PERFORMANCE ANALYSIS OF DWT-

OFDM AND FFT-OFDM SYSTEMS

G. Gowri, G. Uma Maheswari, E. Vishnupriya, S. Prabha, D. Meenakshi, N. R. Raajan
Department of Electronics and Communication Engineering
School of Electrical and Electronics Engineering
SASTRA University
Thanjavur, Tamilnadu, India.
gowrisastra7@gmail.com, umaty91@gmail.com, priyavishnu234@gmail.com, prabhakumaresan@gmail.com,
d.meenakshi2310@gmail.com, nrraajan@ece.sastra.edu
AbstractPerformance of Orthogonal Frequency Division Multiplexing (OFDM) systems using Fast
Fourier Transforms (FFT) and Discrete Wavelet Transform (DWT) are analyzed in this paper. The
performance of DWT-OFDM is assessed by various parameters such as BER, eye diagram and
constellation diagram. BER is the ratio of the no. of bits with error to the total no. of bits transmitted
through the channel. From the Eye diagram, we can find the eye height through which the interference
can be calculated. By varying the different wavelets and the different modulation schemes the BER is
calculated and the eye diagram and the constellation diagram are plotted with the aid of MATLAB-
SIMULINK. In this work, performance of most widely used wavelet based OFDM systems Haar,
Daubechies, Biorthogonal, and Reverse Biorthogonal wavelets are studied and wavelet basis suitable for
optimum OFDM system is investigated. Bit Error Rate (BER) versus Signal to Noise Ratio (SNR) is used
as system parameter.
I ndex Terms OFDM- Orthogonal Frequency Division Multiplexing, DWT- Discrete Wavelet Transform,
FFT-Fast Fourier Transform, BER- Bit Error Rate, SNR- Signal to Noise Ratio, PAPR- Peak to Average
Power Ratio, AWGN Additive White Gaussian Noise
I. INTRODUCTION
OFDM-ORTHOGONAL FREQUENCY DIVISION MULTIPLEXING is a combination of both
modulation and multiplexing technique. Frequency Division Multiplexing (FDM) divides the available
bandwidth into several subcarriers which does not overlap with each other, but the orthogonality in the OFDM
provides the subcarriers to overlap with each other, thus providing high spectral efficiency.
This OFDM technique can be implemented effectively by using Fast Fourier transform (FFT) but it has
got some disadvantages such as a very high Peak to Average Power Ratio (PAPR) and the use of Cyclic Prefix,
which is used to reduce the Inter Symbol Interference (ISI), reduces the spectrum efficiency. In order to
overcome the above disadvantages, we go for the Discrete Wavelet Transform (DWT). OFDM provides an
efficient means to handle high speed data streams over a multipath fading environment. Conventional OFDM
systems use FFT filters for modulation and demodulation. Due to superior spectral containment properties of
wavelet filters, DWT based OFDM systems have improved performance.
II. FAST FOURIER TRANSFORM
The Fourier transform is the most popular transform used to obtain frequency spectrum of signals. Fast
Fourier Transform is an algorithm which is used to find the Discrete Fourier Transform in a faster way and with
reduced complexity. It is best suited for signals whose frequency components do not change with time. The data
generator used is of bit stream and it is modulated using M-PSK or M-ary QAM modulators to map input data
into symbols. These symbols are sent to IFFT block to generate N- parallel data streams, since OFDM transmits
parallel data streams simultaneously over the subcarriers.
For a given finite sequence of data,
x|n], n = u to N -1 (1)
We define the FFT of x[n] i.e., the output in frequency domain as
X|k] = x|n]c
-]k[
2H
N
n
N-1
n=0
, k = u to N -1 (2)
X[k] is the component of the spectrum due to the frequency.
F=
kF
s
N
(3)
The transformed output is added with cyclic prefix to decrease ISI effects of channel. It is usually 25%
of the last part of the original transformed signal.
In the receiver, the reverse operation is performed by removing the cyclic prefix, FFT modulation and
demodulation to obtain the original data back.
G. Gowri et al. / International Journal of Engineering and Technology (IJET)
ISSN : 0975-4024 Vol 5 No 2 Apr-May 2013 1455
The IFFT i.e., the output in discrete domain as
x|n] =
1
N
X|k]c
]k[
2H
N

, n = u to N -1
N-1
k=0
(4)

III. DISCRETE WAVELET TRANSFORM
A wavelet is a small portion of a continuous signal. Discrete wavelet has got its name, since the
samples are discretely sampled and the most important fact is that it provides both time and frequency domain
information. There are several types of wavelet transforms widely used such as HAAR, DAUBECHIES, BI-
ORTHOGONAL, etc. The Wavelet Transform should satisfy orthonormal basis property and the perfect
reconstruction property. The wavelet Transform decomposes the signal into small wavelets by using the wavelet
function and the basis function, so a wavelet transform has got two functions as named above. The filters in the
Discrete Wavelet Transform have got both low pass and high pass filters which actually acts as the Quadrature
Mirror Filters (QMF). The low pass filter coefficients is called as approximated coefficients and the high pass
filter coefficients is called as detailed coefficients.



