JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO.

4, FEBRUARY 15, 2010 423

Approaching the Non-Linear Shannon Limit

Andrew D. Ellis, Jian Zhao, Member, IEEE, and David Cotter

(Invited Paper)

Abstract—We review the recent progress of information theory

in optical communications, and describe the current experimental
results and associated advances in various individual technologies
which increase the information capacity. We confirm the widely
held belief that the reported capacities are approaching the funda-
mental limits imposed by signal-to-noise ratio and the distributed
non-linearity of conventional optical fibres, resulting in the reduc-
tion in the growth rate of communication capacity. We also discuss
the techniques which are promising to increase and/or approach
the information capacity limit.
Index Terms—Information rates, modulation coding, non-linear
optics, wavelength division multiplexing.
Fig. 1. Evolution of telecommunication network capacities in response to
changing access technologies and consumer applications. Circles: Bandwidth
of available access network connection. Candlesticks: Ratio of maximum
I. INTRODUCTION capacity of managed link in core network to access rate. Solid line: Growth
trend for net access rate, Dashed Lines: Approximate bounds of ratio of core to
access capacity.
HE capacity of optical communication links has grown
T exponentially since their introduction in the late 1970s,
and each generation has enabled new methods of communica-
modern technologies currently being researched to approach
tion and services ranging from simple text e-mails through to
this limit. We will also outline the fundamental issues required
the ubiquitous video applications in use today. This continuing
to increase the limits of current optical networks imposed by
growth has been enabled by many individual technological ad-
signal-to-noise ratio and fibre non-linearities. The paper is or-
vances, including third-window distributed-feedback lasers, er-
ganised as follows. In Section II, we briefly review historical
bium-doped fibre amplifiers, wavelength division multiplexing
trends in communication bandwidth provision, before consid-
(WDM), dispersion management, forward error correction and
ering in Section III the classical theoretical predictions of the
Raman amplification. Throughout this technological evolution,
limits to communication capacity of linear transmission chan-
one constant factor has been the use of single mode optical
nels. In Section IV, the more recent development of the use of
fibre, although with evolving designs to control chromatic dis-
information theory to predict the capacity of non-linear channels
persion and non-linearity. However, there is now a growing real-
in optical fibre communications is reviewed. Then in Section V
isation that the continuing bandwidth demand will shortly push
we describe recently proposed technologies including the multi-
the required capacity close to the maximum capacity which has
carrier transmission techniques of orthogonal frequency divi-
been predicted theoretically for such fibres. The economic and
sion multiplexing (OFDM) and coherent WDM, and discuss the
other consequences of demand exceeding capacity are a matter
potential of these technologies to allow significant increases in
of much debate. However, it is generally acknowledged that
the maximum capacity of an installed optical fibre network.
the current network architectures and transmission technologies
will not be capable of meeting the customer bandwidth demand
in the medium term. The time at which demand exceeds supply II. BACKGROUND
may be delayed by changes in network architecture and service The evolution of demand in telecommunications networks
pricing, but is likely to occur within the next decade. may be traced by plotting the bandwidth available to the user
In this paper, we will review the limits to the information (net access rate, circles) against the date of introduction of var-
capacity in optical fibre communications, and investigate the ious access technologies, as shown in Fig. 1, starting from the
introduction of the 1.2 kb/s modem for use in Bulletin Board
Manuscript received May 19, 2009; revised August 10, 2009. First published Systems in 1978 [1], [93] through to Passive Optical Networks
August 21, 2009; current version published February 01, 2010. This work was at contended bit rates up to 10 Gb/s [2] for video and gaming
supported by Science Foundation Ireland under Grant 06/IN/I969, Enterprise
Ireland under Grant CFTD/08/333 and the European Commission under Project applications. The solid line shows a steady growth rate of 15%
PHASORS (FP7-ICT-2007-2 22457). per annum in net access rate. To date overall network capac-
The authors are with the Tyndall National Institute and Department of ities, such as transatlantic link capacities, have increased at a
Physics, University College Cork, Ireland (e-mail: [email protected];
[email protected]; [email protected]). faster rate [3] due to increasing numbers of users with access to
Digital Object Identifier 10.1109/JLT.2009.2030693 advanced communications services.
0733-8724/$26.00 © 2010 IEEE

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 424 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO. 4, FEBRUARY 15, 2010

Fig. 3. (a) Ideal transmitted constellation (continuous) and (b) discrete point
approximation (after [8]).

Fig. 2. Evolution of maximum reported transmission capacity for single wave- if the communication rate satisfies , where is the
length (diamonds, open symbols for optical time division multiplexing), wave-
length division multiplexing (triangles), single and multi-banded OFDM (filled channel capacity, and is given by
circles) and coherent detection (open circles).
(1)

Accompanying the increase in available access rates has been

where is the average signal power and equals ,
a steady increase in the bandwidth within the core of the net-
where is the average energy per bit, the noise spectral
work. The candlestick symbols in Fig. 1 illustrate as a func-
density and the channel bandwidth. The proof presented in
tion of time the ratio of two parameters: the numerator is the
[6] assumes arbitrary line and error correction coding. For a
maximum capacity in the core network which may be inde-
linear channel degraded by additive white Gaussian noise, the
pendently configured (for example, a T1 or T2 carrier in 1978,
optimum constellation in phase and quadrature components of
and a 40 Gb/s SONET wavelength or 100 GbE wavelength in
the optical field may be calculated. The optimum field takes ar-
2012); and the denominator is the available access rate. De-
bitrary continuous values, with the probability of each value fol-
spite the exponential growth in bandwidth demand, this ratio
has remained remarkably constant, representing the continual lowing a Gaussian distribution [5]. Such a continuous biGaus-
design trade off that is made between, on one hand, complexity sian distribution may be emulated in practice by a discrete-point
(favouring coarse bandwidth granularity in the core network), constellation [8], [9], as shown in Fig. 3. In a well designed
and on the other hand, reliability (favouring fine granularity). discrete point constellation the density of points reduces with
Over a period of three decades since the late 1970s to the present distance from the centre of the constellation, in a manner ap-
day, these ratios have consistently fallen within a band of values proaching the optimum distribution. This approximation may
500–5 000 (horizontal dashed lines, Fig. 1). Despite profound be improved further by varying the probability of occupancy of
developments in the underlying technologies over this period, it each point in the constellation.
would appear that the basic cost-driven design trade-offs have Many different constellations may be considered for optical
remained unchanged. Therefore, extrapolating these bands of transmission (as shown for example in Fig. 4), ranging from
values into the future suggests that the network should be able single quadrature formats typically generated with a single
to support 100 Gb/s transport in the core network today [4] and amplitude modulator, including (a) binary phase shift keying
1 Tb/s transport as early as 2017. However, to maintain the cur- (BPSK), (b) amplitude shift keying (ASK) and (c) quaternary
rent core network architecture, this would require a total number ASK (4-ASK), to formats consisting of in-phase and quadra-
of wavelengths deployed similar to today, typically 160, but car- ture components, including (d and e) M-ary phase shift keying
rying information at an information spectral density exceeding (QPSK and 8PSK respectively), (g and h) quadrature amplitude
30 b/s/Hz (which represents an immense technical challenge). shift keying (QAM) constellations (typically generated using a
On the other hand, Fig. 2 illustrates the evolution of the avail- dual parallel Mach Zehnder modulator), and (f) hybrid ampli-
able fibre transmission capacity reported from experiments car- tude phase shift keying (APSK) (typically generated using an
ried out in research laboratories worldwide. Whilst a long term amplitude modulator and a phase modulator in series).
growth trend of around 60% per annum was observed from the To calculate the performance of each constellation, we first
early 1990s, this has saturated recently, prompting a move to- determine the impact of noise on each constellation point. For a
wards the adoption of coherent detection techniques, where the system using coherent detection, the noise and signal are com-
additional degree of freedom (optical phase) is expected to allow bined as a vector addition and the noise is independent of the
for greater capacity increases [5]. However, even with such in- signal amplitude [10], [11]. On the other hand, for direct- and
novations, capacity increases of about 25 times the current ca- differentially-detected signals, the noise level after detection is
pacity, as required around 2017, appear to be very challenging. dependent on the signal intensity [12]. In this paper, we consider
only coherent detection with signal independent noise, which is
III. SHANNON LIMIT the optimum case appropriate to evaluate the performance limit.
The Shannon limit to information capacity on a communica- In the following, we calculate the bit error rate (BER) perfor-
tions link [6], [7] is well known. Shannon proved that reliable mance of a given constellation assuming hard decision detec-
communication over a discrete memory-less channel is possible tion, by calculating the probability that a given transmitted bit

