Jordan University of Science and Technology

Faculty of Engineering
Electrical Engineering Department

Digital Signal Processing for Communications

(EE 768)

Bandwidth-Variable Transceivers based on

Four-Dimensional Modulation Formats
Johannes Karl Fischer, Member, IEEE, Saleem Alreesh, Student Member, IEEE, Robert
Elschner, Member, IEEE,Felix Frey, Markus N¨olle, Carsten Schmidt-Langhorst, and Colja
Schubert.

Majd Shakhaterh
128983
Contents

Abstract 1

List of Acronyms 2

List of Figure 3

List of Table 3

1. Introduction 4

1.1 Related Work 5

1.1.1 Previous work 5

1.1.2 Following work 8

2. FOUR-DIMENSIONAL MODULATION FORMAT 9

3. BVT- ARCHITECTURE 13

4. Future work 17

5. CONCLUSION 17

6. References 18
Abstract

The authors discuss technology options for “bandwidth-variable transceivers”

which are key components for the realization of flexible software-defined optical
networking.

They focus on recent advances in four-dimensional modulation formats, in

modulation format transparent data-aided digital signal processing and the cost
of hardware implementation.

Bandwidth-variable transceivers enable the software controlled adaptation of

physical layer parameters such as transmitted bit rate, spectral efficiency and
transparent reach according to the traffic demands at hand.

1|Page
List of Acronyms
6PolSK 6-ary polarization-shift keying QPSK
ADC analog to digital converter
AWGN additive white Gaussian noise
BER bit error rate
BPS bits per symbol
BPSK binary phase shift keying
BVT Bandwidth-variable transceivers
CM coded modulation
CAZAC constant-amplitude zero-autocorrelation
DAC digital-to-analog converter
DP-QPSK Dual-polarised Quadrature Phase-Shift Keying
DSP Digital Signal Processing
DP I/Q dual-polarization in-phase and quadrature
EON elastic optical networking
FEC forward error correction
4D four-dimensional
FDE frequency-domain equalizer
GN Gaussian noise
HD hard-decision
LDPC low density parity check
PDM-mQAM polarization-division multiplexed m-ary quadrature
amplitude modulation

QAM Quadrature Amplitude Modulation

QPSK Quadrature Phase-Shift Keying
ROSNR required optical signal-to-noise ratio
RS Reed–Solomon
SD soft-decision
SDN software defined networking
SE spectral efficiency
SNR signal-to-noise ratio
SSMF standard single-mode fiber
TDE time-domain equalizer
WDM wavelength-division multiplexed
2|Page
List of Figure

Fig. 1. Constellation constrained capacity of selected. 10

Fig. 2. BER as a function of SNR per symbol. 10

Fig. 3. reach versus net spectral efficiency for Nyquist WDM transmission. 12

Fig. 4. Reach as a function of signal power spectral density. 13

Fig. 5. Bandwidth-variable coherent transceiver based on modulation 15

format flexible data-aided DSP.

Fig.6. Measured Q-factor as a function of transmitted distance. 17

Fig.7. Simulated reach as a function of launch power per channel. 18

List of Table
Table I. It lists the name of the modulation format together with the respective number of
encoded bits per symbol, the required optical signal-to-noise ratio (ROSNR) for an AWGN
channel considering two different FEC scenarios and the resulting net bitrate. 11

3|Page
1.Introduction

BANDWIDTH-VARIABLE transceivers (BVT) are a key enabling technology for

elastic optical networking (EON) [1] and software defined networking [2].

These new networking paradigms require transceivers which operate on a flexible

wavelength grid and which can accommodate traffic needs by flexibly adapting
spectral efficiency, bit rate and reach.

The capabilities and costs of BVTs are defined by the architecture which is
employed to realize the necessary flexibility [3]. the choice of architecture will
also determine the most suitable technology to implement BVTs.

The authors assume that a cost-efficient realization of a BVT operates at a fixed

symbol rate.

4|Page
1.1Related Work

1.1.1Previous work:

The authors discussed the previous work for this paper, there are exist several
approaches to realize flexible variation of the three key parameters of a
transceiver:

A. Adaptive Size of the Modulation Alphabet

Its objective is to have granularity of the reach of data by controlling the
trade-off between Spectral Efficiency (SE) and Bit Rate (BR). And this can
be achieved by employing polarization-division multiplexed m-ary
quadrature amplitude modulation (PDM-mQAM). Where m-ary is
varying size of modulation alphabet [4]– [5], and the variation of m
always results in an even number of encoded bits per symbol. The
resulting modulation formats are referred to as time-domain hybrid
QAM.

