Document [original]

AKULTÄT FÜR

ELEKTROTECHNIK,

INFORMATIK UND

MATHEMATIK

Selwan Khaleel Ibrahim

Study of Multilevel Modulation Formats for

High Speed Digital Optical Communication

Systems

Ph.D. Dissertation

Paderborn, July 2007

AKULTÄT FÜR

ELEKTROTECHNIK,

INFORMATIK UND

MATHEMATIK

Study of Multilevel Modulation Formats

for High Speed Digital Optical

Communication Systems

Zur Erlangung des akademischen Grades

DOKTOR-INGENIEUR (Dr.-Ing.)

der Fakultät für Elektrotechnik, Informatik und Mathematik

der Universität Paderborn

vorgelegte Dissertation

von

M.Sc. Selwan Khaleel Ibrahim

Referent: Prof. Dr.-Ing. Reinhold Noé

Korreferent: Prof. Dr.-Ing. Reinhold Häb-Umbach

Tag der mündlichen Prüfung: 12.07.2007

Paderborn, den 16.07.2007

Diss. EIM-E/233

Study of Multilevel Modulation Formats for High

Speed Digital Optical Communication Systems

A Thesis Submitted to the Faculty of Computer Science

Electrical Engineering and Mathematics

University of Paderborn

in Partial Fulfillment of

the Requirements For The Degree of Doktor-Ingenieur

in Electrical Engineering

(Dr.-Ing.)

M.Sc. Selwan Khaleel Ibrahim

Reviewers:

Prof. Dr.-Ing. Reinhold Noé

Prof. Dr.-Ing. Reinhold Häb-Umbach

Date of Thesis Submission: March 1, 2007

Date of Defense Examination: July 12, 2007

Paderborn, Germany

July 2007

Abstract

Spectrally efficient modulation formats can be used to overcome the problems associated

with limited channels and bandwidth of Dense Wavelength Division Multiplexing (DWDM)

optical systems. Multilevel modulation formats are considered spectrally efficient and can

double the transmission capacity by transmitting more information in the amplitude, phase,

polarization or a combination of all. Here we have studied and evaluated different multilevel

modulation formats in a practical optical transmission system.

Differential Quadrature Phase Shift Keying (DQPSK) doubles the transmission rate by

transmitting more information in the phase of the optical carrier signal. DQPSK receivers

require more complex and costly components including delay interferometers and balanced

detection photodiode receivers. Combining DQPSK with Polarization Division Multiplex

(PolDM) would result into 4 times the transmission rate, but a polarization controller would

be needed which results into a more complex receiver. A 20 Gbit/s DQPSK signal can be

generated by combining two 10 Gbit/s Differential Phase Shift Keying (DPSK) signals in

quadrature phases using a QPSK modulator. The receiver sensitivity for 20 Gbit/s NRZ-

and RZ-DQPSK signals are measured to be -36.8 dBm and -38.3 dBm, respectively. The

chromatic dispersion (CD) tolerance for a 1 dB optical signal to noise ratio (OSNR) penalty

for a 20 Gbit/s RZ-DQPSK signal is ~360 ps/nm. When PolDM is used to combine two

20 Gbit/s DQPSK signals, the receiver sensitivity for 40 (2×2×10) Gbit/s NRZ- and RZ-

DQPSK PolDM signals become -33.8 dBm and -34.7 dBm, respectively.

Conventional Quaternary Intensity Modulation (4-IM) doubles the data rate by trans-

mitting more information in the amplitude of the carrier signal. This can be achieved by

modulating the optical amplitude with an electrical 4-level Amplitude Shift Keying (ASK)

signal. A simple receiver consisting of a single photo diode, three decision circuits and a

decoding logic can be used to receive and extract the original transmitted data. It is also

possible to generate quaternary intensity modulation signals by different means, such as

combining two optical Binary/Duobinary NRZ signals with unequal amplitudes in quadra-

ture phases, or orthogonal polarizations. When optically combining two 10 Gbit/s binary

NRZ-ASK signals with unequal amplitudes in quadrature phases using a QPSK modulator

a 20 Gbit/s 4-constellation point quaternary intensity signal is generated (QASK) having the

same bandwidth as a single 10 Gbit/s binary NRZ-ASK signal. The receiver sensitivity and

the 1 dB CD tolerance are measured to be -21.6 dBm and ~130 ps/nm, respectively.

The spectral efficiency of this quaternary intensity modulation signal can be increased

even more by combining two 10 Gbit/s optical duobinary signals with unequal amplitudes

in quadrature phases using a QPSK modulator resulting in a 20 Gbit/s 9-constellation point

Quaternary Duobinary signal (QDB). When combined in orthogonal polarizations using a

Polarization Beam Combiner (PBC), a 20 Gbit/s 9-constellation point Quaternary Polariza-

tion Duobinary signal (QPolDB) is generated. The receiver sensitivity and CD tolerance of

a 20 Gbit/s QDB signal generated using a duobinary stub filter with a frequency response

dip at 5 GHz, are measured to be -21.2 dBm and ~140 ps/nm, respectively. The receiver sen-

sitivity and CD tolerance of a 20 Gbit/s QPolDB signal generated using the same duobinary

stub filter, are measured to be -20.5 dBm and ~340 ps/nm, respectively. When using another

duobinary stub filter with a frequency response dip at 6 GHz, the receiver sensitivity and CD

tolerance of the 20 Gbit/s QPolDB signal became -18.4 dBm and ~530 ps/nm, respectively.

A polarization and phase insensitive direct detection receiver with a single photodiode

has been used to detect all generated quaternary signals as 4-IM signals. The 20 Gbit/s

QDB and QPolDB quaternary signals that we have reported for the first time, have a nar-

row spectrum with a bandwidth similar to that of a single 10 Gbit/s duobinary modulation

signal. These quaternary intensity modulation formats are attractive for DWDM systems

applied for short reach (several kilometers) to medium reach (several hundred kilometers)

transmission applications (metro applications). However, for long or ultra long haul opti-

cal fiber transmission systems, DQPSK and DQPSK PolDM modulation formats featuring

better receiver sensitivities are potential candidates.

Acknowledgments

First, I would like to thank my doctoral adviser Professor Dr.-Ing. Reinhold Noé, for allow-

ing me to work under his supervision in his internationally recognized group. His guidance

and support were essential for me to gain a great amount of experience and knowledge in

the field of optical communication and high frequency engineering. I would also like to

thank Professor Dr.-Ing. Andreas Thiede and Professor Dr. Franz Rammig for acting as my

supervisors, Professor Dr.-Ing. Reinhold Häb-Umbach for acting as second reviewer, and

Professor Dr.-Ing. Klaus Meerkötter and Professor Dr.-Ing. Rolf Schuhmann for being in

my examination committee.

I am also very grateful to the International Graduate School “Dynamic Intelligent Sys-

tems” at the University of Paderborn for providing me a fellowship in addition to financially

supporting me to attend international conferences that were beneficial for my Ph.D. work. I

would like to thank all the staff of the Graduate school for their help and support, especially

Dr. Eckhard Steffen and Astrid Canisius.

My Special thanks goes to all of my colleagues and technical staff at the Optical Com-

munications and High Frequency Engineering group at the University of Paderborn for their

support and encouragement.

Also I would like to thank Dr. Robert Griffin from Bookham, UK for the loan of the

GaAs/AlGaAs oDQPSK modulator and Dr. Henri Porte from Photline Technologies,

France for providing the LiNbO3oDQPSK modulator.

Finally I would like to thank my small and big family including my beloved parents,

wife, child and all of my friends for their patience and support.

List of Ph.D. Publications

1. Selwan K. Ibrahim, S. Bhandare, D. Sandel, A. Hidayat, A. Fauzi, R. Noé, “Low-Cost,

Signed Online Chromatic Dispersion Detection Scheme Applied to a 2×10 Gb/s RZ-

DQPSK Optical Transmission System”, IEE Proceedings Optoelectronics, October

2006, Volume 153, Issue 5, pp. 235-239.

2. Selwan K. Ibrahim, Suhas Bhandare, and Reinhold Noé, “Performance of 20 Gbit/s

Quaternary Intensity Modulation Based on Binary or Duobinary Modulation in Two

Quadratures With Unequal Amplitudes”, IEEE Journal of Selected Topics in Quantum

Electronics, Vol. 12, No. 4, 2006, pp. 596-602.

3. Selwan K. Ibrahim, Suhas Bhandare, and Reinhold Noé, “20 Gbit/s Quaternary In-

tensity Modulation Based on Duobinary Modulation with Unequal Amplitude in Two

Polarizations”, IEEE Photonics Technology Letters, Vol. 18, No. 14, 2006, pp. 1482-

1484.

4. Selwan K. Ibrahim, Suhas Bhandare, Reinhold Noé, “Narrowband 2×10 Gbit/s Qua-

ternary Intensity Modulation Based on Duobinary Modulation in Two Polarizations

with Unequal Amplitudes”, Optical Fiber Communication Conference (OFC 2006),

Anaheim, CA, USA, March 2006, paper OThI2.

5. Selwan K. Ibrahim, S. Bhandare, H. Zhang, R. Noé, “2×10 Gbit/s Quaternary Inten-

sity Modulation Generation using an Optical QPSK modulator”, Proc. of SPIE, Vol.

6021, Paper 602119, (2005).

6. S. K. Ibrahim, S. Bhandare, H. Zhang, R. Noé, “2×10 Gbit/s Quaternary Inten-

sity Modulation Generation using an Optical QPSK modulator”, Asia-Pacific Optical

Communications Conference (APOC 2005), Nov. 2005, Shanghai, China, Session

5b, paper 6021-44.

7. S. K. Ibrahim, S. Bhandare, R. Noé, “Narrowband 20 Gbit/s Quaternary Intensity

Modulation Generated by Duobinary 10 Gbit/s Modulation in 2 Quadratures”, Proc.

31st European Conference on Optical Communication (ECOC 2005), 25-29 Septem-

ber 2005, Glasgow, Scotland, Th2.6.5, Vol. 4, pp. 909-910.

8. Selwan K. Ibrahim, S. Bhandare, D. Sandel, A. Hidayat, A. Fauzi, R. Noé, “Low-Cost,

Signed Online Chromatic Dispersion Detection Scheme Applied to a 2×10 Gb/s RZ-

DQPSK Optical Transmission System”, Conference digest 7th Optical Fibre Mea-

surement Conference (OFMC 2005), 21-23 September 2005, Teddington, UK, pp.

83-86.

viii

9. S. Bhandare, D. Sandel, B. Milivojevic, A. Hidayat, A. Fauzi, H. Zhang, S. K. Ibrahim,

F. Wüst, and R. Noé, “5.94 Tbit/s, 1.49 bit/s/Hz (40×2×2×40 Gbit/s) RZ-DQPSK

Polarization Division Multiplex C-Band Transmission over 324 km”, IEEE Photonics

Technology Letters, Vol. 17, No. 4, 2005, pp. 914-916.

10. S. Bhandare, D. Sandel, B. Milivojevic, A. Hidayat, A. Fauzi, H. Zhang, S. K. Ibrahim,

F. Wüst, R. Noé, “5.94 Tbit/s (40×2×2×40 Gbit/s) C-Band Transmission over 324

km using RZ-DQPSK Combined with Polarization Division Multiplex”, 6th ITG-

Fachtagung "Photonische Netze", Leipzig, Germany, 2.-3. May 2005, pp. 87-90.

List of Figures

2.1 Schematic of an optical direct detection pre-amplified receiver . . . . . . . 6

2.2 Basic digital modulation formats (ASK, PSK, FSK) . . . . . . . . . . . . . 9

2.3 Principle of Binary NRZ-ASK modulation generation using intensity MZM 10

2.4 ASK signal constellation diagram . . . . . . . . . . . . . . . . . . . . . . 10

2.5 Schematic diagram of an optical DPSK transmission setup . . . . . . . . . 11

2.6 Principle of DPSK modulation generation using a MZM . . . . . . . . . . 12

2.7 DPSK signal constellation diagram . . . . . . . . . . . . . . . . . . . . . . 13

2.8 10 Gbit/s optical RZ-DPSK generation scheme using two MZMs with re-

sulting NRZ-DPSK (left) and 50% RZ-DPSK (right) eye diagrams. . . . . . 14

2.9 DPSKreceiver ................................ 14

2.10 Demodulated 10 Gbit/s NRZ-DPSK (a) and RZ-DPSK (b) intensity eye pat-

terns...................................... 15

2.11 Schematic diagram of a standard Duobinary generation and detection scheme

(top) and a corresponding practical optical Duobinary transmission setup

(bottom).................................... 16

2.12 Principle of Duobinary modulation generation using a MZM and a Duobi-

nary encoder based on the conventional one bit delay and add filter . . . . . 18

2.13 10 Gbit/s optical Duobinary modulation generation using a MZM and a

Duobinary encoder based on a 5th order Bessel filter with cutoff frequency

at 2.8 GHz ................................... 19

2.14 Optical Duobinary signal constellation diagram . . . . . . . . . . . . . . . 19

3.1 Schematic of an optical DQPSK transmission system . . . . . . . . . . . . 22

3.2 Schematic of an optical DQPSK modulator . . . . . . . . . . . . . . . . . 23

3.3 DQPSK signal constellation diagram . . . . . . . . . . . . . . . . . . . . . 24

3.4 NRZ-DQPSK signal eye diagram showing the intensity dips . . . . . . . . 25

3.5 Schematic of an optical DQPSK decoder . . . . . . . . . . . . . . . . . . . 26

3.6 Eye diagram of a demodulated data signal from an NRZ-DQPSK signal . . 26

3.7 Experimental 2×10 Gbit/s DQPSK transmission setup . . . . . . . . . . . 28

3.8 Eye diagrams of a demodulated data signal Iand Qextracted from a 2×10

Gbit/s NRZ-DQPSK signal generated using a Photline DQPSK modulator

with PRBS lengths of 27−1(PRBS-7) (top) and 215 −1(PRBS-15) bottom 29

3.9 BER curves for the Iand Qreceived data streams versus received optical

power for a 2×10 Gbit/s NRZ-DQPSK signal using a Photline oDQPSK

Modulator with PRBS lengths of 27−1(PRBS 7) and 215 −1(PRBS 15) . 30

xLIST OF FIGURES

3.10 Average BER versus received optical power for a 20 Gbit/s NRZ-DQPSK

signal using a Photline oDQPSK Modulator with PRBS lengths of 27−1

(PRBS 7) and 215 −1(PRBS15) ...................... 30

3.11 Eye diagrams of a demodulated data signal Iand Qextracted from a 2×10

Gbit/s RZ-DQPSK signal generated using a Photline DQPSK modulator

with PRBS lengths of 27−1(PRBS-7) (top) and 215 −1(PRBS-15) bottom 31

3.12 BER curves for the Iand Qreceived data streams versus received opti-

cal power for a 2×10 Gbit/s RZ-DQPSK signal using a Photline oDQPSK

Modulator with PRBS lengths of 27−1(PRBS 7) and 215 −1(PRBS 15) . 32

3.13 Average BER versus received optical power for a 20 Gbit/s RZ-DQPSK

signal using a Photline oDQPSK Modulator with PRBS lengths of 27−1

(PRBS 7) and 215 −1(PRBS15)....................... 32

3.14 OSNR needed for a BER of 10−9versus CD for a 20 Gbit/s RZ-DQPSK signal 33

3.15 20 Gbit/s RZ-DQPSK signal intensity measurements after transmission over

different lengths of SSMF corresponding to different values of CD, (a) Back-

to-Back (0 ps/nm), (b) 21.3 km (~362 ps/nm), (c) 32.2 km (~547 ps/nm), and

(d)41.5km(~706ps/nm)........................... 34

3.16 2×2×10 Gbit/s RZ-DQPSK PolDM transmission setup . . . . . . . . . . . 35

3.17 Eye diagrams of 2×2×10 Gbit/s NRZ-DQPSK PolDM signal (left), and

RZ-DQPSK PolDM signal (right) . . . . . . . . . . . . . . . . . . . . . . 35

3.18 BER curves for the Iand Qreceived data streams in both polarizations X

and Y versus the received optical power for 2×2×10 Gbit/s NRZ-DQPSK

PolDMsignalwithPRBS7.......................... 36

3.19 Eye diagram of the demodulated data signals Iand Qfor both polarizations

X and Y extracted from a 2×2×10 Gbit/s NRZ-DQPSK PolDM signal with

PRBS7 .................................... 37

3.20 BER curves for the Iand Qreceived data streams in both polarizations X

and Y versus the received optical power for 2×2×10 Gbit/s NRZ-DQPSK

PolDM signal with PRBS 15 . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.21 Average BER versus received optical power for 2×2×10 Gbit/s NRZ-DQPSK

PolDM signal with PRBS 7 and PRBS 15 . . . . . . . . . . . . . . . . . . 38

3.22 BER curves for the Iand Qreceived data streams in both polarizations X

and Y versus the received optical power for 2×2×10 Gbit/s RZ-DQPSK

PolDMsignalwithPRBS7.......................... 39

3.23 Eye diagram of the demodulated data signals Iand Qfor both polarizations

X and Y extracted from a 2×2×10 Gbit/s RZ-DQPSK PolDM signal with

PRBS7 .................................... 39

3.24 Average BER versus received optical power for 2×2×10 Gbit/s RZ-DQPSK

PolDM signal with PRBS 7 and PRBS 15 . . . . . . . . . . . . . . . . . . 40

3.25 2×2×10 Gbit/s DQPSK PolDM transmitter setup photo . . . . . . . . . . 40

3.26 2×2×10 Gbit/s RZ-DQPSK Receiver . . . . . . . . . . . . . . . . . . . . 41

3.27 2×2×10 Gbit/s RZ-DQPSK PolDM transmission setup photo . . . . . . . 41

4.1 Electrical 4-ary ASK signal generation . . . . . . . . . . . . . . . . . . . . 44

4.2 Principle of optical 4-IM generation using intensity MZM . . . . . . . . . . 45

4.3 Optical 4-IM signal constellation diagram . . . . . . . . . . . . . . . . . . 46

LIST OF FIGURES xi

4.4 Schematic of a 4-IM receiver with decoding logic diagram . . . . . . . . . 46

4.5 20 Gbit/s Quaternary intensity modulation transmission setup . . . . . . . . 47

4.6 BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s (4-IM) signal . . . . . . . . . . . . . . 48

4.7 Average BER versus received optical power for a 20 Gbit/s (4-IM) signal . 48

4.8 Optical QASK (2×10 Gbit/s) generation in an optical QPSK modulator, and

resulting quaternary intensity eye diagram (bottom) with electrical attenua-

torsetting(b=1/2) .............................. 50

4.9 QASK electrical field signal constellation diagram (top-right) generated us-

ing a QPSK modulator driven by binary signals with unequal amplitudes.

The resulting intensity eye diagram (bottom-right) shows the 4 intensity lev-

els (a= 1/√2)................................. 50

4.10 Heterodyned electrical spectrum for a 2×10 Gbit/s QASK signal . . . . . . 51

4.11 BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s QASK signal . . . . . . . . . . . . . . 52

4.12 Average BER versus received optical power for a 20 Gbit/s QASK signal . 52

4.13 OSNR needed for a BER of 10−9versus CD for a 20 Gbit/s QASK signal . 53

4.14 Intensity eye diagrams of a 20 Gbit/s QASK signal measured at the sen-

sitivity edge after transmission over different length of SSMF, (a) (Back-

to-Back) 0 km (0 ps/nm), (b) 5.34 km (~90.78 ps/nm), (c) 10.9 km (~185

ps/nm), (d) 16.24 km (~276 ps/nm) . . . . . . . . . . . . . . . . . . . . . . 53

4.15 Principle of optical QDB modulation generation using duobinary low pass

filtering (LPF) and a QPSK modulator (electrical attenuator setting b = 1/2).

The resulting quaternary intensity eye diagram (bottom) and the electrical

duobinary eye diagram (top) are also shown. . . . . . . . . . . . . . . . . . 54

4.16 QDB electrical field signal constellation diagram (top-right) generated using

a QPSK modulator driven by duobinary signals with unequal amplitudes.

The resulting intensity eye diagram (bottom-right) shows the 4 intensity lev-

els (a= 1/√2)................................. 55

4.17 Heterodyned electrical spectrum of the 20 Gbit/s QDB signal . . . . . . . 56

4.18 BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s QDB signal . . . . . . . . . . . . . . . 56

4.19 Average BER versus received optical power for a 20 Gbit/s QDB signal . . 57

4.20 OSNR needed for a BER of 10−9versus CD for a 20 Gbit/s QDB signal . . 58

4.21 Photo of the 2×10 Gbit/s QDB transmitter based on the Bookham QPSK

modulator................................... 58

4.22 Schematic of the optical QPolDB signal generation scheme (a= 1/√2) . . 59

4.23 QPolDB electrical field signal constellation diagram (top-right) generated

using a polarization division multiplex setup (left) with duobinary signals

(a= 1/√2) .................................. 60

4.24 Schematic of the transmission setup used to generate and detect 2×10 Gbit/s

QPolDBsignals ............................... 61

4.25 Heterodyned electrical spectrum of the 20 Gbit/s QPolDB-6 signal . . . . . 62

4.26 BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s QPolDB-5 signal . . . . . . . . . . . . 62

xii LIST OF FIGURES

4.27 BER curves for the three received eyes (top, middle, and bottom) for a 20

Gbit/sQPolDB-6signal............................ 63

4.28 Average BER versus received optical power for 20 Gbit/s QPolDB-5 and

QPolDB-6signals............................... 63

4.29 Average BER versus received optical power for 20 Gbit/s QPolDB-6 signal

withPRBS7,10and15............................ 64

4.30 OSNR needed for a BER of 10−9versus CD in ps/nm for 20 Gbit/s QPolDB-

5andQPolDB-6signals ........................... 65

4.31 Intensity eye diagrams of the 20 Gbit/s QPolDB-5 signal measured at the

sensitivity edge after transmission over (a) 0 km (back-to-back), (b) 10.9

km, (c) 16.24 km, (d) 26.65 km, and (e) 37.55 km of SSMF . . . . . . . . . 66

4.32 Intensity eye diagrams of the 20 Gbit/s QPolDB-6 signal measured at the

sensitivity edge after transmission over (a) 0 km (back-to-back), (b) 10.9

km, (c) 16.24 km, (d) 26.65 km, (e) 37.55 km, and (f) 41.54 km of SSMF . 67

4.33 Photo of the PolDM transmitter setup used to generate the QPolDB signals 67

4.34 Schematic of the optical QPolASK signal generation scheme (a= 1/√2),

eye diagram of 2×10 Gbit/s QPolASK signal (top-right). . . . . . . . . . . 68

4.35 QPolASK electrical field signal constellation diagram (top-right) generated

using a polarization division multiplex setup with binary signals (a= 1/√2) 69

4.36 Schematic of quaternary DP-ASK transmitter and receiver . . . . . . . . . 70

4.37 DP-ASK signal constellation diagram . . . . . . . . . . . . . . . . . . . . 70

5.1 Average BER curves versus received optical power for 20 Gbit/s quaternary

modulation signals (4-IM, QASK, QDB, QPolDB-5, and QPolDB-6). . . . 72

5.2 OSNR needed for a BER of 10-9 versus CD in ps/nm for 20 Gbit/s quater-

nary modulation signals (QASK, QDB, QPolDB-5, and QPolDB-6). . . . . 73

A.1 Schematic drawing of a Mach-Zehnder interferometer . . . . . . . . . . . . 75

A.2 Mach-Zehnder modulator (push-pull) structure using x−cut LiNbO3. . . 77

A.3 Typical transfer characteristic curve of a Mach-Zehnder modulator . . . . . 77

B.1 Schematic of a DQPSK transmission system . . . . . . . . . . . . . . . . . 79

B.2 QPSK modulator and generated phase states φ(k).............. 80

B.3 Karnaugh-maps for the DQPSK precoder outputs I(k)and Q(k)...... 82

B.4 DQPSK precoder circuit diagram . . . . . . . . . . . . . . . . . . . . . . . 82

C.1 Schematic of a 4-ary ASK receiver . . . . . . . . . . . . . . . . . . . . . 83

C.2 Karnaugh-maps for the 4-ary ASK decoder outputs D1 and D2 . . . . . . . 84

C.3 4-ary ASK decoder circuit diagram . . . . . . . . . . . . . . . . . . . . . . 84

D.1 Schematic of the one bit delay-and-add duobinary encoder versus the stub

filterduobinaryencoder............................ 85

D.2 Photo of constructed electrical stub LPFs, (a) Stub-5, (b) Stub-6 . . . . . . 86

D.3 Frequency response of the electrical stub LPF (Stub-5) and eye diagram of

a generated 10 Gbit/s electrical duobinary signal . . . . . . . . . . . . . . . 87

D.4 Frequency response of the electrical stub LPF (Stub-6), and eye diagram of

a generated 10 Gbit/s electrical duobinary signal . . . . . . . . . . . . . . . 87

D.5 Frequency response of a Stub-5 filter and 5th order Bessel LPF ([email protected]) 88

LIST OF FIGURES xiii

D.6 Frequency response of a Stub-6 filter and 5th order Bessel LPF ([email protected]) 88

xiv LIST OF FIGURES

List of Tables

2.1 Transmitted bit stream “01010011” for DPSK modulation . . . . . . . . . . 13

2.2 Transmitted and received bit stream “01010011” for Duobinary modulation 20

3.1 Mapping of the input signals (Ikand Qk) to the output electric field (Eo) and

generated phase state (φk) of the o(D)QPSK modulator . . . . . . . . . . . 24

3.2 Mapping of data signals (Ukand Vk) to transmitted symbol (dk), and corre-

sponding phase changes (4φk) for DQPSK modulation . . . . . . . . . . 25

3.3 Transmitted and received data bit streams Uk= “01010011” and Vk= “10011010”

for an optical DQPSK system . . . . . . . . . . . . . . . . . . . . . . . . . 27

4.1 Mapping of input binary data to output levels of a 4-ary ASK signal . . . . 45

5.1 Comparison between different modulation formats at 20 Gbit/s . . . . . . . 74

B.1 Mapping of the phase change (4φk) to received data (u(k)and v(k)) for

DQPSKsignals ................................ 80

B.2 Mapping of the input signals (I(k)and Q(k)) to the output electric field

(Eo) and generated phase state (φ(k)) of the o(D)QPSK modulator . . . . . 80

B.3 DQPSK precoder look up table . . . . . . . . . . . . . . . . . . . . . . . . 81