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 ELLIS et al.: APPROACHING THE NON-LINEAR SHANNON LIMIT 425

TABLE I
ERROR PROBABILITIES FOR A FEW COMMON MODULATION FORMATS AS A
FUNCTION OF ELECTRICAL SIGNAL-TO-NOISE RATIO

Fig. 4. Some examples of signal constellations with one (a), (b), two (c), (d)
and three (e)–(h) bits per symbol.

The total BER is then given by the sum of the BER for each bit
multiplied by the probability that this bit is transmitted,
that is

(4)

The transmitted signal power is related to the geometric distri-

bution of the constellation points, such that the mean energy per
bit of the transmitted bit is
Fig. 5. An example of BER calculation. a) Constellation diagram for 4-ASK, (5)
b) Probability of detecting a given field amplitude given that symbol 2 was trans-
mitted.
which in turn gives an electrical signal-to-noise ratio of .
Here, the signal-to-noise ratio is the parameter commonly used
in communication theory, and in an optical system limited
crosses an imaginary boundary (the decision threshold) between by amplified spontaneous emission noise (ASE), represents
it and its nearest neighbour [13]. We use the constellation of the photon number per bit entering the optical pre-amplifier.
Fig. 4(c) as an example. For a coherently detected signal, the electrical signal-to-noise
In this example, also shown in Fig. 5, additive white Gaussian ratio (snr) is related to optical signal-to-noise ratio (OSNR) by
noise gives a Gaussian probability density function for the re- , where and are the transmis-
ceived signal values (Fig. 5(b)). For the second bit , erro- sion capacity and reference noise bandwidth (e.g., 12.5 GHz,
neous detection occurs by crossing either of two decision thresh- corresponding to 0.1 nm at 1550 nm wavelength) respectively,
olds towards its nearest neighbours ( and ). The probability and the ASE noise is assumed to be randomly polarised. BER
of an error for this bit is thus performance results for a few common modulation formats are
shown in Table I, [11], [14], where represents the number
(2) of constellation points bits per symbol) assuming
.
where is the field amplitude of the th constellation point and The performance of each format may then be compared to
is related to the complimentary error function by the Shannon capacity limit by calculating the required snr for
a given BER and error correction code, and calculating the
net information spectral density (number of transmitted bits
(3) per hertz), taking the symbol rate and the number of bits per
symbol into account.
Note that here, Q represents a mathematical function, and Due to the current limitation in the electronic bandwidth, it
should not be confused with the “Q-factor” used in optical is impractical to modulate the full optical bandwidth available.
communications. Note that the constellation points located at Current technologies to achieve the maximum possible infor-
the two ends (furthest from the centre in the general case) have mation throughput involve WDM where the available optical
fewer nearest neighbours and therefore smaller error probabili- bandwidth is split into frequency bands, each of which is mod-
ties. For example, a transmitted would only be erroneously ulated separately. In this case, the information spectral density
detected if it crossed the threshold between itself and . also depends on the combined width of the guard bands

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 426 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO. 4, FEBRUARY 15, 2010

Fig. 7. Illustration of the limitation in the net information capacity as a func-

tion of the number of transmitted bit per symbols for uni-polar M-ASK (circles),
bipolar M-ASK (down triangles), M-PSK (squares) and QAM (diamonds) as-
Fig. 6. Information spectral density of uni-polar ASK (circles), PSK (squares) suming a snr of 15.5 dB.
and QAM (diamonds) showing the maximum system capacity as a function of
electrical signal-to-noise ratio for a BER of 10 in a WDM system with 20%
guard bands between channels. The solid line represents the Shannon theoretical
limit [3], [6]. Exploiting both quadratures with M-PSK further reduces the re-
quired snr (see Fig. 6) and 4 bit per symbol M-PSK is possible
at 15.5 dB snr with appropriate FEC. QAM exhibits still fur-
between WDM channels. Fig. 6 shows the information spec- ther performance enhancement, resulting in 5 bits per symbol
tral density (ISD) as a function of snr in a WDM system with without significant reduction in throughput due to FEC over-
20% guard bands for uni-polar ASK (circles), PSK (squares), head.
and QAM (diamonds) formats. Whilst it is likely that FEC circuits will become available
Whilst the benefits associated with modulation in both which will require less overhead than assumed here, including
quadratures, using either M-PSK or QAM, are apparent from current proprietary FEC circuits, it will still be the case that an
Fig. 6 and the expressions in Table I and recent record spectral optimum ISD will exist for a given snr and modulation format.
density results [15] strong forward error correction (FEC) is It is clear that any required guard band between WDM
essential to enable operation close to the fundamental Shannon channels reduces the ISD. The guard bands may be avoided
limit [16]. Furthermore, the use of higher order modulation by employing orthogonal frequency division multiplexing
formats suggests that the capacity increase is only obtained (OFDM) techniques [19], [20], such as no-guard-interval
at the expense of requiring higher snr and implementation OFDM [21]–[23], coherent WDM [24]–[27] direct detection
complexity. These requirements could be reduced somewhat OFDM [28], [29] and coherent optical OFDM [30]–[35].
by combining functions traditionally performed separately, In all of these multi-carrier systems, the frequency spacing
for example, it has been demonstrated that demodulation and between the orthogonal sub-carriers is equal to the symbol rate
error correction may be performed simultaneously, with some per subcarrier. A typical example of the orthogonal carriers are
performance benefit [17]. shown in Fig. 8, where the peak of the spectrum of a given
The reduction in ISD due to the FEC overhead is shown in channel corresponds to nulls in the spectra of all of the other
Fig. 7 for various modulation formats employing coherent de- sub-channels, and in particular, the first null in the spectrum
tection. In this figure it is assumed that the baseline system was of the adjacent sub-channel. Ideally, matched filters are used to
designed to give a BER of at a 15.5 dB snr (sufficient for separate each sub channel [19], and this may be implemented ef-
direct detection of an on-off keyed signal). The required FEC ficiently using Fast Fourier Transform algorithms for low sub-
overhead for error-free operation is calculated and then sub- channel data rates (e.g., 100 Mb/s), with the digital signal pro-
tracted from the net information capacity. A simplified approx- cessing (DSP) complexity scaling approximately linearly with
imation was used to calculate the FEC overhead, where each the total capacity ( , where is the channel number)
FEC was assumed to require 7% overhead [18] for every [28]–[33]. However, for a system with a high symbol rate per
of BER to be corrected. For example, we assumed an overhead channel (e.g., 40 Gb/s), the practical implementation of precise
of 21% for the correction of a BER of . From Fig. 7, matched filters proves difficult, and may be approximated in the
it is shown that the calculated overhead results in a negligible optical domain using asymmetric Mach Zehnder interferome-
decrease in capacity for 2-bit per symbol uni-polar signal with ters [23], [24] or with simple digital filters [22]. The impact of
coherent detection. However, as the number of bits per symbol any residual crosstalk may then be minimised using appropriate
is increased, the BER degradation increases, requiring larger optimisation of the relative phases of each sub-channel [25] or
overheads, and, eventually, the required additional FEC over- cancelled using post-detection signal processing [26], [36]. In
head outstrips the additional capacity offered by an extra bit per all cases, the net result is the straightforward generation of a
symbol, at a fixed snr. By changing to bipolar, the required snr signal with a capacity per polarisation equal to the number of
is greatly reduced, allowing, in this example, 3 bits per symbol. bits per symbol (or ), with the potential suitability for

