INVITED
PAPER
100G and Beyond Transmission
Technologies for Evolving
Optical Networks and Relevant
Physical-Layer Issues
The authors outline technologies that are enabling next-generation optical fiber
communication systems with channels that support 100-Gb/s and higher rates.
By Ezra Ip, Philip Ji, Eduardo Mateo, Yue-Kai Huang, Lei Xu,
Dayou Qian, Neng Bai, and Ting Wang
|
As 100-Gb/s= digital coherent systems enter
compensation; optical time division multiplexing; optical signal
commercial deployment, an effort is underway to uncover the
processing; orthogonal frequency division multiplexing; single
technologies that will enable the next-generation optical fiber
communication systems. We envisage that future optical trans-
mode fibers; space division multiplexing
ABSTRACT
port will be software-defined, enabling flexible allocation of
bandwidth resources, with dynamically adjustable per-channel
I. INTRODUCTION
Digital Object Identifier: 10.1109/JPROC.2012.2183329
Optical fiber forms the backbone of modern telecommunication systems. For decades, optical fiber provided greater
signal bandwidth than could be utilized. However, with
global Internet traffic demand continuing to grow at
40% per year [1]Vdriven by services such as video
sharing, high-definition television on demand, and cloud
computingVengineers are faced with new challenges.
For network operators, the ultimate parameter of interest is the cost per bit-per-second.kilometer (b/s.km).
The challenge is to increase system capacity-distance at a
rate substantially higher than linear with hardware cost
and operating cost. Early fiber optic systems were based on
time-division multiplexing (TDM), which relied on increasing the speed of optoelectronics. When the achievable
bandwidths of electronics saturated [2], wavelengthdivision multiplexing (WDM) enabled continued capacity
growth by transmitting parallel channels of information
over the same fiber [2]–[4]. Then in the 1990s, inline
erbium-doped amplifiers revolutionized fiber optic networks by enabling unrepeatered transmission over transcontinental and transoceanic distances [5]. Since the
2000s, the primary driver of capacity-distance growth has
been improved spectral efficiency using advanced modulation formats and signal detection schemes [6]–[8]. The
0018-9219/$31.00 Ó 2012 IEEE
Vol. 100, No. 5, May 2012 | Proceedings of the IEEE
data rates based on instantaneous traffic demand and qualityof-service requirements, leading to unprecedented network
agility. Software-defined transponders will have the programmability to adopt various modulation formats, coding rates,
and the signal bandwidth based on the transmission distance
and type of fiber. Digital signal processing will become
increasingly ubiquitous and sophisticated, capable of compensating all types of channel impairments, enabling advanced
forward error correction coding, and performing functions
previously handled poorly by optical analog hardware such as
spectrum shaping and demultiplexing of optical channels.
KEYWORDS
|
Coherent detection; digital signal processing;
error correction coding; fiber optic communications; modulation formats; multimode fibers; multicore fibers; nonlinear
Manuscript received July 25, 2011; revised November 18, 2011; accepted
December 31, 2011. Date of publication March 16, 2012; date of current version
April 18, 2012.
E. Ip, P. Ji, E. Mateo, Y.-K. Huang, L. Xu, D. Qian, and T. Wang are with NEC
Laboratories America, Princeton, NJ 08540 USA (e-mail: ezra.ip@nec-labs.com).
N. Bai is with the CREOL, University of Central Florida, Orlando, FL 32816 USA.
1065
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
most advanced commercialized systems today operate at
100 Gb/s per transponder and use dual-polarization quadriphase shift keying (DP-QPSK) as the transmission format, with coherent detection at the receiver and digital
signal processing to compensate transmission impairments
[9]–[11].
The question then is the following: What new technologies will be employed for next-generation systems, which
will not only require higher data rate per channel, but will
require flexibility to adapt to constantly changing traffic
condition on the network?
Improving capacity-distance product ultimately depends on doing one or more of the following: 1) increasing
spectral efficiency; 2) increasing the total available bandwidth; and 3) reducing the amount of signal distortion
accumulated per unit distance. Realizing these goals requires different technologies with different challenges as
well as tradeoffs. For example, spectral efficiency may be
increased by using larger signal constellations that transmit more bits per symbol [12]–[14]. However, this will
render a system less tolerant to noise and other distortions,
thus reducing the distance between repeaters. A promising
proposal to increase spectral efficiency without sacrificing
sensitivity per bit is to improve network bandwidth utilization. Current systems operating on a fixed ITU grid with
guard bands between adjacent channels. There have been
suggestions to move toward a Bgridless[ network [15]–
[18], where optical transmission use Bsuperchannels[ can
enable an ultrawideband signal to continuously occupy an
allotted bandwidth, removing the spectrally wasteful guard
bands [19]–[23]. To reduce accumulated signal distortion,
it is possible to improve the link hardware by using shorter
fiber spans, lower noise amplifiers, distributed Raman
amplification [24], and advanced fibers with ultralow loss
and large effective area [25], [26]. Recently, digital signal
processing techniques including nonlinear compensation
[27], [28] and advanced forward error-correction coding
[29] has been proposed to reduce distortion or improve
signal sensitivity. Ultimately, the achievable spectral
efficiency-distance product is constrained by fiber nonlinearity. It has been shown that unlike most transmission
media where capacity can be increased by using higher
power, a Bnonlinear Shannon’s Limit[ exists in optical
fiber [30], [31]. While the total bandwidth available for
WDM transmission can be increased somewhat by removing water absorption in glass and using new optical amplifiers for previously unused bands, fundamental physical
limits exist. To improve capacity per fiber beyond that
achievable with single-mode fibers, radical new ideas such
as using multicore fibers and mode-division multiplexing
using multimode fibers have been suggested [32]–[34].
Nevertheless, the progress witnessed in the field of
optical communications has been immense [35]–[39].
Only a decade ago, the record capacity of single-mode fiber
was 6.4 Tb/s [40]. The combination of advanced modulation formats, coherent detection, and digital signal pro1066
Proceedings of the IEEE | Vol. 100, No. 5, May 2012
cessing has seen record capacity increasing sixteen-fold to
101 Tb/s by 2011 [39], at an average rate of 32% per year,
barely in keeping with demand growth.
As of today, 100-Gb/s coherent systems have already
been successfully deployed in the US, Japan, and Europe.
There has been much discussion on the modulation formats, signal generation schemes, digital signal processing
algorithms, and optical components for beyond-100G
systems [41]–[46]. In this paper, we will review the major
technologies that are likely to play a role in shaping nextgeneration ultrahigh-capacity optical networks, comparing
their pros and cons. We will also investigate the latest
advances in such areas as nonlinear compensation and
spatial multiplexing that may play a role in future systems,
where it is assumed that continual improvement in the
computational efficiency of silicon will continue to exhibit
Moore’s Law growth, enabling digital signal processing
(DSP) options that are beyond reach of current 100G
systems [47]. The format of this paper is as follows. In
Section II, we review the advantages of coherent detection
and digital signal processing, and in particular, the emergence of an all-digital platform providing flexibility, reconfigurability, and network agility. In Section III, we discuss
transmission technologies, including O-TDM and multicarrier Bsuperchannel[ transmission using O-OFDM. The
benefits and limitations of each technology are reviewed.
Section IV discusses nonlinear compensation techniques,
while Section V discusses advance forward error correction (FEC). Both of these sections highlight recent advances aimed at reducing algorithmic complexity so these
techniques may be implementable in real-time in nextgeneration systems. Finally, Section VI covers SDM, including recent results using MMF and MCF, comparing
their benefits and limitations.
I I. DIGITAL T RANSMITTER
A N D RE CE I VE R
Coherent detection and DSP were the key enabling technologies in the development of 100G systems [48]–[52].
Next-generation systems will likely continue this trend,
with DSP playing an even more ubiquitous role, where
advanced algorithms will be used for compensating fiber
impairments and perform other signal processing functions that are impractical to perform in analog hardware.
A. Digital Receiver
The advantage of digital coherent receivers stems from
the ability to arbitrarily manipulate the electric fields in
the two signal polarizations. In conjunction with sampling
by analog-to-digital converters (ADCs) above Nyquist rate,
the digitized waveform retains all the information of the
analog optical waveform (Fig. 1). Therefore, any operation
traditionally performed by analog hardware can be duplicated in digital software. Consider chromatic dispersion
(CD) compensation (CDC), for example. Traditionally,
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
Fig. 1. Digital transmitter and receiver.
CD must be compensated optically using dispersion
compensation fiber (DCF) before direct or interferometric
detection in order to recover the signal with penalty. The
sensitivity of these systems to uncompensated dispersion
scales as the square of the baud rate [53]. Hence, optical
CDC faces very stringent requirements at high baud rates.
In legacy 40G DPSK systems, transponders required dedicated CDC modules to fine-tune residual CD (due to fiber
dispersion slope) at its particular wavelength. However,
CD is a linear time-invariant impairment and can be compensated efficiently by DSP [7]. A digital coherent receiver
enables the same transponder to be used for detecting any
WDM channel since the amount of CDC to be applied can
be dynamically changed by setting the coefficients of the
digital CD equalizer [9]. Digital CDC also has other advantages. First, DCF incurs loss. By eliminating the DCF,
signal-to-noise ratio (SNR) can be increased. Second, dispersion unmanaged transmission can significantly reduce
nonlinear penalties due to the walk-off effect [54]. Removing the DCF thus increases the nonlinear threshold and the
achievable capacity. When CD is compensated in the frequency domain via a frequency-domain equalizer (FDE),
algorithmic complexity only scales logarithmically with residual dispersion. Hence, digital CDC scales well with
increasing transmission distance.
Another major benefit of digital coherent receivers is
that it facilitates channel demultiplexing. In traditional
WDM systems, the channel of interest is selected by optical filtering devices such as optical interleavers, arrayed
waveguide gratings (AWGs) [55], and wavelength-selective
switches (WSSs) [56]. Optical filters have finite frequency
roll-off, necessitating guard bands between adjacent WDM
channels, reducing bandwidth utilization and hence
spectral efficiency. With a digital coherent receiver, optical filters are no longer necessary. It is possible to tune a
local oscillator (LO) laser near the carrier frequency of the
channel of interest and use a digital filter to extract the
signal. Digital filters have sharper roll-off and have more
controllable characteristics compared to analog hardware
filters. DSP-based channel selection without optical
filtering in WDM systems has been studied in [57].
Finally, a digital coherent receiver enables novel
modulation formats that are difficult to detect using
only analog hardware. One example is electrical OFDM
(E-OFDM), where information is transmitted over a set of
orthogonal Bsubcarriers[ generated via a fast Fourier
transform (FFT). This modulation format and its unique
system advantages are discussed in Section IV-C [58], [59].
