P. 1
a1

a1

|Views: 23|Likes:

More info:

Published by: ਸ. ਇੰਦਰਦੀਪ ਸਿੰਘ on Aug 02, 2012
Copyright:Attribution Non-commercial

Availability:

Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less

07/14/2013

pdf

text

original

620 J. OPT. COMMUN. NETW./VOL. 3, NO. 8/AUGUST 2011 Mesleh et al.

On the Performance of Different OFDM
Based Optical Wireless Communication
Systems
Raed Mesleh, Hany Elgala, and Harald Haas
Abstract—This paper analyzes the performance of indoor
orthogonal frequency division multiplexing (OFDM) optical
wireless communication systems in the presence of light
emitting diode (LED) nonlinear distortions. There are several
forms of optical OFDM using intensity modulation [7th
Int. Symp. on Communication Systems Networks and Digital
Signal Processing (CSNDSP), 2010, pp. 566–570]. In this paper,
DC-biased optical OFDM (DCO-OFDM) and asymmetrically
clipped optical OFDM (ACO-OFDM) are considered. ACO-
OFDM produces a half-wave symmetry time signal at the
output of the OFDM modulator by special assignment of
subcarriers, thus allowing signal clipping at the zero level and
avoiding the need for DC bias at the expense of data rate
reduction. DCO-OFDM assigns data to all possible subcarriers
to increase the data rate. However, half-wave symmetry signals
cannot be achieved and a high DC bias is needed to convert
the bipolar signal to a unipolar signal before modulating the
LED intensity. This paper considers a practical LED model and
studies the performance of both systems in terms of average
electrical OFDM signal power versus bit error ratio in the
presence of an additive white Gaussian noise (AWGN) channel.
In addition, DC power consumption and the transmitted
optical power for the two systems are compared. The analytical
results are validated through Monte Carlo simulations and
the obtained results demonstrate close match. It is shown
that LED clipping has significant impact on the performance
of both systems and an optimum system design should take
into account the OFDM signal power, DC-bias point, and LED
dynamic range.
Index Terms—ACO-OFDM; DCO-OFDM; Direct detection; In-
tensity modulation; LED nonlinearity; OFDM; Optical wireless
communication.
I. INTRODUCTION
S
hort-range communication is one of the most relevant
as well as diversified fields of endeavor in wireless
communications. As such, it has been a subject of intense
research and development worldwide, particularly in the last
decade. Concepts such as wireless social networks, inter-car
Manuscript received March 17, 2011; revised June 22, 2011; accepted July 1,
2011; published July 29, 2011 (Doc. ID 144119).
Raed Mesleh (e-mail: raed.mesleh@ieee.org) is with the Electrical Engineer-
ing Department and SNCS Research Center, University of Tabuk, P.O. Box
6592-2, 47315-4031 Tabuk, Saudi Arabia.
Hany Elgala is with Jacobs University Bremen, Campus Ring 1, Research I,
28759 Bremen, Germany.
Harald Haas is with Jacobs University Bremen, Campus Ring 1, Research I,
28759 Bremen, Germany, and is also with the Institute for Digital Communica-
tions, Joint Research Institute for Signal and Image Processing, The University
of Edinburgh, Edinburgh EH9 3JL, UK.
Digital Object Identifier 10.1364/JOCN.3.000620
communications, home and office networking, wireless grids,
and personal communications heavily rely on short-range
wireless links.
In recent years, interest in optical wireless communication
(OWC) as a promising complementary technology for radio
frequency (RF) in short-range communications has gained new
momentum [1,2]. The abundance of unregulated bandwidth
at the optical frequencies has prompted interest in optical
wireless (OW) technology as a viable candidate to cope with
the future demand of indoor wireless access generated through
real-time bandwidth-intensive applications such as voice over
IP (VoIP), streaming video and music, and network attached
storage (NAS). Also, there are certain locations where the
use of RF based systems is ill-advised, such as in hospitals
and installations containing navigation equipment such as
in airplanes. The OWC technology does not suffer from such
restrictions.
An OWC communication system relies on optical radiation
to convey information in free space. The transmitter/source
converts the electrical signal to an optical signal and the
receiver/detector converts the optical power into electrical
current. OWC indoor applications prefer the use of light
emitting diodes (LEDs) as light sources due to relaxed
safety regulations, low cost, and high reliability compared
to laser diodes (LDs). Simple and low cost optical carrier
modulation and demodulation are usually achieved through
intensity modulation (IM) with direct detection (DD). The
desired waveform is modulated onto the instantaneous power
of the optical carrier and the detector generates a current
proportional to the received instantaneous power, i.e., only
the intensity of the optical wave is detected and there is no
frequency or phase information [3].
The performance of OWC systems depends on the propa-
gation and type of system used. The basic system types fall
into diffuse or line of sight (LOS) systems. In LOS systems,
high data rates of the order of Gbits/s can be achieved [4], but
the systems are vulnerable to blockage/shadowing because of
their directionality. In a diffuse OWC system, several paths
from source to receiver exist, which makes the systems robust
to blockage/shadowing. However, the path losses are high
and multipaths create inter-symbol interference (ISI) which
limits the achievable data rate [3,5]. A promising solution to
combat multipath distortion and boost the data rate without
any bandwidth or power expansion is by using an OFDM
(orthogonal frequency division multiplexing) technique.
OFDM for OWC systems was proposed in [6–11] to
support high data rates through parallel transmission of high
1943-0620/11/080620-09/$15.00 © 2011 Optical Society of America
Mesleh et al. VOL. 3, NO. 8/AUGUST 2011/J. OPT. COMMUN. NETW. 621
S(n) =
1

N
N−1

k=0
_
ℜ(s(k)) cos
_

k
N
n
_
−ℑ(s(k)) sin
_

k
N
n
__
+j
1

N
N−1

k=0
_
ℜ(s(k)) sin
_

k
N
n
_
+ℑ(s(k)) cos
_

k
N
n
__
,
. ¸¸ .
=0
n =0, 1, . . . , N−1, (2)
s =
_
0 s
0
0 s
1
· · · 0 s
N/4−1
0 s

N/4−1
0 · · · s

1
0 s

0
_
T
. (3)
Box 1.
order multi-level quadrature amplitude modulation (M-QAM)
symbols on orthogonal subcarriers. OFDM systems are able
to support high data rates without the need for complex
channel equalizers and the time-varying channel can be
easily estimated using frequency-domain channel estimation.
In general, the output of the OFDM modulator is complex
and bipolar. In IM optical systems, quadrature modulation is
not possible. This means that the baseband signal must be
real. Also, the OFDM signal envelope variations are utilized
to intensity modulate the LED and the bipolar signal must be
converted to a unipolar signal. Therefore, OFDM as commonly
used in RF communications must be modified.
This paper analyzes the performance of two existing OFDM
modulation techniques suitable for OWC communication. The
performance of these systems is simulated via Monte Carlo
simulations and validated analytically in the presence of an
additive white Gaussian noise (AWGN) channel and LED
clipping distortions. Clipping at the maximum allowable AC
voltage (corresponding to the maximum allowable forward
current) and at the LED turn-on voltage (TOV) obtained from
the LED data sheet are considered.
The remainder of this paper is organized as follows.
Section II discusses the considered OWC OFDM techniques.
Performance analysis along with the LED model are presented
in Section III. Simulation and analytical results are presented
in Section IV. Finally, Section V concludes the paper.
II. OPTICAL WIRELESS OFDM TECHNIQUES
A general system model of indoor OWC OFDM systems is
depicted in Fig. 1.
The information stream, q, is first parsed into a block of
complex data symbols denoted by x. The complex symbols
are drawn from QAM/PSK (phase shift keying)/PAM (pulse
amplitude modulation) constellations. These complex symbols
are then mapped onto the following vector:
_
s
k
_
N−1
k=0
=
_
0 {x
k
}
N/2−1
k=
1
0 {x

k
}
1
k=N/2−1
_
, (1)
where (·)

denotes the complex conjugate. The Hermitian
symmetry property of the vector s is needed to create a real
output signal that is used to modulate the LED intensity.
The OFDM modulator applies an N-point inverse fast Fourier
transform (IFFT) on the vector s and adds a cyclic prefix (CP)
creating the real-time signal S. The output of the discrete IFFT
can be written as shown in Eq. (2), given in Box 1, where
j =

