Dut Masters Degree in Engineering Dessertation - Final 04 Oct 2020

Comparative Analysis, Design and Optimisation of a
48 Channel DWDM System using Various Design

Parameters
Submitted in the fulfilment for the requirements of the degree of

Master of Engineering in Electronic Engineering
in the Faculty of Engineering and the Built Environment at the
Durban University of Technology
Jerrin Zachariah Mohan
20919951
2019
PLAGIARISM DECLARATION
1. I know and understand that plagiarism is using another person’s work and pretending
it is one’s own, which is wrong.
2. This document is my own work.
3. I have appropriately referenced the work of other people I have used.
4. I have not allowed, and will not allow anyone to copy my work with the intention of
passing it off as his/her own work.
5. This document has not been submitted, in whole or part, for a degree to any other
University or Institution.
______________________ _____________________
Jerrin Mohan Date
Student No.: 20919951
Approved for Final Submission by:
______________________ ______________________
Supervisor: Dr. O. A. Sokoya Date
i
ABSTRACT
In the current era, there is an ever-growing demand for data hungry applications and services
that need large amounts of bandwidth to send digital information at very high speeds. In order
to meet this challenge for higher bandwidth capacity, Dense Wave Division Multiplexing
(DWDM) is used as the strategy to transmit multiple high-bit rate channels at extremely
narrow channel spacings over a single fiber core. However, this gives rise to detrimental
transmission impairments such as linear effects and non-linear effects. The dissertation
minimises the impairments by optimally designing a new DWDM system that produces a
detectable and acceptable quality of signal at the receiver.
In this dissertation, a comparative analysis is performed on the simulative design of a 48-

channel DWDM system that has a 25 Gb/s bit rate and a 100 km transmission distance. The
research mitigates the effects of transmission impairments such that an error-free matched
communication link is produced for equally spaced (ES) channels of 100 GHz, 50 GHz, 25 GHZ
and 12.5 GHz and 6.25 GHz. Various design parameters are used to create the comparative
analysis model to optimise the 48 channel DWDM network. The design is simulated using the
Optisystem simulation platform and the signal analysis is based on the bit error rate (BER) and
quality (Q) factor of the received signal’s eye diagrams.
It is established in the desertion that modified networks with matched active components has
ES frequency channels that are aligned to each other and has a higher optical signal to noise
ratio (OSNR) than mismatched networks. The maximum signal power and OSNR of the 3-
erbium doped fiber amplifier (EDFA)-post symmetric compensation technique is always higher
than the 1-EDFA post compensation technique for all channel spacings in any type of network.
Modified duobinary return to zero (MDRZ) when compared to non-return to zero (NRZ) and
return to zero (RZ) has a greater dispersion tolerance, higher fiber non-linearity tolerance and
a higher acceptable signal transmission over longer distances with the least amount of errors.
The optimised design parameter configurations produce the highest signal performance
(highest Q factor > 6 and lowest BER > 10-9) and the highest bandwidth efficiency for the RZ
Modulation (at 100 GHz, 50 GHz and 25 GHz channel spacings) and MDRZ Modulation (at 12.5
and 6.25 GHz channel spacing).
ii
ACKNOWLEDGEMENTS
My sincere thanks go to my supervisor, Dr Oludare Ayodeji Sokoya for his constant support,
continuous motivation and careful guidance with the topic of this dissertation. His mentorship
and many encouragements during the past three years have given me the strength to endure
any challenge.
I would like to thank my wife, Anitta, for being the rock in my life who has been my constant
support and source of encouragement throughout this dissertation.
My parents, Mr and Mrs. Mohan, my in-laws, Mr and Mrs Eapen and finally my siblings, Jithin,
Anoop and Navya are owed many thanks for their support and their keen interest in my
wellbeing.
Special thanks go to my uncle and aunt, Mr and Mrs K.T. Jain. They were the two teachers in
my life that made me love physics, electronics and in particular – fiber optics. They have always
supported me in whatever challenges I have taken up in my life and my MEng was no
exception.
Thanks, is also owed to Transnet Freight Rail for their much-appreciated support in allowing
the time and resources to do hours of research for this MEng within their premises.
Last but not the least I would like to give thanks unto the LORD; for He is good and His mercy
endures for ever.
iii
TABLE OF CONTENTS
PLAGIARISM DECLARATION ................................................................................................. i
ABSTRACT ........................................................................................................................... ii
ACKNOWLEDGEMENTS ...................................................................................................... iii
TABLE OF CONTENTS.......................................................................................................... iv
LIST OF FIGURES .............................................................................................................. viii
LIST OF TABLES ...............................................................................................................xviii
LIST OF ACRONYMS ......................................................................................................... xix
.......................................................................................................... xx
GENERAL INTRODUCTION TO THE RESEARCH IN DENSE WAVE DIVISION

MULTIPLEXING ................................................................................................................... 1
1.1 Introduction........................................................................................................... 1
1.2 Research Motivation .............................................................................................. 1
1.3 Research Objectives ............................................................................................... 2
1.4 Research Methodology .......................................................................................... 2
1.5 Publications ........................................................................................................... 3
1.6 Original Research Contributions ............................................................................. 4
1.7 Dissertation Overview ............................................................................................ 5
LITERATURE REVIEW......................................................................................... 7
2.1 Introduction........................................................................................................... 7
2.2 Evolution of Fundamental Concepts in Fiber Optical Communication ...................... 7
2.3 Advantages of DWDM .......................................................................................... 10
2.4 Performance limitation factors affecting DWDM systems...................................... 11
2.4.1 Linear effects ................................................................................................ 11
2.4.2 Non-linear effects.......................................................................................... 12
iv
2.5 Exploring the limitation and extent of existing research ........................................ 15
2.6 DWDM System Design Components ..................................................................... 22
2.6.1 WDM Transmitter ......................................................................................... 22
2.6.1.1 Pseudo-Random Bit Sequence Generator (PRBS) .................................... 23

2.6.1.2 Light Source........................................................................................... 24
2.6.1.3 Modulation scheme type of transmitting light source ............................. 24
2.6.1.4 Extinction Ratio ..................................................................................... 24
2.6.1.5 Line coding Modulation format used in WDM Transmitter: NRZ and RZ... 25
2.6.2 Continuous Wave (CW) Laser Array with Equally Spaced (ES) channels ........... 27
2.6.2.1 Advanced Line coding Modulation format: MDRZ ................................... 28

2.6.3 WDM Multiplexer (MUX) ES and WDM De-multiplexer (DEMUX) ES ............... 30
2.6.3.1 Bessel Filter ........................................................................................... 30

2.6.3.2 Rectangular Filter................................................................................... 31
2.6.4 Optical Link ................................................................................................... 32
2.6.4.1 Fiber optic transmission medium ........................................................... 32

2.6.4.2 Optical Amplification ............................................................................. 33
2.6.5 Optical Receiver ............................................................................................ 34
2.7 Theoretical Concepts for the Design of the Optical Transmission Link .................... 35
2.8 Measuring system performance ........................................................................... 37
2.8.1 Optical Signal to Noise Ratio (OSNR) .............................................................. 37
2.8.2 Quality (Q) factor .......................................................................................... 38
2.8.3 Bit Error Rate (BER) ....................................................................................... 38
2.8.4 Overall Signal Performance and Bandwidth Efficiency .................................... 39
2.9 Summary ............................................................................................................. 40
DESIGN OF A 48 X 25Gb/s DWDM SYSTEM...................................................... 41
3.1 Introduction......................................................................................................... 41
3.2 Design Methodology ............................................................................................ 41
3.3 Design Parameters ............................................................................................... 43
v
3.3.1 Set Design Parameters................................................................................... 43
3.3.1.1 Design of the Transmitter block.............................................................. 43

3.3.1.2 Design of the Optical Transmission Link .................................................. 44
3.3.1.3 Design of the Receiver Block .................................................................. 46
3.3.2 Varied design parameters .............................................................................. 46
3.3.2.1 100 GHz channel spacing.......................................................................... 46

3.3.2.2 50 GHz channel spacing............................................................................ 47
3.3.2.3 25 GHz channel spacing............................................................................ 47
3.3.2.4 12.5 GHz channel spacing ......................................................................... 47
3.3.2.5 6.25 GHz channel spacing ......................................................................... 48
3.4 Summary ............................................................................................................. 53
SIMULATION RESULTS FOR THE 48 CHANNEL DWDM NETWORKS ................... 54
4.1 Introduction......................................................................................................... 54
4.2 The 48 Channel x 25 Gbps at 100 GHz Spacing using NRZ ...................................... 55
4.3 The 48 Channel x 25 Gbps at 100 GHz Spacing using RZ ......................................... 59
4.4 The 48 Channel x 25 Gbps at 50 GHz Spacing using NRZ ........................................ 63
4.5 The 48 Channel x 25 Gbps at 50 GHz Spacing using RZ ........................................... 68
4.6 The 48 Channel x 25 Gbps at 25 GHz Spacing using NRZ ........................................ 73
4.7 The 48 Channel x 25 Gbps at 25 GHz Spacing using RZ ........................................... 82
4.8 The 48 Channel x 25 Gbps at 12.5 GHz Spacing using NRZ ..................................... 91
4.9 The 48 Channel x 25 Gbps at 12.5 GHz Spacing using RZ ........................................ 98
4.10 The 48 Channel x 25 Gbps at 12.5 GHz Spacing using MDRZ ................................ 105
4.11 The 48 Channel x 25 Gbps at 6.25 GHz Spacing using NRZ ................................... 106
4.12 The 48 Channel x 25 Gbps at 6.25 GHz Spacing using RZ ...................................... 111
4.13 The 48 Channel x 25 Gbps at 6.25 GHz Spacing using MDRZ ................................ 116
4.14 Summary ........................................................................................................... 117
DISCUSSION OF RESULTS FOR THE 48 CHANNEL DWDM NETWORKS ............. 118
5.1 Introduction....................................................................................................... 118

vi
5.2 Comparison of OSNR for Post Compensation Technique (with 1 EDFA) vs Post
Symmetric Compensation Technique (with 3 EDFAs) ........................................... 119
5.3 Signal performance comparison of NRZ, RZ and MDRZ at various channel spacings for
maximum optimisation of a 48 Channel DWDM system ...................................... 122
5.3.1 Comparison of Post Compensation Technique with 1 EDFA vs Post Symmetric