B. Digital Transmitter
DSP can also be used at the transmitter to improve
system performance. In a mirror image of the receiver,
DSP is used to compute the desired waveform at a sampling rate above the Nyquist criterion. Digital-to-analog
converters (DACs) are then used to produce the electrical
drive signals for the optical modulators.
One of the benefits of transmitter-side DSP is spectral
shaping. In the case of E-OFDM, DSP is required for computing the inverse FFT (IFFT). However, DSP can also be
used to generate raised-cosine (RC) pulses in single-carrier
systems. RC pulses have spectral properties than optical
return-to-zero pulses that are traditionally used as they are
easy to produce in analog hardware. In particular, an RC
pulse with 0% roll-off is the sinc pulse (Nyquist pulse
shape) and is the most spectrally efficient pulse shape
since it occupies a bandwidth of Rs when modulated at a
baud rate of Rs . The sinc function has infinite tails and is a
noncausal function, therefore while it is difficult to
generate in analog hardware, it is simple to implement
in DSP. Single-carrier signals with Nyquist pulse shaping
are as spectrally efficient as OFDM and can have advantages such as better nonlinear performance in legacy
dispersion-managed systems, where reduction in signal
amplitude fluctuation can be advantageous. Additionally,
single-carrier transmission eliminates the need for frame
synchronization, simplifying receiver DSP. Single-carrier
transmission using DP-16QAM with Nyquist pulse
shaping at 112 Gb/s has been demonstrated in real time
in [60].
Another benefit of digital transmitters is that it enables
channel impairment predistortion. For example, it has
been shown that splitting nonlinearity compensation
Vol. 100, No. 5, May 2012 | Proceedings of the IEEE
1067
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
(Section V) between transmitter and receiver can lead to
improvement in overall system performance [61], [62].
In summary, a software-defined optical system with
DSP at both transmitter and receiver enables the most
agile platform. Not only does it allow channel impairments
be compensated by powerful DSP algorithms, it also enables new signal (de)multiplexing paradigms. In addition,
the channel data rate, modulation format, and coding
scheme can all be programmed by network management in
response to changing channel conditions. This enables
network flexibility since optical signals may be routed
anywhere, irrespective of the distance, fiber type, or the
number of reconfigurable optical add/drop multiplexers
(ROADMs) transited in correspondence with adjusting the
modulation format and coding scheme to provide reliable
end-to-end connection at the highest data rate possible.
III . TRANSMISSION TECHNOLOGIES
In this section, we review transmission technologies that
are considered prime candidates for 100G beyond-100G
systems. We assume the use of coherent detection, which
allows information to be encoded in amplitude and phase
of both fiber polarizations. The transmitted signal assumes
the canonical form
xðtÞ ¼
Nc
dX
1
2 e1
X
xn;m pðt nTs Þ expðj!m tÞ
(1)
n¼bN2c c m¼1
where xn;m ½xn;m ; yn;m T is a two-dimensional complexvalued symbol transmitted at the mth subcarrier in the
nth symbol period, !m is the frequency of the mth subcarrier (relative to the channel’s carrier frequency), Ts is
the symbol period, and pðtÞ is the pulse shape. This formulation is sufficiently general to describe all the signaling
formats to follow. It is apparent that the data rate of xðtÞ
can be increased by: 1) using larger signal constellations
for xn;m ; 2) increasing the symbol rate Rs ¼ 1=Ts ; or
3) increasing the number of subcarriers Nc . We will investigate each of these methods and their associated benefits
and limitations.
A. Higher-Order Modulation
One method to increase capacity is to use larger signal
constellations. The spectral efficiency and power efficiency of modulation formats has been studied in [7]. For
digital coherent systems, dual-polarization M-ary quadrature-amplitude modulation (DP-MQAM) has become the
modulation format of choice. By utilizing all the degrees of
freedom available for encoding information, DP-MQAM
achieves the best spectral and power efficiency amongst all
easy-to-generate modulation formats like dual-polarization
M-ary phase-shift keying (DP-MPSK). In the presence of
1068
Proceedings of the IEEE | Vol. 100, No. 5, May 2012
nonlinearity, power efficiency is especially important because of the existence of an optimal launch power at which
capacity is maximized. Using a modulation format with
low SNR requirement increases the achievable capacity.
However, data rate cannot be increased indefinitely by
higher-order modulation because of the nonlinear
Shannon’s limit [31]. The highest spectral efficiencies
achieved to date in experimental systems are 12.4 b/s/Hz
for single-channel transmission [13] and 11 b/s/Hz for
WDM transmission [39].
B. Time-Division Multiplexing (TDM)
Another method to increase capacity is to increase the
baud rate, which increases signal bandwidth without
changing spectral efficiency [63]. However, as the bandwidths of electronic components such as signal generators,
driver amplifiers, and electrooptic modulators at the transmitter and photodiodes, transimpedance amplifiers, and
ADCs at the receiver are much narrower than the multiTHz bandwidth of a fiber’s transparent region, the data
rate achievable using electrical TDM (E-TDM) is limited.
1) Optical Time-Division Multiplexing: It is possible to
overcome the bandwidth limitation of electronics by leveraging optical signal processing techniques. In optical TDM
(O-TDM), short-duration pulses can be generated either
using a pulsed laser [64] or by concatenating optical modulators [65]. Pulsed compressors based on highly nonlinear
fiber can be used to further reduce the duty cycle of the
generated pulses [66], [67]. The optical pulse train is split
into parallel paths; in each path, data is modulated using
standard Mach–Zehnder (MZ) in-phase/quadrature (I/Q)
modulators (Fig. 2). The independently modulated data
paths are then combined with appropriate mutual delays to
Fig. 2. O-TDM (a) transmitter and (b) receiver.
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
create an O-TDM signal. At the receiver, the signal is
combined with a pulsed LO using a standard coherent
receiver front end consisting of a polarization-and-phase
diversity hybrid followed by photodetectors. Only when
signal pulses are aligned with the LO pulses will an
electrical output be produced. In theory, an N OTDM
system can increase the achievable baud rate by N.
The disadvantage of O-TDM is that the optical spectrum is dependent on the quality of the pulses produced by
the pulse generator, which cannot be easily controlled. It is
possible to use optical filters to truncate the optical spectrum to the desired bandwidth. However, this can cause
intersymbol interference (ISI), which may require a large
number of taps to equalize if a finite impulse response
(FIR) filter is used, or require maximum likelihood sequence estimation (MLSE).
C. Optical Orthogonal
Frequency-Division Multiplexing
Another method to increase capacity is to transmit
data over multiple subcarriers. Suppose in (1), pðtÞ ¼
rectðt=Ts Þ is a rectangular pulse with duration Ts , and the
subcarrier frequencies are !m 2m=Ts where bNc =2c m G dNc =2e 1, with bxc and dxe being the nearest integers above and below x. Then, it can be shown that the set
of basis functions m ðtÞ ¼ rectðt=Ts Þ expðj2ðm=Ts ÞtÞ are
orthogonal over any continuous time interval t G þ
Ts of duration Ts . OFDM has been studied in the context of
mobile communications and is an important technique for
combating multipath fading. Recently, OFDM has received
interest in optical fiber transmission as a means of spectral
shaping, reducing the guard band requirement between
neighboring channels and thus increasing bandwidth utilization and spectral efficiency. In optical Bsuperchannels,[
optical techniques are used to generate optical subcarriers.
Consider the system shown in Fig. 3, where the transmitter has a tone generatorVusually a pulsed laser or an
overdriven MZ modulatorVwhich produces phase-locked
optical tones. In contrast with O-TDM where the pulses
are manipulated in the time domain, optical OFDM
(O-OFDM) manipulates them in the frequency domain.
The phase-locked tones are first demultiplexed using an
AWG or a WSS, followed by parallel data modulation,
followed by recombining.
Two types of optical OFDM have been described:
single-carrier modulation on each optically generated subcarrier (O-OFDM/SC), and electrical OFDM on each
optically generated subcarrier (O-OFDM/E-OFDM).
O-OFDM/SC has the advantage of not requiring a fully
digital transmitter. Apart from the tone generator,
O-OFDM/SC uses identical transmitters and receivers as
developed for current-generation 100G coherent systems
[19]. Hence, O-OFDM/SC is an attractive candidate for
system upgrade. The data rate of the superchannel can be
scaled by simply increasing the number of optically
generated subcarriers and single-carrier transponders. To
Fig. 3. O-OFDM (a) transmitter and (b) receiver.
detect a subcarrier of interest, the receiver tunes a
continuous-wave (CW) LO near its center frequency and
uses a standard coherent receiver front end to downconvert
the optical signal to electrical baseband. Assuming the baud
rate of the subcarriers is equal to the subcarrier spacing
(also known as zero-guard-band OFDM) [19], residual CD
must be first compensated using a frequency-domain
equalizer (FDE) to restore orthogonality between the
subcarriers. Assume the DSP sampling rate is M times the
subcarrier symbol rate, the subcarrier of interest can be
demultiplexed digitally by initializing an adaptive timedomain equalizer (TDE) with coefficients equal to the first
column of an M-point FFT matrix. The TDE coefficients can
thereafter be adapted using any commonly used algorithm
for single-carrier systems such as constant modulus
algorithm (CMA) or decision-directed algorithm (DD)
[68]. Carrier recovery and symbol detection follows. Thus,
apart from the possibility of higher oversampling requirement and the need to properly initialize the adaptive TDE,
O-OFDM/SC is detected using the same DSP algorithm as
current-generation 100G coherent systems. If the bandwidth of the coherent receiver front end is sufficiently
wide, it is possible to detect multiple subcarriers per transponder. Consider Fig. 4 where the LO is placed midway
between two subcarriers. The upper and lower subcarriers
can be shifted to DC by DSP before invoking the
impairment compensation algorithms described above.
O-OFDM/SC has practical problems, however. These
include the following.
i) When Nc is very large, it is difficult to ensure all
the tones generated by the tone generator have
equal power. It is often necessary to use a WSS as
a power equalizer, followed by an EDFA to boost
Vol. 100, No. 5, May 2012 | Proceedings of the IEEE
1069
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
2)
Fig. 4. Demodulating multiple subcarriers per transponder.
all the subcarriers to the required power level,
causing transmitter-side noise.
ii) Perfectly rectangular pulses are impossible to
generate due to their infinite bandwidth. When
imperfect rectangular pulses are used, loss of
orthogonality between the subcarriers results in
interchannel crosstalk (ICI). To ensure good orthogonality, the baud rate per optical subcarrier
needs to be much lower than the bandwidth of the
transmitter hardware.
iii) The receiver, likewise, requires high bandwidth to
ensure orthogonality of the subcarriers before demultiplexing. In particular, large signal constellations are sensitive to ICI and filtering distortions.