B. Rate-Adaptive Coding
Rate Adaptive schemas for optical communication are used to request a
specific reach by using Forward Error Correction (FEC) codes [6]– [10]. In
such schemes, the FEC code rate and m-ary size of the modulation
alphabet are adjusted to optimally support the desired net bit rate over
a requested reach.

5|Page
 Various rate adaptive schemes for optical communications are
discussed, each with a different trade-off between performance and
implementation complexity:

1. A rate-adaptive scheme based on Hard Decision (HD) decoding

uses concatenated Reed–Solomon (RS) codes at a fixed symbol rate and fixed
[7] or adaptive [8] size of the modulation alphabet. By varying the
modulation format between PDM-4QAM, PDM-8QAM and PDM-16QAM,
and uses signal-to-noise Ratio (SNR) per symbol ranging from below 0 dB up
to 16 dB to achieve net bit rate from below 20 Gb/s to 200 Gb/s; respectively.

2. A rate-adaptive scheme based on Soft Decision (HD) decoding

The scheme was extension to Hard Decision Decoding, where Combines the
inner low-density parity check (LDPC) code and outer RS code [9]. This improves
SNR per symbol to −3 dB and up to below 14 dB at net bit rates of 20 Gb/s and
200 Gb/s, respectively.

3. A rate-adaptive scheme based on Non-binary LDPC coded

modulation (CM)

By using 6 different code rates and PDM-4QAM, PDM-8QAM and PDM-

16QAM constellations at a fixed symbol rate of 40 GBd, it allows variation of
the net bit rate and SE from 107 Gb/s to 293 Gb/s and 2.67 bit/symbol to 7.33
bit/symbol, respectively [10].

C. Four-Dimensional Modulation and Coding

To achieve a finer granularity by using time-domain hybrid QAM and the
application of alternative modulation formats in four-dimensional (4D)
signal space.

6|Page
 Theoretically, limits small bit error ratio (BER) which increased spectral
efficiency (SE) and power efficiency compared to PDM-mQAM signals
[11], [12].

Four-Dimensional (4D) signal formats such as:

• mSP-QAM (m-ary set partitioned QAM), where m is the size of the modulation
alphabet in four dimensions.

we can have derived regular 4D constellations based on cubic four-

dimensional lattices from standard PDM-mQAM by applying Ungerb¨ock’s
set partitioning scheme in four dimensions [13].so, they are ideally suited for
simple implementation using standard commercial off-the-shelf optical
components such as: integrated dual-polarization in-phase and quadrature
(DP I/Q) modulators and polarization-diversity coherent receivers.

This modulation formats encode odd numbers of bits per symbol and thus
enable a granularity of 1 bit/symbol for flexible adaptation of SE, bit rate and
reach when combined with PDM-mQAM.

• 6Po1SK (6-ary polarization-shift keying) QPSK.

is generated by modulating six different states of polarization (±S1, ±S2

and ±S3 in Stokes space) with quadrature phase-shift keying (QPSK) signals
[14]. This modulation format encodes 4.5 bit/symbol and theoretically
shows better asymptotic power efficiency compared to PDM-QPSK [11].

• Four-Dimensional Modulation Format Experiments.

several 4D modulation formats have already been demonstrated

experimentally.

For instance, by using the polarization-switched quadrature phase-shift

keying (PS-QPSK) format, record sensitivity at a bit error ratio (BER) of 10-3
was demonstrated for signals with bit rates of 84 Gb/s and 112 Gb/s [9].

7|Page
 Further experimental demonstrations of 4D modulation formats include 6-
ary polarization-shift keying (6PolSK) QPSK [10] at 126 Gb/s and m-ary set
partitioned (mSP) QAM with m=32 and m=128 at 140 Gb/s and 196 Gb/s,
respectively [11].
Impressive record capacity × distance and spectral efficiency × distance
products for transmission over single-mode fiber were achieved by
interleaving PDM-16QAM symbols and 128SP-QAM symbols in the time
domain (like the time-domain hybrid QAM format) [12], [13]. Furthermore,
these results were enabled by applying a maximum a-posteriori (MAP)
decoder in combination with an FEC base d on low-density parity check
(LDPC) codes and iterative decoding.