Compensation Technique with 3 EDFAs ....................................................... 128
5.3.2 Optimisation of DWDM System by varying different parameters for RZ and NRZ
Modulation formats .................................................................................... 136
5.3.3 Optimisation of 12.5 GHz and 6.25 GHz Channel spacing using MDRZ ........... 146
5.4 Summary ........................................................................................................... 148
CONCLUSION AND RECOMMENDATIONS...................................................... 151
6.1 Introduction....................................................................................................... 151
6.2 Chapter Conclusions........................................................................................... 151
6.3 Recommendations ............................................................................................. 153
6.4 Current Potential of Designed DWDM Network Model and Future Work ............. 153
REFERENCES ................................................................................................................... 155
vii
LIST OF FIGURES
Figure 2.1: Timeline and Overview of DWDM Communication System [3] ............................. 9
Figure 2.2: Illustration of FWM on a three channel WDM system [3]................................... 14
Figure 2.3: A 48 port WDM Transmitter and the Block Diagram of a single channel in a WDM
Transmitter ..................................................................................................... 23
Figure 2.4: Illustration of the differences between NRZ and RZ Signals [35] ......................... 27
Figure 2.5: 48 port CW Laser Array and Design components of MDRZ modulation subsystem
....................................................................................................................... 28
Figure 2.6: Internal setup of Optical Receiver ..................................................................... 35
Figure 3.1: Block Diagram Design of 48 Channel DWDM system with bit rate of 25 Gb/s and
various other design parameters ...................................................................... 49
Figure 3.2: 48 Channel DWDM Network Design using either NRZ or RZ Modulation ............ 50
Figure 3.3: Post Compensated Optical Link Design with 1 EDFA .......................................... 51
Figure 3.4: Post Symmetric Optical Link Design with 3 EDFAs .............................................. 51
Figure 3.5: 48 Channel DWDM Network Design using MDRZ Modulation ............................ 52
Figure 4.1: Eye Diagram of 100 GHz Channel Spacing Using NRZ, 1 EDFA, MUX BW = 80 GHz
and DEMUX BW = 80 GHz ................................................................................ 55
Figure 4.2: Q Factor vs BER Graph of 100 GHz Spacing Using NRZ, 1 EDFA, MUX BW = 80 GHz
and DEMUX BW = 80 GHz ................................................................................ 55
Figure 4.3: Eye Diagram of 100 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz.............................. 56
Figure 4.4: Q factor vs BER Graph of 100 GHz Spacing Using NRZ, Post Symmetric Dispersion
Figure 4.6: Q Factor vs BER of 100 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3
EDFAs, MUX BW = 80 GHz and DEMUX BW = 50 GHz ........................................ 57
viii
EDFAs, MUX BW = 50 GHz and DEMUX BW = 80 GHz ........................................ 58
Figure 4.9: Eye Diagram of 100 GHz Channel Spacing Using RZ, 1 EDFA, MUX BW = 80 GHz and
DEMUX BW = 80 GHz ....................................................................................... 59
Figure 4.10: Q Factor vs BER of 100 GHz Channel Spacing Using RZ, 1 EDFA, MUX BW = 80 GHz
and DEMUX BW = 80 GHz .............................................................................. 59
Figure 4.11: Eye Diagram of 100 GHz Channel Spacing Using RZ, Post Symmetric Dispersion
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz ........................... 60
Figure 4.12: Q factor vs BER of 100 GHz Spacing Using RZ, Post Symmetric Dispersion with 3
EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz...................................... 60
Figure 4.14: Q Factor vs BER of 100 GHz Channel Spacing Using RZ, Post Symmetric Dispersion
Figure 4.16: Q Factor vs BER of 100 GHz Spacing Using RZ, Post Symmetric Dispersion with 3
Figure 4.17: Eye Diagram of 50 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz
and DEMUX BW = 80 GHz .............................................................................. 63
Figure 4.18: Q factor vs BER of 50 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 63
ix
Figure 4.26: Q Factor vs BER of 50 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion
Figure 4.27: Eye Diagram of 50 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 68
Figure 4.28: Q factor vs BER of 50 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 68
Figure 4.29: Eye Diagram of 50 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with
3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz ................................... 69
Figure 4.34: Q factor vs BER Diagram of 50 GHz Spacing Using RZ, Post Symmetric Dispersion
Figure 4.37: Eye Diagram of 25 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz
and DEMUX BW = 80 GHz .............................................................................. 73
Figure 4.38: Q factor vs BER of 25 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 73
x
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 12.5 GHz......................... 76
EDFAs, MUX BW = 80 GHz and DEMUX BW = 12.5 GHz ................................... 76
with 3 EDFAs, MUX BW = 12.5 GHz and DEMUX BW = 80 GHz......................... 78
EDFAs, MUX BW = 12.5 GHz and DEMUX BW = 80 GHz ................................... 78
with 3 EDFAs, ER = 5dB, Rectangular Filter, MUX BW = 80 GHz and DEMUX BW =
25 GHz .......................................................................................................... 80
EDFAs, ER = 5dB, Rectangular Filter, MUX BW = 80 GHz and DEMUX BW = 25 GHz
..................................................................................................................... 80
with 3 EDFAs, ER = 5dB, Rectangular Filter, MUX BW = 25 GHz and DEMUX BW =
25 GHz .......................................................................................................... 81
xi
..................................................................................................................... 81
Figure 4.55: Eye Diagram of 25 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 82
Figure 4.56: Q Factor vs BER of 25 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 82
3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 12.5 GHz ................................ 85
EDFAs, MUX BW = 80 GHz and DEMUX BW = 12.5 GHz ................................... 85
3 EDFAs, MUX BW = 12.5 GHz and DEMUX BW = 80 GHz ................................ 87
Figure 4.68: Q factor vs BER of 25 GHz Channel Spacing Using RZ, Post Symmetric Dispersion
xii
3 EDFAs, ER = 5dB, Rectangular Filter, MUX BW = 80 GHz and DEMUX BW = 25
GHz ............................................................................................................... 89
..................................................................................................................... 89
3 EDFAs, ER = 5dB, Rectangular Filter, MUX BW = 25 GHz and DEMUX BW = 25
GHz ............................................................................................................... 90
..................................................................................................................... 90
Figure 4.73: Eye Diagram of 12.5 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz
and DEMUX BW = 80 GHz .............................................................................. 91
Figure 4.74: Q factor vs BER of 12.5 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 91
Figure 4.75: Eye Diagram of 12.5 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion
Figure 4.76: Q factor vs BER of 12.5 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3
Figure 4.78: Q Factor vs BER of 12.5 GHz Spacing Using NRZ, Post Symmetric Dispersion with
3 EDFAs, MUX BW = 12.5 GHz and DEMUX BW = 80 GHz ................................ 95
xiii
with 3 EDFAs, MUX BW = 3.125 GHz and DEMUX BW = 80 GHz....................... 97
EDFAs, MUX BW = 3.125 GHz and DEMUX BW = 80 GHz ................................. 97
Figure 4.87: Eye Diagram of 12.5 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz
and DEMUX BW = 80 GHz .............................................................................. 98
Figure 4.88: Q Factor vs BER of 12.5 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz..................................................................................... 98
Figure 4.89: Eye Diagram of 12.5 GHz Channel Spacing Using RZ, Post Symmetric Dispersion
Figure 4.90: Q Factor vs BER of 12.5 GHz Spacing Using RZ, Post Symmetric Dispersion with 3
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 12.5 GHz....................... 100
EDFAs, MUX BW = 80 GHz and DEMUX BW = 12.5 GHz ................................. 100
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 6.25 GHz....................... 101
EDFAs, MUX BW = 80 GHz and DEMUX BW = 6.25 GHz ................................. 101
Figure 4.96: Q factor vs BER of 12.5 GHz Spacing Using RZ, Post Symmetric Dispersion with 3
xiv
with 3 EDFAs, MUX BW = 3.125 GHz and DEMUX BW = 80 GHz..................... 104
Figure 4.100: Q Factor vs BER Diagram of 12.5 GHz Spacing Using RZ, Post Symmetric
Dispersion with 3 EDFAs, MUX BW = 3.125 GHz and DEMUX BW = 80 GHz.. 104
Figure 4.101: Eye Diagram of 12.5 GHz Channel Spacing Using MDRZ, Post Symmetric
Figure 4.102: Q Factor vs BER of 12.5 GHz Spacing Using MDRZ, Post Symmetric Dispersion
with 3 EDFAs, MUX BW = 3.125 GHz and DEMUX BW = 80 GHz .................. 105
Figure 4.103: Eye Diagram of 6.25 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80
GHz and DEMUX BW = 80 GHz ................................................................... 106
Figure 4.104: Q factor vs BER of 6.25 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz ................................................................................ 106
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz ....................... 107
Figure 4.106: Q factor vs BER of 6.25 GHz Spacing Using NRZ, Post Symmetric Dispersion with
3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz ............................... 107
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 6.25 GHz .................... 108
3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 6.25 GHz ............................ 108
with 3 EDFAs, MUX BW = 6.25 GHz and DEMUX BW = 80 GHz .................... 109
3 EDFAs, MUX BW = 6.25 GHz and DEMUX BW = 80 GHz ............................ 109
3 EDFAs, MUX BW = 3.125 GHz and DEMUX BW = 80 GHz .......................... 110
Figure 4.113: Eye Diagram of 6.25 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz
and DEMUX BW = 80 GHz .......................................................................... 111
Figure 4.114: Q Factor vs BER of 6.25 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz ................................................................................ 111
xv
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz ....................... 112
EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz.................................. 112
with 3 EDFAs, MUX BW = 80 GHz and DEMUX BW = 6.25 GHz .................... 113
EDFAs, MUX BW = 80 GHz and DEMUX BW = 6.25 GHz ............................... 113
with 3 EDFAs, MUX BW = 6.25 GHz and DEMUX BW = 80 GHz .................... 114
EDFAs, MUX BW = 6.25 GHz and DEMUX BW = 80 GHz ............................... 114
EDFAs, MUX BW = 3.125 GHz and DEMUX BW = 80 GHz ............................. 115
Figure 4.123: Eye Diagram of 6.25 GHz Channel Spacing Using MDRZ, Post Symmetric
Figure 4.124: Q Factor vs BER of 6.25 GHz Spacing Using MDRZ, Post Symmetric Dispersion
Figure 5.1: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 100GHz spacing .. 129
Figure 5.2: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 50GHz spacing .... 131
Figure 5.3: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 25GHz spacing .... 132
Figure 5.4: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 12.5 GHz spacing 134
Figure 5.5: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 6.25 GHz spacing 135
Figure 5.6: Comparison of Signal Performance of 100GHz spacing for Various MUX/DEMUX
BW Configurations......................................................................................... 137
Figure 5.7: Comparison of Signal Performance of 50GHz spacing for Various MUX/DEMUX BW
Configurations ............................................................................................... 139
Configurations ............................................................................................... 141
xvi
Configurations ............................................................................................... 142
Figure 5.10: Comparison of Signal Performance of 12.5GHz spacing for Various MUX/DEMUX
BW Configurations ...................................................................................... 144
Figure 5.11: Comparison of Signal Performance of 6.25GHz spacing for Various MUX/DEMUX
BW Configurations ...................................................................................... 145
Figure 5.12: Comparison of Signal Performance of 12.5GHz spacing using various Line Coding
Modulation Formats .................................................................................... 147
Figure 5.13: Comparison of Signal Performance of 6.25GHz spacing using various Line Coding
Modulation Formats .................................................................................... 147
xvii
LIST OF TABLES
Table 2.1: Differences between NRZ and RZ ....................................................................... 26

