OPTIMIZATION OF OPTICAL SSB-OFDM SYSTEM WITH
DIRECT DETECTION FOR APPLICATION IN
METROPOLITAN/REGIONAL NETWORKS
Bruno M. M. Meireles and João L. Rebola
Department of Information Science and Technology, Instituto Universitário de Lisboa (ISCTE-IUL), Lisboa, Portugal
Group of Research on Optical Fiber Telecommunication Systems, Instituto de Telecomunicações, Lisboa, Portugal
Keywords: Direct-detection, Noise accumulation, Optical add-drop multiplexers, OPTICAL communications, optical
fiber, Orthogonal frequency division multiplexing (OFDM).
Abstract: Due to its spectral efficiency, dispersion robustness, simplicity and transparency to electrical modulation
format, orthogonal frequency division multiplexing (OFDM) is a promising technique that allows to
increase the capacity and to make the upgrade of optical telecommunications systems already installed. In
this work, the study and optimization of OFDM transmission in an optical communication system using
direct detection with a short and medium range (metropolitan and regional networks) for a target bit rate of
10 Gbps will be performed exhaustively. The optimization is carried out taking into account the impact of
noise accumulation and of signal distortion resulting from bandwidth narrowing introduced by the cascade
of add-drop multiplexing nodes on the network performance. Optical transmission of single sideband
OFDM signals is considered and the carrier-to-signal power ratio of the OFDM signal is optimized in order
to achieve the best network performance.
1 INTRODUCTION
Optical frequency division multiplexing (OFDM) is
a digital modulation technique that allows
transmitting multiple signals simultaneously in a
high-speed data channel. The data are separated into
narrow parallel subcarriers, where frequency
overlapping is possible without intercarrier
interference since all subcarriers are orthogonal
(Shieh, 2010). Hence, OFDM makes a more
efficient use of the spectrum than conventional
frequency division multiplexing. Different
modulation schemes, such as Quadrature Amplitude
Modulation (QAM) or phase shift-keying, can be
chosen for the different subcarriers independently
(Shieh, 2010). Moreover, since OFDM has a long
symbol period allows practically eliminating the
intersymbol interference due to fiber channel
dispersion (Shieh, 2010).
Another advantage of using OFDM is that
electrical channel equalization is done by a single
tap equalizer, while single carrier systems require
more complex electrical equalization techniques
such as adaptive equalization (Schmidt, 2008),
(Lowery, 2007).
An important drawback of OFDM is the signal
large peak-to-average power ratio (PAPR). OFDM
being a superposition of a higher number of
modulated subchannels signals may exhibit a high
instantaneous signal peak power (Hanzo, 2004). In
optical communications, high PAPR can impose
higher signal distortion in the presence of fiber
nonlinear effects (Shieh, 2008).
Currently there are two types of optical detection
used in OFDM transmission: coherent optical
OFDM (CO-OFDM) detection and direct-detection
OFDM (DD-OFDM). The use of direct detection is
widespread in existing optical communication
systems. Although, it has lower sensitivity than CO-
OFDM systems (Shieh, 2008, 2010), its simplicity
and lower cost, makes direct detection suitable for
metropolitan/regional networks, which do not
require state-of-the-art components as long haul
networks do.
Furthermore, the upgrade of installed links to
transmit OFDM signals is carried out mainly on the
electronic part of the communication link (through
133
M. M. Meireles B. and L. Rebola J..
OPTIMIZATION OF OPTICAL SSB-OFDM SYSTEM WITH DIRECT DETECTION FOR APPLICATION IN METROPOLITAN/REGIONAL NETWORKS.
DOI: 10.5220/0003488701330138
In Proceedings of the International Conference on Data Communication Networking and Optical Communication System (OPTICS-2011), pages
133-138
ISBN: 978-989-8425-69-0
Copyright
c
2011 SCITEPRESS (Science and Technology Publications, Lda.)
the use of digital signal processing to perform the
Fast Fourier Transform (FFT) and its inverse) and
not on the existing optical components (Shieh,
2010).