It is necessary to use not only high-bandwidth
components, but also large DSP oversampling
ratios. For example, while O-OFDM/SC with
DP-QPSK per subcarrier achieves acceptable
performance at 2 oversampling, and has been
demonstrated for a subcarrier baud rate of 25 GHz
[69], O-OFDM/SC with DP-16QAM per subcarrier requires 4 to 6 oversampling and has been
demonstrated at a subcarrier baud rate of only
12.5 GHz [70]. Thus, O-OFDM/SC may have significantly increased DSP requirement than nonfrequency-overlapping WDM channels at high
spectral efficiencies.
Since (i) favors using smaller Nc , while (ii) and (iii) favors
larger Nc , it is difficult to optimize all three conditions at
once. In practice, most experimental systems use baud
rates between 12.5 and 25 GHz per optical subcarrier, and
the highest spectral efficiency achieved using AO-OFDM/
SC is only 7 b/s/Hz [71]. Nevertheless, AO-OFDM/SC is an
effective technique for scaling up the per-channel data
rate. Indeed, optical transmission treating the entire
C-band as a single continuous Bsuperchannel[ has been
demonstrated using this technique [22].
It is possible to alleviate the oversampling problem by
using O-OFDM/E-OFDM, where DACs are used to generate orthogonal subcarriers. The advantages of E-OFDM
include the following.
1) Ease of generating a large number of electrical
subcarriers.
1070
Proceedings of the IEEE | Vol. 100, No. 5, May 2012
Low baud rate per subcarrier improves their mutual orthogonality, hence reducing ICI.
3) High oversampling ratio relative to the subcarrier
baud rate reduces ICI.
4) DSP simplifies nonlinear operations such as insertion and removal of the cyclic prefix for intersymbol interference (ISI) mitigation.
5) A digital transmitter enables much more precise
over the transmitted waveform, enabling better
quality signal constellations compared to using an
analog transmitter. Also, when a large number of
electrical subcarriers are used, the signal spectrum is nearly rectangular, reducing the guard
band requirement even if the carriers of neighboring channels are not phase-locked.
In particular, E-OFDM demodulation is equivalent to
detecting multiple O-OFDM/SC subcarriers in one
transponder: The digital FFT operation is equivalent to
initializing adaptive TDEs with different tap values to
extract the different subcarriers. Provided the number of
electrical subcarriers Nc is large, the frequency response
over the bandwidth of a subcarrier will not change
substantially, enabling frequency-domain compensation
of CD and polarization-mode dispersion (PMD) via 2 2
matrices (Bsingle-tap[ filter) at each electrical subcarrier
[72]. In dispersion unmanaged systems, however, the Nc
required to keep the cyclic prefix overhead reasonable can
be very large due to the channel’s long impulse response.
Consequently, the received signal is sometimes treated as a
single-carrier signalVi.e., the known CD of the channel is
first compensated using an FDE. The cyclic prefix then
only has to account for a short channel impulse response
caused by PMD and other statistical effects, reducing the
number of electrical subcarriers. This scheme is also
referred to as Breduced-guard interval[ OFDM (RGIOFDM). An efficient implementation that combines an
FDE with OFDM demultiplexer was studied in [73].
If the frequency spacing between electrical subcarriers
is an integer fraction of the frequency spacing between
phase-locked optical subcarriers, it is possible to overlap
the E-OFDM of adjacent channels, while all the subcarriers remain detectable [59].
Still, the O-OFDM/E-OFDM technique has drawbacks.
These include the following.
1) Using a large number of subcarriers may render
the system more sensitive to laser phase noise.
2) E-OFDM detection requires frame synchronization, increasing DSP complexity.
3) In legacy dispersion-managed systems, E-OFDM
may be more sensitive to nonlinearity. This is
because when Nc is large, the amplitude of xðtÞ
becomes Gaussian with large peak-to-averagepower ratio (PAPR). It is well known that
dispersion-managed systems favor constant amplitude modulation formats like M-ary PSK.
Indeed, one alternative is to use DSP to generate
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
Nyquist-shaped pulses: Blocks of Nc symbols are
generated using FFT and IFFT, with optional cyclic
prefix for each block like in OFDM [74].
Ultimately, if the cost of extra DSP and DACs at the
transmitter is not an issue, O-OFDM/E-OFDM has the best
potential of all the transmission technologies discussed.
The highest spectral efficiency and highest capacity
reported for a WDM system in a single-core single-mode
fiber was achieved using this technique [39].
IV. NONLINEARI TY COMPENSATION
The capacity of optical fiber is ultimately limited by the
Kerr nonlinearity, where refractive index changes with
field intensity, causing localized phase shift proportional
to power as the signal propagates. In the absence of noise,
a single-channel signal is limited by self-phase modulation (SPM), whereas WDM systems are limited by crossphase modulation (XPM) and four-wave mixing (FWM)
[30], [53].
Optical phase conjugation (OPC) has been proposed
for the comprehensive compensation of fiber impairments.
Based on spectral inversion of the signal, dispersion and
nonlinear distortions experienced in the part of the link
before OPC are reversed (compensated) in the subsequent
part. Typically, the OPC operation is located in the middle
of the link so that signal distortions are canceled out
provided that group delay and nonlinear phase shift occur
in a symmetric fashion with respect to the OPC location.
OPC has been experimentally demonstrated for dispersion
compensation [75] as well as for the compensation of
nonlinear effects [76].
Since the nonlinear Schrödinger equation (NLSE) for
signal propagation is a deterministic equation, SPM, XPM,
and FWM can also be compensated by using DSP to solve
in inverse NLSE (iNLSE). This is known as digital backpropagation (DBP). Ultimate system capacity is then
limited by statistical and nondeterministic nonlinear interaction between signal and noise.
In currently deployed 100G systems, there has been no
attempt to digitally compensate nonlinearity due to the
algorithmic complexity of nonlinearity compensation
(NLC) algorithms. However, as DSP capability improves,
and as future systems seek to achieve the highest capacity
possible, NLC may well become practical. Consider doubling system capacity by using DP-16QAM. In the linear
regime, DP-16QAM requires around 7 dB higher SNR than
DP-QPSK [7]. As a result, the longest transmission distance reported for this format without NLC is only
3123 km for single-channel transmission [77] and 2000 km
for WDM [59]. However, for DP-16QAM to be practically
deployed over nationwide terrestrial networks, it needs to
reach at least 1500 km with system margin. One way to
realize this may be to use nonlinear compensation. The
system impact of NLC is shown in Fig. 5, showing typical characteristic curves for no NLC with self-phase
Fig. 5. Effect of NLC on the performance of fiber transmission system.
modulation compensation and cross-phase modulation
compensation.
The vertical axis labeled BQ[ denotes the quality of the
signal constellation, defined as signal power divided by the
mean error vector magnitude. If the error vector is
Gaussian, the relationship between Q and capacity is well
known. Nonlinear compensation pushes the onset of the
nonlinear regime toward higher powers, increasing the
optimal launch power and the ultimate performance
achievable. The increase in Q depends on the dispersion
map, signal modulation format, and the signal’s power
spectrum.
A. Intrachannel Nonlinearity Compensation
The simplest NLC is SPM compensation (SPMC),
which only compensates the nonlinear effects of a channel
on itself. In a digital coherent receiver, this means replacing the CDC with SPMC, which jointly compensates CD
and SPM. Typically, solving the iNLSE involves dividing
the total channel into multiple steps such that the chromatic dispersion and nonlinear phase accumulated at each
step are small. The overall complexity of the algorithm
depends on the total number of steps, or inversely with the
step size. In general, CD affects a signal much more
strongly than nonlinearity. Hence, if the most accurate
solution of the NLSE is desired, the required step size is
mostly determined by the characteristic dispersion length
[28]. However, the computational requirement may be
impractically large for long haul systems. A variety of simplified DBP have been proposed that trade off algorithmic
complexity with performance.
For dispersion managed systems, one option is to use
multisubband filtered DBP, which exploits the walk-off
effect of dispersion whereby a given frequency packet of
the signal experiences stronger nonlinear effects from frequencies closer to it than frequencies far away. Thus, by
partitioning the signal into subbands and calculating the
nonlinear effect accurately with filtering, larger step sizes
can be used. This technique has been studied in [78] and
[79]. Recently, it was found that in dispersion managed
systems where Nspan spans are identical (or near identical),
it is possible to use a Bfolded DBP[ that collapses Nspan
Vol. 100, No. 5, May 2012 | Proceedings of the IEEE
1071
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
Fig. 6. Improved NLC algorithm by using a low pass filter (LPF)
in the nonlinear step.
spans of the actual link into a single equivalent span with
Nspan times the nonlinearity [80], [81].
For dispersion unmanaged systems, it is possible to use
filtered DBP. Consider the iterative algorithm shown in
Fig. 6.
Since the nonlinear phase accumulated in one step is
proportional to the integral of signal intensity, when large
steps are taken, dispersion causes the high frequencies of
the power envelope to be misestimated. Such wrong estimation contributes to the system noise making NLC
inefficient. It has been found that by removing those components, the step size can be significantly increased while
keeping a reasonable performance of NLC. Consequently,
we can modify DBP by low-pass-filtering the signal intensity before doing phase derotation as shown in Fig. 6.
Filtered DBP has been studied in [82], which considered a
time-domain LPF. In [83], the same approach is implemented in the frequency domain. Both techniques are
equivalent, and indicate the DBP can enable step sizes as
large as four fiber spans per step.
Although dispersion unmanaged transmission requires
higher algorithmic complexity than dispersion managed
transmission irrespective of whether LE or SPMC is used,
dispersion unmanaged systems always achieve superior
performance due to enhanced walk-off and reduction in
path loss.
B. Interchannel Nonlinearity Compensation
In WDM systems, XPM and FWM are the dominant
effects. Hence, the performance improvement achievable
with SPMC alone is quite limited. To increase nonlinearity
tolerance in WDM systems, interchannel nonlinearities
need to be compensated. Fig. 5 shows a typical performance curve of SPMC only and with interchannel NL
compensation.
Digital coherent systems provide a means of reconstructing the full electric field of the propagating signal.
Suppose N WDM channels copropagate from transmitter
to receiver. It is possible to use parallel receivers driven by
a bank of local oscillators to recover equivalent baseband
electric fields in the neighborhood of each LO. If the LO’s
are further phase-locked, the WDM field can be digitally
reconstructed and then digitally backpropagated as if it
1072
Proceedings of the IEEE | Vol. 100, No. 5, May 2012
was a single channel using the techniques outlined in
Section V. This Btotal-field[ NLSE (T-NLSE) algorithm
will simultaneously compensate SPM, XPM, and FWM.