−1 is the imaginary unit, ℜ(·) denotes the real part of
a complex number, and ℑ(·) denotes the imaginary part of
a complex number. The CP is needed to avoid inter-carrier
interference (ICI) as well as inter-block interference (IBI) by
converting the linear convolution with the channel into a
circular one. The resulting time signal is used to modulate the
intensity of an LED. However, the intensity cannot be negative
and the bipolar time signal S must be converted to unipolar
before modulating the LED intensity.
Based on the bipolar–unipolar conversion process, two OWC
OFDM techniques are reported in the literature, namely,
ACO-OFDM [9–12] and DCO-OFDM [6–8]. Details of each
OFDM technique will be discussed in what follows.
A. ACO-OFDM Technique
In ACO-OFDM, only odd subcarriers are modulated, as
given in Eq. (3) given in Box 1, where (·)
T
denotes the transpose
of a vector or a matrix. The values of the first and the N/2
subcarriers must be real to ensure that the output consists of
only real values [8,10]. Therefore, the achieved data rate for an
ACO-OFDM system is given by
R
{ACO}
=
_
N/4−1
N+N
g
_
B log
2
M bits/s, (4)
where B denotes the bandwidth, N
g
denotes the number of
guard subcarriers used in the CP, and M is the modulation
order. Since s contains data on the odd subcarriers only, the
OFDM modulator produces a half-wave symmetry real-time
signal S. An example of S assuming N = 16 is depicted in
Fig. 2. The half-wave symmetry of S means that the same
information in the first N/2 samples is repeated in the second
half of the OFDM symbol. As a consequence, the negative part
can be clipped without any loss of information. This clipping
produces a unipolar signal. The intermodulation caused by
clipping occurs only in the even subcarriers and does not affect
the data-carrying odd subcarriers. However, it reduces the
amplitude by a factor of two [10]. The unipolar signal is then
converted to an analog signal through the digital-to-analog
converter (D/A) and used to modulate the intensity of an LED.
At the receiver, the photodiode (PD) detects the transmitted
intensity and the analog signal is converted back to digital.
622 J. OPT. COMMUN. NETW./VOL. 3, NO. 8/AUGUST 2011 Mesleh et al.
PD (Photo diode)
LED (Light emiting diode)
Data
sink
De-modulator
Data
subcarrier
extraction
OFDM
demodulator
A/D AWGN
Data
source
Modulator
Subcarrier
assignment
OFDM
modulator
Bipolar–unipolar
conversion
D/A
q(k)
q(k)
x (k) s (k) S (n)
S
~
(n)
x (k)
Y (n) s (k)
Fig. 1. (Color online) Indoor OWC OFDM system model.
The received signal before the analog-to-digital converter (A/D)
is given by
Y=h⊗S+w, (5)
where the ⊗ denotes linear convolution, h =
[h(0) h(1) · · · h(L−1)]
T
is the L-path impulse response
of the optical channel which is modeled as discussed in detail
in [13, Eq. (13)], and w is an AWGN that represents the sum
of the receiver thermal noise and shot noise due to ambient
light with overall noise power denoted by σ
2
n
[3, Eq. (21)]. It
is important to note that the noise is added in the electrical
domain; hence, the received signal Y can be negative as well as
positive. So, unlike the transmitted signal, the received signal
is bipolar instead of unipolar. The CP is first removed and the
linear convolution is converted to circular convolution,
˜
Y=h
˜
S+ ˜ w, (6)
where ˜ w,
˜
Y, and
˜
S are the signals without the CP and the
denotes circular convolution. To demodulate the signal, an
N-point fast Fourier transform (FFT) is taken, yielding
ˆ y =Hˆ s+ ˆ w, (7)
where H is an N×N diagonal matrix whose diagonal elements
are the N-point FFT of h and ˆ y, ˆ s, and ˆ w are the N-point FFTs
of
˜
Y,
˜
S, and ˜ w respectively. The odd subcarriers are extracted
from ˆ y to yield
ˆ y
o
=H
o
ˆ s
o
+ ˆ w
o
, (8)
where H
o
is an N/2 × N/2 diagonal matrix whose diagonal
contains the odd elements of the diagonal of H, ˆ y
o
and ˆ w
o
are the vectors composed of the odd elements of ˆ y and ˆ w,
respectively, and ˆ s
o
is
ˆ s
o
=
1
2
_
ˆ s
0
ˆ s
1
· · · ˆ s
N/4−1
ˆ s

N/4−1
ˆ s

N/4−2
· · · ˆ s

0
_
T
.
(9)
A zero-forcing (ZF) equalizer is used to mitigate the effect of
the channel,
ˆ x =2
_
H
H
o
H
o
_
−1
H
H
o
ˆ y
o
, (10)
where (·)
H
and (·)
−1
are the Hermitian transpose and the
inverse of a vector or a matrix, respectively. A hard decision
is made on the estimated symbols ˆ x to retrieve the transmitted
information bits.
B. DCO-OFDM Technique
The second technique assigns data to all odd and even
subcarriers as follows [6–8]:
s =
_
s
0
s
1
. . . s
N/2−1
0 s

N/2−1
· · · s

1
s

0
_
T
.
(11)
Hence, the achieved data rate for the DCO-OFDM system is
given by
R
{DC−biased}
=
_
N/2−1
N+N
g
_
B log
2
M bits/s. (12)
The OFDM modulator is applied to s producing the real-time
signal S. An example of S for N = 16 is depicted in Fig. 2.
The time signal is bipolar and DC bias is needed to shift
the negative values to positive values before modulating
the LED intensity. After adding the DC bias, all values
which create a voltage drop across the LED that is less
than the TOV are clipped. The DC-bias value depends on
the LED characteristics, the OFDM signal envelope which
depends on the considered modulation order, and the number
of OFDM subcarriers and can significantly affect system
performance [14–16]. The processing of the time signal is
similar to the above discussed signal processing of ACO-OFDM
and is skipped here for the sake of brevity.
III. PERFORMANCE ANALYSIS
A. LED Model
In this paper, a white LED (Golden DRAGON, ZW W5SG)
from OSRAM is considered [17]. The maximum allowed
forward current is 500 mA. At 350 mA, the recommended
biasing point, the LED has a maximum luminous flux of 71 lm,
Mesleh et al. VOL. 3, NO. 8/AUGUST 2011/J. OPT. COMMUN. NETW. 623
Time samples
A
m
p
l
i
t
u
d
e

(
V
)
DCO–OFDM
Time samples
A
m
p
l
i
t
u
d
e

(
V
)
ACO–OFDM
–0.06
–0.04
–0.02
0
0.02
0.04
0.06
0.08
2 4 6 8 10 12 14 16
2 4 6 8 10 12 14 16
0.1
–0.05
0
0.05
0.1
Fig. 2. (Color online) The output OFDM time signals for the
ACO-OFDM and DCO-OFDM systems for N = 16. Half-wave
symmetry signals are achieved for ACO-OFDM. For DCO-OFDM,
however, a DC bias must be used.
a typical luminous intensity of 17 cd, and a 120

viewing angle
at 50% of the luminous intensity. In this paper, the relation
between the forward voltage across the LED and the current
through the LED is modeled through a polynomial using the
least-square curve fitting technique.
Each LED has a minimum threshold value known as
the TOV, c
l
, which is the onset of current flow and light
emission [18]. The TOV is considered at 100 mA, which
corresponds to a 2.75 V drop across the LED. This is the
minimum allowed forward current according to the data sheet.
The LED outputs light that is linear with the drive current.
However, thermal aspects causing a drop in the electrical-
to-optical conversion efficiency must be considered [18].
Hence, AC/pulsed currents must be adjusted according to the
manufacturer’s data sheet to ensure that the LED chip does
not overheat, in order to avoid degradation in output light or, in
the worst case, total failure. The maximum allowed AC/pulsed
current is considered to be 1 A, which corresponds to a 4 V drop
across the LED, i.e., the maximum forward voltage, c
u
, of the
measured data sheet transfer curve.
A polynomial of the second degree shows a good fit to the
LED transfer characteristic, as shown in Fig. 3. The curve
shows the nonlinear behavior of the LED using the developed
polynomial for forward voltage amplitudes in the range from
2.75 V up to 4 V.
A predistorter uses the LED inverse characteristics as a
nonlinear compensator to condition the OFDM signal prior
to the LED modulation [19]. Through predistortion, a linear
response curve is achieved over a large range of the input
signal amplitudes. However, the region which can be linearized
is limited. The maximum input amplitude that will be
modulated linearly depends upon the maximum permissible
AC/pulsed current through the LED. In Fig. 4, the value
of the TOV is subtracted from the values of the forward
voltages, i.e., a 0 V corresponds to the 2.75 V. The dashed
10
–1
10
0
Forward voltage (V)
F
o
r
w
a
r
d

c
u
r
r
e
n
t

(
A
)
Data sheet values
Polynomial curve
(a)
(b)
2.5 3 3.5 4 4.5
10
3
5
10
2
5
10
1
mA
I
F
V
F
OHL02520
2.0 2.5 3.0 3.5 4.0 4.5
V
Fig. 3. (Color online) (a) Discrete data sheet values and the LED
polynomial curve. (b) The data sheet curve.
curve in Fig. 4 shows the nonlinear behavior of the LED
using the developed polynomial for forward voltage amplitudes
in the range from 0 V up to 1.25 V. Therefore, and for the
considered LED, forward voltages below 0 V and above 1.25 V
are clipped, i.e., corresponding to the TOV and the voltage at
the maximum allowed AC/pulsed current. For the considered
LED, predistortion linearizes the LED response over the range
from 0.25 V up to 1.25 V. The solid curve in Fig. 4 illustrates
the linearized V–I relation. Therefore, forward voltages below
0.25 V and above 1.25 V are clipped.
Finally, regarding the optical power, most data sheets of
white LEDs provide only the photometric power, namely,
luminous flux in lumens or the luminous intensity measured
in candela, which are useful metrics for illumination design.
Luminous flux measures the light’s luminosity as perceived
by the human eye, whose sensitivity varies for different
wavelengths. However, the radiometric power in watts is a
624 J. OPT. COMMUN. NETW./VOL. 3, NO. 8/AUGUST 2011 Mesleh et al.
x: 0.25
y: 0.1107
Forward voltage (V)
x: 1.25
y: 1.011
F
o
r
w
a
r
d

c
u
r
r
e
n
t

(
A
)
LED curve (polynomial model)
LED linearized curve
–0.2
0
0.2
0.4
0.6
0.8
1
1.2
0 0.2 0.4 0.6 0.8 1 1.2
Fig. 4. (Color online) The V–I dashed curve using the developed LED
polynomial function after subtracting the TOV from the values of the
forward voltages across the LED. The linearized V–I solid curve with
the predistorter.
I–Radiant flux
Forward current (A)
R
a
d
i
a
n
t