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 ELLIS et al.: APPROACHING THE NON-LINEAR SHANNON LIMIT 427

[62], [63], or via emulation of back propagation using look-up

tables [64].
Assuming ideal compensation of all intra-channel effects
other than noise, cross-phase modulation (XPM), which causes
multiplicative noise, appears to be the principal source of
impairments that fundamentally limits the information capacity
of an optical communication system. XPM induces random
fluctuations in the target channel which are exponentially
Fig. 8. Illustration of overlapping modulation sidebands of OFDM signal. related to the intensity of the neighbouring WDM channels. It
has been shown that the intensity scale for these fluctuations is
given by [50], [65]
ultrahigh total capacities (between 300 and 1 080 Gb/s and be-
yond [37]–[39]) which are difficult to achieve using single car-
rier modulation.
(6)

IV. NON-LINEAR LIMITS where and are the channel bandwidth, local dis-
persion, WDM channel spacing and fibre non-linear coefficient
The above discussion applies equally to optical fibre, wire- respectively. is the number of WDM channels and is
less and copper based transmission systems, and in the absence the non-linear effective length of the system given by
of any further signal degradation, performance approaching the , for a system with lumped amplifiers where is
Shannon limit would be possible using forward error correc- the loss coefficient and is the number of amplifiers. Here we
tion. Wireless systems, particularly those employing OFDM, have assumed equally spaced channels of equal intensities. A re-
experience non-linearity due to the saturation characteristics of lated amount of information is lost from the channel of interest
power amplifiers [40]. On the other hand, periodically-ampli- due to the random crosstalk induced by XPM. The net effect is
fied optical fibre based systems are characterised by distributed to reduce the information capacity of a coherently detected (CD)
non-linear effects in the fibre itself. The most predominant system to
non-linear effect arises from the intensity dependent refractive
index (Kerr effect) and results in a number of phenomena
such as self-phase modulation [41], cross-phase modulation
[42] and inter- [43] and intra-channel [44] four wave mixing.
Whilst many techniques to mitigate the impact of non-linearity
have been developed, including most significantly dispersion (7)
management [45]–[49], the impact of these non-linearities in
terms of the information theoretical limits have only recently where is the average signal power per channel. (for a
been addressed [50]–[52]. The key simplification introduced system with discrete amplifiers) is equal to ,
by Mitra and Stark [50] was to equate a non-linear communi- where is the number of fibre spans, is the amplifier gain
cation channel to a linear channel with multiplicative noise, for and is the spontaneous emission noise factor. Note that this
which analytical results can be obtained. It was found that, in equation is applicable to OFDM or coherent WDM techniques,
contrast to linear channels with additive noise, the capacity of and in a conventional WDM system, the capacity is reduced by
a non-linear channel does not grow indefinitely with increasing a factor of , where is the channel spacing in the fre-
signal power, but has a maximal value. This is a fundamental quency domain. The same approach has been employed to deter-
feature which distinguishes non-linear communication chan- mine the reduction in information capacity when the dominant
nels from linear ones. In making use of this new analytical non-linearity is four-wave-mixing [5], where the phase condi-
approach, it is assumed that any deterministic effects, such as tion in the fibre is different. The capacity bound is given by (7)
chromatic dispersion and self-phase modulation, which depend with the non-linear intensity replaced by , where
only on the channel of interest, may be fully compensated. This
compensation may take the form of efficient modifications of
the transmitted or received signals based on prior knowledge
of the signal format itself. For example, reduction in dispersion
penalties are observed using pre-chirp [53], [54] or electronic
dispersion compensation [55]–[57], whilst non-linear penalties (8)
may be lowered by reducing phase noise (or timing jitter) by
modulating the received signal with a phase proportional to The general form of these predictions is confirmed by indepen-
the received intensity [58]–[60]. For multi-level formats, these dent analysis [66]. Following the same general argument, it may
techniques may also be applied predicatively at the transmitter be expected that the maximum ISD of a system employing di-
[61]. Full non-linearity compensation may be applied at the rect detection (DD) is similarly degraded. Thus, starting from
expense of complexity, either by optical phase conjugation the linear ISD limit [5], [67], we find (for high OSNR) that

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 428 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO. 4, FEBRUARY 15, 2010

Fig. 9. Examples of predicted information spectral density limits per polarisa-

tion for linear transmission with coherent (long dashes) and direct (dot-dashed) Fig. 10. Maximum reported information capacity as a fraction of the maximum
detection and for non-linear transmission including XPM for coherent (solid) information capacity (7) for the system configuration reported versus the year
and direct (short dashes) detection. Detailed parameters are shown in Table II. for transmission distances up to 2 000 km.

TABLE II
SIMULATION PARAMETERS USED FOR FIG. 9 ONWARDS, UNLESS OTHERWISE ISD for any given system configuration. Since the inception of
SPECIFIED. VALUES ARE SELECTED TO INDICATE GENERAL TRENDS AND DO
NOT REPRESENT ACTUAL SYSTEM DESIGNS optical communication systems, advances in various individual
technologies have enabled these limits to be approached, and
the ratio of ISDs for numerous reported transmission system
experiments to the maximum values are shown in Fig. 10. Much
of the progress is attributed to improvements in modulation
efficiency, adoption of WDM and the subsequent reduction in
channel spacing. The reason for the recent reduction in the rate
of growth of bit rate distance product (Fig. 2) also becomes
apparent, as we observe in Fig. 10 that experimental measure-
ments had already exceeded 50% of the theoretical maximum
information capacity by 2008 [69]. This imminent limit to
growth in the information capacity is stimulating renewed
interest in techniques for mitigating the effect of non-linearity,
in order to allow operation at higher launch powers, and hence
higher overall capacities (Fig. 9).
Experimental data is also compared to the theoretical limits in
Fig. 11, but, this time, as a function of transmission distance for
direct detection (upper) and coherent detection (lower). Again,
we readily observe that recent experimental results are within a
factor of two of the maximum ISD predicted by (7).
(9) For the longest transoceanic systems ( km and based
on direct detection [60], [70], [71]), whilst the achieved ISDs are
Fig. 9 depicts the XPM-limited ISD versus transmitted power usually modest, below 1 b/s/Hz, the results closely approach the
density for coherent and direct detection. The information limits Shannon limit. In these papers, the dispersion maps were opti-
in the linear channels are also plotted for comparison. In the mised to minimize the impact of optical non-linearity; this was
figure, a high local dispersion coefficient value and a non- achieved in various ways: by removing all dispersion compen-
quasi-phase-matched dispersion map are assumed to avoid res- sation from the system [70]; by combining multiple fibre types
onances in the non-linear response of the system [43], [47]. within the transmission span with different properties [71] or
High local dispersion values can help to reduce the interac- by destroying quasi-phase matching [43], [44] by carefully bal-
tion between the WDM signals and ASE by minimising the ancing broadband dispersion compensation (fibre based) and
phase matching. The figure shows the increase in maximum using a periodic group delay compensation device [60]. With
ISD achieved by using coherent detection, and the effect of fibre the benefit of hindsight, we can say that each of the dispersion
non-linearity at higher transmitted powers preventing indefinite management schemes above follow the predictions of (7). In ad-
growth in the channel capacity. For this particular example, the dition, these results partially benefited from a degree of non-lin-
effect of XPM becomes prominent at transmitted power densi- earity compensation, either through the use of guiding filters to
ties beyond 0.01 W/THz, and a maximum ISD of 6 b/s/Hz is combat intra-channel self-phase modulation [60], [72], or via
predicted. A similar value was reported in recent numerical sim- midspan spectral inversion [62], [70].
ulations [68]. Coherent detection can greatly increase the limits to infor-
It is relatively straightforward to calculate the maximum mation capacity, and has attracted much interest recently for
launch power from (7) above, and thus predict the maximum long haul transmission [33]. In order to approach the maximum