1.1.2Following work:

the following work starting by motivating the transition from fixed grid networks
to elastic optical networks. Once the fixed spectrum grid is broken down, many new
technologies that are just now crossing the boundary between research and
product development will enable a much more flexible network.

This process has already started with elastic reconfigurable optical add-drop
multiplexers (ROADMs), which will not carry significant extra cost and will enable
the EON. Then transceivers will become more flexible, fueled by the need to
support both high and low demands depending on the required reach.

These transceivers may eventually allow for adaptive use of resources, flexible use
of spectrum, and a flexible relationship between client technologies such as IP and
the optical layer.

8|Page
2.FOUR-DIMENSIONAL MODULATION FORMATS

in this section the authors are comparing of several 4D modulation formats to

regular PDM-mQAM formats with quadratic (m = 16, 64) and cross (m = 32)
constellations.

We have twelve bits were mapped on the: PS-QPSK, PS-BPSK ,128SP-QAM, 32SP-
QAM,6PolSK-QPSK, PDM-8QAM, PDM-16QAM, PDM-32QAM, PDM-64QAM, PDM-
QPSK,512SP-QAM and 2048SP-QAM.

The constellation constrained capacity for an additive white Gaussian noise

(AWGN) channel can be determined by calculating the mutual information
between transmitted and received symbols. For a discrete-input complex-output
channel with modulation alphabet χ, the mutual information can be expressed as
[41], where X is the transmitted signal field and Y is the received noisy signal field

𝑃(Y|X)(𝑦|𝑎)
I (X; Y) = ∑𝑎𝜖𝑋 𝑃X(𝑎) ∫ 𝑃(Y|X)(𝑦|𝑎) log 2 𝑑𝑦
𝑃Y (𝑦)

where X is the transmitted signal field and Y is the received noisy signal field,
which are both four-dimensional in our case.
The mutual information is then determined by four- dimensional integration over
the four quadrature components of the received optical field.

9|Page
 Fig. 1. Constellation constrained capacity of selected polarization multiplexed
and four-dimensional modulation formats for an AWGN channel.

Fig. 1 shows the resulting constellation constrained capacity as a function of the

SNR per symbol, where Es is the average energy per 4D symbol and N0 is the one-
sided noise power spectral density.

Fig. 2. BER as a function of SNR per symbol for selected polarization multiplexed
and four-dimensional modulation formats.

Fig. 2 shows the BER obtained from Monte Carlo simulations as a function of SNR
per symbol for an AWGN channel . For the 4D modulation formats, symbol decision
at the receiver was performed based on the minimum Euclidean distance in 4D

10 | P a g e
space. It can be observed that the 4D modulation formats significantly improve the
sensitivity at low BERs.

For instance, at a BER of 5 × 10−5 the PS-QPSK format shows almost the same
sensitivity as PDM-BPSK although it encodes one more bit per symbol.

However, contemporary FEC operates on significantly larger BER values with

thresholds1 of BER = 3.8 × 10−3 for HD FEC with 7% overhead [42] and BER = 2× 10−2
for soft-decision FEC with 23% overhead. Unfortunately, compared to PDM-mQAM
formats, the performance gain of the considered 4D modulation formats degrades
with decreasing SNR, i.e. at a BER of 10−1 the mSP-QAM formats show the same
sensitivity as the respective higher-order PDM-mQAM formats, while encoding one
bit less per symbol. This loss of performance in the low-SNR regime is due to the
non-Gray mapping of the 4Dformats and the resulting burst errors.

Table I. It lists the name of the modulation

format together with the respective number
of encoded bits per symbol, the required
optical signal-to-noise ratio (ROSNR) for an
AWGN channel considering two different FEC
28GBd and 32GBd, and the resulting net
bitrate.

By choosing the twelve-modulation format

and keeping the symbol rate constant, the
gross SE can be tuned between 2 bit/symbol
and 12 bit/symbol, resulting in net bit rates
between 50 Gb/s and 300 Gb/s and a range in
ROSNR spanning more than 15 dB.

1
The FEC threshold is defined as the pre-FEC BER where the signal after FEC is considered as error free. Commonly,
a post-FEC BER < 10−15 is considered as “error free.”