In this work, the performance of an optical DD-
OFDM system is studied and optimized, with the
objective of using this signalling scheme in
metropolitan/regional networks (with about 600 km
of total length), where the accumulation of noise and
bandwidth narrowing introduced by a cascade of
optical nodes can have a significant impact on the
network performance.
2 SYSTEM DESCRIPTION
2.1 OFDM Transmitter
Figure 1: OFDM transmitter architecture.
Figure 1 shows the OFDM transmitter architecture
scheme used in this work. The input serial data is
first mapped to symbols using QAM. Then, the data
symbols are converted into N parallel blocks of data
(N subcarriers). N blocks of zeros are inserted in the
middle of the data subcarriers, making a total of 2N
parallel subcarriers within one OFDM symbol. The
introduction of these zeros (oversampling), as well
as the low pass filter (LPF) before the IQ modulator,
is used to reduce the aliasing (Alves, 2009). Then an
inverse-FFT is applied to the subcarriers resulting in
a time domain waveform that contains a
superposition of all 2N subcarriers. After, cyclic
prefix and guard interval are inserted. The resulting
waveform is then modulated by an IQ modulator to a
radio frequency (RF) carrier with frequency f
RF
(Lowery, 2006). Then, the two signal I and Q
components are added and the OFDM electrical
signal is obtained.
Before optical transmission, the OFDM electrical
signal needs to be modulated onto the optical
domain using an external modulator (Shieh, 2010).
In this work, a Mach-Zehnder modulator (MZM)
operating in linear and nonlinear regime will be
considered. In the linear regime, no signal distortion
is introduced by the MZM. In the nonlinear regime,
the bias voltage of the MZM should be chosen to
improve the MZM linearity and minimize signal
distortion. Hence, the MZM is biased at the
quadrature point (Leibrich, 2009). Then, the optical
carrier power (in relation to the OFDM signal
power) is controlled by the single-side band (SSB)
filter, in order to improve the sensitivity of the
OFDM reception (Lowery, 2007). After SSB
filtering, the OFDM signal is transmitted through a
singlemode fiber (SMF).
2.2 OFDM Receiver
Figure 2: OFDM receiver architecture.
Figure 2 shows the DD-OFDM receiver architecture.
The optical receiver includes an optical amplifier, an
optical filter followed by a PIN photodetector and by
an OFDM electrical receiver. At the OFDM
electrical receiver, the baseband signal is recovered
using an IQ demodulator with a LPF in each
demodulation arm. The resulting baseband signal is
sampled using an A/D converter, then cyclic prefix
and guard interval are removed and a FFT is applied
to the resulting signal.
After FFT, the zero subcarriers are removed and
each subcarrier is equalized in order to compensate
for phase and amplitude distortion due to
transmission (Agrawal, 2004). Equalization is
performed by applying the inverse of the estimated
channel response using training sequences (Lowery,
2006). After equalization, each sub-channel is
demapped and the original bits are recovered. At this
point, the DD-OFDM system performance is
evaluated by estimating the bit error rate (BER).
2.3 System Performance Evaluation
In order to perform the system performance
evaluation, two distinct methods are used: BER
estimated using Monte Carlo simulation and direct
error count (DEC), which is named BER
DEC
and is
defined per subcarrier k by (Alves, 2010).