C.1 Mapping of the input levels and the 3 received patterns to the decoded out-

puts for a 4-ary ASK receiver . . . . . . . . . . . . . . . . . . . . . . . . . 84

xvi LIST OF TABLES

Contents

Abstract iii

Acknowledgments v

List of Ph.D. Publications vii

List of Figures ix

List of Tables xv

Contents xvii

1 Introduction 1

1.1 Background.................................. 1

1.2 Motivation................................... 3

1.3 Organization of the Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Digital Modulation Formats for Optical Communication Systems 5

2.1 Introduction.................................. 5

2.2 Performance Evaluation Parameters . . . . . . . . . . . . . . . . . . . . . 5

2.2.1 Receiver Sensitivity . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.2.2 Chromatic Dispersion Tolerance . . . . . . . . . . . . . . . . . . . 6

2.2.3 SpectralEfficiency .......................... 7

2.3 Digital Optical Modulation Formats . . . . . . . . . . . . . . . . . . . . . 8

2.3.1 Binary Intensity Modulation (IM) . . . . . . . . . . . . . . . . . . 9

2.3.2 Differential Phase Shift Keying (DPSK) . . . . . . . . . . . . . . . 11

2.3.2.1 DPSK Precoder . . . . . . . . . . . . . . . . . . . . . . 11

2.3.2.2 DPSK Transmitter . . . . . . . . . . . . . . . . . . . . . 12

2.3.2.3 DPSK Receiver . . . . . . . . . . . . . . . . . . . . . . 14

2.3.3 Duobinary Modulation (DB) . . . . . . . . . . . . . . . . . . . . . 15

2.3.3.1 Duobinary Precoder . . . . . . . . . . . . . . . . . . . . 16

2.3.3.2 Duobinary Encoder . . . . . . . . . . . . . . . . . . . . 17

2.3.3.3 Duobinary Decoder . . . . . . . . . . . . . . . . . . . . 18

3 Differential Quadrature Phase Shift Keying (DQPSK) 21

3.1 Introduction.................................. 21

3.2 Differential Quadrature Phase Shift Keying (DQPSK) . . . . . . . . . . . . 21

xviii CONTENTS

3.2.1 DQPSKPrecoder........................... 22

3.2.2 DQPSK Optical Encoder . . . . . . . . . . . . . . . . . . . . . . . 22

3.2.3 DQPSK Optical Decoder . . . . . . . . . . . . . . . . . . . . . . . 26

3.3 DQPSK Transmission System at 2×10Gbit/s ................ 27

3.4 DQPSK PolDM Transmission System at 2×2×10 Gbits/s . . . . . . . . . . 34

4 Quaternary Intensity Modulation 43

4.1 Introduction.................................. 43

4.2 Conventional Quaternary Intensity Modulation (4-IM) . . . . . . . . . . . 43

4.2.1 Generation of Conventional 2×10 Gbit/s Quaternary 4-IM Signals . 44

4.2.2 Quaternary 4-IM Signal Detection . . . . . . . . . . . . . . . . . . 45

4.2.3 Experimental Transmission System for 2×10 Gbit/s Quaternary Mod-

ulation ................................ 46

4.3 Quaternary Amplitude Shift Keying (QASK) . . . . . . . . . . . . . . . . 49

4.3.1 Generation and Detection of 2×10 Gbit/s QASK Signals . . . . . . 49

4.3.2 Experimental Results Measured for 2×10 Gbit/s QASK . . . . . . 51

4.4 Quaternary Duobinary Modulation (QDB) . . . . . . . . . . . . . . . . . . 54

4.4.1 Generation and Detection of 2×10 Gbit/s QDB signals . . . . . . . 54

4.4.2 Experimental Results Measured for 2×10 Gbit/s QDB . . . . . . . 55

4.5 Quaternary Polarization Duobinary Modulation (QPolDB) . . . . . . . . . 59

4.5.1 Generation and Detection of 2×10 Gbit/s QPolDB Signals . . . . . 59

4.5.2 Experimental Results Measured for 2×10 Gbit/s QPolDB . . . . . 60

4.6 Other Quaternary Multilevel Modulation Formats . . . . . . . . . . . . . . 68

4.6.1 Quaternary Polarization Amplitude Shift Keying (QPolASK) . . . . 68

4.6.2 Quaternary Differential-Phase ASK Modulation (DP-ASK) . . . . . 69

5 Results Discussion and Conclusion 71

5.1 Introduction.................................. 71

5.2 ResultsDiscussion .............................. 71

5.3 Conclusion .................................. 74

A Optical Intensity Mach-Zehnder Modulator (MZM) 75

B DQPSK Precoder 79

C 4-ary ASK Decoder 83

D Duobinary Filters 85

Bibliography 89

Chapter 1

Introduction

1.1 Background

The growth increase in the number of Internet users and the high demand for multimedia

applications such as audio and video will cause a need to upgrade the existing Internet back-

bone communication networks to operate at higher transmission rates. Currently, virtually

most or all of the telephone conversations, cellular phone calls, and Internet packets must

pass through an optical fiber communication network between the source and destination.

While initial deployment of optical fiber networks were mainly for long-haul or subma-

rine transmission, optical fiber networks are currently in virtually all metro networks [1].

These optical communication networks widely use Dense Wavelength Division Multiplexed

(DWDM) transmission systems to increase the transmission capacity by transmitting differ-

ent data streams on different wavelengths (channels). It is possible to increase the through-

put of such DWDM transmission systems by using wider optical bandwidths per channel so

that the data rate per channel can be increased, or by using advanced modulation formats

with higher spectral efficiency (SE) that can be used to transmit more information using the

same bandwidth, or by using a combination of both mentioned methods [2, 3, 4, 5, 6]. Using

wider optical bandwidths per channel will lead to a limited number of DWDM channels due

to a limitation by the spectral bandwidth of optical amplifiers and fiber transmission lines

[5]. Higher spectral efficiency can be achieved by doubling the transmission capacity by

transmitting more information in the amplitude, phase, polarization or a combination of all

[2, 6, 7]. Multilevel modulation formats are considered as advanced modulation formats

that can be used to overcome the problems associated with limited channels and bandwidth

of DWDM systems by lowering the signaling speed by carrying several information bits on

a single symbol. Multilevel modulation formats are widely used in highly efficient electri-

cal communication systems such as wireless communications [5]. Some examples of such

optical advanced multilevel modulation formats that can double the data rate are:

1. Conventional Quaternary Intensity Modulation (4-IM) or also called 4-ary ASK can

double the data rate by transmitting more information in the amplitude of the carrier

signal by modulating the amplitude with an electrical 4-level ASK signal. To receive

and extract the original transmitted data, a single photo diode, three decision circuits

and a decoding logic is needed at the receiver side [8, 9, 10, 11, 12].

2. Differential Quadrature Phase Shift Keying (DQPSK) doubles the data rate by trans-

2Introduction

mitting information in two quadrature phases. To receive and extract the original

transmitted data, two balanced receiver structures each similar to a Differential Phase

Shift Keying (DPSK) receiver are needed at the receiver side. Each receiver consists

of two photo diodes and a Mach Zehnder Delay Interferometer (MZDI) [13, 14, 15].

3. Differential Polarization Phase Shift Keying (DPolPSK) doubles the data rate by

transmitting information in the carrier phase and two orthogonal polarizations. To

receive and extract the original transmitted data, a balanced receiver structure, consist-

ing of two photodiodes and a MZDI (DPSK receiver) in addition to a similar receiver

used for the 4-IM signal are needed [16, 17].

4. Differential Phase Amplitude Shift Keying (DP-ASK) doubles the data rate by trans-

mitting information in both the phase and amplitude of the carrier signal. To receive

and extract the original transmitted data, a balanced receiver structure, consisting of

two photo diodes and a MZDI (DPSK receiver) to detect one data stream in addition

to a single photo diode (ASK receiver) to detect the other transmitted data stream are

needed at the receiver [18, 19].

Combining DQPSK with Polarization Division Multiplex (PolDM) would improve the spec-

tral efficiency but would require a more complex receiver. The multilevel 4-IM modulation

format can be considered as a cost-effective single-channel implementation of high-capacity

links and can result in increased capacity when combined with DWDM. However, its in-

herent degradation of the receiver sensitivity suggests that it is not suitable for ultra-long

distance transmission. Therefore, it can be suitable for short reach (several kilometers) to

medium reach (several hundreds kilometers) transmission applications [5]. A quaternary

4-ary (4-IM) modulation signal can also be generated using different generation techniques

than the conventional method by:

1. Optically combining two binary intensity modulated signals with unequal amplitudes

in quadrature phases using a QPSK modulator (QASK) [20, 21, 6]. To receive and

extract the original transmitted data, the same receiver structure of the conventional

4-IM method is used.

2. Optically combining two binary intensity modulated signals with unequal amplitudes

in orthogonal polarizations using a polarization beam splitter (QPolASK) [22, 23].

To receive and extract the original transmitted data, the same receiver structure of the

conventional 4-IM method is also used.

All of the previous mentioned methods double the transmission capacity and are spectrally

efficient to one extent, but still do not have the best spectral efficiency. Another limitation

for high speed optical transmission systems is the chromatic dispersion (CD) in single mode

fibers which is the optical frequency dependence of the group delay. It is possible to over-

come the CD limitation by using other modulation formats that have higher CD tolerance

such as Duobinary modulation. Duobinary modulation outperforms all the other previously

mentioned quaternary modulation formats in term of CD tolerance. Although Duobinary

modulation does not double the transmission rate, it is also spectrally efficient [24, 25, 26].

Combining Duobinary modulation and Quaternary modulation could result in even a higher

spectral efficient modulation format. It is possible to reduce the spectral bandwidth of the

1.2 Motivation 3

QASK and QPolASK modulation formats to an extent where it outperforms all the other

mentioned multilevel modulation formats in terms of spectral efficiency, by using Duobinary

modulation instead of binary modulation in the generation scheme. A 9-constellation point

optical quaternary intensity signal can be generated by combining two optical Duobinary

signals with unequal amplitudes in quadrature phases using a QPSK modulator resulting

in a Quaternary Duobinary signal (QDB)/(9-QAM QPSK) [27, 28, 21, 6] or in orthogonal

polarizations using a Polarization Beam Combiner (PBC) resulting in a Quaternary Polar-

ization Duobinary signal (QPolDB)/(9-QAM PolDM) [29, 7, 6], respectively. Both of these

modulation formats have higher spectral efficiency compared to the binary, Duobinary, and

the other quaternary modulation formats. The QPolDB format had higher CD tolerance

compared to binary modulation and other quaternary modulation formats operating all at

the same bit rate.

1.2 Motivation

The motivation of the work performed in this dissertation was to implement different op-

tical quaternary modulation formats at 10 Gbaud and 20 Gbit/s data rate, measure their

performance regarding to their sensitivities, CD tolerance, and spectral bandwidth and com-

pare them with each other and to other intensity modulation schemes at the same bit rate.

The quaternary modulation formats used throughout this work were DQPSK, DQPSK with

PolDM, 4-IM, QASK, and the newly proposed QDB and QPolDB. All these modulation

formats are known to be advanced modulation formats with higher spectral efficiency com-

pared to binary Non-Return-to-Zero amplitude modulation (NRZ-ASK). The experimental

results obtained in this thesis gives the reader or designer useful information that can be

used to select between different modulation formats a suitable one that can match the re-

quirements of a specific optical transmission system for short to medium reach transmission

applications.

1.3 Organization of the Thesis

The thesis is organized into five main chapters as follows:

•The first chapter is a general introduction to the work performed in this thesis and the

motivation.

•The second chapter in its first part gives a brief description of some of the perfor-

mance parameters used to compare between different modulation formats. The second

part introduces some digital modulation formats that are used for optical communi-

cation systems such as binary (NRZ-ASK), differential phase shift keying (DPSK),

and Duobinary modulation (DB). For each of these modulation formats a brief expla-

nation will be given on their generation and detection schemes and expected signal

constellation diagram. Measured eye diagrams at 10 Gbit/s obtained by experimental

implementation of each modulation format will also be presented. These modulation

formats will be useful to understand the multilevel modulation formats mentioned

later in chapters three and four.

4Introduction

•The third chapter will introduce the optical DQPSK modulation format describing

the generation and detection scheme, and the expected signal constellation diagram.

Experimental data will be presented for 20 Gbit/s DQPSK and 40 Gbit/s DQPSK

PolDM transmission system showing the receiver sensitivity and eye diagrams. Also

the CD tolerance for 20 Gbit/s RZ-DQPSK will be evaluated.

•The fourth chapter will introduce different quaternary modulation formats imple-

mented at 20 Gbit/s including conventional 4-IM, QASK, QDB, and QPolDB. The

generation and detection scheme and the expected signal constellation diagram for

each of these modulation formats will be explained. Also the practical hardware sys-

tem setup and the experimental measured results obtained such as the eye diagrams,

receiver sensitivity, and measured CD tolerances will be presented for the quaternary

modulation formats. The last section of this chapter will introduce and briefly explain

some other previously reported quaternary modulation formats such as QPolASK, and

DP-ASK.

•The fifth chapter is the last chapter of the thesis that will discusses the results ob-

tained from chapter three and four with a full comparison based on the performance

parameters given in chapter two. A conclusion of the work is given in this chapter.

Chapter 2

Digital Modulation Formats for Optical

Communication Systems

2.1 Introduction

This chapter gives a short introduction to the basic concepts and performance evaluation

parameters used to compare the different modulation formats presented in this thesis. The

parameters covered in this chapter are the receiver sensitivity, chromatic dispersion tol-

erance, and spectral efficiency (SE). The last section of this chapter will introduce some

standard digital optical modulation formats such as binary amplitude shift keying (ASK),

Differential phase shift keying (DPSK), and Duobinary modulation (DB). The generation

and detection schemes for each modulation format will be explained in details showing the

corresponding signal constellation and eye diagrams. These modulation formats are useful

to fully understand the multilevel modulation formats later mentioned in chapters three and

four.

2.2 Performance Evaluation Parameters

The commonly used linear parameters that will be used to compare the performance of

different optical transmission systems based on different modulation formats mentioned in

this thesis are the receiver sensitivity, dispersion, and spectral efficiency. A brief explanation

for each parameter will be given in the following:

2.2.1 Receiver Sensitivity

An important parameter to indicate the receiver performance in an optical transmission sys-

tem is called the receiver sensitivity. It is usually defined as the minimum average received

optical power for which the Bit Error Rate (BER) of the optical receiver is 10−9. Figure 2.1

shows the main components of a simple optical pre-amplified direct detection receiver used

for an optical transmission system consisting of an optical pre-amplifier, optical band-pass

filter (BPF), a photodiode, an electrical amplifier and an electrical low-pass filter (LPF).

The BER depends on the signal to noise ratio (SNR), which depends on various noise

sources that degrade the received optical signal. The important noise contributions are the

6Digital Modulation Formats for Optical Communication Systems

Electrical

Amplifier

Photo

detector

Optical

Pre-Amp.

BPF LPF

Received

Optical

signal

Received

Electrical

signal

Figure 2.1: Schematic of an optical direct detection pre-amplified receiver

Amplified Spontaneous Emission (ASE), Shot noise and Thermal noise [30]. It should

be noted that the photons at the output of the optical amplifier obey noncentral or central

negative bionomial distributions for transmitted one or zero, respectively. If there is no

gain, these become Poisson distributions. If the gain is infinite, these become chi-square

distributions [31]. The receiver sensitivity can also be degraded due to the fiber dispersion

that leads to power penalties and depends on both the bit rate and the fiber length. The

performance of an optical receiver can also be measured using the eye diagram. The closing

of the eye is a measure of degradation in receiver performance and is associated with a

corresponding increase in the BER [32].

2.2.2 Chromatic Dispersion Tolerance

Chromatic dispersion (CD) is one of the most basic characteristics of optical fibers [33].

It is also known as group-velocity dispersion (GVD) which is caused by the wavelength

dependence of the refractive index. In optical transmission systems when an optical carrier

is modulated and transmitted over a fiber, the modulated signal occupies a certain optical

spectral bandwidth. Signal from different parts of the spectrum propagates with different

speed due to chromatic dispersion [33, 1]. When operating at higher bit rates over long

lengths of optical fiber, the chromatic dispersion causes broadening of a single transmitted

pulse in addition to interference between adjacent pulses which is known as inter-symbol-

interference (ISI), hence producing distortions in the received signals [34, 35].

To understand better the chromatic dispersion, we will use the complex field transfer

function of an optical fiber with a length Lin the frequency domain

H(ω) = ejϕ =e−jβ(ω)L,(2.1)

where β(ω)is the propagation constant , ωis the optical angular frequency and the

attenuation has been neglected. It is useful to approximate its phase ϕby a truncated Taylor

series around the carrier frequency ω0,

ϕ=−β(ω)L=−n(ω)ω

cL=−(β0+β1(ω−ω0) + 1

2β2(ω−ω0)2+...)L, (2.2)

where cis the speed of light in vacuum and n(ω)is the refractive index. At ω0the

propagation constant β(ω)assumes the value β0, and its first and second derivatives with

respect to ωare β1, and β2, respectively as follows:

2.2 Performance Evaluation Parameters 7

β1=β0=1

υg

=dβ

dωω=ω0

,(2.3)

β2=β00 =d

dω 1

υg!=d2β

dω2ω=ω0

,(2.4)

where β0is the inverse of the group velocity υg,β00 is the first order dispersion, λis the

optical wavelength, and ω= 2πc/λ. The group delay is:

τg=−ϕ0=dϕ

dω = (β0+ (ω−ω0)β00)L. (2.5)

It is a linear function of ω. Its derivative with respect to wavelength λand length Lis

the chromatic dispersion coefficient:

D=d2τg

dλ.dL =d

dλ 1

υg!=−2πc

λ2β00 =−2π

λ2 2d¯n

dω +ωd2¯n

dω2!,(2.6)

in the units of ps/(km.nm) [36, 37, 1, 32]. The wavelength dependence of Dis governed

by the frequency dependence of the mode index ¯n. Standard single mode fiber (SSMF) has

a dispersion coefficient from 16 to 19 ps/(nm.km) at the low-loss window of 1550 nm. Due

to chromatic dispersion in optical fiber, the level of intersymbol interference that limits the

maximum transmission distance depends on the chromatic dispersion index factor γ[34, 38]

given as :

γ=R2LDλ2

πc ,(2.7)

where Ris the data rate of the system, and Lis the transmission length. It can be

seen that the limit of transmission length due to the waveform distortion by the chromatic

dispersion of the fiber is inversely proportional to the chromatic dispersion and the square

of the system data rate. The chromatic dispersion tolerance curve shows the required optical

signal-to-noise ratio OSNR [dB/0.1nm] for a BER = 10−9versus the residual chromatic

dispersion. The OSNR [dB/0.1nm] is usually measured with an Optical Spectrum Analyzer

(OSA) by subtracting the noise power (dBm) measured with the OSA resolution bandwidth

(RB) set to 0.1 nm from the signal peak power (dBm) measured with the OSA RB set to

1 nm. Normally the CD tolerance of a modulation format corresponds to the maximum

residual dispersion where the OSNR penalty is 1dB [39].

2.2.3 Spectral Efficiency

The throughput of a dense wavelength-division multiplexed (DWDM) transmission system

can be increased by using a wider optical bandwidth, increasing the spectral efficiency SE,

or by some combination of both [2]. Utilizing a wider bandwidth typically requires ad-

ditional amplifiers and other optical components. So a better solution is to efficiently use

the available optical bandwidth by reducing the channel spacing to increase the number of

channels within a fixed optical bandwidth. This can be achieved by reducing the spectral

occupancy for the modulation signal by using alternative advanced modulation formats with

8Digital Modulation Formats for Optical Communication Systems

high spectral efficiency [4, 3, 10, 40, 41, 42]. The spectral efficiency (SE) limit in a DWDM

system is defined as:

SE =C

4f,(2.8)

where 4fis the channel spacing and Cis the capacity per channel. Cand Shave

units of bits per second (b/s) and b/s/Hz, respectively. Advanced modulation formats such

as multilevel modulation formats have higher spectral efficiency due to the narrower optical

spectrum with respect to that of binary NRZ signal with identical bit rate [40, 39, 43].

2.3 Digital Optical Modulation Formats

For designing digital optical communication links, there exist a wide variety of modulation

formats to choose from. The electric field of the optical carrier is given by [32]:

E(t) = ˆeAe−j(ωt+φ).(2.9)

Four properties of an optical signal can be modulated: Ais the amplitude of the optical

field, φis the optical phase, ωis the optical angular frequency, and ˆe is the polarization

vector of the laser source [44, 45, 46, 47]. Each of these parameters can be modulated by an

electrical binary baseband signal q(t):

q(t) = ∞

i=−∞

Iiq(t−iTb),(2.10)

with the i−th information coefficient I∈[0,1] and the baseband pulse shape q(t)

delayed by multiples of the bit period Tb. Depending on which parameter of the laser source

is modulated, the modulation is mainly differentiated as: amplitude shift keying (ASK)

[48, 24, 49, 50], frequency shift keying (FSK) [51, 52, 53, 54], phase shift keying (PSK) [55,

56, 49, 57, 58, 15, 59], or polarization shift keying (PolSK) [60, 61, 62]. Figure 2.2 shows

an electrical binary data stream “01010011” used to modulate the amplitude, frequency, and

phase of an optical carrier signal, resulting in the generation of the standard optical digital

modulation formats ASK, PSK and FSK respectively.

Most systems today use the binary amplitude modulation format amplitude-shift-keying

(ASK) because it is simple and cheap to implement. Recently, and in the previous years,

novel advanced modulation formats with improved performance with respect to binary mod-

ulation have been suggested and investigated. However, these advanced modulation formats

may add further complexity to the transmitter and receiver hardware of the optical transmis-

sion system [45, 46, 63, 15]. Some of these advanced modulation formats such as Duobinary

and DPSK feature enhanced robustness to chromatic dispersion, optical filtering, and/or

nonlinearities [15].

Other advanced modulation formats such as multilevel modulation formats can even

have higher spectral efficiency than standard binary modulation due to the doubling of the

transmission capacity by transmitting more information in the amplitude, phase, polarization

or a combination of all [6]. Some of the previous published multilevel modulation formats

2.3 Digital Optical Modulation Formats 9

0 1 0 1 0 0 1 1 Electrical

Binary

Data

Optical

Carrier

Signal

ASK

PSK

FSK

Figure 2.2: Basic digital modulation formats (ASK, PSK, FSK)

are quaternary intensity modulation (4-IM) or also called 4-ary ASK [8, 9, 10, 11, 12], dif-

ferential quadrature phase-shift-key (DQPSK) [14, 15, 13], Differential Polarization-Phase-

Shift Keying (DPolPSK) [16, 17], Quaternary polarization amplitude-shift-keying (QPo-

lASK) [23, 22], Differential-phase amplitude-shift-keying (DP-ASK) [18, 19], and quater-

nary amplitude-shift-keying (QASK) [20, 28, 6, 21].

In the following sections the principle of operation will be given for generating and de-

tecting standard optical modulation formats such as Binary Intensity Modulation (IM), Dif-

ferential Phase Shift Keying (DPSK), and Duobinary Modulation (DB). These modulation

formats will be used as a base to generate more advanced quaternary modulation formats

such as DQPSK, 4-IM, QASK, QDB, and QPolDB that will be explained in more details in

the following chapters.

2.3.1 Binary Intensity Modulation (IM)

Binary intensity modulation (IM) or also called Amplitude Shift Keying (ASK) is the most

common used modulation format in the deployed optical transmission systems currently

available today. This modulation technique is well known from classical telecommunica-

tion theory [64, 65, 42]. It is also known as On-Off-keying (OOK) because it is a result of

switching ON and OFF the amplitude of an optical carrier signal. This can be achieved by

directly modulating the current of the laser source or by using an external optical modulator

to modulate the amplitude of the laser carrier signal [66]. The most common generation

scheme is externally modulating the laser signal using an intensity Mach-Zehnder Modu-

lator (MZM) biased at the quadrature point of the modulator power transfer function, and

driven by an electrical binary NRZ-ASK signal with peak-to-peak (p-p) amplitude of Vπ

as shown in figure 2.3. More details and explanation on the principle of operation of the

intensity Mach-Zehnder modulator are given in appendix A.

10 Digital Modulation Formats for Optical Communication Systems

Laser

Electrical

Binary NRZ

Data Stream

MZM

Electrical

NRZ-ASK

Optical Intensity Modulated Signal (NRZ-ASK)

Intensity

levels

Output

Intensity

Input

Voltage

Quadrature Bias point

Figure 2.3: Principle of Binary NRZ-ASK modulation generation using intensity MZM

The intensity eye diagram of a detected 10 Gbit/s binary NRZ-ASK signal is also shown

in figure 2.3 (left). Figure 2.4 shows the resulting constellation diagram of the optical ASK

signal consisting of the OFF state “0” and the ON state “1”.

Re(E)

Im(E)

0 1

Figure 2.4: ASK signal constellation diagram

The ASK modulation format can also be characterized by the extinction ratio (ER) which

is defined as the ratio between the intensities of the ON and OFF states [35]:

ER = 10 log ION

IOF F .(2.11)

The binary NRZ-ASK signal can be detected using a simple photo diode at the receiver

(direct detection) similar to the receiver shown in figure 2.1. The spectrum of a binary NRZ-

ASK signal contains a carrier and has a bandwidth equal to twice the bit rate ( 2

Tb), where the

bandwidth will be defined as the null-to-null bandwidth of the main lobe spectrum [42].

2.3 Digital Optical Modulation Formats 11

2.3.2 Differential Phase Shift Keying (DPSK)

Differential-phase-shift-keying (DPSK) signal format is a promising technique to improve

the system performance of optical communication systems [55]. In the DPSK modulation

format, the information is carried by the phase of the optical carrier signal. The optical

power in a DPSK signal appears in each bit slot, with the binary data encoded as either a 0

or πoptical phase shift between adjacent bits. In NRZ-DPSK the optical power can occupy

the entire bit slot. In RZ-DPSK the optical power can appear as an optical pulse. RZ-

DPSK can be implemented by using a DPSK modulator in combination with an RZ pulse

carver [55, 47]. The most benefit of DPSK signals when compared to the conventional ASK

format is the high receiver sensitivity due to the ~3dB lower OSNR required to achieve a

given BER [67, 55, 59, 68]. DPSK can be used to extend the transmission distance, reduce

optical power requirements, and relax component specifications [15]. A critical element in

any DPSK system is the demodulator, which converts the phase modulation of the received

signal into intensity modulation [68]. The advantage of DPSK over ASK is obtained at the

expense of increased complexity and cost in the transmitter and receiver structure, such as a

need of a differential encoder at the transmitter, a delay- interferometer and its stabilization

at the receiver side and two photodiodes to construct a balanced receiver [55, 59]. An

extension of DPSK to Differential quadrature phase-shift keying (DQPSK) should enable

higher spectral efficiency and greater chromatic-dispersion tolerance [69, 70].

