Improving Nonlinear Precompensation in Direct-Detection Optical OFDM
Communications Systems
Liang Du, Arthur Lowery
Monash University, Clayton, Victoria 3800, Australia; arthur.lowery@eng.monash.edu.au
Abstract
Carrier boosting at the receiver enables direct detection optical OFDM (DDO-OFDM) to outperform coherent
OOFDM in the nonlinear limit. Boosting also improves the effectiveness of nonlinearity precompensation
substantially.
Introduction
Optical OFDM uses multiple subcarriers and
electronic equalisation to compensate for almost
unlimited dispersion in fibre links: transmission rates
of 121Gbps over 1009 km have been achieved per
WDM channel [1]. OFDM is resilient to both noise
and dispersion but the close packing of the
subcarriers causes strong nonlinear mixing between
them. This means that Optical OFDM operates
optimally at channel powers less than -5 dBm [2] at
10 Gbit/s.
Nonlinearity precompensation can mitigate
nonlinear effects in coherent optical OFDM (CO-
OFDM) systems [3],[4]; an improvement in received
electrical signal quality of over 10 dB is possible for
low dispersion fibres [4]. Postcompensation has
also been demonstrated to mitigate for fibre
nonlinearity for CO-OFDM systems [5]. CO-OFDM
requires a sophisticated optical receiver with a
narrow-linewidth local oscillator: Direct-Detection
OOFDM (DDO-OFDM) transmits a carrier along
with the subcarriers and so only needs a photodiode
for reception. However, because the carrier
propagates along with the signal, there is far more
nonlinear mixing, so nonlinearity precompensation
is less effective [6]. Previously we showed that
boosting the strength of the carrier relative to the
sidebands just before the photodiode can improve
the noise performance of DDO-OFDM systems [7].
Boosting also allows the carrier to be transmitted at
a lower level, thus potentially reduces the strength
of nonlinear mixing in the fibre.
This paper shows using simulations that carrier
boosting can also greatly improve the nonlinear
performance of DDO-OFDM systems, to exceed
coherent systems. Additionally, nonlinearity
precompensation becomes much more effective
with carrier boosting.
Nonlinear compensation
The transmitter is very similar to [6], where the 5-
GHz upper sideband is positioned 5-GHz above the
optical carrier as shown in Fig. 1. 10 Gbit/s is
transmitted using 4-QAM and 512 subcarriers. For
precompensation, phase modulation is applied to
the subcarriers and carrier, in proportion to the
instantaneous optical power in the sideband [6].
Figure 1: Spectrum of DDO-OFDM after
precompensation phase modulation.
Phase modulation in the time domain can
completely compensate for the intermixing of the
OFDM subcarriers in dispersionless fibre; however,
in dispersive fibre the waveform evolves during
propagation, so that the precompensation is less
effective. The amount of precompensation is
generally reduced to find an optimum signal quality
at the receiver. We are interested in low-dispersion
fibres as these have a much poorer nonlinear
performance and so greatly benefit from nonlinearity
compensation [3]. In the following examples, a
4000-km system with of 2 ps/nm/km fibre was
simulated using VPItransmissionMaker. Amplifier
noise was turned off to isolate the nonlinear effects
as we are interested in the nonlinear limit. Span
lengths of 80-km were used with an attenuation of
0.2 dB/km. At the input of every span, the optical
power was re-amplified to -3 dBm.
The optical signal was also amplified just before the
photodiode receiver, and the carrier boosted with a
1-GHz band-pass optical filter with a variable stop-
band attenuation. The optimum transmitted carrier
level without boosting is when the carrier power
equals the sideband power: we also investigated
suppressing (cutting) the carrier by a further 5 dB
at the transmitter.
Results
Fig. 2 plots received signal quality, Q, averaged
over all subcarriers [2] versus the precompensation
P.4.08
ECOC 2008, 21-25 September 2008, Brussels, Belgium
Vol. 5 - 147
1 978-1-4244-2228-9/08/$25.00 (c) 2008 IEEE
effective length, TxLeff [4]. Without
precompensation, DDO-OFDM outperforms
coherent OOFDM by 2 dB using 10-dB boost (, y
axis). With precompensation all of the DDO-OFDM
systems are improved (e.g. 3.5 dB benefit without
boosting, ), but the best improvement is obtained if
carrier boosting is also used (7.5 dB benefit, ,
peak). If the carrier power is cut by 5-dB at the
transmitter and then boosted at the receiver, (,
peak) an additional 0.5-dB benefit is available.