cos(2f
RF
t)
sin(2f
RF
t)
sin(2f
RF
t)
cos(2f
RF
t)
3
N
/
2+1
OPTICS 2011 - International Conference on Optical Communication Systems
134

number of bit errors
total number of transmitted bits
DEC
BER k
(1)
and the BER obtained from the error vector
magnitude (EVM), named BER
EVM
and given by


2
2
11 3 1
4
log 1
EVM
RMS
M
BER k Q
MMEVMk





(2)
where M-QAM mapping per subcarrier is assumed
and EVM
RMS
is the root mean square of the EVM
defined by

 

2
() ()
1
()
1
s
s
N
ll
oi
l
RMS
N
l
i
l
sksk
EVM k
sk
(3)
In equation (3), N
s
is the number of OFDM symbols
transmitted per OFDM frame and

()l
i
sk
and

()l
o
sk
are the M-QAM symbols corresponding to
the k
th
subcarrier of the l
th
OFDM symbol of the ideal
constellation and the constellation obtained at the
equalizer output, respectively (Alves, 2010).
The BER of the OFDM symbol is obtained by
averaging the BERs obtained for all subcarriers.
Although the effect of signal distortion on the
system performance is not accurately taken into
account using the BER
EVM
method, when noise
accumulation is dominant over signal distortion, the
BER
EVM
provides very precise estimates of the
system performance (Alves, 2010).
3 RESULTS
All results have been obtained considering an
OFDM signal with the parameters shown in Table 1
for a target bit rate of 10 Gbit/s and are based on the
work done by Lowery (2007).
Table 1: Simulation parameters.
Modulation
Nº of
transmitted bits
per OFDM
symbol
Nº of subcarriers
Bit rate per
OFDM stream
D
b
4 QAM 1024 512 10 Gbps
OFDM symbol
duration - T
S
Guard interval
G
i
Cyclic prefix
OFDM signal
bandwidth
102.4 ns 7.4206 ns 6.4 ns
5 GHz
Figure 3 a) shows the power spectral density
(lowpass equivalent representation) of the double-
sideband (DSB) OFDM signal at the MZM output
centered at the RF carrier of 7.5 GHz, with a
bandwidth of 5 GHz.
In a preliminary study, the bandwidths of the
electrical filters of the OFDM transmitter and
receiver have been optimized and 6
th
order Bessel
filters with a bandwidth of 4 GHz and 2.9 GHz,
respectively, have been found as optimum and used
throughout this work.
3.1 Optical SSB Filter Optimization
In this section, the bandwidth of the optical SSB
(OSSB) filter is optimized in order to suppress the
optical carrier power and improve the DD-OFDM
system performance (Lowery, 2007). Furthermore,
the use of SSB signalling (in comparison with DSB),
allows to decrease the distortion introduced by the
fiber chromatic dispersion during optical
transmission. The carrier-to-signal power ratio
(CSPR) is defined as CSPR = P
c
/P
OFDM
(Alves,
2010), where P
c
is optical carrier power and P
OFDM
is the SSB OFDM signal power. For illustrative
purposes, Figure 3 b) shows the OFDM signal
spectrum after the OSSB filter for a CSPR of 0 dB.
a) b)
Figure 3: Power spectral density of the OFDM signal. a)
DSB signal at the modulator output. b) SSB signal after
filtering for a CSPR of 0 dB.
The optimization is performed by estimating the
performance of the DD-OFDM system (using the
BER
DEC
and BER
EVM
) as a function of the CSPR (by
varying the OSSB filter bandwidth) for the optical
signal-to-noise ratios (OSNRs) of 11 dB, 15 dB and
20 dB. A 2
nd
order supergaussian optical filter is
considered at the optical receiver. The optimization
is performed for an external modulator working in
the linear (Figure 4) and nonlinear (Figure 5)
regimes.
Figure 4 shows that for an OSNR of 11 and 15
dB, the optimum CSPR is 0 dB as predicted in the
works by Lowery (2006, 2007) and Jansen (2007).
