Simulation of WiMAX Systems
Onsy Abdel Alim Hiba S. Abdallah
Prof., Faculty of Engineering, Eng., Faculty of Engineering,
Elect. Eng. Dept., Elect. Eng. Dept.,
Beirut Arab University, Beirut Arab University,
Lebanon Lebanon
onsy20@hotmail.com hiba abdallah@yahoo.com
ABSTRACT-WiMAX (Worldwide Interoperability for
Microwave Access) has the potential to impact all forms of
telecommunications. In a fixed wireless communication, WiMAX
can replace the telephone company's copper wire networks, the
cable TV's coaxial cable infrastructure while offering Internet
Service Provider services. In its mobile variant, WiMAX has the
potential to replace cellular networks. However, it still facing
real challenge for low complexity and efficient system
implementation. It supports NLOS environment with high data
rate transmission and high mobility up to 125km/hr.
In this paper, two main issues are discussed. The first one
presents models for simulating OFDM WiMAX system in
Simulink including channel estimation and equalization
subsystems in MATLAB functions. Next, the effect of channel
estimation error on the performance of MIMO VBLAST
receivers in uncorrelated Rayleigh flat fading channels is
investigated.
keywords
WiMAX, MIMO, VBLAST, OFDM, Channel
Estimation, Equalization, Doppler Effect
I. BACKGROUND
WiMAX is based on Wireless Metropolitan Area Networking
(WMAN) standards developed by the IEEE 802.16 group and
adopted by both IEEE and ETSI HIPERMAN group. It provides very
high data throughput over long distance in a point-tomultipoint and
line of sight (LOS) or non-line of sight (NLOS) environments.
WiMAX can provide seamless wireless services up to 20 or 30 miles
away from the base station. The IEEE 802.16 group subsequently
produced 802.16a, which include NLOS applications in the 2GHz-
11GHz band, using an orthogonal frequency division multiplexing
(OFDM)-based physical layer. The WiMAX forum has two different
system profiles: a new standard in 2004 based on IEEE 802.16-2004,
OFDM PHY, and is called the fixed system profile [1], the other one
based on IEEE 802.16e-2005 scalable OFDMA PHY, called the
mobility system profile [1]. The Mobile WiMAX specification
defines multiple-input multiple-output (MIMO) option, which is a
key feature in mobile WiMAX.
OFDMA allows multiple-antenna operations to be performed on
vector-flat sub-carriers. Complex equalizers are not required to
compensate for frequency selective fading. OFDMA therefore, is
very well suited to support smart antenna technologies. In fact,
IMOOFDM/OFDMA is envisioned as the foundation for next-
generation broadband communication systems. This paper is
organized as follows. The end-to-end WiMAX system model
978-1-4244-1754-4/08/$25.00 ©2008 IEEE
11
Azza M. Elaskary
Eng., National Center for
Radiation and Technology,
(NCRRT)
Egypt
azza48@hotmail.com
including channel estimation and equalization to facilitate evaluation
of performance in single-input single-output fixed/mobile system is
introduced in section II. An overview of MIMO and MIMO WiMAX
is presented in section III, while in section IV the performance of
different MIMO receivers under the effect of imperfect channel
estimation is simulated. In section V, we state some conclusions and
possible future work.
II. FIXED/MOBILE WIMAX USING SINGLE
ANTENNA
The WiMAX physical layer is based on OFDM. OFDM is the
transmission scheme of choice to enable high-speed data, video, and
multimedia communications. The WiMAX forum has two different
system profiles: one based on IEEE 802.16-2004, OFDM PHY,
called the fixed system profile; the other one based on IEEE 802.16e-
2005 scalable OFDMA PHY, called the mobility system profile. In
this section, we briefly describe these standards, the channel
estimation and equalization approaches that are used in this paper,
the WiMAX system simulation and finally results and discussion are
provided.
A. WIMAX PHYSICAL LAYER
Fixed WiMAXOFDM-PHY
For this version the FFT size is fixed at 256, which 192
subcarriers used for carrying data, 8 used as pilot subcarriers for
channel estimation and synchronization purposes, and the rest used
asguard band subcarriers. Since the FFT size is fixed, the subcarrier
spacing varies with channel bandwidth. When larger bandwidths are
used, the subcarrier spacing increases, and the symbol time
decreases. Decreasing symbol time implies that a larger fraction
needs to be allocated as guard time to overcome delay spread.