Coherent systems (+) allow a greater degree of
precompensation, so offer the ultimate performance.
4
6
8
10
12
14
16
18
0 5 10 15 20
Effective Precompensation Length/Span, TxLeff [km/span]
E
l
e
c
t
r
i
c
a
l
S
i
g
n
a
l
Q
u
a
l
i
t
y
Q
[
d
B
]
Cut
0 dB 0 dB
5 dB 5 dB
Boost
0 dB 5 dB
5 dB 15 dB
0 dB 10 dB
Coherent
With no
precompensation,
only the net boost
= (boost - cut)
affects the signal
quality
DDO-OFDM
Figure 2: Effectiveness of precompensation for
boosted and unboosted systems.
Fig. 2 also shows that without precompensation, no
benefit is produced by cutting the transmitted carrier
power and boosting it by the same amount at the
receiver. This is surprising as reducing the carrier
power greatly reduces the power of nonlinear
intermodulation products involving the carrier. The
explanation is that the majority these products lie
out of the OFDM signal band [6]; so they will only
affect the electrical signal if they can mix back into
the electrical signal bandwidth upon detection, by
mixing with optical tones away from the carrier. A
boosted carrier makes these unwanted electrical
signals weak compared with the wanted signal [8],
as in a coherent receiver.
Fig. 3 plots the qualities of each subcarrier within
the signal band. The middle subcarriers have the
most intermodulation products fall upon them, so
they generally have the worst Qs [9]. In contrast,
the lower-frequency subcarriers are degraded most
in DDO-OFDM systems without boost [6]. With 20-
dB carrier boost, however, the shape of the DDO-
OFDM curve becomes almost identical to that of the
coherent system. Because the coherent systems
shape is completely explained by intermodulation
within the subcarrier band [9], it is likely that carrier-
boosted DDO-OFDM systems have the same
method of degradation.
5 6 7 8 9 10
2
4
6
8
10
Frequency of Subcarrier [GHz]
Q
[
d
B
]
DD with no Rx carrier boost
DD with 5dB Rx carrier boost
DD with 20dB Rx carrier boost
Coherent
Figure 3: Individual subcarrier performance without
precompensation.
Fig. 3 also shows that the 20-dB boosted direct
detection system is superior to the coherent system
by nearly 2-dB over all subcarriers. This is
surprising as the coherent system has perfect
rejection of all out of band frequencies. The reason
is that in DDO-OFDM, the optical carrier
accumulates the same phase modulation as the
subcarriers (in the zero-dispersion limit), so some
nonlinear phase errors are cancelled upon
detection. Coherent systems, however, perform
better with optimal precompensation because in
direct-detection systems, the transmitted carrier will
ultimately walk-off from the sideband in dispersive
fibres.
Conclusion
We have shown that without precompensation, a
DDO-OFDM system with carrier boosting prior to
the photodiode offers a 2-dB improvement over CO-
OFDM systems across all subcarriers when fibre
nonlinearity dominates. When the precompensation
is optimal, carrier boosted direct detection system
performs within 3.5-dB of a coherent system.
Acknowledgements
This work is supported by the Australian Research
Councils Discovery funding scheme (DP 0772937).
References
1. Jansen, L.S. et al., OFC, (2008), paper PDP2
2. Lowery, A.J. et al., J. Lightwave Technol., 25
(2007), 131
3. Lowery, A.J., Photon. Technol. Lett., 19 (2007),
1556
4. Lowery, A.J., Opt. Express, 15 (2007), 12965
5. Shieh, W. Opt. Express, 15 (2007), 9936
6. Du, L. et al., Opt. Express, (2008) in press
7. Lowery, A.J., OFC, (2008), paper OMM4
8. Lowery, A.J., Opt. Express, 16 (2008), 860
9. Lowery, A.J., Opt. Express, 15 (2007), 13282
P.4.08
ECOC 2008, 21-25 September 2008, Brussels, Belgium
Vol. 5 - 148
2 978-1-4244-2228-9/08/$25.00 (c) 2008 IEEE