f
l
u
x

(
W
)
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0.5
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Fig. 5. (Color online) The relation between the forward current
through the LED and the radiated power in watts.
more relevant parameter for wireless transmission. Therefore,
for the considered LED, luminous fluxes in watts are
determined. The luminous efficacy (measured in lumens per
watt) is used for this purpose. The luminous efficacy measures
the amount of luminous flux achieved with a given amount
of radiant flux. The theoretical limit for white LEDs is
300 lm/W [20], which is assumed in this paper for conducting
simulations. The relation between the forward current through
the LED and the radiated power in watts is shown in Fig. 5.
This curve is obtained from the data sheet curve relating the
current through the LED to the luminous flux normalized
to the luminous flux at 350 mA, i.e., 71 lm at 350 mA and
300 lm/W luminous efficacy.
Denoting the polynomial equation relating the forward
voltage V and the forward current I shown in Fig. 4 as f
1
(V)
and the polynomial equation relating the output optical power
p
o
and I as f
2
(I), the output optical power for any input voltage
is calculated as follows:
p
o
= f
2
( f
1
(V)) W. (13)
B. Optical OFDM Techniques Performance Analysis
The OFDM signal exhibits a high peak-to-average power
ratio (PAPR). The high PAPR values in OFDM stem from
the superposition of a large number of usually statistically
independent sub-channels that can constructively sum up
to high signal peaks in the time domain. Hence, careful
adjustment of the OFDM signal envelope is required to
control distortion levels. Additionally, the LED should be
biased at an optimum biasing point to ensure minimum
distortion, while fulfilling power consumption and cell coverage
requirements [15,16,19,21,22]. The analysis in this paper
provides a simple analytical investigation of the effect of these
parameters on the performance of optical OFDM techniques.
An OFDM signal is the sum of N independent subcarriers.
For large values of N (N > 10), the ensemble average of S can
be accurately modeled as a Gaussian random process (due to
the central limit theorem) with zero mean and variance σ
2
equal to the total OFDM electrical signal power. Thereby, the
probability that, at any given time instance, S
n
takes the value
z is given by
p(S
n
= z) = p(z) =
1
_
2πσ
2
exp
_

z
2

2
_
. (14)
Simulation results show that the OFDM signal constella-
tions, in the presence of an LED, experience amplitude and
phase distortions in a random fashion [23]. According to the
Bussgang theorem, the amplitude distortion of a Gaussian
input results in an uncorrelated additive noise [24]. Hence, by
modeling this nonlinearity induced noise as Gaussian noise, a
simple definition of the effective SNR, ρ, as a function of the
nonlinearity induced noise power, σ
2
clip
, is given by
ρ =
OFDM signal power
Effective noise power
=
σ
2
σ
2
n

2
clip
, (15)
where σ
2
clip
is given by
σ
2
clip

2
uc

2
lc
, (16)
with σ
2
uc
being the noise component due to the clip-
ping/attenuation of the upper peaks of the OFDM signal, and
σ
2
lc
being the noise component due to the clipping/attenuation
of the lower peaks of the OFDM signal.
In ACO-OFDM, the LED is biased at the value of c
l
and the
negative part of the signal is clipped, producing a noise that is
orthogonal to the signal; hence, σ
2
lc
=0. The signal peaks larger
than the value of c
u
are also clipped, which produces an overall
Mesleh et al. VOL. 3, NO. 8/AUGUST 2011/J. OPT. COMMUN. NETW. 625
clipping distortion power given by
σ
2
clip
=
_

c
u
(z −c
u
)
2
p(z) dz. (17)
For the DCO-OFDM system, signal levels above c
u
and
below c
l
are clipped, producing an overall clipping noise power
given by
σ
2
clip
=
_

c
u
(z −c
u
)
2
p(z) dz+
_
c
l
−∞
_
z −c
l
_
2
p(z) dz. (18)
The overall noise power is then obtained by adding the clipping
noise power and the thermal noise power as given in Eq. (15).
The bit error ratio (BER) for any M-QAM constellation can
be calculated as given in Eq. (19) (see Box 2) [25], where
U is the number of constellation points on the x-axis and
J is the number of constellation points on the y-axis for
any rectangular QAM. For instance, U = 4 and J = 2 for a
rectangular 8-QAM.
IV. SIMULATION RESULTS
The performances of ACO-OFDM and DCO-OFDM systems
in the presence of LED clipping distortion are analyzed
analytically and through Monte Carlo simulations. The LED
model discussed in the previous section is considered. In all
results, the x-axis represents the average electrical OFDM
signal power before modulating the LED. The OFDM signal
power is varied from 0 dBm to 30 dBm and an AWGN power of
−10 dBm is assumed. As a result, the simulated electrical SNR
range is from 10 dB to 40 dB, which is within the reported SNR
values for indoor OWC systems [22,26,27]. In all analyses, a
channel bandwidth of B =20 MHz, a number of subcarriers of
N =1024, and a number of guard interval subcarriers of N
g
=4
are considered [3,8]. The simulated channel parameters are the
same as the configuration A parameters shown in [13, Table 1].
The BER performance for different QAM modulation orders
is depicted in Fig. 6 for ACO-OFDM and in Fig. 7 for
DCO-OFDM.
By observing the performance of both systems, the following
trends can be observed:
• In both figures, dashed lines represent analytical results,
whereas solid lines represent Monte Carlo simulation
results. Both results demonstrate a very close match for
a wide range of SNRs and different modulation orders.
For Monte Carlo simulation results, the average BERs
for at least 10
6
transmitted bits are calculated for each
SNR value. The number of simulated bits increases with
increasing SNR.
• At low OFDM signal powers, the performance is noise
dominant, while clipping noise dominates at high signal
power. In all considered systems, it can be found that there
is an optimum power. The discontinuity in some of the BER
curves indicates that the simulated BER is less than 10
−8
for the corresponding OFDM signal powers. The optimum
value depends on the considered LED characteristics and
modulation order. Therefore, system design must consider
10
–8
10
–7
10
–6
10
–5
10
–4
10
–3
10
–2
10
–1
10
0
Electrical OFDM average signal power (dBm)
B
i
t

e
r
r
o
r

r
a
t
i
o
8-QAM–ana
8-QAM–sim
16-QAM–ana
16-QAM–sim
32-QAM–ana
32-QAM–sim
64-QAM–ana
64-QAM–sim
256-QAM–ana
256-QAM–sim
0 5 10 15 20 25 30
Fig. 6. (Color online) ACO-OFDM BER performance for different
QAM modulation orders.
10
–8
10
–7
10
–6
10
–5
10
–4
10
–3
10
–2
10
–1
10
0
Electrical OFDM average signal power (dBm)
B
i
t

e
r
r
o
r

r
a
t
i
o
4-QAM–ana
4-QAM–sim
8-QAM–ana
8-QAM–sim
16-QAM–ana
16-QAM–sim
32-QAM–ana
32-QAM–sim
64-QAM–ana
64-QAM–sim
256-QAM–ana
256-QAM–sim
0 5 10 15 20 25 30
Fig. 7. (Color online) DCO-OFDM BER performance for different
QAM modulation orders.
the LED characteristics and optimize the transmitted
signal power to avoid significant performance degradation.
• The performance enhancement for all systems, when
increasing the power from 0 dBm to the optimum value, can
be explained by the increase in the SNR and the absence of
signal clipping at low amplitudes.
• Similarly, the performance degradation when increasing
the signal power beyond the optimum value is due to
clipping distortion.
• The optimum value for the ACO-OFDM system appears
at larger signal power as compared to the DCO-OFDM
system for the same modulation order. This behavior can be
explained by noting that the ACO-OFDM system utilizes a
larger LED dynamic range as compared to the DCO-OFDM
system since it requires a lower DC-bias value and the
clipping distortion is only due to clipping of the upper peaks.
However, the achieved data rate for the same modulation
order is not the same for both systems. For instance,
626 J. OPT. COMMUN. NETW./VOL. 3, NO. 8/AUGUST 2011 Mesleh et al.
BER=
_
¸
¸
¸
¸
_
¸
¸
¸
¸
_

M−1

M log
2

M
erfc
__

2(M−1)
_
, square M−QAM
1
log
2
(U×J)
_
U−1
U
erfc
_
_

U
2
+J
2
−2
_
+
J −1
J
erfc
_
_

U
2
+J
2
−2
__
, M=U×J rectangular QAM.
(19)
Box 2.
ACO-OFDM with 64-QAM modulation order achieves a
data rate of about 29.77 Mbps while the DCO-OFDM
system achieves a data rate of 59.65 Mbps for the same
modulation order. If the ACO-OFDM system was designed
to achieve the same data rate as the DCO-OFDM system, a
constellation order of M = 4069 would be needed, which is
impractical.
In addition, DC power consumption and the average
output optical power are not the same either, as shown in
Fig. 8 for ACO-OFDM and in Fig. 9 for the DCO-OFDM
system. ACO-OFDM requires less DC power since it
needs biasing at the TOV only. However, for visible light
communication (VLC), larger DC-bias values are needed
to fulfill the illumination requirements and the DC power
consumption will be comparable for ACO-OFDM and
DCO-OFDM. The transmitted optical power is, however,
higher for DCO-OFDM since the DC bias shifts the signal
to higher values.
• ACO-OFDM transmitted optical power increases with
increase of the input electrical power and starts to flatten at
larger input power values (>24 dBm in Fig. 8). In contrast,
the transmitted average optical power of the DCO-OFDM
system starts to decay with increase of the input OFDM
signal power. This is because the DCO-OFDM system is
biased at the average LED value and, due to clipping
distortions, larger input signal leads to larger clipping
distortion and more loss in transmitted optical power.
• High modulation orders, such as 256-QAM and higher, are
very sensitive to clipping distortions and increasing signal
powers beyond the optimum value becomes futile.
The analysis in this paper allows a simple method
to optimize system parameters for OFDM optical wireless
transmission techniques. For example, the considered DC-bias
system above uses a bias point b =0.625 V, which corresponds
to the recommended bias point for this LED from the
manufacturer. However, this might not be the optimum bias
point for communication. In Fig. 10, the BERs for the 16-QAM
DC-bias system with different bias points are depicted. It is
shown that the b =0.525 V bias point performs slightly better
than the considered b = 0.625 V bias point and with reduced
DC power consumption.
V. CONCLUSIONS
This paper has analyzed the performance of existing indoor
OWC OFDM transmission techniques. Existing techniques are
categorized based on the way data is assigned to subcarriers
and the bipolar–unipolar conversion process. The ACO-OFDM
Electrical OFDM signal power (dBm)
A
v
e
r
a
g
e