A standard DPSK transmission system is shown in figure 2.5. It consists of a CW laser

source, a precoder circuit, an electrical modulator driver amplifier, a DPSK transmitter (op-

tical modulator) which encodes the data in the phase of the optical signal, and a DPSK

receiver which demodulates the received data [59].

Laser

Electrical

Binary NRZ

Data Stream

Electrical

Binary signal

Optical DPSK signal

k-1

Precoder

-1

Fiber

DPSK

Reciever

Received

Binary data

DPSK

Transmitter

Figure 2.5: Schematic diagram of an optical DPSK transmission setup

A description for each block of the optical DPSK setup will be given in the following

subsections:

2.3.2.1 DPSK Precoder

A precoder must be used at the transmitter side in order to recover the original transmitted

data at the receiver side by using a differential detection receiver with out the need of a

special decoding circuit [1, 47]. Generally, the precoder differentially encodes the original

binary bit sequence using a logic XOR gate with a feedback tap with one bit delay. The

precoding function can be written as:

12 Digital Modulation Formats for Optical Communication Systems

bk=ak⊕bk−1,(2.12)

where ak∈ {0,1}is the original transmitted binary data sequence, bk∈ {0,1}is the

precoded binary sequence, and ⊕is the logic instruction “XOR”. The encoded symbol bkis

then transmitted in a bipolar fashion ck∈ {−1,+1}.

2.3.2.2 DPSK Transmitter

A commonly used NRZ-DPSK transmitter setup consists of an external MZM biased at

minimum transmission and driven with a precoded binary data with twice the switching

voltage required for ASK modulation 2Vπas shown in figure 2.6.

Laser MZM

Electrical

Binary

Signal

Optical DPSK Signal

Intensity

levels

Output

Intensity

Input

Voltage

Minimum

Transmission

Bias Point

Electrical

Binary NRZ

Data Stream

Precoder

-1 +1

Figure 2.6: Principle of DPSK modulation generation using a MZM

Figure 2.6 also shows the eye diagram (bottom-right) for a 10 Gbit/s electrical binary

NRZ signal obtained from a pseudo-random bit sequences (PRBS) pattern generator and the

resulting 10Gbit/s optical DPSK signal eye diagram (top-left) detected using a photodiode.

A phase modulator (PM) can also be used instead of the intensity MZM by only modulating

the phase of the optical field, which results in a constant-envelope optical signal [15]. A

drawback of this scheme is that the waveform ripples of the electrical driving signals are

converted into phase modulation and will eventually degrade the receiver sensitivity. This

is not the case with the MZM since the phase of the optical field changes its sign when a

transition is occurred through the minimum point of the MZMs power transmission curve,

therefore, the two neighboring intensity transmission maxima points will have an opposite

phase, resulting in a near-perfect πphase shift independent of the drive voltage swing [59,

15]. Table 2.1 illustrates the different digital signals in a DPSK transmitter, for an input bit

pattern ak= ”01010011”. For differential encoding a reference bit is needed to initiate the

encoding process. This reference bit could arbitrarily be set to logic “1” or logic “0”. In the

table the reference bit bkat instant k=−1was set to logic “0”.

2.3 Digital Optical Modulation Formats 13

Table 2.1: Transmitted bit stream “01010011” for DPSK modulation

Time instant k-1 0 1 2 3 4 5 6 7

Transmitted data ak0 1 0 1 0 0 1 1

Diff. Encoded data bk0 0 1 1 0 0 0 1 0

Bipolar encoded ck-1 -1 +1 +1 -1 -1 -1 +1 -1

I/P voltage to MZM ±Vπ−Vπ−Vπ+Vπ+Vπ−Vπ−Vπ−Vπ+Vπ−Vπ

O/P Electrical Field ±E+E+E−E−E+E+E+E−E+E

Transmitted phase φ0 0 π π 0 0 0 π0

Phase difference |φ|0π0π0 0 π π

The optical spectrum of a DPSK signal has the same bandwidth of an ASK signal but

without containing any carrier frequency component. Figure 2.7 shows the resulting signal

constellation diagram of the DPSK signal.

Re(E)

Im(E)

-1 1

Figure 2.7: DPSK signal constellation diagram

For the generation of RZ-DPSK signals, the transmitter setup will consist of two external

modulators as shown in figure 2.8. The first modulator is used for phase modulation which is

mostly based on a MZM biased at minimum transmission and driven with precoded binary

data with a voltage swing of 2Vπ[59, 15] thus generating an NRZ-DPSK signal as shown

in figure 2.8 (Top-Left). The second MZM is used as a pulse carver and is driven by a

sinusoidal signal. When the pulse carver MZM is biased at the quadrature point (similar to

ASK) and driven with a clock frequency signal with voltage swing amplitude of Vπ, a 50%

RZ-DPSK signal will be generated as shown in figure 2.8 (Top-Right). Different duty cycle

ratios can be obtained for the RZ-DPSK signals depending on the driving conditions of the

pulse carver such as the modulation frequency, bias position, and driving voltage [59].

It is also noticed that the generated NRZ-DPSK signal shown in Figure 2.8 has a non-

constant optical intensity. This is due to the fact that the signal intensity returns to zero at

every transition between marks “1” and spaces “0”, which results in a residual amplitude

modulation. The width of the resulting intensity dips depends on the driving signal band-

width and voltage [59, 55]. However, since DPSK encodes information in the optical phase

rather than in the intensity, these dips are of reduced importance, especially for RZ-DPSK,

where the pulse carver cuts out the amplitude-modulation-free center portions of the bits

only, and thus largely eliminates any residual dips [15].

14 Digital Modulation Formats for Optical Communication Systems

Laser MZM

10Gbit/s optical NRZ-DPSK Signal

10 Gbit/s PRBS Electrical

Binary NRZ signal

RZ-Mod

10Gbit/s optical RZ-DPSK Signal

10 GHz Electrical

Clock signal

Figure 2.8: 10 Gbit/s optical RZ-DPSK generation scheme using two MZMs with resulting

NRZ-DPSK (left) and 50% RZ-DPSK (right) eye diagrams.

2.3.2.3 DPSK Receiver

The most commonly used DPSK receiver is shown in figure 2.9 . It consists of an optical

pre-amplifier, an optical filter, a Mach-Zehnder interferometer with a 1-bit delay in one arm,

and a dual photodiode balanced receiver [71, 68, 59, 55, 15, 44]. The critical component of

the DPSK receiver is the one bit delay Mach-Zehnder delay-interferometer (MZDI) which

acts as the demodulator.

MZDI

Optical

Pre-Amplifier

BPF

1-bit delay

Data

Photodiode

Balanced receiver

DPSK

Signal

Figure 2.9: DPSK receiver

The one bit delay MZDI can be constructed using two 3dB 2×2couplers. One coupler

is used at the input and the other at the output [44]. The two optical branches connecting the

two couplers have unequal optical path lengths. The difference in the optical path lengths

correspond to one bit duration. If the input optical field to the MZDI shown in figure 2.9 is

E(t), the output optical fields at both the output arms of MZDI are E1(t)and E2(t)[44, 72]

as follows:

E1(t) = 1

2[E(t) + E(t−T)] (2.13)

E2(t) = −j1

2[E(t)−E(t−T)] .(2.14)

The MZDI lets two adjacent bits interfere with each other at its output ports. This in-

terference leads to the presence of power at a MZDI output port when there is constructive

interference between two adjacent bits, or the absence of power when there is destructive

2.3 Digital Optical Modulation Formats 15

interference between two adjacent bits. Therefore, the previous bit in a DPSK encoded bit

stream acts as the phase reference for demodulating the current bit [55, 15]. At the MZDI

outputs, the two output ports will carry identical, but logically inverted data streams under

DPSK modulation. The optically demodulated signal measured at the constructive port is

similar to Duobinary modulation, and at the destructive port is similar to alternate-mark

inversion (AMI) modulation signal [15, 73, 74, 59]. After the MZDI, direct detection can

be achieved by using a single photo-diode (single ended detection) to receive the signal by

detecting it at one of the interferometer output ports, or by using two photo receivers (bal-

anced detection) to receive the signal by detecting the difference between the constructive

and destructive interferometer output ports [56, 68]. For balanced detection the photocurrent

difference is proportional to the difference of the intensities at the two photodiode outputs:

|E1(t)|2−|E2(t)|2=Re (E∗(t)E(t−T)) .(2.15)

Figure 2.10 shows the eye diagram patterns for a demodulated 10 Gbit/s NRZ-DPSK

and RZ-DPSK signal at the output of a DPSK receiver with balanced detection. The MZDI

used in the receiver had a differential delay of 100 ps. The two high speed photodiodes

used for balanced detection had approximately equal fiber length. An electrical RF variable

delay line was used to fine tune the path delay difference between the outputs of the two

photodiodes. An electrical differential amplifier was used to measure the current difference

between the two photodiodes.

Figure 2.10: Demodulated 10 Gbit/s NRZ-DPSK (a) and RZ-DPSK (b) intensity eye pat-

terns

2.3.3 Duobinary Modulation (DB)

Waveform distortion due to fiber chromatic dispersion in high-speed optical transmission

systems is a serious problem. Optical Duobinary modulation is an effective way to avoid

such distortion resulting in an increased dispersion tolerance compared to binary NRZ mod-

ulation format [75]. Duobinary modulation is also spectrally efficient due to the reduced

occupied spectral bandwidth which is about half the spectrum bandwidth of a conventional

binary NRZ intensity modulation (IM) [24, 25, 76]. The cost of an optical transmission link

over long-haul distances can be reduced by using Duobinary modulation due to the better

CD tolerance and therefore no need of dispersion compensation which adds to the total cost

[47, 77]. An optical Duobinary signal can be generated by modulating the optical carrier

signal with a three level electrical Duobinary signal. The most common type of optical

Duobinary modulation is the Amplitude Modulation Phase Shift Keying (AM-PSK) type

16 Digital Modulation Formats for Optical Communication Systems

[75]. Duobinary is interesting for Dense Wavelength Division Multiplexing (DWDM) ap-

plications, since it has been optimized to reduce the channel bandwidth [47]. It should also

be noted that Duobinary and DPSK have similarities. The DPSK demodulator which is a

Mach-Zehnder delay interferometer (MZDI) can be viewed as a narrow-band optical filter

with sinusoidal shape. If the MZDI is used at the transmitter side after the intensity Mach-

Zehnder modulator, the signal obtained at the output of the MZDI constructive port has most

of the features of a Duobinary signal [47]. A standard Duobinary transmitter and receiver

consist of a precoder, Duobinary encoder, and decoder. The corresponding practical optical

Duobinary transmission setup consists of the precoder logical circuit, a Duobinary filter as

an encoder, an electrical modulator driver amplifier used to amplify to electrical Duobinary

signal to drive the optical MZM, and a decoder to extract the data from the demodulated

optical Duobinary signal as shown in figure 2.11.

| |

k-1

-1

Precoder Encoder LPF + AC Coupling Decoder

Laser

Electrical

Binary NRZ

Data Stream

MZM

Precoder

Amplifier

Fiber

Photodiode

Duobinary Filter

Electrical

Duobinary signal

Optical

Duobinary signal

Figure 2.11: Schematic diagram of a standard Duobinary generation and detection scheme

(top) and a corresponding practical optical Duobinary transmission setup (bottom)

A description for each block of the optical Duobinary setup will be given in the following

subsections:

2.3.3.1 Duobinary Precoder

The Duobinary encoder at the transmitter side scrambles the original binary data; there-

fore an electrical precoder is needed to be inserted before the encoder to recover the original

transmitted binary data at the receiver side using a conventional NRZ receiver. This precoder

normally is a differential encoder which is also used for a DPSK system [47]. The differ-

ential encoder (precoder) is used to avoid recursive decoding in the receiver thus avoiding

error propagation and reducing hardware complexity [76, 78, 75]. The precoder is basically

constructed using a logic XOR gate with a feedback tap with one bit delay. The precoding

function for Duobinary coding can be written as:

bk= ¯ak⊕bk−1,(2.16)

where akis the transmitted binary data sequence, bkis the precoded binary sequence,

and ⊕is the logic instruction “XOR”. In a laboratory environment, most of the optical

2.3 Digital Optical Modulation Formats 17

Duobinary transmission experiments are performed without a precoder. This is due to the

properties of the commonly used pseudo-random bit sequences (PRBS) where the precoder

output bit stream is in fact a time-delayed version of the input PRBS bit stream. Therefore,

the information is maintained and the functionality of the precoder can be omitted. Also

the transmitted PRBS pattern is inverted at the input of the precoder in order to receive a

non-inverted PRBS at the receiver [26, 76, 45].

2.3.3.2 Duobinary Encoder

To generate an optical Duobinary signal two common types of Duobinary encoding schemes

are well known, a low-pass filtering (LPF) method [24, 79, 80] and the conventional method

by using an electrical one-bit delay and add method [24, 25, 26]. The electrical LPF used to

generate a Duobinary signal usually has approximately a 3dB bandwidth equal to quarter of

the data bit rate. This LPF filter is equivalent to a one-bit delay-and-add operation similar

to the conventional scheme. The Duobinary encoder is mainly used to convert an electrical

binary NRZ signal to a three-level electrical Duobinary signal, producing the logic levels

{0, 1, 2} by adding the current bit to the previous bit [78] as shown in figure 2.11 based on

the following equation:

ck=bk+bk−1.(2.17)

The DC offset is removed from the electrical Duobinary signal ckusing a DC-block ca-

pacitor resulting in the electrical Duobinary signal dkwith the normalized levels {-1 , 0, +1}.

To generate an optical Duobinary signal in practice an intensity MZM biased at minimum

transmission is used and driven by an electrical Duobinary signal with the peak-to-peak

swing voltage (Vp−p) equal to 2Vπ. The generated electric fields are {−E, 0, +E} which

corresponds to the normalized intensities {0, 1} as shown in figure 2.12. Figure 2.12 shows

the optical Duobinary generation method using the conventional one bit delay-and-add en-

coding scheme. The eye diagrams shown in figure 2.12 are for a 10Gbit/s electrical binary

NRZ signal obtained from a PRBS pattern generator (bottom-left), the generated electrical

10Gbit/s Duobinary signal (bottom-right), and finally the resulting 10 Gbit/s optical Duobi-

nary signal detected using a photodiode (top-left).

The ON state corresponding to the normalized intensity “1” can have one of the two

optical phases, 0 and πdepending on the input levels “-1” and “+1” of the electrical Duobi-

nary encoded signal, respectively. The OFF state corresponds to the intensity “0” and de-

pends on the input level “0” of the electrical Duobinary encoded signal [25]. Figure 2.13

shows another optical Duobinary modulation generation scheme based on driving an in-

tensity MZM biased at transmission minimum with an electrical Duobinary signal with a

voltage swing amplitude of 2Vπgenerated using a 5th order Bessel LPF with a cutoff fre-

quency at 2.8 GHz. The measured eye diagrams shown in figure 2.13 are for an electrical

10Gbit/s binary NRZ signal (left) generated using a PRBS pattern generator, an electrical

Duobinary signal (bottom-right) generated using the Bessel filter, and finally the intensity

eye diagram of the generated 10 Gbit/s optical Duobinary signal (top-right). It can be no-

ticed that the intensity profile of the conventional delay-and-add type optical Duobinary

signal [24, 25, 26] shown in figure 2.12 (top-left) is more similar to that of a binary NRZ

signal, while the intensity diagram shown for optical Duobinary signal using the LPF gen-

eration method shown in figure 2.13 (top-right) [24, 79, 80] leaves a small fraction of light

18 Digital Modulation Formats for Optical Communication Systems

Laser MZM

Electrical

Duobinary

Signal

Optical Duobinary Signal (DB)

Intensity

levels

Output

Intensity

Input

Voltage

Minimum

Transmission

Bias Point

10Gbit/s Electrical

Binary PRBS

Data Stream

Conventional

Duobinary delay-

and-add filter

-1 0 +1

Figure 2.12: Principle of Duobinary modulation generation using a MZM and a Duobinary

encoder based on the conventional one bit delay and add filter

within “0” symbols, which incorporate πphase shifts in their center. Due to this property,

this type of Duobinary modulation has also been named Phase-Shaped Binary Transmis-

sion (PSBT) [47, 81, 82]. The optical Duobinary signal generated using the LPF method

suppresses the optical spectrum side-lobes better than the Duobinary one-bit delay and add

method. Therefore it tolerates more chromatic dispersion. A drawback of the LPF method

generated optical Duobinary signal is that it has worse eye opening, poor tolerance to noise,

and worse back-to-back receiver sensitivity compared to the one-bit-delay and add method

[24, 48, 83, 84, 82, 47].

The optical spectrum of a Duobinary signal has two special characteristics. One is

that most of the signal power is concentrated within a narrow bandwidth approximately

half of the binary NRZ-ASK IM signal spectrum. Such a narrow band signal can be ex-

pected to achieve high tolerance to chromatic dispersion [79]. The other is that the optical

Duobinary signal has no carrier frequency component in contrast with the binary NRZ-ASK

IM signal[26]. Figure 2.14 shows the resulting signal constellation diagram of the optical

Duobinary signal.

2.3.3.3 Duobinary Decoder

A simple decoder can be used to detect the received Duobinary signals by using a square-law

device which can be a photo-diode used for detecting binary intensity modulated signals [78,

26]. The photodiode will convert the received optical Duobinary signal dkwith electrical

field levels {-E, 0, +E} to an electrical binary signal |dk|2with normalized amplitudes {0,

1}. Due to the inverter and precoder used at the transmitter side the demodulated electrical

binary signal will be identical to the original transmitted binary data stream [26, 76, 45].

It is also possible invert the data at the receiver side instead of the transmitter side. Table

2.2 illustrates the different digital signals in a Duobinary system, for an input bit pattern ak

2.3 Digital Optical Modulation Formats 19

10Gbit/s Electrical

Binary PRBS

Data Stream

Laser MZM

Duobinary 5th

order Bessel LPF

Optical Duobinary Signal (DB)

Electrical Duobinary Signal

-1

Figure 2.13: 10 Gbit/s optical Duobinary modulation generation using a MZM and a Duobi-

nary encoder based on a 5th order Bessel filter with cutoff frequency at 2.8 GHz

Re(E)

Im(E)

-1 0 1

Figure 2.14: Optical Duobinary signal constellation diagram

20 Digital Modulation Formats for Optical Communication Systems

= “01010011”. For differential encoding a reference bit is needed to initiate the encoding

process. This reference bit could arbitrarily be set to logic “1” or logic “0”. In table 2.2 the

reference bit bkat instant k=−1was set to logic “0”. To prove the precoding, encoding,

and decoding functions are correct, it is shown in table 2.2 that the received data |dk|2is

identical to the original transmitted data ak.

Table 2.2: Transmitted and received bit stream “01010011” for Duobinary modulation

Time instant k-1 0 1 2 3 4 5 6 7

Transmitted data ak0 1 0 1 0 0 1 1

Inverted Data ¯ak1 0 1 0 1 1 0 0

Diff. Encoded Data bk0 1 1 0 0 1 0 0 0

Duobinary Encoded ck1 2 1 0 1 1 0 0

Transmitted Duobinary dk0 +1 0 -1 0 0 -1 -1

I/P voltage to MZM ±Vπ0+Vπ0−Vπ0 0 −Vπ−Vπ

O/P Electrical Field ±E0−E0+E0 0 +E+E

Optical Power (Intensity) 0 E20E20 0 E2E2

Received data |dk|20 1 0 1 0 0 1 1

Chapter 3

Differential Quadrature Phase Shift

Keying (DQPSK)

3.1 Introduction

In this chapter, the Differential Quadrature Phase Shift Keying (DQPSK) modulation format

will be introduced. DQPSK is a multilevel (quaternary) modulation format that can be used

for long haul optical communication systems. The following sections of this chapter will

cover the generation and detection scheme of DQPSK. Experimental data will be presented

for 20 Gbit/s DQPSK and 40 Gbit/s DQPSK PolDM transmission system showing the re-

ceiver sensitivity and eye diagrams. Also the CD tolerance for 20 Gbit/s RZ-DQPSK will

be evaluated.

3.2 Differential Quadrature Phase Shift Keying (DQPSK)

Differential quadrature phase-shift keying (DQPSK) is one of the interesting advanced mod-

ulation formats that has received intense study and is seen as an alternative modulation for-

mat for high bit rate long haul optical transmission systems [85, 69, 13, 70, 86, 87, 88].

DQPSK is a four-level (quaternary) phase modulation format and can be considered as an

extension of the DPSK modulation format introduced in the previous chapter. For DQPSK

two bits are transmitted for each symbol. Each transmitted symbol is mapped into four

(quaternary) possible phase change transitions. Since two bits are transmitted for each sym-

bol, the symbol rate is half of the bit rate which results in reduced spectral occupancy and

bandwidth requirements for the transmitter and receiver components. Another advantage of

DQPSK are the extended chromatic dispersion and PMD tolerances [15]. The high spectral

efficiency of DQPSK can be even doubled by using polarization multiplexing [44, 89]. For

DQPSK, the required OSNR to reach a given BER compared to DPSK measured at the same

bit rate, is increased by about 1-2 dB [90, 15, 91]. Combined with RZ coding its robustness

against non linear effects is increased because the intensity is not modulated by the data but

is rather modulated by a pulse carver [44]. The main components needed to build an optical

DQPSK (oDQPSK) system [13, 69, 92] are the digital precoder, optical encoder and the

optical decoder as shown in the schematic in figure 3.1.

A description for each block of the optical DQPSK system will be given in the following

22 Differential Quadrature Phase Shift Keying (DQPSK)

Laser

Precoder Encoder

Fiber

Decoder

k-1

oDQPSK

Figure 3.1: Schematic of an optical DQPSK transmission system

subsections:

3.2.1 DQPSK Precoder

Due to the differential nature of decoding in DQPSK, a precoding function is required to

provide a direct mapping of the data from the input to the output [13] taking care that the

received data streams are identical to the original transmitted data streams. The precoder

for the DQPSK transmitter can operate at a clock rate which is half of the total transmission

data rate. The precoder block shown in figure 3.1 has four inputs and two outputs. Two of

the four inputs are the input data streams (Ukand Vk) and the other two are the time delayed

version of the precoder outputs (Ik−1and Qk−1). The two output precoded signals (Ikand

Qk) are used to drive the optical encoder resulting in the transmitted symbol (dk). Every

transmitted symbol is coded into one of four possible phase levels, representing one of the

four combinations of the two transmitted signals Ikand Qk. The input signals Ukand Vk

can be demultiplexed from a signal operating at the total transmitted bit rate, or can be two

independent data inputs at half the total transmitted bit rate. For the DQPSK encoder based

on the parallel MZM transmitter, the operation of the precoder is described by the following

set of logic equations [13, 93, 1]:

Ik=UkVkIk−1+UkVkQk−1+UkVkIk−1+UkVkQk−1(3.1)

Qk=UkVkQk−1+UkVkIk−1+UkVkQk−1+UkVkIk−1.(3.2)

A more detailed explanation on how to derive the precoder equations is given in appendix

B. For DQPSK experimental test measurements, the precoding is normally not implemented

in hardware [45]. By transmitting a binary PRBS data stream and a time-shifted binary

PRBS data stream by using an optical DQPSK modulator, the expected received pattern at

the receiver side can be calculated by software means. These calculated patterns are then

used to program a programmable bit error detector (BER) with the corresponding expected

received patterns enabling BER measurements for the received data.

3.2.2 DQPSK Optical Encoder

A preferred and most widely used implementation of the optical encoder is the oDQPSK

modulator based on two parallel MZMs as shown figure 3.2 [13, 86].

3.2 Differential Quadrature Phase Shift Keying (DQPSK) 23

Laser

Modulator

Input

Bias

o(D)QPSK Modulator

λ/4 Path

Length

Difference

Modulator

Output

Optical DQPSK signal

Electrical binary

NRZ data stream (I)

Electrical binary

NRZ data stream (Q)

Optical DPSK signal (I)

Optical DPSK signal (Q)

Figure 3.2: Schematic of an optical DQPSK modulator

The oDQPSK modulator consists of two parallel DPSK modulators that are integrated

together in order to achieve phase stability [15]. These two DPSK modulators are mainly

two Mach-Zehnder modulators (MZMs), placed in two arms of another interferometer that

forms a Mach-Zehnder superstructure. The superstructure has quadrature control electrodes

in both arms for phase control enabling an adjustment of the optical phase difference π

2be-

tween the upper and lower arms. Each of the MZMs are biased for minimum transmission

and driven by a Binary NRZ data signal with peak-to-peak amplitude of 2Vπ. Two binary

NRZ data streams in-phase Iand quadrature Qcan be transmitted simultaneously. One bi-

nary NRZ data signal Iwith a full 2Vπvoltage swing generates in one MZM the in-phase

optical field Re(E)with normalized field amplitudes {-1, +1} which is similar to a differ-

ential phase shift keying (DPSK) signal. The other arm MZM is driven with another binary

NRZ data signal Qwith also a full 2Vπvoltage swing therefore generating the quadrature

optical field component Im(E)with normalized field amplitudes {-1, +1}, which is also

similar to a DPSK signal. The DPSK signal generated in one arm is combined with a π

phase shifted version of the other DPSK signal in the other arm. Since the output field

of the oDQPSK modulator is Eo=Re(E) + jIm(E), the generated fields at the output

are {±1±j} resulting in a four-level phase modulated signal with resulting phase values

within {π

4,3π

4,5π

4,7π

4} and labeled { “00”, “10”, “11”, “01”} respectively. The mapping of

the input signals (Ikand Qk) to the output electric field (Eo) and generated phase state (φk)

of the optical (D)QPSK modulator is shown in table 3.1 . The modulation resulting signal

constellation diagram is shown in figure 3.3.