Table 5.1: OSNR Comparison of Post Compensation Technique with 1 EDFA vs Post Symmetric
compensation Technique with 3 EDFAs [2] ....................................................... 120
Table 5.2: Results of 100GHz channel spacing with various design parameters and modulation
types ............................................................................................................... 122
Table 5.3: Results of 50GHz channel spacing with various design parameters and modulation
types ............................................................................................................... 123
Table 5.4: Results of 25 GHz channel spacing with various design parameters and modulation
types ............................................................................................................... 124
Table 5.5: Results of 12.5 GHz channel spacing with various design parameters and modulation
types ............................................................................................................... 126
Table 5.6: Results of 6.25 GHz channel spacing with various design parameters and modulation
types ............................................................................................................... 127
xviii
LIST OF ACRONYMS
Acronyms Description
ATM Asynchronous Transfer Mode
BER Bit Error Rate
BW Bandwidth
CD Chromatic Dispersion
CW Continuous Wave
DCF Dispersion-Compensating Fiber
DEMUX Demultiplexer
DSF Dispersion-Shifted Fiber
DWDM Dense Wavelength Division Multiplexing
EDFA Erbium-Doped Fiber Amplifier
ER Extinction Ratio
ES Equally Spaced
FDM Frequency Division Multiplexing
FEC Forward Error Correction
FWM Four-Wave Mixing
GUI Graphical User Interface
IP Internet Protocol
ITU International Telecommunication Union
LAN Local Area Network
LED Light-Emitting Diode
MAN Metropolitan Area Network
MDRZ Modified Duobinary Return to Zero
MOA Mode of Operation
MQW Multiquantum Well
MZ Mach-Zehnder
MUX Multiplexer
xix
Acronyms Description
NRZ Nonreturn To Zero
NZDSF Nonzero Dispersion Shifted Fiber
OC Optical Carriers
O/E/O Optical/Electrical/Optical
OSNR Optical Signal-To-Noise Ratio
PCM Pulse Code Modulation
PDH Plesiochronous Digital Hierarchy
PMD Polarization Mode Dispersion
PSD Power Spectral Density
RoF Radio over fiber
RZ Return to Zero
SBS Stimulated Brillouin Scattering
SDH Synchronous Digital Hierarchy
SMF Single-Mode Fiber
SOA Semiconductor Optical Amplifier
SPM Self-Phase Modulation
SRS Stimulated Raman Scattering
STM Synchronous Transport Module
TDM Time Division Multiplexing
WDM Wavelength Division Multiplexing
XPM Cross Phase Modulation
xx
GENERAL INTRODUCTION TO THE RESEARCH IN DENSE WAVE
DIVISION MULTIPLEXING
1.1 Introduction
The explosion in the demand for information brought about by the internet is creating
enormous needs for capacity expansion in the next generation telecommunication networks.
The demand for increasing bandwidth (voice, video, and data traffic) continues to drive the
demand for ongoing evolution in fiber optical networks. Even though optical networks were
widely regarded as the ultimate solution to the bandwidth needs of future communication
systems, the telecommunications industry is still struggling to keep up with the growth pace
of bandwidth requirements. To fulfil the enormously increasing capacity requirements,
especially in long haul optical link communications, it had become imperative to increase the
number of channels as well as the data rate per channel. Therefore, to reach the future
customer demands, more capacity is needed, so carriers had three possible solutions: Firstly,
install a new fiber infrastructure with or secondly, invest in new Time Division Multiplexing
(TDM) technology to achieve faster bit rates or thirdly, deploy dense wavelength division
multiplexing (DWDM). This dissertation focuses on the latter.
1.2 Research Motivation

DWDM is an extension of wavelength division multiplexing (WDM) optical networking systems
and its capability to transmit multiple ultra-high bit rate channels at narrow channel spacings
is an attractive approach to meet the challenges of current optical communication systems
and thereby increase the capacity of future optical networks.
In this dissertation, the capability and limitations of DWDM systems to scale WDM channels
to even more high dense dimensions are exploited to further increase the transmission
capacity gains of existing communication systems. The use of line encoding modulation
formats to improve the quality of signal (QoS) is explored by minimising the performance
degradations caused by the reductions in channel spacings. The availability and versatility of
the Optisystem platform is utilised to simulate and evaluate the transmission of a such a
DWDM signal for various design parameters. Evaluation of signal quality and performance
1
limitations is possible due to eye diagram and bit error rate (BER) tools found within the
Optisystem platform.
The literature review shows that there is very little research that has been done by other
researchers in finding out the impact of varying the channel spacings of a 48-channel x 25Gbps
DWDM system from 100 GHz, 50 GHz, 25 GHz, 12.5 GHz, and 6.25 GHz with modulation
formats of non-return to zero (NRZ), return to zero (RZ) and modified duobinary return to zero
(MDRZ) for the novel post symmetric compensation optical link setup. Furthermore, the
international telecommunication union (ITU) standard supported by ITU-T G.694.1 only
recommends fixed channel spacings ranging from 12.5 GHz to 100 GHz as stated in [2, 18, 50].
Therefore, there is an untapped channel spacing of 6.25 GHz that is used in this dissertation
to further expand the channel capacity of the ITU frequency grid and for that purpose, it is
essential to design and analyse a pilot network that is able to test the limitations of its
transmission parameters.
1.3 Research Objectives

1. Simulate and evaluate the performance analysis of the 48 Channel DWDM system with
a bit rate of 25 Gb/s.
2. Simulate and evaluate the effect of varying the channel spacing (100 GHz, 50 GHz, 25
GHZ, 12.5 GHz and 6.25 GHz) with respect to the above bit rate.
3. Simulate and evaluate the effect of varying the modulation format (NRZ, RZ, MD-RZ)
with respect to the above bit rate and channel spacings to optimise signal quality.
4. Validate the most optimal specification from the given parameters of the 48 channel
DWDM system that provides the best spectral bandwidth efficiency and signal
performance.
1.4 Research Methodology

A quantitative research approach was taken to collect information from various literature
reviews and other related work done previously. This helped synthesise a comparative
research study, which justified the research approach and the design parameters used to
optimise a 48-channel DWDM network. The research approach was as follows:
1. A 32 channel DWDM design model was recreated and verified on the Optisystem
platform using the parameters from existing literature reviews. This served as a pilot
2
to help one understand the procedure to design a DWDM network using the
Optisystem simulation platform. The 32-channel model was then modified to design a
novel comparative study that helped understand the transmission capabilities and
limitations of increasing bit rates at different transmission distances when using the
NRZ and RZ modulation formats in combination with a 1-erbium doped fiber amplifier
(EDFA) post compensation technique [1].
2. A novel comparative study of a 48 channel DWDM network was done using
mismatched active components. This helped understand the increasing and decreasing
effects of linear and non-linear losses for various channel spacings using the NRZ and
RZ modulation formats in combination with a novel 3-EDFA post symmetric dispersion
compensation technique [2].
3. The parameters collected from the calculations, literature reviews and the previous
comparative studies were then used to simulate numerous 48-channel DWDM
network models using various design parameters. In this dissertation, the comparative
study and optimisation of the 48 channel DWDM networks was done using matched
active components for each of the varied parameters. The simulated results helped
validate the most optimal design specification that gave the best signal performance
when using the NRZ, RZ and MDRZ modulation formats at very narrow channel
spacings in combination with the 3-EDFA post symmetric dispersion compensation
technique.
4. In each of the above comparative studies, the optical signal-to-noise ratio (OSNR),
received power and optical link loss calculations were done manually and then verified
using the Optisystem platform.
5. The performance of the signals for each channel was evaluated by the analysis of the
quality factor (Q) and bit error rate (BER) of eye diagrams.
1.5 Publications
Parts of the work presented in this dissertation have been presented and submitted by the
Author at two conferences:
1. J.Z. Mohan and O. A. Sokoya, "Comparison of NRZ and RZ in a 32x5Gbps and a 32x10Gbps
DWDM Networks for 100km and 500km," Proceedings of The Southern Africa
3
Telecommunication Networks and Applications Conference (SATNAC), Hermanus,
Western Cape, South Africa, 2018, pp. 142-148
2. J.Z. Mohan and O. A. Sokoya, "Performance Analysis of a 48 x 25Gb/s DWDM system at
Various Channel Spacings with NRZ and RZ," Proceedings of The Southern Africa
Telecommunication Networks and Applications Conference (SATNAC), Ballito, KwaZulu
Natal, South Africa, 2019, pp. 172-177
1.6 Original Research Contributions

