You are on page 1of 4

2015 IEEE 6th Control and System Graduate Research Colloquium, Aug.

10 - 11, UiTM, Shah Alam, Malaysia

Comparison between LDPC Codes and QC-LDPC


Codes in Term of PAPR in OFDM System with
Different Encoding Techniques
Ezmin Abdullah Azlina Idris
Electrical Engineering Electrical Engineering
University Technology Mara University Technology Mara
Shah Alam, Selangor Shah Alam, Selangor
nimze_275@yahoo.com azlina831@yahoo.com

Abstract—Peak to Average Power Ratio (PAPR) is known to be sinusoid mixing to form signal to be transmitted [1][2]. The
a traditional problem in Orthogonal Frequency Division large peaks of the signal bring to other problems such as the
Multiplexing (OFDM). The peak value of power signals bring to probability of energy spilling to the adjacent channel, signal
many other problems, thus the implementation of OFDM system distortion and error rate performance degradation as well as
in many wireless applications are growing slowly. There are
increasing the amplifier cost significantly [3].
many techniques have been discussed to reduce PAPR in OFDM
systems and one of them is PAPR reduction through coding. In There are many techniques have been discussed in order to
this paper, a comparison between two coding; Low Density reduce high PAPR in OFDM system and one of the technique
Parity Check Codes (LDPC) codes and Quasi Cyclic LDPC (QC- is through coding. Coding techniques do not cause the signal
LDPC) codes has been discussed and additionally two types of distortion like clipping and filtering technique [1] which
encoding technique; G-Matrix and Approximate Triangular that resulting in high bit error rate (BER). This technique enjoys the
is used in LDPC/QC-LDPC has been compared. The results show benefits of PAPR reduction and also the error correcting
that PAPR reduced about 13% in case of QC-LDPC Codes from properties [4]. The previous investigation regarding the coding
the original OFDM system and 8% in case of encoding process
technique has been conducted in [4] and [5] utilizing Turbo
using Approximate Triangular from G-Matrix. The comparison
result between LDPC and QC-LDPC can be the proof that QC-
codes and LDPC codes. Simulation results in [5] reported that
LDPC codes are better than LDPC codes while the second result LDPC codes have led to better performance compared to Turbo
shows that encoding process also contributes a significant impact codes. There is another promising coding technique which has
to PAPR reduction. These results could lead to further a low computational complexity than LDPC Codes called
investigation in the future. Quasi-Cyclic LDPC Codes (QC-LDPC) [6]. In order to encode
the data, there is two type of encoding techniques has been
Index Terms—Orthogonal Frequency Division Multiplexing applied in the construction of codeword in LDPC and QC-
(OFDM), Low Density Parity Check (LDPC), Quasi Cyclic Low LDPC codes. First is G-Matrix encoding and the other one is
Density Parity Check Codes (QC-LDPC), Power-Average-Power-
Approximate Triangular encoding. Approximate Triangular
Ratio (PAPR).
encoding gives a significant low computational complexity
compared to G-Matrix encoding[7]. However, as far as being
reported, these techniques has not yet been compared in term
I. INTRODUCTION
of PAPR reduction.
The demand for wireless communication system has led to In this paper, the PAPR comparison between LDPC codes
the development of efficient, reliable and also high speed and QC-LDPC codes is conducted. Further, two types of
wireless communication in order to provide people with encoding techniques that has been used to encode LDPC/QC-
sophisticated and ubiquity services. OFDM is a multicarrier LDPC which is G-Matrix [8] and Approximate Triangular [9]
modulation technique that is practically easier to implement in is compared. The comparison between these two techniques is
a digital communications network. This technique gives a conducted in order to determine its effect to the PAPR
significant increase in data rates, robustness in frequency reduction. The OFDM system model is illustrated in Fig.1. (*)
selective fading, high spectral efficiency as well as low indicates the focus in this study.
computational complexity. Due to these advantages, OFDM
has been widely used in the most well-known high data
communication standards such as Long Term Evolution (LTE)
cellular, IEEE 802.16 Worldwide Interoperability for
Microwave Access (WiMAX) and Digital Video Broadcasting
[1]. Fig. 1. OFDM system block diagram with LDPC/QC-LDPC Codes.
However, OFDM systems suffer from high PAPR problem
caused by the nature of modulation process which involving

