Content uploaded by Xiang Liu
Author content
All content in this area was uploaded by Xiang Liu on Dec 03, 2013
Content may be subject to copyright.
IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 15, NO. 4, APRIL 2003 617
Comparison of Return-to-Zero Differential
Phase-Shift Keying and
ON–OFF Keying in
Long-Haul Dispersion Managed Transmission
Chris Xu, Xiang Liu, Linn F. Mollenauer, Fellow, IEEE, and Xing Wei
Abstract—Performance of return-to-zero (RZ) differential
phase-shift keying (DPSK) in ultralong-haul dense wavelength-di-
vision-multiplexing (WDM) dispersion managed transmission is
studied experimentally and compared with conventional
ON–OFF
keying (OOK) in a 10-Gb/s system. We show that, while OOK
out-performs phase-shift keying in a low spectral efficiency WDM
system, the performance of DPSK is comparable to OOK at
10-Gb/s transmission with a spectral efficiency of 0.2. Further-
more, RZ DPSK is advantageous in a high spectral efficiency (e.g.,
0.4) system and our numerical simulation results show superior
performanceof DPSK at10 Gb/s with25-GHz channel separation.
Index Terms—Amplitude shift keying, differential phase-shift
keying (DPSK), optical fiber communication, optical solitons,
phase-shift keying (PSK).
D
IFFERENTIAL phase-shift keying (DPSK) has been
studied for fiber optic transmissions in the past [1]–[7].
Nonetheless, applications of DPSK in dense wavelength-di-
vision-multiplexing (WDM) ultralong-haul (ULH) optical
communications have only recently attracted a lot of interests
[8]–[11]. It was also realized lately that return-to-zero (RZ)
DPSK has several added advantages [3], [8], [9]. The ability of
DPSK to eliminate cross-phase modulation (XPM) penalties
and its significant improvement in receiver sensitivity when
employing balanced receivers [3] have generated many new
excitements and impressive transmission results. Furthermore,
implementation of DPSK with direct detection is straightfor-
ward at 10 and 40 Gb/s [4], [6], [9]. It is, however, also realized
that nonlinear phase noise caused by amplitude fluctuations
and self-phase modulation (SPM) poses new limitations on any
PSK system. Since SPM and XPM depend on the intensity,
amplified spontaneous emission (ASE) noise or nonlinear
interactions caused amplitude fluctuations will translate into
phase noise through both SPM and XPM. It is previously
known that a PSK system is fundamentally limited by ASE and
SPM-induced nonlinear phase noise, the Gordon–Mollenauer
effect [12]. In this letter, we show that the relative performances
of
ON–OFF keying (OOK) and PSK depend on the spectral
efficiency of the system. While OOK out-performs PSK at low
spectral efficiencies, the performance of DPSK is comparable
to OOK at 10-Gb/s transmission with a spectral efficiency of
Manuscript received November 15, 2002; revised December 23, 2002.
C. Xu, X. Liu, and L. F. Mollenauer are with Bell Laboratories, Lucent Tech-
nologies, Holmdel, NJ 07733 USA (e-mail: cx10@cornell.edu).
X. Wei is with Bell Laboratories, Lucent Technologies, Murray Hill, NJ
07974 USA.
Digital Object Identifier 10.1109/LPT.2003.809317
Fig. 1. Experimental setup. For simplicity, only half of the transmitter
is shown. AWG: array waveguide grating. Inset: schematic of RZ DPSK
encoding. “
” and “ ” signs indicate the phases of the pulses.
0.2. Furthermore, RZ DPSK is advantageous in a high spectral
efficiency (e.g.,
0.4) system and our numerical simulation
results show superior performance of DPSK at 10 Gb/s with
25-GHz channel separation.
Transmission experiments were performed in an all-Raman
amplified, dispersion managed system (Fig. 1). The WDM
sources consist of 64 DFB lasers in a single
25 nm band
(1555–1580 nm), spaced at 50 GHz. Odd and even channels
are modulated independently by two transmitters consisting of
two LiNbO
Mach–Zehnder modulators (MZM) in series. The
first modulator of each transmitter driven by a 5-GHz clock is
a pulse carver (PC) generating
33 duty-cycle RZ pulses.
