Journal of Engineering Technology and Applied Sciences, 2017 e-ISSN: 2548-0391

Received 5 May 2017, Accepted 10 December 2017 Vol. 2, No. 3, 137-144

Published online 30 December 2017 doi: will be added

Citation: Babaoglu, A., Muratoglu Curuk, S., "Exigency of Channel Estimation in an OFDM System:
A Simulation Based Study". Journal of Engineering Technology and Applied Sciences 2 (3) 2017 : 137-144.

EXIGENCY OF CHANNEL ESTIMATION IN AN OFDM

SYSTEM: A SIMULATION BASED STUDY

Ayfer Babaoğlua*, Selva Muratoğlu Çürükb

a*
Department of Electrical and Electronics Engineering, Faculty of Engineering and Natural Science,
Iskenderun Technical University, İskenderun/Hatay, Turkey (corresponding author)
ayfer@emtamuhendislik.com.tr
b
Department of Electrical and Electronics Engineering Faculty of Engineering and Natural Science,
Iskenderun Technical University, İskenderun/Hatay, Turkey
selva.curuk@iste.edu.tr

Abstract
The popular multicarrier modulation technique, Orthogonal Frequency Division Multiplexing
(OFDM), is used for high rate wireless communication systems, as well as for other inexplicit
multipath channel applications, such as powerline and underwater communications. Final performance
of an OFDM system depends highly on the channel because variations in the characteristics of the
channel result with a degradation in the performance. In this study, a modular OFDM system which is
modeled using Matlab/Simulink program is given. The proposed model is used for performance
evaluation. The variations in the performance of the OFDM system are analyzed by investigating the
Bit Error Rate (BER). A comparison is made for two modulation schemes, namely phase-shift keying
(M-PSK) and quadrature amplitude modulation (M-QAM), under the additive white Gaussian noise
assumption. It has been shown that for a safe communication on a heavy Rayleigh channel especially,
channel estimation is a necessity.

Keywords: Channel estimation, Matlab, Orthogonal Frequency Division Multiplexing, Simulink.

1. Introduction
Orthogonal Frequency Division Multiplexing (OFDM), the well-known widely deployed
multicarrier techniques, is a modulation scheme that splits the total transmission bandwidth
into a number of subchannels with orthogonal subcarriers in order to cope with multipath
channel effects and transmit the symbols. Although OFDM applications had a limited use for
a long time due to implementation complexity, with the developments in electronics it has
been a popular topic in the field of communications. Nowadays, with its powerful anti-fading
capacity and data transfer rate, the OFDM technology is used by many communication
services in wireless communications as well as in other multipath channel applications, such
as powerline and underwater communications and is considered as the core technology of
next-generation broadband communications.

The performance of an OFDM system is indeed a well-researched topic and a huge amount of
paper related to this scope can be found in the literature. Also there are simulation models;
some noteworthy ones are presented in [1-7]. Studies on modelling systems use highly
specialized mathematical tools such as Matlab, Simulink [8], Maple, and Mathematica in the
literature. These software packages provide built-in libraries and standard building blocks for
using in developing of systems. The most popular tool is Matlab/Simulink in communication
area which give the advantage of easily modeling channel, modulation, and demodulation
with different parameters. For example, in [1] the performance of OFDM is tested for two
digital modulation techniques using Matlab/Simulink toolbox. In [2], authors built an OFDM
simulation model under environment of Simulink. Similarly in paper [3], a Simulink based
simulation system is implemented under Additive White Gaussian Noise channel (AWGN) to
study the performance analysis. The authors of [4] have proposed the design of an OFDM
system, using Simulink modeling tool and communication block set with PSK digital
modulation schemes. Mapping techniques for OFDM are simulated in [5]. The combination
of MIMO and OFDM system is given in [6]. The authors of [7] presents a simulation model
of OFDM communication system based on again Simulink.

OFDM systems are widely used in different areas under varying channel conditions, with
various transceiver designs and different modulation schemes. Thus modular simulation
models of specific cases may still be essential for performance analysis. These models
provide a platform for further studies of OFDM systems. Further simulation models may also
be used for educational demonstration and experimental purposes offering highlighting and
studying specific concept and phenomena. In the first step of this study, a simulation model of
an OFDM system with Matlab/Simulink is presented. Matlab and Simulink have been chosen
as the modeling tool because of their flexibility and mathematical capabilities. Next, the
performance of the modeled system is tested under Rayleigh channel and with AWGN
assumption. The performance of the OFDM transmitter and receiver is investigated for
Rayleigh fading channel and with AWGN; and the variations in performance caused by
channel are observed. The results concerning two typical digital modulation schemes of
OFDM, namely M-QAM (M-ary Quadrature Amplitude Modulation) and M-PSK (M-ary
Phase Shift Keying) are presented. Bit Error Rate (BER) versus Signal to Noise Ratio (SNR)
curves are used for comparing the results of M-QAM and M-PSK modulations. It is shown
that channel estimation and correction is a critical issue in all OFDM receivers.