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 ELLIS et al.: APPROACHING THE NON-LINEAR SHANNON LIMIT 429

are required [4]. In this case, a direct detection system which

also uses a form of OFDM to increase the ISD may be used [27],
where a dispersion map optimised for 40 Gb/s transmission was
used to minimise the impact of non-linearity [49], [73]. With co-
herent detection, fibre non-linearity was successfully managed
over an in-line compensated map by using a RZ-shaped constant
intensity modulation format, and the ISD was increased further
using 8-PSK [74].
Over even shorter distances, the improved OSNR enables the
use of yet higher-order modulation formats. Although a wide
variety of formats have been studied, the highest ISDs are typ-
ically associated with single side-band [75], OFDM [35], and
large constellation QAM systems [15], [76], the common fea-
ture of which is the efficient use of the available spectrum by
minimising duplication of information.

V. INCREASING THE INFORMATION CAPACITY LIMIT

We observed in Section IV that, for both direct and coherent
detection, the latest experimental results are falling on a line ap-
proximately a factor of two away from the ultimate limits pre-
dicted by the Shannon limit incorporating the effects of cross-
Fig. 11. Maximum reported information spectral density versus transmission phase modulation. In the few cases where this gap appears to
distance for experimentally reported data (dots) and theoretical limits (lines) have been closed, the experiments mitigated the fibre non-lin-
with EDFA-amplified systems with 40 km repeater spacing (dotted line) and 80
km spacing (solid line) (other parameters as per Table II). earity [60], [70] or employed larger effective area fibre (note
that the reference calculations assumed a fixed value of these
parameters) [71], [77].
In addition to implementing the various modulation formats,
impairment mitigation techniques, and FEC technologies to ap-
proach the limit, it is also desirable to increase the maximum
information spectral density by taking optimum values for the
parameters in (6–8), including critical fibre characteristics (loss,
dispersion and non-linear coefficient), the channel spacing, the
effective amplifier noise figure and finally the number of cas-
caded links. The capabilities of optically multiplexed OFDM
[39], or coherent WDM [27], [37] to generate phase coherent
high capacity signals from a single source, suggests one way
to extend the theoretical ISD limit. That is, by increasing the
channel spacing, if compensation of the linear and non-linear
Fig. 12. Theoretical channel information spectral density limits versus power impairments may be applied across the entire OFDM spectrum,
spectral density in a 10 THz bandwidth, plotted for different values of the
channel bandwidth: 1 GHz 100 GHz, 1 THz, 3 THz, 5 THz and 10 THz an increased fraction of the spectrum may be treated as a single
(dashing length varies from 1 GHz-longest to 5 THz shortest, and 10 THz is channel [78], [79]. Through (7) we may expect that this would
the solid line) with other parameters as per Table II. reduce the impact of XPM enabling the information capacity
limit to be enhanced.
In these systems, for a fixed amplifier bandwidth, e.g., 10
information spectral density in such system, the dispersion is THz, the non-linear ISD limit arising from XPM is dependent
maintained at its maximum value throughout the transmission on the channel granularity, as shown in Fig. 12, which depicts
to minimise non-linear effects, and the information capacity of the theoretical capacity limit in a 10 THz bandwidth for different
the transmitted channel may be increased by using OFDM to channel spacing (or occupied bandwidth per channel in OFDM
minimise the spacing of sub-carriers, and by using QPSK mod- systems). From this figure, it can be seen that the maximum in-
ulation on each sub-carrier. Similar results were obtained using formation capacity is increased as the channel bandwidth in-
134 WDM channels, each of which carried OFDM with QPSK creases due to the anticipated dependence of information ca-
sub-carriers [22]. In both cases, coherent detection not only fun- pacity with channel bandwidth (6). The curve for 10 THz corre-
damentally increased the information capacity limit by reducing sponds to the one-channel case without the possibility of XPM,
the information loss from noise and allowing access to a second and this is the same as the Shannon limit in a linear channel. The
quadrature, but also enabled the compensation of the accumu- bandwidth per channel is typically less than 100 GHz in present
lated chromatic dispersion using DSP. commercially available systems; a limit imposed both by stan-
For shorter transmission distances more suited to inland net- dardisation and also by the available modulation and detection
works, transmission reaches in the region of 1 000 to 1 500 km bandwidths.

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 430 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO. 4, FEBRUARY 15, 2010

Fig. 13. Theoretical channel ISD for various values of the amplifier noise figure
(dotted: 4.5 dB NF, dashed: 3 dB NF, solid: 0 dB NF), other parameters as per
Table II.
Fig. 14. Maximum ISD of various fibre types. See Table III for fibre parameters
and Table II for other parameters.
The logarithmic dependence of (1) and (7) with respect to the
noise power spectral density reduces the benefit from amplifier
TABLE III
noise figure reductions, as showing in Fig. 13, such that a 1 dB FIBRE PARAMETERS USED FOR FIG. 14 (EST: ESTIMATED FROM EFFECTIVE
reduction in noise figure results in substantially less than a 1 dB AREA OR FRACTION OF LIGHT PROPAGATING IN MEDIUM, *: ASSUMED VALUE
increase in the maximum ISD. In Fig. 13, we consider the ef- FOR CHROMATIC DISPERSION, $: PREDICTED VALUE)