The second MZM is the data modulator (DM), imposing data
either as phase modulation (biased at the null point) or as
intensity modulation (biased at the mid-point). Thus, switching
between DPSK and OOK format can be accomplished by
simply changing the bias point of the MZM. An optical 3-dB
combiner combines the even and odd channels. The combined
channels are sent to a precompensation module (precomp) with
a dispersion of approximately
300 ps/nm. The recirculating
loop comprises six spans ofTrueWaveReduced Slope[(TWRS)
ps/nm/km] fiber. Each span consists of 100 km of
TWRS followed by nearly slope matching dispersion compen-
sation fiber (DCF). The resulting residual dispersion per span
ranges from 12 to 22 ps/nm. A dynamic gain equalizing filter
(DGEF) is placed within the loop after the six spans to equalize
channel powers. Two discrete Raman amplifiers are also used
in the loop to compensate for losses from the DGEF itself,
the acoustooptic switches (AOS) as well as a 3-dB coupler.
Our measurement showed that the loop has an average loss of
32.4 dB 100 km. A tunable bandpass filter (BPF) is used to
1041-1135/03$17.00 © 2003 IEEE
618 IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 15, NO. 4, APRIL 2003
Fig. 2. RZ DPSK at low and high channel power. The path-averaged powers
are indicated.
select the wavelength channel at the receiver end. The signal
is then sent through a post dispersion compensation coil (post-
comp). A delay interferometer (DI) with a 1-bit delay converts
the incoming RZ DPSK signal into intensity-modulated signals
at its two output ports. These signals are differentially detected
by balanced photodiodes. The bit error rate (BER) is then
measured on the regenerated data. Forward error correction
was not employed in our measurements.
One key limitation in any ULH PSK system is the amplitude
noise to phase noise conversion via nonlinear effects such as
SPM and XPM [12]. Such nonlinear transmission penalty in-
creases strongly with the increase of the signal power. Fig. 2
shows RZ DPSK transmission at two different signal power
levels. Clearly, the system performance is significantly better
at the lower transmission power, with an error-free (defined as
) transmission distance of 5400 km. The total
nonlinear phase shift due to SPM is estimated to be
1 radian
at 5000 km with
12.9 dBm/channel, which is close to the
optimum value for ULH PSK transmission [12]. The perfor-
mance degradation at higher channel powersoriginatesfrom the
Gordon–Mollenauer effect. It is well known that a balanced re-
ceiver for DPSK provides close to 3-dB improvements in re-
ceiver sensitivity for a system limited by ASE beat noise [3].
This improvement, however, is reduced when the Gordon–Mol-
lenauer effect becomes dominant [13]. This is confirmed by our
experiments in which we compared a balanced receiver with a
single ended receiver at different signal power levels at a trans-
missiondistanceof 4200 km. A 3-dBadvantageforthebalanced
receiver is obtained at
12.9 dBm/channel, while the advantage
reduces to
1.5 dB at 10.5 dBm/channel. Thus, it is essential
to be in the quasi-linear transmission region in order to obtain
the 3-dB receiver sensitivity advantage for DPSK.
In principle, the SPM mediated Gordon–Mollenauer effect
is a single-channel transmission penalty and is independent
of WDM transmission. The generalized, XPM mediated,
Gordon–Mollenauer effect is a concern in dense WDM.
Nonetheless, its effect is relatively small at a spectral efficiency
of 0.2 [14]. Thus, we expect the relative transmission perfor-
mance of DPSK when compared with OOK to improve as the
spectral efficiency of the system increases. Furthermore, the
improvement in receiver sensitivity for DPSK with balanced
detection allows error-free transmission at significantly lower
signal powers (even lower pulse energies), leading to additional
reduction in nonlinear transmission penalty, which becomes
Fig. 3. Single-channel RZ-OOK and RZ DPSK transmission. The
path-averaged powers are
12.9 dB and 10.5 dBm for DPSK and OOK,
respectively.