This corresponds to an OSSB filter bandwidth of 5
GHz. For an OSNR of 20 dB, it is possible to
observe that the optimum value is shifted to higher
CSPR values and therefore higher OSSB filter
bandwidths. This shift of the CSPR for higher
OSNRs has been also observed in the work by
OPTIMIZATION OF OPTICAL SSB-OFDM SYSTEM WITH DIRECT DETECTION FOR APPLICATION IN
METROPOLITAN/REGIONAL NETWORKS
135
Jansen (2007) and is attributed to the higher
influence of the intermixing of OFDM subcarriers
after the photodetection [the photodetection mixing
products are described in detail by Lowery (2008)]
on the performance, when the amplified spontaneous
emission (ASE) noise power is lower.
Figure 4: BER
EVM
and BER
DEC
(diamonds) as a function of
the CSPR for the OSNRs of 11 dB (blue), 15 dB (red) and
20 dB (green), for a linear external modulator and for the
receiver optical filter bandwidths of 100 GHz (dashed
lines) and 30 GHz (continuous lines).
Figure 4 shows also a good agreement between
the error probabilities estimated using the BER
DEC
and the BER
EVM
for lower OSNRs, in accordance
with the results presented in the work by Alves
(2010). Furthermore, it shows that the performance
of the DD-OFDM system is improved by using the
receiver optical filter with a bandwidth of 30 GHz,
due to higher ASE noise power reduction.
Figure 5 shows the BER as a function of the
CSPR, for the nonlinear external modulator, for
different modulation indexes m. It is possible to
observe that the optimum CSPR value is always near
0 dB, although, for higher modulation indexes, a
slight shift of the optimum CSPR is observable. It is
also possible to observe that, as the bandwidth of the
SSB filter increases (higher CSPR) and for higher
modulation indexes, the distortion due to the
external modulation non-linear regime starts to
impose the system performance. For lower
modulation indexes, the external modulator is
operating near the linear regime and the noise is the
main performance impairment, once the distortion
induced by the external modulator on the OFDM
signal is negligible.
Moreover, it is possible to infer that for a DSB-
OFDM signal (which corresponds to the rightmost
CSPR depicted for each modulation index), there
exists an optimum value for the modulation index,
which leads to the best system performance. This
conclusion is in agreement with the results presented
in the work by Alves (2010).
Figure 5 shows again a good agreement between
the estimates of the BER obtained using the BER
DEC
and the BER
EVM
.
Figure 5: BER
EVM
(dashed lines) and BER
DEC
(diamonds)
as a function of the CSPR for the model of the nonlinear
external modulator for different modulation indexes (m)
and an OSNR of 11 dB.
3.2 OFDM Signal Transmission along a
Cascade of Optical Nodes
In this section, the transmission of OFDM signals
along an optical communication system with several
fiber spans is studied. The impact of noise
accumulation and of bandwidth narrowing along the
several spans on the performance of the DD-OFDM
system will be investigated. Figure 6 shows the
scheme considered for the OFDM transmission
system, with a special emphasis on the optical node
configuration. The optical fiber considered in this
investigation is a standard singlemode fiber with a
dispersion parameter D
λ
= 17 ps/(nm.km) and an
attenuation of α =0.2 dB/km. Linear transmission
along the fiber is considered.
The optical node is simply modelled by an ideal
inline optical amplifier with gain G, which adds
ASE noise to the signal (operation performed in the
simulation by a noise generator) and an optical filter
to reduce the noise power. In this scheme, the noise
added by each erbium-doped fiber amplifier (EDFA)
accumulates along the link. Notice also that the
signal distortion can be enhanced at the
photodetector’s input due to the cascade of optical
filters. It is also assumed that the optical amplifier
exactly compensates the fiber attenuation. In a first
approach, we also considered that fiber dispersion is
completely compensated by the electrical equalizer.
OPTICS 2011 - International Conference on Optical Communication Systems
136
Figure 6: OFDM transmission system with several spans.
Each span is composed by a fiber followed by the optical
node constitute by an EDFA and an optical filter.