 The dissertation extends the work done in [1, 2, 13, 16 – 29] by performing an
investigation of a 32 channel and a 48 Channel DWDM system at different bit rates (5,
10, and 25 Gb/s), for different channel spacings (6.25 GHz, 12.5 GHz, 25 GHZ, 50 GHz
and 100 GHz), different transmission distances (100 km and 500 km) and different
modulation formats (NRZ, RZ, MDRZ).
 The dissertation modifies the existing OSNR and received power equations for the
current design setups. It explores the effects of three modulation formats (NRZ, RZ and
MDRZ), transmission impairments caused by non-linear effects such as Four wave
mixing (FWM) and the impact of using a novel dispersion compensation technique (i.e.
a post symmetric compensation technique with three EDFAs) on the current design
setup. Furthermore, the dissertation implements novel ideas to combat the effects of
FWM and thereby enhance signal quality by modifying certain design parameters of
the above-mentioned channel spacings. The post symmetric compensation technique
with the three EDFAs is a modified setup of the post compensation technique shown
in [1]. This all then leads to the following research contributions:
 Improves the QoS of the existing 48 channel networks operating at 100 GHz
channel spacing to the point that there are no bit errors received by the optical
receiver for any channel within that network i.e. achieves a BER = 0 for the
received signals.
 Produces an acceptable QoS for each of the signals transmitted in 100 GHZ, 50
GHZ, 25 GHZ, 12.5 GHz and 6.25 GHZ channel spacings where Quality factor >
6 and BER > 10-9 and in each instance, the optical receiver is able to detect each
of these signals successfully.
4
 Produces an acceptable QoS for the 6.25 GHz channel spacing with the least
amount of errors since there was very little research done by other researchers
at this channel spacing. This allows the potential for the ITU Frequency grid to
be expanded to 6.25 GHz channel spacing.
1.7 Dissertation Overview

This dissertation is divided into six chapters.
Chapter 1 enumerates some of the challenges faced in an optical communication networks

due to future bandwidth requirements that demands growth in optical communication
systems. The motivation that led to the topic choice of this research is described. A research
plan showing the direction and approach of the research is listed and explains with regards to
the intermediate objectives and the research methodology. The original contribution to the
engineering body of knowledge is enumerated and explained. The publications that resulted
due to the research of this dissertation is also mentioned.
Chapter 2 is a literature survey on fiber optic communication with the focus being on DWDM
fiber optic communication systems. The evolution of fundamental concepts that led to the
development of DWDM communication is explained. The advantages and performance
limitations factors affecting DWDM system is described. A small sample of all the literature
surveys done for this research is shown to understand the limitations of DWDM systems and
the extent of work done by other researchers. The fundamental building blocks, components
and its design properties used for this research are also explained.
Chapter 3 analyses the performance of a 48 x 25 Gb/s DWDM system at various channel

spacings (100 GHz, 50 GHz, 25 GHz, 12.5 GHz and 6.25 GHz) using NRZ and RZ modulation
formats with a multiplexer (MUX) and demultiplexer (DEMUX) that has equally spaced (ES)
channels. Various design parameters such as the bandwidth of the MUX and DEMUX, type of
filter used in the MUX and DEMUX, the extinction ratio (ER) of the WDM transmitter and the
MDRZ advanced modulation format are compared to optimise the DWDM system for
bandwidth efficiency. The simulated results of the designed networks are analysed, compared
and discussed for signal performance variations in the above-mentioned design parameters.
Optimised design parameters are found for each channel spacing.
5
Chapter 4 illustrates the simulation results of the 48 channel DWDM system in the form of eye
diagrams that represented the quality of the received signals. The results for all 48 channels
for each type of network used in this dissertation is tabulated and represented on a Q factor
vs BER chart. It also illustrates the eye diagram for the channel with the highest Q factor for
analysis.
Chapter 5 analyses and discusses the results obtained in chapter 4. The results in chapter 4
are tabulated in this chapter for the purpose of comparing the effect of varying the design
parameters on the signal performance of each type of network.
In Chapter 6, the final conclusions and future recommendations are drawn in this dissertation.
6
LITERATURE REVIEW
2.1 Introduction
In this chapter, a literature review is done to obtain the necessary background information on
the evolution of transmission media and the fundamental concepts in fiber optic
communications with the main focus being on dense wave division multiplexing (DWDM)
technology. Literature reviews were compiled from credible online sources, various
textbooks, equipment specification datasheets, equipment manufacturer manuals, research
journals and conference article entries. Also, later in this chapter, the literature for the design
and performance measuring criteria of DWDM system components used in this dissertation is
expatiated.
2.2 Evolution of Fundamental Concepts in Fiber Optical Communication

The rapid growing increase in the usage of new applications, broadband data services,
advanced internet traffic, changing patterns of usage and redistribution of services such as
data, voice and video always causes a bandwidth bottleneck in telecommunications networks.
This meant that the nature of communication systems is always in a state of ongoing evolution
[3]. Fiber optic communication started replacing analogue electrical communication (copper
cables) for long haul communication between the 1950s and 1960s with the introduction of
pulse code modulation (PCM) [3]. PCM involved sampling, quantizing, and coding analogue
telephone voice signals to produce a compressed binary digital signal [4], [5]. Fiber optic
communication systems used optical multiplexing techniques that made use of carrier waves
in specific regions of the optical spectrum to modulate light waves that allowed the
transmission of digital signals up to a few gigahertz or gigabits per second [6]. The earliest
fiber systems operated around 850 nm, the first window in the optical spectrum. A second
window (S band), at 1310 nm, started being used due to its lower attenuation characteristics.
The third window (C band) at 1550 nm had an even lower optical loss. Today, a fourth window
(L band) near 1625 nm is under development and early deployment [3], [6].
By the 1970s and 1980s, optical communication systems used hierarchical multiplexing
techniques such as Plesiochronous Digital Hierarchy (PDH) and Synchronous Digital Hierarchy
7
(SDH) respectively. PDH was a four-group multiplexing system that used multiple stages of
multiplexing or de-multiplexing to move from a lower bit rate channel to a higher bit rate
channel or vice versa. PDH was not fully synchronous and was plesiochronous in nature, which
according to the ITU-T G.702 recommendations [8], meant that signals were transmitted
nominally at the same bit rate but were not synchronized to each other by a common master
clock. The maximum data rate for a PDH carrier signal was an E4 (European standard, level 4),
which was 140 Mb/s [4], [7]. Unlike PDH, SDH used a repeated hierarchy of fixed length
frames, which carried isochronous traffic channels that were able to encompass asynchronous
transfer mode (ATM) frames, PDH frames, IP Packets and ethernet frames. SDH eliminated
mountains of multiplexers by allowing a single stage of multiplexing and de-multiplexing to
transfer multiple digital bit streams synchronously over an optical fiber cable using a single
master clock thereby reducing the hardware complexities of PDH technology. The maximum
data rate for SDH carrier signal was a STM256 (Synchronous Transport Module, level 256),
which was 40 Gb/s [7].
PDH and SDH were both time division multiplexed (TDM) systems that increased the capacity
of a long haul transmission link by slicing the time into smaller intervals so that the bits from
multiple input sources can be transmitted on a single trunked wavelength carrier (either 1310
nm or 1550 nm) for each fiber core in a single mode fiber (SMF) optic cable link, effectively
increasing the number of bits transmitted per second [5], [7]. This meant that for a full duplex
communication system, a pair of fiber cores would be needed to send and receive a signal
between two terminal equipment. However, to increase the bandwidth and channel capacity
of the network, one would need to increase the number of fibers cores used. And whenever
the bandwidth in the existing fiber cores of a fiber cable were exhausted, one would need to
replace the fiber optic infrastructure with a fiber cable that had a larger number of fiber cores,
which ultimately was a costly affair [3].
The need to increase bit rates and bandwidths without changing the existing fiber optic
infrastructure led to the development of wavelength division multiplexing (WDM) in the late
1980s using two widely spaced wavelengths in the 1310nm and 1550nm regions. This
technology was able to propagate two carrier waves, each having its own wavelength, to
transmit and receive signals over a single fiber core [3].
8
WDM formed the foundation building block that led to the development of Dense Wavelength
Division Multiplexing (DWDM), which had been used since 1999 [5]. In DWDM, several carrier
waves can be transmitted over multiple wavelengths that are on the same fiber core with
minimal losses and interactions between each wavelength. Each carrier wave has its own
frequency that corresponded to a different wavelength. DWDM uses a transmitter that had
different light sources to emit different wavelengths of light pulses. These wavelengths are
narrowly spaced apart at a typical frequency spacing of 100 GHz or less in the C and L
frequency bands. These emitted signals are then coupled as a composite signal and injected
into a single optical fiber core by a multiplexer (MUX). After transmission on the fiber, a de-
multiplexer (DEMUX) separates the composite signal into the different wavelength signals.
The DEMUX then sends each of these signals towards its own photodetector at the receiving
end. See Figure 2.1 [3].
Figure 2.1: Timeline and Overview of DWDM Communication System [3]
9
DWDM technology eases the bandwidth bottleneck by increasing channel capacity and bit
rate without needing to change the existing fiber optic infrastructure to increase bandwidth.
However, due to the “Data Tsunami” era, there is a greater demand to increase transmission
distance and channel capacity of data traffic at even higher bit rates, smaller channel spacings
and higher quality of service (QoS). This gives rise to linear effects (attenuation and dispersion)
and non-linear effects. In DWDM, four-wave mixing (FWM) is the most critical type of non-
linear effect. All these unwanted effects must be managed and optimised for efficient spectral
efficiency by line encoding modulation formats, which has a direct effect on the accuracy and
performance of a DWDM signal. The focus of this dissertation is to use three type of line
encoding modulations formats namely, non-return to zero (NRZ), return to zero (RZ) and
Modified duobinary return to zero (MDRZ) to minimise the effects of attenuation, dispersion
and FWM when decreasing channel spacings from 100 GHz to 6.25 GHz.
2.3 Advantages of DWDM