However, the step size will need to be short enough to
follow the fastest variation of the optical field, so the
number of steps required (Fig. 6) can be very large. In
presence of dispersion, however, FWM is often negligible
compared to XPM. It is then possible to compensate XPMonly via a Bcoupled[ NLSE (C-NLSE) approach, which
treats the spectral slices recovered by the parallel receivers
as N independent signals mutually coupled by nonlinearity. Unlike T-NLSE, the C-NLSE approach does not require
phase-locked LO’s. The step size only needs to be shorter
than the walkoff length between furthest channels. If the
per-channel bandwidth is small relative to channel spacing, C-NLSE will be much more efficient than T-NLSE.
The techniques used to simplify SPMC can likewise be
applied for interchannel NLC. For example, a walk-off
factorization method was recently introduced for singlepolarization systems [84] and for polarization multiplexing
[85]. This method incorporates the walkoff effect in the
nonlinear calculation (see Fig. 6) to increase step size and
is analogous to the filtered DBP method for SPMC. For
dispersion-compensated links, it is similarly possible to
measure the total instantaneous power of all the channels
using a slow photodiode, and then derotate each channel
by a phase proportional to the total power. This Bpartial[
XPM compensation (P-XPMC) method, which factorizes
the channel as a lumped nonlinear rotation and lumped
CD, is analogous to an earlier partial SPMC method developed for single-channel transmission [86]. P-XPMC was
recently demonstrated for WDM where all the channels
transmit OFDM [87]. P-XPMC can also be used to improve
system performance in mixed line rate systems where
legacy 10-Gb/s on–off-keying (OOK) channels impart large
nonlinear penalties on coherent channels [88]. Although
P-XPMC is numerically simple, the major drawback is that
it works only for very limited dispersion maps and requires
neighboring channels to have fluctuating power to obtain
appreciable performance improvement.
Ultimately, interchannel NLC represents the last
frontier in DSP tools to increase capacity toward the nonlinear Shannon’s limit. Real-time implementation of interchannel NLC remains a distant possibility at present. Even
if DSP advances were to enable interchannel NLC, the
performance improvement will be severely degraded if
channels are added or dropped mid-link in a terrestrial
network [89]. However, interchannel NLC may find application in point-to-point networks such as transoceanic
submarine links.
V. FORWARD E RROR
CORRECTION CODING
A major limitation for nearly all recent ultrahigh data-rate
transmission experiments is that the uncoded bit error rate
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
(BER) can be as high as 103 even for back-to-back [21],
[59], [63], [65], [90]. This is because in the push toward
ever higher spectral efficiency, larger signal constellations,
more aggressive filtering, and even overlapping carriers
are used, rendering a system more sensitive to impairments such as signal generator imperfections, timing
offsets and frequency response imbalance between I and Q
channels (at both transmitter and receiver), laser phase
noise, and misadjustment error due to adaptive filters
having to track time-varying channels. Back-to-back (BtB)
BER can often be as high as 104 to 103 , necessitating
strong forward error correction codes (FEC) to achieve
BER acceptable for practical systems.
However, even without BtB system imperfections,
nonlinearity precludes using arbitrarily high signal power
to obtain the required BER. Fig. 5 showed that a maximum
achievable Q-factor exists, as well as the need to maximize
data throughput at this optimum launch power. Achieving
capacity requires using large constellations in conjunction
with high-redundancy FEC codes [31]. Thus, FECs with
high coding gains and Bhigh BER threshold[ are one of the
key enabling technologies for next-generation optical communication systems.
The FEC schemes that have been proposed for highspeed optical transmission can be categorized into several
generations. First-generation FECs were based on various
hard-decision binary codes such as BCH and RS codes [29],
[91]–[94]. In particular, the RS(255,239) code with 7%
overhead used in the ITU-T G.975 [95] and ITU-T G.709
[96] standards had a pre-FEC BER threshold around 104 .
The second-generation FECs were based on various code
concatenation schemes, such as the concatenation of two
Reed Solomon (RS) codes [e.g., RSð255; 239Þ þ RSð255;
233Þ], or the concatenation of an RS code with a convolutional code [97]. Recently reported continuously interleaved BCH enhanced FEC codes have BER threshold as
high as 4:5 103 , which is the highest reported to date
for 7% overhead hard-decision (HD) FEC codes [98].
With the development of powerful DSPs, it has become
possible to implement soft-decision (SD) FEC codes, which
have higher coding gain compared to HD-FEC codes, but
have higher decoding complexity. Third-generation softdecision FEC have used turbo-product codes [29], [94] and
LDPC codes, with BER thresholds around 102 [99]–[102].
To further improve system tolerance to fiber channel impairments and to support optical communication data rates
beyond 100-Gb/s per channel, fourth-generation FECs
based on LDPC-coded modulation with various joint
detection/equalization soft-decision decoding schemes
have been proposed [103]–[105]. Codes on graphs, such
as turbo codes and LDPC codes, have thus revolutionized
communications and are becoming standard in many
applications.
LDPC codes were invented by Gallager in the 1960s,
and are linear block codes whose parity check matrix has
low density of 1’s [99]. An iterative LDPC decoder based
on the sum-product algorithm (SPA) has been shown to
achieve performance only 0.0045 dB away from the
Shannon limit [103]. Recently, LDPC codes received much
interest in the context of optical fiber communication
[100]–[102], leading to rapidly improved understanding of
its various aspects, such as the design of good codes with
high coding gains and large girths resulting in very low
error floors ðG 1015 Þ [102]; the design of multidimensional LDPC-coded modulation formats with improved
receiver sensitivity and spectral efficiency [104]–[106];
and complexity reduction algorithms that enable tradesoffs between coding gain performance and decoding
complexity [107].
Future advances in LDPC codes will likely include the
following:
1) using concatenated LDPC codes to obtain even
higher BER threshold, building on the results from
concatenated HD-FECs [108], [109];
2) integrating channel equalization with softdecoding through turbo equalization schemes
[102], [110];
3) design of new codes based on nonbinary alphabets
that are better matched to modulation formats
encountered in digital coherent systems.
Nonbinary LDPC-coded modulation was recently demonstrated to achieve high coding gain and high tolerance
to optical fiber impairments [106]. The parity-check
matrix of nonbinary q-ary LDPC codes can be constructed
by assigning nonzero elements from the Galois (or finite)
field of order q to the 1’s in the corresponding binary LDPC
code. Compared to binary LDPC-coded modulation, nonbinary LDPC-coded modulation has several advantages:
1) reduction in the number of encoders and decoders: For 2m -QAM (where m > 1), m binary
LDPC encoders/decoders can be replaced by a
single 2m -ary encoder/decoder;
2) elimination of the block (de-)interleavers needed
for binary-to-nonbinary and nonbinary-to-binary
alphabet conversion;
3) integration of the a posteriori probability (APP)
demapper and LDPC decoder into a single block,
eliminating the need for iterating extrinsic information between the APP demapper and LDPC
decoder.
Whereas 1) reduces system complexity and algorithmic
complexity, 2) and 3) reduce system latency. In addition, it
has been demonstrated that nonbinary LDPC-coded modulation schemes can provide higher coding gains than its
binary counterparts [106]. Fig. 7 shows the measured BtB
BER results for three different rate-0.8 4-ary LDPC codes,
illustrating BER thresholds as high as 3 102 .
Indeed, with increasingly powerful DSPs, sophisticated
SD-LDPC schemes can be expected to play an important
role in next-generation systems by further increasing BER
threshold, thereby improving system margin and enabling
the transmission of high-spectral-efficiency modulation
Vol. 100, No. 5, May 2012 | Proceedings of the IEEE
1073
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
Fig. 7. Measured BtB BER for three different rate-0.8, (3,15)-regular,
4-ary LDPC codes, namely LDPC(16935,13548), LDPC(34665,27732),
and LDPC(69945,55956) [111].
formats that are otherwise not feasible with existing FEC
codes.
VI . ADVANCED FIBER DESIGNS
As previous sections noted, irrespective of improvements
in modulation format, multiplexing technology, nonlinear
compensation technique, or FEC technology used, channel
capacity has an ultimate limit due to nonlinear interaction
between signal and noise. To keep pace with exponentially
increasing bandwidth demand will also require new fiber
designs.
A. Nonlinearity Reduction
One method of increasing capacity-distance product is
to use ultra-large effective area fiber (ULAF). This is of
particular interest to submarine transmission whereby for
fixed spectral efficiency, system reach is approximately
inversely proportional to fiber effective area. However,
increasing the effective area of the propagating mode also
reduces mode confinement, making the fiber more sensitive to macro-bending loss. Recently, it was proposed to
transmit signals using the fundamental mode of a Bfew
mode fiber[ (FMF). Provided mode coupling between the
fundamental to higher-order mode(s) is low; the system
will suffer little excess loss [112]. The advantages of FMF
are: 1) with larger core area, the fundamental mode in
FMF is better confined than that of single-mode ULAF;
2) the effective area of the fundamental mode in FMF can
be even larger than that of ULAF. An attractive aspect of
FMF is that it requires no new amplifier or transponder
design. FMF merely replaces the SMF: At the beginning of
every span, signal from an SMF is center-launched into the
FMF, while at the end of the span, the FMF is centercoupled back into SMF. Long-haul transmission over >
5000 km using FMF has already been demonstrated [113].
1074
Proceedings of the IEEE | Vol. 100, No. 5, May 2012
B. Space Division Multiplexing
In terrestrial networks, the primary interest is to
increase system capacity. Increasing fiber effective area is
less meaningful. Consider Shannon’s limit on information
capacity: Even without nonlinearity, achievable capacity
scales only logarithmically with power. Thus, using ever
larger constellations to increase data rates is not a costeffective solution. A more power-efficient method of increasing capacity is to transmit information over parallel
channels. An energy consumption analysis for SDM can be
found in [114]. It is notable that previous technological
breakthroughs have included WDM transmission, which is
parallelization in frequency, and coherent receivers, which
enables parallel transmission using all the degrees of freedom available in a fiber, namely, the amplitude and phase
of the two signal polarizations. With the capacity of singlemode fibers nearly fully exhaustedVindeed, 100 Tb/s was
achieved experimentally using C+L band transmission at
11 b/s/Hz [39]Vfuture systems will require new paradigms in transmission, such as space-division multiplexing
(SDM). The simplest SDM method is to use multiple
fibers. This requires parallel transmitters, fibers, amplifiers, and receivers. System complexity will scale approximately linearly with capacity, so cost reduction per bit will
only be achieved by minimizing the cost of inline amplifiers and transponders.