WiMAX allows a wide range of guard times that allow system
designers to make appropriate trade-offs between spectral efficiency
and delay spread robustness. For maximum delay spread robustness,
a 25 percent guard can be used, which can accommodate delay
spreads up to 16 pts when operating in a 3.5MHz channel and up to 8
pts when operating in a 7MHz channel as shown in table 1 [1].

Table 1. Table Represent WiMAX Phy. Layer Specifications
Mobile WiMAX0FDMA-PHY
Mobile WiMAX is intended for the 2.3 GHz, 2.5 GHz and
3.5GHz spectra. The system is defined so that the user can travel at
speeds between 0-125 km/h. The theoretical upper limit for the bit
rate in WiMAX, given a bandwidth of 1OMHz, is 31 Mbps in
downlink and 23 Mbps in uplink [1]. The base stations have a typical
coverage up to an 8 km radius in a NLOS environment. In Mobile
WiMAX, the FFT size is scalable from 128 to 2,048. Here, when the
available bandwidth increases, the FFT size is also increased such
that the subcarrier spacing is always 10.94 kHz. The subcarrier
spacing of 10.94 kHz was chosen as a good balance between
satisfying the delay spread and Doppler spread requirements for
operating in mixed fixed and mobile environments. This subcarrier
spacing can support delay-spread values up to 200 pts and Vehicular
mobility up to 125 kmph when operating in 3.5 GHz. Also, a
subcarrier spacing of 10.94 kHz implies that 128, 512, 1024, and
2,048 FFT are used when the channel bandwidth is 1.25MHz, 5MHz,
10MHz, and 20MHz, respectively. The OFDM symbol frequency
domain description in WiMAX is illustrated in Fig. 1 [1].
Figure I OFDMA Sub-Carrier Structure
In Fig. 1, total of 256 subcarriers are spread to four parts.
There are 192 subcarriers for data transmission, 8 subcarriers for
pilot tone, 52 subcarriers for guard bands, and 1 subcarrier for DC
in every OFDM symbol.
B. CHANNEL ESTIMATION & EQ UALIZATION
There are two maj or kinds of channel estimators that are
found in literature, namely pilot assisted and blind estimation. In
training sequence methods or non-blind methods, the transmitted
data and
training sequences known to the receiver are embedded into the
frame and sent through the channel. Main advantages of training
sequences are the conceptual simplicity of the method, which is
facilitated in some new communication standards that includes
12
frames for sending training sequences. Obviously, the drawback
is because the time-slots occupied in the transmission of these
training sequences reduces throughput.
In blind methods, mathematical or statistical properties of
transmitted data are used. They are bandwidth efficient however,
they are significantly slow to converge and require important
computational capacity. A mixture of these two, where a blind
method with limited training symbols is used, is called semi-blind
technique. The semi-blind methods use information from both
training sequence and statistical properties of the transmitted
signal, which makes them more robust than the blind methods
while they still require less training compared to the non-blind
methods. Traditional one-dimensional channel estimation
techniques for the OFDM systems can be summarized as follows:
Least Squares (LS), Minimum Mean squared Error (MMSE) and
Linear MMSE (LMMSE). LS estimators are very simple to
constitute, but they suffer from MSE in low SNR conditions.
MMSE, based on time domain estimators, are high complexity
estimators that provide good performance in sampled-spaced
channels, but limited performance in non-sample spaced channels
and high SNR conditions. The third one, LMMSE provides good
performance in both sampled and non-sampled channels [1].
C. WiMAXSYSTEMSIMULATION
MATLAB and Simulink are used for modeling theWiMAX
OFDM physical layer [1,4]. The WiMAX End-to-End Model is
shown in figure 2. In the transmitter, the input output data vector
is read in and written to MATLAB workspace after each major
function block then it is compared to the standard test vector and
it was identical.