It is worth recalling that one of the main goals of
this work is to study OFDM transmission in a
metropolitan/regional optical network, which has
about 600 km of distance coverage. So, it is
important to estimate the DD-OFDM system
performance for the optical link depicted in Figure 6,
and to determine if the commitment for the network
coverage length is accomplished.
Figure 7: BER
EVM
as a function of the number of sections
used in the optical link obtained for an OSNR of 25 dB
(diamonds) and 20 dB (continuous lines) for different
receiver optical filter bandwidths. The error probability
corresponding to the FEC limit is also depicted.
Figure 7 shows the error probability estimated
using the BER
EVM
as a function of the number of
sections for different receiver optical filters. The
MZM is assumed to be working in the linear regime
at the optimum CSPR. Figure 7 shows that the
probability of error increases as the number of
optical sections increases, since the OSNR is
degrading with the number of sections. For the
optical filters with a bandwidth of 10 GHz (blue)
and 20 GHz (red), the distortion suffered by the
OFDM signal (due the smaller optical filter
bandwidth and severe bandwidth narrowing due to
the cascade of optical nodes) is dominant over the
noise and no distinction is found between the error
probabilities estimated for both OSNRs of 20 dB
and 25 dB (measured at the output of the first optical
node). For higher optical bandwidths (larger than
30 GHz), the signal distortion introduced by the
optical filtering is not so significant and noise
accumulation along the optical link plays an
increasing role on degrading the network
performance. Similar conclusions were found for the
nonlinear MZM model, as long as it is working in
lower modulation indexes (near the linear regime).
Only BER
EVM
results are presented, since when
noise is the dominant factor that impairs the system
performance (over signal distortion), BER
EVM
provides very accurate BER results (Alves, 2010).
To achieve the objective of covering a
metropolitan/regional network with a length of about
600 km and by assuming that each fiber section is
about 80 km long, approximately 8 sections are
needed to cover that distance [by considering that
the error probability is bounded to the value of 10
3
,
which corresponds to the limit for forward error
correction (FEC) (Shieh, 2010)]. From Figure 7, one
can conclude that, for an OSNR of 25 dB, the goal
of covering the typical maximum distance of
metropolitan/regional network is accomplished.
Then, the impact of optical fiber transmission on
the system performance is evaluated to confirm if
the distortion imposed by the optical fiber dispersion
is indeed compensated by the equalizer and can be
considered negligible or not. The results obtained
are shown in Figure 8.
Figure 8: BER
EVM
as a function of the optical fiber length
in the optical link. The error probability corresponding to
the FEC limit is also depicted.
Figure 8 shows that for an OSNR of 20 dB and
for a fiber length up to 600 km, the OFDM system
error probability is always below 10
10
. Therefore, it
is possible to state that the distortion imposed by the
SMF is sufficiently compensated by the equalizer.
The small oscillations observed in the BER are
10
15
OPTIMIZATION OF OPTICAL SSB-OFDM SYSTEM WITH DIRECT DETECTION FOR APPLICATION IN
METROPOLITAN/REGIONAL NETWORKS
137
attributed to the imperfect estimation of the
equalizer coefficients for each fiber length. A more
accurate coefficients estimation would lead to a less
oscillatory BER variation with the fiber length. For
an OSNR of 15 dB, as the noise power is higher, the
influence of the imperfect estimation of the equalizer
coefficients is not visible, and the BER variation
with the fiber length shows practically no
oscillations for fiber lengths below 1400 km. After
1400 km, the effect of fiber dispersion on the signal
can no longer be compensated by the equalizer, due
to the insufficient guard interval duration of the
OFDM signal (Shieh, 2010) and a peak of the error
probability occurs [as shown in Figure 8]. This
means that, it is possible to reach a fiber length of
approximately 1400 km without significant
distortion added by the SMF on the OFDM
transmission system. In conjunction with the results
presented in Figure 7, it can be stated that the
distortion introduced by the optical fiber along the
target distance of about 600 km is negligible, and
that noise accumulation and signal distortion due to
bandwidth narrowing along a chain of optical
multiplexing nodes are the dominant factors causing
the performance degradation for typical distances of
metropolitan/regional networks.