An alternative strategy is to have SDM within a single
strand of fiber. Two such schemes have been proposed.
These are: 1) multicore fiber (MCF), and 2) multimode
fiber (MMF).
C. Multimode Fibers
In mode-division multiplexing (MDM), the spatial
modes of an MMF are used as parallel channels. Due to
bending and fiber perturbations, spatial modes can couple
during propagation. To eliminate crosstalk from mode
coupling, multiple-input–multiple-output (MIMO) signal
processing is required. Naively, an N N MIMO system
will require N times the algorithmic complexity to detect
per mode compared to N noncoupling modes. However,
there are strategies that can reduce complexity. First, conventional 50- and 62.5-m MMF used for short-reach
applications (with > 100 modes) are ill-suited for MDM.
Reducing the total number of modes to only two or three
Bmode groups[Vi.e., using an FMFVwill make mode
coupling more manageable, and MIMO detection numerically tractable. Second, the number of filter taps required
to equalize each input–output mode-pair depends on the
maximum differential modal group delay (DMGD) between
the propagating modes. If an FMF can be designed to have
low DMGD, corresponding to making the MIMO channel
matrix more Bfrequency-flat,[ mode demultiplexing complexity can be reduced. MDM transmission using FMF has
already been reported [32]–[34]. However, extending the
reach of MDM systems to multiple fiber spans will require
the development of multimode fiber amplifiers.
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
D. Multicore Fibers
An alternative to using multimode fibers is to use a
multicore fiber [115], where a strand of fiber contains multiple single-mode cores. Mode coupling between the cores
can be reduced by increasing their mutual spacing, using
heterogeneous cores, as well as mode-confinement techniques such as refractive index Btrenching.[ A disadvantage
of MCF is that outer cores may have higher loss due to
coupling into cladding modes. MCF generally requires
using larger cladding diameter, and they are sensitive to
bending. In addition, the information density per unit area
in MCF is lower than in FMF, where the parallel channels
are physically confined in the same core. However, if some
mode coupling can be tolerated, it is possible to reduce the
pitch spacing and use homogeneous cores, using MIMO to
compensate for the resulting crosstalk. In this sense, the
MCF becomes a special case of a FMF [116]. High-coupling
MCF may have an advantage over a FMF if the DMGD
between the cores can be smaller than the DMGD between
the different propagating modes in FMF. However, inline
amplification in MCFs remains a challenge. It is possible to
use a fanout to couple optical signals from the MCF cores to
multiple single-mode fibers, and then amplify each SMF
with standard erbium-doped fiber amplifiers (EDFAs) or
Raman amplifiers, followed by another fanout to couple the
signals back into the MCF. This strategy, however, has no
cost benefit compared to SDM using multiple fibers. On the
other hand, unless the amplifier pump can be efficiently
coupled only to the cores of an MCF, any pump power
leaked into the cladding will reduce efficiency. SDM transmission using MCF has been reported in [117]–[119]. In
particular, the highest capacity achieved to date on a single
strand of fiber was in a seven-core MCF [119].
Ultimately, both FMF and MCF face numerous technological challenges. The practicability of these schemes
relies on future improvements in large-scale photonic integration and (especially) improvements in digital signal
processors. Only by integrating SDM transmitters, inline
amplifiers, and receivers to process multiple cores or
modes simultaneously within a single device, and making
the required DSP feasible, will the cost of FMF and MCF
transmission grow less than linearly with capacity, and
thus become competitive than deploying multiple singlemode fibers.
REFERENCES
[1] R. T. Tkach, BScaling optical
communications for the next decade and
beyond,[ Bell Labs Tech. J., vol. 14, no. 4,
pp. 3–9, Winter 2010.
[2] A. H. Gnauck, R. W. Tkach, A. R. Chraplyvy,
and T. Li, BHigh-capacity optical
transmission systems,[ J. Lightw. Technol.,
vol. 26, no. 9, pp. 1032–1045, May 2008.
[3] C. A. Brackett, BDense wavelength
division multiplexing networks: Principles
and applications,[ IEEE J. Sel. Areas in
Commun., vol. 8, no. 6, pp. 948–964,
Aug. 1990.
VI I. CONCLUSION
The convergence of coherent receiver, ultrahigh-speed
digital signal processors, tunable lasers, and spectrum variable wavelength selective switches has enabled the possibility of software-defined optical transponders. Future
optical networks will no longer be constrained by a fixed
wavelength grid, but will have the ability to dynamically
allocate network resources on a heterogeneous basis to different users in response to instantaneous traffic demands.
One of the key enabling technologies that could be
deployed in the near-term is the optical Bsuperchannel,[
which uses optical signal processing techniques to create
orthogonal optical subcarriers that can be modulated and
detected via parallel transmitter and receiver architectures.
As a result, per-channel data rates will no longer be constrained by the bandwidth of electronics, allowing continuous data modulation over vastly increased bandwidths.
Digital transmitters will also enable advanced spectral
shaping, reducing the need for guard bands between
neighboring channels, thus increasing bandwidth utilization and spectral efficiency. With precise DSP control over
the transmitted waveform, signal constellations generated
by digital transmitters will have higher fidelity, enabling
larger constellation sizes and extending the system reach
of high spectral efficiency modulation formats.
Advanced forward error correction based on lowdensity parity check codes and reduced-complexity decoding algorithms will enable next-generation systems to have
higher coding gains and higher bit-error rate thresholds,
resulting in system performance closer to channel capacity.
Increasingly powerful DSPs may also enable nonlinear
compensation. Already, reduced-complexity intrachannel
nonlinearity compensation algorithms have algorithmic
complexities only a few times higher than chromatic dispersion compensation only. In the more distant future,
interchannel nonlinearity may also be possible for pointto-point systems to increased system reach.
Ultimately, continued exponential growth in bandwidth demand will require new transmission paradigms.
When the capacity of single-core single-mode fibers becomes exhausted, space-division multiplexing will be required. Multicore and multimode fibers are a technology
for the far future, with numerous technological challenges
waiting to be solved. h
[4] K. Emura and M. Shikada, BHigh capacity
optical networking technologies,[ presented
at the Conf. Lasers Eletro-Opt. (CLEO),
Seoul, Korea, 1999, Paper TutC.
[5] E. Desurvire, D. Bayart, B. Desthieux, and
S. Bigo, Erbium-Doped Fiber Amplifiers:
Device and System development.
New York: Wiley, 2002.
[6] R. A. Linke and A. H. Gnauck,
BHigh-capacity coherent lightwave
systems,[ J. Lightw. Technol., vol. 6,
no. 11, pp. 1750–1769, Nov. 1988.
[7] E. Ip, A. P. T. Lau, D. J. F. Barros, and
J. M. Kahn, BCoherent detection in optical
fiber systems,[ Opt. Exp., vol. 16, no. 2,
pp. 753–791, Jan. 2008.
[8] P. J. Winzer, BModulation and multiplexing
in optical communication systems,[ IEEE
LEOS Newslett., vol. 23, no. 1, pp. 4–10, 2009.
[9] T. J. Xia, G. Wellbrock, B. Basch, S. Kotrla,
W. Lee, T. Tajima, K. Fukuchi, M. Cvijetic,
J. Sugg, Y. Ma, B. Turner, C. Cole, and
C. Urricariet, BEnd-to-end native IP Data
100G single carrier real time DSP coherent
detection transport over 1520-km field
deployed fiber,[ presented at the Opt. Fiber
Commun. Conf. (OFC), San Diego, CA,
2010, Paper PDPD4.
Vol. 100, No. 5, May 2012 | Proceedings of the IEEE
1075
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
[10] M. Birk, P. Gerard, R. Curto, L. Nelson,
X. Zhou, P. Magill, T. J. Schmidt,
C. Malouin, B. Zhang, E. Ibragimov,
S. Khatana, M. Glavanovic, R. Lofland,
R. Marcoccia, G. Nicholl, M. Nowell, and
F. Forghieri, BReal-time single-carrier
Coherent 100 Gb/s PM-QPSK field
trial,[ J. Lightw. Technol., vol. 29, no. 4,
pp. 417–425, Feb. 2011.
[11] OIF, 100G Ultra Long Haul DWDM
Framework Document, Jun. 30, 2009.
[12] M. Nakazawa, S. Okamoto, T. Omiya,
K. Kasai, and M. Yoshida, B256 QAM
(64 Gbit/s) coherent optical transmission
over 160 km with an optical bandwidth
of 5.4 GHz,[ presented at the Opt. Fiber
Commun. Conf. (OFC), San Diego, CA,
2010, Paper OMJ5.
[13] S. Okamoto, K. Toyoda, T. Omiya, K. Kasai,
M. Yoshida, and M. Nazazawa, B512 QAM
(54 Gbit/s) coherent optical transmission
over 150 km with an optical bandwidth of
4.1 GHz,[ presented at the Eur. Conf. Opt.
Commun. (ECOC), Torino, Italy, 2010,
Paper PD2.3.
[14] M.-F. Huang, D. Qian, and E. Ip,
B50.53-Gb/s PDM-1024QAM-OFDM
transmission using pilot-based phase noise
mitigation,[ presented at the Opto-Electron.
Commun. Conf. (OECC), Kaohsiung,
Taiwan, 2011, Paper PDP.
[15] M. Jinno, H. Takara, and B. Kozicki,
BDynamic optical mesh networks: Drivers,
challenges and solutions for the future,[
presented at the Eur. Conf. Opt. Commun.
(ECOC), Vienna, Austria, 2009, Paper 7.7.4.
[16] A. N. Patel, P. N. Ji, J. P. Jue, and T. Wang,
BRouting, wavelength assignment, and
spectrum allocation in transparent
flexible optical WDM (FWDM) networks,[
presented at the Conf. Photon. Switching,
Monterey, CA, 2010, Paper PDPWG2.
[17] K. Christodoulopoulos, I. Tomkos, and
E. Vargarigos, BSpectrally/bitrate flexible
optical network planning,[ presented at the
Eur. Conf. Opt. Commun. (ECOC), Torino,
Italy, 2010, Paper We.8.D.3.
[18] S. Gringeri, B. Basch, V. Shukla, R. Egorov,
and T. J. Xia, BFlexible architectures for
optical transport nodes and networks,[ IEEE
Commun. Mag., vol. 48, no. 7, pp. 40–50,
Jul. 2010.