24 Differential Quadrature Phase Shift Keying (DQPSK)

Table 3.1: Mapping of the input signals (Ikand Qk) to the output electric field (Eo) and

generated phase state (φk) of the o(D)QPSK modulator

IkQkEoφk

0 0 +1 + jπ

0 1 +1 −j7π

1 0 −1 + j3π

1 1 −1−j5π

)Im()Re( EjEE

-1 0 +1

Re(E)

-1

Im(E)

Re(E)

Im(E)

“00”

“01”

“10”

“11”

-1+j

-1-j+1-j

+1+j

Laser

o(D)QPSK Modulator

MZM

MZM π/2

2×V

Figure 3.3: DQPSK signal constellation diagram

3.2 Differential Quadrature Phase Shift Keying (DQPSK) 25

Figure 3.4 shows a generated 2×10 Gbit/s NRZ-DQPSK signal eye diagram generated

using the same DQPSK modulator given in figure 3.2. It can be noticed that the NRZ-

DQPSK signal contains residual intensity dips at the transition between two symbols. The

width of the intensity dips depends on the drive signal voltage and bandwidth [15]. The

strong intensity dip is a result of the optical power drop to zero when both Iand Qchange

value at the same time (e.g., ”00” ←→ ”11”). The smaller intensity dip is a result of

the optical power drop to half when either Ior Qchanges value (e.g., ”00” ←→ ”01” or

”00” ←→ ”10”) [45]. Since DQPSK modulation encodes the information in the optical

phase rather than in the amplitude, the intensity dip has negligible effect on the transmitted

data, especially for RZ-DQPSK signals where the pulse carver largely eliminates the resid-

ual dips [15]. The spectrum of the generated DQPSK signal is just as broad as for a single

DPSK modulation signal although the capacity is doubled [45].

Power drops to half when either I and Q change value Power drops to zero when both I and Q change value

P/2

Re(E)

Im(E)

“00”

“10”

“01”

“11”

Im(E)

“00”

“10”

“01”

“11”

Re(E)

“00” “11”

“00” “01”

“00” “10”

Figure 3.4: NRZ-DQPSK signal eye diagram showing the intensity dips

With the use of a digital precoder before the optical encoder (DQPSK modulator) at the

transmitter side the original input data signals (Ukand Vk) corresponding to a transmitted

symbol (dk) will map to the phase changes (4φk) as shown in table 3.2 . The phase change

4φkcorresponds to the phase difference between the current phase (φk) generated by (Ik

and Qk) driving the DQPSK modulator and the previous phase (φk−1) generated by (Ik−1

and Qk−1) driving the DQPSK modulator. It contains the transmitted information that can

be extracted at the receiver side by the DQPSK decoder which is used to demodulate the

received signal.

Table 3.2: Mapping of data signals (Ukand Vk) to transmitted symbol (dk), and correspond-

ing phase changes (4φk) for DQPSK modulation

UkVkdk4φk

1 1 0 0

1 0 1 π

0 0 2 π

0 1 3 3π

26 Differential Quadrature Phase Shift Keying (DQPSK)

3.2.3 DQPSK Optical Decoder

The DQPSK signals can be decoded optically using an optical delay and add interferometer

structure. To simultaneously receive the two transmitted data streams, the decoder needs two

Mach-Zehnder delay interferometers (MZDI) and two balanced detectors which is similar

to using two DPSK demodulators as shown in figure 3.5 [13]. Both interferometers need a

1 bit delay in one of their arms. The differential optical phase between the interferometer

arms is to be set to +π/4and −π/4for the upper and lower branches respectively in order to

receive the two transmitted data streams [15]. If the input signal to the MZDI has the form

Eoe−j(ωo+4φk), the output signals after balanced detection for each MZDI are proportional to

[cos (4φk) + sin (4φk)] and [cos (4φk)−sin (4φk)], respectively where 4φkis the phase

difference between consecutive bits [13, 45, 44, 94]. The balanced detection used after each

interferometer provides a 3dB improvement in the receiver sensitivity [14].

1-bit delay

MZDI

Balanced

Receiver 1

Data

Stream 1

Data

Stream 2

Input

Signal

(DQPSK)

1-bit delay

MZDI Balanced

Receiver 2

-π/4

+π/4

Figure 3.5: Schematic of an optical DQPSK decoder

The received phase modulated DQPSK signal is converted into an amplitude modulated

signal after demodulation. Figure 3.6 shows the eye diagram of a demodulated 2×10 Gbit/s

NRZ-DQPSK signal after balanced detection. Standard clock and data recovery circuits can

be used to extract the clock and data from the demodulated signal [44, 13, 45]. DQPSK

demodulators suffer from stability problems due to small temperature drift in the interfer-

ometer may cause a phase change between the two arms. Therefore, the interferometer

needs to be temperature stabilized [45].

Figure 3.6: Eye diagram of a demodulated data signal from an NRZ-DQPSK signal

Table 3.3 illustrates the different digitally precoded, optically encoded and optically

decoded DQPSK signals for two transmitted binary data streams Uk“01010011” and Vk

3.3 DQPSK Transmission System at 2×10 Gbit/s 27

“10011010” in the optical DQPSK system shown in figure 3.1. The precoder function is

based on the precoder equations 3.1 and 3.2. The optical encoder is based on the parallel

MZM oDQPSK modulator shown in figure 3.2. The decoder is based on the two MZDI and

two balanced receivers as shown in figure 3.5. For the precoder both the initial values of

Ikand Qkat time instant k=−1were set to logic “0”. It can be seen that the received

data is identical to the transmitted data, therefore verifying that the precoding and decoding

functions are correct.

Table 3.3: Transmitted and received data bit streams Uk= “01010011” and Vk= “10011010”

for an optical DQPSK system

Time instant k-1 0 1 2 3 4 5 6 7

Binary data Uk0 1 0 1 0 0 1 1

Binary data Vk1 0 0 1 1 0 1 0

Precoded data Ik0 0 0 1 1 1 0 0 0

Precoded data Qk0 1 0 1 1 0 1 1 0

I/P voltage to MZM (I) -Vπ-Vπ-Vπ+Vπ+Vπ+Vπ-Vπ-Vπ-Vπ

I/P voltage to MZM (Q) -Vπ+Vπ-Vπ+Vπ+Vπ-Vπ+Vπ+Vπ-Vπ

O/P Electrical Field E0+1+j +1-j +1+j -1-j -1-j -1+j +1-j +1-j +1+j

Transmitted phase φkπ

7π

5π

3π

7π

Phase difference 4φ3π

2π03π

2π0π

Received data uk0 1 0 1 0 0 1 1

Received data vk1 0 0 1 1 0 1 0

3.3 DQPSK Transmission System at 2×10 Gbit/s

Figure 3.7 shows the details of an experimental DQPSK transmission setup used to generate

and detect 2×10 Gbit/s NRZ-, RZ-DQPSK signals. NRZ-DQPSK signals are generated

when the RZ modulator is bypassed. RZ-DQPSK signals can be generated by placing the RZ

modulator (pulse carver) before or after the DQPSK modulator. RZ coding is widely used

in optical communication systems due to its superior performance and improved receiver

sensitivity [45, 16]. RZ-DQPSK is known to have an approximately 2 dB advantage in the

OSNR over NRZ-DQPSK [45].

A 192.5 THzDFB laser was used at the transmitter. A fiber-pigtailed LiNbO3oDQPSK

Modulator (from Photline) with a structure similar to that given in figure 3.2 was used to

generate the DQPSK signals. A pattern generator was used to generate a 10 Gbit/s Pseudo

Random Bit Sequence (PRBS) Binary NRZ data stream. The 10 Gbit/s NRZ data stream

was split and delayed by 31 bit durations to emulate two decorrelated patterns (Iand Q).

Each MZM modulator in both arms of the oDQPSK modulator was biased at minimum

transmission and driven by the two 10 Gbit/s binary NRZ data stream patterns (Iand Q)

with each signal having a 2Vπsignal peak-to-peak voltage amplitude corresponding to 10V

(Vπ≃5V). No DQPSK precoder was implemented. The intensity eye diagram of the

28 Differential Quadrature Phase Shift Keying (DQPSK)

10Gbit/s binary

NRZ data stream

RZ-DQPSK

DFB

Laser Modulator

output

QPSK Modulator

(I) Data

Stream 1

(Q) Data

Stream 2

Fiber

MZDI

Demodulator

100 ps

Data

BER detector

10 GHz

RZ Mod.

192.5 THz

DEMUX 192.5 THz

31 bit

Delay

Pre-Amp. EDFA

Oscilloscope

APC

EDFA

BPF

NRZ-DQPSK

Figure 3.7: Experimental 2×10 Gbit/s DQPSK transmission setup

generated 2×10 Gbit/s NRZ-DQPSK signal is shown in figure 3.7 (top-left). The DQPSK

modulator was followed by a LiNbO3Mach-Zehnder modulator. The latter was biased

at the quadrature point and driven by a 10 GHz clock signal with peak-to-peak voltage

amplitude of Vπto generate 50% duty cycle Return-to-Zero (RZ) pulses, thereby completing

a 2×10 Gbit/s RZ-DQPSK signal. The intensity eye diagram of the generated 2×10 Gbit/s

RZ-DQPSK signal is shown in figure 3.7 (top-right).

The receiver employed an Erbium Doped Fiber Amplifier (EDFA) as an optical pream-

plifier followed by a 40-channel Dense Wavelength Division Multiplexing (DWDM) Arrayed-

Waveguide Grating (AWG) Demultiplexer (DEMUX) with 100 GHzchannel spacing and

Gaussian top characteristics. The DEMUX acted as a narrow bandpass optical filter to

remove the broadband optical noise. After the DEMUX another EDFA operated in an auto-

matic power control (APC) mode was used to amplify the received signal to a level where

it can be detected with high sensitivity. The received signal was then split to the demodu-

lator and an optical front end based on a p-type intrinsic n-type (PIN) photodiode followed

by a Trans-Impedance Amplifier (PIN-TIA). This photo-receiver was used to stabilize a

feedback loop (not shown) that controlled the pump current of the last EDFA for automatic

power control (APC). The other part of the received RZ-DQPSK signal after the last BPF

was connected to the demodulator which was an integrated Planar Lightwave Circuit (PLC)

Mach-Zehnder interferometer (from NEL) with a 100 ps (1 bit) differential delay. The NEL

PLC MZDI was based on silica waveguide technology and was temperature stabilized using

a PI controller. By adjusting the differential optical phase between the interferometer arms

by controlling the differential micro-heaters of the MZDI, it was possible to select either of

the transmitted patterns Iand Q. The demodulator outputs were connected to two optical

front ends (PIN-TIA) (from NORTEL) having a bandwidth (BW) of 11 GHz and photodi-

ode responsitivity (R) of 0.88 A/W. The outputs of the two optical front ends were then

subtracted by connecting them to a high sensitivity differential limiting amplifier iT3011

(from iTerra). The outputs of the differential amplifier were connected to an oscilloscope

for signal monitoring and a bit error rate (BER) detector for measuring the BER and re-

ceiver sensitivity. To allow BER measurements, the error detector was programmed with

the expected data sequence received at the output of the demodulator.

3.3 DQPSK Transmission System at 2×10 Gbit/s 29

Figure 3.8 shows the demodulated eye diagram measured at the output of the differential

amplifier for the Iand Qchannels extracted from a 2×10 Gbit/s NRZ-DQPSK signal gen-

erated with PRBS lengths of 27−1and 215 −1for the top (PRBS-7) and bottom (PRBS-15)

eyes, respectively.

I (PRBS-7) Q

I (PRBS-15) Q

Figure 3.8: Eye diagrams of a demodulated data signal Iand Qextracted from a 2×10 Gbit/s

NRZ-DQPSK signal generated using a Photline DQPSK modulator with PRBS lengths of

27−1(PRBS-7) (top) and 215 −1(PRBS-15) bottom

The measured BER curves versus the received optical power measured at the input of

the optical preamplifier for both the received patterns Iand Qare plotted in figure 3.9 for

a 2×10 NRZ-DQPSK signal generated with PRBS lengths of 27−1(PRBS 7) and 215 −1

(PRBS 15). Figure 3.10 shows the average BER of the received 2×10 NRZ-DQPSK data

streams Iand Qversus the received optical power for both PRBS 7 and PRBS 15. From the

curve in figure 3.10 the measured receiver sensitivity for the 20 (2×10) Gbit/s NRZ-DQPSK

signal at a BER of 10−9was ~-36.8 dBm and ~-36.4 dBm for both PRBS 7 and PRBS 15,

respectively.

30 Differential Quadrature Phase Shift Keying (DQPSK)

-46 -45 -44 -43 -42 -41 -40 -39 -38 -37 -36 -35

-11

-9

-7

-5

-3

Received Optical Power [dBm]

BER

PRBS 7 (I)

PRBS 7 (Q)

PRBS 15 (I)

PRBS 15 (Q)

Figure 3.9: BER curves for the Iand Qreceived data streams versus received optical power

for a 2×10 Gbit/s NRZ-DQPSK signal using a Photline oDQPSK Modulator with PRBS

lengths of 27−1(PRBS 7) and 215 −1(PRBS 15)

-46 -45 -44 -43 -42 -41 -40 -39 -38 -37 -36 -35

-11

-9

-7

-5

-3

Received Optical Power [dBm]

Average BER

PRBS 7

PRBS 15

Figure 3.10: Average BER versus received optical power for a 20 Gbit/s NRZ-DQPSK

signal using a Photline oDQPSK Modulator with PRBS lengths of 27−1(PRBS 7) and

215 −1(PRBS 15)

3.3 DQPSK Transmission System at 2×10 Gbit/s 31

Figure 3.11 shows the demodulated eye diagram measured at the output of the differ-

ential amplifier for the Iand Qchannels extracted from a 2×10 Gbit/s RZ-DQPSK signal

generated with PRBS lengths of 27−1and 215 −1, for the top (PRBS-7) and bottom (PRBS-

15) eyes respectively.

I (PRBS-7) Q

I (PRBS-15) Q

Figure 3.11: Eye diagrams of a demodulated data signal Iand Qextracted from a 2×10

Gbit/s RZ-DQPSK signal generated using a Photline DQPSK modulator with PRBS lengths

of 27−1(PRBS-7) (top) and 215 −1(PRBS-15) bottom

The measured BER curves versus the received optical power measured at the input of

the optical preamplifier for both the received patterns Iand Qare plotted in figure 3.12 for

a 2×10 Gbit/s RZ-DQPSK signal generated with PRBS lengths of 27−1(PRBS 7) and

215 −1(PRBS 15). Figure 3.13 shows the average BER of the received 2×10 Gbit/s RZ-

DQPSK data streams Iand Qversus the received optical power for both PRBS 7 and PRBS

15. From the curve in figure 3.13 the measured receiver sensitivity for the 20 (2×10) Gbit/s

RZ-DQPSK signal at a BER of 10−9was ~-38.3 dBm and ~-37.3 dBm for both PRBS 7 and

PRBS 15, respectively. In [14] the receiver sensitivity for 2×10 Gbit/s RZ-DQPSK signal

with PRBS length of 29−1was measured ~-34.3 dBm.

From our measurements RZ-DQPSK had an improved receiver sensitivity over NRZ-

DQPSK by about (1-1.5) dB. It has been reported that the receiver sensitivity would be

improved by 2-3 dB depending on the exact implementation, but for theoretically optimal

DQPSK signals the OSNR sensitivity difference between RZ and NRZ pulse shaping is

limited to about 1dB [95].

32 Differential Quadrature Phase Shift Keying (DQPSK)

-47 -46 -45 -44 -43 -42 -41 -40 -39 -38 -37 -36

-11

-9

-7

-5

-3

Received Optical Power [dBm]

BER

PRBS 7 (I)

PRBS 7 (Q)

PRBS 15 (I)

PRBS 15 (Q)

Figure 3.12: BER curves for the Iand Qreceived data streams versus received optical

power for a 2×10 Gbit/s RZ-DQPSK signal using a Photline oDQPSK Modulator with

PRBS lengths of 27−1(PRBS 7) and 215 −1(PRBS 15)

-47 -46 -45 -44 -43 -42 -41 -40 -39 -38 -37 -36

-11

-9

-7

-5

-3

Received Optical Power [dBm]

Average BER

PRBS 7

PRBS 15

Figure 3.13: Average BER versus received optical power for a 20 Gbit/s RZ-DQPSK signal

using a Photline oDQPSK Modulator with PRBS lengths of 27−1(PRBS 7) and 215 −1

(PRBS 15)

3.3 DQPSK Transmission System at 2×10 Gbit/s 33

Figure 3.14 is the chromatic dispersion (CD) tolerance curve showing the required opti-

cal signal-to-noise ratio (OSNR) for a BER of 10−9versus the residual chromatic dispersion.

A fiber-pigtailed Bookham GaAs/AlGaAs oDQPSK modulator was used to generate the

DQPSK signals in the RZ-DQPSK setup shown in figure 3.7. The measured CD tolerance

for the 20 Gbit/s RZ-DQPSK signal corresponding to a 1-dB OSNR penalty at a BER of

10−9was measured to be ~360 ps/nm.

0100 200 300 400 500 600 700

Chromatic Dispersion [ps/nm]

OSNR [dB/0.1nm]

Figure 3.14: OSNR needed for a BER of 10−9versus CD for a 20 Gbit/s RZ-DQPSK signal

Figure 3.15 shows the intensity measurements versus time for a 2×10 Gbit/s RZ-DQPSK

signal measured before the MZDI Demodulator of figure 3.7 after different lengths of stan-

dard single mode fiber, corresponding to different values of chromatic dispersion.

34 Differential Quadrature Phase Shift Keying (DQPSK)

Figure 3.15: 20 Gbit/s RZ-DQPSK signal intensity measurements after transmission over

different lengths of SSMF corresponding to different values of CD, (a) Back-to-Back (0

ps/nm), (b) 21.3 km (~362 ps/nm), (c) 32.2 km (~547 ps/nm), and (d) 41.5 km (~706 ps/nm)

3.4 DQPSK PolDM Transmission System at 2×2×10 Gbits/s

Combining DQPSK modulation and Polarization division multiplex (PolDM) results in a

doubled transmission rate, improved spectral efficiency and chromatic dispersion tolerance

[89, 96, 44, 97, 98, 99, 100, 87]. Figure 3.16 shows an RZ-DQPSK PolDM setup operating

at 2×2×10 Gbit/s. At the transmitter a MZM biased at the quadrature point and driven with a

10 GHz clock signal was used to generate 50% RZ pulses. The optical signal is then split into

two polarizations (X and Y) through the polarization beam splitter (PBS), and each signal is

then modulated by a Photline DQPSK modulator driven by two 10 Gbit/s PRBS data streams

with 31 bit mutual delay resulting in the generation of an optical DQPSK signal similar to

that generated in section 3.3.2. The two resultant DQPSK signals having a mutual delay of

22 bit are then combined in orthogonal polarizations in a polarization beam combiner (PBC).

The resultant signal is a 2×2×10 Gbit/s RZ-DQPSK PolDM signal. If the RZ modulator is

removed or switched off, NRZ-DQPSK PolDM signal will be generated. Figure 3.17 shows

the intensity eye diagrams measured at the output of the DQPSK PolDM setup for 2×2×10

Gbit/s NRZ-DQPSK PolDM signals (left), and RZ-DQPSK PolDM signals (right).

3.4 DQPSK PolDM Transmission System at 2×2×10 Gbits/s 35

DFB

Laser

QPSK Modulator A

Fiber

MZDI

Demodulator

100 ps

Data

Oscilloscope

192.5 THz

DEMUX 192.5 THz

Pre-Amp. EDFA

BER Detector

APC

EDFA

BPF

10 Gbit/s binary NRZ

data stream B

31 bit

Delay

Iy Qy

10 Gbit/s binary

NRZ data stream A

31 bit

Delay

Ix Qx

QPSK Modulator B

PBS PBC

2×2×10 Gbit/s

DQPSK signal

with PolMux

Polarizer

RZ Mod.

10 GHz

Clock

Figure 3.16: 2×2×10 Gbit/s RZ-DQPSK PolDM transmission setup

Figure 3.17: Eye diagrams of 2×2×10 Gbit/s NRZ-DQPSK PolDM signal (left), and RZ-

DQPSK PolDM signal (right)

36 Differential Quadrature Phase Shift Keying (DQPSK)

To extract the information from the received DQPSK PolDM signals a DQPSK receiver

similar to that in section 3.3 was used with a polarization controller and polarizer placed

before the last APC EDFA in order to select the received polarization as shown in figure

3.16. In a practical implementation, a PBS would be used instead of the Polarizer in order to

receive the signals in both polarizations simultaneously where each signal will be then con-

nected to a separate DQPSK receiver to extract the transmitted information. The measured

BER curves versus the received optical power measured at the input of the optical pream-

plifier for the received patterns Iand Qin both polarizations X and Y for 2×2×10 Gbit/s

NRZ-DQPSK PolDM signal with PRBS 7 are plotted in figure 3.18. Figure 3.19 shows the

demodulated eye diagrams measured at the sensitivity limit for the Iand Qchannels in both

polarizations X and Y extracted from the 2×2×10 Gbit/s NRZ-DQPSK PolDM signal with

PRBS 7.

-42 -41 -40 -39 -38 -37 -36 -35 -34 -33 -32

-11

-9

-7

-5

-3

Received Optical Power [dBm]

BER

I-X

Q-X

I-Y

Q-Y

Figure 3.18: BER curves for the Iand Qreceived data streams in both polarizations X and

Y versus the received optical power for 2×2×10 Gbit/s NRZ-DQPSK PolDM signal with

PRBS 7

The measured BER curves versus the received optical power measured at the input of

the optical preamplifier for the received patterns Iand Qin both polarizations X and Y for

2×2×10 Gbit/s NRZ-DQPSK PolDM signal with PRBS 15 are plotted in figure 3.20. Figure

3.21 shows the average BER of the received data streams Iand Qin both polarizations

versus the received optical power for 2×2×10 Gbit/s NRZ-DQPSK PolDM signal with

PRBS 7 and PRBS 15. From the curve in figure 3.21 the measured receiver sensitivity for

the 2×2×10 Gbit/s NRZ-DQPSK PolDM signal at a BER of 10−9for PRBS 7 and PRBS

15, was ~-33.8 dBm and ~-32.4 dBm, respectively.

3.4 DQPSK PolDM Transmission System at 2×2×10 Gbits/s 37

I-X Q-X

I-Y Q-Y

Figure 3.19: Eye diagram of the demodulated data signals Iand Qfor both polarizations X

and Y extracted from a 2×2×10 Gbit/s NRZ-DQPSK PolDM signal with PRBS 7

-42 -41 -40 -39 -38 -37 -36 -35 -34 -33 -32 -31

-11

-9

-7

-5

-3

Received Optical Power [dBm]

BER

I-X

Q-X

I-Y

Q-Y

Figure 3.20: BER curves for the Iand Qreceived data streams in both polarizations X and

Y versus the received optical power for 2×2×10 Gbit/s NRZ-DQPSK PolDM signal with

PRBS 15

38 Differential Quadrature Phase Shift Keying (DQPSK)

-42 -41 -40 -39 -38 -37 -36 -35 -34 -33 -32 -31

-11

-9

-7

-5

-3

Received Optical Power [dBm]

Average BER

PRBS 7

PRBS 15

Figure 3.21: Average BER versus received optical power for 2×2×10 Gbit/s NRZ-DQPSK

PolDM signal with PRBS 7 and PRBS 15

When the RZ modulator is used, the measured BER curves versus the received optical

power measured at the input of the optical preamplifier for the received patterns Iand Q

in both polarizations X and Y for 2×2×10 Gbit/s RZ-DQPSK PolDM signal with PRBS 7

are plotted in figure 3.22. Figure 3.23 shows the demodulated eye diagrams measured at the

sensitivity limit for the Iand Qchannels in both polarizations X and Y extracted from the

2×2×10 Gbit/s NRZ-DQPSK PolDM signal with PRBS 7.

Figure 3.24 shows the average BER of the received data streams Iand Qin both polar-

izations versus the received optical power for 2×2×10 Gbit/s RZ-DQPSK PolDM signal.

From the curve in figure 3.24 the measured receiver sensitivity for the 2×2×10 Gbit/s RZ-

DQPSK PolDM signal at a BER of 10−9with PRBS 7 and PRBS 15 was ~-34.7 dBm and

~-34.3 dBm, respectively. In [87] the receiver sensitivity for 2×2×10 Gbit/s RZ-DQPSK

PolDM signal at a BER of 10−9with PRBS 15 was measured ~-30.4 dBm.

3.4 DQPSK PolDM Transmission System at 2×2×10 Gbits/s 39

-42 -41 -40 -39 -38 -37 -36 -35 -34 -33

-11

-9

-7

-5

-3

Received Optical Power [dBm]

BER

I-X

Q-X

I-Y

Q-Y

Figure 3.22: BER curves for the Iand Qreceived data streams in both polarizations X and

Y versus the received optical power for 2×2×10 Gbit/s RZ-DQPSK PolDM signal with

PRBS 7

I-X Q-X

I-Y Q-Y

Figure 3.23: Eye diagram of the demodulated data signals Iand Qfor both polarizations X

and Y extracted from a 2×2×10 Gbit/s RZ-DQPSK PolDM signal with PRBS 7

40 Differential Quadrature Phase Shift Keying (DQPSK)

-42 -41 -40 -39 -38 -37 -36 -35 -34 -33

-11

-9

-7

-5

-3

Received Optical Power [dBm]

Average BER

PRBS 7

PRBS 15

Figure 3.24: Average BER versus received optical power for 2×2×10 Gbit/s RZ-DQPSK

PolDM signal with PRBS 7 and PRBS 15

Figure 3.25 shows a photo of the 2×2×10 Gbit/s DQPSK PolDM transmitter setup based

on two Photline DQPSK modulators and two polarization beam splitters (PBS). Figures

3.26 show the main components used to construct the experimental 2×2×10 Gbit/s DQPSK

PolDM receiver setup including the 2×10 Gbit/s DQPSK receiver. Figure 3.27 shows a

photo of the full 2×2×10 Gbit/s RZ-DQPSK PolDM transmission setup that was used to

perform the BER measurements.

Figure 3.25: 2×2×10 Gbit/s DQPSK PolDM transmitter setup photo

3.4 DQPSK PolDM Transmission System at 2×2×10 Gbits/s 41

Figure 3.26: 2×2×10 Gbit/s RZ-DQPSK Receiver

Figure 3.27: 2×2×10 Gbit/s RZ-DQPSK PolDM transmission setup photo

42 Differential Quadrature Phase Shift Keying (DQPSK)

Chapter 4

Quaternary Intensity Modulation

4.1 Introduction

This chapter will introduce several types of multilevel quaternary modulation formats used

for 2×10 (20 Gbit/s) optical communication systems. These multilevel modulation formats

are Conventional Quaternary Intensity Modulation (4-IM), Quaternary intensity modula-

tion based on combining two binary signals with unequal amplitudes in quadrature phases

(QASK), Quaternary intensity modulation based on combining two Duobinary signals with

unequal amplitudes in quadrature phases (QDB), and Quaternary intensity modulation based

on combining two duobinary signals with unequal amplitudes in orthogonal polarizations

(QPolDB). The generation and detection scheme for each modulation format will be ex-

plained in details with an evaluation of their performances based on experimental results

obtained from their implementation in a practical 20 Gbit/s optical transmission system. The

last section of this chapter will briefly cover other reported quaternary modulation formats

Quaternary Polarization ASK (QPolASK) and Quaternary Differential-Phase ASK Modu-

lation (DP-ASK).