11 | P a g e
 Fig. 3. (a) a symbol rate of 28 GBd assuming HD-FEC with and BER = 3.8 × 10−3 threshold, (b) a symbol
rate of 32 GBd assuming SDFEC with and BER = 2 × 10−2 threshold.

the figure above reach versus net spectral efficiency for Nyquist WDM
transmission of selected polarization-multiplexed and four-dimensional
modulation formats over SSMF at two different BER.

The first one shows in fig.3 (a) the reach at optimum launch power into the fiber
as obtained by the GN model when assuming a symbol rate of 28 GBd and a HD-
FEC threshold of a BER of 3.8 × 10−3. The reach varies between 15120 km and 800
km at gross SEs of (2-12) bit/s/Hz, respectively. The solid black lines denote the
minimum and maximum SE× reach products.

12 | P a g e
The second one in fig.3 (b) The reach for a symbol rate of 32 GBd and a SD-FEC
threshold of 2 × 10−2. Due to the lower ROSNR (cp. Table I) the reach is increased
compared to Fig. 3(a). It is noticeable that the 4D modulation formats do not gain
as much reach as the PDM-mQAM formats.

The GN model has been verified for many system configurations both by means of
numerical simulations [43] and by transmission experiments [44].

to verify that the GN model also yields reliable analytical results in the case of 4D
modulation formats, the authors conducted a set of numerical simulations
considering Nyquist WDM transmission over a dispersion uncompensated standard
single-mode fiber (SSMF) link.

Fig. 4. Reach as a function of signal power spectral density for Nyquist WDM transmission of selected
polarization-multiplexed (red) and four-dimensional (blue) modulation formats at a threshold of BER =
10−3. Shown are results of the analytical Gaussian noise (GN) model (solid lines) and numerical
simulations (symbols) for the twelve modulation formats.

Fig. 3 shows the achieved reach at a BER of 10−3. Symbols denote simulation
results and solid lines are obtained from the GN model. For all considered
modulation formats, the results show that the GN model provides a good

13 | P a g e
estimate of the reach, hence verifying the application of the GN model to reach
predictions for transmission of the considered 4D modulation formats over
dispersion uncompensated SSMF links.

For the considered 4D modulation formats, the results also verify the assertion
that in the Nyquist WDM case, the optimum signal power spectral density is
independent of modulation format and symbol rate [45].

14 | P a g e
3.BVT- ARCHITECTURE

Fig. 4: Bandwidth-variable coherent transceiver based on modulation format flexible data-aided DSP. ASIC:
application-specific integrated circuit, ADC: analog-to-digital converter, BPD: balanced photo-detetctor, CD:
chromatic dispersion, DAC: digital-to-analog converter, DP: dual-polarization, LS: laser source, PBS:
polarization beam splitter, PR: polarization rotator.

Fig. 5 depicts a BVT consisting of the DSP in the electronics part and the optical
transmitter and receiver components.

The required optical components are:

• two tunable laser sources for the transmitter side and as local oscillator
(or just a single laser source if independent tuning is not required).
• a DP I/Q modulator.
• a coherent polarization diversity frontend.

and this component commercially available in highly integrated small form factor
packages [46].

The DSP is performed by application specific integrated circuits (ASIC). Due to the
required high-speed interfaces to the digital-to-analog converters (DAC) and
analog-to-digital converters (ADC) these are usually mixed-signal ASICs, which
integrate the DACs and ADCs on the same chip as the DSP functionality. BVTs
require a rate adaptive data input/output interface, which is a challenging design
task.
15 | P a g e
The Transmitter
At the transmitter side, the modulation format is determined after FEC encoding
by the applied bit-to-symbol mapping. For each modulation format supported by
the BVT, a custom mapper is required. Before applying digital pulse shaping
filters, the header sequences required for the data-aided DSP at the receiver are
periodically inserted into the payload signal.

The header contains special sequences to perform different functions, e.g.

Sequences for frame synchronization, for correction of the frequency offset
between signal carrier frequency and local oscillator and for estimation of the
channel [47].

Finally, the signal is predistorted to compensate for non-ideal transfer

characteristics of the DACs, the driver amplifiers and the nonlinear transfer
characteristic of the DP I/Q modulator.

The Receiver
In the receiver side DSP, I/Q imbalances of the optical frontend are corrected first.
After blind compensation of chromatic dispersion (CD), the exact position of the
header is determined by frame synchronization. Several algorithms for frame
synchronization are available, mainly originating from wireless communications
[47].