DWDM has the ability to provide potentially limitless transmission capacity, which is
dependant only by the properties of the equipment that make up the system. The money
invested on the existing fiber infrastructure can not only be retained, but can also be
optimised. This is because as demands increase, additional capacity can be added, either by
simple equipment upgrades or by increasing the number of wavelengths on the fiber, thereby
increasing the bandwidth without the need for high costly upgrades. Besides Bandwidth,
DWDM’s most enthralling technical advantages were as follows [3]:
• Transparency – DWDM provisions for TDM and other data formats such as ATM,
Gigabit Ethernet, PDH and SDH fiber channels with open interfaces transparently over
its shared architectural physical layer [3].
• Cost-effective integration – By using DWDM as a transportation layer for TDM, existing
SDH equipment investments can be preserved. Not only can one integrate existing SDH
equipment to DWDM equipment effortlessly, but one can also eliminate layers of SDH
multiplexing equipment that would be needed for future SDH requirements by
interfacing directly to DWDM equipment from ATM and packet switches [3].
• Scalability – DWDM leverages the large amount of dark fibers found within many
metropolitan and enterprise networks to swiftly meet the rapid demand for capacity
on point-to-point links and on spans of existing SDH rings [3].
10
• Dynamic provisioning – DWDM provides fast, simple, and dynamic provisioning of
network connections, which equips service providers with the capability to deliver a
high-bandwidth of services to customers within a matter of days [3].
2.4 Performance limitation factors affecting DWDM systems

Fiber optic transmission presents several limitations that must be dealt with when designing
a DWDM system. These fall into two broad categories namely linear effects and non-linear
effects. These effects had several causes, not all of which affect DWDM. The following section
only discusses those causes that were relevant to DWDM.
2.4.1 Linear effects

• Attenuation – This is the deterioration of the signal strength, or the loss of optical
power as the signal is transmitted through the fiber. Attenuation in a fiber optic link
occurs due to the internal and external factors of the fiber optic link.
 Internal factors such as Rayleigh scattering and absorption are caused by
density variations and impurities found within the fiber optic glass of the cable,
respectively. These factors are directly dependant on the length of the fiber
and the wavelength of the light, therefore an increase in these parameters will
increase the attenuation. In this dissertation, the attenuation due to internal
factors of the fiber will be found in the non-zero dispersion shifted fiber
(NZDSF) and the dispersion compensated fiber (DCF). This attenuation loss is
measured in decibels per kilometre (dB/km) [3], [9].
 External factors such as link loss mechanisms are caused by the insertion loss
of active components and the noise figure of the optical amplifier in the fiber
optic link. Active components that exhibit loss in this dissertation include the
Lithium-niobate Mach–Zehnder Modulator (Linb MZM), WDM transmitter,
WDM MUX and DEMUX, connector and the optical receiver. The attenuation
(or loss) of these factors, for a given wavelength, is defined as the ratio
between the input power and the output power of the fiber being measured.
It is expressed in decibels (dB) [3], [9].
• Chromatic dispersion – This is the spreading phenomena of light pulses as they
travelled down the fiber. Chromatic dispersion (CD) occurs in DWDM because a single
11
light pulse that is transmitted through a fiber core is composed of multiple
wavelengths, each travelling at different speeds through the fiber. The interaction
between the dissimilar propagation speeds widens the light pulse when it arrives at
the receiver, which in turn causes a reduction in the optical signal-to-noise ratio
(OSNR) and an increase in the number of bit errors. The CD coefficient (referred as D)
of a given fiber is the normalised value that represents the relative arrival delay (in ps)
of two wavelength components separated by one nanometre (nm) per each kilometre
(km) of transmission, expressed in ps/(nm x km) [3], [9]. For the ITU G.655D
compliant [10] non-zero dispersion shifted fiber (NZDSF) used in this dissertation, the
chromatic dispersion coefficient (𝐷𝐷) for a centre wavelength of 1550 nm is chosen
using the expression shown below in (2.1), which is taken from [10]. The expression in
(2.1) is chosen since the wavelength range used for the design in this dissertation is
from 1460 nm to 1550 nm:
7.00
⎧𝐷𝐷𝑚𝑚𝑚𝑚𝑚𝑚 (𝜆𝜆 ) , (𝜆𝜆 − 1460) − 4.20⎫
⎪ 90 ⎪
𝐷𝐷 (𝜆𝜆 )1460𝑛𝑛𝑛𝑛 𝑡𝑡𝑡𝑡 1550𝑛𝑛𝑛𝑛 = (2.1)
⎨ 2.91 ⎬
⎪𝐷𝐷𝑚𝑚𝑚𝑚𝑚𝑚 (𝜆𝜆 ) , (𝜆𝜆 − 1460) + 3.29⎪
⎩ 90 ⎭
• Polarization Mode Dispersion

Unlike CD, Polarization mode dispersion (PMD) is the smearing of a signal, which is
caused by an interaction between pulses that happens over time. This interaction
happens due to the ovality of the fiber shape from external stressors or because of the
lack of maintenance done to create a consistent birefringence pattern along the length
in the fiber’s perpendicular polarisation modes during the manufacturing process.
Birefringence is the phenomenon where light is divided into two different pathways.
Most SMF have two polarisation states, one is vertical and the other is horizontal [3],
[11]. For this dissertation, we assume that PMD is negligible.
2.4.2 Non-linear effects

As opposed to linear effects that can be compensated, non-linearities are increasingly
additive effects formed by the continuous interaction of the light signal with the fiber
medium through which it travels, which results in deleterious changes to the light
signal. The high-power level and the large number of optical channels used in the
12
DWDM system are the main causes to non-linear effects, which are the fundamental
limiting factors that determines the total quantity of data that can be transmitted in
optical fiber. There are different types of nonlinear effects namely, stimulated Brillouin
scattering (SBS), stimulated Raman scattering (SRS), self-phase modulation (SPM),
Cross-phase modulation (XPM), and four-wave mixing (FWM). In DWDM, FWM is the
most critical of these types [3], [11].
• Stimulated Brillouin Scattering (SBS) and Stimulated Raman Scattering (SRS)

SBS is a backscattering phenomenon that is evident when an induced periodic change
in the fiber refractive index causes a loss of power and scattering of signal, however it
only occurs when a few channels are transmitted. SRS is a scattering effect (formed
only when using the Raman optical amplifier) that causes a transfer of signal power
from a shorter wavelength to a longer wavelength signal; and in the process of doing
so, scatters the light in all directions. SBS and SRS are not applicable for the design used
in this dissertation since many channels are amplified using the EDFA instead of the
Raman [3], [11].
• Self-Phase Modulation (SPM) and Cross-Phase Modulation (XPM)

SPM produces a time varying phase change in the same channel and XPM produces a
phase change on another channel; both SPM and XPM are formed by high signal
intensities that changes the fiber’s refractive index and causes spectral broadening of
the optical signal [3], [11].
• Four Wave Mixing (FWM)

FWM is an interference phenomenon that degrades the performance of a DWDM
system because it produces a number of new unwanted signal frequency (𝑓𝑓𝑁𝑁 ≡ 𝑓𝑓𝑖𝑖𝑖𝑖𝑖𝑖 )
from a combination of three different signal frequencies 𝑓𝑓𝑖𝑖 , 𝑓𝑓𝑗𝑗 and 𝑓𝑓𝑘𝑘 as shown below
in [3], [12]:
( 𝑓𝑓𝑁𝑁 ≡ 𝑓𝑓𝑖𝑖𝑖𝑖𝑖𝑖 = 𝑓𝑓𝑖𝑖 ± 𝑓𝑓𝑗𝑗 ± 𝑓𝑓𝑘𝑘 ) (2.2)
where i, j and k vary from 1 to N and N is the total number of DWDM channels.
13
This creates a large number of unwanted FWM components (𝑁𝑁𝐹𝐹𝐹𝐹𝐹𝐹 ), which can be
calculated using the expression shown in (2.3), which is taken from [12]:
𝑁𝑁 2. (𝑁𝑁 − 1)
𝑁𝑁𝐹𝐹𝐹𝐹𝐹𝐹 = (2.3)
2
The new unwanted channels are ghost channels, which are formed when any
combination of the three different channels produced a fourth channel. Due to the
high-power levels, ghost channels may either be a new sideband frequency to one of
the other frequencies or it may overlap with one of the other frequencies [3]. The
effect of FWM on a 3 channel WDM system is illustrated in [3] as an example. By using
the equation above, for a 3 channel (λ1, λ2, λ3) system, NFWM = 9.
Figure 2.2: Illustration of FWM on a three channel WDM system [3]
The three different channels induce a fourth channel known as ghost channel. Due to
high power levels, FWM effects produce several ghost channels (some of which
overlap actual signal channels), depending on the number of actual signal channels.
Similarly, it could be deduced that for a 48 channel DWDM system used in this
dissertation, there are a potential of 54144 unwanted FWM frequency components or
ghost channels that will degrade system performance as the channel spacing is
decreased. Therefore, FWM is one of the most adverse nonlinear effects in DWDM
systems. FWM effects limit the channel capacity of DWDM systems and it increases
due to the following factors [3], [12]:
 Increase in the length of the fiber
 Increase in the number of channels
 Increase in power levels
 Decrease in chromatic dispersion
 Decrease in channel spacings
14
Dispersion-shifted fiber (DSF) is one of the types of fibers that is greatly affected by the
effects of FWM due to its zero-dispersion property near 1550nm, making it unsuitable
for DWDM systems. Manufacturers subsequently created the non-dispersion-shifted
fiber (NZDSF) with the above point in mind because this fiber had a certain amount of
chromatic dispersion (CD) that occurs around 1550 nm region. This leads to different
wavelengths having different group velocities, and thereby causes a reduction in the
non-linear effects of FWM [3].
2.5 Exploring the limitation and extent of existing research