t
r
a
n
s
m
i
t
t
e
d

o
p
t
i
c
a
l

p
o
w
e
r

(
W
)
ACO-DC power consumption
ACO clipping at saturation voltage
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0.22
0.24
0 5 10 15 20 25 30
Fig. 8. (Color online) ACO-OFDM DC power consumption and the
average output optical power for different input OFDM electrical
signal powers.
0 5 10 15 20 25 30
0.3
0.31
0.32
0.33
0.34
0.35
0.36
Electrical OFDM average signal power (dBm)
A
v
e
r
a
g
e

t
r
a
n
s
m
i
t
t
e
d

o
p
t
i
c
a
l

p
o
w
e
r

(
W
)
DCO-AC output optical power–ideal
DCO-DC power consumption
Fig. 9. (Color online) DCO-OFDM DC power consumption and the
average output optical power for different input OFDM electrical
signal powers.
OWC technique produces a half-wave symmetry time signal
and allows clipping at the zero level, thus reducing DC power
consumption by avoiding the need for a DC bias. These
advantages are achieved at the expense of a major reduction
in data rate as compared to DCO-OFDM systems. The
performance of these systems in the presence of LED clipping
clearly highlights the significant dependence on the considered
Mesleh et al. VOL. 3, NO. 8/AUGUST 2011/J. OPT. COMMUN. NETW. 627
10
–15
10
–10
10
5
10
0
SNR (dB)
B
i
t

e
r
r
o
r

r
a
t
i
o
0 5 10 15 20 25 30
b = 0 .625 V
b = 0 .725 V
b = 0.525 V
b = 0.425 V
Fig. 10. (Color online) DCO-OFDM optimum bias point analysis. A
16-QAM constellation is considered and the BER versus electrical
OFDM signal power is calculated for different bias points.
modulation order and other system parameters, such as bias
point value. It was shown that AWGN noise dominates at low
SNR values and clipping distortion dominates at large SNR
values. Therefore, system design should consider LED clipping
effects and should optimize the OFDM signal power and the
considered modulation order.
The analysis in this paper allows a simple method to
optimize system parameters. An example was shown where
an optimum DC-bias point is found for the considered LED
in this paper. It is shown that the optimum DC-bias point
for communication is different from the optimum DC-bias
point suggested by the manufacturer, which mainly considers
optimum brightness, long lifetime, and minimum temperature.
ACKNOWLEDGMENT
Dr. Raed Mesleh gratefully acknowledges the support for
this work from the Deanship of Scientific Research at the
University of Tabuk.
The authors acknowledge the support for this work from
the German Federal Ministry of Economics and Technology
(BMWi) as part of the Lufo 2
nd
Call project SINTEG.
REFERENCES
[1] S. Randel, F. Breyer, S. C. J. Lee, and J. W. Walewski, “Advanced
modulation schemes for short-range optical communications [in-
vited paper],” IEEE J. Sel. Top. Quantum Electron., vol. 16, no. 5,
pp. 1280–1289, 2010.
[2] A. M. Street, P. N. Stavrinou, D. C. O’brien, and D. J. Edwards,
“Indoor optical wireless systems—a review,” Opt. Quantum Elec-
tron., vol. 29, no. 3, pp. 349–378, Mar. 1997.
[3] J. M. Kahn and J. R. Barry, “Wireless infrared communications,”
Proc. IEEE, vol. 85, no. 2, pp. 265–298, Feb. 1997.
[4] M. Akbulut, C. Chen, M. Hargis, A. Weiner, M. Melloch, and
J. Woodall, “Digital communications above 1 Gb/s using 890-nm
surface-emitting light-emitting diodes,” IEEE Photon. Technol.
Lett., vol. 13, no. 1, pp. 85–87, Jan. 2001.
[5] J. M. Kahn, W. J. Krause, and J. B. Carruthers, “Experimental
characterization of non-directed indoor infrared channels,” IEEE
Trans. Commun., vol. 43, no. 234, pp. 1613–1623, Feb.–Mar.–Apr.
1995.
[6] Y. Tanaka, T. Komine, S. Haruyama, and M. Nakagawa, “In-
door visible light data transmission system utilizing white LED
lights,” IEICE Trans. Commun., vol. E86-B, no. 8, pp. 2440–2454,
Aug. 2003.
[7] M. Z. Afgani, H. Haas, H. Elgala, and D. Knipp, “Visible light
communication using OFDM,” in Proc. of the 2nd Int. Conf. on
Testbeds and Research Infrastructures for the Development of
Networks and Communities (TRIDENTCOM), IEEE, Barcelona,
Spain, Mar. 1–3, 2006, pp. 129–134.
[8] H. Elgala, R. Mesleh, H. Haas, and B. Pricope, “OFDM visible
light wireless communication based on white LEDs,” in Proc. of
the 64th IEEE Vehicular Technology Conference (VTC), Dublin,
Ireland, Apr. 22–25, 2007.
[9] J. Armstrong and B. J. C. Schmidt, “Comparison of asymmet-
rically clipped optical OFDM and DC-biased optical OFDM in
AWGN,” IEEE Commun. Lett., vol. 12, no. 5, pp. 343–345, May
2008.
[10] J. Armstrong and A. Lowery, “Power efficient optical OFDM,”
Electron. Lett., vol. 42, no. 6, pp. 370–372, Mar. 2006.
[11] J. Armstrong, “OFDM for optical communications,” J. Lightwave
Technol., vol. 27, no. 3, pp. 189–204, 2009.
[12] S. C. J. Lee, S. Randel, F. Breyer, and A. M. J. Koonen,
“PAM-DMT for intensity-modulated and direct-detection optical
communication systems,” IEEE Photon. Technol. Lett., vol. 21,
no. 23, pp. 1749–1751, Dec. 2009.
[13] J. Barry, J. Kahn, W. Krause, E. Lee, and D. Messerschmitt,
“Simulation of multipath impulse response for indoor wireless
optical channels,” IEEE J. Sel. Areas Commun., vol. 11, no. 3,
pp. 367–379, Apr. 1993.
[14] R. Mesleh, H. Elgala, and H. Haas, “An overview of indoor
OFDM/DMT optical wireless communication systems,” in 7th Int.
Symp. on Communication Systems Networks and Digital Signal
Processing (CSNDSP), July 21–23, 2010, pp. 566–570.
[15] H. Elgala, R. Mesleh, and H. Haas, “A study of LED nonlinearity
effects on optical wireless transmission using OFDM,” in Proc.
6th IEEE Int. Conf. on Wireless and Optical Communications
Networks (WOCN), Cairo, Egypt, Apr. 28–30, 2009.
[16] H. Elgala, R. Mesleh, and H. Haas, “Non-linearity effects and
predistortion in optical OFDM wireless transmission using
LEDs,” Int. J. Ultra Wideband Commun. Syst., vol. 1, no. 2,
pp. 143–150, 2009.
[17] OSRAM GmbH, Datasheet: ZW W5SG Golden DRAGON White
LED, Aug. 2010 [Online]. Available: http://www.osram.de.
[18] E. Schubert, Light-Emitting Diodes, 1st ed. Cambridge Univ.
Press, 2003.
[19] H. Elgala, R. Mesleh, and H. Haas, “Modeling for predistortion
of LEDs in optical wireless transmission using OFDM,” in Proc.
of the IEEE 10th Int. Conf. on Hybrid Intelligent Systems (HIS),
Shenyang Liaoning, China, Aug. 12–14, 2009.
[20] L. Zyga, “White LEDs with super-high luminous efficacy could
satisfy all general lighting needs,” PhysOrg.com, Aug. 2010 [On-
line]. Available: http://www.physorg.com/news202453100.html.
[21] I. Neokosmidis, T. Kamalakis, J. W. Walewski, B. Inan, and
T. Sphicopoulos, “Impact of nonlinear LED transfer function on
discrete multitone modulation: Analytical approach,” J. Light-
wave Technol., vol. 27, no. 22, pp. 4970–4978, 2009.
[22] K.-D. Langer, J. Grubor, O. Bouchet, M. El Tabach, J. Walewski,
S. Randel, M. Franke, S. Nerreter, D. O’Brien, G. Faulkner,
628 J. OPT. COMMUN. NETW./VOL. 3, NO. 8/AUGUST 2011 Mesleh et al.
I. Neokosmidis, G. Ntogari, and M. Wolf, “Optical wireless
communications for broadband access in home area networks,”
in Proc. of the 10th Anniversary Int. Conf. Transparent Optical
Networks (ICTON 08), Athens, Greece, June 22–26, 2008, vol. 4,
pp. 149–154.
[23] H. Elgala, R. Mesleh, and H. Haas, “Practical considerations for
indoor wireless optical system implementation using OFDM,”
in Proc. of the IEEE 10th Int. Conf. on Telecommunications
(ConTel), Zagreb, Croatia, June 8–10, 2009.
[24] J. Bussgang, “Cross correlation function of amplitude-distorted
Gaussian signals,” Tech. Rep. 216, Research Laboratory for Elec-
tronics, Massachusetts Institute of Technology, Cambridge, MA,
Mar. 1952.
[25] K. Cho and D. Yoon, “On the general BER expression of one-
and two-dimensional amplitude modulations,” IEEE Trans. Com-
mun., vol. 50, no. 7, pp. 1074–1080, July 2002.
[26] J. Grubor, S. Randel, K. Langer, and J. Walewski, “Bandwidth
efficient indoor optical wireless communications with white light
emitting diodes,” in Proc. of the 6th Int. Symp. on Communication
Systems, Networks and Digital Signal Processing, Graz, Austria,
June 23–25, 2008, vol. 1, pp. 165–169.
[27] D. O’Brien, G. Parry, and P. Stavrinou, “Optical hotspots speed
up wireless communication,” Nat. Photonics, vol. 1, pp. 245–247,
2007.
Raed Mesleh (S’00–M’08) received his B.Sc.
degree in communication engineering from
Yarmouk University, Irbid, Jordan, in 2000,
his M.Sc. degree in communication technology
from Ulm University, Ulm, Germany, in 2004,
and his Ph.D. degree in electrical engineering
from Jacobs University, Bremen, Germany,
in 2007. In October 2010, he joined the
University of Tabuk in Saudi Arabia where he
is now an Assistant Professor in the Electrical
Engineering Department and a researcher at the Sensor Networks
and Cellular System (SNCS) research center. From September 2007
until September 2010 he was with the School of Electrical Engineering
and Computer Science, Jacobs University, Bremen, Germany, where
he worked as a Postdoctoral Fellow. His main research interests are
in wireless communication and optical wireless communication with
particular focus on MIMO techniques. He is a co-inventor of seven
patents, three of which are already granted, in the field of RF and
optical wireless communication systems.
Dr. Mesleh received a tuition waiver for his Ph.D. study at Jacobs
University along with a three-year scholarship. In 2003, he received
the award for the best student performance in the M.Sc. course at
Ulm University (LEG award), and, in 2000, he received the award for
the best graduate communication engineer in year 2000 from Yarmouk
University. Dr. Mesleh joined Orange in Amman, Jordan (2000–2001),
as a System Engineer and worked as a Consultant for Nokia in Saudi
Arabia from 2001 to 2002. Dr. Mesleh serves as a technical reviewer for
several IEEE transaction journals and international conferences and
he is a TPC member for several international conferences in wireless
communications.
Hany Elgala (S’02–M’10) received his Ph.D.
degree fromJacobs University Bremen in 2010.
He received his B.Sc. degree in electronics and
communications from Ain-shams University
in 2000 and completed his M.Sc. degree in
microsystems engineering from Furtwangen
University in 2003. He is currently a Postdoc-
toral Fellow at the Department of Electrical
Engineering and Computer Science at Jacobs
University. Since 2005 he has been conducting
research on indoor optical wireless communication. His main research
interests are in the areas of communication systems, digital signal
processing, and circuit design.
Harald Haas (SM’98–AM’00–M’03) received
his Ph.D. degree from the University of
Edinburgh in 2001. From 2001 to 2002 he
was a Research Project Manager at Siemens
in Munich. Prof. Haas joined Jacobs University
Bremen in 2002 as an Associate Professor,
before returning in 2007 to The University
of Edinburgh where he now holds a personal
Chair of Mobile Communications.
His research interests are in interference
coordination and mitigation in wireless networks, the spatial modula-
tion multiple antenna concept, and optical wireless communication.
He holds more than 15 patents in the area of wireless
communications, and has published more than 170 conference and
journal papers. He has co-authored a book entitled Next Generation
Mobile Access Technologies: Implementing TDD with Cambridge
University Press. Professor Haas was an invited speaker at the
TEDGlobal Conference in 2011.