fect of reducing the amplifier noise figure from a typical value

of 4.5 dB to the quantum limit of 3 dB, confirming that this of-
fers only a small increase in the maximum ISD. In this example,
the ISD limit may be increased by a further 1 b/s/Hz by using
phase sensitive amplification for which the theoretical minimum
noise figure is 0 dB. Note that, in this case, we must consider the
quantum nature of light and that the photon number distribution
is fundamentally broadened by periodic attenuation and ampli-
fication. The net effect of these quantum processes is that the
signal-to-noise ratio is improved by a factor of 2 by moving from
a quantum-limited phase insensitive amplifier to a quantum-lim-
ited phase sensitive amplifier [12], [80], [81]. Similarly modest
increases in capacity are expected from systems operating with
distributed amplification, where the OSNR is maximised [8].
Whilst the increase in net ISD is small, of greater interest is Fibre designs, on the other hand, offers substantially greater
the required total launch power for a given ISD. For an ISD of scope for performance improvement. Indeed many transmis-
5.85 b/s/Hz, according to Fig. 13, a system comprising 4.5 dB sion records were attributed to the fibre designs [71], [84] in
noise figure amplifiers would require a launched power spectral addition to optimised transmission formats. Fig. 14 illustrates
density of around 14 mW/THz. This would result in a system the predicted maximum performance for a number of measured
operating in a non-linear transmission regime, requiring proper solid core fibres, all of which demonstrate a maximum infor-
design to minimise inter-channel non-linearity, and full com- mation capacity between 6 and 8 b/s/Hz, assuming optimum
pensation of intra-channel non-linearity. On the other hand, the dispersion management and full compensation of intra-channel
use of ideal phase sensitive amplifiers allows the operation of non-linearity. Fig. 14 also shows for comparison speculative
the same system in a linear transmission regime, with a launch prediction of the performance of a hollow core photonic crystal
power spectral density of only 3.5 mW/THz, representing a sub- fibre (PCF), where the non-linear coefficient has been reduced
stantial energy saving, even when the reduced power efficiency in proportion to the estimated fraction of the optical signal
of phase sensitive amplifiers [82] is taken into account. Practical propagating in glass (1%) and the predicted minimum loss has
deployment of such phase sensitive amplifiers requires further been assumed [85]. Preliminary transmission measurements
development to realise fibre to fibre noise figures approaching [86] have been performed at a wavelength of 1550 nm. In this
the assumed 0 dB, and the development of systems to ensure region, whilst the loss will be higher than that for conventional
that the useable gain bandwidth of a phase sensitive amplifier fibres, it might be expected that the significantly reduced
approaches that of the phase insensitive parametric amplifiers non-linearity would enable the development of a hybrid system,
[83] with PCF used in sections experiencing high optical powers,
We may therefore conclude from Fig. 13 that the overall ca- and low loss fibre elsewhere. Development of the necessary
pacity could in principle be increased by 1–2 b/s/Hz by only transponder and amplifier technology [e.g., 87, 88] for opera-
modifying the transponders and amplifier sites, offering signif- tion in the midinfrared region, accompanied by achievement
icant practical advantage. of the predicted low loss performance around 1900 nm, should

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 ELLIS et al.: APPROACHING THE NON-LINEAR SHANNON LIMIT 431

enable the development of long reach communication with [10] J. M. Kahn and E. Ip, “Principles of digital coherent receivers for op-
information capacities above 10 b/s/Hz. The overall capacity tical communications,” in Proc. OFC 2009, San Diego, ISA, 2009,
2009, Paper OTuG5.
of the fibre will of course be restricted by the wavelength range [11] E. Ip, A. P. T. Lau, D. J. F. Barros, and J. M. Khan, “Coherent detection
over which the fibre wave-guiding is effective. Reductions in optical fiber systems,” Opt. Exp., vol. 16, no. 2, pp. 753–791, 2008.
in the effective non-linear coefficient are also expected for [12] E. Desurvire, Erbium-Doped Fiber Amplifiers. Hoboken, NJ: Wiley,
2002.
multi-mode fibre systems [89], and proportional increases in [13] S. Haykin, Digital Communications. Hoboken, NJ: Wiley, 1988.
maximum ISD may be expected if appropriate measures are [14] J. G. Proakis, Digital Communications, 4th ed. New York: McGraw-
taken to accommodate the bandwidth limitations. Hill, 2000.
[15] M. Nakazawa, “Challenges to FDM-QAM coherent transmission with
ultrahigh spectral efficiency,” in Proc. ECOC 2008, Brussels, 2008,
VI. CONCLUSION Paper Tu1E1.
[16] S. Y. Chung, G. D. Forney, T. J. Richardson, and R. Urbanke, “On
Communication capacity has shown a remarkable exponen- the design of low density parity check codes within 0.0045 dB of the
tial growth over more than 30 years, with the overall capacity Shannon limit,” IEEE Commun. Lett., vol. 5, no. 2, pp. 58–60, 2001.
of the core of the network closely tracking the user demand. [17] B. Zhou, L. Zhang, J. Kang, O. Huang, Y. Y. Tai, S. Lin, and M. Xu,
“Non-binary LDPC codes versus Reed–Solomon codes,” in Proc. Inf.
In this paper we have discussed the recent reduction in the Theory Appl. Workshop 2008, San Diego, CA, 2008, pp. 175–184.
growth rate of communication capacity in laboratory systems [18] G.709: Interfaces for the Optical Transport Network (OTN) [Online].