Additional functions such as Inverse Fast Fourier Transformed
(IFFT) and cyclic prefix addition to the OFDM Symbol have been
added and modeled to ensure compatibility. The frequency
change due to the Doppler Effect depends on the relative motion
between the source and receiver. WiMAX specifications are
considered to implement the channel. In our simulation, the "low
value" Doppler frequency is z70 Hz corresponds to a mobile
speed of 20 Km/hr at 3.5GHz carrier frequency and a "high
value" Doppler shift corresponding to the operation at 3.5 GHz
for a velocity of 120 Km/hr is z400 Hz.
In the simulation, the channel has been modeled as a Multipath
Rayleigh fading channel with additive white Gaussian noise
(AWGN), at Simulink multipath fading box, the calculated
Doppler shift has been inserted. The receiver implementation is
an inversion of all the transmitter functions with addition of
channel estimation and equalization parts. The channel estimation
based on comb type pilot arrangement is studied through different
algorithms for both estimating channel at pilot frequencies and
interpolating the channel. The estimation of channel is based on
LS and LMS while the channel interpolation is done using linear
interpolation, second order interpolation, low-pass interpolation,
and spline cubic interpolation [1]. The effect of different
interpolation approaches based on LS and LMS were tested with
low and high relative velocities.
 Figure 2 WiMAX End-to-End Simulink Model

Figure5. BER vs. SNR in dB for LMS with small Doppler

shift (low relative velocity)
Figure3. BER vs. SNR in dB for LS and LMS various interpolation criteria in
fixed conditions

Figure6. BER vs. SNR in dB for LS with high Doppler shift

Figure 4. BER vs. SNR in dB for LS with small Doppler shift (high relative velocity)
(low relative velocity)

Figure7. BER vs. SNR in dB for LMS with high Doppler shift
(high relative velocity)

13
 D. RESULTS AND DISCUSSION
We have simulated the proposed scheme for WiMAX
communication system specifications. The AWGN and Multipath
Doppler shift models are used as testing environment. Comparing
the performances of all schemes by measuring BER versus SNR
with setup as shown in Fig.5, the results are shown in figures 3 to
7. Figure 3 shows that Channel estimation based on LS algorithm,
with the linear interpolation, the second order interpolation, the
spline cubic interpolation and the low-pass interpolation,
respectively perform about the same as LMS algorithm in fixed
conditions. From figures 4 to 7, it is clear that LS estimator
performance degrades at high relative velocity and LMS has
better performance in mobile case. In general, the channel
estimator that performed the best in terms of lowest BER is the
Lowpass interpolation algorithm.
III. MULTIPLE ANTENNAS IN WiMAX
SYSTEMS
A. MIMO
Demand for capacity in wireless communications has been
rapidly increasing worldwide. Nevertheless, the available
spectrum is limited and capacity needs cannot be met without a
significant increase in spectral efficiency. Advances in channel
coding make it feasible to approach the Shannon capacity limit in
systems with single antenna links (SISO-Single-Input Single-
Output).
As a solution to the high capacity requirements of future
wireless systems, a lot of awareness has been drawn to multiple-
input multiple-output (MIMO) wireless systems with multiple
antennas at both the transmit and receive sides by increasing the
number of antennas at both the transmitter and the receiver [2]. In
general, capacity grows linearly with the number of transmit
antennas, Nt, as long as the number of receive antennas, Nr,
satisfies the condition Nr > Nt as shown in Fig.8. As a result,
MIMO systems are excellent candidates for highdata-rate future
mobile systems, such as 3G, WiMAX and beyond.
t3~~~~~~~-
D ,~~~~ J I
S __,-------
1- 4 4- T|-0-
r* w~1 4-
0 I I I
Figure 8. Outage capacity for various antenna configuration
B. MIMO WIMAX
One of the WiMAX system profiles is the simple Space-
Time coding (STC) scheme proposed by Alamouti for transmit
diversity on the downlink. In the IEEE 802. 16e-2005
specifications, this scheme is referred to as "Matrix A" which
improves the reliability of data transmission for mobile modems.
A MIMO cell site sends multiple, redundant copies of a data
stream to the receiver to increase the probability that some of
theom mav suerviveblthe pyicalpath betnmwenframnmisiodnandm
reception in a good enough state to allow reliable decoding. These
conditions are most likely to occur while the terminal is highly
mobile, with rapid signal fading and multipath reception.