ACO-OFDM [9–12] and DCO-OFDM [6–8]. Clipping at the maximum allowable AC voltage (corresponding to the maximum allowable forward current) and at the LED turn-on voltage (TOV) obtained from the LED data sheet are considered. ACO-OFDM Technique In ACO-OFDM. it reduces the amplitude by a factor of two [10]. q. (4) II. Section II discusses the considered OWC OFDM techniques. NETW. Based on the bipolar–unipolar conversion process. The resulting time signal is used to modulate the intensity of an LED. However. the OFDM modulator produces a half-wave symmetry real-time signal S. VOL. The Hermitian symmetry property of the vector s is needed to create a real output signal that is used to modulate the LED intensity. Therefore. (3) given in Box 1. OPT. as given in Eq. The intermodulation caused by clipping occurs only in the even subcarriers and does not affect the data-carrying odd subcarriers. where (·)T denotes the transpose of a vector or a matrix. . These complex symbols are then mapped onto the following vector: sk k=0 = 0 N −1 N /2 {xk }k= −1 1 0 {x∗ }1 = N /2−1 . quadrature modulation is not possible. . two OWC OFDM techniques are reported in the literature. Performance analysis along with the LED model are presented in Section III. 8/AUGUST 2011/J. The half-wave symmetry of S means that the same information in the first N/2 samples is repeated in the second half of the OFDM symbol. Simulation and analytical results are presented in Section IV. As a consequence. where j = −1 is the imaginary unit. creating the real-time signal S. . N g denotes the number of guard subcarriers used in the CP. 1. (2). . the achieved data rate for an ACO-OFDM system is given by R {ACO} = N/4 − 1 B log2 M N + Ng bits/s. N − 1. The output of the discrete IFFT can be written as shown in Eq.10]. given in Box 1. An example of S assuming N = 16 is depicted in Fig. Finally. The complex symbols are drawn from QAM/PSK (phase shift keying)/PAM (pulse amplitude modulation) constellations. The performance of these systems is simulated via Monte Carlo simulations and validated analytically in the presence of an additive white Gaussian noise (AWGN) channel and LED clipping distortions. In general. (2) s= 0 s0 0 s1 ··· 0 s N /4−1 0 s∗ /4−1 N 0 ··· s∗ 1 0 s∗ 0 T . The unipolar signal is then converted to an analog signal through the digital-to-analog converter (D/A) and used to modulate the intensity of an LED. The information stream. The values of the first and the N/2 subcarriers must be real to ensure that the output consists of only real values [8. In IM optical systems. only odd subcarriers are modulated. The remainder of this paper is organized as follows. order multi-level quadrature amplitude modulation (M-QAM) symbols on orthogonal subcarriers.Mesleh et al. namely. the negative part can be clipped without any loss of information. 1. the OFDM signal envelope variations are utilized to intensity modulate the LED and the bipolar signal must be converted to a unipolar signal. COMMUN. and ℑ(·) denotes the imaginary part of a complex number. 3. This paper analyzes the performance of two existing OFDM modulation techniques suitable for OWC communication. k k (1) where (·)∗ denotes the complex conjugate. (3) Box 1. Section V concludes the paper. Therefore. Since s contains data on the odd subcarriers only. ℜ(·) denotes the real part of a complex number. This means that the baseband signal must be real. OFDM systems are able to support high data rates without the need for complex channel equalizers and the time-varying channel can be easily estimated using frequency-domain channel estimation. the photodiode (PD) detects the transmitted intensity and the analog signal is converted back to digital. At the receiver. A. OFDM as commonly used in RF communications must be modified. is first parsed into a block of complex data symbols denoted by x. 621 S (n) = k k ℜ (s (k)) cos 2π n − ℑ (s (k)) sin 2π n N N N k=0 1 N −1 k k +j ℜ (s (k)) sin 2π n + ℑ (s (k)) cos 2π n N N N k=0 =0 1 N −1 . However. . This clipping produces a unipolar signal. and M is the modulation order. Details of each OFDM technique will be discussed in what follows. The CP is needed to avoid inter-carrier interference (ICI) as well as inter-block interference (IBI) by converting the linear convolution with the channel into a circular one. Also. 2. The OFDM modulator applies an N-point inverse fast Fourier transform (IFFT) on the vector s and adds a cyclic prefix (CP) where B denotes the bandwidth. NO. the intensity cannot be negative and the bipolar time signal S must be converted to unipolar before modulating the LED intensity. n = 0. the output of the OFDM modulator is complex and bipolar. O PTICAL W IRELESS OFDM T ECHNIQUES A general system model of indoor OWC OFDM systems is depicted in Fig.