[19] B. Zhu, X. Liu, S. Chandrasekhar,
D. W. Peckham, and R. Lingle,
BUltra-long-haul transmission of 1.2-Tb/s
multicarrier no-guard-interval CO-OFDM
superchannel using ultra-large-area fiber,[
Photon. Technol. Lett., vol. 22, no. 11,
pp. 826–828, Jun. 2010.
[20] X. Yi, N. K. Fontaine, R. P. Scott, and
S. J. Ben Yoo, BTb/s coherent optical OFDM
systems enabled by optical frequency
combs,[ J. Lightwave Technol., vol. 28, no. 14,
pp. 2054–2061, Jul. 2010.
[21] T. Xia, G. Wellbrock, Y. Huang, E. Ip,
M. Huang, Y. Shao, T. Wang, Y. Aono,
T. Tajima, S. Murakami, and M. Cvijetic,
BField experiment with mixed line-rate
transmission (112-Gb/s, 450-Gb/s, and
1.15-Tb/s) over 3,560 km of installed
fiber using filterless coherent receiver
and EDFAs only,[ presented at the Opt.
Fiber Commun. Conf. (OFC), Los Angeles,
CA, 2011, Paper PDPA3.
[22] D. Hillerkuss, T. Schellinger, R. Schmogrow,
M. Winter, T. Vallatis, R. Bonk,
A. Marculescu, J. Li, M. Dreschmann,
J. Meyer, S. Ben Ezra, N. Narkiss,
B. Nebendahl, F. Parmigiani, P. Petropoulos,
B. Resan, K. Weingarten, T. Ellermeyer,
1076
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
J. Lutz, M. Möller, M. Huebner, J. Becker,
C. Koos, W. Freude, and J. Leuthold,
BSingle-source optical OFDM transmitter
and optical FFT receiver demonstrated at
line rates of 5.4 and 10.8 Tbit/s,[ presented
at the Opt. Fiber Commun. Conf. (OFC),
San Diego, CA, 2010, Paper PDPC1.
T. J. Xia, BOptical channel capacityVFrom
Mb/s to Tb/s and beyond,[ Opt. Fiber
Technol., Jun. 2011, Invited Paper.
M. Islam, BRaman amplifiers for
telecommunications,[ IEEE J. Sel.
Top. Quantum Electron., vol. 8, no. 3,
pp. 548–559, May/Jun. 2002.
K. Nagayama, M. Kakui, M. Matsui,
T. Saitoh, and Y. Chigusa, BUltra-low-loss
(0.1484 dB/km) pure silica core fibre
and extension of transmission distance,[
Electron. Lett., vol. 38, no. 20, pp. 1168–1169,
Sep. 2002.
J.-X. Cai, Y. Cai, C. Davidson, D. Foursa,
A. Lucero, O. Sinkin, W. Patterson,
A. Pilipetskii, G. Mohs, and N. Bergano,
BTransmission of 96 100G pre-filtered
PDM-RZ-QPSK channels with 300%
spectral efficiency over 10 608 km and
400% spectral efficiency over 4368 km,[
presented at the Opt. Fiber Commun. Conf.
(OFC), San Diego, CA, 2010, PDPB10.
X. Li, X. Chen, G. Goldfarb, E. Mateo,
I. Kim, F. Yaman, and G. Li, BElectronic
post-compensation of WDM transmission
impairments using coherent detection and
digital signal processing,[ Opt. Exp., vol. 16,
no. 2, pp. 881–888, Jan. 2008.
E. Mateo, L. Zhu, and G. Li, BImpact of XPM
and FWM on the digital implementation
of impairment compensation for WDM
transmission using backward propagation,[
Opt. Exp., vol. 16, no. 20, pp. 16 124–16 137,
Sep. 2008.
T. Mizuochi, BRecent progress in forward
error correction and its interplay with
transmission impairments,[ IEEE J. Sel.
Topics Quantum Electron., vol. 12, no. 4,
pp. 544–554, Jul.–Aug. 2006.
P. P. Mitra and J. B. Stark, BNonlinear limits
to the information capacity of optical fiber
communications,[ Nature, vol. 411, no. 6841,
pp. 1027–1030, Jun. 2001.
R.-J. Essiambre, G. Kramer, P. J. Winzer,
G. J. Foschini, and B. Goebel, BCapacity
limits of optical fiber networks,[ J. Lightw.
Technol., vol. 28, no. 4, pp. 662–701,
Feb. 2010.
A. Amin, A. Li, S. Chen, X. Chen,
G. Gao, and W. Shieh, BDual-LP11 mode
4 4 MIMO-OFDM transmission over a
two-mode fiber,[ Opt. Exp., vol. 19, no. 17,
pp. 16672–16679, Aug. 2011.
C. Kobele, M. Salsi, D. Sperti, P. Tran,
P. Brindel, H. Mardoyan, S. Bigo, A. Boutin,
F. Verluise, P. Sillard, M. Astruc, L. Provost,
F. Cerou, and G. Charlet, BTwo mode
transmission 2 100-Gb/s over 40-km long
prototype few-mode fiber using LCOS-based
programmable mode multiplexer and
demultiplexer,[ Opt. Exp., vol. 19, no. 17,
pp. 16593–16600, Aug. 2011.
S. Randel, R. Ryf, A. Sierra, P. J. Winzer,
A. H. Ghanuck, C. A. Bolle, R.-J. Essiambre,
D. W. Peckham, A. McCurdy, and R. Lingle,
B6 56-Gb/s mode-division multiplexed
transmission over 33-km few-mode fiber
enabled by 6 6 MIMO equalization,[
Opt. Exp., vol. 19, no. 17, pp. 16 601–16 611,
Aug. 2011.
A. H. Gnauck, G. Charlet, P. Tran,
P. Winzer, C. Doerr, J. Centanni,
E. Burrows, T. Kawanishi, T. Sakamoto, and
Proceedings of the IEEE | Vol. 100, No. 5, May 2012
[36]
[37]
[38]
[39]
[40]
[41]
[42]
[43]
[44]
[45]
[46]
[47]
[48]
[49]
K. Higuma, B25.6-Tb/s WDM transmission
of polarization-multiplexed RZ-DQPSK
signals,[ J. Lightw. Technol., vol. 26, no. 1,
pp. 79–84, Jan. 2008.
X. Zhou, J. Yu, M.-F. Huang, Y. Shao,
T. Wang, P. Magill, M. Cvijetic, L. Nelson,
M. Birk, G. Zhang, S. Ten, H. B. Matthew,
and S. K. Mishra, BTransmission of 32-Tb/s
capacity over 580 km using RZ-shaped
PDM-8QAM modulation format and
cascaded multimodulus blind equalization
algorithm,[ J. Lightw. Technol., vol. 28, no. 4,
pp. 456–465, Feb. 2010.
X. Zhou, J. Yu, M.-F. Huang, Y. Shao,
T. Wang, L. Nelson, P. Magill, M. Birk,
P. I. Borel, D. W. Peckham, R. Lingle, and
B. Zhu, B64-Tb/s, 8 b/s/Hz, PDM-36QAM
transmission over 320 km using both
pre- and post-transmission digital signal
processing,[ J. Lightw. Technol., vol. 29,
no. 4, pp. 571–577, Feb. 2011.
A. Sano, H. Masuda, T. Kobayashi,
M. Fujiwara, K. Horikoshi, E. Yoshida,
Y. Miyamoto, M. Matsui, M. Mizoguchi,
H. Yamazaki, Y. Sakamaki, and H. Ishii,
B69.1-Tb/s (432 171-Gb/s) C- and extended
L-band transmission over 240 km using
PDM-16-QAM modulation and digital
coherent detection,[ presented at the Opt.
Fiber Commun. Conf. (OFC), San Diego,
CA, 2010, PDPB7.
D. Qian, M.-F. Huang, E. Ip, Y.-K. Huang,
Y. Shao, J. Hu, and T. Wang, B101.7-Tb/s
(370 294-Gb/s) PDM-128QAM-OFDM
transmission over 3 55-km SSMF
using pilot-based phase noise mitigation,[
presented at the Opt. Fiber Commun.
Conf. (OFC), Los Angeles, CA, 2011,
Paper PDPB5.
K. Emura, BMulti-Terabit/s DWDM;
technologies and perspectives,[ presented
at the Opt. Fiber Commun. Conf. (OFC),
Los Anaheim, CA, 2001, Paper ML1-1.
P. J. Winzer, BBeyond 100G ethernet,[ IEEE
Commun. Mag., vol. 48, no. 7, pp. 26–30,
Jul. 2010.
K. Roberts, D. Beckett, D. Boertjes,
J. Berthold, and C. Laperle, B100G and
beyond with digital coherent signal
processing,[ IEEE Commun. Mag., vol. 48,
no. 7, pp. 63–69, Jul. 2010.
J. Yu and X. Zhou, BUltra-high capacity
DWDM transmission system for 100G and
beyond,[ IEEE Commun. Mag., vol. 48, no. 3,
pp. S56–S64, Mar. 2010.
M. Camera, B.-E. Olsson, and G. Bruno,
BBeyond 100 G: System implications towards
400 G and 1 T,[ presented at the Eur.
Conf. Opt. Commun. (ECOC), Torino, Italy,
2010, Paper Th.10.G.1.
M. W. Chbat and S. Spaelter, B From 100 G
to 1000 G: Is there a straight road ahead,[
presented at the Eur. Conf. Opt. Commun.
(ECOC), Torino, Italy, 2010, Paper Th.9.G.2.
D. W. Boertjes, BAgile subsystems for
coherent systems beyond 100 G,[ presented
at the Opt. Fiber Commun. Conf. (OFC),
Los Angeles, CA, 2011, Paper OThE1.
E. Roza, BSystems-on-chip: What are the
limits,[ Elect. Commun. Eng. J., pp. 249–255,
Dec. vol. 13, no. 6, 2001.
D.-S. Ly-Gagnon, S. Tsukamoto, K. Katoh,
and K. Kikuchi, BCoherent detection of
optical quadrature phase-shift keying
signals with carrier phase estimation,[
J. Lightw. Technol., vol. 24, no. 1, pp. 12–21,
Jan. 2006.
S. J. Savory, BDigital filters for coherent
optical receivers,[ Opt. Exp., vol. 16, no. 2,
pp. 804–817, Jan. 2008.
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
[50] T. Pfau, S. Hoffmann, O. Adamczyk,
R. Peveling, V. Hereath, M. Porrmann, and
R. Noé, BCoherent optical communication:
Towards realtime systems at 40 Gbit/s
and beyond,[ Opt. Exp., vol. 16, no. 2,
pp. 866–872, Jan. 2008.
[51] X. Liu, S. Chandrasekhar, and A. Leven,
BDigital self-coherent detection,[ Opt. Exp.,
vol. 16, no. 2, pp. 792–803, Jan. 2008.
