance of interference
cancellation diagrams of 64-QAM modulator
with   0.05

Figure 3.17. BER performance of interference
cancellation diagrams of 64-QAM modulator
with   0.15


17

Figure 3.18. BER performance of interference cancellation diagrams of 64-QAM modulator with   0.3

With the 256-QAM modulator, the results of BER performance simulations
of the three methods are shown in Figure 3.19, Figure 3.20 and Figure 3.21:

Figure 3.19. BER performance of interference cancellation Figure 3.20. BERperformance of interference cancellation
diagrams of 256-QAM modulator with   0.05
diagrams of 256-QAM modulator with   0.15

Figure 3.21.BER performance of interference cancellation diagrams of 256-QAM modulator with   0.3


18

Based on the simulated results, author tabulate to calculate the gain with

diargrams of the 4-QAM, 16-QAM and 256-QAM at BER = 10-3
Table 3.1. Compare the gain of 4-QAM interference cancellation scheme
Method
Do not use
ICI
interference
cancellation
Self –
Cancellation
Maximum
likelihood
Extended
KalmanFilter

  0.05

Gain

23dB

  0.15

Gain

26 dB

  0.3

Gain

42 dB

22.5dB

0.5 dB

24.5 dB

1.5 dB

28.5 dB

13.5 dB

20dB

3 dB

20 dB

6 dB

23 dB

20 dB

22.5dB

0.5 dB

22.5 dB

3.5 dB

22.5 dB

19.5 dB

Table 3.2. Compare the gain of 16-QAM interference cancellation scheme
Method
Do not use
ICI
interference
cancellation
Self –
Cancellation
Maximum
likelihood
Extended
Kalman Filter

  0.05

Gain

31 dB

  0.15

Gain

  0.3

Gain

40 dB

The ICI is great
Cancel ICI
interference but
BER is large

22.5 dB

8.5 dB

32 dB

8 dB

19 dB

12 dB

21 dB

19 dB

22.5 dB

22.5 dB

20 dB

11 dB

20 dB

20 dB

20 dB

20 dB

Table 3.4. Compare the gain of 256-QAM interference cancellation scheme
Method
Gain
  0.05
Do not use
ICI
40 dB
interference
cancellation
Self –
20 dB
20 dB
Cancellation
Maximum
18 dB
22 dB

likelihood
Extended
17.5 dB 22.5 dB
Kalman Filter
Look at the results we see:

  0.15

Gain

  0.3

Gain

The ICI is great

Cancel ICI interference but BER is large
18.5

22.5 dB

17.5

17.5 dB


19

- With the higher order modulator (the larger the binary symbol size), the
greater the frequency offset  , the gain increases significantly when using the ML

method and the EKF method. The EKF has the best cancellation performance
when the higher the modulation level, the greater the frequency offset; as in the
256-QAM modulator and the frequency offset   0.3 , EKF method gives a gain
of about 5 dB at BER = 10-3 compared to the ML method.
- The Self – Cancellation method is only effective when the frequencyoffset
 is low. But when the greater the frequencyoffset  , the SC method is almost
impossible to cancel ICI interference.
General comment:
- The Self – Cancellation method does not completely cancel ICI
interference from vicinity subcarriers and the effect of ICI interference that is not
completely cancelled will increase as the frequency shifting and symbol size
increase. The Self – Cancellation method does not require too complexity
hardware and software, however it reduces bandwidth efficiency when there are
two redundancy per subcarrier.
- The closest method has the same disadvantage as the Self – Cancellation
method. However, the BER performance is better because the frequency shifting
estimate is more accurate. The technical implementation of this method is less
complicated than the Self – Cancellation method.
- Extended Kalman Filter method does not reduce bandwidth efficiency
when frequency offset can be estimated from the beginning of the data series in
each OFDM frame. This method is more complicated than two above methods.
Three above methods reduce ICI interference. The choice of which method
depends on the specific application. With the new generation of digial video
broadcasting –terrestrial, high-carrier mode, high-level modulation and mobile
reception (high frequency offset), it is proposed to use the Extended Kalman Filter
(EKF) to eliminate the ICI interference in the systemwill give the best BER
performance.
CHAPTER 4: APPLICATION OF SOFT DECODING TO IMPROVE BIT
ERROR PROBABILITY FOR LINK CODE
4.1. LDPC code and LDPC link code

4.2. SISO soft decoding algorithms
Classification of Soft-Input Soft-Output algorithms (SISO) are shown in
Figure 4.2.


20

Figure 4.2. Classification of SISO algorithms
Comparing the above SISO decoding algorithms, according to [75]
comments on complexity: MAP algorithm has the highest complexity. Log-MAP
decoding algorithm is three times as complex as SOVA algorithm but much lower
than MAP algorithm, The Max-Log-MAP algorithm is twice as complex as the
SOVA algorithm. About quality: The quality of MAP algorithm is the best, the
quality of Log-MAP algorithm is not as good asthat of MAP algorithm, the quality
of Max-Log-MAP is lower than that of Log-MAP algorithm. Finally, the quality
of SOVA algorithm is the worst.
4.3. Application of MAP decoding algorithm for LDPC code
The MAP algorithm involves decoding algorithms of the maximum
likelihood values (ML) to minimize bit error probability and this is the optimal
method for estimating the state and output of Markov processes in white noise
conditions.
The MAP decoding algorithm that the author constructs here is a soft
decision decoder using channel information.
The decoding algorithm consists of two steps:
Horizontal step: update rmn ( x )
The rmn ( x ) quantities that concern with the m check bits will be updated and
passed as the message to the bit nodes that have been checked by the m check
nodes. This operation is performed with all test nodes.
Vertical step: update qmn ( x )
The quantities qmn ( x ) that concern with the n bits are updated and passed as