S. an N-point fast Fourier transform (FFT) is taken. NO. and w are the N-point FFTs ˜ ˜ ˜ of Y. ˆ x = 2 HH H o o −1 HH yo . a white LED (Golden DRAGON. ˆ respectively. NETW. and the number of OFDM subcarriers and can significantly affect system performance [14–16]. An example of S for N = 16 is depicted in Fig. COMMUN. the achieved data rate for the DCO-OFDM system is given by R {DC−biased} = N/2 − 1 B log2 M N + Ng bits/s. At 350 mA. A hard decision ˆ is made on the estimated symbols x to retrieve the transmitted information bits. and w is an AWGN that represents the sum of the receiver thermal noise and shot noise due to ambient light with overall noise power denoted by σ2 [3. The odd subcarriers are extracted ˆ from y to yield ˆ ˆ ˆ y o = Ho s o + w o . ˆ s N /4−1 ˆN s∗ /4−1 ˆN s∗ /4−2 ··· ˆ0 s∗ T . the OFDM signal envelope which depends on the considered modulation order.622 J. ˜ ˜ ˜ Y = h S + w. The time signal is bipolar and DC bias is needed to shift the negative values to positive values before modulating the LED intensity. The DC-bias value depends on the LED characteristics. It n is important to note that the noise is added in the electrical domain. s. s N /2−1 0 s∗ /2−1 N ··· s∗ 1 s∗ 0 T . P ERFORMANCE A NALYSIS A. the received signal is bipolar instead of unipolar. The received signal before the analog-to-digital converter (A/D) is given by Y = h ⊗ S + w. (13)]. Eq./VOL. the recommended biasing point. After adding the DC bias. the received signal Y can be negative as well as positive. the LED has a maximum luminous flux of 71 lm. hence. unlike the transmitted signal.. OPT. To demodulate the signal. respectively. The maximum allowed forward current is 500 mA. and w respectively. ZW W5SG) from OSRAM is considered [17].. (8) where H o is an N/2 × N/2 diagonal matrix whose diagonal ˆ ˆ contains the odd elements of the diagonal of H. Subcarrier Data Modulator assignment source x (k) q (k) OFDM modulator s (k) Bipolar–unipolar D/A conversion ~ S (n) S (n) Data sink q (k) De-modulator x (k) Data subcarrier extraction OFDM demodulator s (k) A/D AWGN Y (n) LED (Light emiting diode) PD (Photo diode) Fig. h = [h(0) h(1) · · · h(L − 1)]T is the L-path impulse response of the optical channel which is modeled as discussed in detail in [13. 3. and s o is ˆ so = 1 ˆ s0 2 ˆ s1 ··· The OFDM modulator is applied to s producing the real-time signal S. (9) A zero-forcing (ZF) equalizer is used to mitigate the effect of the channel. yielding ˆ y = Hˆ + w. (Color online) Indoor OWC OFDM system model. (6) B. 2. (21)]. (5) where (·)H and (·)−1 are the Hermitian transpose and the inverse of a vector or a matrix. III. and S are the signals without the CP and the denotes circular convolution. y o and w o ˆ ˆ are the vectors composed of the odd elements of y and w. (11) Hence. So. 8/AUGUST 2011 Mesleh et al. LED Model In this paper. 1. Y. Eq. s ˆ (7) where H is an N × N diagonal matrix whose diagonal elements ˆ ˆ ˆ are the N-point FFT of h and y. o ˆ (10) . (12) ˜ ˜ ˜ where w. The processing of the time signal is similar to the above discussed signal processing of ACO-OFDM and is skipped here for the sake of brevity. where the ⊗ denotes linear convolution. all values which create a voltage drop across the LED that is less than the TOV are clipped. DCO-OFDM Technique The second technique assigns data to all odd and even subcarriers as follows [6–8]: s = s0 s1 . The CP is first removed and the linear convolution is converted to circular convolution.

3. 3. However.25 V are clipped.05 0 –0.75 V up to 4 V.04 – 0. total failure.5 Fig.25 V. 8/AUGUST 2011/J. c l .e. corresponding to the TOV and the voltage at the maximum allowed AC/pulsed current. The TOV is considered at 100 mA. A predistorter uses the LED inverse characteristics as a nonlinear compensator to condition the OFDM signal prior to the LED modulation [19]. The dashed 102 5 101 2. as shown in Fig. i.1 2 4 6 10 –1 8 10 Time samples 12 14 16 2.06 0.05 0.0 V 4. the maximum forward voltage.0 3. 623 DCO–OFDM 0.25 V and above 1.06 (a) 10 0 Amplitude (V) 2 4 6 8 10 Time samples ACO–OFDM 12 14 16 Forward current (A) Data sheet values Polynomial curve 0. A polynomial of the second degree shows a good fit to the LED transfer characteristic. COMMUN. namely. in order to avoid degradation in output light or.75 V drop across the LED. a 0 V corresponds to the 2. forward voltages below 0. however.e. whose sensitivity varies for different wavelengths.75 V.25 V.e. (Color online) (a) Discrete data sheet values and the LED polynomial curve. For the considered LED. 3.0 2. i. which corresponds to a 4 V drop across the LED.5 4 Forward voltage (V) OHL02520 4.5 4. the value of the TOV is subtracted from the values of the forward voltages.08 0. This is the minimum allowed forward current according to the data sheet.1 Amplitude (V) 0. In Fig.25 V up to 1.02 – 0. NO. a DC bias must be used. The maximum input amplitude that will be modulated linearly depends upon the maximum permissible AC/pulsed current through the LED. The solid curve in Fig. The LED outputs light that is linear with the drive current. (Color online) The output OFDM time signals for the ACO-OFDM and DCO-OFDM systems for N = 16.02 0 – 0. Each LED has a minimum threshold value known as the TOV. AC/pulsed currents must be adjusted according to the manufacturer’s data sheet to ensure that the LED chip does not overheat. VOL. 4. luminous flux in lumens or the luminous intensity measured in candela. Luminous flux measures the light’s luminosity as perceived by the human eye. which are useful metrics for illumination design.. Therefore. Hence. the relation between the forward voltage across the LED and the current through the LED is modeled through a polynomial using the least-square curve fitting technique. 4 illustrates the linearized V–I relation. The maximum allowed AC/pulsed current is considered to be 1 A. For DCO-OFDM.5 VF Fig. NETW. i. (b) The data sheet curve.. Through predistortion. 2. forward voltages below 0 V and above 1. predistortion linearizes the LED response over the range from 0. 4 shows the nonlinear behavior of the LED using the developed polynomial for forward voltage amplitudes in the range from 0 V up to 1. regarding the optical power. c u . However. which corresponds to a 2. curve in Fig. and a 120◦ viewing angle at 50% of the luminous intensity. Therefore. which is the onset of current flow and light emission [18]. However. OPT. in the worst case. Half-wave symmetry signals are achieved for ACO-OFDM. and for the considered LED.5 3 3.5 3.04 0. of the measured data sheet transfer curve. the region which can be linearized is limited.. In this paper. Finally. The curve shows the nonlinear behavior of the LED using the developed polynomial for forward voltage amplitudes in the range from 2. (b) IF 103 mA 5 a typical luminous intensity of 17 cd. most data sheets of white LEDs provide only the photometric power.25 V are clipped. thermal aspects causing a drop in the electricalto-optical conversion efficiency must be considered [18].Mesleh et al. the radiometric power in watts is a . a linear response curve is achieved over a large range of the input signal amplitudes.

2 0./VOL. Thereby. ρ .9 1 p (S n = z) = p (z) = exp − .2 0 –0.25 0. in the presence of an LED. experience amplitude and phase distortions in a random fashion [23].2 0.8 0. a simple definition of the effective SNR. 4. the ensemble average of S can be accurately modeled as a Gaussian random process (due to the central limit theorem) with zero mean and variance σ2 equal to the total OFDM electrical signal power. Optical OFDM Techniques Performance Analysis The OFDM signal exhibits a high peak-to-average power ratio (PAPR). σ2 . This curve is obtained from the data sheet curve relating the current through the LED to the luminous flux normalized to the luminous flux at 350 mA. at any given time instance. (15) where σ2 is given by clip σ2 = σ2 + σ2 . 5. The luminous efficacy measures the amount of luminous flux achieved with a given amount of radiant flux. the amplitude distortion of a Gaussian input results in an uncorrelated additive noise [24]. Hence. by modeling this nonlinearity induced noise as Gaussian noise..7 0.3 0. (13) 1 0.4 Radiant flux (W) 0. producing a noise that is orthogonal to the signal. luminous fluxes in watts are determined.5 0.05 0. the output optical power for any input voltage is calculated as follows: p o = f 2 ( f 1 (V )) W.624 J. σ2 = 0. the LED is biased at the value of c l and the negative part of the signal is clipped.6 0.2 LED curve (polynomial model) LED linearized curve x: 1. hence. Additionally. Therefore.4 0.4 0. NETW. and σ2 being the noise component due to the clipping/attenuation lc of the lower peaks of the OFDM signal. The analysis in this paper provides a simple analytical investigation of the effect of these parameters on the performance of optical OFDM techniques. For large values of N (N > 10). In ACO-OFDM. (Color online) The V–I dashed curve using the developed LED polynomial function after subtracting the TOV from the values of the forward voltages across the LED. An OFDM signal is the sum of N independent subcarriers. According to the Bussgang theorem. 5. i. which produces an overall . COMMUN. 71 lm at 350 mA and 300 lm/W luminous efficacy.011 p o and I as f 2 (I). the LED should be biased at an optimum biasing point to ensure minimum distortion.2 0. 8/AUGUST 2011 Mesleh et al. for the considered LED. S n takes the value z is given by 1 2πσ2 z2 2σ2 0 0. careful adjustment of the OFDM signal envelope is required to control distortion levels. (14) Forward current (A) Simulation results show that the OFDM signal constellations.5 0.16.6 0. The signal peaks larger lc than the value of c u are also clipped.25 y: 0. 1.2 x: 0. I–Radiant flux 0. which is assumed in this paper for conducting simulations.25 y: 1. Hence.45 0. Denoting the polynomial equation relating the forward voltage V and the forward current I shown in Fig. the probability that.15 0. NO.4 0. The theoretical limit for white LEDs is 300 lm/W [20].3 0.22]. while fulfilling power consumption and cell coverage requirements [15. The high PAPR values in OFDM stem from the superposition of a large number of usually statistically independent sub-channels that can constructively sum up to high signal peaks in the time domain.19. OFDM signal power Effective noise power σ2 σ2 + σ2 n clip more relevant parameter for wireless transmission. The relation between the forward current through the LED and the radiated power in watts is shown in Fig. The linearized V–I solid curve with the predistorter. as a function of the nonlinearity induced noise power. 4 as f 1 (V ) and the polynomial equation relating the output optical power = .e. (Color online) The relation between the forward current through the LED and the radiated power in watts.8 Forward current (A) 0. is given by clip ρ= Fig.35 0. OPT.21.6 0.1 0.8 1 1. The luminous efficacy (measured in lumens per watt) is used for this purpose. uc clip lc (16) with σ2 being the noise component due to the clipuc ping/attenuation of the upper peaks of the OFDM signal.2 Forward voltage (V) Fig. 3.1 0.1107 B.