978-1-4673-6716-5/15/$31.00 ©2015 IEEE 23


2015 IEEE 6th Control and System Graduate Research Colloquium, Aug. 10 - 11, UiTM, Shah Alam, Malaysia

II. LDPC CODES CONSTRUCTION ܲଵଵ ܲଵଶ … ܲଵ(௞ିଵ) ܲଵ௞


LDPC codes which first invented by Gallager [8] are sparse ‫=ܪ‬൦ ܲ ଶଵ
ܲଶଶ … ܲଶ(௞ିଵ) ܲଶ௞ ൪ (2)
block codes with parity check matrices, H, that contain 0’s and ‫ڭ‬ ‫ڭ ڭ ڭ ڭ‬
few 1’s. The matrix is a parity check matrix, M × N where ܲ ௝ଵ
ܲ௝ଶ … ܲ௝(௞ିଵ) ܲ௝௞
N > ‫ ܰ = ܯ ݀݊ܽ ܯ‬െ ‫ܭ‬. These properties define the total
Where a ij ȯ^«/– ’`[11]. Parameters k and j
number of the codeword is N bits, message bits is M and K is
are related to the resultant overall code rate R by
the parity bits. There are some important parameters that need
to be considered in order to construct LDPC codes such as
5•– j / k (3)
size of H, weight distribution for each row and column and the
minimum cycle degree in biparte graph [10].
The example of parity-check matrix H of the QC-LDPC codes
LDPC codes are often represented in Tanner Graph (Fig.2)
which utilizing circulant sub-matrices is shown as follows
which consist of two set of vertices, N vertices for the
codeword bits (bit nodes) and M vertices for parity check
‫ܫ‬ ‫ܫ … ܫ ܫ‬ … ‫ܫ‬
equations (check nodes). There are edges between bit nodes ‫ۍ‬0 ‫ې‬
‫(ܲ … ܲ ܫ‬௝ିଶ) ‫( ܲ ڮ‬௞ିଶ)
and check nodes corresponding parity check equation. The ‫ێ‬ ‫ۑ‬
number of 1’s in parity check matrix, H is equal to the number
‫ݍ(ܪ‬, ݆, ݇) = ‫ێ‬0 0 ‫ܲ … ܫ‬ଶ(௝ିଶ) ‫ܲ ڮ‬ଶ(௞ିଶ) ‫ۑ‬ (4)
of edges in Tanner Graph [10]. ‫ڭێ‬ ‫ڭ … ڭ ڭ‬ ‫ڮ‬ ‫ڭ‬ ‫ۑ‬
‫ۏ‬0 0 ‫ ڮ‬0 ‫(ܲ ڮ ܫ‬௝ିଵ)(௞ିଵ) ‫ے‬

where I x is a PxP identity matrix with rows cyclically shifted.

IV. ENCODING TECHNIQUES


In this paper, there are 2 types of encoding technique has
been compared in QC-LDPC codes. The first technique is
generator matrix (G-Matrix) and the second technique is the
approximate upper triangular matrix (Approximate Triangular).
1 1 0 1 0 0
‫ = ܪ‬቎0 1 1 0 1 0቏ A. G-Matrix
1 0 0 0 1 1
0 0 1 1 0 1 The generator matrix, G, is obtained while defining the H
matrix as G*H=0. Firstly, H will be put into this form Eq. (5),
Fig. 2. The Tanner Graph and its correspoding Parity Check Matrix, H by performing Gauss-Jordan elimination
with 6-cycle [10].
‫ܣ[ = ܪ‬, ‫ܫ‬௡ି௞ ] (5)
III. QC-LDPC CODES CONSTRUCTION
where A is a (n-k) x k binary matrix and ‫݊ܫ‬െ݇ is the size n-k
For the case of high encoding complexity of LDPC code, identity matrix. So, the generator matrix becomes
it is preferable that most of the parity bits are computable
using sparse operation, leading to approximately linear
‫ܫ[ = ܩ‬௞, ‫] ்ܣ‬ (6)
encoding complexity. Random LDPC code can be encoded
with almost linear cost named as QC-LDPC code matrices
The bits are encoded through the generator matrix using
consist of the blocks of circularly shifted identity matrices
Eq. (7),
[11], where a codeword QC-LDPC code can be constructed by
ܿ =‫ܩכݑ‬ (7)
applying circulant matrices [7]. Let P below be the ‫ܮ × ܮ‬
circulant permutation matrices, given by
where C is a codeword and u is message bits. A codeword c is
valid if the subsequent parity check equation is true,
0 1 0‫ڮ‬ 0
‫ ۍ‬0 0 1‫ڮ‬ 0‫ې‬
‫ێ‬ ‫ۑ‬ ‫= ܿכܪ‬0 (8)
ܲ = ‫ڭێ‬ ‫ڭ‬ ‫ڭ‬ ‫ۑڭ‬ (1)
‫ ێ‬0 0 0‫ڮ‬ 1‫ۑ‬
‫ۏ‬1 0 0‫ڮ‬ 0‫ے‬ B. Approximate Triangular Form
Rather than using G-Matrix for encoding, the method of the
Let ܲ௜ stand for the circulant permutation matrix, which back institution by transforming parity check matrix, H into
is derived from the identity matrix I after the latter one is approximate upper triangular can reduce encoding complexity
shifted to WKHULJKWE\LWLPHV ”L”/). The zero matrix is significantly [12]. The transformation is done through
defined as 3’ whereas the resulting parity check matrix H of permutations process of rows and columns. Firstly, H matrix is
size (݆ ή ‫ ݇( × )ܮ‬ή ‫ )ܮ‬is denoted in Eq. (2). divided into six blocks as in Eq. (9).