and have shown that this is because the reported capacities are Available: http://www.itu.int/rec/T-REC-G.709/en downloaded from
[19] R. W. Chang, “Synthesis of band-limited orthogonal signals for multi-
approaching the fundamental limits imposed by signal-to-noise channel data transmission,” Bell Syst. Tech. J., vol. 45, pp. 1775–1796,
ratio [6] and the distributed non-linearity of conventional 1966.
[20] R. R. Mosier and R. G. Clabaugh, “Kineplex, a bandwidth efficient
optical fibres [50].
binary transmission system,” AIEE Trans., vol. 76, pp. 723–728, 1958.
In the short term, capacity growth by a factor of ap- [21] H. Sanjoh, E. Yamada, and Y. Yoshikuni, “Optical orthogonal fre-
proximately two may be expected by each of the following quency division multiplexing using frequency/time domain filtering
for high spectral efficiency up to 1 bit/s/Hz,” in Proc. OFC’02, Ana-
approaches: compensation of intra-channel non-linearity either
heim, 2002, Paper ThD1.
through link design or signal processing at the terminals; [22] A. Sano, E. Yamada, H. Masuda, E. Yamazaki, T. Kobayashi, E.
increase in the total bandwidth (and capacity) of each channel Yoshida, Y. Miyamoto, S. Matsuoka, R. Kudo, K. Ishihara, Y. Taka-
tori, M. Mizoguchi, K. Okada, K. Hagimoto, H. Yamazaki, S. Kamei,
using phase coherent optical multiplexing; optimisation of the 2
and H. Ishii, “13.4-Tb/s (134 111-Gb/s/ch) no-guard-interval co-
OSNR through careful link design; adoption of distributed and herent OFDM transmission over 3 600 km of SMF with 19-ps average
phase sensitive amplifiers and optimisation of the amplifier PMD,” in Proc. ECOC’08, Brussels, 2008, Paper Th3E1.
spacing. However, with the underlying long term rate of ca- [23] K. Takiguchi, M. Oguma, T. Shibata, and H. Takahashi, “Optical
OFDM demultiplexer using silica PLC based optical FFT circuit,” in
pacity growth equal to a doubling every two years, it appears Proc. OFC 2009, San Diego, CA, 2009, Paper OWO3.
inevitable that unless capacity demand saturates, or network [24] A. D. Ellis and F. C. G. Gunning, “Filter strategies for coherent WDM,”
in Proc. Emerging Technol. Opt. Sci., Cork, Jul. 26th–29th, 2004.
architectures are devised which radically alter the capacity [25] A. D. Ellis and F. C. G. Gunning, “Spectral density enhancement
demands placed on the core network, new transmission media using coherent WDM,” IEEE Photon. Technol. Lett., vol. 17, no. 2,
will be essential within the next two decades. pp. 504–506, 2005.
[26] J. Zhao and A. D. Ellis, “Performance improvement using a novel MAP
detector in coherent WDM systems,” in Proc. ECOC’08, 2008, Paper
ACKNOWLEDGMENT Tu1.D.2.
[27] T. Healy, F. C. G. Gunning, E. Pincemin, B. Cuenot, and A. D. Ellis, “1
The authors would like to thank Dr. S. Ibrahim for useful 200 km SMF (100 km spans) 280 Gbit/s coherent WDM transmission
discussions. using hybrid Raman/EDFA amplification,” ECOC’07. Berlin, 2007,
paper Mo1.3.5.
[28] A. J. Lowery and J. Armstrong, “Orthogonal frequency division mul-
REFERENCES tiplexing for dispersion compensation of long-haul optical systems,”
[1] L. Wood and D. Blankenhdn, “State of the BBS nation: Behold Opt. Exp. 2006, vol. 14, pp. 2079–2084, 2006.
lowly bulletin board now encompassing the globe,” Byte, vol. 15, pp. [29] B. J. C. Schmidt, Z. Zan, L. B. Du, and A. J. Lowery, “100 Gbit/s trans-
298–304, Jan. 1990. mission using single band direct detection optical OFDM,” in Proc.
[2] 10 Gb/s Ethernet Passive Optical Network [Online]. Available: OFC’09, San Diego, 2009, Paper PDPC4.
http://www.ieee802.org/3/av Apr. 20, 2009, IEEE P802.3av Task [30] I. B. Djordjevic and B. Vasic, “100-Gb/s transmission using orthogonal
Force, downloaded frequency-division multiplexing,” IEEE Photon. Technol. Lett., vol. 18,
[3] E. B. Desurvire, “Capacity demand and technology challenges for no. 15, pp. 1576–1578, 2006.
lightwave systems in the next two decades,” J. Lightw. Technol., vol. [31] S. L. Jansen, I. Morita, and H. Tanaka, “10-Gb/s OFDM with conven-
24, no. 12, pp. 4697–4710, 2006. tional DFB lasers,” in Proc. ECOC’07, Berlin, 2007, Paper Tu. 2.5.2.
[4] P. Magill, “System technologies for 100G transport networks,” in Proc. [32] W. Shieh, “High spectral efficiency coherent optical OFDM for 1 Tb/s
OFC 2009, San Diego, 2009, Paper OThR1 (2009). ethernet transport,” in Proc. OFC 2009, San Diego, CA, 2009, Paper
[5] J. M. Kahn and K.-P. Ho, “Spectral efficiency limits and modula- OWW1.
tion/detection techniques for DWDM systems,” IEEE J. Sel. Topics [33] S. L. Jansen, I. Morita, N. Takeda, and H. Tanaka, “20-Gb/s OFDM
Quantum Electron., vol. 10, no. 2, pp. 259–272, Mar./Apr. 2004. transmission over 4 160-km SSMF enabled by RF-pilot tone phase
[6] C. E. Shannon, “A mathematical theory of communication,” Bell Syst. noise compensation,” in Proc. OFC Conf. 2007, Anaheim, 2007, Paper
Tech. J., vol. 27, pp. 379–423, 623–656, 1948. PDP 15.
[7] C. E. Shannon and W. Weaver, The Mathematical Theory of Commu- [34] H. Takahashi, A. Al Amin, S. L. Jansen, I. Morita, and H. Tanaka,
nication. Chicago: Univ. Illinois Press, 1963. “DWDM transmission with 7.0 bit/s/Hz spectral efficiency using
[8] R.-J. Essiambre, “Capacity limits of fiber-optic communication sys- 2
8 65.1 Gbit/s coherent PDM OFDM signals,” in Proc. OFC 2009,
tems,” in Proc. OFC, San Diego, ISA, 2009, 2009, Paper OThL1. San Diego, 2009, Paper PDPB7.
[9] N. Kikuchi, K. Mandai, K. Sekine, and S. Sasaki, “Incoherent 32-level [35] X. Yi, W. Shieh, and Y. Ma, “Phase noise on coherent optical OFDM
optical multilevel signalling technoloigies,” J. Lightw. Technol., vol. systems with 16-QAM and 64-QAM beyond 10 Gb/s,” in Proc.
26, no. 1, pp. 150–157, 2008. ECOC’07, Berlin, 2007, Paper Tu5.2.3.