Exploiting the known training sequences, it is a simple task to compensate the

frequency offset and estimate the channel transfer function to perform
polarization demodulation and compensation of polarization mode dispersion as
well as residual CD. While blind time-domain equalizers require filter update
strategies matched to the specific modulation format [48], the data-aided
approach for channel estimation is inherently independent of the modulation
format of the payload symbols.

16 | P a g e
After clock recovery, the carrier phase must be recovered. Although there exist
algorithms for carrier phase recovery which are suitable for arbitrary PDM-mQAM
formats and most 4D modulation formats [49], [50], the required hardware effort
may be too costly when a large size of the modulation alphabet is considered.

Finally, decision and demapping are performed before FEC decoding. Like the
transmitter, a custom decision and demapping stage is required for each
supported modulation format.

The two following figures show examples for a performance comparison between
blind and data-aided algorithms:

Fig.6. Measured Q-factor as a function of transmitted distance for 20 ×126 Gb/s 6PolSK-QPSK
WDM transmission with blind and data-aided equalization.

Fig. 5 shows experimental results for 28-GBd 6PolSK-QPSK signals [25], [45],
CAZAC sequences with a length of 16 symbols were used [51]. The header was
repeated every 215 payload symbols with a total overhead of 1.17%, the
modulation format transparent data-aided DSP shows similar performance as the
format dependent blind DSP.
17 | P a g e
 Fig. 7. Simulated reach as a function of launch power per channel for 34 × 288 Gb/s PDM-
16QAM Nyquist WDM transmission over the fiber link

Fig. 6 shows a result of a transmission simulation using 36-GBd Nyquist-PDM-

16QAM signals. with same simulation parameters. The channel was assumed to
be static (no polarization drift) and laser phase noise was neglected. For the data-
aided DSP, CAZAC sequences with a length of 64 symbols were used.

The simulated reach with both blind TDE and data-aided FDE lies within ±1 span
of the GN model prediction. This shows that under the idealized assumption of a
dispersive and nonlinear static channel, the transmission reach does not depend
on the actual equalization scheme.

The results indicate that the use of data-aided DSP is a viable solution for a cost-
efficient BVT implementation with good performance for the following reasons:

• The same data-aided algorithm block provides stable performance for all
considered formats, simplifying design and test of the implementation.
• The internal feed-forward structure of the data-aided equalizer enables
straight-forward real-time implementation and minimizes performance
losses due the required hardware parallelization.
• The use of an FDE reduces the hardware complexity in comparison to a
TDE.

18 | P a g e
4.Future work

The authors can be making more experiment on different type of modulation like
8D modulation and using other types of transmission channels to achieve a finer
granularity.

5.CONCLUSION

In this paper the authors discussed the following concepts:

• 4-D modulation formats in the context of BVTs and their application in

flexible networks.
• Adaption of the modulation format and constellation size is a simple and
hardware-efficient scheme for flexibility in bit rate, SE and reach.
• Format-flexible DSP in the coherent receiver is the key enabling technology
for such a scheme.
• The benefits of data-aided DSP for carrier frequency recovery and channel
estimation.
• Data-aided functions combined with a FDE are a promising solution for
resource-efficient implementation of format-flexible DSP.

19 | P a g e
6.References

[1] O. Gerstel, M. Jinno, A. Lord, and S. J. Ben Yoo, “Elastic optical networking: A new dawn for the
optical layer?,” IEEE Commun. Mag., vol. 50, no. 2, pp. S12–S20, Feb. 2012.

[2] S. Gringeri, N. Bitar, and T. J. Xia, “Extending software defined network principles to include
optical transport,” IEEE Commun. Mag., vol. 51, no. 3, pp. 32–40, Mar. 2013.

[3] K. Roberts and C. Laperle, “Flexible transceivers,” presented at the 38th Eur. Conf.Opt.
Commun.,Amsterdam, TheNetherlands, Sep. 2012, PaperWe.3.A.3.

[4] G. Bosco, V. Curri, A. Carena, P. Poggiolini, and F. Forghieri, “On the performance ofNyquist-
WDM terabit superchannels based on PM-BPSK, PM-QPSK, PM-8QAM or PM-16QAM subcarriers,” J.
Lightw. Technol., vol. 29, no. 1, pp. 53–61, Jan. 2011.

[5] P. J. Winzer, “High-spectral-efficiency optical modulation formats,” J. Lightw. Technol., vol. 30,
no. 24, pp. 3824–3835, Dec. 2012.