4 CONCLUSIONS
In this paper, it has been shown that it is possible to
cover a metropolitan/regional optical network using
an optical OFDM system with direct detection at the
bit rate of 10 Gbit/s. The optical OFDM system has
been optimized in order to attain the best network
performance and it has been shown that the error
probability is still below the FEC limit after 8
sections of optical nodes (for a typical distance of
about 600 km). This conclusion was obtained for an
OSNR of 25 dB and for optical filters (inside the
optical node) with a bandwidth above 30 GHz. ASE
noise accumulation and signal distortion resulting
from bandwidth narrowing play a significant role on
achieving this limit. We have also found that the
distortion introduced by the optical fiber along the
network link length is negligible.
The consideration of a more realistic model for
the optical multiplexing node based on, for example,
reconfigurable optical add-drop multiplexers and the
study of the impact of the detuning of the optical
filters (inside the optical nodes) on the network
performance are left for future work.
REFERENCES
Agrawal G., 2004, Lightwave Technology – Components
and Devices, John Wiley & Sons, New Jersey, 1
st
edition.
Alves, T., Cartaxo, A., 2009, “Performance degradation
due to OFDM-UWB radio signal transmission along
dispersive single-mode fiber”, IEEE Photonics
Technology Letters, vol. 21, no. 3, pp. 158-160, Feb.
Alves, T., Cartaxo, A., 2010, “Analysis of methods of
performance evaluation of direct-detection orthogonal
frequency division multiplexing communication
systems”, Fiber and Integrated Optics, vol. 29, no. 3,
pp. 170-186, May.
Hanzo, L., Münster, M., Choi, B., Keller, T., 2004, OFDM
and MC-CDMA for broadband multi-user
communications, WLANs and broadcasting, IEEE
Press, Chichester, 1
st
edition.
Jansen, S., Morita, I., Tanaka, H., 2007, Carrier-to-signal
power ratio in fiber-optic SSB-OFDM transmission
systems, International Conference on Electrical and
Computer Engineering IEICE, Mongaguá, Santos,
Brazil.
Leibrich, J., Ali, A., Paul, H., Rosenkranz, W.,
Kammeyer, K., 2009, “Impact of modulator bias on
the OSNR requirement of direct-detection optical
OFDM”, IEEE Photonics Technology Letters, vol. 21,
no. 15, pp. 1033-1035, Aug.
Lowery, A., Armstrong, J., 2006, “Orthogonal-frequency-
division multiplexing for dispersion compensation of
long-haul optical systems”, Optics Express, vol. 14,
no. 6, pp. 2079-2084, Mar.
Lowery, A., Du, L., Armstrong, J., 2007, “Performance of
optical OFDM in ultralong-haul WDM lightwave
systems”, Journal of Lightwave Technology, vol. 25,
no. 1, pp. 131-138, Jan.
Lowery, A., 2008, “Amplified-spontaneous noise limit of
optical OFDM lightwave systems”, Optics Express,
vol. 16, no. 2, pp. 860-865, Jan.
Schmidt, B., Lowery, A., Armstrong, J., 2008,
“Experimental demonstrations of electronic dispersion
compensation for long-haul transmission using direct-
detection optical OFDM”, Journal of Lightwave
Technology, vol. 26, no. 1, pp. 196-203, Jan.
Shieh, W., Bao, H., Tang, Y., 2008, “Coherent optical
OFDM: theory and design”, Optics Express, vol. 16,
no. 2, pp. 841-859, Jan.
Shieh, W., Djordjevic, I., 2010, Orthogonal Frequency
Division Multiplexing for Optical Communications,
Academic Press, San Diego, 1
st
edition.
OPTICS 2011 - International Conference on Optical Communication Systems
138