[52] C. R. S. Fludger, T. Duthel, D. Borne,
C. Schulien, E.-D. Schmidt, Wuth,
J. Geyer, E. De Man, G.-D. Khoe, and
H. de Waardt, BCoherent equalization and
POLMUX-RZ-DQPSK for robust 00-GE
transmission,[ J. Lightw. Technol., vol. 26,
no. 1, pp. 64–72, Jan. 2008.
[53] G. P. Agrawal, Fiber-Optic Communication
Systems, 3rd ed. New York: Wiley, 2002.
[54] N. Kikuchi, K. Sekine, and S. Sasaki,
BAnalysis of cross-phase modulation (XPM)
effect on WDM transmission performance,[
Electron. Lett., vol. 33, no. 8, pp. 653–654,
Apr. 1997.
[55] H. Takahashi, S. Suzuki, K. Kato, and
I. Nishi, BArrayed waveguide grating for
wavelength division multi/demultiplexer
with nanometer resolution,[ Electron. Lett.,
vol. 26, no. 2, pp. 87–88, Jan. 1990.
[56] G. Baxter, S. Frisken, D. Abakoumov,
H. Zhou, I. Clarke, A. Bartos, and S. Poole,
BHighly programmable wavelength selective
switch based on liquid crystal on silicon
switching elements,[ presented at the Opt.
Fiber Commun. Conf. (OFC), Anaheim, CA,
2006, Paper OTuF2.
[57] L. E. Nelson, S. L. Woodward, S. Foo,
M. Moye, D. Beckett, M. O’Sullivan, and
P. Magill, BDetection of a single 40 Gb/s
polarization-multiplexed QPSK channel
with a real-time intradyne receiver in the
presence of multiple WDM channels,[
J. Lightw. Technol., vol. 28, no. 20,
pp. 2933–2943, Oct. 2010.
[58] J. Armstrong, BOFDM for optical
communications,[ J. Lightw. Technol.,
vol. 27, no. 3, pp. 189–204, Feb. 2009.
[59] X. Liu, S. Chandrasekhar, B. Zhu,
P. J. Winzer, A. H. Gnauck, and
D. W. Peckham, B448-Gb/s
reduced-guard-interval CO-OFDM
transmission over 2000 km of
ultra-large-area fiber and five 80-GHz-grid
ROADs,[ J. Lightw. Technol., vol. 29, no. 4,
pp. 483–490, Feb. 2011.
[60] R. Schmogrow, M. Winter, M. Meyer,
D. Hillerkuss, B. Nebendahl, J. Meyer,
M. Dreschmann, M. Huebner, J. Becker,
C. Koos, W. Freude, and J. Leuthold,
BReal-time Nyquist pulse modulation
transmitter generating rectangular shaped
spectra of 112 Gbit/s 16QAM signals,[
presented at the Signal Process. Photon.
Commun. (SPPCom), Toronto, Canada,
2011, Paper SPMA5.
[61] A. J. Lowery, BFiber nonlinearity pre-and
post-compensation for long-haul optical links
using OFDM,[ Opt. Exp., vol. 20, no. 15,
pp. 12965–12970, Oct. 2007.
[62] S. Zhang, L. Xu, P. Y. Kam, C. Yu, and
T. Wang, BPerformance investigation of
joint SPM compensation in a long-haul
coherent dual-polarization QPSK
system,[ presented at the Eur. Conf.
Opt. Commun. (ECOC), Torino, Italy,
2010, Paper P3.15.
[63] P. J. Winzer, A. H. Gnauck,
S. Chandrasekhar, S. Draving, J. Evangelista,
and B. Zhu, BGeneration and 1,200-km
transmission of 448-Gb/s ETDM
56-Gbaud PDM 16-QAM using a single
[64]
[65]
[66]
[67]
[68]
[69]
[70]
[71]
[72]
[73]
[74]
[75]
[76]
I/Q modulator,[ presented at the Eur. Conf.
Opt. Commun. (ECOC), Torino, Italy, 2010,
Paper PD2.2.
S. Kawanishi, H. Takara, K. Uchiyama,
I. Shake, and K. Mori, B3-Tbit/s
(160-Gbit/s 19 channel) optical TDM
and WDM transmission experiment,[
Electron. Lett., vol. 35, no. 10, pp. 826–827,
May 1999.
Y.-K. Huang, E. Ip, M.-F. Huang,
B. Zhu, P. N. Ji, Y. Shao, D. W. Peckham,
R. Lingle, Y. Aono, T. Tajima, and T. Wang,
B10 456-Gb/s DP-16QAM transmission
over 8 100 km of ULAF using coherent
detection with a 30-GHz analog-to-digital
converter,[ presented at the Opto-Electron.
Commun. Conf. (OECC), Sapporo, Japan,
2010, Paper PDP3.
T. Yamamoto, E. Yoshida, K. R. Tamura,
K. Yonenaga, and M. Nakazawa, B640-Gbit/s
optical TDM transmission over 92 km
through a dispersion-managed fiber
consisting of single-mode fiber and reverse
dispersion fiber,[ IEEE Photon. Technol. Lett.,
vol. 12, no. 3, pp. 353–355, Mar. 2000.
T. Richter, E. Palushani,
C. Schmidt-Langhorst, M. Nölle, R. Ludwig,
and C. Schubert, BSingle wavelength
channel 10.2 Tb/s TDM-data capacity using
16-QAM and coherent detection,[ presented
at the Opt. Fiber Commun. Conf. (OFC),
Los Angeles, CA, 2011, Paper PDPA9.
E. Ip and J. M. Kahn, BFiber impairment
compensation using coherent detection and
digital signal processing,[ J. Lightw. Technol.,
vol. 28, no. 4, pp. 502–519, Feb. 2010.
J. Yu, Z. Dong, X. Xiao, Y. Xia, S. Shi, C. Ge,
W. Zhou, N. Chi, and Y. Shao, BGeneration,
transmission and coherent detection of
11.2 Tb/s (112 100Gb/s) single source
optical OFDM superchannel,[ presented at
the Opt. Fiber Commun. Conf. (OFC),
Los Angeles, CA, 2011, Paper PDPA6.
Z. Wang, Y.-K. Wang, E. Ip, P. R. Prucnal,
and T. Wang, BPerformance investigation
of polarization-multiplexed 16-QAM
using all-optical OFDM transmission
and digital coherent detection,[ presented at
the Opt. Fiber Commun. Conf. (OFC),
Los Angeles, CA, 2011, Paper OMS5.
Y.-K. Huang, E. Ip, M.-F. Huang, Z. Wang,
Y. Shao, Y. Aono, and T. Tajima, B1.5 Tb/s
super-channel transmission over 1200 SSMF
using all-optical OFDM and digital coherent
receiver,[ presented at the Asia Commun.
Photon. Conf. (ACP), Shanghai, China, 2010,
Paper PD1.
S. L. Jansen, I. Morita, T. C. W. Schenk, and
H. Tanaka, B121.9-Gb/s PDM-OFDM
Transmission with 2 b/s/Hz Spectral
Efficiency over 1,000 km of SSMF,[ J.
Lightw. Technol., vol. 27, no. 3, pp. 177–188,
Feb. 2009.
A. Tolmachev and M. Nazarathy,
BLow-complexity multi-band polyphase
filter bank for reduced-guard-interval
coherent optical OFDM,[ presented at the
Signal Process. Photon. Commun.
(SPPCom), Toronto, Canada, 2011, Paper
SPMB3.
W. Shieh, X. Yi, Y. Ma, and Q. Yang,
BCoherent optical OFDM: Has its
time come?[ J. Opt. Netw., vol. 7, no. 3,
pp. 234–255, 2008.
A. Yariv, D. Fekete, and D. M. Pepper,
BCompensation for channel dispersion by
nonlinear optical phase conjugation,[ Opt.
Lett., vol. 4, no. 2, pp. 52–54, Feb. 1979.
S. Watanabe and M. Shirasaki, BExact
compensation for both chromatic dispersion
[77]
[78]
[79]
[80]
[81]
[82]
[83]
[84]
[85]
[86]
[87]
[88]
[89]
and Kerr effect in a transmission fiber using
optical phase conjugation,[ J. Lightwave
Technol., vol. 14, no. 3, pp. 243–248,
Mar. 1996.
T. Kobayashi, A. Sano, H. Masuda,
K. Ishihara, E. Yoshida, Y. Miyamoto,
H. Yamazaki, and T. Yamada, B160-Gb/s
polarization-multiplexed 16-QAM long-haul
transmission over 3,123 km using digital
coherent receiver with digital PLL based
frequency offset compensator,[ presented at
the Opt. Fiber Commun. Conf. (OFC)
San Diego, CA, 2010, Paper OTuD1.
R. Weidenfeld, M. Nazarathy, R. Noe, and
I. Shpantzer, BVolterra nonlinear
compensation of 100G coherent OFDM
with baud-rate ADC, tolerable complexity
and low intra-channel FWM/XPM error
propagation,[ presented at the Opt. Fiber
Commun. Conf. (OFC), San Diego, CA,
2010, Paper OTuE3.
E. Ip, N. Bai, and T. Wang, BComplexity
versus performance tradeoff in fiber
nonlinearity compensation using
frequency-shaped, multi-subband
backpropagation,[ presented at the Opt.
Fiber Commun. Conf. (OFC), Los Angeles,
CA, 2011, Paper OThF4.
J. K. Fischer, C.-A. Bunge, and K. Petermann,
BEquivalent single-span model for
dispersion-managed fiber-optics
transmission systems,[ J. Lightw. Technol.,
vol. 27, no. 16, pp. 3425–3432, Aug. 2009.
L. Zhu and G. Li, BFolded digital backward
propagation for dispersion managed
fiber-optic transmission,[ Opt. Exp.,
vol. 19, no. 7, pp. 5953–5959, Mar. 2011.
L. Li, Z. Tao, L. Dou, W. Yan, S. Oda,
T. Tanimura, T. Hoshida, and
J. C. Rasmussen, BImplementation efficient
nonlinear equalizer based on correlated
digital backpropagation,[ presented at the
Opt. Fiber Commun. Conf. (OFC),
Los Angeles, CA, 2011, Paper OWW3.
L. B. Du and A. J. Lowery, BImproved single
channel backpropagation for intra-channel
fiber nonlinearity compensation in long-haul
optical communication systems,[ Opt. Exp.,
vol. 18, no. 16, pp. 17075–17088, Aug. 2010.