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 432 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO. 4, FEBRUARY 15, 2010

[36] T. Miki and H. Ishio, “Viabilities of the wavelength-division-multi- [58] X. Liu, X. Wei, R. E. Slusher, and C. J. McKinstrie, “Improving trans-
plexing transmission system over an optical fiber cable,” IEEE Trans. mission performance in differnetioal phase-shift-keyed systems by use
Commun., vol. 26, no. 7, pp. 1082–1087, 1978. of lumped nonlinear phase-shift compensation,” Opt. Lett., vol. 27, pp.
[37] A. D. Ellis, F. C. G. Gunning, B. Cuenot, T. C. Healy, E. Pincemin, 1616–1618, 2002.
and M. Rukosueva, “Towards 1TbE using coherent WDM,” in Proc. [59] K. Kikuchi, “Electronic post-compensation for nonlinear phase fluc-
OECC/ACOFT 2008, Sydney, Australia, 2008, Paper WeA-1. tuations in a 1000-km 20-Gbit/s optical quadrature phase-shift keying
[38] F. C. G. Gunning, T. Healy, X. Yang, and A. D. Ellis, “0.6 Tbit/s ca- transmission system using the digital coherent receiver,” Opt. Exp., vol.
pacity and 2 bit/s/Hz spectral efficiency at 42.6 Gsymbol/s using a 16, no. 2, pp. 889–896, 2008.
single DFB laser with NRZ coherent WDM and polarisation multi- [60] L. F. Mollenauer, A. Grant, X. Liu, X. Wei, C. Xie, I. Kang, and C.
plexing,” in Proc. CLEO Eur. 2007, Munich, Germany, 2007, paper 2
Doerr, “Demonstration of 109 10 G dense WDM over more than 18
CI8-5. 000 km using novel, periodic-group-delay complemented dispersion
[39] Y. Ma, Q. Yang, Y. Tang, S. Chen, and W. Shieh, “1 Tb/s per channel compensation and dispersion managed solitons,” in Proc. ECOC 03,
coherent optical OFDM transmission with subwavelength bandwidth Rimini, 2003, Post-deadline Paper Th4.3.4.
access,” in Proc. OFC’09, San Diego, CA, 2009, Paper PDPC1. [61] C. Xu, X. Liu, and X. Wei, “Differential phase-shift keying for high
[40] J. Armstrong, “Peak-to-average power reduction for OFDM by re- spectral efficiency optical transmissions,” IEEE J. Sel. Topics Quantum
peated clipping and frequency domain filtering,” Electron. Lett., vol. Electron., vol. 10, no. 2, pp. 281–293, Mar./Apr. 2004.
38, no. 5, pp. 246–248, 2002. [62] W. Pieper, C. Kurtze, R. Schnabel, D. Bruer, R. Ludwig, and K. Pe-
[41] D. J. Malyon, T. Widdowson, E. G. Bryant, S. F. Carter, J. V. Wright, termann, “Nonlinearity insensitive standard fibre transmission based
and W. A. Stallard, “Demonstration of optical pulse propagation over on optical phase conjugation in a semiconductor amplifier,” Electron.
10000 km of fibre using recirculating loop,” Electron. Lett., vol. 27, no. Lett., vol. 30, no. 9, pp. 724–725, 1992.
2, pp. 120–121, 1991. [63] S. Watanabe and M. Shirasaki, “Exact compensation for both chromatic
[42] H. J. Thiele, R. I. Killey, and P. Bayvel, “Influence of fibre dispersion dispersion and Kerr effect in a transmission fiber using optical phase
and bit rate on cross-phase-modulation-induced distortion in amplified conjugation,” J. Lightw. Technol., vol. 14, no. 3, pp. 243–248, Mar.
optical fibre links,” Electron. Lett., vol. 34, no. 21, pp. 2050–2051, 1996.
1998. [64] K. Roberts, C. Li, L. Strawczynski, M. O’Sullivan, and I. Hardcastle,
[43] A. D. Ellis and W. A. Stallard, “Four wave mixing in ultra long trans- “Electronic precompensation of optical nonlinearity,” IEEE Photon.
mission systems incorporating linear amplifiers,” IEE Colloquium, No. Technol. Lett., vol. 18, no. 2, pp. 403–405, 2006.
159, 1990. [65] J. B. Stark, P. P. Mitra, and A. Sengupta, “Information capacity for
[44] R.-J. Essiambre, B. Mikkelsen, and G. Raybon, “Intra-channel cross- optical communication channels,” Opt. Fiber Technol., vol. 7, pp. 275–,
phase modulation and four-wave mixing in high-speed TDM systems,” 2001.
Electron. Lett., vol. 35, no. 18, pp. 1576–1578, 1999. [66] B. Wu and E. Narimanov, “Information capacity of nonlinear fiber-
[45] A. D. Ellis, J. D. Cox, D. Bird, J. Regnault, J. V. Wright, and W. A. optical systems,” in Proc. 2005 Quantum Electron. Laser Sci. Conf.
Stallard, “5 Gbit/s soliton propagation over 350 km with large periodic (QELS), 2005, Paper JThE74.
dispersion coefficient perturbations using erbium doped fibre amplifier [67] A. Mecozzi and M. Shtaif, “On the capacity of intensity modulated
repeaters,” Electron. Lett., vol. 27, no. 10, pp. 878–, 1991. systems using optical amplifiers,” IEEE Photon. Technol. Lett., vol. 13,
[46] I. Morita, K. Tanaka, N. Edagawa, and M. Suzuki, “Impact of the dis- pp. 1029–1031, 2001.
persion map on long-haul 40 Gbith single-channel soliton transmission [68] R.-J. Essiambre, G. J. Foschini, P. J. Winzer, G. Kramer, and E. C. Bur-
with periodic dispersion compensation,” in Proc. OFC’99, San Diego, rows, “The capacity of fiber-optic communication systems,” in Proc.
1999, Paper FD1. OFC2008, San Diego, 2008, paper OTuE1.
[47] P. V. Mamyshev and L. F. Mollenauer, “Pseudo-phase-matched [69] H. Goto, M. Yoshida, T. Omiya, K. Kasai, and M. Nakazawa, “Po-
four-wave mixing in soliton wavelength-division multiplexing trans- larisation and frequency division multiplexed 1G symbol/s, 64 QAM
mission,” Opt. Lett., vol. 21, no. 6, pp. 396–398, 1996. coherent optical transmission with 8.6 b/s/Hz spectral efficiency over
[48] N. J. Smith and N. J. Doran, “Modulational instabilities in fibers 160 km,” IEICE Electron. Exp., vol. 5, no. 18, pp. 776–781, 2008.
with periodic dispersion management,” Opt. Lett., vol. 21, no. 8, pp. [70] S. L. Jansen, D. van den Borne, C. Climent, M. Serbay, C.-J. Weiske, H.
570–572, Apr. 1996. Suche, P. M. Krummrich, S. Spalter, S. Calabro, N. Hecker-Denschlag,
[49] E. Pincemin, A. Tan, A. Bezard, A. Tonello, S. Wabnitz, J.-D. Ania- P. Leisching, W. Rosenkranz, W. Sohler, G. D. Khoe, T. Koonen, and
Castañòn, and S. Turitsyn, “Robustness of 40 Gb/s ASK modulation 2 2
H. de Waardt, “10 200 km 22 2 l0 Gbit/s RZ-DQPSK dense WDM
formats in the practical system infrastructure,” Opt. Exp., vol. 14, no. transmission without inline dispersion compensation through optical
25, pp. 12049–12062, 2006. phase conjugation,” in Proc. OFC’05, Anaheim, 2005, Paper PDP28.
[50] P. P. Mitra and J. B. Stark, “Nonlinear limits to the information capacity [71] L. Becouarn, G. Vareille, P. Pecci, and J. F. Marcerou, “3 Tbit/s trans-
of optical fibre communications,” Nature, vol. 411, pp. 1027–1030, mission (301 DPSK channels at 10.709 Gb/s) over 10270 km with a
2001. record efficiency of 0.65(bit/s)/Hz,” in Proc. ECOC 03, Rimini, 2003,
[51] L. G. L. Wegener, B. M. L. Povinelli, A. G. Green, P. P. Mitra, J. Post-deadline Paper Th4.3.2.
B. Stark, and P. B. Littlewood, “The effect of propagation nonlineari- [72] Y. Kodama and S. Wabnitz, “Reduction of soliton interaction forces
ties on the information capacity of WDM optical fiber systems: Cross- by bandwidth limited amplification,” Electron. Lett., vol. 27, no. 21,
phase modulation and four-wave mixing,” Physica D: Nonlinear Phe- pp. 1931–1932, 1991.
nomena, vol. 189, no. 1–2, pp. 81–99, 2004. [73] J. Zhao, A. D. Ellis, F. C. G. Gunning, P. Frascella, S. K. Ibrahim, and
[52] R. J. Essiambre, G. J. Foschini, P. J. Winzer, and G. Kramer, “Ex- R. Weerasuriya, “High-capacity transmission using coherent WDM
ploring capacity limits of fibre-optic mommunication systems,” in (invited),” OECC 2009 Hong Kong, 2009.
Proc. ECOC 2008, Brussels, 2008, Paper We1E1,. [74] M. Seimetz, L. Molle, D.-D. Gross, B. Auth, and R. Freund, “Co-
[53] A. D. Ellis, S. J. Pycock, D. A. Cleland, and C. H. F. Sturrock, “Disper- herent RZ-8 PSK transmission at 30 Gbit/s over 1200 km employing
sion compensation in a 450 km transmission system employing stan- homodyne detection with digital carrier phase estimation,” in Proc.
dard fibre,” Electron. Lett., vol. 28, no. 10, p. 954, 1992. ECOC’07, Berlin, 2007, Paper We8.3.4.
[54] D. McGhan, C. Laperle, A. Savehenko, L. Chuandong, G. Mak, and [75] A. H. Gnauck, G. Charlet, P. Tran, P. Winzer, C. Doerr, J. Centanni,
M. O’Sullivan, “5120 km RZ-DPSK transmission over G652 fiber at E. Burrows, T. Kawanishi, T. Sakamoto, and K. Higuma, “25.6-Tb/s
10 Gb/s with no optical dispersion compensation,” in Proc. OFC 2005, C+L-band transmission of polarization-multiplexed RZ-DQPSK sig-
Anaheim, 2005, paper PDP27. nals,” in Proc. OFC 2007, Anaheim, 2007, Paper PDP19.
[55] J. H. Winters and R. D. Gitlin, “Electronic signal processing techniques [76] M. Yoshida, H. Goto, K. Kasai, and M. Nakazawa, “64 and 128 co-
in long haul fiber optic systems,” IEEE Trans. Commun., vol. 38, no. herent QAM optical transmission over 150 km using frequency-sta-
9, pp. 1439–1453, 1990. bilized laser and heterodyne PLL detection,” Opt. Exp., vol. 16, pp.
[56] M. E. McCarthy, J. Zhao, A. D. Ellis, and P. Gunning, “Full-field elec- 829–840, 2008.
tronic dispersion compensation of 10 Gbit/s OOK without Raman am- 2
[77] I. Morita and N. Edagawa, “50 GHz-spaced 64 42.7 Gbit/s transmis-
2
plification over 4 124 km field-installed SSF,” J. Lightw. Technol., sion over 8200 km using pre-filtered CS-RZ DPSK signal and EDFA
vol. 27, no. 23, pp. 5327–5335, Dec. 2009. Repeaters,” in Proc. ECOC’03, Rimini, 2003, Paper Th4.3.1.
[57] M. G. Taylor, “Coherent detection method using DSP for demodulation [78] A. J. Lowery, “Fiber nonlinearity pre- and post-compensation for long-
of signal and subsequent equalization of propagation impairments,” haul optical links using OFDM,” Opt. Exp., vol. 15, no. 20, pp. 12965–,
IEEE Photon. Technol. Lett., vol. 16, no. 2, pp. 674–676, 2004. 2007.