Figure 1.FILTER SPLITTING
The input is first passed through the low pass filter g(n) and its output is the approximated coefficient and then
the same input passed through the high pass filter h(n) produces the output as the detailed coefficient. This is
how the signal is decomposed by the wavelet. The mathematical expression for this decomposition is described
below.
F1(k) = _ x|n]g|2k -n]

n=0
(5)
F2(k) = _ x|n]b|2k -n]

n=0
(6)
IV. STRUCTURE OF FFT OFDM AND DWT OFDM


Fig.2. FFT WITH CYCLIC PREFIX
In the above block diagram data generator generates the data which is modulated and mapped into the
symbol and the serial to parallel converter converts the high speed data into low speed parallel data stream then
the inverse transform is performed into the modulated symbol then the cyclic prefix is added to it and then it is
passed over the Additive White Gaussian Noise (AWGN) channel and at the receiver the inverse operation is
performed.
G. Gowri et al. / International Journal of Engineering and Technology (IJET)
ISSN : 0975-4024 Vol 5 No 2 Apr-May 2013 1456


Figure 3.DWT TRANSMITTER
The high input data stream is converted to low parallel data stream by using the serial to parallel conversion and
it is transposed then it is passed through the low pass and the high pass then finally the output of it is the OFDM
symbol which has to be transmitted over the AWGN channel.


Figure 4.DWT RECEIVER
Generally zero padding in time domain increases the number of samples whereas zero padding in the frequency
domain results in an increased sampling rate in the time domain. In DWT Transmitter zero padding is done and
at the receiver zero is discarded.
V. SIMULATION RESULTS
Table I.BER FOR BPSK MODULATION
Eb/N0 FFT HAAR DB4
BIOR
5.5
RBIOR
3.3
0 0.5004 0.4148 0.5010 0.5052 0.5051
3 0.5003 0.3777 0.5005 0.5039 0.5030
5 0.5003 0.3481 0.4997 0.5005 0.4997
10 0.4998 0.2426 0.4996 0.4992 0.4995
20 0.4998 0.0117 0.4995 0.4992 0.4991
50 0.4996 0 0.4995 0.4991 0.4990
100 0.4996 0 0.4995 0.4989 0.4985



G. Gowri et al. / International Journal of Engineering and Technology (IJET)
ISSN : 0975-4024 Vol 5 No 2 Apr-May 2013 1457
Table II.BER FOR DBPSK MODULATION
Eb/No FFT HAAR DB4
BIOR
5.5
RBIOR
3.3
0 0.5021 0.4874 0.4998 0.5016 0.5003
3 0.5020 0.4766 0.4984 0.5004 0.5006
5 0.5020 0.4622 0.4999 0.5020 0.5022
10 0.5017 0.3888 0.5 0.4967 0.4983
20 0.5016 0.0388 0.4950 0.5014 0.4971
50 0.5010 0 0.4948 0.5003 0.4916
100 0.5010 0 0.4938 0.5001 0.4937
Table III.BER FOR QPSK MODULATION
Eb/N0 FFT HAAR DB4
BIOR
5.5
RBIOR
3.3
0 0.7493 0.6506 0.7519 0.7584 0.7535
3 0.7493 0.6088 0.7530 0.7570 0.7524
5 0.7493 0.5710 0.7563 0.7566 0.7520
10 0.7493 0.4227 0.7541 0.7574 0.7529
20 0.7395 0.0245 0.7519 0.7567 0.7540
50 0.7397 0 0.7495 0.7545 0.7487
100 0.7396 0 0.7479 0.7551 0.7509
Table IV.BER FOR DQPSK MODULATION
Eb/N0 FFT HAAR DB4
BIOR
5.5
RBIOR
3.3
0 0.7489 0.7298 0.7343 0.7409 0.7546
3 0.7460 0.7143 0.7185 0.7310 0.7539
5 0.7465 0.6957 0.7069 0.7219 0.7498
10 0.7463 0.5983 0.6570 0.6819 0.7352
20 0.7455 0.0947 0.5377 0.5447 0.6725
50 0.7454 0 0.5206 0.5145 0.5254
100 0.7450 0 0.5207 0.5143 0.5243
Table V.BER FOR OQPSK MODULATION
Eb/N0 FFT HAAR DB4
BIOR
5.5
RBIOR
3.3
0 0.7495 0.7415 0.7446 0.7465 0.7525
3 0.7494 0.7415 0.7441 0.7464 0.7510
5 0.7486 0.7407 0.7439 0.7462 0.7476
10 0.7483 0.7401 0.7428 0.7457 0.7475
20 0.7389 0.7398 0.7421 0.7454 0.7468
50 0.7381 0.7387 0.7417 0.7449 0.7467
100 0.7375 0.7385 0.7410 0.7441 0.7460
Table VI.BER FOR 16 QAM MODULATION
Eb/No FFT HAAR DB4
BIOR
5.5
RBIOR
3.3
0 0.9401 0.9287 0.9361 0.9381 0.9403
3 0.9410 0.9257 0.9359 0.9379 0.9399
5 0.9399 0.9223 0.9355 0.9379 0.9386
10 0.9388 0.9108 0.9352 0.9375 0.9380
20 0.9385 0.8364 0.9350 0.9370 0.9363
50 0.9396 0.4709 0.9345 0.9367 0.9378
100 0.9396 0.4709 0.9343 0.9365 0.9378