a message to the check nodes, including the n bit node.
Consider the case of binary LDPC code:
4.3.1. Horizontal step: update rmn ( x )
For qml  qml ( 0 )  qml ( 1 ) and rml  rml ( 0 )  rml ( 1 )
According to [72], we have: rmn  qmn
(4.6)
nN m ,n


21

For each other element O (m, n) of H, calculate the product qmn , along the m
row, subtract the value of n column. It is therefore called horizontal step.
Use condition rmn ( 0 )  rmn ( 1 )  1 , there is a link between rmn , rmn ( 0 ) and
rmn ( 1 ) calculated by the following formula:
1  rmn
1  rmn
, rmn ( 1 ) 
(4.7)
rmn ( 0 ) 
2
2
4.3.2. Vertical step : update qmn ( x )
Theorem 4.1 [72]: For a cn bit which relate to the check states, if the

independent check bits are:
qn ( x )  P( cn  x | rn )  P( z m  0 | cn  x ,r )

(4.8)

mM n

Where  is normalized constant.
We have:
rmn ( x )  P( zm  0 | cn  x ,r )
Using (4.9) and theorem 4.1 we can write:
qn ( x )  P( cn  x | r )  rmn ( x )

(4.9)
(4.10)

mM n

Abbreviate : qmn  P( cn  x | zm  0,m  M n ,m ,r )
According to theorem 4.1, it is possible to write:
qmn ( x )  P( cn  x | r )  rmn ( x )

(4.11)
(4.12)

mM nm

From the product of (4.12) calculated under the column of matrix H (along
the check bits), the update qmn ( x ) is called the vertical step of the decoding
algorithm.
4.3.3. Initialization and completion of decoding
The iterative decoding algorithm is initialized by setting qmn x   Pn x  , with
qn ( x ) determined by the formula:
qn ( x)   n Pn ( x)  rmn ( x)
mM n

(4.13)

Where  n was chosen to qn 0  qn 1  1. Posterior probability q n ( x ) is
used to make decision for x , x  0,1.
Perform temporary decision:
If qn 1  0.5 , establish cn 1  1
If qn ( 1 )  0.5 , establish cn ( 1 )  0
Decoding ends when Hcˆ  0 , that means all check bits are satisfied
simultaneously. Conversely, if the number of iterations is less than the maximum
number of iterations, repeat from the horizontal step; if the number of iterations is
greater than or equal to the maximum number of iterations, the error is reported .
This is a break-down that goes beyond code error correction capability with the
number of that loops.


22

From the above analyzes of the loop decoding procedure for the LDPC code,
we have the decoding algorithm flowchart shown in Figure 4.4.

Figure 4.4. Algorithm flowchart of loop decoding for binary LDPC code
4.4. Effeciency of improve bit error probability when using soft decoding for
LDPC code compared to hard decoding
4.4.1. Simulation block diagram
The simulation block diagram is shown in Figure 4.5.

Figure 4.5. Simulation block diagram

23

4.4.2. Simulation results
Matlab Simulink simulation software was used to evaluate soft loop decoding
using MAP algorithm with hard decision decoding with LDPC code (16200,
8100), ie code rate 1 2 , with effect of Gaussian noise. The simulation’s result is
shown in Figure 4.7.

Figure 4.7.Evaluate soft loop decoding and hard decision
decoding with LDPC code (16200, 8100)
From the simulation’s result in Figure 4.7, it can be seen that: with bit error
rate of 10-4, the gain of soft decoding compared to the hard decoding for LDPC
code (16200, 8100) is 1.2dB. That means that when applying the MAP decoding
algorithm to the OFDM system, the decoding result is better than hard decoding,
thereby significantly improving the quality of the system. Especially, with Digial
Video Broadcasting – Second Generation Terrestrial (DVB-T2) system that is
currently using hard decoding algorithm (bit flipping algorithm) to decode the
LDPC code, it is necessary to consider replacing the soft decoding algorithm for
this system. However, in order to apply the MAP decoding algorithm to the
present Digial Video Broadcasting –Terrestrial system, it is necessary to replace
the microprocessing chips and the memory capacity to ensure that the system is
not delayed during processing in real time. Because, with hard decoding, the
signal passes through the demodulator will make a decision right away about the
information bits before send them to the decoder. Whereas, with soft decoding, the
demodulator does not self-determine the obtained bits information obtained, but
will send its information to the decoder in a suitable structure so that the decoder
makes the final decision more accurately, ie, the system’s quality will be better.
But then the system will be delayed because it takes some time to process and
make the final decision on the obtained bits information. That means that when