Figure 9.Comparison of Alamouti/MRC with 2x2 spatial
multiplexing [3]
The second multiple antenna profile included in WiMAX
systems is the 2x2 MIMO technique based on the so called
"Matrix B". This system performs spatial multiplexing (SM) and
does not offer any diversity gain from the transmit side. But it
does offer a diversity gain of 2 on the receiver side when detected
using Maximum-Likelihood (ML) detection [2]. SM system sends
each data frame only once allowing full independent usage of
antennas, however it gives limited benefit and is not always the
best transmission scheme for a given BER target. Without
naturally occurring multi-path, MIMO Matrix B signal can
destructively combine, resulting in no signal at the receiver.
Natural multi-path is common in highly cluttered environments,
such as dense urban areas.
The tradeoff between Matrix A and Matrix B will be left to
future investigations. Figure 10 shows the results on an
uncorrelated Rayleigh fading channel when the Alamouti/MRC
scheme uses 16-QAM and the spatial multiplexing scheme uses
QPSK (4 bits per symbol period in both cases). It can be observed
that the zero-forcing (ZF) receiver does not exploit the diversity
of the spatial multiplexing scheme and that the slope of its BER
4
1
-
curve is only half that of the Maximum Likelihood (ML) receiver
as shown in Fig.9. The other major observation is that the slope of
the Alamouti/MRC scheme is twice as large as that of the spatial
multiplexing ML receiver, which is due to the diversity factor of
4 for the former and of 2 for the latter [3].
IV. MIMO WITH IMPERFECT CHANNEL ESTIMATION
An example of a pure multiplexing scheme is the Vertical-
Bell Labs layered space-time (V-BLAST). The main idea of the
V-BLAST architecture is to split the information bit stream into
several substreams and transmit them in parallel using a set of
transmit antennas (the number of transmit antennas equals the
number of substreams) at the same time and frequency. At the Rx
end, each Rx antenna sees all the transmitted signals, which are
mixed due to the nature of the wireless propagation channel.
Using appropriate signal processing at the receiver, these signals
can be separated so that the matrix wireless channel is
14
transformed into a set of virtual parallel independent channels
(provided that the multipath is rich enough). Previous researches
analyzed tight closed-form BER approximation of MIMO ZF
receiver in an uncorrelated Rayleigh flat fading channels under
the assumption that perfect CSI is available at the receiver.
However, noisy channel estimates arise in any practical system,
4srls N:
=
-- -N
--
..........
rst=....
--- -; f-X
~~~~~~~..............
..
........... . .. , ...
..- --- -- --;,------ -.--- ----.-- .
;;~~ ~ ~~~~~~~~
~X5 ;B---S------;--X;;X-;U-UL;0--
2
LH..N..pS..'_t. - t
Figure 1O. BER performance comparison of QPSK MIMO V-BLAST ZF receiver with Nt = 2, Nr = 3 versus Nt = 2,Nr
4 and Nt = 2. Nr = 6 for 0%. 1%. 5% nmse error
so it is important to evaluate the performance of MIMO receivers
in the presence of imperfect CSI.
The standard baseband system model is given by:
r=Hs+n, Ki) |
where s denotes the transmitted data symbol vector consisting of
Nt symbols each with a constellation size M and E[SSH]=ES Nt I,
while n N(O, No INt) is the additive Gaussian noise vector, and
-
ES is the transmitted symbol energy per transmit antenna. In order
to retrieve the transmitted data symbol vector at the receiver,
accurate knowledge of the complex channel gain matrix H is
required. However, in practice, imperfect channel estimates arise
in any practical system. The noisy channel state information is
modeled as [2]:
H = 1 --e H+eQ.
Where eQ is the estimation error that is uncorrelated with H,
the entries of Q are independent and identically distributed
(i.i.d.) zeromean complex Gaussian random variables with
unity variance. The factor p = 1 - e2 represents the correlation
coefficient between the actual channel gain and its estimate,
which is assumed to be the same for all channel gains.