G. Gowri et al. / International Journal of Engineering and Technology (IJET)
ISSN : 0975-4024 Vol 5 No 2 Apr-May 2013 1458
EYE HEIGHT
The eye height is inversely proportional to the interference level. So, higher the eye height, lower is the
interference level. The maximum height is for the HAAR Wavelet, which has got the maximum height under the
BPSK Modulation Scheme. So, the interference level is less for DWT-OFDM when compared with the FFT-
OFDM System. Below are the eye diagrams and constellation plots for various wavelet transforms.
Table VII. EYE HEIGHT
EYE HEIGHTS FFT HAAR DB4 BIOR 5.5 RBIOR 3.3
BPSK 0.171 0.043 0.028 0.031 0.032
DBPSK 0.180 0.043 0.035 0.023 0.025
QPSK 0.188 0.019 0.038 0.020 0.038
DQPSK 0.3 0.017 0.031 0.030 0.031
OQPSK 0.195 0.019 0.017 0.034 0.025
16-QAM 1 0.094 0.219 0.156 0.188
64-QAM 1.875 0.25 0.266 0.313 0.125
Figure 5.EYE DIAGRAM FOR BPSK BIOR 5.5

Figure 6.EYE DIAGRAM FOR BPSK DB4

Figure 7.EYE DIAGRAM FOR BPSK HAAR

Figure 8.EYE DIAGRAM FOR BPSK RBIOR 3.3



G. Gowri et al. / International Journal of Engineering and Technology (IJET)
ISSN : 0975-4024 Vol 5 No 2 Apr-May 2013 1459
Figure 9.CONSTELLATION DIAGRAM FOR BPSK BIOR 5.5

Figure 10.CONSTELLATION DIAGRAM FOR BPSK DB4

Figure 11.CONSTELLATION DIAGRAM FOR BPSK HAAR

Figure 12.CONSTELLATION DIAGRAM FOR BPSK RBIOR 3.3












G. Gowri et al. / International Journal of Engineering and Technology (IJET)
ISSN : 0975-4024 Vol 5 No 2 Apr-May 2013 1460
Table VIII. BER VARYING WAVELET TRANSFORMS ON BOTH SIDES
Eb/N0 HAAR RBIOR 3-3 BIOR 5-5
0 0.4758 0.4991 0.4976
3 0.4725 0.4978 0.4979
5 0.4690 0.4984 0.4976
10 0.4633 0.5022 0.5005
20 0.4950 0.5032 0.4983
50 0.5018 0.5029 0.4983
100 0.5018 0.5029 0.4984

VI. CONCLUSION
The DWT-OFDM has a very great advantage over the FFT-OFDM and the performance parameters that we
have considered proves the same. Especially, the HAAR wavelet under the BPSK modulation provides a very
good platform for the wireless communication with the minimum BER, less INTER SYMBOL
INTERFERENCE (ISI), and generally with the less PAPR. The wavelet transform has been varied in the
transmitter and at the receiver. The DAUBECHIES 4 has been used at the receiver to take the forward transform
and the minimum BER is obtained when the HAAR wavelet is used at the transmitter at an SNR of 10db. So,
with this, we can conclude that the aim of our paper is to provide the wavelet, with which the best performance
in wireless communication can be obtained.

VII. REFERENCES

[1] W. Saad, N. El-Fishawy, S. EL-Rabaie, and M. Shokair, An Efficient Technique for OFDM System Using Discrete Wavelet
Transform, 5th International Conference, GPC 2010, Hualien, Taiwan, pp 533-541, May 10-13, 2010
[2] Shiferaw Y., Comparative performance study on wavelet based orthogonal frequency division multiplexing (OFDM) using different
wavelets, Masters thesis, Addis Ababa University (2007)
[3] Abdullah K., Hussain Z.M., Studies on DWT-OFDM and FFT-OFDM systems, International Conference on Communication,
Computer and Power (ICCCP 2009), February 15-18 (2009)
[4] W. Saad, N. El-Fishawy, S. EL-Rabaie, and M. Shokair, An Efficient Technique for OFDM System Using Discrete Wavelet
Transform,
[5] N.R.Raajan, B.Monisha, M.Ram Kumar, A.Jenifer Philomina, M.V.Priya, D.Parthiban, S.Suganya, Design and Implementation of
Orthogonal Wavelet Division Multiplexing (OHWDM) with Minimum Bit Error Rate, Trendz in Information Sciences and Computing
(TISC), 2011 3rd International Conference, pp 122 127, 8-9 DEC 2011
[6] Paul Guanming Lin, OFDM SIMULATION in MATLAB, Bachelors project, California Polytechnic State University, June 2010
[7] Communication systems, 4th edition, Simon Haykin, John Wiley and Sons, Inc.
G. Gowri et al. / International Journal of Engineering and Technology (IJET)
ISSN : 0975-4024 Vol 5 No 2 Apr-May 2013 1461