4.2 Conventional Quaternary Intensity Modulation (4-IM)

An electrical binary ASK signal has an amplitude switching between two levels “0” and “1”,

while an electrical multilevel ASK signal (M-ary ASK) has an amplitude switching between

different levels where each level can represent two or more bits of information [101]. The

number of levels required for each transmitted symbol is M = 2n, where nis the number of

bits transmitted for each symbol. If the number of bits transmitted for each symbol is two

(n= 2), then an electrical 4-ary ASK signal (M=4) will result. This electrical 4-level ASK

signal can be used to drive an optical intensity modulator resulting in the generation of an

optical quaternary intensity modulation signal (4-IM) [10, 11, 8, 9, 12].

An advantage of using optical 4-IM signals is that it needs electrical and optical compo-

nents operating at half the bit rate when compared to a binary signal operating at the same

bit rate, which may reduce the cost of the required components of an optical transmission

system. Also a 4-IM signal has a narrower spectrum width than a binary signal (reduced

by a factor of 2), resulting in improved spectral efficiency which can be useful in a DWDM

transmission system [20].

44 Quaternary Intensity Modulation

A disadvantage of using optical 4-IM signals is that it is has an eye closure penalty due

to the more closely spaced levels than binary modulation. This makes it more sensitive

to noise and distortion than a binary signal, therefore requiring larger optical power at the

receiver side [101, 10].

A conventional quaternary 4-IM signal generation and detection scheme used for a dig-

ital optical communication system will be explained in the following subsections.

4.2.1 Generation of Conventional 2×10 Gbit/s Quaternary 4-IM Sig-

nals

A conventional quaternary 4-IM signal is generated by modulating a laser carrier signal with

an electrical 4-level ASK signal (4-ASK) by directly modulating the laser current or using

an external intensity modulator. The electrical 4-ASK signal can be obtained by adding

two electrical binary signals with unequal amplitudes. In order to obtain an equally spaced

electrical 4-ASK signal, the amplitude of one of the original electrical Binary NRZ signals

(Data stream 1) is halved by using an electrical attenuator with 6 dB attenuation, then com-

bined with the second electrical Binary NRZ signal (Data stream 2) in an electrical power

combiner as shown in figure 4.1 (left). The eye diagrams shown in figure 4.1 are; Data

stream 1 (top-left) which is a 10 Gbit/s electrical binary NRZ-ASK signal, Data stream 2

(bottom) which is also a 10 Gbit/s electrical binary NRZ-ASK but with half the amplitude

of data stream 1, and finally the generated 10 Gbaud (2×10 Gbit/s) electrical 4-ary ASK

signal (top-right).

Electrical Binary

NRZ Data Stream (1)

Electrical Binary

NRZ Data Stream (2) 1/2

Combiner

Electrical 4-ary

ASK signal

Figure 4.1: Electrical 4-ary ASK signal generation

The mapping between the input data and corresponding output levels for the 4-ary ASK

signal generated in figure 4.1 is shown in table 4.1.

In practice the original binary data streams should be gray encoded before generating

the quaternary signal. The gray encoder is useful because the most likely error to occur in

the received quaternary signal is between the adjacent levels of the received signal which

only differ by one binary digit, therefore the gray encoder would force only one bit error

when decision errors are made between adjacent levels [101, 65, 64]. Figure 4.2 shows a

common method to generate an optical quaternary intensity signal (4-IM) by driving an ex-

ternal intensity Mach-Zehnder modulator (MZM) biased at the quadrature point (half power

4.2 Conventional Quaternary Intensity Modulation (4-IM) 45

Table 4.1: Mapping of input binary data to output levels of a 4-ary ASK signal

Data Stream 1 Data Stream 1 Output Level

0 0 0

0 1 1

1 0 2

1 1 3

point) with an electrical 4-ary signal having a Vπpeak-to-peak voltage amplitude [10, 20].

Since the intensity transfer function of a MZM modulator normally has a cos2(x)transfer

characteristics (See Appendix A), it is necessary to optimally adjust the amplitude ratio of

the two transmitted electrical binary data streams to obtain an equally spaced 4-level optical

quaternary intensity signal (4-IM) by adjusting the attenuator (b) shown in figure 4.2.

Laser

Binary NRZ

Data Stream 1

Binary NRZ

Data Stream 2

MZM

Combiner

Electrical

4-ary signal

4-level Optical Modulated Signal (4-IM)

Intensity

levels

Output Intensity

Input Voltage

Vπ

Figure 4.2: Principle of optical 4-IM generation using intensity MZM

The eye diagrams shown in figure 4.2 are for a 2×10 Gbit/s electrical 4-ary ASK signal

(bottom-right) and the resulting optical 2×10 Gbit/s quaternary intensity modulation signal

(4-IM) (top-left). The 4-IM signal constellation diagram is shown in figure 4.3. The optical

field Ehas four different constellation points with normalized field amplitudes {0, 1, √2,

√3}, which results into four different intensity levels {“0”, “1”, “2”, “3”}.

4.2.2 Quaternary 4-IM Signal Detection

Figure 4.4 shows a schematic of a four level 4-IM receiver [10]. A single photodiode (di-

rect detection) is used at the receiver to convert the optical quaternary intensity modulated

signal (4-IM) to a 4-ary electrical signal. The received quaternary signal contains three eye

46 Quaternary Intensity Modulation

Re(E)

Im(E)

Figure 4.3: Optical 4-IM signal constellation diagram

openings corresponding to three different patterns {Q1, Q2, Q3} representing the bottom,

middle, and top eyes respectively. These three patterns are separated at the receiver by three

decision circuits (D flip-flops). Suitable decoding is then needed to recover the two original

transmitted data streams from the three detected patterns [101, 10, 6]. The decoding logic

functions used to extract Data stream 1 and Data stream 2 are:

Data Stream 1 = Q2(4.1)

Data Stream 2 = Q3+Q1Q2.(4.2)

The schematic of the decoding logic circuit is also shown in the figure 4.4. A more detailed

explanation showing how to derive the receiver logic equations and decoder circuit is given

in appendix B.

Decision

“2”↔“3”

Clock

Decision

“1”↔“2”

Decision

“0”↔“1”

Decoding Logic

Data

Stream 1

Data

Stream 2

Intensity

Level

0 0 0

1 0 1

2 1 0

3 1 1 Data

Stream 1

Data

Stream 2

Quaternary

Optical Signal

(4-IM)

Figure 4.4: Schematic of a 4-IM receiver with decoding logic diagram

4.2.3 Experimental Transmission System for 2×10 Gbit/s Quaternary

Modulation

Figure 4.5 shows the details of an experimental transmission setup used to generate and

detect a 2×10 Gbit/s quaternary 4-IM signal. The transmitter (Tx) used in the figure is

based on the conventional 4-IM generation scheme shown in figure 4.2. A 193.5 THz DFB

laser was used. A pattern generator was used to generate a 10 Gbit/s pseudo random bit

sequence (PRBS) binary NRZ data stream with a PRBS data length of 27−1. The 10

4.2 Conventional Quaternary Intensity Modulation (4-IM) 47

Gbit/s binary NRZ data stream was split and delayed by 31 bit durations to emulate two

decorrelated patterns (Data stream 1 and Data stream 2). Both data streams were amplified

using 10 GHz optical modulator drivers from JDS-UNIPHASE. The amplitude of one of

the electrical binary NRZ signals (Data stream 1) was attenuated and then combined with

the second electrical binary NRZ signal (Data stream 2) in an electrical power combiner,

therefore generating a 4-ary electrical signal with a peak-to-peak voltage amplitude of Vπ.

No Gray encoding was implemented. The electrical 4-ary signal was used to drive a fiber-

pigtailed LiNbO3Mach-Zehnder intensity modulator (MZM) biased at the quadrature point

(half power point) to generate an optical quaternary intensity signal. Figure 4.2 (top-left)

shows the 4-level eye diagram of the generated optical quaternary intensity modulated sig-

nal, detected at the modulator output at 10 Gbaud (20 Gbit/s). The generated 2×10 Gbit/s

4-IM signal spectrum contains a carrier and is just as broad as for a single binary 10 Gbit/s

NRZ-ASK modulation signal although the capacity is doubled.

The receiver employed in figure 4.5 is similar to that for receiving DQPSK signals in

figure 3.7, but without the MZDI and balanced receiver. Also the error detector was pro-

grammed, using different thresholds, to receive all the three patterns, corresponding to the

top, middle, and bottom eye diagram. Proper BER averaging is performed to represent the

mean BER of the received patterns.

10 Gb/s Binary

NRZ Data stream 1

10 Gb/s Binary

NRZ Data stream 2

193.5 THz

VOA

EDFA

Preamplifier

BER

AWG DEMUX

EDFA

Electrical

Amplifier PIN-TIA

Fiber

EDFA

APC

DFB

Figure 4.5: 20 Gbit/s Quaternary intensity modulation transmission setup

The BER of the three received eye patterns (top, middle, and bottom) versus the received

power measured at the optical preamplifier input is shown in figure 4.6.

The average BER versus the received power measured at the optical preamplifier input

is shown in figure 4.7. The measured receiver sensitivity for the 2×10 Gbit/s 4-IM signal

with a PRBS length of 27−1at a BER of 10−9is ~-13 dBm.

48 Quaternary Intensity Modulation

-31 -29 -27 -25 -23 -21 -19 -17 -15 -13 -11

-11

-9

-7

-5

-3

Optical Power [dBm]

BER

Top

Middle

Bottom

Figure 4.6: BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s (4-IM) signal

-31 -29 -27 -25 -23 -21 -19 -17 -15 -13 -11

-11

-9

-7

-5

-3

Optical Power [dBm]

Average BER

Figure 4.7: Average BER versus received optical power for a 20 Gbit/s (4-IM) signal

4.3 Quaternary Amplitude Shift Keying (QASK) 49

4.3 Quaternary Amplitude Shift Keying (QASK)

In the conventional quaternary intensity modulation (4-IM) generation scheme explained in

the previous section, an optical 4-IM signal was generated by modulating a continuous light-

wave carrier signal with an electrical 4-ary ASK signal that was obtained by combining an

electrical binary signal with another one with reduced amplitude in an electrical RF power

combiner. Such configuration results into transforming any amplitude distortion in the orig-

inal electrical binary signals into significant distortion in the optical 4-IM signal [20]. This

amplitude distortion can result in receiver sensitivity degradation. It is possible to suppress

the amplitude distortion by using a different generation scheme by optically combining one

binary signal and another intensity-halved binary signal in two quadrature phases generating

an optical quaternary ASK signal (QASK) [20, 28, 6, 21]. This generation scheme has been

proposed in [20] to generate a 2×5 Gbit/s QASK signal by using two parallel MZM modu-

lators, an optical phase control section, and an optical intensity control section all integrated

on lithium niobate (LiNbO3) substrate. In the following subsections a 2×10 Gbit/s QASK

signal has been generated with a similar method, but by using an integrated GaAs/AlGaAs

Optical QPSK modulator based on two parallel MZM modulators, and an optical phase

control section, without using any optical intensity control section. The optical QPSK mod-

ulator was driven by two 10 Gbit/s binary NRZ data streams with unequal amplitudes, to

generate an optical 4-constellation-point 20 Gbit/s quaternary intensity signal (QASK). The

receiver sensitivity for the proposed QASK signal generated at 2×5 Gbit/s [20] and 2×10

Gbit/s [6, 21] was found to be improved when compared to a the conventional 4-IM signal

at the same bit rate. The generation scheme, receiver sensitivity and CD tolerance of a 20

Gbit/s QASK modulation format will be discussed in the following subsections.

4.3.1 Generation and Detection of 2×10 Gbit/s QASK Signals

The optical QPSK modulator shown in figure 4.8 contains two Mach-Zehnder modulators

(MZMs), placed in the two arms of another interferometer that forms a Mach-Zehnder super-

structure. The superstructure has quadrature control electrodes in both arms for phase trim-

ming. Throughout the experiments in this section and in the next section, a fiber-pigtailed

Bookham GaAs/AlGaAs o(D)QPSK modulator [92, 13] was used.

The two MZMs are driven by two electrical 10 Gbit/s binary NRZ-ASK signals. The

drive amplitude of the NRZ-ASK signal in one of the MZMs equals Vπ. This generates an

in-phase optical field Re(E)with normalized field amplitudes {0, +1}. The other MZM

is driven by another binary NRZ-ASK signal with voltage amplitude of Vπ/2. This results

into the generation of a quadrature optical field Im(E)with normalized field amplitudes

{0, +a}. The total output electric field is Eo=Re(E) + jIm(E)with the generated

fields are {0, +ja, +1, and +1 + ja} corresponding to four different constellation points as

shown in figure 4.9. The four constellation points results into four different intensities {0,

a2, 1, 1+a2}. For a= 1/√2, corresponding to 6 dB of electrical attenuation (b = 1/2), the

intensities are equidistant and are now labeled as “0”, “1”, “2”, “3” as shown in figure 4.9

(bottom-right). To receive the original transmitted patterns from the QASK signal, decoding

is needed identical to that of the 4-IM decoder shown in figure 4.4.

50 Quaternary Intensity Modulation

DFB

Laser

Modulator

Input

Modulator

Output

Bias

oQPSK Modulator

λ/4 Path Length

Difference

QASK

10 Gb/s Binary

NRZ Data stream 1

10 Gb/s Binary

NRZ Data stream 2

Figure 4.8: Optical QASK (2×10 Gbit/s) generation in an optical QPSK modulator, and

resulting quaternary intensity eye diagram (bottom) with electrical attenuator setting (b =

1/2)

)Im()Re( EjEE

0 +1

Re(E)

Im(E)

Laser

Binary NRZ

Data Stream 1

MZM

MZM π/2

Binary NRZ

Data Stream 2

Re(E)

Im(E)

1+ja

“3”

“2”

“1”

“0”

2/1=a

“3”

“2”

“1”

“0”

Figure 4.9: QASK electrical field signal constellation diagram (top-right) generated using a

QPSK modulator driven by binary signals with unequal amplitudes. The resulting intensity

eye diagram (bottom-right) shows the 4 intensity levels (a= 1/√2).

4.3 Quaternary Amplitude Shift Keying (QASK) 51

4.3.2 Experimental Results Measured for 2×10 Gbit/s QASK

The experimental transmission setup used to generate and detect a 2×10 Gbit/s quaternary

QASK signal was similar to the setup used for generating and detecting the 4-IM signal

shown in figure 4.5 in the previous section. The transmitter used in the setup is the same

given in figure 4.8. The 10 Gbaud (2×10 Gbit/s) QASK signal represents a 4-IM and its

spectrum is just as broad as for a single binary 10 Gbit/s NRZ-ASK modulation signal

although the capacity is doubled. Figure 4.10 shows the heterodyned electrical spectrum of

the 10 Gbaud optical QASK signal. There exists a carrier component in the spectrum which

is identical to that of a 4-IM signal.

7.5 10 12.5 15.0 17.5 20.0 22.5 25.0 27.5

-45

-40

-35

-30

-25

-20

-15

-10

-5

Frequency [GHz]

Spectral Power [dBm]

Figure 4.10: Heterodyned electrical spectrum for a 2×10 Gbit/s QASK signal

Figure 4.11, shows the BER curves for the three received eyes (top, middle, and bottom)

of the 2×10 Gbit/s QASK signal.

The average BER versus the received optical power measured at the optical preamplifier

input is shown in figure 4.12. The measured receiver sensitivity for the 2×10 Gbit/s QASK

signal with a PRBS 27−1at a BER of 10−9is ~-21.6 dBm.

The chromatic dispersion (CD) tolerance was measured for the generated 20 Gbit/s

QASK signal. Figure 4.13 shows the OSNR after the optical preamplifier needed for a BER

of 10−9versus CD. An optical attenuator was used to vary the OSNR. The 1-dB tolerance

at a BER of 10−9was found to be ~130 ps/nm for 20 Gbit/s QASK.

Figure 4.14 shows the intensity eye diagrams of a 20 Gbit/s QASK signal at the sensi-

tivity edge after transmission over 0, 5.34, 10.9, and 16.24 km of SSMF.

52 Quaternary Intensity Modulation

-36 -34 -32 -30 -28 -26 -24 -22 -20

-11

-9

-7

-5

-3

Optical Power [dBm]

BER

Top

Middle

Bottom

Figure 4.11: BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s QASK signal

-36 -34 -32 -30 -28 -26 -24 -22 -20

-11

-9

-7

-5

-3

Optical Power [dBm]

Average BER

Figure 4.12: Average BER versus received optical power for a 20 Gbit/s QASK signal

4.3 Quaternary Amplitude Shift Keying (QASK) 53

050 100 150 200 250 300

Chromatic Dispersion [ps/nm]

OSNR [dB/0.1nm]

Figure 4.13: OSNR needed for a BER of 10−9versus CD for a 20 Gbit/s QASK signal

Figure 4.14: Intensity eye diagrams of a 20 Gbit/s QASK signal measured at the sensitivity

edge after transmission over different length of SSMF, (a) (Back-to-Back) 0 km (0 ps/nm),

(b) 5.34 km (~90.78 ps/nm), (c) 10.9 km (~185 ps/nm), (d) 16.24 km (~276 ps/nm)

54 Quaternary Intensity Modulation

4.4 Quaternary Duobinary Modulation (QDB)

In the previous section on QASK we used an optical QPSK modulator, driven by two binary

10 Gbit/s NRZ data streams with unequal amplitudes, to generate an optical 4-constellation-

point 20 Gbit/s quaternary amplitude shift keying (QASK) signal, which has an identical

spectral bandwidth of a 4-IM signal. It is possible to further increase the spectral efficiency

by replacing the binary data streams by duobinary data streams in order to generate an opti-

cal 9-constellation-point 20 Gbit/s quaternary duobinary (QDB) signal [28, 27, 6, 21]. This

QDB signal has a spectral bandwidth which is about half of that for 4-IM and QASK. The

generation scheme, receiver sensitivity and CD tolerance of the proposed QDB modulation

format at 20 Gbit/s will be discussed in the following subsections.

4.4.1 Generation and Detection of 2×10 Gbit/s QDB signals

The same QASK generation scheme shown in figure 4.8 was used after replacing the bi-

nary data streams with duobinary data streams as shown in figure 4.15. Both MZMs were

biased at minimum transmission. Since differential encoding was not available, two mutu-

ally delayed PRBS 10 Gbit/s binary NRZ electrical signals were lowpass-filtered (LPF) to

generate the duobinary signals. The lowpass filters were constructed using open stubs with

single-path delays of 50 ps. Each stub filter (LPF) responds to an impulse by two impulses

of equal height and 100 ps mutual delay, thereby forming an idealized duobinary one-bit-

delay and add filter [25, 24]. The simulated frequency response of the LPF stub used in this

experiment had a ~ 35dB dip at the frequency 5 GHz. The frequency response of this stub

filter (stub-5) and a comparison with the frequency response of a duobinary Bessel LPF is

given in appendix D.

DFB

Laser Modulator

Input

Modulator

Output

Bias

oQPSK Modulator

λ/4 Path Length

Difference

Differential

encoding

LPF LPF

Differential

encoding

QDB

10 Gb/s Binary

NRZ Data stream 1

10 Gb/s Binary

NRZ Data stream 2

Figure 4.15: Principle of optical QDB modulation generation using duobinary low pass

filtering (LPF) and a QPSK modulator (electrical attenuator setting b = 1/2). The resulting

quaternary intensity eye diagram (bottom) and the electrical duobinary eye diagram (top)

are also shown.

4.4 Quaternary Duobinary Modulation (QDB) 55

A duobinary signal with a full 2Vπvoltage swing generates in one MZM the in-phase

optical field Re(E)with normalized field amplitudes {-1, 0, +1}. The other arm MZM

is driven with a duobinary signal with a Vπvoltage swing and generates the quadrature

component Im(E)with normalized field amplitudes {−a, 0, +a}. The total output field

is Eo=Re(E) + jIm(E). The generated QDB signal constellation diagram is shown in

figure 4.16 (top-right). The generated fields are {0, ±ja,±1, ±1±ja}, corresponding

to 9-constellation points and are mapped into four different intensities {0, a2, 1, 1 + a2}.

Illustrating the case a= 1/√2(b = 1/2), the intensities are again labeled “0”, “1”, “2”, “3”

corresponding to the 4 levels of the generated 10 Gbaud (20 Gbit/s) QDB signal as shown in

figure 4.16 (bottom-right). Not only differential encoding is needed at the transmitter side,

but also 4-IM decoding at the receiver side.

)Im()Re( EjEE

-1 0 +1

Re(E)

-a

Im(E)

Laser

Duobinary

Data Stream 1

MZM

MZM π/2

Duobinary

Data Stream 2

2/1=a

“3”

“2”

“1”

“0”

Re(E)

Im(E)

+ja

“3”

“2”

“3” “3”

“3”

“2”

“1”

“0”

“1”

-ja +1-ja

-1-ja

+1+ja

-1+ja

-1 0

Figure 4.16: QDB electrical field signal constellation diagram (top-right) generated using

a QPSK modulator driven by duobinary signals with unequal amplitudes. The resulting

intensity eye diagram (bottom-right) shows the 4 intensity levels (a= 1/√2).

4.4.2 Experimental Results Measured for 2×10 Gbit/s QDB

The experimental transmission setup used to generate and detect the 2×10 Gbit/s quaternary

QDB signal is similar to that used for 4-IM and QASK signals as shown in figure 4.5. The

transmitter used in the setup is the QDB transmitter given in figure 4.15. The spectrum

of the generated 10 Gbaud (2×10 Gbit/s) QDB signal is just as broad as a single 10 Gbit/s

duobinary modulation signal although the capacity is doubled. Its spectral bandwidth is also

equivalent to half of that for 2×10 Gbit/s 4-IM and QASK signals. Figure 4.17 shows the

heterodyned electrical spectrum of the 10 Gbaud optical QDB signal. There exists no carrier

in the spectrum which is the case for duobinary modulation.

Figure 4.18, shows the BER curves for the three received eyes (top, middle, and bottom)

of the 2×10 Gbit/s QDB signal. The average BER curve versus the received optical power

measured at the input of the receiver optical preamplifier is shown in figure 4.19. The

measured receiver sensitivity for 20 Gbit/s QDB with a PRBS 27−1at a BER of 10−9is

~-21.2 dBm.

56 Quaternary Intensity Modulation

9 11 13 15 17 19 21 23

-40

-35

-30

-25

-20

-15

-10

-5

Frequency [GHz]

Spectral Power [dBm]

Figure 4.17: Heterodyned electrical spectrum of the 20 Gbit/s QDB signal

-37 -35 -33 -31 -29 -27 -25 -23 -21 -19

-11

-9

-7

-5

-3

Optical Power [dBm]

BER

Top

Middle

Bottom

Figure 4.18: BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s QDB signal

4.4 Quaternary Duobinary Modulation (QDB) 57

-37 -35 -33 -31 -29 -27 -25 -23 -21 -19

-11

-9

-7

-5

-3

Optical Power [dBm]

Average BER

Figure 4.19: Average BER versus received optical power for a 20 Gbit/s QDB signal

The chromatic dispersion (CD) tolerance was measured for the generated 20 Gbit/s QDB

signal. Figure 4.20 shows the OSNR after the optical preamplifier needed for a BER of

10−9versus CD. An optical attenuator was used to vary the OSNR. The 1-dB tolerance at a

BER of 10−9was found to be ~140 ps/nm for 20 Gbit/s QDB. Although 20 Gbit/s QDB can

tolerate CD up to ~140 ps/nm at the expense of an OSNR penalty of 1 dB, it can also tolerate

up to ~250 ps/nm for an OSNR penalty of 1.8 dB, while 20 Gbit/s QASK can tolerate CD

up to ~130 ps/nm at the expense of an OSNR penalty of 1 dB, it can only tolerate up to ~250

ps/nm at the expense of an OSNR penalty of 5.8 dB. Therefore QDB signals can tolerate

more CD than QASK.

Figure 4.21 shows a photo of the 2×10 Gbit/s QDB transmitter setup using an optical

QPSK modulator (Bookham), duobinary stub LPFs, and electrical modulator drivers (JDS

Uniphase).

58 Quaternary Intensity Modulation

050 100 150 200 250 300

Chromatic Dispersion [ps/nm]

OSNR [dB/0.1nm]

Figure 4.20: OSNR needed for a BER of 10−9versus CD for a 20 Gbit/s QDB signal

Figure 4.21: Photo of the 2×10 Gbit/s QDB transmitter based on the Bookham QPSK

modulator

4.5 Quaternary Polarization Duobinary Modulation (QPolDB) 59

4.5 Quaternary Polarization Duobinary Modulation (QPolDB)

In the previous section the spectral efficiency for 20 Gbit/s quaternary intensity modula-

tion was improved by generating a 9-constellation point quaternary duobinary signal (QDB)

based on combining two 10 Gbit/s duobinary signals with unequal amplitudes in quadra-

ture phases using a QPSK modulator [27, 6, 21]. It is possible to generate a similar 9-

constellation point quaternary signal by combining two duobinary signals with unequal

amplitudes in orthogonal polarizations instead of quadrature phases by using a polariza-

tion division multiplex setup (PolDM). This 9-constellation point quaternary polarization

duobinary signal (QPolDB) [6, 29, 7] does not require that the two duobinary combined

in orthogonal polarizations must have quadrature phases. A generated 20 Gbit/s QPolDB

modulation signal has high spectral efficiency similar to QDB signals and has better CD tol-

erance compared to standard NRZ-ASK binary modulation, RZ-DQPSK, 4-IM, QASK, and

QDB signals all operating at the same bit rate (20 Gbit/s). The generation scheme, receiver

sensitivity and CD tolerance of the proposed QPolDB modulation format at 20 Gbit/s will

be discussed in the following subsections.

4.5.1 Generation and Detection of 2×10 Gbit/s QPolDB Signals

The schematic of the transmitter used to generate QPolDB signals is shown in figure 4.22.