E. F. Mateo, F. Yaman, and G. Li, BEfficient
compensation of inter-channel nonlinear
effects via digital backward propagation
in WDM optical transmission,[ Opt. Exp.,
vol. 18, no. 14, pp. 15 144–15 154, Jul. 2010.
E. F. Mateo, X. Zhou, and G. Li, BImproved
digital backward propagation for the
compensation of inter-channel nonlinear
effects in polarization-multiplexed
WDM systems,[ Opt. Exp., vol. 19, no. 2,
pp. 570–583, Jan. 2011.
K.-P. Ho and J. M. Kahn, BElectronic
compensation technique to mitigate
nonlinear phase noise,[ J. Lightw. Technol.,
vol. 22, no. 3, pp. 779–783, Mar. 2004.
L. Du and A. Lowery, BExperimental
demonstration of XPM compensation for
CO-OFDM systems with periodic dispersion
maps,[ presented at the Opt. Fiber Commun.
Conf. (OFC), Los Angeles, CA, 2011,
Paper OWW2.
G. Charlet, BThe impact and mitigation
of nonlinear effects in coherent optical
transmission,[ presented at the Opt. Fiber
Commun. Conf. (OFC), San Diego, CA,
2009, Paper NThB4.
E. Ip, BNonlinear compensation
using backpropagation for
polarization-multiplexed transmission,[
J. Lightw. Technol., vol. 28, no. 6,
pp. 939–951, Mar. 2010.
Vol. 100, No. 5, May 2012 | Proceedings of the IEEE
1077
Ip et al.: 100G and Beyond Transmission Technologies for Evolving Optical Networks
[90] X. Liu, S. Chandrasekhar, and P. J. Winzer,
B728-Gb/s CO-OFDM transmission over
800-km ULAF using 64-QAM subcarrier
modulation and single-step coherent
detection with 4 80-Gsamples/s ADCs,[
presented at the Asia Commun. Photon.
Conf. (ACP), Shanghai, China, 2010,
Paper PD2.
[91] Telecommunication Standardization Sector:
Forward Error Correction for Submarine
Systems, ITU, Geneva, Switzerland,
Rec. G.975, 1996.
[92] Telecommunication Standardization Sector:
Forward Error Correction for High Bit
Rate DWDM Submarine Systems, ITU,
Rec. G. 975.1, Feb. 2004.
[93] T. Mizuochi, Y. Miyata, T. Kobayashi,
K. Ouchi, K. Kuno, K. Kubo, K. Shimizu,
H. Tagami, H. Yoshida, H. Fujita, M. Akita,
and K. Motoshima, BForward error
correction based on block turbo code
with 3-bit soft decision for 10 Gb/s optical
communication systems,[ IEEE J. Sel.
Top. Quantum Electron., vol. 10, no. 2,
pp. 376–386, Mar./Apr. 2004.
[94] O. A. Sab and V. Lemarie, BBlock turbo
code performances for long-haul DWDM
optical transmission systems,[ presented at
the Opt. Fiber Commun. Conf. (OFC),
Baltimore, MD, 2000, Paper ThS5-1.
[95] Forward Error Correction for Submarine
Systems, International Telecommunication
Union, Telecommunication Standardization
Sector, G. 975, Oct. 2000.
[96] Interfaces for the Optical Transport Network,
International Telecommunication Union,
Telecommunication Standardization Sector,
G. 709, Mar. 2003.
[97] K. Seki, K. Mikami, M. Baba, A. Katayama,
H. Tanaka, Y. Hara, M. Kobayashi, and
N. Okada, BSingle-chip FEC codec LSI using
iterative CSOC decoder for 10Gb/s long-haul
optical transmission systems,[ presented at
the Custom Integr. Circuits Conf. (CICC),
Orlando, FL, 2002, Paper 9-3-1.
[98] F. Chang, K. Onohara, and T. Mizuochi,
BForward error correction for 100 G
transport networks,[ IEEE Commun. Mag.,
vol. 48, no. 3, pp. S48–S55, Mar. 2010.
[99] R. G. Gallager, Low Density Parity Check
Codes. Cambridge, MA: MIT Press, 1963.
[100] B. Vasic and I. B. Djordjevic, BLow-density
parity check codes for long haul optical
communications systems,[ IEEE Photon.
[101]
[102]
[103]
[104]
[105]
[106]
[107]
[108]
[109]
[110]
Technol. Lett., vol. 14, no. 8, pp. 1208–1210,
Aug. 2002.
I. B. Djordjevic, L. Xu, T. Wang, and
M. Cvijetic, BGLDPC codes with
Reed-Muller component codes suitable for
optical communications,[ IEEE Commun.
Lett., vol. 12, no. 9, pp. 684–686, Sep. 2008.
I. B. Djordjevic, M. Arabaci, and L. Minkov,
BNext generation FEC for high-capacity
communication in optical transport
networks,[ J. Lightw. Technol., vol. 27, no. 16,
pp. 3518–3530, Aug. 2009.
S.-Y. Chung, G. D. Forney, Jr.,
T. J. Richardson, and R. Urbanke, BOn the
design of low-density parity-check codes
within 0.0045 dB of the Shannon Limit,[
IEEE Commun. Lett., vol. 5, no. 2, pp. 58–60,
Feb. 2001.
I. B. Djordjevic, M. Cvijetic, L. Xu, and
T. Wang, BUsing LDPC-coded modulation
and coherent detection for ultra high-speed
optical transmission,[ J. Lightw. Technol.,
vol. 25, no. 11, pp. 3619–3625, Nov. 2007.
H. G. Batshon, I. B. Djordjevic, L. Xu, and
T. Wang, BModified hybrid subcarrier/
amplitude/phase/polarization LDPC-coded
modulation for 400 Gb/s optical
transmission and beyond,[ Opt. Exp.,
vol. 18, no. 13, pp. 14108–14113, Jun. 2010.
M. Arabaci, I. B. Djordjevic, L. Xu, and
T. Wang, BFour-dimensional nonbinary
LDPC-coded modulation schemes for ultra
high-speed optical fiber communication,[
IEEE Photon. Technol. Lett., vol. 23, no. 18,
pp. 1280–1282, Sep. 2011.
I. B. Djordjevic, L. Xu, and T. Wang, BOn the
reduced- complexity of LDPC decodes for
ultra-high-speed optical transmission,[ Opt.
Exp., vol. 18, no. 22, pp. 23 371–23 377,
Oct. 2010.
N. Kamiya and S. Shiori, BConcatenated
QC-LDPC and SPC codes for 100Gbps ultra
long-haul optical transmission systems,[
presented at the Opt. Fiber Commun. Conf.
(OFC), San Diego, CA, 2010, Paper OThL2.
K. Onohara, Y. Miyata, T. Sugihara, K. Kubo,
H. Yoshida, and T. Mizuochi, BSoft decision
FEC for 100G transport systems,[ presented
at the Opt. Fiber Commun. Conf. (OFC),
San Diego, CA, 2010, Paper OThL1.
I. B. Djordjevic, L. L. Minkov, L. Xu, and
T. Wang, BSuppression of fiber nonlinearities
and PMD in coded-modulation schemes
with coherent detection by using turbo
[111]
[112]
[113]
[114]
[115]
[116]
[117]
[118]
[119]
equalization,[ J. Opt. Commun. Netw., vol. 1,
no. 6, pp. 555–564, Nov. 2009.
S. Zhang, M. Arabaci, F. Yaman,
I. B. Djordjevic, L. Xu, T. Wang, Y. Inada,
T. Ogata, and Y. Aoki, BExperimental
demonstration of non-binary LDPC coded
modulation for ultra-long-haul transmission
systems,[ Opt. Exp., vol. 19, no. 20,
pp. 19 042–19 049, Sep. 2011.
F. Yaman, N. Bai, B. Zhu, T. Wang, and
G. Li, BLong distance transmission in
few-mode fibers,[ Opt. Exp., vol. 18, no. 12,
pp. 13 250–13 257, Jun. 2010.
F. Yaman, N. Bai, Y.-K. Huang, M.-F. Huang,
B. Zhu, T. Wang, and G. Li, B10 112 Gb/s
PDM-QPSK transmission over 5032 km in
few-mode fibers,[ Opt. Exp., vol. 18, no. 20,
pp. 21 342–21 349, Sep. 2010.
P. J. Winzer, BEnergy-efficient optical
transport capacity scaling through spatial
multiplexing,[ IEEE Photon. Technol. Lett.,
vol. 23, no. 13, pp. 851–863, Jul. 2011.
Y. Kokubun and M. Koshiba, BNovel
multi-core fibers for mode division
multiplexing: Proposals and design
principle,[ ICICE Elect. Exp., vol. 6, no. 8,
pp. 522–528, Apr. 2009.
C. Xia, N. Bai, I. Ozdur, X. Zhou, and G. Li,
BSupermodes for optical transmission,[
Opt. Exp., vol. 19, no. 17, pp. 16 653–16 664,
Aug. 2011.
T. Hayashi, T. Taru, O. Shimakawa,
T. Sasaki, and E. Sasaoka, BDesign and
fabrication of ultra-low crosstalk and low-loss
multicore fiber,[ Opt. Exp., vol. 19, no. 17,
pp. 16 576–16 592, Aug. 2011.
J. Sakaguchi, Y. Awaji, N. Wada, A. Kanno,
T. Kawanishi, T. Hayashi, T. Taru,
T. Kobayashi, and M. Watanabe, B109-Tb/s
(7 97 172-Gb/s SDM/WDM/PDM)
QPSK transmission through 16.8-km
homogeneous multi-core fiber,[ presented
at the Opt. Fiber Commun. Conf. (OFC),
Los Angeles, CA, 2011, PDPB6.
B. Zhu, T. F. Taunay, M. Fishteyn, X. Liu,
S. Chandrasekhar, M. F. Yan, J. M. Fini,
E. M. Monberg, and F. V. Dimarcello,
B112-Tb/s space-division multiplexed
DWDM transmission with 14-b/s/Hz
aggregate spectral efficiency over a 76.8-km
seven-core fiber,[ Opt. Exp., vol. 19, no. 17,
pp. 16 665–16 671, Aug. 2011.
ABOUT THE AUTHORS
Ezra Ip, photograph and biography not available at the time of
publication.
Lei Xu, photograph and biography not available at the time of
publication.
Philip Ji, photograph and biography not available at the time of
publication.
Dayou Qian, photograph and biography not available at the time of
publication.
Eduardo Mateo, photograph and biography not available at the time of
publication.
Neng Bai, photograph and biography not available at the time of
publication.
Yue-Kai Huang, photograph and biography not available at the time of
publication.
Ting Wang , photograph and biography not available at the time of
publication.
1078
Proceedings of the IEEE | Vol. 100, No. 5, May 2012