[6] M. Arabaci, I. B. Djordjevic, R. Saunders, and R. M. Marcoccia,“Polarization-multiplexed rate-

adaptive nonbinary-quasi-cyclic-LDPCcoded multilevel modulation with coherent detection for
optical transport networks,” Opt. Exp., vol. 18, no. 3, pp. 1820–1832, Feb.2010.

[7] G.-H. Gho, L. Klak, and J. M. Kahn, “Rate-adaptive coding for optical fiber transmission systems,”
J. Lightw. Technol., vol. 29, no. 2, pp. 222–233, Jan. 2011.

[8] G.-H. Gho and J. M. Kahn, “Rate-adaptive modulation and coding for optical fiber transmission
systems,” J. Lightw. Technol., vol. 30, no. 12, pp. 1818–1828, Jun. 2012.

[9] G.-H. Gho and M. Kahn, “Rate-adaptive modulation and low-density parity-check coding for
optical fiber transmission systems,” IEEE/OSA J. Opt. Commun. Netw., vol. 4, no. 10, pp. 760–768,
Oct. 2012.

[10] M. Arabaci, I. B. Djordjevic, L. Xu, and T. Wang, “Nonbinary LDPCcoded modulation for rate-
adaptive optical fiber communication without bandwidth expansion,” IEEE Photon. Technol. Lett.,
vol. 24, no. 16, pp. 1402–1404, Aug. 2012.

[11] E. Agrell andM. Karlsson, “Power-efficient modulation formats in coherent transmission

systems,” J. Lightw. Technol., vol. 27, no. 22, pp. 5115–5126, Nov. 2009.

[12] M. Karlsson and E. Agrell, “Spectrally efficient four-dimensional modulation,” presented at the
Opt. Fiber Commun. Conf., Los Angeles, CA,USA, Mar. 2012, Paper OTu2C.1.

[13] L. D. Coelho and N. Hanik, “Global optimization of fiber-optic communication systems using
four-dimensional modulation formats,” presented at the 37th Eur. Conf. Opt. Commun., Geneva,
Switzerland, Sep. 2011,Paper Mo.2.B.4.

20 | P a g e
[14] H. B¨ulow, “Polarization QAM modulation (POL-QAM) for coherent detection schemes,”
presented at the Opt. Fiber Commun. Conf., San Diego, CA, USA, Mar. 2009, Paper OWG2.

[15] M. Sj¨odin, P. Johannisson, H. Wymeersch, P. A. Andrekson, and M. Karlsson, “Comparison of

polarization-switched QPSK and polarization-multiplexed QPSK at 30 Gbit/s,” Opt. Exp., vol. 19, no.
8,pp. 7839–7846, Apr. 2011.

[16] M. Karlsson and E. Agrell, “Which is themost power-efficient modulation format in optical
links?,” Opt. Exp., vol. 17, no. 13, pp. 10814–10819, Jun. 2009.

[17] P. Serena, A. Vanucci, and A. Bononi, “The performance of polarization switched-QPSK (PS-
QPSK) in dispersion managedWDMtransmissions,” presented at the 36th Eur. Conf. Opt. Commun.,
Torino, Italy, Sep. 2010,Paper Th.10.E.2.

[18] L. E. Nelson, X. Zhou, N. Mac Suibhne, A. D. Ellis, and P. Magill,“Experimental comparison of

coherent polarization-switched QPSK to polarization-multiplexed QPSK for 10 × 100 km WDM
transmission,”Opt. Exp., vol. 19, no. 11, pp. 10849–10856, May 2011.

[19] D. S. Millar, D. Lavery, S. Makovejs, C. Behrens, B.C. Thomsen,P. Bayvel, and S. J.

Savory,“Generation and long-haul transmission ofpolarization-switched QPSK at 42.9 Gb/s,” Opt.
Exp., vol. 19, no. 10,pp. 9296–9302, May 2011.

[20] J. K. Fischer, L. Molle, M. N¨olle, D.-D. Groß, and C. Schubert, “Experimental investigation of 28-
GBd polarization-switched quadrature phaseshift keying signals,” presented at the 37th Eur. Conf.
Opt. Commun., Geneva, Switzerland, Sep. 2011, Paper Mo.2.B.1.

[21] J. K. Fischer, L. Molle, M. N¨olle, D.-D. Groß, C. Schmidt-Langhorst,and C. Schubert´,

“Experimental investigation of 84-Gb/s and 112-Gb/s polarization-switched quadrature phase-shift
keying signals,” Opt. Exp., vol. 19, no. 26, pp. B667–B672, Dec. 2011.