MZM

PBS

Laser

QPolDB

Differential

Encoding LPF

10 Gbit/s Binary

Data Stream 2 Differential

Encoding LPF

10 Gbit/s Binary

Data Stream 1

Figure 4.22: Schematic of the optical QPolDB signal generation scheme (a= 1/√2)

The laser signal is split into two branches where each branch consists of two Mach-

Zehnder modulators biased at minimum transmission and driven by 10 Gbit/s electrical

duobinary signals having peak-to-peak voltage amplitude of 2Vπ. In one of the branches the

generated optical duobinary signal carries an electric field Exwith normalized amplitudes

of {-1, 0, +1}. The optical duobinary signal in the other branch is optically attenuated by

3 dB (a= 1/√2) resulting in an electric field Eywith normalized electric field amplitudes

of {−a, 0, +a}. The two resultant duobinary optical signals Exand Eycarried in the two

branches are then combined with orthogonal polarizations using a polarization beam splitter

(PBS). Due to the uncertain phase relationship of the two polarizations, the Jones vector

of this QPolDB signal is E=



ejφEy

, where φis the unknown and unimportant phase

difference. The QPolDB 9-point signal constellation diagram is shown in figure 4.23. The

resulting intensities are {0, a2, 1, 1+a2} which corresponds to the symbols “0”, “1”, “2”, “3”

60 Quaternary Intensity Modulation

respectively. To receive the transmitted data at the receiver, not only differential encoding is

needed at the transmitter side, but also 4-IM decoding at the receiver side.













-1 0 +1

-a

(0,a)

“3”

“2”

“3” “3”

“3”

“2”

“1”

“0”

“1”

(0,-a)(1,-a)

(-1,-a)

(1,a)

(-1,a)

(1,0)

(-1,0) (0,0)

MZM

PBS

Laser

10 Gbit/s Duobinary

Data Stream 2

10 Gbit/s Duobinary

Data Stream 1

2/1=a

Figure 4.23: QPolDB electrical field signal constellation diagram (top-right) generated using

a polarization division multiplex setup (left) with duobinary signals (a= 1/√2)

4.5.2 Experimental Results Measured for 2×10 Gbit/s QPolDB

The experimental setup used to generate and detect a 10 Gbaud (2×10 Gbit/s) QPolDB sig-

nal is shown in figure 4.24. For simplicity, only one optical duobinary signal was generated

by passing a 10 Gbit/s binary data stream through a duobinary stub LPF having a frequency

response dip at 5 GHz(Stub-5), similar to that used for QDB generation. The differential

encoding was neither implemented nor needed because a 27−1PRBS data stream was

used. The generated electrical duobinary signal had a 2Vπpeak-to-peak voltage swing and

was used to drive a MZM biased at minimum transmission. At the output of the modula-

tor an optical duobinary signal (Duobinary-5) was generated with an intensity eye diagram

shown in figure 4.24 (top-left). Another duobinary stub LPF, producing impulses spaced

by 83 ps and having a frequency response dip at 6 GHz, was used instead of the previous

stub filter which resulted in the generation of the optical duobinary signal (Duobinary-6)

with the intensity eye diagram shown in figure 4.24 (top-right). The frequency response

of this filter (Stub-6) and a comparison with the frequency response of a duobinary Bessel

LPF is given in appendix D. The generated optical duobinary signal is passed through an

optical amplifier, polarization controller, and a polarizer. The output signal of the polarizer

is split by a coupler into two branches. One of the branches, carrying the duobinary signal

in the x-polarization Ex, delayed the signal by 31 symbol periods in an additional certain

length of standard single mode fiber (SSMF) to decorrelate the two PRBS patterns. The

signal in the orthogonal y-polarization Eyin the other branch was optically attenuated by 3

dB (a= 1/√2). The two electric fields Exand Eyare orthogonally recombined in a sub-

sequent PBS resulting in a 4-level quaternary polarization duobinary signal QPolDB. The

3 dB attenuation in the Eybranch was used to obtain equal eye openings in the intensity

domain for the QPolDB signal.

4.5 Quaternary Polarization Duobinary Modulation (QPolDB) 61

MZM

DFB

Laser

QPolDB

Differential

Encoding

LPF

10 Gbit/s Binary

PRBS Data Stream

Polarizer

Polarization

Controller

31 bit delay

PBS

193.5 THz

VOA

EDFA

(Pre-Amp.)

BER

AWG

DEMUX

(BPF)

EDFA

Electrical

Amplifier

PIN-TIA

Fiber

EDFA

(APC)

QPolDB-5 QPolDB-6

Duobinary-5 Duobinary-6

EDFA

Figure 4.24: Schematic of the transmission setup used to generate and detect 2×10 Gbit/s

QPolDB signals

The eye diagrams of the QPolDB signal generated using the two stub filters are shown

in figure 4.24 (bottom). The extension -5 or -6 of the modulation format acronym QPolDB

denotes the dip frequency of the respective stub filter in GHz. Later it will be seen that

the QPolDB-5 signals yield better sensitivity but worse CD tolerance than the QPolDB-6

signals. Similar experience with differently shaped duobinary signals has been reported in

[102, 24, 83, 82]. Figure 4.25 shows the heterodyned electrical spectrum of the QPolDB-6

signal. There is no carrier, and the bandwidth is equivalent to that of a single Duobinary-6

signal. The spectrum of the QPolDB-5 signal looks the same and has the same bandwidth

as the heterodyned QDB spectrum shown in figure 4.17.

Similar to the experimental setup used in the previous sections the receiver employed an

optical preamplifier followed by a 100 GHzspaced, 40-channel, C-band dense wavelength

division multiplexed (DWDM) demultiplexer (DeMux). This DeMux is of the Gaussian

type and acts as a narrow bandpass optical filter. A variable optical attenuator (VOA) fol-

lowed by a polarization controller placed before the optical preamplifier is used to vary the

optical signal to noise ratio (OSNR) for sensitivity and CD tolerance measurements. The

polarization controller was used to compensate the polarization dependent loss (PDL) of

the optical preamplifier. For an automatic power control (APC), the detected photocurrent

of an optical front end, a PIN photodiode integrated with a transimpedance amplifier (PIN-

TIA), is stabilized by a feedback loop (not shown) that controls the pump current of the last

62 Quaternary Intensity Modulation

10 12.5 15.0 17.5 20.0 22.5 25.0

-25

-20

-15

-10

-5

Frequency [GHz]

Spectral Power [dBm]

Figure 4.25: Heterodyned electrical spectrum of the 20 Gbit/s QPolDB-6 signal

erbium-doped fiber amplifier (EDFA). An electrical amplifier is used to amplify the received

signal before it feeds an oscilloscope or a BER detector. The error detector was programmed

with three different expected patterns corresponding to the top, middle, and bottom eye di-

agrams. To detect these patterns, different decision threshold settings of the error detector

are used. Proper bit-error-rate (BER) averaging is performed to represent the mean BER

of the received patterns. Figure 4.26 and figure 4.27, show the BER curves for the three

received eyes (top, middle, and bottom) of the 20 Gbit/s QPolDB signals using LPF stubs

with 5−GHz dip (QPolDB-5), and 6−GHz dip (QPolDB-6) respectively.

-33 -31 -29 -27 -25 -23 -21 -19

-11

-9

-7

-5

-3

Optical Power [dBm]

BER

Top

Middle

Bottom

Figure 4.26: BER curves versus received optical power for the three received eyes (top,

middle, and bottom) for a 20 Gbit/s QPolDB-5 signal

4.5 Quaternary Polarization Duobinary Modulation (QPolDB) 63

-32 -30 -28 -26 -24 -22 -20 -18 -16

-11

-9

-7

-5

-3

Optical Power [dBm]

BER

Top

Middle

Bottom

Figure 4.27: BER curves for the three received eyes (top, middle, and bottom) for a 20

Gbit/s QPolDB-6 signal

The average BER curves versus the received optical power measured at the optical

preamplifier input for 20 Gbit/s QPolDB-5 and QPolDB-6 signals with PRBS length of

27−1are shown in figure 4.28. The measured receiver sensitivity using a PRBS length of

27−1at an average BER of 10−9for the 20 Gbit/s QPolDB-5 and QPolDB-6 signals are

~-20.5 dBm, and ~-18.4 dBm, respectively.

-33 -31 -29 -27 -25 -23 -21 -19 -17

-11

-9

-7

-5

-3

Optical Power [dBm]

Average BER

QPolDB-5

QPolDB-6

Figure 4.28: Average BER versus received optical power for 20 Gbit/s QPolDB-5 and

QPolDB-6 signals

Figure 4.29 shows the average BER versus the received optical power measured at the

64 Quaternary Intensity Modulation

optical preamplifier input for 20 Gbit/s QPolDB-6 signal with PRBS length of 27−1(PRBS

7), 210 −1(PRBS 10) , and 215 −1(PRBS 15). The measured receiver sensitivity at an

average BER of 10−9for the 20 Gbit/s QPolDB-6 signal with PRBS lengths of 210 −1

(PRBS 10) and 215 −1(PRBS 15) are ~-18 dBm, and ~-13 dBm, respectively.

-31 -29 -27 -25 -23 -21 -19 -17 -15 -13 -11

-11

-9

-7

-5

-3

Received Optical Power [dBm]

Average BER

PRBS 7

PRBS 10

PRBS 15

Figure 4.29: Average BER versus received optical power for 20 Gbit/s QPolDB-6 signal

with PRBS 7, 10 and 15

4.5 Quaternary Polarization Duobinary Modulation (QPolDB) 65

The chromatic dispersion (CD) tolerance was measured for the generated 20 Gbit/s

QPolDB signals (QPolDB-5 and QPolDB-6). Figure 4.30 shows the measured OSNR after

the optical preamplifier needed for a BER of 10−9versus CD for both 20 Gbit/s QPolDB-5

and QPolDB-6 signals. An optical attenuator was used to vary the OSNR. The 1-dB tol-

erances at a BER of 10−9are ~340ps/nm and ~530ps/nm, for QPolDB-5 and QPolDB-6,

respectively.

0100 200 300 400 500 600 700 800

Chromatic Dispersion [ps/nm]

OSNR [dB/0.1nm]

QPolDB-5

QPolDB-6

Figure 4.30: OSNR needed for a BER of 10−9versus CD in ps/nm for 20 Gbit/s QPolDB-5

and QPolDB-6 signals

Figure 4.31 and figure 4.32 show the intensity eye diagrams at the sensitivity edge (BER

=10−9) for 20 Gbit/s QPolDB-5 and QPolDB-6 signals respectively after transmission over

different lengths of SSMF (D~ 17 ps/nm.km). It can be seen that by increasing the fiber

length the chromatic dispersion increases and the quality of the QPolDB signal degrades.

Since QPolDB-6 has better CD tolerance than QPolDB-5 it was possible to measure the

intensity eye diagram at the sensitivity limit for the QPolDB-6 signal after transmission

over 41.54 km of SSMF (~706ps/nm) as shown in figure 4.31 (f).

Figure 4.33 shows a photo of the polarization division multiplex (PolDM) setup used to

generate the 20 Gbit/s QPolDB signals. It consists of 3 polarization controllers, a polarizer,

a 3dB coupler, 2 optical attenuators, and a polarization beam splitter (PBS).

66 Quaternary Intensity Modulation

Figure 4.31: Intensity eye diagrams of the 20 Gbit/s QPolDB-5 signal measured at the

sensitivity edge after transmission over (a) 0 km (back-to-back), (b) 10.9 km, (c) 16.24 km,

(d) 26.65 km, and (e) 37.55 km of SSMF

4.5 Quaternary Polarization Duobinary Modulation (QPolDB) 67

Figure 4.32: Intensity eye diagrams of the 20 Gbit/s QPolDB-6 signal measured at the

sensitivity edge after transmission over (a) 0 km (back-to-back), (b) 10.9 km, (c) 16.24 km,

(d) 26.65 km, (e) 37.55 km, and (f) 41.54 km of SSMF

Figure 4.33: Photo of the PolDM transmitter setup used to generate the QPolDB signals

68 Quaternary Intensity Modulation

4.6 Other Quaternary Multilevel Modulation Formats

In addition to the quaternary modulation formats covered in the previous sections, there exist

also other quaternary multilevel formats that use various combinations of phase- polarization-

and amplitude-shift keying to achieve high spectral efficiency [19, 103, 18]. In the following

subsections we will briefly cover some of the other reported quaternary modulation formats

such as quaternary polarization amplitude shift keying (QPolASK) [23, 22] and quaternary

differential phase amplitude shift keying (DP-ASK) [18, 19]. Both of these modulation for-

mats double the transmission rate and have higher spectral efficiency than binary NRZ-ASK

signals operating at the same bit rate.

4.6.1 Quaternary Polarization Amplitude Shift Keying (QPolASK)

Quaternary polarization amplitude shift keying modulation (QPolASK) is generated by opti-

cally combining two binary intensity modulated signals NRZ-ASK with unequal amplitudes

in orthogonal polarizations [23, 22]. The resulting modulation format is similar to the QASK

[20], but orthogonal polarizations are used to generate the 4-level signal instead of quadra-

ture phases. The schematic of the QPolASK transmitter is similar to the QPolDB transmitter

shown in figure 4.22 but with replacing the duobinary signals with binary signals as shown

in figure 4.34.

MZM

PBS

Laser QPolASK

NRZ-ASK Binary

Data Stream 2

NRZ-ASK Binary

Data Stream 1

Figure 4.34: Schematic of the optical QPolASK signal generation scheme (a= 1/√2), eye

diagram of 2×10 Gbit/s QPolASK signal (top-right).

The laser signal is split into two branches where each branch consists of a Mach-Zehnder

modulator biased at the quadrature point and driven by an electrical binary NRZ-ASK sig-

nals having peak-to-peak voltage amplitude of Vπ. In one of the branches the generated op-

tical binary NRZ-ASK signal carries an electric field Exwith normalized field amplitudes

of {0, +1}. The optical binary NRZ-ASK signal in the other branch is optically attenuated

by 3 dB (a= 1/√2) resulting in an orthogonal electric field Eywith normalized field am-

plitudes {0, +a}. The two generated optical signals Exand Eycarried in the two branches

are then combined with orthogonal polarizations in a polarization beam splitter (PBS) gen-

erating a quaternary polarization amplitude shift keying modulation signal (QPolASK). The

intensity eye diagram of a generated 2×10 Gbit/s QPolASK signal is shown in figure 4.34

(top-right). The QPolASK signal constellation diagram is given in figure 4.35. The QPo-

lASK signal can be detected as a 4-IM with intensities {0, a2, 1, 1 + a2} corresponding to

4.6 Other Quaternary Multilevel Modulation Formats 69

the levels labeled “0”, “1”, “2”, “3”. Similar decoding as in 4-IM is needed at the receiver

side. The spectrum of the QPolASK signal is expected to contain a carrier component and

equivalent to the 4-IM and the QASK spectrum. The spectral bandwidth of the QPolASK is

about twice the bandwidth of the QPolDB signal. The measured receiver sensitivity for the

QPolASK modulation format at 20 Gbit/s is ~-21 dBm.













0 +1

MZM

PBS

Laser

NRZ-ASK Binary

Data Stream 2

NRZ-ASK Binary

Data Stream 1

2/1=a

“3”

“2”

“1”

“0”

(0,a)(1,a)

(1,0)

(0,0)

Figure 4.35: QPolASK electrical field signal constellation diagram (top-right) generated

using a polarization division multiplex setup with binary signals (a= 1/√2)

4.6.2 Quaternary Differential-Phase ASK Modulation (DP-ASK)

It has been reported in [18, 19] on a quaternary differential-phase ASK modulation for-

mat based on independent DPSK and ASK modulation. The schematic of the DP-ASK

transmitter and receiver setup are shown in figure 4.36. DP-ASK can be used to transmit

independently two binary 10 Gbit/s data streams on the same optical carrier frequency by

using two cascaded MZM modulators, one configured for ASK modulation and the other

configured for DPSK modulation [18].

The DP-ASK signal constellation diagram in the optical domain is represented by four

signal points on the real axis of the electric field as shown in figure 4.37 having two different

amplitudes, aand b(b > a > 0), and two possible phase angles 0 and π[19]. The extinction

ratio (ER) of the ASK branch (ER =b/a) must be optimized (~9.5 dB) in order to obtain

equal receiver sensitivity for both the ASK and DPSK signals [18].

To receive the two transmitted data streams, the DP-ASK signal is split into two signals.

One part enters a standard ASK receiver based on a single photodiode, and the other part

enters a standard DPSK receiver based on a MZDI and a balanced detection photodiode

receiver. One drawback of this method is the complicated receiver design and extra cost due

to the need of both a DPSK balanced receiver and an ASK receiver. The reported receiver

sensitivity and optical bandwidth for the DP-ASK modulation format at 20 Gbit/s were ~-30

dBm, and ~ 20 GHz respectively [18].

70 Quaternary Intensity Modulation

MZM

Laser

DP-ASK

Differential

Encoding

NRZ-ASK Binary

Data Stream 1 NRZ-ASK Binary

Data Stream 2

Fiber

Data

Stream 1

Data

Stream 2

MZDI

DPSK Balanced Receiver

MZM

DPSK Mod. ASK Mod.

ASK Receiver

Figure 4.36: Schematic of quaternary DP-ASK transmitter and receiver

Re(E)

Im(E)

-a ab

-b

Figure 4.37: DP-ASK signal constellation diagram

Chapter 5

Results Discussion and Conclusion

5.1 Introduction

This chapter will discuss the experimental results obtained for the multilevel optical mod-

ulation formats obtained from chapter 3 and 4. These modulation formats are DQPSK,

DQPSK PolDM, 4-IM, QASK, QDB, QPolDB-5 and QPolDB-6. A comparison between

the different quaternary modulation formats with standard binary and duobinary modulation

operating at 20 Gbit/s will also be given in a table in the following section. The advantages

and disadvantages of each modulation format will also be highlighted. A brief conclusion

of the overall work will be given in the last section of this chapter.

5.2 Results Discussion

The measured receiver sensitivities at an average BER of 10−9for the quaternary modula-

tion formats NRZ-DQPSK, RZ-DQPSK, 4-IM, QASK, QDB, QPolDB-5, and QPolDB-6

are -36.8 dBm, -38.3 dBm, -12.8 dBm, -21.6 dBm, -21.2 dBm, -20.5 dBm, and -18.4 dBm,

respectively for 2×10 Gbit/s transmitted data with a PRBS length of 27−1. Obviously

the RZ-DQPSK modulation format achieved the best receiver sensitivity compared to the

other modulation formats. The 4-IM had the worst receiver sensitivity. For 2×2×10 Gbit/s

DQPSK PolDM, the receiver sensitivities at an average BER of 10−9for NRZ-DQPSK

PolDM and RZ-DQPSK PolDM signals are -33.8 dBm and -34.7 dBm, respectively with

a PRBS length of 27−1. Figure 5.1 shows the BER curves obtained for all the quater-

nary multilevel modulation format experiments performed in chapter 4 (4-IM, QASK, QDB,

QPolDB-5, and QPolDB-6) with a PRBS length of 27−1.

The effects of very long PRBS pattern lengths (≥223 −1) could not be assessed since

the expected patterns could not be uploaded to the BER detector with the available soft-

ware and memory size. It is believed that the sensitivity for the QDB and QPolDB signals

would be degraded when used with higher PRBS lengths since they are based on duobinary

modulation which is known to be degraded when used with longer PRBS patterns [83, 84].

To confirm that conclusion it was possible to generate and detect a QPolDB-6 signal using

2×10 Gbit/s data streams with a PRBS length of 210−1and 215 −1resulting into a measured

average receiver sensitivity of ~-18 dBm and ~-13 dBm, respectively. It may be possible to

improve the receiver sensitivity of the QPolDB signals by optimizing the generated duobi-

72 Results Discussion and Conclusion

-38 -34 -30 -26 -22 -18 -14 -10

-11

-9

-7

-5

-3

Optical Power [dBm]

Average BER

4-IM

QASK

QDB

QPolDB-5

QPolDB-6

Figure 5.1: Average BER curves versus received optical power for 20 Gbit/s quaternary

modulation signals (4-IM, QASK, QDB, QPolDB-5, and QPolDB-6).

nary signals by filtering it with a narrow bandwidth optical filter [83, 84, 104, 105]. As

for DQPSK signals, the average receiver sensitivity for 2×10 Gbit/s DQPSK signals with a

PRBS length of 215 −1, is measured to be ~-36.4 dBm and ~-37.3 dBm for NRZ and RZ

pulse shaping, respectively. As for 2×2×10 Gbit/s DQPSK PolDM signals with a PRBS

length of 215 −1, the average receiver sensitivity is measured ~-32.5 dBm and ~-34.2 dBm

for NRZ and RZ pulse shaping, respectively.

In terms of chromatic dispersion tolerance, figure 5.2 shows the OSNR needed for an av-

erage BER of 10−9versus CD in ps/nm for 20 Gbit/s quaternary modulation signals (QASK,

QDB, QPolDB-5, and QPolDB-6). The measured 1 dB CD tolerance of the multilevel mod-

ulation formats at 20 Gbit/s for RZ-DQPSK, QASK, QDB, QPolDB-5, and QPolDB-6 are

360 ps/nm, 130 ps/nm, 140 ps/nm, 340 ps/nm, and 530 ps/nm respectively. The CD toler-

ance measurement was not possible for the 4-IM signal due to high noise and poor receiver

sensitivity.

The QPolDB-6 modulation format had the best CD tolerance compared to the other

modulation formats while the QASK scheme had the worst CD tolerance. This is due to the

reduced bandwidth of the duobinary modulation used for the QPolDB signals. The analog

QDB signal using duobinary signals with the QPSK modulator had a better sensitivity but

a worse CD tolerance. The reduced CD tolerance of the QDB scheme may be understood

from the fact that CD affects the phase. On the other hand QPolDB is of course affected

by polarization-dependent loss, since unbalanced optical powers in the orthogonal polariza-

tions result in unequal eye openings at the receiver side. The QDB signal is also affected

by the fact that one of the Mach-Zehnder modulators is driven with a Vπvoltage swing. In

the presence of electrical intersymbol interference this results in a non-optimal optical gen-

erated duobinary signal in the modulator. A solution to this problem is to place an optical

attenuator in one of the QPSK modulator branches [20] and to drive both Mach-Zehnder

modulators with a full 2Vπvoltage swing, where optical duobinary signal quality is best.

5.2 Results Discussion 73

0100 200 300 400 500 600 700

Chromatic Dispersion [ps/nm]

OSNR [dB/0.1nm]

QASK

QDB

QPolDB-5

QPolDB-6

Figure 5.2: OSNR needed for a BER of 10-9 versus CD in ps/nm for 20 Gbit/s quaternary

modulation signals (QASK, QDB, QPolDB-5, and QPolDB-6).

The modulation amplitudes in all the quaternary generation schemes were adjusted to opti-

mize the 3 eye openings for the quaternary signal. Any change in the modulation amplitudes

will cause unequal eye openings in the quaternary signal. It was also found that the 5−GHz

stub generated QPolDB-5 signals have better sensitivity but worse CD tolerance than the

QPolDB-6 signals generated by using the 6−GHz stub. This is due to the fact that two

different methods were used to generate the two duobinary signals. The QPolDB-5 signals

were similar to the duobinary signals generated by the conventional one-bit-delay and add

method [24, 25, 26]. The QPolDB-6 signals were similar to the duobinary signals generated

by the LPF method. It has been reported in [24, 82, 83] that the duobinary signals based on

the LPF method better suppresses the optical spectrum side-lobes, tolerates more CD than

the one bit delay method, but has a worse back-to-back sensitivity than the one-bit-delay

and add method. A similar behavior was obtained for the QPolDB-5 and QPolDB-6 signals.

Table 5.1 compares different modulation formats at 20 Gbit/s in terms of receiver sen-

sitivity, CD tolerance and optical bandwidth occupancy. For comparability all quaternary

modulation experiments in table 5.1 refer to a PRBS length of 27−1.

From table 5.1 it is can be concluded that the 20 Gbit/s standard binary NRZ-ASK sig-

nal had the largest spectral bandwidth and therefore has worse spectral efficiency than all

the other compared multilevel modulation formats operating at the same bit rate. The best

spectral bandwidth was for QDB, QPolDB, and DQPSK PolDM signals. In terms of CD

tolerance, DQPSK PolDM, duobinary modulation, then QPolDB-6 modulation had the best

CD tolerance compared to all the other modulation formats operating at the same bit rate. In

terms of receiver sensitivity, RZ-DQPSK and RZ-DQPSK PolDM had the best receiver sen-

sitivity. The merits of DQPSK and DQPSK PolDM signals come at the cost and complexity

of the extra components at the receiver (interferometer-based balanced photoreceiver with

two photodiodes, polarization controller).

74 Results Discussion and Conclusion

Table 5.1: Comparison between different modulation formats at 20 Gbit/s

Mod. Format (@ 20 Gbit/s) Sensitivity (dBm) CD tol. (ps/nm) BW (GHz) Reference

NRZ-ASK -31.6 ~260 ~40 [48]

Duobinary -28.5 600 ~20 [24]

NRZ-DQPSK -36.8 - ~20 (measured)

RZ-DQPSK -38.3 360 >20 (measured)

NRZ-DQPSK PolDM ~-36.8 - ~10 (scaled)

RZ-DQPSK PolDM ~-37.7 ~1440 >10 (scaled)

4-IM -12.8 - ~20 [6]

QASK -21.6 130 ~20 [6]

QDB -21.2 140 ~10 [6]

QPolDB-5 -20.5 340 ~10 [6]

QPolDB-6 -18.4 530 ~12 [6]

QPolASK -21 - ~20 (measured)

DP-ASK -30 - ~20 [18]

5.3 Conclusion

The two duobinary QDB and QPolDB schemes which require only a single photodiode

as a receiver are believed to represent intensity modulation with the narrowest reported

spectrum reported to date ( 10GHz). The polarization division multiplex based duobinary

QPolDB-6 scheme doubles the transmission capacity and is believed to feature the largest

reported CD tolerance (530 ps/nm) for 20 Gbit/s intensity modulation with such a narrow

spectrum ( 12GHz), which makes it attractive for DWDM systems applied for short reach

(several kilo meters) to medium reach (several hundreds kilometers) transmission applica-

tions (metro applications). However, for long or ultra long haul optical fiber transmission

systems, RZ-DQPSK modulation (with reported receiver sensitivity -38.3 dBm and CD tol-

erance 360 ps/nm at 20 Gbit/s) and Duobinary modulation (with reported receiver sensitivity

-28.5 dBm and CD tolerance 600 ps/nm at 20 Gbit/s) are the potential candidates. Also RZ-

DQPSK PolDM is a promising modulation format for ultra long haul DWDM optical fiber

transmission systems with channels operating at high data rates ≥40 Gbit/s.