24
2015 IEEE 6th Control and System Graduate Research Colloquium, Aug. 10 - 11, UiTM, Shah Alam, Malaysia

TABLE I. The PAPR value has decreased 6.9% in case of


‫ܣ‬ ‫ܤ‬ ܶ LDPC codes and 13% in case of QC-LDPC Codes from
‫=ܪ‬ቂ ቃ (9)
‫ܥ‬ ‫ܦ‬ ‫ܧ‬ the original OFDM system.

where matrix T is a lower triangular matrix. Encoding will be


formulated using the A, B, C, D, E and F blocks in the H
sparse matrix. Parity bits must be calculated using the
formulas ܴଵ = FCAܵଵ் and ܴଶ = [C (Aܵଵ் ) +Bܴଵ் ] where ܵଵ is
the information bit. The encoded data is built by this equation

ܿ‫ ݀ݎ݋ݓ݁݀݋‬1 = ܵଵ ܴଵ ܴଶ (10)

V. PEAK TO AVERAGE POWER RATIO


When N signals are added with the same phase in the N-
point IFFT stage to produce OFDM symbol, the peak value
of the same signals can be very high as compared the
Fig. 3. CCDF of PAPR original OFDM without coding, LDPC Codes and
average of the whole system. Mathematically, PAPR QC-LDPC Codes (64 QAM, Coding rate 1/2)
definition is given as
­° Ppeak ½° TABLE I. Comparison of PAPR between original OFDM with no coding,
PAPR 10 log ® ¾ (11) LDPC Codes and QC-LDPC Codes at 10ିଶ clip rate
°̄ Pavg °¿ OFDM System Improvement
PAPR in dB
where P peak is the maximum power of OFDM symbol, and (64QAM, 128IFFT) percentage (%)
P avg is the average power [1]. Mathematically, No Coding 10 -
LDPC Codes 9.3 6.9
­ max x(t) 2 ½ QC-LDPC Codes 8.7 13
° °
PAPR 10 log ® ¾ (12)
2
° E x(t ) °
¯ ¿ In Fig.4, it is shown that Approximate Triangular better than
where max|x(‫|)ݐ‬2 is maximum signal power and E[|x(‫|)ݐ‬2] is G-Matrix encoding used by QC-LDPC in OFDM system. The
the average signal power. Average signal power of OFDM percentage of PAPR reduction is 8.33%. This result is possible
system calculated by because of the effect of the lower computational complexity in
the encoding process of Approximate Triangular. TABLE II
Sum of the magnitude of all OFDM symbols shows the numerical result of PAPR CCDF at 10ିଶ clip rate.
‫=ܧ‬ No. of OFDM symbols
(13)

CCDF is one of the most used to check how often the


PAPR execute the threshold value [2].CCDF can be
expressed by

3U 3$35!; í3 3$35; (14)