Fig. 4. Dense WDM (10 Gb/s at 50-GHz channel separation) RZ-OOK and
RZ DPSK transmission. The path-averaged powers are indicated.
more important in high spectral efficiency systems. Fig. 3
compares low spectral efficiency DPSK and OOK transmission
(by turning
OFF two neighboring channels on each side of the
channel being measured). It was previously known that only
the effects caused by adjacent channels need to be considered
in a high channel count WDM system [15]. OOK clearly
out-performed DPSK in such an extreme case. We note that the
signal power (
18 fJ/pulse or 10.5 dBm/channel) used for the
RZ-OOK transmission approximately satisfies thecondition for
dispersion-managed soliton(with
ps/nm/km). In fact,
even better “single” channel OOK performance can be obtained
by increasing the signal power, such as demonstrated in many
soliton experiments. The results for WDM transmission are
markedly different. Fig. 4 shows the transmission of DPSK and
OOK at 10 Gb/s with a channel separation of 50 GHz. Com-
paring with the data shown in Fig. 3, WDM DPSK performance
is essentially the same as that of the single channel, while the
performance of WDM OOK is significantly reduced. We have
measured many channels within the wavelength band and the
results showed that DPSK performances are comparable or
slightly better than OOK at a spectral efficiency of 0.2. The
variations in channel performance
1dB can be explained by
the OSNR variations across the band. Our experiments clearly
showed that transmission penalty resulting from WDM is not
significant when using DPSK format.
Post dispersion compensation is important for optimum per-
formance of the transmission. High tolerance for variations of
dispersion compensation is essential for a robust and low-cost
system. We have measured RZ DPSK performance at variable
XU et al.: COMPARISON OF RZ DPSK AND OOK IN LONG-HAUL DISPERSION MANAGED TRANSMISSION 619
Fig. 5. Numerical simulation results of dense WDM (10 Gb/s at 25-GHz
channel separation) RZ DPSK transmission with a 25-dB back-to-back
factor. The path-averaged power is 14.5 dBm/channel.
amounts of post dispersion compensation at 4200 km. We
found a dispersion range of greater than
315 ps/nm for a
1-dB penalty, which is much larger than that of a dispersion
managed soliton system. Intuitively, because DPSK transmis-
sion is essentially in the “linear” region, the function of post
dispersion compensation in DPSK is to restore the pulse width.
While in addition to restoring pulse width, post dispersion
compensation also compensates some nonlinear transmission
penalties in an OOK system. For example, soliton timing jitter
can be reduced in part by using appropriate amount of post
dispersion compensation.
We have extended the comparison of DPSK and OOK at
10 Gb/s and 25-GHz channel separations through numerical
simulations in the absence of experimental data. Our numer-
ical simulations have been providing valuable insights in DPSK
performance, with predictions mostly confirmed by our exper-
iments and other recent results [9]. Fig. 5 shows the numerical
simulation results of RZ DPSKat 10 Gb/s with 25-GHz channel
separation. The modeling parameters are similar to our exper-
imental setup, except that the path-averaged power is further
reduced. An error-free transmission distance of
4500 km is
predicted. The reach is longer than an OOK system. (For ex-
ample, timing jitter caused by soliton collisions alone will limit
the system reach to
3500 km for a dispersion managed soliton
system at 10 Gb/s and 25-GHz channel spacing [15].)
We note that there are advantages by trading XPM penalty
(dominant in OOK) with SPM penalty (dominant in DPSK).
SPM in a PSK system is bit-pattern independent, while XPM in
an OOK system is bit-pattern dependent. Ways of compensating
the Gordon–Mollenauer effect have already been proposed to
enhance the performance of a PSK system [16]. The absence of
significant XPM penalty also allows polarization division mul-
tiplexing in a DPSK system, further increasing the spectral effi-
ciency and capacity of a system [17].
In summary, we have shown that OOK out-performs PSK in
a low spectral efficiency system. However, at 10-Gb/s transmis-
sion with 50-GHz channel separation, RZ-OOK and RZ DPSK
performances are comparable. We further argue that RZ DPSK
is advantageous in a high spectral efficiency system and our nu-
merical simulation results showed superior performance of RZ
DPSK at 10 Gb/s with 25-GHz channel separation.
A
CKNOWLEDGMENT
The authors would like to thank A. Grant, A. Chraplyvy,
C. Mckenstrie, S. Hunsche, R. Giles, R. Slusher, A. Gnauck,
P. Winzer, H. Kim, and D. Fishman for valuable discussions,
and C. Doerr for providing the DI and DGEF.
R
EFERENCES
[1] R. A. Linke and A. H. Gnauck, “High-capacity coherent lightwave sys-
tems,” J. Lightwave Technol., vol. 6, pp. 1750–1769, Nov. 1988.