[22] M. N¨olle, J. K. Fischer, L. Molle, C. Schmidt-Langhorst, and C. Schubert,“Influence of channel

spacing on ultra long-haul WDM transmission of 8 × 112 Gb/s PS-QPSK signals,” presented at the
Opt. Fiber Commun.Conf., Los Angeles, CA, USA, Mar. 2012, Paper OTh3A.4.

[23] J. K. Fischer, S. Alreesh, R. Elschner, F. Frey, C. Meuer, L. Molle, C. Schmidt-Langhorst, T.

Tanimura, and C. Schubert, “Experimental investigation of 126-Gb/s 6PolSK-QPSK signals,”
presented at the 38th Eur. Conf. Opt. Commun., Amsterdam, The Netherlands, Sep. 2012, Paper
We.1.C.4.

[24] F. Buchali and H. B¨ulow, “Experimental transmission with POLQAM and PS-QPSK modulation
format using a 28-Gbaud 4-D transmitter,” presented at the 8thEur.Conf.Opt.Commun.,Amsterdam,
TheNetherlands, Sep. 2012, Paper We.3.A.1.

[25] J. K. Fischer, S. Alreesh, R. Elschner, F. Frey, C. Meuer, L. Molle, C. Schmidt-Langhorst, T.

Tanimura, and C. Schubert, “Experimental investigation of 126-Gb/s 6PolSK-QPSK signals,” Opt. Exp.,
vol. 20, no. 26,pp. B232–B237, Dec. 2012.

21 | P a g e
[26] H. B¨ulow, T. Rahman, F. Buchali, W. Idler, and W. Kuebart, “Transmission of 4-D modulation
formats at 28-Gbaud,” presented at the Opt. Fiber Commun. Conf., Anaheim, CA, USA, Mar. 2013,
Paper JW2A.39.

[27] T. Tanimura, S. Alreesh, J. K. Fischer, C. Schmidt-Langhorst, F. Frey,C. Meuer, R. Elschner, L.

Molle, and C. Schubert, “Nonlinear transmission of 6PolSK-QPSK signals using coded modulation and
digital back propagation,” presented at the Opt. Fiber Commun. Conf., Anaheim, CA,USA, Mar. 2013,
Paper OTu3B.3.

[28] H. B¨ulow, X. Lu, and L. Schmalen, “Experimental analysis of single carrier POLQAM (6Pol-QPSK)
transmission with soft-decoding,” presented at the 39th Eur. Conf. Opt. Commun., London, UK., Sep.
2013, Paper Mo.4.D.2.

[29] J. Renaudier, O. Bertran-Pardo, A. Ghazisaeidi, P. Tran, H. Mardoyan, P. Brindel, A. Voicila, G.

Charlet, and S. Bigo, “Experimental transmission of Nyquist pulse shaped 4-D coded modulation
using dual polarization 16QAM set-partitioning schemes at 28 Gbaud,” presented at the Opt. Fiber
Commun. Conf., Anaheim, CA, USA, Mar. 2013, Paper OTu3B.1.

[30] H. Zhang, H. G. Batshon, D. G. Foursa, M. Mazurczyk, J.-X. Cai, C. R. Davidson, A. Pilipetskii, G.

Mohs, and N. S. Bergano, “30.58 Tb/s transmission over 7,230 km using PDM half 4D-16QAM coded
modulation with 6.1 b/s/Hz spectral efficiency,” presented at the Opt. Fiber Commun. Conf.,
Anaheim, CA, USA, Mar. 2013, Paper OTu2B.3.

[31] H. Zhang, J.-X. Cai, H. G. Batshon, M. Mazurczyk, O. V. Sinkin,D. G. Foursa, A. Pilipetskii, G.

Mohs, and N. S. Bergano, “200 Gb/s and dual-wavelength 400 Gb/s transmission over transpacific
distance at 6 b/s/Hz spectral efficiency,” presented at the Opt. Fiber Commun. Conf.,Anaheim, CA,
USA, Mar. 2013, Postdeadline Paper PDP5A.6.