The need for a DWDM system is application dependent with varying parameters such as signal
data rate and number of users. The research of this dissertation led to the publication of two
conference articles.
The first article in [1] dealt with the analysis of a 32 channel DWDM network for bit rates of 5
Gb/s and 10 Gb/s, a channel spacing of 100 GHz and transmission distances of 100km and
500km. The Q factor of the NRZ and RZ modulation formats were compared in combination
with a post compensation technique that used 1 EDFA. The obtained results showed that the
NRZ modulation technique performed better in producing a higher quality of signal for long
haul DWDM links. Long haul links have non-linear losses that affected the transmitted signals
because it produced noise. In the 5 Gb/s DWDM network for small distances such as 100 km,
the RZ technique had a better performance due to the higher Q-factor and lower BER for short
distances. However, for the 5 Gb/s network at 500 km and the 10 Gb/s DWDM networks for
both 100km and 500km, NRZ modulation was the preferred choice because of its low BER,
high Q-factor and higher eye heights.
The second article in [2] dealt with the performance analysis of a 48 channel DWDM network,
which had mismatched active components, that evaluated the effect of linear and non-linear
losses for channel spacings of 100 GHz, 50 GHz, 25 GHZ and 12.5 GHz and 6.25 GHz when using
the NRZ and RZ modulation formats in combination with a post symmetric dispersion
compensation technique that used 3 EDFAs. The modifications of the optical link by adding
symmetric EDFAs significantly improved signal quality because it compensated for the
increased dispersion and in turn also lowered the effects of non-linearities to a certain extent
for smaller channel spacings. Non-linear effects of FWM was mostly supressed at a 100GHz
15
channel spacing for NRZ. The obtained results showed that the NRZ modulation technique at
a channel spacing of 100GHz performed better than RZ. Lower channel spacings for both NRZ
and RZ for 50GHz, 25GHz, 12.5GHz and 6.25GHz were highly susceptible to non-linear losses
that produced noise.
In addition to the literature surveys done by the author in the two published conference
articles researched in [1] and [2]; the following represents some of the many other literature
surveys done to explore the extent of the work done by other researchers in the topic of point-
to-point DWDM systems for long-haul communications:
Subash and Babu [13] performed an iterative analysis to obtain an optimized DWDM system
in terms of data rate, channel spacing and number of channels. The system was designed and
analysed for three channel numbers of 16, 32 and 64. For each of these three channels, the
output was analysed by varying the bit rate by 5 Gb/s, 10 Gb/s and 20 Gb/s and channel
spacing by 30 GHz, 40 GHz and 50 GHz.
Kaur and Sharma [16] presented an intensive review of the DWDM and its hard-core analysis.
The most exponential changes in communication engineering had taken place in the last
several decades with the advent of DWDM to explore the untapped potential of unlimited
bandwidth on a fiber pair. Originally applied to solve the problem of fiber saturation in long-
distance networks, DWDM was becoming the leading player in all-topical networks. DWDM
was insensitive to bit rate and permits communication networks to enhance its transmission
capacity without installing new fibers. This was done at a decreased cost and with much
greater flexibility than other communication systems. DWDM installs narrowly spaced
channels onto the same fiber and improve capacity on existing fiber routes by factors of 16,
32, 40, or more.
Antil et al. [17] provided an overview of DWDM networks and its current technologies.
Analysis of applications and roles of current network protocols in the future DWDM
frameworks were also provided. The article covered the functions and applications of DWDM
system components. The operation of each component was discussed individually. The
interaction of all the components associated with a DWDM system was critical to system
design. Fiber optic terminology like attenuation, dispersion and optical signal to noise ratio
(OSNR) were factors that measure the quality of the optical signal. These key factors help to
16
design and optimise a DWDM system. During the design process from transmitter to receiver,
the strength (power level) and quality (OSNR and dispersion) of every optical signal must be
maintained. Optical amplifiers extended the reach of DWDM systems by overcoming losses
owing to attenuation, but could cause OSNR problems. Dispersion compensation devices and
dispersion compensated fibres reduced the amount of dispersion of a span, hence it increased
the transmission distance.
Sajjan et al. [18] gave a detailed description about link design and optical power budget
calculations in a DWDM network. Power budget analysis involved the description of gain
accumulation in the signal due to erbium-doped fiber amplifiers (EDFAs), attenuation
calculation owing to fiber and insertion losses caused by reconfigurable optical add-drop
multiplexers (ROADMs) and dispersion compensation modules (DCMs). For DWDM network
design, it was very necessary to obtain optimised power values at the EDFA nodes that were
well within the range specified. Power management was an important task to achieve optical
receive power for better bit error rate (BER). Output power at the optical amplifier (OA)
correlated to the number of channels passing through them. Dispersion compensation was
also very much necessary so that signal was efficiently detectable at the receiver. Placement
of DWDM components in a network required the analysis of the distance between two nodes,
type of data rate, technical specifications of the component, their gain and insertion losses.
DWDM network setup required various network components to make the system work.
Optimal optical power requirements were the major constraints in these kinds of networks.
The DWDM system considered in this article was designed to carry 80 channels in a 1550nm
band. This provided 50 GHz channel spacing (approx. 0.4 nm), with 80 channels in a single
mode fiber. DWDM channel spectrum had been defined in ITU-T recommendation G.694.1.
The power management in a DWDM network was dependent on the number of channels
which were supported by the setup. The amplifiers used to boost the signal power were fixed
gain amplifiers. The power measurement at each component of DWDM system was calculated
separately. Calculations were accounted to the 80-channel DWDM system and calculations
were done with respect to the individual channels. Insertion loss of the network elements and
attenuation through the fiber link were considered during calculation. The relationship
between the optical power at input and output of each component in DWDM system were
discussed and was made feasible to carry all 80 channels without violating the power
17
requirements and constraints for the amplifiers as well as other components used in the
network.
Saxena and Saxena [19] studied the relationship and calculations between different
parameters related optical strength of light signal. They established that Optical Power
debugging played an important role for smooth and efficient functioning of a DWDM system.
They then evaluated the ideal and actual optical power of the light signal at input points of
DWDM components. Experimental results show that at many points, the actual value of
optical power of light signal differed from ideal value, which caused a failure of the network
design and led to traffic outage and hardware fault of DWDM components. They had also
described difference between ideal and actual value with the help of attenuation. Therefore,
the article concluded the importance of optical power debugging for better performance of
DWDM system having fixed gain amplifiers.
Moses and Lakshmy [20] aimed to find the most efficient amplifier with respect to
transmission and to identify the significant degradation factors that influenced transmission
distance. Raman-EDFA showed better performance than any other amplifiers mentioned. The
main factor that had limited transmission was amplifier-produced noise, the amount of which
was more in the case of the semiconductor optical amplifier (SOA). The main factor that held
down a transmission system with a discrete Raman amplifier was inter channel crosstalk,
which was a product of the Four Wave Mixing (FWM) non-linear effect. It was also necessary
to note that the difference in accumulated inter channel crosstalk was mostly explained by
the signal intensity difference at the amplifier output. At the same time, it was established
that the non-linear response of the SOA and the small effective area together with the high
non-linear coefficient of the discrete Raman amplifier made a serious impact on the quality of
transmission and limited the total achievable length of the link. The parameters of SOA were
adjusted so that it produced higher amplification with less signal distortions.
Kaur and Sarangal [21] investigated the performance of a 32-channel DWDM system with
post-dispersion compensation using dispersion compensated fiber (DCF) at different bit rates
(10, 20 and 40 Gbps). The simulation results showed that if the DCFs were incorporated in
system through the dispersion compensation technique, then the DWDM systems had good
18
signal performances, low bit error rates and were able to fully exploit the high-speed data
rates.
The analysis and design of a 32-Channel with 40 Gb/s point-to-point DWDM system was
proposed by Pal and Revathi [22]. By varying the modulation format, it was seen that the Q-
factor, bit error rate (BER), eye height, threshold performance gave the best result in the
return-to-zero (RZ) modulation scheme, then non-return-to-zero (NRZ), and finally Mach
Zehnder (MZ) modulation. MZ modulator had high extinction ratio and high-speed
performance. Varying channel frequency from 100 GHz to 50 GHz produced inter-symbol
interference (ISI). For single mode fiber (SMF), the DCF length was less. It produced very good
BER and Q factor, as well as good eye height. At 50 km SMF and 10 km DCF, it produced
optimum results.
Cho and Kim [23] proposed the simulation and optimization of a 32-channel DWDM system
for different modulation formats, channel spacings and line widths in the 1550 nm
transmission window using Optisystem 12.0. On the criterion of Q factor and BER, it was
observed that RZ modulation format, channel spacing 150 MHz and line width 0.3 MHz gave
the optimised performance. It was also observed that DCF and EDFA should also be used at
the appropriate place for the optimum performance.
Chaudhary et al. [24] demonstrated an error free transmission of a 32 x 45 Gb/s DWDM system
that had a channel spacing of 200 GHz with distance of 100 km based on DWDM/EDFA signals;
NRZ modulation format of the optical signal was used. The dispersion management was fully
treated by DCF as a compensator with the in-line optical amplifiers such as EDFA to improve
the optical signal-to-noise ratio (OSNR), and reduce the nonlinear effects in transmission
system. The optimum fiber length to this system was 100 km transmission distance with the
average power of -10 dBm, and the average noise power was -37 dBm for all 32-channels. This
meant that the DWDM system had a good performance, and fully exploited the high speed,
low error rate, availability of multiple channels on a single fiber, and the major contribution
was the development of the multi-destination communication over the light wave DWDM
system. This scheme was very attractive for the upgrading of current optical networks.
Expenditures may be reduced by adding more bandwidth to the 45 Gb/s DWDM system and
19
increasing the capacity of existing systems. Finally, the 32 x 45 Gb/s channels were successfully
transmitted using DWDM/EDFA architecture.
Senthamizhselvan et al. [25] analysed the performance of a DWDM based fiber optic
communication system at different modulation schemes, various power level and different
number of data channels. It was established in this article that non-linear optical effects
degraded the performance of DWDM systems. These included cross phase modulation (XPM),
self-phase modulation (SPM), four-wave mixing (FWM), stimulated brillouin scattering (SBS)
and stimulated Raman scattering (SRS). Analysis was performed on a DWDM system with NRZ,
RZ modulation schemes and DCF. The performance of RZ modulation was shown to be better
than NRZ modulation. The bit error rate was greatly reduced by using RZ modulation. The SMF
signal dispersion was compensated by using DCF with dispersion coefficient of -83.75
ps/nm/km. The DCF along with SMF was used for length of 100km at 1550nm to reduce the
dispersion of optical signal. The performance of improved detected signals was evaluated by
the analysis of quality factor and BER. The simulation studies were carried out using
Optisystem platform from Optiwave.
Singh and Bhagat [26] performed an analysis of a DWDM transmission system using different
modulation formats (NRZ and RZ), power levels (5, 8, 10, 12, 15 and 20 dBm), data
transmission rates (2.5, 5 and 10 Gb/s) and a number of information channels (8, 16, 24, and
32) at the input. A single mode fiber of length 100 km at 1550 nm wavelength was used along
with a dispersion compensating fiber to reduce dispersion. The performance of the signal
received at the receiver was analysed based on the quality factor and BER. It was found that
the presence of non-linear effects in optical fiber, namely FWM, SPM, XPM, SRS, and SBS,
degraded the performance of the DWDM transmission system. From the results presented, it
may be concluded that RZ modulation format showed a better performance as compared to
that of NRZ modulation format. This is because the quality of signal received in the case of RZ
modulation format was better than that of NRZ modulation format. Moreover, it could be
concluded that by varying the number of input channels in a DWDM system, the quality of the
signal received at the receiver end remained the same and showed negligible degradation.
Furthermore, as the transmission power level was increased up to 10 dBm, the quality of the
received signal improved, but as the transmission power level was further increased the
quality of the received signal degraded due to the increased efficiency of non-linear effects.
20
An increase in the data transmission rate degraded the performance of the received signal.
Thus, it may be concluded that for faithful transmission in case of DWDM transmission system,
the RZ modulation format must be used and the transmitting power levels should be kept
close to 10 dBm for each input channel.
Kaur et al. [27] investigated an 8 channel DWDM system for a fiber based on optical add/drop
multiplexers (OADMs) at 1550 nm window at different data rates i.e. 5, 10, 15 Gb/s. In
addition, the system was examined at different distances i.e. 800, 960, 1120, 1200, 1440 kms.
Finally, the system was examined with a frequency of 193.1 THz at different channel spacing
of 20, 40, 80, 100 GHz. Various performance parameters such as different data rates,
distances, and channel spacings were varied to investigate the system performance in terms
of the BER and quality factor. The comparison of output power at different channel spacings,
distances and data rate were made and an improved eye diagram was observed at a maximum
distance of 1440 kms.
Rajalakshmi at al. [28] simulated a 32-channel DWDM network using the RZ and NRZ
modulation formats at the data rate of 40 Gb/s. The transmitter used was a 32-channel WDM
transmitter at a frequency spacing of 100 GHz. The transmission loop was a 50km SMF optical
link with 10 km of DCF and two EDFAs. The receiver was a 32-channel WDM demultiplexer
and BER testers. The important feature of EDFA was the ability to pump the devices at
different wavelength, which was very suitable for DWDM. It had a very low coupling loss,
compatible–sized fiber transmission medium and a very low dependence of gain on light
polarization. EDFAs were highly transparent to signal format and bit rate. They were immune
to interference effects between different channels, thus only EDFAs were used for long
distance transmission. The simulated results were tested using BER analysers. It was analysed
for different modulation formats with PIN receivers using an Optisystem simulation tool.
Choudhary et al. [29] performed a comparative analysis of a 16-channel DWDM optical