The rest of the paper is organized as follows. Section 2 briefly describes the Simulink OFDM
system model used in simulations. Simulation results are presented in Section 3. Section 4
follows with conclusions.

2. OFDM system model

OFDM is the popular technique for frequency selective multipath channels. An OFDM signal
is the sum of N independent QAM/PSK signals modulated onto orthogonal subcarriers. In a
frequency selective band, selecting the bandwidth of each subchannel small enough results
with flat fading subchannels. Thus, in an OFDM system the bandwidth of each subchannel is
small compared to the coherence bandwidth of the channel. That is, the symbol duration is
large compared to the channel time dispersion and the subcarriers experience flat fading.

138
Figure 1 displays the conventional transceiver of an OFDM system modeled with Matlab /
Simulink, which is used in our simulations. The amplitudes and/or phases of the carriers are
calculated based on the modulation scheme, namely M-PSK, or M-QAM. Then, a serial-to-
parallel buffer segments the information sequence into frames, where the bits in the frames
are split into N groups, i.e. each subcarrier is assigned data that will be transmitted. Then an
OFDM block follows. Here the spectrum is converted back to its time domain signal using an
Inverse Fourier Transform. The Inverse Fast Fourier Transform (IFFT) performs this
transformation and provides a simple way of ensuring the orthogonality of subcarriers. The
guard period may be added to the beginning of each symbol. The use of a cyclic prefix is also
optional, which both preserves the orthogonality of the tones and eliminates intersymbol
interference (ISI) between consecutive OFDM symbols.

Figure 1. Block diagram of an OFDM system modeled with Matlab/Simulink

The generated signal waveform then passes through the Rayleigh channel with AWGN. This
channel parameters can easily be adjusted for a wide variety. The channel output is fed to the
receiver. The receiver basically does the reverse operation of the transmitter. First, the guard
period and cyclic prefix are removed from the received signal, if they have been added. The
FFT of each symbol is then taken to find the original transmitted spectrum. Finally, according
to modulation scheme, the phase angle and/or amplitude of each carrier is evaluated and
converted back to the data word. Finally the data words are then combined to obtain the
original transmitted data. The performance is analyzed by comparing the transmitted and
received data and calculating BERs.

3. Simulation results
In this section, we analyze the performance of the OFDM simulator explained in the previous
section under the assumption of AWGN. The number of subcarriers in each OFDM symbol is
selected to be 128, and frames are sent through the channel with modulation M-PSK and M-
QAM. The BER results are calculated after 1000000 bits.

139
First we compare the performance of OFDM system with PSK and QAM modulation schemes
with various modulation levels. For this part, we assume that channel effects are removed
completely and we analyze the performance of modulation with different levels only. Figure 2
and Figure 3 give BER versus SNR graphs for PSK modulation with M = 2, 4 and 8 levels
and QAM modulation with M = 4, 8 and 16 levels for 0 - 30 dB SNR range, respectively. As
expected, for both modulation schemes as SNR increases the performance gets better. Also as
the level increases the BER increases, since the signals in the cancellation diagram get closer.

Figure 2. BER vs SNR graph, PSK modulation, M=2, 4, 8

Figure 3. BER vs SNR graph, QAM modulation, M=4, 8, 16

140
In an OFDM system, the generated waveform is said to be disturbed during the transmission
through the channel, since the channel introduces amplitude and phase shifts due to frequency
selective and time varying nature. The question is that if the receiver needs to take into
account these unknown changes to acquire the original bits or not. Therefore we investigate
the performance of the OFDM system under the Rayleigh fading channel and AWGN
assumption. The transmitted signals are 4-PSK modulated. Two multipath Rayleigh fading
channels are generated for the simulation. The first one is modeled as a two tapped delay line,
with path delays [0 10-4] and path gains [0 0.4]. The second channel is a five tapped
multipath, a heavier one, with path delays [0 10-4 2.10-5 3.10-6 4.10-7] and path gain vector [0
0.4 0.3 0.2 0.1]. The constellation diagrams related to AWGN channel only, and Rayleigh
channels with 2-tapped and 5 tapped multipath are given in Figure 4, Figure 5 and Figure 6,
respectively. All the plots are given for SNR = 30 dB and maximum Doppler shift is selected
to be very low (10-8 Hz) in order to see the effect of multipath only. It is easily seen from the
figures that the performance loss is intolerable even for slight multipath case. (The BER
performance is expected to be very bad compared to AWGN channel.) Thus the OFDM
receiver needs estimation and equalizing, i.e, the channel factors should be estimated and
compensated prior the detector in the case of multipath.

Figure 4. Constellation diagram, 4-PSK, AWGN channel, SNR=30 dB.