VI. SIMULATION AND RESULTS
The simulation is carried out using Matlab software.
The input data that produce ½ coding rate of LDPC Code
and QC-LDPC is randomly generated. One Hundred of
OFDM signals is considered to calculate the PAPR Fig. 4. CCDF of PAPR QC-LDPC Codes using G-Matrix encoding and
CCDF. The simulation includes; IFFT number, n= 128 Approximate Triangular encoding.(64 QAM, Coding rate 1/2)
and modulation scheme, m. The simulation result is
shown in CCDF plots. Fig.3 shows the PAPR comparison TABLE II. Comparison of PAPR between QC_LDPC using G-Matrix
encoding and Approximate Triangular encoding at
between original OFDM system with no coding, OFDM
10ିଶ clip rate
system with LDPC Codes and OFDM system with QC-
OFDM System Improvement
LDPC Codes. PAPR in dB
(64QAM, 128IFFT) percentage (%)
In Fig.3, at the PAPR CCDF clip rate of 10ିଶ , the
OFDM system using QC-LDPC Codes shows the best G-Matrix 9.6 -
reading. It is then followed by OFDM system using Approximate-
8.8 8.33
LDPC Codes and lastly follow by original OFDM system Triangular
with no coding. The numerical comparison is shown in

25
2015 IEEE 6th Control and System Graduate Research Colloquium, Aug. 10 - 11, UiTM, Shah Alam, Malaysia

VII. CONCLUSION
QC-LDPC Codes is the better selection of coding
technique in order to reduce high PAPR with 13%
improvement compare to LDPC codes with only 6.9%
improvement. Further investigation on the encoding technique
used in QC-LDPC Codes for OFDM system, Approximate
Triangular encoding shows a better result compare to G-Matrix
encoding. This finding concludes that a lower complexity of
encoding process also contributes a significant effect in PAPR
reduction.
Furthermore, the construction of the QC-LDPC codes with
Approximately Triangular is also easier to implement with low
computational complexity. These advantages will encourage
researchers to utilize this simple approach in the future.

REFERENCES
[1] S. Hee Han and J. Hong Lee, “An Overview Peak-to-
Average Power Ratio Reduction Techniques For
Multicarrier Transmission,” in IEEE Wireless
Communications, no. April, pp. 56–65, 2005.
[2] S. H. Wang, J. C. Sie, C. P. Li, and Y. F. Chen, “A Low-
Complexity PAPR Reduction Scheme for OFDMA Uplink
Systems,” IEEE Trans. Wirel. Commun., vol. 10, no. 4, pp.
1242–1251, Apr. 2011.
[3] M. Sabbaghian, Y. Kwak, B. Smida, and V. Tarokh, “Near
Shannon Limit and Low Peak to Average Power Ratio
Turbo Block Coded OFDM,” IEEE Trans. Commun., vol.
59, no. 8, pp. 2042–2045, 2011.
[4] O. Daoud and M. Al-kaidi, “Reducing The Peak-to-Average
Power Ratio Using Turbo Coding,” IEEE Proceeding-
Commun., vol. 153, no. 6, pp. 818–821, 2006.
[5] O. Alani and O. Daoud, “Reducing the PAPR by utilisation
of the LDPC code,” IET Commun., vol. 3, no. 4, pp. 520–
529, 2009.
[6] A. Yahya, F. Ghani, R. Badlishah, and R. Malook, “An
Overview of Low Density Parity Check Codes,” J. Appl.
Sci., vol. 10, no. 17, pp. 1910–1915, 2010.
[7] M. P. C. Fossorier, S. Member, and A. Preliminaries,
“Quasi-Cyclic Low-Density Parity-Check Codes From
Circulant Permutation Matrices,” IEEE Trans. Inf. Theory,
vol. 50, no. 8, pp. 1788–1793, 2004.
[8] R. G. Gallager, “Low-density parity-check codes,” IEEE
Trans. Info. Theory, vol. 8, no. 1, pp. 21–28, 1962.
[9] C. Di, T. J. Richardson, S. Member, and R. L. Urbanke,
“Weight Distribution of Low-Density Parity-Check Codes,”
vol. 52, no. 11, pp. 4839–4855, 2006.
[10] S. J. Johnson, “Introducing Low-Density Parity-Check
Codes,” 2006.
[11] S. Myung, K. Yang, and J. Kim, “Quasi-cyclic LDPC codes
for fast encoding,” IEEE Trans. Inf. Theory, vol. 51, no. 8,
pp. 2894–2901, 2005.
[12] T. J. Richardson, “Efficient Encoding of Low-Density
Parity-Check Codes,” vol. 47, no. 2, pp. 638–656, 2001.

26

You might also like