communication system for different dispersion compensation schemes, namely pre-, post-
and mix-dispersion compensation schemes using DCF. They also used different modulation
formats, namely non-return-to-zero (NRZ), carrier suppressed return-to-zero (CS-RZ),
duobinary return-to-zero (DRZ) and modified duobinary return-to-zero (MDRZ) at different bit
rates of 10 Gb/s, 20Gb/s and 40 Gb/s for the comparison. They observed that at a high bit
21
rate, the MDRZ format gave a better performance than the other formats. The mix-dispersion
compensation scheme showed a better performance as compared to pre- and post-
compensation schemes. It gave better results at high bit rate (40Gb/s) when used with the
MDRZ modulation format. The results were based on the quality factor, BER and eye opening
of the received signal.
It can be seen by the work done in [13, 16 – 29] that there was very little research done for
DWDM systems operating at very narrow channel spacings of 12.5 GHz and 6.25 GHz channel
spacings with a 25 Gb/s bit rate. This dissertation focuses on achieving an optimised system
performance for a 48 channel DWDM system model that has a bit rate of 25 Gb/s for channel
spacings of 6.25 GHz, 12.5 GHz, 25 GHz, 50 GHz and 100 GHz.
2.6 DWDM System Design Components

The design of the DWDM communication system comprises mainly of nonlinear devices and
non-gaussian noise sources. Therefore, the analysis of such a communication system would
have been tremendously time-intensive since it is a highly multifaceted system. As a result,
these tasks are performed efficiently and effectively with the help of an optical
communication system design platform.
Optisystem is used as the design platform since it is an innovative optical communication

system simulation package with a graphical user interface (GUI). It helped virtually design,
simulate, test, and optimise the DWDM optical link in the physical layer of a broad spectrum
based on the realistic modelling of a DWDM fiber optic network. Optisystem version 7.0 is
used for this design.
The various components that determines the performance characteristics of the designed
DWDM system is explained in the following sections.
2.6.1 WDM Transmitter

DWDM systems require multiple transmitters for each wavelength channel and different
parameters for each one of them. In order to reduce the time consumption of designing
multiple transmitters, a WDM transmitter is used. The WDM Transmitter array encapsulated
different components that comprised the design of multiple WDM transmit channels. This
22
allows the selection of different line coding modulation formats and laser modulation scheme
types for multiple channels in one single component [29].
Figure 2.3: A 48 port WDM Transmitter and the Block Diagram of a single channel in a WDM Transmitter
Figure 2.3 shows a 48 port WDM transmitter on the left and the equivalent block diagram of
a single channel (which in this case is the 1st channel) of the WDM Transmitter on the right.
The most important components and parameters that make up the WDM transmitter is shown
in the sections that follow.
2.6.1.1 Pseudo-Random Bit Sequence Generator (PRBS)

The first stage is the PRBS; which creates a bit sequence that is designed to approximate the
characteristics of random data. A different seed of leading and trailing zeros is generated for
each bit sequence for each WDM channel [29]. The WDM transmitter had a built-in PRBS that
automatically generates a random seed of data for each channel port. When using the external
PRBS generator, one needs to set the individual bit rate parameter of each channel, whereas
with the WDM transmitter, one is able to set and apply one-bit rate for all channels.
23
2.6.1.2 Light Source
There are two types of light sources namely, light emitting diodes (LEDs) and laser diodes [29],
[3]. A laser diode is the preferred light source since it is a pure light source that provides
coherent energy with little distortion. Even though lasers are more expensive than LEDs, laser
diodes are the preferred choice because they offer a better performance than LEDs. It has
much higher transmission speeds, is able to use precise wavelengths, has a narrower spectrum
width, enough power, and has the ability to control chirp (the change in frequency of a signal
over time) [3]. Chirp is undesirable because a pulse with a broad range of wavelengths is
susceptible to dispersion through the optical fiber transmission [3].
2.6.1.3 Modulation scheme type of transmitting light source

There are two modulation schemes for the laser diode, namely a direct modulated laser (DML)
and an external modulated laser (EML) [29]. For DML, the bias current applied to the laser is
itself modulated by an electrical data signal. This variance of the DML’s electrical current
(input signal) causes an undesirable wavelength chirp effect, which in turn causes an excessive
chromatic dispersion at high speeds. For EML, the bias current to the laser is held constant,
and the constant light output from the laser is modulated by the electrical data signal. This
provides a constant level of optical power for the light-wave output data signal [31].
Compared to a DML, an EML is the preferred choice since it limited chirp effects and has a
smaller chromatic dispersion with a stable wavelength under high-speed operation. This is
because the input signal to the laser section is not modulated and therefore did not change
[32]. C-band wavelengths with a centre frequency of 1550nm are assigned to the CW lasers to
modulate the data.
2.6.1.4 Extinction Ratio