141
Figure 5. Constellation diagram, 4-PSK, 2-tapped delay channel with AWGN, SNR=30 dB.

Figure 6. Constellation diagram, 4-PSK, 5-tapped delay channel with AWGN, SNR=30 dB.

142
The estimation and equalization methods that can be used depend on the channel and the
desired complexity of the receiver. There are two basic channel estimation methods in the
literature: The Least Square (LS) algorithm and the Linear Minimum Mean Square Error
(LMMSE) algorithm [9, 10]. The goal of LS estimator is to minimize the square distance
between the received and the original signal. The LS estimation method is simple, but since it
does not use the related features of the channel, the accuracy of the estimation is low. The
LMMSE estimator minimizes the estimation mean square errors and has a better accuracy.
This algorithm needs additional information, such as SNR and channel characteristics and has
a higher complexity. Although the LMMSE estimator is more complex, we suggest using it
because of its performance especially for heavy multipath channel conditions.

4. Conclusions
We built an OFDM simulation model using interactive environment of Matlab/Simulink tool
which gives opportunity for studying the OFDM system and it can be enlarged easily because
of the modular structure. Using this model, the performance of OFDM system is analyzed for
two typical digital modulation techniques, namely M-PSK and M-QAM. The results obtained
for BER versus SNR graphs show low BER values for larger value of SNR, as expected. It is
also observed that although high value of M increases spectrum efficiency, this modulation
schemes are easily affected by noise. Thus, OFDM systems with higher M-ary modulation
schemes may be used for large capacity applications but with the cost of an increase in BER.
We have also considered the performance of the OFDM system under Rayleigh fading
channel. Simulation results show that the receiver is in need of channel estimation and
equalizing blocks, with a cost of the complexity, for a satisfactory performance. In the
Simulink model proposed, according to actual needs, a suitable channel estimation algorithm
can be added. Although it will increase the complexity, preferring LMMSE estimator is
advised because there is a great degradation in the performance of the receiver especially
under 5-tap Rayleigh channel and it will provide better performance.

References
[1] Patel, Jigisha N. and Upena D. Dalal. “A Comparative Performance Analysis of OFDM
using MATLAB Simulation with M-PSK and M-QAM Mapping,” International
Conference on Computational Intelligence and Multimedia Applications, Sivakasi,
Tamil Nadu, India, December 13-15, 2007.
[2] Zhang, Jun and Zhongpei Zhang, “Simulation and Analysis of OFDM System Based on
Simulink,” International Conference on Communications, Circuits and Systems,
International Conference Center, UESTC, Chengdu, China, July 28-30, 2010.
[3] T. P., Surekha, T. Ananthapadmanabha and C. Puttamadappa, “Modeling and
Performance Analysis of QAM-OFDM System with AWGN Channel”, Third Pacific-
Asia Conference on Circuits, Communications and System, Wuhan, China, July 17-18,
2011.
[4] Pokle, S. B. and K. D. Kulat Fiete, “ MATLAB Simulation of a Wireless
Communication System using OFDM Principle”, IETE Technical Review, 23:3 (2006),
187-198. (Published online: 26 Nov 2015)
[5] Mahor, Devesh and Kamlesh Kumar Yadav, “Investigating Mapping Technique Over
Statistical Parameter under Noisy Channel for Wireless Communication”, 5th

143
 International Conference on Confluence - The Next Generation Information Technology
Summit, Noida, India, Sept.25-26, 2014.
[6] Ogale, Archana, Shubhangi Chaoudhary and A. J. Patil, “Performance Evaluation of
MIMO-OFDM System using Matlab / Simulink With Real Time Image Input”, 10th
IEEE and IFIP International Conference on Wireless and Optical Communications
Networks, Bhopal, India, July 26-28, 2013.
[7] Qiang Li, Tong and Li Gang Xu, “Study and Simulation Analysis of OFDM System
Based on Simulink”, 3rd International Conference on Mechatronics and Intelligent
Materials: Advanced Materials Research, 706:708 (2013) 1967-1970.
[8] http://www.mathworks.com/help/simulink/, The Matlab Simulink website.
[9] Khlifi, Abdelhakim and Ridha Bouallegue, “Performance Analysis of LS and LMMSE
Channel Estimation Techniques for LTE Downlink Systems”, International Journal of
Wireless & Mobile Networks: Networking and Internet Architecture, 3: 5 (2011), 141-
149. (Published online: 7 Nov 2011)
[10] Zettas, Spiridon, Pavlos. I. Lazaridis and Zaharias D. Zaharis, et al. “Performance
Comparison of LS, LMMSE and Adaptive Averaging Channel Estimation (AACE) for
DVB-T2”, 10th IEEE International Symposium on Broadband Multimedia Systems and
Broadcasting, Ghent, Belgium, June 17-19, 2015.

144