Appendix A

Optical Intensity Mach-Zehnder

Modulator (MZM)

Several types of external optical intensity modulators have been developed over the past

several decades for optical fiber communication applications. Most of the modern wide-

bandwidth modulators are based on the linear electro-optic (EO) effect (also known as

Pockels effect) which depends on the applied electric field which makes the modulator a

voltage-controlled device [66]. EO effect denotes the change of the optical refractive in-

dex in nonlinear optical (NLO) crystals due to the presence of the electric field. The index

change leads to change in the optical phase, which can be converted into intensity modula-

tion in a Mach-Zehnder interferometer. Lithium Niobate (LiNbO3) is the most widely used

material for manufacturing electro-optic devices, including phase modulators and Mach-

Zehnder intensity modulators (MZM). The Mach-Zehnder interferometer is the most popu-

lar device for implementing optical intensity modulation using EO effect. Figure A.1 shows

a schematic drawing of such an interferometer .

Arm 1

Arm 2

Splitter Combiner

Optical Input Optical Output

Figure A.1: Schematic drawing of a Mach-Zehnder interferometer

At its optical input port, there is an optical splitter that divides the input optical power

into two equal portions. The divided power propagates in two separate waveguides that

are often called “two arms”. In a MZM, at least one of these two arms is designed as an

EO waveguide, along which the optical phase can be modulated by an applied voltage. If

the optical waves are in phase after propagating through the two arms, they combine as a

single mode in the output optical combiner, which results in a maximum intensity output;

whereas if the optical waves are out of phase after propagating through the two arms, they

76 Optical Intensity Mach-Zehnder Modulator (MZM)

combine as a higher order spatial mode near the optical combiner, therefore most of the

optical power becomes an unguided wave beyond the combiner and the output intensity is

minimum [44, 66]. The optical field amplitude Eoat the output of the MZM can be generally

represented by [44, 66, 1, 35]:

Eo=1

√2E1ejΦ1+E2ejΦ2,(A.1)

where (E1, E2) and (Φ1,Φ2) represent the optical field amplitudes and optical phase

delays in the two arms respectively. The output optical power Pois:

Po=|Eo|2=1

2h|E1|2+|E2|2+ 2 |E1||E2|cos (Φ1−Φ2)i.(A.2)

The phase difference Φ1−Φ2consists of two parts (Φ1−Φ2= ∆Φ + Φ0): one is

the path difference Φ0at zero applied voltage and the other is the phase difference ∆Φ due

to the applied voltage. Assuming that the optical splitter splits the input electric field Ei

into two equal electric fields of Ei/√2each E1=E2=Ei/√2and by neglecting the

waveguide losses and ignoring the arm path delays and constant phase shifts (Φ0= 0 ). The

resulting output electric field of the MZM structure shown in figure A.1 when only arm one

is modulated (Φ2= 0 ,Φ1= ∆Φ) can be written as:

Eo=Ei

2ejΦ1+ 1=Ei

2ejπ V(t)

Vπ+ 1=Ei"cos π

V(t)

Vπ!#ejπ

V(t)

Vπ,(A.3)

where V(t)is the applied voltage to the electrode of arm one, and Vπis the voltage value

at which the voltage-induced phase difference reaches π(or 180o) [66, 1]. However, in ad-

ditional to the change of the optical intensity, the input and output relation of Eq. A.3 also

accompanies with phase modulation of ejφ where φ=π

V(t)

Vπ. The ratio of the phase to am-

plitude modulation is called the chirp coefficient [1]. In addition to chirp, the unmodulated

path (lower arm) of figure A.1 reduces the modulation efficiency of the modulator. When

the microwave electrodes are properly designed, both paths of the Mach-Zehnder modulator

can be modulated to improve the modulator efficiency. Figure A.2 shows the waveguide

structure for a Mach-Zehnder modulator using x−cut LiNbO3crystal. The two paths of

the Mach-Zehnder modulator are phase-modulated with opposite phase shifts in a push-pull

structure. For this type of MZM structure, the Vπis reduced by half with respect to the

MZM structure with only one modulated arm.

Because the two arm paths are modulated with opposite phases of ±πV (t)

2Vπso that Φ1=

−Φ2and the output electric field of the MZM (push-pull) structure is:

Eo=Ei

2ejΦ1+ejΦ2=Ei

2ejπ V(t)

2Vπ+e−jπ V(t)

π=Ei"cos π

V(t)

Vπ!#.(A.4)

From Eq. A.4, it can be concluded that no phase modulation accompanies the amplitude

modulation and therefore this single-drive x-cut push-pull modulator has zero-chirp. In

terms of optical intensity for both of the previous mentioned structures of the Mach-Zehnder

modulator (single arm and push-pull), the transfer characteristic equation (transfer function

Optical Intensity Mach-Zehnder Modulator (MZM) 77

signal ground

ground

Optical waveguides

Figure A.2: Mach-Zehnder modulator (push-pull) structure using x−cut LiNbO3

or optical intensity transmission) as a function of the applied voltage from equations A.3

and A.4 can be written as:

|Eo|2

|Ei|2= cos2 π

V(t)

Vπ!=1

2"1 + cos πV(t)

Vπ!#.(A.5)

The transfer function of the MZM is a nonlinear sinusoidal function as shown in figure

A.3. The voltage to turn the modulator from minimum to maximum transmission points is

Vπ. The input and output transfer characteristic of the MZM modulator may provide 0 and

πphase modulation to an optical signal. For V(t)from 0 to Vπ,Eoand Eihave the same

phase. For V(t)from Vπto 2Vπ,Eoand Eihave opposite phases [1]. Because of some

amount of constant shift (Φ06= 0), the driving voltage V(t)when equal to zero (V(t)=0)

in a practical modulator is not necessarily corresponding to the maximum transmission point

as shown in figure A.3.

Transmission

Input

Voltage (V)

Minimum

Transmission

Point

Maximum

Transmission

Point

Positive inflection

Point

Negative inflection

Point

Figure A.3: Typical transfer characteristic curve of a Mach-Zehnder modulator

Phase modulation provided by zero-chirp modulators have the advantage that amplitude

jitter does not give phase jitter, i.e., variation of drive voltage does not transfer to varia-

78 Optical Intensity Mach-Zehnder Modulator (MZM)

tion of output phase. However, the phase modulation is limited to 0 and π, equivalent to

±1. For phase-modulation operation, to minimize loss and obtain optimal amplitude jitter

compression, the modulator should operate between two maximum transmission points for

a peak-to-peak voltage swing of 2Vπ[1]. Currently, most commercially available LiNbO3

amplitude modulators have a Vπfrom 4 to 6 V. In additional to LiNbO3, semiconductor

materials also have electro-optical effect and can be used to fabricate external modulators

[1, 66]. For LiNbO3MZMs, the most remarkable advantages are the lowest optical loss

and the highest optical power handling capability. Among all types of MZMs, LiNbO3

modulators are still considered as the devices with the best performance. The main dis-

advantage for LiNbO3MZMs is the large size. It also has the bias-drifting issue, which

requires bias control circuit. Also LiNbO3modulators are polarization sensitive, difficult

to be integrated with other components and have a higher fabrication cost in terms of large

volume production.

As for semiconductor GaAs MZMs, its performance is now getting closer to that of

LiNbO3. Also GaAs MZMs have smaller chip sizes, less bias-drifting problem, lower fab-

rication cost and can be integrated with a wide range of components, such as lasers, semi-

conductor optical amplifiers, photodetectors, passive optical circuits and even electronic

drivers. However, GaAs MZMs have a slightly smaller modulation efficiency and higher

optical loss (several dB higher) in comparison to LiNbO3MZMs [66].

Appendix B

DQPSK Precoder

A schematic of a DQPSK transmission setup is shown in figure B.1. The DQPSK signal

carries the information to transmit the two binary data streams U(k)and V(k)on the optical

phase of the laser carrier signal. To receive the two data streams, the DQPSK signal is

demodulated by two interferometers at the receiver side as shown in figure B.1 (right).

Laser

Precoder oDQPSK Mod.

Fiber

DQPSK Receiver

(Decoder + Balanced detection)

U(k)

V(k)

I(k)

Q(k)

I(k-1)

Q(k-1)

oDQPSK

Signal

1-bit delay

MZDI

Balanced

Receiver 1 Data

Stream 1

u(k)

Data

Stream 2

v(k)

1-bit delay

MZDI Balanced

Receiver 2

-π/4

+π/4

DQPSK Transmitter

(Precoder + Optical Encoder)

I(k)=f(U(k),V(k),I(k-1),Q(k-1))

Q(k)=f(U(k),V(k),I(k-1),Q(k-1))

)sin()cos()(

ku φφ ∆+∆=

)sin()cos()(

kv φφ ∆−∆=

))f(I(k),Q(kk=)(φ

)(k(k)∆

1−−= φφφ

Figure B.1: Schematic of a DQPSK transmission system

Both interferometers need a 1 bit delay in one of their arms. The differential optical

phase between the interferometer arms is to be set to +π/4and −π/4for the upper and

lower branches respectively in order to receive the two transmitted data streams [15]. The

received phase difference ∆φ=φ(k)−φ(k−1) is decoded by the DQPSK receiver to two

data bit outputs (u(k)and v(k)) corresponding to the two original transmitted bits (U(k)and

V(k)) by the following relations [13, 94, 45, 44]:

u(k) = cos(4φk) + sin(4φk)(B.1)

v(k) = cos(4φk)−sin(4φk).(B.2)

80 DQPSK Precoder

The mapping of the received phase difference (4φk) to the received data streams (u(k)

and v(k)) are shown in table B.1.

Table B.1: Mapping of the phase change (4φk) to received data (u(k)and v(k)) for DQPSK

signals

Phase difference (4φk)u(k)v(k)

0 1 1

π/21 0

π0 0

3π/20 1

A precoding function is required for the DQPSK system to provide a direct mapping

of the data from the input at the transmitter side to the output at the receiver side after

demodulation [13, 1]. The precoder block shown in figure B.1 (left) has four inputs where

two of the four inputs are the input data streams (U(k)and V(k)) and the other two are the

time delayed version of the precoder outputs (I(k−1) and Q(k−1)). The outputs of the

precoder (I(k)and Q(k)) are used to drive an optical (D)QPSK modulator (optical encoder)

to generate different phase states φ(k) = f(I(k), Q(k)) as shown in figure B.2.

)Im()Re( EjEE

Re(E)

Im(E)

“00”

“01”

“10”

“11”

-1+j

-1-j+1-j

+1+j

o(D)QPSK Modulator

Laser

MZM

MZM π/2

2×V

(π/4)

(3π/4)

(5π/4) (7π/4)

Figure B.2: QPSK modulator and generated phase states φ(k)

The mapping of the optical (D)QPSK modulator input signals (I(k)and Q(k)) to the

output electric field (Eo) and generated phase state φ(k)are shown in table B.2.

Table B.2: Mapping of the input signals (I(k)and Q(k)) to the output electric field (Eo) and

generated phase state (φ(k)) of the o(D)QPSK modulator

I(k)Q(k)O/P Electric Field EoPhase state (φ(k))

0 0 +1 + j π/4

0 1 +1 −j7π/4

1 0 −1 + j3π/4

1 1 −1−j5π/4

Based on the DQPSK receiver decoder equations B.1 and B.2 and by using the informa-

tion obtained from table B.1 and B.2, it is possible to generate a lookup table that maps the

DQPSK Precoder 81

precoder input signals (U(k),V(k),I(k−1), and Q(k−1)) to the two output precoded

signals (I(k)and Q(k)) as shown in table B.3. From table B.3 and by using the Karnaugh

maps shown in figure b.3 it is possible to find the simplified precoder logic equations that

represent I(k)and Q(k)as follows [13, 1, 93]:

I(k) = U(k)V(k)I(k−1)+U(k)V(k)Q(k−1)+U(k)V(k)I(k−1)+U(k)V(k)Q(k−1)

(B.3)

Q(k) = U(k)V(k)Q(k−1)+U(k)V(k)I(k−1)+U(k)V(k)Q(k−1)+U(k)V(k)I(k−1).

(B.4)

Table B.3: DQPSK precoder look up table

∆φkU(k)V(k)I(k−1) Q(k−1) φ(k−1) I(k)Q(k)φ(k)

π0 0 0 0 π/41 1 5π/4

π0 0 0 1 7π/41 0 3π/4

π0 0 1 0 3π/40 1 7π/4

π0 0 1 1 5π/40 0 π/4

3π/20 1 0 0 π/40 1 7π/4

3π/20 1 0 1 7π/41 1 5π/4

3π/20 1 1 0 3π/40 0 π/4

3π/20 1 1 1 5π/41 0 3π/4

π/21 0 0 0 π/41 0 3π/4

π/21 0 0 1 7π/40 0 π/4

π/21 0 1 0 3π/41 1 5π/4

π/21 0 1 1 5π/40 1 7π/4

0 1 1 0 0 π/40 0 π/4

0 1 1 0 1 7π/40 1 7π/4

0 1 1 1 0 3π/41 0 3π/4

0 1 1 1 1 5π/41 1 5π/4

The set of logic equations (B.3 and B.4) describe the operation of the precoder used

for a DQPSK system using on an optical encoder based on the parallel MZM oDQPSK

modulator as shown in figure B.2. The precoder equations can be realized using the logic

circuit diagram shown in figure B.4 . It is also possible to use different circuits than in figure

B.4 that are based on different logic gate arrangements that also follow the truth table B.3.

82 DQPSK Precoder

1001

1100

0110

0011

00 01 11 10

U(k)V(k)

I(k-1)Q(k-1)

I(k) Q(k)

1100

0110

0011

1001

00 01 11 10

U(k)V(k)

I(k-1)Q(k-1)

Figure B.3: Karnaugh-maps for the DQPSK precoder outputs I(k)and Q(k)

I(k-1)

U(k)

V(k)

I(k)

Q(k)

Q(k-1)

Precoder

Figure B.4: DQPSK precoder circuit diagram

Appendix C

4-ary ASK Decoder

Figure C.1 shows a schematic of a 4-ary receiver [10, 101]. The input to the receiver is a

4 level quaternary signal (figure C.1 (left)) that contains three eye openings corresponding

to three different patterns {Q1, Q2, Q3} representing the bottom, middle, and top eyes

respectively. These three patterns {Q1, Q2, Q3} are separated by three decision circuits (D

flip-flops) with the reference voltage on each flip set at the decision levels V1 (0↔1), V2

(1↔2), and V3 (2↔3), respectively. Suitable decoding is then needed to recover the two

original transmitted data streams from the three detected patterns [101, 10, 6].

Level

Decision

“2”↔“3”

Clock

Decision

“1”↔“2”

Decision

“0”↔“1”

Data

Stream 1

(D1)

Data

Stream 2

(D2)

Quaternary 4-level Signal

tDecoding Logic

Level D1 D2

0 0 0

1 0 1

2 1 0

3 1 1

Decision

Threshold

Figure C.1: Schematic of a 4-ary ASK receiver

The mapping of the four input signal levels {0, 1, 2, 3} to the three received patterns

{Q1, Q2, Q3} and the two decoded data streams {D1, D2} that represent the recovered

transmitted patterns are shown in table C.1.

From table C.1 and by using the Karnaugh maps shown in figure C.2 it is possible to

find the simplified decoder logic equations that represent the received data streams D1 and

D2 as follows:

D1 = Q2 (C.1)

D2 = Q3 + Q1Q2.(C.2)

The schematic of the decoding logic circuit is shown in figure C.3.

84 4-ary ASK Decoder

Table C.1: Mapping of the input levels and the 3 received patterns to the decoded outputs

for a 4-ary ASK receiver

Level Q1 Q2 Q3 D1 D2

0 00000

1 10001

2 11010

3 11111

00 01 11 10

11×0

×××0

Q2Q3

01×1

×××0

00 01 11 10

Q2Q3

D1 D2

Figure C.2: Karnaugh-maps for the 4-ary ASK decoder outputs D1 and D2

Figure C.3: 4-ary ASK decoder circuit diagram

Appendix D

Duobinary Filters

There exits two common methods to generate duobinary signals. They are; the electrical

low-pass filtering (LPF) method [24, 79, 80] and the conventional electrical one-bit delay

and add method [24, 25, 26]. The electrical LPF used in generating a duobinary signal is

usually based on an electrical LPF having a 3 dB bandwidth of about 1

4of the of the bit

rate [106, 78, 105]. The one-bit delay-and-add method can be constructed by splitting the

electrical signal by an RF power divider and delaying one of the paths by 1 bit delay and

then combining the two signals together in a RF power combiner. It is also possible to

implement the one-bit delay-and-add duobinary encoder by using a transmission line with

an open circuit (O.C.) stub connected parallel to it (Duobinary Stub filter) as shown in figure

D.1.

k-1

Duobinary one-bit delay

and add Encoder

O.C.

Discontinuity at z=0

Duobinary Stub filter Encoder

= b

+ b

k-1

= V

-j2βl

= V

-j2βl

k-1

-j2βl

Figure D.1: Schematic of the one bit delay-and-add duobinary encoder versus the stub filter

duobinary encoder

At the discontinuity, represented in figure D.1 by an open stub connected in parallel

across the transmission line, the total voltages on either side of the junction must be identical.

Therefore at the plane z=0, taken as a local reference plane Vf+Vr=Vt+ 0 since there

in no wave reflected from the matched load back to the discontinuity, where Vf,Vrand

Vtare the forward, reverse and transmitted traveling wave voltages, respectively. With the

transmission coefficient defined as T=Vt/Vf, it follows that T= 1 + Γ, where Γis

the voltage reflection coefficient just to the left of the open stub. Assuming lossless lines

Γ=Γle−j2βl, where Γl=ZO.C.−Zo

ZO.C.+Zo = 1 is the reflection coefficient at the load side of the

86 Duobinary Filters

open circuit stub, β=2π

λ= 2π(f

υ)is the phase coefficient, λis the wavelength, fis the

frequency, lis the length of the open stub, υ=c/√εris the phase velocity in the medium

with the relative permittivity εr,ZO.C. =ZL=∞is the O.C. load impedance, and Zois

the characteristic impedance [107]. Therefore the transmitted voltage Vt=Vf+Vfe−j2βl =

Vf+Vfe−j2πf(2l

υ).

It was possible to realize a Duobinary Encoder for a 10 Gbit/s data signal by constructing

a stub filter using open stubs with single-path delays of 50 ps ( l

υ= 50ps) by making the

length of the O.C. stub (l=υ×50 = c

√εr×50 ≃8.15mm) where c= 2.997925×108m/s is

the speed of light in free space, and εr= 3.38 is the dielectric constant for the material used

to build the microstrip transmission lines (RO4003 from Rogers). This stub filter responds

to an impulse by two impulses of equal height and 100 ps mutual delay (1 bit =2l

υ= 100ps),

thereby forming an idealized duobinary one-bit-delay and add filter as shown in figure D.2

(a). The simulated frequency response of the LPF stub used in this experiment had a ~

35 dB dip at the frequency 5 GHz (Stub-5) as shown in figure D.3. The eye diagram for

a generated 10 Gbit/s electrical duobinary signal shown in figure D.3 is measured after

filtering a 10 Gbit/s electrical binary NRZ-ASK signal by the constructed stub (Stub-5).

(a) (b)

Figure D.2: Photo of constructed electrical stub LPFs, (a) Stub-5, (b) Stub-6

Another duobinary stub LPF, producing impulses spaced by 83 ps (2l

υ= 83ps) and

having a frequency response dip at 6 GHz (Stub-6), was also constructed as shown in figure

D.2 (b). The frequency response of the LPF stub (Stub-6) is shown in figure D.4. The

eye diagram shown in figure D.4 is for a 10 Gbit/s electrical duobinary signal generated

by filtering a 10 Gbit/s electrical binary NRZ-ASK signal by this stub (Stub-6). Several

versions of the Duobinary stub filters with slightly different lengths were built and tested in

order to optimize the performance and response of the filter.

Duobinary Filters 87

0 2 4 6 8 10 12

-40

-35

-30

-25

-20

-15

-10

-5

Frequency [GHz]

S21 [dB]

Figure D.3: Frequency response of the electrical stub LPF (Stub-5) and eye diagram of a

generated 10 Gbit/s electrical duobinary signal

0 2 4 6 8 10 12

-45

-40

-35

-30

-25

-20

-15

-10

-5

Frequency [GHz]

S21 [dB]

Figure D.4: Frequency response of the electrical stub LPF (Stub-6), and eye diagram of a

generated 10 Gbit/s electrical duobinary signal

88 Duobinary Filters

Figure D.5 and D.6 show the frequency response of a 5th order Bessel LPF with a 3

dB corner frequency at 2.8 GHz [82] plotted together with the frequency response of the

Stub-5 and Stub-6 filters respectively. From figure D.6 it can be noticed that the 3 dB

corner frequency of the Stub-6 was approximately the same as the Bessel filter, therefore the

duobinary signal generated using the Stub-6 filter will perform more similar to a duobinary

signal generated using the Bessel filter.

0246810 12

-45

-40

-35

-30

-25

-20

-15

-10

-5

Frequency [GHz]

S21 [dB]

Stub-5

Bessel

Figure D.5: Frequency response of a Stub-5 filter and 5th order Bessel LPF ([email protected])

0246810 12

-45

-40

-35

-30

-25

-20

-15

-10

-5

Frequency [GHz]

S21 [dB]

Stub-6

Bessel

Figure D.6: Frequency response of a Stub-6 filter and 5th order Bessel LPF ([email protected])

Bibliography

[1] K.-P. Ho, Phase-Modulated Optical Communication Systems, Springer, 2005, ISBN

0-387-24392-5.

[2] J. M. Kahn and K.-P. Ho, “Spectral efficiency limits in DWDM systems”, in 31st

European Conference on Optical Communication (ECOC 2005), Glasgow, vol. 4,

pp. 843 – 846, September 2005.

[3] K. Petermann and S. Randel, “Strategies for spectrally efficient optical fiber commu-

nication systems with direct detection”, in International Conference on Transparent

Optical Networks (ICTON 2003), Warsaw, Poland, vol. 2, pp. 58 – 63, 2003.

[4] H. Louchet, K. Petermann, A. Robinson, and R. Epworth, “On the spectral informa-

tion distribution in optical fibers”, in IEEE Lasers and Electro-Optics Society Annual

Meeting (LEOS 2004), Rio Grande, Puerto Rico, vol. 1, pp. 17 – 18, 2004.

[5] N. Kikuchi, “Amplitude and phase modulated 8-ary and 16-ary multilevel signal-

ing technologies for high-speed optical fiber communication”, in SPIE Proceedings,

vol. 6021, pp. 17 – 18, 2005.

[6] S. K. Ibrahim, S. Bhandare, and R. Noé, “Performance of 20 Gbit/s Quaternary In-

tensity Modulation Based on Binary or Duobinary Modulation in Two Quadratures

With Unequal Amplitudes”, IEEE Journal of Selected Topics in Quantum Electron-

ics, vol. 12, no. 4, pp. 596 – 602, 2006.

[7] S. K. Ibrahim, S. Bhandare, and R. Noé, “20 Gbit/s Quaternary Intensity Modulation

Based on Duobinary Modulation with Unequal Amplitude in Two Polarizations”,

IEEE Photonics Technology Letters, vol. 18, no. 14, pp. 1482 – 1484, 2006.

[8] B. Wedding, W. Idler, B. Franz, W. Pöhlmann, and E. Lach, “40 Gbit/s quaternary dis-

persion supported transmission over 31 km standard singlemode fibre without optical

dispersion compensation”, in 24th European Conference on Optical Communication

(ECOC 1998), Madrid, Spain, vol. 1, pp. 523 – 524, September 1998.

[9] S. Walklin and J. Conradi, “A 10 Gb/s 4-ary ASK Lightwave System”, in 23rd Euro-

pean Conference on Optical Communication (ECOC 1997), Edinburgh, UK, pp. 255

– 258, September 1997.

[10] S. Walklin and J. Conradi, “Multilevel signaling for increasing the reach of 10 Gb/s

lightwave systems”, IEEE Journal of Lightwave Technology, vol. 17, no. 11, pp. 2235

– 2248, November 1999.

90 BIBLIOGRAPHY

[11] H. Suzuki, J. Kani, T. Nakamura, M. Teshima, and K. Iwatsuki, “High-Spectral-

Efficiency DWDM Access Network Supporting 10-Gbps Access Services”, in Opti-

cal Fiber Communications Conference (OFC 2003), Atlanta, Georgia, USA, vol. 2,

pp. 551 – 552, March 2003.

[12] A. Wonfor, J. K. White, E. E. Coulson, R. V. Penty, and I. H. White, “Uncooled oper-

ation of a 40Gb/s directly modulated multi-level laser for datacoms applications”, in

29th European Conference on Optical Communication (ECOC 2003), Rimini, Italy,

Tu4.5.6, September 2003.

[13] R. A. Griffin and A. C. Carter, “Optical differential quadrature phase-shift key

(oDQPSK) for high capacity optical transmission”, in Optical Fiber Communications

Conference (OFC 2002), Anaheim, California, USA, WX6, pp. 367 – 368, March

2002.

[14] C. Wree, J. Leibrich, J. Eick, and W. Rosenkranz, “Experimental investigation of re-

ceiver sensitivity of RZ-DQPSK modulation format using balanced detection”, in Op-

tical Fiber Communications Conference (OFC 2003), Atlanta, Georgia, USA, vol. 2,

pp. 456 – 457, March 2003.

[15] A. H. Gnauck and P. J. Winzer, “Optical Phase-Shift-Keyed Transmission”, IEEE

Journal of Lightwave Technology, vol. 23, no. 1, pp. 115 – 130, January 2005.

[16] Y. Han and G. Li, “Differential Polarization-Phase-Shift Keying without Using Po-

larization Control”, in , Anaheim, CA, USA, JWA39, March 2005.

[17] Y. Han and G. Li, “Direct detection differential polarization-phase-shift keying based

on Jones vector”, Optics Express, vol. 12, no. 24, pp. 5821 – 5826, November 2004.

[18] X. Liu, X. Wei, Y.-H. Kao, J. Leuthold, C. R. Doerr, and L. F. Mollenauer, “Qua-

ternary differential-phase amplitude-shift-keying for DWDM transmission”, in 29th

European Conference on Optical Communication (ECOC 2003), Rimini, Italy, vol. 4,

pp. 1010 – 1011, September 2003.