[32] R. Rios-M¨uller, J. Renaudier, O. Bertran-Pardo, A. Ghazisaeidi, P. Tran,G. Charlet, and S. Bigo,

“Experimental comparison between hybrid-QPSK/8QAM and 4D-32SP-16QAM formats at 31.2
GBaud using Nyquist pulse shaping,” presented at the 39th Eur. Conf. Opt. Commun.,London, U.K.,
Sep. 2013, Paper Th.2.D.2.

[33] H. Sun, R. Egorov, B. E. Basch, J. McNicol, and K.-T. Wu, “Comparison of twomodulation formats
at spectral efficiency of 5 bits/dual-pol symbol,”presented at the 39th Eur. Conf. Opt. Commun.,
London, U.K., Sep. 2013,Paper Th.2.D.3.

[34] H. G. Batshon, I. Djordjevic, and T. Schmidt, “Ultra high speed optical transmission using
subcarrier-multiplexed four-dimensional LDPC coded modulation,” Opt. Exp., vol. 18, no. 19, pp.
20546–20551, Sep. 2010.

[35] M. Arabaci, I. B. Djordjevic, L. Xu, and T. Wang, “Four-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.
[36] Y. Zhang, M. Arabaci, and I. B. Djordjevic, “Evaluation of fourdimensional nonbinary LDPC-coded
modulation for next-generationlong-haul optical transport networks,” Opt. Exp., vol. 20, no. 8, pp.
9296–9301, Apr. 2012.

22 | P a g e
[37] Y. Zhang, M. Arabaci, and I. B. Djordjevic, “Rate-adaptive fourdimensional nonbinary LDPC-
coded modulation for long-haul optical transport networks,” presented at the Opt. Fiber Commun.
Conf., Los Angeles, CA, USA, Mar. 2012, Paper JW2A.46.

[38] I. B. Djordjevic and Y. Zhang, “Adaptive software-defined coded modulation for ultra-high speed
optical transport,” in Proc. SPIE 8915 Photon.North, Ottawa, ON, Canada, Jun. 2013, pp. 89150E-1–
89150E-12.

[39] L. Beygi, E. Agrell, J. M. Kahn, and M. Karlsson, “Rate-adaptive coded modulation for fiber-optic
communications,” J. Lightw. Technol., vol. 32,no. 2, pp. 333–343, Jan. 2014.

[40] L. Beygi, E. Agrell, J. M. Kahn, and M. Karlsson, “Coded modulation for fiber-optic networks,”
IEEE Signal Process. Mag., vol. 31, no. 2, pp. 93–103, Mar. 2014.

[41] R.-J. Essiambre, G. Kramer, P. J. Winzer, G. J. Foschini, and B. Goebel,“Capacity limits of optical
fiber networks,” J. Lightw. Technol., vol. 28,no. 4, pp. 662–701, Aug. 2010.

[42] International Telecommunication Union, Telecommunication Standardization Sector (ITU-T),

Recommendation G.975.1, Appendix I.9, Feb.2004.

[43] A. Carena, V. Curri, G. Bosco, P. Poggiolini, and F. Forghieri, “Modeling of the impact of
nonlinear propagation effects in uncompensated optical coherent transmission links,” J. Lightw.
Technol., vol. 30, no. 10, pp. 1524–1539, May 2012.

[44] E. Torrengo, R. Cigliutti, G. Bosco, A. Carena, V. Curri, P. Poggiolini, A. Nespola, D. Zeolla, and F.
Forghieri, “Experimental validation of an analytical model for nonlinear propagation in
uncompensated optical links,”Opt. Exp., vol. 19, no. 26, pp. B790–B798, Dec. 2011.

[45] S. Alreesh, J. K. Fischer, P. Wilke-Berenguer, and C. Schubert, “Blind adaptive equalization for
6PolSK-QPSK signals,” presented at the 39th Eur. Conf. Opt. Commun., London, U.K., Sep. 2013,
Paper Mo.4.D.3.

[46] P. Poggiolini, “TheGNmodel of non-linear propagation in uncompensated coherent optical

systems,” J. Lightw. Technol., vol. 30, no. 24, pp. 3875–3879, Dec. 2012.

[47] A. Beling et al., Proc. OFC’11, OML5(2011)

[48] R. Elschner et al., Opt. Express 20, 28786(2012)

[49] D. S. Millar et al., Opt. Express 19, 8533(2011)

[50] T. Pfau et al., J. Lightwave Technol. 27,989 (2009)

[51] X. Zhou, IEEE Photonics Technol. Lett. 22,1051 (2010)

23 | P a g e