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.
 ELLIS et al.: APPROACHING THE NON-LINEAR SHANNON LIMIT 433

[79] E. Yamazaki, F. Inuzuka, K. Yonenaga, A. Takada, and M. Koga, Andrew Ellis was born in Underwood, U.K., in 1965. He received the B.Sc.
“Compensation of interchannel crosstalk induced by optical fiber degree in physics with a minor in mathematics from the University of Sussex,
nonlinearity in carrier phase-locked WDM system,” IEEE Photon. Brighton, U.K., in 1987. He received the Ph.D. degree in electronic and elec-
Technol. Lett., vol. 19, no. 1, pp. 9–11, Jan. 2007. trical engineering from The University of Aston in Birmingham, Birmingham,
[80] L. Thylen, E. Berglind, and O. Nilsson, “Channel capacity of optical U.K., in 1997 for his study on all optical networking beyond 10 Gbit/s.
fibres,” Private Communication, Jun. 2002. He previously worked for British Telecom Research Laboratories as a Se-
[81] H. A. Haus and Y. Yamamoto, “Quantum circuit theory of phase sen- nior Research Engineer investigating the use of optical amplifiers and advanced
sitive linear systems,” IEEE J. Quantum Electron., vol. QE-23, no. 2, modulation formats in optical networks and the Corning Research Centre as a
Senior Research Fellow where he led activities in optical component character-
pp. 212–221, Feb. 1987.
isation. Currently, he heads the Transmission and Sensors Group at the Tyndall
[82] S. Oda, H. Sunnerud, and P. A. Andrekson, “High efficiency and high
National Institute in Cork, Ireland, where he is also a member of the Department
output power fiber-optic parametric amplifier,” Opt. Lett., vol. 32, no.
of Physics, University College Cork. He research interests include the evolution
13, pp. 1776–1778, 2007. of core and metro networks, and the application of photonics to sensing. He has
[83] J. M. C. Boggio, C. Lundström, J. Yang, H. Sunnerud, and P. A. An- published over 100 journal papers and over 20 patents in the field of photonics.
drekson, “Double-pumped FOPA with 40 dB flat gain over 81 nm band- Dr. Ellis is a member of the Institute of Physics and the Institute of Engi-
width,” in Proc. ECOC 2008, Brussels, 2008, Paper Tu.3B5. neering Technology, and is a Chartered Physicist. He acts as a reviewer for IEEE
[84] G. Charlet, M. Salsi, H. Mardoyan, P. Tran, J. Renaudier, S. Bigo, M. JOURNAL OF LIGHTWAVE TECHNOLOGY and IEEE PHOTONICS TECHNOLOGY
Astruc, P. Sillard, L. Provost, and F. Cérou, “Transmission of 81 chan- LETTERS.
nels at 40 Gbit/s over a transpacific-distance erbium-only link, using
PDM-BPSK modulation, coherent detection, and a new large effective
area fibre,” in Proc. ECOC’08, Brussels, 2008, Paper Th3E3.
[85] P. J. Roberts, F. Couny, H. Sabert, B. Mangan, D. Williams, L. Farr, Jian Zhao received the B.Eng. degree from University of Science and Tech-
M. Mason, A. Tomlinson, T. Birks, J. Knight, and P. St. J. Russell, nology of China (USTC) in 2002, and the M.Phil. and Ph.D. degrees from the
“Ultimate low loss of hollow-core photonic crystal fibres,” Opt. Exp., Chinese University of Hong Kong (CUHK) in 2004 and 2007, respectively.
vol. 13, pp. 236–244, 2005. During 2002–2004, he worked on supercontinuum-based photonic devices
[86] K. Tajima, J. Zhou, K. Nakajima, and K. Kurokawa, “Ultra low loss for multi-wavelength optical networks. In his Ph.D. program (2004–2007), his
and long length photonic crystal fiber,” in Proc. OFC’04, Los Angeles, research interests included electronic signal processing and advanced modula-
2004, Paper WI7. tion formats in optical communication and wavelength-division multiplexing
[87] R. M. Percival, D. Szebesta, C. P. Seltzer, S. D. Perin, S. T. Devey, and passive optical networks. He joined the Photonic Systems Group at the Tyn-
M. Louka, “A 1.6- m pumped 1.9- m thulium-doped fluoride fiber dall National Institute as a Postdoctoral Research Scientist in August 2007. His
laser and amplifier of very high efficiency,” IEEE J. Quantum Electron., current research interests include electronic signal processing in optical com-
vol. 31, no. 3, pp. 489–493, Mar. 1995. munications and spectrally-efficient multi-carrier optical transmission systems.
[88] A. Krier and Y. Mao, “High performance uncooled InAsSbP/InGaAs He has published more than 30 technical papers in peer-reviewed international
photodiodes for the 1.8–3.4 m wavelength range,” Infrared Phys.
journals and conferences and 1 patent.
Dr. Zhao is an Enterprise Ireland principal investigator. He was the recipient
Technol., vol. 38, no. 7, pp. 397–403, 1997.
of the First Prize of Outstanding Student Scholarship of USTC.
[89] Z. Tong, Q. Yang, Y. Ma, and W. Shieh, “21.4 Gb/s coherent optical
OFDM transmission over 200 km multimode fiber,” in Proc. OECC/
ACOFT 2008, Syndey, 2008, Paper PDP5.
[90] M. Bigot-Astruc, F. Gooijer, N. Montaigne, and P. Sillard, “Trench-
assisted profiles for large-effective-area single-mode fibers,” in Proc. David Cotter graduated from Southampton University, Southampton, U. K.,
and received the Ph.D. degree in non-linear optics in 1976. He was awarded a
ECOC’08, Brussels, 2008, Paper Mo.4.B1.
D.Sc. degree by Southampton University in 1996 for contributions to the field
[91] C. Rasmussen, T. Fjelde, J. Bennike, L. Fenghai, S. Dey, B. Mikkelsen,
of non-linear optics and its applications in quantum electronics, semiconductor
P. Mamyshev, P. Serbe, P. van der Wagt, Y. Akasaka, D. Harris, D. physics and telecommunications.
Gapontsev, V. Ivshin, and P. Reeves-Hall, “DWDM 40 G transmis- He heads the Photonic Systems Group at the Tyndall National Institute in
sion over trans-pacific distance (10 000 km) using CSRZ-DPSK, en- Cork, Ireland, where he is also a member of the Department of Physics, Univer-
hanced FEC, and all-Raman-amplified 100-km ultrawave fiber spans,” sity College Cork. Formerly, he was with Corning Research Centre, Ipswich, U.
J. Lightw. Technol., vol. 21, no. 1, pp. 203–207, 2004. K., British Telecom Research Laboratories in Ipswich, U. K., the Max-Planck
[92] S. Ten, “Advanced fibers for submarine and long-haul applications,” in Institute for Biophysical Chemistry in Goettingen, Germany, and the Atomic
Proc. LEOS 2004, San Francisco, CA, vol. 2, pp. 543–544, paper WJ2. Energy Research Establishment in Harwell, U. K.
[93] C. K. Ramaiah, “Bulletin board systems for libraries,” DESIDOC Bull. Prof. Cotter is a Fellow of the Institute of Physics, a Chartered Physicist and
Inf. Technol., vol. 15, no. 4, pp. 23–31, 1995. Chartered Engineer, U.K.

Authorized licensed use limited to: NATIONAL INSTITUTE OF TECHNOLOGY SRINAGAR. Downloaded on January 01,2022 at 18:10:27 UTC from IEEE Xplore. Restrictions apply.