The BER performance for different QAM modulation orders is depicted in Fig. while clipping noise dominates at high signal power. (Color online) ACO-OFDM BER performance for different QAM modulation orders. 6. signal levels above c u and below c l are clipped. • The performance enhancement for all systems. Therefore. (18) –5 –6 The overall noise power is then obtained by adding the clipping noise power and the thermal noise power as given in Eq. NO. which is within the reported SNR values for indoor OWC systems [22. In all results. The LED model discussed in the previous section is considered. However. VOL. The optimum value depends on the considered LED characteristics and modulation order. the x-axis represents the average electrical OFDM signal power before modulating the LED. For instance. when increasing the power from 0 dBm to the optimum value. the LED characteristics and optimize the transmitted signal power to avoid significant performance degradation.27].26. (Color online) DCO-OFDM BER performance for different QAM modulation orders. U = 4 and J = 2 for a rectangular 8-QAM. a channel bandwidth of B = 20 MHz. can be explained by the increase in the SNR and the absence of signal clipping at low amplitudes. (15). Table 1]. By observing the performance of both systems. dashed lines represent analytical results. The discontinuity in some of the BER curves indicates that the simulated BER is less than 10−8 for the corresponding OFDM signal powers. the following trends can be observed: • In both figures. As a result. 8/AUGUST 2011/J. • At low OFDM signal powers. 10 10 –7 –8 8-QAM–ana 8-QAM–sim 16-QAM–ana 16-QAM–sim 32-QAM–ana 32-QAM–sim 64-QAM–ana 64-QAM–sim 256-QAM–ana 256-QAM–sim 0 5 10 15 20 25 Electrical OFDM average signal power (dBm) 30 Fig. (19) (see Box 2) [25]. the performance degradation when increasing the signal power beyond the optimum value is due to clipping distortion.Mesleh et al. the achieved data rate for the same modulation order is not the same for both systems. 625 clipping distortion power given by σ2 = clip ∞ 10 0 cu (z − c u )2 p (z) dz. The number of simulated bits increases with increasing SNR. In all considered systems. The simulated channel parameters are the same as the configuration A parameters shown in [13. it can be found that there is an optimum power. For instance. NETW. system design must consider 10 0 10 –1 10 –2 10 –3 10 –4 10 –5 10 –6 10 –7 10 –8 4-QAM–ana 4-QAM–sim 8-QAM–ana 8-QAM–sim 16-QAM–ana 16-QAM–sim 32-QAM–ana 32-QAM–sim 64-QAM–ana 64-QAM–sim 256-QAM–ana 256-QAM–sim Bit error ratio 0 5 10 15 20 25 Electrical OFDM average signal power (dBm) 30 Fig. 6 for ACO-OFDM and in Fig. COMMUN. 7 for DCO-OFDM. IV. 7. and a number of guard interval subcarriers of N g = 4 are considered [3. the performance is noise dominant. The OFDM signal power is varied from 0 dBm to 30 dBm and an AWGN power of −10 dBm is assumed. This behavior can be explained by noting that the ACO-OFDM system utilizes a larger LED dynamic range as compared to the DCO-OFDM system since it requires a lower DC-bias value and the clipping distortion is only due to clipping of the upper peaks. a number of subcarriers of N = 1024. • The optimum value for the ACO-OFDM system appears at larger signal power as compared to the DCO-OFDM system for the same modulation order. . For Monte Carlo simulation results. the average BERs for at least 106 transmitted bits are calculated for each SNR value. (17) 10 10 Bit error ratio –1 –2 For the DCO-OFDM system. The bit error ratio (BER) for any M-QAM constellation can be calculated as given in Eq. whereas solid lines represent Monte Carlo simulation results. producing an overall clipping noise power given by σ2 = clip ∞ 10 10 10 10 –3 –4 cu (z − c u )2 p (z) dz+ cl −∞ z − cl 2 p (z) dz. the simulated electrical SNR range is from 10 dB to 40 dB. where U is the number of constellation points on the x-axis and J is the number of constellation points on the y-axis for any rectangular QAM.8]. Both results demonstrate a very close match for a wide range of SNRs and different modulation orders. • Similarly. In all analyses. OPT. 3. S IMULATION R ESULTS The performances of ACO-OFDM and DCO-OFDM systems in the presence of LED clipping distortion are analyzed analytically and through Monte Carlo simulations.

In addition. the considered DC-bias system above uses a bias point b = 0.626 J. 10. which corresponds to the recommended bias point for this LED from the manufacturer.32 DCO-AC output optical power–ideal DCO-DC power consumption 0.1 0. however. If the ACO-OFDM system was designed to achieve the same data rate as the DCO-OFDM system.525 V bias point performs slightly better than the considered b = 0. Box 2. This is because the DCO-OFDM system is biased at the average LED value and. However. 8/AUGUST 2011 Mesleh et al. the transmitted average optical power of the DCO-OFDM system starts to decay with increase of the input OFDM signal power. higher for DCO-OFDM since the DC bias shifts the signal to higher values. a constellation order of M = 4069 would be needed. 0./VOL. OPT. BER =            M −1 M log2 M erfc 3ρ .34 0. thus reducing DC power consumption by avoiding the need for a DC bias.06 0 5 10 15 20 Electrical OFDM signal power (dBm) 25 30 ACO-DC power consumption ACO clipping at saturation voltage Fig. NO. are very sensitive to clipping distortions and increasing signal powers beyond the optimum value becomes futile.625 V bias point and with reduced DC power consumption. 8). for visible light communication (VLC). 3. It is shown that the b = 0. due to clipping distortions. 8.35 Average transmitted optical power (W) 0. ACO-OFDM with 64-QAM modulation order achieves a data rate of about 29. • High modulation orders. COMMUN. 8 for ACO-OFDM and in Fig. 9. (Color online) ACO-OFDM DC power consumption and the average output optical power for different input OFDM electrical signal powers. larger DC-bias values are needed to fulfill the illumination requirements and the DC power consumption will be comparable for ACO-OFDM and DCO-OFDM. this might not be the optimum bias point for communication. However.16 0. In contrast.65 Mbps for the same modulation order. M = U × J rectangular QAM. The performance of these systems in the presence of LED clipping clearly highlights the significant dependence on the considered .22 Average transmitted optical power (W) 0. which is impractical. DC power consumption and the average output optical power are not the same either. 9 for the DCO-OFDM system.625 V. The transmitted optical power is.08 0. Existing techniques are categorized based on the way data is assigned to subcarriers and the bipolar–unipolar conversion process. 2 (M − 1) 3ρ square M − QAM + U −1 1 erfc log2 (U × J) U U 2 + J2 − 2 J −1 erfc J 3ρ U 2 + J2 − 2 (19) .36 0.18 0. as shown in Fig. For example. 0. In Fig.12 0. These advantages are achieved at the expense of a major reduction in data rate as compared to DCO-OFDM systems. C ONCLUSIONS This paper has analyzed the performance of existing indoor OWC OFDM transmission techniques. larger input signal leads to larger clipping distortion and more loss in transmitted optical power.24 0. such as 256-QAM and higher. 0.3 0 5 10 15 20 Electrical OFDM average signal power (dBm) 25 30 Fig. The ACO-OFDM OWC technique produces a half-wave symmetry time signal and allows clipping at the zero level. • ACO-OFDM transmitted optical power increases with increase of the input electrical power and starts to flatten at larger input power values (> 24 dBm in Fig.2 0. the BERs for the 16-QAM DC-bias system with different bias points are depicted. NETW.14 0.33 The analysis in this paper allows a simple method to optimize system parameters for OFDM optical wireless transmission techniques.31 0. (Color online) DCO-OFDM DC power consumption and the average output optical power for different input OFDM electrical signal powers. ACO-OFDM requires less DC power since it needs biasing at the TOV only. V.77 Mbps while the DCO-OFDM system achieves a data rate of 59.