Parameter e E [0,1] is the measure of quality of channel state
information. A smaller "e" indicates higher correlation and
hence better CSI quality. The normalized mean square error
(nmse) of the channel estimation is defined as
where ij ij h^ h , represent the (i,j)th element of H and H^
respectively. It can be easily shown that the normalized mean
mnse- (tw-1eIz
square is related to e by:
A. SIMULATION RESULTS
The performance of V-BLAST ZF receiver in the presence of
different degrees of CSI imperfection through simulation is
presented.
.D R S t<. N rA S E.- 1 9,., N M '-3 E--
N, 2 P"
hi... 2 --
--------------
.2. N --E5 r-Al--2 P-4 --13
- ..._
.. ..
.. ....
_ .. _.
The effect of the number of transmit antennas Nt on the BER
performance of VBLAST-ZF QPSK in the presence of imperfect
CSI is investigated with D =Nt-Nr. Fig. 10 and fig.1 1 show that as
Nt increases the BER gets worse with the same value of D and
nmse, while the BER performance only depends on D and SNR
when perfect CSI is assumed. Intuitively the reason for the
dependence on Nt is that the inter-stream interference cannot be
cancelled perfectly in the presence of channel estimation error. As
Nt increases, the number of data streams increases. Given the
same transmit power per antenna, the source of the interference
increases. As a result the BER performance gets worse as Nt
increases.
Also, when there is no channel estimation error, as Eb/No -* cc,
(2 BER -, o, while an error floor is expected at high SNR due to
the channel estimation error. Moreover as Nt and/or e increases,
the SNR value corresponding to the starting point of the error
floor decreases. The previous results are compared with the
results obtained using the pilot channel estimation as shown in
Figure 12 [2]. It shows that, for low SNR (-5dB), the
performance of V-BLAST ZF using pilot estimation is equivalent
to 5% nmse error, when the SNR increases, the BER performance
approaches that using perfect CSI.
V. CONCLUSIONS AND FUTURE WORK
In the first part, WiMAX top level simulink with all system
details have been implemented for simulation purpose. Channel
estimation with different interpolation approaches for
fixed/mobile OFDM system with parameters from WiMAX
)I
standards have been illustrated. The results showed that the
Doppler shift had a greater impact on the performance between
the different channel estimators and interpolation approaches. In
the second part, the performance of MIMO VBLAST ZF
receivers over uncorrelated Rayleigh flat fading channels in the
presence of channel estimation error is investigated. It is found
that contrary to the case when there is no channel estimation
15
error, the BER performance is not only a function of D = Nr- [3] www.mathwork.com Recorded Webinar "From a Wireless Standards
Document to an Executable Model using MATLAB and Simulink" July
Nt, but also a function of Nt. 14, 2005.
Besides, for the same value of D and error "e" the BER [4] B.Muquet, E.Biglieri, H. Sari: "Mimo Techniques for Mobile WiMAX
performance gets worse as Nt increases. Another impact of Systems", SEQUANS Commu. White paper, 2007.
imperfect channel estimation is that when SNR is high, BER does [5] 0. Abdel Alim, A. El Naggary:"Performance of MIMO Antenna
not approach zero but approaches an error floor whose value Techniques on IEEE802.16E", ICICT2007, Dec. 16-18, Cairo, Egypt.
depends on Nt, Nrand e.
For Future work, RF planning for different geographical
environments is to be considered for WiMAX systems
implementation [5]. An adaptive technique for switching between
Spatial Multiplexing and Diversity techniques in MIMO OFDM
as defined in the standard WiMAX systems will be investigated.
Also, the effect of imperfect channel estimation will be
considered

w3LA

Figure 11. BER performance of 16QAM MIMO V-BLAST ZF

receiver with D=2 and D =4.

Figure 12 BER performance of V-BLAST ZF receiver for QPSK

modulated signal for 2x2 system.

REFERENCES

[1] 0. Abdel Alim, N. Elboghdadly, M.Ashour, A. Elaskary:" Channel

Estimation and Equalization for Fixed/Mobile OFDM WiMAX System
in Simulink", Mobilware'08, Feb. 12-15, 2008, Austria.
[2] H. S. Abdallah, I. A. Ghaleb, S. Abou. Chahine: "The Effect of
Imperfect Channel Estimation on the Performance of Layered Space
Time Receivers", IEEE ISSPIT'07, Dec. 15-18, 2007, Cairo, Egypt.

16