[19] M. Ohm and J. Speidel, “Quaternary Optical ASK-DPSK and Receivers With Direct

Detection”, IEEE Photonics Technology Letters, vol. 15, no. 1, pp. 159 – 161, 2003.

[20] T. Nakamura, J.-I. Kani, M. Teshima, and K. Iwatsuki, “A quaternary amplitude

shift keying modulator for suppressing initial amplitude distortion”, IEEE Journal

of Lightwave Technology, vol. 22, no. 3, pp. 733 – 738, March 2004.

[21] S. K. Ibrahim, S. Bhandare, H. Zhang, and R. Noé, “2×10 Gbit/s Quaternary Intensity

Modulation Generation using an Optical QPSK modulator”, in SPIE Proceedings,

vol. 6021, December 2005.

[22] M. I. Hayee, M. C. Cardakli, A. B. Sahin, and A. E. Willner, “Doubling of

bandwidth utilization using two orthogonal polarizations and power unbalancing in

a polarization-division-multiplexing scheme”, IEEE Photonics Technology Letters,

vol. 13, no. 8, pp. 881 – 883, August 2001.

BIBLIOGRAPHY 91

[23] M. I. Hayee, M. C. Cardakli, and A. E. Willner, “Power unbalanced polarization

division-multiplexing for efficient bandwidth utilization”, in Conference on Lasers

and Electro-Optics (CLEO 1999), Baltimore, MD, USA, CTuQl, pp. 181 – 182, May

1999.

[24] T. Ono, Y. Yano, and K. Fukuchi, “Demonstration of high-dispersion tolerance of 20-

Gbit/s optical duobinary signal generated by a low-pass filtering method”, in Optical

Fiber Communications Conference (OFC 1997), Dallas, USA, ThH1, pp. 268 – 269,

February 1997.

[25] K. Yonenaga, S. Kuwano, S. Norimatsu, and N. Shibata, “Optical duobinary transmis-

sion system with no receiver sensitivity degradation”, IEE Electronic Letters, vol. 31,

no. 4, pp. 302 – 304, Feb. 1995.

[26] K. Yonenaga and S. Kuwano, “Dispersion-Tolerant Optical Transmission System Us-

ing Duobinary Transmitter and Binary Receiver”, IEEE Journal of Lightwave Tech-

nology, vol. 15, no. 8, pp. 1530 – 1537, August 1997.

[27] S. K. Ibrahim, S. Bhandare, and R. Noé, “Narrowband 20 Gbit/s Quaternary Intensity

Modulation Generated by Duobinary 10 Gbit/s Modulation in 2 Quadratures”, in 31st

European Conference on Optical Communication (ECOC 2005), Glasgow, Scotland,

vol. 4, pp. 909 – 910, September 2005.

[28] S. K. Ibrahim, S. Bhandare, H. Zhang, and R. Noé, “2×10 Gbit/s Quaternary Intensity

Modulation Generation using an Optical QPSK modulator”, in Asia-Pacific Optical

Communications Conference (APOC 2005), Shanghai, China, 6021-44, November

2005.

[29] S. K. Ibrahim, S. Bhandare, and R. Noé, “Narrowband 2×10 Gbit/s Quaternary Inten-

sity Modulation Based on Duobinary Modulation in Two Polarizations with Unequal

Amplitudes”, in Optical Fiber Communication Conference (OFC 2006), Anaheim,

CA, USA, OThI2, March 2006.

[30] G. Jacobsen, Noise in Digital Optical Transmission Systems, Artech House, 1994,

ISBN 0-89006-695-7.

[31] R. Noé, “Optical amplifier performance in digital optical communication systems”,

Electrical Engineering, vol. 83, pp. 15 – 20, February 2001.

[32] G. P. Agrawal, Fiber-Optic Communication Systems, Wiley, 1992, ISBN 0-471-

54286-5.

[33] A. E. Willner, “Chromatic Dispersion and Polarization-Mode Dispersion: Managing

Key Limitations in Optical Communication Systems”, Optics and Photonics News,

OPN Trends, OSA, vol. 13, no. 2, pp. 116 – 121, March 2002.

[34] A. F. Elrefaie, R. E. Wagner, D. A. Atlas, and D. G. Daut, “Chromatic Dispersion

Limitations in Coherent Lightwave Transmission Systems”, IEEE Journal of Light-

wave Technology, vol. 6, no. 5, pp. 704 – 709, May 1988.

92 BIBLIOGRAPHY

[35] A. Hodzic, Investigations of high bit rate optical transmission systems employing a

channel data rate of 40 Gb/s, Ph.D. thesis, Technische Universität Berlin, Berlin,

Germany, 2004.

[36] R. Noé, D. Sandel, S. Bhandare, F. Wüst, B. Milivojevic, and V. Mirvoda, “Chro-

matic dispersion monitoring by synchronous signal arrival time detection in the clock

recovery PLL”, Journal of Optical Networking, OSA, vol. 3, no. 8, pp. 589 – 600,

2004.

[37] P. J. Corvini and T. L. Koch, “Computer Simulation of High-Bit-Rate Optical Fiber

Transmission Using Single-Frequency Lasers”, IEEE Journal of Lightwave Technol-

ogy, vol. 5, no. 11, pp. 1591 – 1595, November 1987.

[38] S. Ryu, Coherent Lightwave Communication Systems, Artech House, 1995, ISBN

0-89006-612-4.

[39] M. Ohm and J. Speidel, “Optimal Receiver Bandwidths, Bit Error Probabilities and

Chromatic Dispersion Tolerance of 40Gbit/s Optical 8-DPSK with NRZ and RZ Im-

pulse Shaping”, in Optical Fiber Communications Conference (OFC 2005), Ana-

heim, CA, USA, OFG5, 2005.

[40] M. W. Chbat and D. Penninckx, “High-spectral-efficiency transmission systems”,

in Optical Fiber Communications Conference (OFC 2005), Baltimore, USA, TuJ1,

pp. 134 – 136, March 2000.

[41] J. M. Kahn and K.-P. Ho, “Spectral efficiency limits and modulation/detection tech-

niques for DWDM systems”, IEEE Journal of Selected Topics in Quantum Electron-

ics, vol. 10, no. 2, pp. 259 – 272, March/April 2004.

[42] S. Haykin, Communication Systems, Wiley, Fourth edition, 2001, ISBN 0-471-17869-

[43] J. D. Ralston, J. M. Kahn, and K.-P. Ho, “Advanced modulation and signal pro-

cessing techniques for 40 Gb/s optical transmission systems”, in SPIE Proceedings,

vol. 4872, pp. 24 – 31, 2002.

[44] B. Milivojevic, Study of Optical Differential Phase Shift Keying Transmission Tech-

niques at 40 Gbit/s and beyond, Ph.D. thesis, University of Paderborn, EIM-E, Pader-

born, Germany, 2005.

[45] T. Tokle, Optimised Dispersion Management and Modulation Formats for High

Speed Optical Communication Systems, Ph.D. thesis, Research Center COM, Tech-

nical University of Denmark, Denmark, 2004.

[46] P. J. Winzer and R.-J. Essiambre, “Advanced optical modulation formats”, in 29th

European Conference on Optical Communication (ECOC 2003), Rimini, Italy,

vol. 4872, pp. 1002 – 1003, 2003.

[47] S. Bigo, G. Charlet, and E. Corbel, “What has hybrid phase/intensity encoding

brought to 40 Gbit/s ultralong-haul systems?”, in 30th European Conference on Op-

tical Communication (ECOC 2004), Stockholm, Sweden, pp. 872 – 875, 2004.

BIBLIOGRAPHY 93

[48] T. Kataoka, Y. Miyamota, K. Hagimoto, and K. Noguchi, “20 Gbit/s long distance

transmission using a 270 photon/bit optical preamplifier receiver”, IEE Electronic

Letters, vol. 30, no. 9, pp. 715 – 716, April 1994.

[49] M. Daikoku, N. Yoshikane, and I. Morita, “Performance comparison of modulation

formats for 40 Gbit/s DWDM transmission systems”, in Optical Fiber Communica-

tions Conference (OFC 2005), Anaheim, CA, USA, OFN2, March 2005.

[50] A. Hodzic, B. Konrad, and K. Petermann, “Alternative Modulation Formats in N×40

Gb/s WDM Standard Fiber RZ-Transmission Systems”, IEEE Journal of Lightwave

Technology, vol. 20, no. 4, pp. 598 – 607, April 2002.

[51] W. Idler, A. Klekamp, R. Dischler, and B. Wedding, “Advantages of Frequency Shift

Keying in 10 Gb/s Systems”, in IEEE Lasers and Electro-Optics Society Annual

Meeting (LEOS 2004), Rio Grande, Puerto Rico, FD3, 2004.

[52] T. Sakamoto, T. Kawanishi, T. Miyazaki, and M. Izutsu, “Novel modulation scheme

for optical continuous-phase frequency-shift keying”, in Optical Fiber Communica-

tions Conference (OFC 2005), Anaheim, CA, USA, OFG2, March 2005.

[53] B. Wedding, R. Jung, C. Haslach, and H. Söhnle, “10.7 Gbit/s FSK transmission

with 61 dB power budget”, in 29th European Conference on Optical Communication

(ECOC 2003), Rimini, Italy, Th1.5.5, September 2003.

[54] J. Mo, Y. Dong, Y. Wen, S. Takahashi, Y. Wang, and C. Lu, “Optical Minimum-Shift

Keying Modulator for High Spectral Efficiency WDM Systems”, in 31st European

Conference on Optical Communication (ECOC 2005), Glasgow, Scotland, Th1.2.3,

September 2005.

[55] A. Gnauck, “40-Gb/s RZ-Differential Phase Shift Keyed Transmission”, in Opti-

cal Fiber Communications Conference (OFC 2003), Atlanta, Georgia, USA, ThE1,

March 2003.

[56] A. H. Gnauck, S. Chandrasekhar, J. Leuthold, and L. Stulz, “Demonstration of 42.7-

Gb/s DPSK Receiver With 45 Photons/Bit Sensitivity”, IEEE Photonics Technology

Letters, vol. 15, no. 1, pp. 99 – 101, January 2003.

[57] A. Hirano, Y. Miyamoto, and S. Kuwahara, “Performances of CSRZ-DPSK and RZ-

DPSK in 43-Gbit/s/ch DWDM G.6S2 Single-Mode-Fiber Transmission”, in Opti-

cal Fiber Communications Conference (OFC 2003), Atlanta, Georgia, USA, vol. 2,

pp. 454 – 456, March 2003.

[58] G. Bosco, A. Carena, V. Curri, R. Gaudino, and P. Poggiolini, “Modulation Formats

Suitable for Ultrahigh Spectral Efficient WDM Systems”, IEEE Journal of Lightwave

Technology, vol. 10, no. 2, pp. 454 – 456, March/April 2004.

[59] H. Kim, “Differential Phase Shift Keying for 10-Gb/s and 40-Gb/s Systems”, in IEEE

Lasers and Electro-Optics Society Annual Meeting (LEOS 2004), Rio Grande, Puerto

Rico, ThC1, pp. 13 – 14, 2004.

94 BIBLIOGRAPHY

[60] A. Carena, V. Curri, R. Gaudino, N. Greco, P. Poggiolini, and S. Benedetto, “Polar-

ization modulation in ultra-long haul transmission systems: a promising alternative

to intensity modulation”, in 24th European Conference on Optical Communication

(ECOC 1998), Madrid, Spain, WdA24, pp. 429 – 430, September 1998.

[61] A. S. Siddiqui, S. G. Edirisinghe, J. J. Lepley, J. G. Ellison, and S. D. Walker,

“Dispersion-tolerant transmission using a duobinary polarization-shift keying trans-

mission scheme”, IEEE Photonics Technology Letters, vol. 14, no. 2, pp. 158 – 160,

February 2002.

[62] S. G. Edirisinghe, J. J. Lepley, and A. S. Siddiqui, “Polarization shaped duobinary

transmission scheme”, IEEE Photonics Technology Letters, vol. 13, no. 11, pp. 1245

– 1247, November 2001.

[63] A. H. Gnauck, X. Liu, X. Wei, D. M. Gill, and E. C. Burrows, “Comparison of

Modulation Formats for 42.7-Gb/s Single-Channel Transmission Through 1980 km

of SSMF”, IEEE Photonics Technology Letters, vol. 16, no. 3, pp. 1245 – 1247,

March 2004.

[64] J. G. Proakis, Digital Communications, McGraw-Hill, Second edition, 1989, ISBN

0-07-050937-9.

[65] J. G. Proakis and M. Salehi, Communication Systems Engineering, Prentice Hall,

1994, ISBN 0-13-158932-6.

[66] G. L. Li and P. K. L. Yu, “Optical Intensity Modulators for Digital and Analog Ap-

plications”, IEEE Journal of Lightwave Technology, vol. 21, no. 9, pp. 2010 – 2030,

September 2003.

[67] W. A. Atia and R. S. Bondurant, “Demonstration of Return-to-zero Signaling in Both

OOK and DPSK Formats to Improve Receiver Sensitivity in an Optically Preampli-

fied Receiver”, in IEEE Lasers and Electro-Optics Society Annual Meeting (LEOS

1999), San Francisco, CA, USA, TuM3, pp. 226 – 227, 1999.

[68] I. Lyubomirsky and C.-C. Chien, “DPSK Demodulator Based on Optical Discrimina-

tor Filter”, IEEE Photonics Technology Letters, vol. 17, no. 2, pp. 492 – 494, February

2005.

[69] R. A. Griffin, R. L. Johnstone, R. G. Walker, J. Hall, S. D. Wadsworth, K. Berry,

A. C. Carter, and M. J. Wale, “10 Gb/s optical differential quadrature phase shift key

(DQPSK) transmission using GaAs/AlGaAs integration”, in Optical Fiber Commu-

nications Conference (OFC 2002), Anaheim, California, USA, FD6, March 2002.

[70] C. Wree, J. Leibrich, and W. Rosenkranz, “RZ-DQPSK format with high spectral effi-

ciency and high robustness toward fiber nonlinearities”, in 28th European Conference

on Optical Communication (ECOC 2002), Copenhagen, Denmark, 9.6.6, September

2002.

[71] E. A. Swanson, J. C. Livas, and R. S. Bondurant, “High Sensitivity Optically Pream-

plified Direct Detection DPSK Receiver with Active Delay-Line Stabilization”, IEEE

Photonics Technology Letters, vol. 6, no. 2, pp. 263 – 265, February 1994.

BIBLIOGRAPHY 95

[72] G. Bosco and P. Poggiolini, “The Impact of Receiver Imperfections on the Perfor-

mance of Optical Direct-Detection DPSK”, IEEE Journal of Lightwave Technology,

vol. 23, no. 2, pp. 842 – 848, February 2005.

[73] D. Penninckx, H. Bissessur, P. Brindle, E. Gohin, and F. Bakhti, “Optical differential

phase shift keying (DPSK) direct detection considered as a duobinary signal”, in

27th European Conference on Optical Communication (ECOC 2001), Amsterdam,

The Netherlands, We.P.40, pp. 456 – 457, September 2001.

[74] X. Wei, X. Liu, S. Chandrasekhar, A. H. Gnauck, G. Raybon, J. Leuthold, and P. J.

Winzer, “40 Gb/s duobinary and modified duobinary transmitter based on an opti-

cal delay interferometer”, in 28th European Conference on Optical Communication

(ECOC 2002), Copenhagen, Denmark, 9.6.3, September 2002.

[75] T. Ono, Y. Yano, K. Fukuchi, T. Ito, H. Yamazaki, M. Yamaguchi, and K. Emura,

“Characteristics of Optical Duobinary Signals in Terabit/s Capacity, High-Spectral

Efficiency WDM Systems”, IEEE Journal of Lightwave Technology, vol. 16, no. 5,

pp. 788 – 797, May 1998.

[76] W. Kaiser, T. Wuth, M. Wichers, and W. Rosenkranz, “Reduced Complexity Optical

Duobinary 10-Gb/s Transmitter Setup Resulting in an Increased Transmission Dis-

tance”, IEEE Photonics Technology Letters, vol. 13, no. 8, pp. 884 – 886, August

2001.

[77] G. Charlet, J.-C. Antona, S. Lanne, P. Tran, W. Idler, M. Gorlier, S. Borne,

A. Klekamp, C. Simonneau, L. Pierre, Y. Frignac, M. Molina, F. Beaumont, J.-

P. Hamaide, and S. Bigo, “6.4Tb/s (159×42.7Gb/s) Capacity over 21×100 km us-

ing Bandwidth-limited Phase-Shaped Binary Transmission”, in 28th European Con-

ference on Optical Communication (ECOC 2002), Copenhagen, Denmark, PD4.1,

September 2002.

[78] W. Rosenkranz, “High Capacity Optical Communication Networks - Approaches for

Efficient Utilization of Fiber Bandwidth”, in First Joint Symposium on Opto- & Mi-

croelectronic Devices and Circuits (SODC 2000), Nanjing, China, pp. 106 – 107,

April 2000.

[79] A. J. Price and N. L. Mercier, “Reduced bandwidth optical intensity modulation with

improved chromatic dispersion tolerance”, IEE Electronic Letters, vol. 31, no. 1,

pp. 58 – 59, January 1995.

[80] A. J. Price, L. Pierre, R. Uhel, and V. Havard, “210 km Repeaterless 10 Gb/s Trans-

mission Experiment Through Nondispersion-Shifted Fiber Using Partial Response

Scheme”, IEEE Photonics Technology Letters, vol. 7, no. 10, pp. 1219 – 1221, Octo-

ber 1995.

[81] D. Penninckx, “Dispersion-tolerant modulation techniques for optical communica-

tions”, in 24th European Conference on Optical Communication (ECOC 1998),

Madrid, Spain, vol. 1, pp. 509 – 510, September 1998.

96 BIBLIOGRAPHY

[82] D. Penninckx, “Effect of electrical filtering of duobinary signals on the chromatic

dispersion transmission limitations”, in 24th European Conference on Optical Com-

munication (ECOC 1998), Madrid, Spain, vol. 1, pp. 537 – 538, September 1998.

[83] H. Kim and C. X. Yu, “Optical Duobinary Transmission System Featuring Improved

Receiver Sensitivity and Reduced Optical Bandwidth”, IEEE Photonics Technology

Letters, vol. 14, no. 8, pp. 1205 – 1207, August 2002.

[84] H. Kim, C. X. Yu, and D. T. Neilson, “Demonstration of Optical Duobinary Transmis-

sion System Using Phase Modulator and Optical Filter”, IEEE Photonics Technology

Letters, vol. 14, no. 7, pp. 1010 – 1012, July 2002.

[85] H. Kim and P. J. Winzer, “Robustness to Laser Frequency Offset in Direct-Detection

DPSK and DQPSK Systems”, IEEE Journal of Lightwave Technology, vol. 21, no. 9,

pp. 1887 – 1891, September 2003.

[86] P. S. Cho, V. S. Grigoryan, Y. A. Godin, A. Salamon, and Y. Achiam, “Transmission

of 25-Gb/s RZ-DQPSK Signals With 25-GHz Channel Spacing Over 1000 km of

SMF-28 Fiber”, IEEE Photonics Technology Letters, vol. 15, no. 3, pp. 473 – 475,

March 2003.

[87] C. Wree, N. Hecker-Denschlag, E. Gottwald, P. Krummrich, J. Leibrich, E.-D.

Schmidt, B. Lankl, and W. Rosenkranz, “High Spectral Efficiency 1.6-b/s/Hz Trans-

mission (8×40 Gb/s With a 25-GHz Grid) Over 200-km SSMF Using RZ-DQPSK

and Polarization Multiplexing”, IEEE Photonics Technology Letters, vol. 15, no. 9,

pp. 1303 – 1305, September 2003.

[88] N. Yoshikane and I. Morita, “1.14 b/s/Hz spectrally-efficient 50×85.4 Gb/s trans-

mission over 300 km using copolarized CS-RZ DQPSK signals”, in Optical Fiber

Communications Conference (OFC 2004), Los Angeles, CA, USA, PDP38, Febru-

ary 2004.

[89] S. Bhandare, D. Sandel, B. Milivojevic, A. Hidayat, A. A. Fauzi, H. Zhang, S. K.

Ibrahim, F. Wüst, and R. Noé, “5.94-Tb/s 1.49-b/s/Hz (40×2×2×40 Gb/s) RZ-

DQPSK Polarization-Division Multiplex C-Band Transmission Over 324 km”, IEEE

Photonics Technology Letters, vol. 17, no. 4, pp. 914 – 916, April 2005.

[90] G. Kramer, A. Ashikhmin, A. J. van Wijngaarden, and X. Wei, “Spectral Efficiency of

Coded Phase-Shift Keying for Fiber-Optic Communication”, IEEE Journal of Light-

wave Technology, vol. 21, no. 10, pp. 2438 – 2445, October 2003.

[91] A. H. Gnauck, P. J. Winzer, C. Dorrer, and S. Chandrasekhar, “Linear and Nonlinear

Performance of 42.7-Gb/s Single-Polarization RZ-DQPSK Format”, IEEE Photonics

Technology Letters, vol. 18, no. 7, pp. 883 – 885, April 2006.

[92] R. A. Griffin, “Integrated DQPSK Transmitters”, in Optical Fiber Communications

Conference (OFC 2005), Anaheim, CA, USA, OWE3, March 2005.

[93] K.-P. Ho and H.-W. Cuei, “Generation of Arbitrary Quadrature Signals Using One

Dual-Drive Modulator”, IEEE Journal of Lightwave Technology, vol. 23, no. 2,

pp. 764 – 770, February 2005.

BIBLIOGRAPHY 97

[94] J. M. Gené, M. Soler, R. I. Killey, and J. Prat, “Investigation of 10-Gb/s Optical

DQPSK Systems in Presence of Chromatic Dispersion, Fiber Nonlinearities, and

Phase Noise”, IEEE Photonics Technology Letters, vol. 16, no. 3, pp. 924 – 926,

March 2004.

[95] D. van den Borne, S. L. Jansen, E. Gottwald, P. M. Krummrich, P. Leisching, G. D.

Khoe, and H. de Waardt, “A robust modulation format for 42.8-Gbit/s long-haul

transmission: RZ-DPSK or RZ-DQPSK?”, in 7th ITG Fachtagung on Photonic Net-

works, Leipzig, Germany, pp. 51 – 56, April 2006.

[96] S. Bhandare, D. Sandel, B. Milivojevic, A. Hidayat, A. Fauzi, H. Zhang, S. K.

Ibrahim, F. Wüst, and R. Noé, “5.94 Tbit/s (40×2×2×40 Gbit/s) C-Band Trans-

mission over 324 km using RZ-DQPSK Combined with Polarization Division Mul-

tiplex”, in 6th ITG Fachtagung on Photonic Networks "Photonische Netze", Leipzig,

Germany, pp. 87 – 90, May 2005.

[97] D. van den Borne, S. L. Jansen, E. Gottwald, P. M. Krummrich, G. D. Khoe, and

H. de Waardt, “1.6-b/s/Hz Spectrally Efficient 40×85.6-Gb/s Transmission Over

1,700 km of SSMF Using POLMUX-RZ-DQPSK”, in Optical Fiber Communica-

tions Conference (OFC 2006), Anaheim, CA, USA, PDP34, March 2006.

[98] Y. Zhu, K. Cordina, N. Jolley, R. Feced, H. Kee, R. Rickard, and A. Hadjifotiou, “1.6

bit/s/Hz orthogonally polarized CSRZ-DQPSK transmission of 8×40 Gbit/s over 320

km NDSF”, in Optical Fiber Communications Conference (OFC 2004), Los Angeles,

CA, USA, TuF1, February 2004.

[99] B. Milivojevic, A. F. Abas, A. Hidayat, S. Bhandare, D. Sandel, R. Noé, M. Guy, ,

and M. Lapointe, “1.6-b/s/Hz 160-Gb/s 230-km RZ-DQPSK Polarization Multiplex

Transmission With Tunable Dispersion Compensation”, IEEE Photonics Technology

Letters, vol. 17, no. 2, pp. 495 – 497, February 2005.

[100] P. S. Cho, G. Harston, C. J. Kerr, A. S. Greenblatt, A. Kaplan, Y. Achiam, G. Levy-

Yurista, M. Margalit, Y. Gross, and J. B. Khurgin, “Investigation of 2-b/s/Hz 40-Gb/s

DWDM Transmission Over 4×100 km SMF-28 Fiber Using RZ-DQPSK and Polar-

ization Multiplexing”, IEEE Photonics Technology Letters, vol. 16, no. 2, pp. 656 –

658, February 2004.

[101] S. S. Walklin, Multilevel Signaling for Increasing the Capacity of High-Speed Optical

Communication Systems, Ph.D. thesis, University of Alberta, Canada, 1997.

[102] J.-P. Elbers, H. Wernz, H. Griesser, C. Glingener, A. Faerbert, S. Langenbach, N. Sto-

janovic, C. Dorschky, T. Kupfer, and C. Schulien, “Measurement of the dispersion

tolerance of optical duobinary with an MLSE receiver at 10.7 Gb/s”, in Optical Fiber

Communications Conference (OFC 2005), Anaheim, CA, USA, OThJ4, March 2005.

[103] S. Hayase, N. Kikuchi, K. Sekine, and S. Sasaki, “Proposal of 8-state per symbol

(binary ASK and QPSK) 30-Gbit/s optical modulation/demodulation scheme”, in

29th European Conference on Optical Communication (ECOC 2003), Rimini, Italy,

Th2.6.4, September 2003.

98 BIBLIOGRAPHY

[104] I. Lyubomirsky and C.-C. Chien, “Tailoring the Duobinary Pulse Shape for Optimum

Performance”, IEEE Journal of Lightwave Technology, vol. 23, no. 11, pp. 3732 –

3736, November 2005.

[105] I. Lyubomirsky and C.-C. Chien, “Ideal Duobinary Generating Filter for Optically

Amplified Systems”, IEEE Photonics Technology Letters, vol. 18, no. 4, pp. 598 –

600, February 2006.

[106] M. Wichers and W. Rosenkranz, “Optical duobinary modulation schemes using a

Mach-Zehnder transmitter for lightwave systems”, in International Conference on

Transparent Optical Networks (ICTON 1999), Kielce, Poland, We.B.1, pp. 15 – 18,

June 1999.

[107] E. H. Fooks and R. A. Zakarevicius, Microwave Engineering Using Microstrip Cir-

cuits, Prentice Hall, 1990, ISBN 0-13-691650-3.