Datasheet: ZW W5SG Golden DRAGON White LED. 29. vol. R.osram. and D. Haruyama. M. no. 2. 1. vol. no.physorg. 1997. Available: http://www. OPT. W. F. 27. on Hybrid Intelligent Systems (HIS). [12] S. no.525 V b = 0.. [11] J. 6th IEEE Int. and T. 10 Bit error ratio 5 10 –10 b = 0 . pp. 349–378. no. Walewski. 3. 8. Haas. C. E. W. Langer. Apr. 3. Carruthers. Aug. 2010 [Online].” in 7th Int. Light-Emitting Diodes. Lett. 2009. 2009. pp. vol.” in Proc. pp. 1280–1289. G. Melloch.” IEEE Commun. 2009. Lee. [13] J. Jan. Tanaka.com. J. Elgala. pp. 42. pp. vol. 2009. H. “Wireless infrared communications. Haas. pp. Elgala. Elgala. 4970–4978. M. Mesleh. Available: http://www. 2003. system design should consider LED clipping effects and should optimize the OFDM signal power and the considered modulation order. [16] H. Syst. and H. 6. H. vol. pp. “Advanced modulation schemes for short-range optical communications [invited paper].–Mar.” Proc. Raed Mesleh gratefully acknowledges the support for this work from the Deanship of Scientific Research at the University of Tabuk. B. 2010. . C. Nerreter..” Int... 11. Z. Walewski. Kamalakis. and minimum temperature. July 21–23. [3] J. 1993. pp. 85–87. Akbulut. 265–298. vol. Ireland. The analysis in this paper allows a simple method to optimize system parameters. O. “Impact of nonlinear LED transfer function on discrete multitone modulation: Analytical approach. [9] J. 2010 [Online]. Breyer. Lightwave Technol. Conf. The authors acknowledge the support for this work from the German Federal Ministry of Economics and Technology (BMWi) as part of the Lufo 2nd Call project SINTEG. of the 2nd Int. 28–30. which mainly considers optimum brightness. Lett. no. W. Messerschmitt. Neokosmidis. Edwards.” in Proc. “Visible light communication using OFDM. Haas. and H. Breyer. 5. pp. no. no. 1997. pp. Press. A CKNOWLEDGMENT Dr. 2440–2454. no. Quantum Electron. J. R. Ultra Wideband Commun. [20] L. J. Dublin. M. “Indoor optical wireless systems—a review.” J. Randel. Grubor. Aug. M. Conf. 5. Walewski. Symp. J. Armstrong. [15] H. Mesleh.” in Proc. Schubert. “Comparison of asymmetrically clipped optical OFDM and DC-biased optical OFDM in AWGN.725 V b = 0.” IEEE Photon. Knipp. [21] I. Therefore. J. Kahn. NETW. S. A. Armstrong and B.de. “An overview of indoor OFDM/DMT optical wireless communication systems.” IEEE Trans. and H. O’Brien. 22. J. Lett. pp. on Wireless and Optical Communications Networks (WOCN). Komine. S.625 V b = 0 . no.. E86-B. 3. El Tabach. S. R. Street. Commun. Mesleh. Stavrinou. N. Koonen. Haas. 1. 367–379. on Communication Systems Networks and Digital Signal Processing (CSNDSP). [14] R. J. Dec. Lee. 566–570. “Non-linearity effects and predistortion in optical OFDM wireless transmission using LEDs. NO. vol. “Experimental characterization of non-directed indoor infrared channels. 23. “Power efficient optical OFDM. R EFERENCES [1] S. 27. “Modeling for predistortion of LEDs in optical wireless transmission using OFDM. IEEE. Aug. no. and D. Woodall. 2009. [18] E. 1–3. 143–150. J. Zyga. T. Armstrong and A.. Weiner. 1613–1623. Afgani. Lowery. 2010. “White LEDs with super-high luminous efficacy could satisfy all general lighting needs. Barry. on Testbeds and Research Infrastructures for the Development of Networks and Communities (TRIDENTCOM). “Digital communications above 1 Gb/s using 890-nm [19] H. Sel. “A study of LED nonlinearity effects on optical wireless transmission using OFDM. [2] A.” PhysOrg. Aug. Apr. Lightwave Technol. Sel. Technol. Krause. C. Hargis. [4] M. H.. Areas Commun.. An example was shown where an optimum DC-bias point is found for the considered LED in this paper. [22] K. [5] J. 13.. and A. and M. 12–14. (Color online) DCO-OFDM optimum bias point analysis. no. long lifetime. J. D. 21. Randel. M. Bouchet. 12. Mar. 16. VOL. China.html. Top. 8/AUGUST 2011/J. Barcelona. M. 234. W.” IEEE J. [6] Y. 43.–Apr. B. such as bias point value. Inan. COMMUN. pp. 30 10 –15 0 5 10 15 SNR (dB) 20 25 [8] H.425 V [7] M. F. 1995. vol. 2001.Mesleh et al. 2. A 16-QAM constellation is considered and the BER versus electrical OFDM signal power is calculated for different bias points. vol.-D. Barry.com/news202453100. M. 2009. Cambridge Univ. modulation order and other system parameters.” IEICE Trans. 3. Conf. “OFDM visible light wireless communication based on white LEDs. Kahn and J. Commun. vol. vol. Egypt. and J.” IEEE Photon.” in Proc. pp. J. Mar. and D. and H. 189–204. of the 64th IEEE Vehicular Technology Conference (VTC).. Mesleh. 2003. 343–345. T. 129–134. Pricope. pp. 370–372. 2006. Kahn. Fig. R. Chen. [17] OSRAM GmbH. 10. “OFDM for optical communications. J. Elgala. 2006. 85. May 2008. 1st ed.. Spain. Quantum Electron. Haas. Feb. Faulkner. of the IEEE 10th Int.” J. Schmidt. Lett. D. 627 10 0 surface-emitting light-emitting diodes. “PAM-DMT for intensity-modulated and direct-detection optical communication systems. vol. “Simulation of multipath impulse response for indoor wireless optical channels. R.” Opt.” IEEE J. Haas. C.” Electron. Lee. Apr. O’brien. no. Sphicopoulos. Technol. S. Cairo. Shenyang Liaoning. Feb. “Indoor visible light data transmission system utilizing white LED lights. IEEE. H. C. Mesleh. Elgala. S. 22–25. vol. It was shown that AWGN noise dominates at low SNR values and clipping distortion dominates at large SNR values. [10] J.. M. and J. It is shown that the optimum DC-bias point for communication is different from the optimum DC-bias point suggested by the manufacturer. and B. 2007. Randel. Elgala. Nakagawa. no. 1749–1751. and J. pp. P. Mar. Franke. Krause.

Germany. Jordan. Hany Elgala (S’02–M’10) received his Ph. He holds more than 15 patents in the area of wireless communications. vol. Harald Haas (SM’98–AM’00–M’03) received his Ph. pp. three of which are already granted. Germany. Cho and D. on Communication Systems. Grubor. Haas joined Jacobs University Bremen in 2002 as an Associate Professor. digital signal processing. Bremen. Prof. “Optical hotspots speed up wireless communication. G. He received his B. Dr. Mar. [26] J. in the field of RF and optical wireless communication systems. and J. pp. 50. G. July 2002. Stavrinou. Ulm. 4.628 J. Raed Mesleh (S’00–M’08) received his B. course at Ulm University (LEG award). in 2004. of the IEEE 10th Int. vol. pp. 3. Mesleh serves as a technical reviewer for several IEEE transaction journals and international conferences and he is a TPC member for several international conferences in wireless communications. [27] D. He has co-authored a book entitled Next Generation Mobile Access Technologies: Implementing TDD with Cambridge University Press. His main research interests are in wireless communication and optical wireless communication with particular focus on MIMO techniques.” Nat. Rep.D. His main research interests are in the areas of communication systems. Germany. [23] H. Photonics. Mesleh joined Orange in Amman. Bremen. Austria. where he worked as a Postdoctoral Fellow. vol. Professor Haas was an invited speaker at the TEDGlobal Conference in 2011. Massachusetts Institute of Technology. Conf. Networks and Digital Signal Processing. R. Transparent Optical Networks (ICTON 08). Dr. “Practical considerations for indoor wireless optical system implementation using OFDM. Elgala. . degree from Jacobs University Bremen in 2010. Cambridge. NO. 1952. Conf. Mesleh. Irbid. degree in communication technology from Ulm University. “On the general BER expression of oneand two-dimensional amplitude modulations. Commun. Mesleh et al. Zagreb. 8/AUGUST 2011 I. on Telecommunications (ConTel). “Bandwidth efficient indoor optical wireless communications with white light emitting diodes. he joined the University of Tabuk in Saudi Arabia where he is now an Assistant Professor in the Electrical Engineering Department and a researcher at the Sensor Networks and Cellular System (SNCS) research center. no. Athens. and has published more than 170 conference and journal papers. 2009. 2008. degree from the University of Edinburgh in 2001. MA. K. [25] K. He is a co-inventor of seven patents. Walewski. Parry. In 2003. degree in microsystems engineering from Furtwangen University in 2003. “Cross correlation function of amplitude-distorted Gaussian signals. Mesleh received a tuition waiver for his Ph. 2007. in 2000. Yoon. Langer. Dr. 149–154. degree in electronics and communications from Ain-shams University in 2000 and completed his M. Croatia. and P. In October 2010.D. and./VOL. From September 2007 until September 2010 he was with the School of Electrical Engineering and Computer Science.Sc. 1. he received the award for the best student performance in the M. Haas. in 2007. He is currently a Postdoctoral Fellow at the Department of Electrical Engineering and Computer Science at Jacobs University. Wolf. June 22–26.D. and M. Neokosmidis. the spatial modulation multiple antenna concept. 216. in 2000. and his Ph. he received the award for the best graduate communication engineer in year 2000 from Yarmouk University. Jordan (2000–2001). June 23–25. [24] J. June 8–10.Sc. 1. 7.Sc.Sc.D. degree in electrical engineering from Jacobs University. COMMUN.” Tech. S.. 245–247. his M. From 2001 to 2002 he was a Research Project Manager at Siemens in Munich. Graz. Since 2005 he has been conducting research on indoor optical wireless communication. NETW. of the 10th Anniversary Int. 1074–1080.” in Proc. as a System Engineer and worked as a Consultant for Nokia in Saudi Arabia from 2001 to 2002. Symp.Sc. vol. 165–169. 2008. OPT. before returning in 2007 to The University of Edinburgh where he now holds a personal Chair of Mobile Communications. Research Laboratory for Electronics. degree in communication engineering from Yarmouk University. Randel. of the 6th Int. and optical wireless communication.” IEEE Trans. His research interests are in interference coordination and mitigation in wireless networks. study at Jacobs University along with a three-year scholarship. Jacobs University.” in Proc. Bussgang. pp. and H.” in Proc. Greece. “Optical wireless communications for broadband access in home area networks. and circuit design. Ntogari. O’Brien.

You're Reading a Free Preview

Download
scribd
/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->