[2] J.-K. Rhee, D. Chowdhury, K. S. Cheng, and U. Gliese, “DPSK
10 Gb/s transmission modeling on 5 90 km terrestrial system,” IEEE
Photon. Technol. Lett., vol. 12, pp. 1627–1629, Dec. 2000.
[3] W. A. Atia and R. S. Bondurant, “Demonstration of return-to-zero sig-
naling in both OOK and DPSK formats to improve receiver sensitivity
in an optically preamplified receiver,” presented at the LEOS’99, San
Francisco, CA, 1999, Paper TuM3.
[4] M. Hanna, H. Porte, J.-P. Goedgebuer, and W. T. Rhodes, “Performance
assessment of DPSK soliton transmission system,” Electron. Lett., vol.
37, pp. 644–646, 2001.
[5] H. Nishizawa, Y. Yamada, Y. Shibata, and K. Habara, “10-Gb/s op-
tical DPSK packetreceiver proof againstlarge power fluctuation,” IEEE
Photon. Technol. Lett., vol. 11, pp. 733–735, June 1999.
[6] M. Rohde, C. Caspar, N. Heimes, M. Konitzer, E.-J. Bachus, and N.
Hanik, “Robustness of DPSK direct detection transmission format in
standard fiber WDM systems,” Electron. Lett., vol. 36, pp. 1483–1484,
2000.
[7] P. Shum, H. Ghafouri-shiraz, and S. F. Yu, “Analysis of DPSK soliton
transmission system,” Opt. Laser Technol., vol. 29, pp. 411–414, 1997.
[8] J. Leibrich, C. Wree, and W. Rosenkranz, “CF-RZ-DPSK for suppres-
sion of XPM on dispersion-managed long-haul optical WDM transmis-
sion on standard single-mode fiber,” IEEE Photon. Technol. Lett., vol.
14, pp. 155–157, Feb. 2002.
[9] A. H. Gnauck, G. Raybon, S. Chandrasekhar, J. Leuthold, C. Doerr,
L. Stul, A. Agarwal, S. Banerjee, D. Grosz, S. Hunsche, A. Kung, A.
Marhelyuk, D. Maywar, M. Movassaghi, X. Liu, C. Xu, X. Wei, and
D. M. Gill, “2.5 Tb/s (64
42.7 Gb/s) transmission over 40 100 km
NZDSF using RZ-DPSK format and all-Raman-amplified spans,” pre-
sented at the OFC’2002, Anaheim, CA, 2002, Postdeadline Paper FC-2.
[10] V. S. Grigoryan, P. S. Cho, and I. Shpantzer, “Nonlinear penalty reduc-
tion of RZ-DPSK versus RZ-OOK modulation format in fiber commu-
nications,” in Proc. ECOC’2002, 2002, p. 3.
[11] T. Miyano, M. Fukutoku, K. Hattori, and H. Ono, “Suppression of
degradation induced by SPM.XPM
GVD in WDM transmission using
a bit-synchronous intensity modulated DPSK signal,” presented at the
OECC’00, Chiba, Japan, 2000, Paper 14D3-3.
[12] J. P. Gordon and L. F. Mollenauer, “Phase noise in photonics com-
munications systems using linear amplifier,” Opt. Lett., vol. 15, pp.
1351–1355, 1990.
[13] H. Kim and A. Gnauck, “Experimental investigation ofthe performance
limitationofDPSKsystemsduetononlinearphasenoise,”IEEEPhoton.
Technol. Lett., vol. 15, pp. 320–322, Feb. 2003.
[14] X. Liu, X. Wei, R. Slusher, and C. J. McKinstrie, “Improving transmis-
sion performance in PSK systems by lumped nonlinear phase shift com-
pensation,” Opt. Lett., vol. 27, pp. 1626–1628, 2002.
[15] C. Xu, C. Xie, and L. F. Mollenauer, “Analysis of soliton collisions in a
wavelength-division-multiplexed dispersion-managed soliton transmis-
sion system,” Opt. Lett., vol. 27, pp. 1303–1305, 2002.
[16] C. Xu and X. Liu, “Post-nonlinearity compensation with data driven
phase modulators in phase shift keying transmission,” Opt. Lett., vol.
27, pp. 1629–1631, 2002.
[17] C. Xu, X. Liu, and X. Wei, “Ultra-long haul DWDM transmission with
differentialphase shift keyingdispersion managed soliton,” presented at
the ECOC’2002, Copenhagen, Denmark, 2002, Paper 1.1.5.