Extinction Ratio (𝐸𝐸𝑟𝑟 ) is a measure of the quality of the laser. This defines the power penalty
ratio of the laser, which is the ratio of the optical power measured at the centre of the
transmitted 1 level state (𝑃𝑃1) to that of the optical power measured at the transmitted 0 level
state (𝑃𝑃0) and is measured in decibels (dB) [33], [34]. The expression shown below is taken
from [33]:
𝑃𝑃1
𝐸𝐸𝑟𝑟 = 10. log10 � � (2.4)
𝑃𝑃0
24
The transmitted 1-level state is defined as when the light pulse is ‘on’ and produces a ‘high’
state. Transmitted 0-level state is defined as when there is ‘no light pulse’ and produces a ‘low
state’. Even though in an ideal working environment, the 0-level state should have no power,
however the laser needs a small amount of power to be emitted at the 0-level state for it to
be biased, thus 𝑃𝑃0 > 0. Usually for an optical transmitter, the Extinction Ratio or transmission
power is kept high enough for the transmission signal to be distinguished at the receiver [33],
[34]. In this dissertation, a very high 30 dB extinction ratio is used in combination with a Bessel
filter, which makes the power penalty close to 0 for most of the network designs. Signal
propagation over length of fiber in a DWDM system causes dispersion and this significantly
decreases the extinction ratio. However, the 4th order Bessel filter does not require for a
maximum or minimum channel extinction ratio to be set for its operation. For any other order
of the Bessel filter, an extinction ratio > 10dB would suffice [33].
The effect of varying the extinction ratio in combination with a rectangular filter is evaluated
for a 25 GHz channel spacing for the purpose of eliminating four wave mixing (FWM). This is
because a rectangular filter had the property to eliminate frequency harmonics without
affecting lower harmonics and signal propagation over a length of fiber in a DWDM system.
Signal propagation through a fiber causes dispersion. This dispersion significantly decreases
the extinction ratio. Therefore, in theory if the extinction ratio is also decreased, it should
increase the power and dispersion in the system. Therefore, in theory, an increase in
dispersion should decrease the effects of FWM.
2.6.1.5 Line coding Modulation format used in WDM Transmitter: NRZ and RZ
For a digital laser to transmit analogue information signals in the electrical domain over an
optical passband domain, these analogue signals have to be converted to a digital format by
means of using a line coding technique. The digital format consists of a sequence of voltage
pulses that is either a high (logic level 1) or a low (logic level 0 – no pulse). The digital signal
transmission rate is the bit rate of the signal and had the unit, bits per second (b/s). Each bit
time slot had the period T with its unit in seconds (s), where T = 1/bit rate [35]. The duty
cycle (dc) of the pulse is a fraction of the bit width (Tb) of the data over the time slot period
(T), where 𝑑𝑑𝑐𝑐 = 𝑇𝑇𝑏𝑏 / T [35]. Two common methods of coding data information in a digital
signal are non-return to zero (NRZ) and return to zero (RZ) coding [35]. Modulation of a signal
format using line coding techniques in a DWDM system is important to eliminate the direct
25
current (DC) component of a signal. This decreases the dependence on the low end of the
frequency spectrum in a baseband signal where there is very little or no transmission
capability at and near zero frequency (DC). For this reason, the polar format of the NRZ and
RZ is used. This is shown in Figure 2.4 [35]. The zero-mean and vanishing spectral density point
at zero frequency is the spectral null (Δ fnull) at DC [36].
The differences between NRZ and RZ are further explained in Table 2.1 and are illustrated in
Figure 2.4:
Table 2.1: Differences between NRZ and RZ
NRZ RZ
1. The definition of NRZ and RZ differs with regards to their modulation
techniques as stated in [35].
NRZ is a non-linear modulation technique RZ is a linear modulation form of pulse
in which the pulse does not return to amplitude modulation, in which the
zero, but rather remains on when pulse returns to the zero state when
successive ones are transmitted. transmitting a one.
2. The baseband signal power spectral density (PSD) of NRZ and RZ is shown in
[35]. The RZ pulse width is half the NRZ pulse width, however, the RZ pulse
spectral width is much wider than the NRZ spectral width because:
The bit width of NRZ is the same as its The bit width of RZ is half of the time slot
time slot width, where T = Tb and dc = 1. width, where Tb = T/2 and dc = 0.5.
The spectral width to the first null is The spectral width to the first null is
Δ fnull = 1/ T. Δ fnull = 2/ T.
3. The transmit signal spectrum for NRZ and RZ is shown in [37] and is centred
at the carrier frequency of the light source, which in this dissertation is set at
193 THz for 1550 nm wavelength. RZ has a wider spectral width and is less
tolerant to dispersion than NRZ because:
NRZ transmitter has a finite bandwidth. RZ transmitter has a larger bandwidth.
4. The mode of operation (MOA) for NRZ and RZ is shown in [37]. The
transmission of a long string of 1s and 0s produces a condition in which the
receiver may lose its amplitude reference for optimum discrimination and
26
transitioning between successive received signals of 1s and 0s. This means
that the effects of fiber losses are more reduced in RZ than NRZ because:
With NRZ, its MOA never demands for a With RZ, its MOA demands for a return
return to a zero-level state before a new to a zero-level state before a new set of
set of 1s and 0s are transmitted. Minimal 1s and 0s were transmitted, which
transitioning and discrimination between maximises the discrimination and eases
the successive transmission of a long the transitioning between each state.
string of either 1s or 0s.
Figure 2.4: Illustration of the differences between NRZ and RZ Signals [35]
2.6.2 Continuous Wave (CW) Laser Array with Equally Spaced (ES) channels
The CW Laser Array ES is an array of CW lasers and is equivalent to the conventional CW Laser
Array component. The emission frequencies are equally spaced (ES). However, the CW Laser
Array ES model is easier to set up for WDM systems, because it only requires the initial laser
emission frequency and spacing. The signal output power is the same for all the output signals.
This type of laser is used as the optical source when there is a need to use advanced
modulation techniques such as the modified duo binary return to zero (MDRZ) line coding
modulation format. This is because the properties of the WDM transmitter in Optisystem does
not support the usage of the MDRZ format. For this purpose, each channel of the CW laser
27
array has to be connected to its own MDRZ subsystem. The left part of Figure 2.5 shows the
48th channel of the CW laser array connected to its MDRZ subsystem. The internal design
components that make up the MDRZ subsystem is shown on the right.
Figure 2.5: 48 port CW Laser Array and Design components of MDRZ modulation subsystem
2.6.2.1 Advanced Line coding Modulation format: MDRZ

The selection of the most optimum signal format for a transmission system is of great
importance to maintain a low channel crosstalk penalty. This is because the modulation
format used, bandwidth of the specified format and the channel spacing are closely
interrelated. An advanced modulation formatting technique is used to enhance bandwidth
utilisation of the DWDM system while sustaining a high quality of service (QoS) [38]. The desire
to achieve an acceptable system performance for narrow channel spacings (< 25 GHz) for the
optical modulation bandwidths with transmission rates (≥ 25 Gb/s) leads to the choice of using
the modified duobinary return to zero (MDRZ), which is an advanced modulation format that
implements forward error correction (FEC) [38]. FEC code is a system of adding redundant
data (parity data) to an information signal, such that it can be recovered by a receiver even
when a number of errors (up to the capability of the code being used) are introduced, either
during the process of transmission, or on reception [39].
28
MDRZ effectively suppresses the discrete frequency tones of RZ signal spectrum and provides
the advantage of a smaller timing jitter and amplitude distortion. Moreover, when compared
to NRZ and RZ, MDRZ signals provides smooth operation even at relatively higher average
channel powers. MDRZ format has a much narrower optical bandwidth over the NRZ and RZ
formats, which leads to a greater chromatic dispersion tolerance and higher fiber non-linearity
tolerance. It thus should be able to achieve an acceptable signal transmission over longer
distances. In this format, the phases of two groups of “ones” that wrap an isolated “zero” are
flipped, leading to a reduced generation of ghost pulses caused by intra-channel four-wave
mixing (FWM) [38].
In order to produce the MDRZ modulation format, it requires the use of two optical Lithium
Niobate Mach-Zehnder modulators (Linb MZM); one generated NRZ duo-binary signal and the
other carved the NRZ data to a RZ by means of an electrical signal [38]. The MZM produces an
intensity modulating signal that uses a simple drive circuit for the modulating voltage [40].
The MZM structure consists of an input optical branch, which could split the incoming light
into two arms followed by an output optical branch [40].
The designed diagram of the MDRZ subsystem is shown in Figure 2.5. Each channel of the CW
laser array is connected to the input of the first Linb MZM within each MDRZ Subsystem. The
first step to create the MDRZ subsystem is to generate the NRZ duobinary signal, and this is
achieved by combining a pseudo random bit sequence generator to a duobinary precoder, the
NRZ generator and an electrical time delay-and subtractor circuit. The pseudo random bit
sequence generator used in this dissertation has a bit rate of 25 Gb/s and is connected to a
duobinary precoder with a delay of 1 bit. The precoder avoids recursive decoding in the
receiver, error propagation and reduces hardware complexity. The precoder is composed of
an exclusive-or gate with a 1-bit delayed feedback path. The output of the NRZ duobinary
signal block is split and fed into the two independent optical arms of the first Linb MZM, which
subsequently is recombined by the output optical branch to feed the input port of the second
Linb MZM. In the second Linb MZM, an electrical signal is split and fed to both optical arms to
control the degree of interference at the output optical branch and therefore controls the
output intensity. The second MZM is driven by a sinusoidal electrical signal with 40 GHz
frequency and phase of −90°. In the MDRZ, the phase between 0 and π for the “1” bits are
alternated and the phase of all the “0” bits are kept constant, which introduces a 180° phase
29
variation between all the consecutive “1s”. This leads to the suppression of the duobinary
carrier signal. The optical power and phase of the modulator output are determined in
response to the modulating voltage waveforms. The modulator’s transfer function relates the
effective drive voltage to the applied drive voltage [38]. The output of the 2nd Linb MZM is
then connected to the WDM MUX.
2.6.3 WDM Multiplexer (MUX) ES and WDM De-multiplexer (DEMUX) ES

The MUX takes transmitted optical wavelengths from multiple fibers of the WDM transmitter
and converges them into one beam that is sent over the optical link. At the receiving end of
the system, the DEMUX separates the received beam from the optical link into its wavelength
components and couples them to individual fibers, which are then sent to its own
photodetector. Demultiplexing must be done before the light is detected, because
photodetectors are inherently broadband devices that could not selectively detect a single
wavelength [3]. The WDM MUX ES and WDM DEMUX ES components used in this dissertation
respectively multiplexes and demultiplexes a user-defined number of WDM signal channels
[29]. The centre frequencies of the internal filters for both the MUX and DEMUX are equally
spaced (ES). The WDM MUX ES and WDM DEMUX ES are both equivalent to the conventional
WDM MUX and WDM DEMUX components respectively [29]. However, the WDM MUX ES and
WDM DEMUX ES are easier to set up for WDM systems, since it only requires the filter centre
frequency and the spacing. The most important parameters that could be set for the MUX and
DEMUX is the centre frequency of the first filter (1550nm), frequency spacing between
adjacent filters (either 100 GHz, 50 GHz, 25 GHz, 12.5 GHz, or 6.25 GHz), 3 dB filter bandwidth
(which in this case refers to a range of values depending on the frequency spacing used),
insertion loss of the DEMUX (5dB), and the internal optical filter type (which in this case is
either a Bessel or rectangular filter).
2.6.3.1 Bessel Filter

The Bessel filter used in the MUX, DEMUX and photodetector is a type of low pass linear filter
that provides a maximally flat phase response and almost constant propagation group delay
across the entire pass band of the frequency spectrum. This preserves the wave shape of the
transmitted filtered signal in the pass band. However, it offers a slower transition from pass-
band to stop-band than any other forms of filters of the same order [11]. All harmonics outside
the pass band are either eliminated or reduced in power level, which is dependent on the
30
order of the Bessel transfer function. Bessel filter has a Bessel frequency transfer function of
𝐻𝐻 (𝑁𝑁 ) , which is given by [30] and is shown in (2.5):
( 2𝑁𝑁 )!
� ( 2𝑛𝑛 . 𝑁𝑁! ) �
𝐻𝐻 (𝑁𝑁 ) = 𝛼𝛼 . � � (2.5)
𝐵𝐵𝑁𝑁
where 𝛼𝛼 is the insertion loss parameter, N is the order parameter of the filter, and 𝐵𝐵𝑁𝑁 is an
Nth-order Bessel polynomial expression is shown in (2.6) with the following form:
𝑘𝑘
( ) ( )
⎛ ⎛ ⎛ 𝑓𝑓 − 𝑓𝑓𝑐𝑐 . �� 2𝑁𝑁 − 1 . 𝑙𝑙𝑙𝑙2��⎞⎞ ⎞
𝑁𝑁
(2𝑁𝑁 − 𝑘𝑘 )!
𝐵𝐵𝑁𝑁 = � ⎜� ⎜ �.⎜ ⎟ (2.6)
⎜ �2 (𝑁𝑁−𝑘𝑘) . 𝑘𝑘!. �(𝑁𝑁 − 𝑘𝑘 )!�� ⎜𝑗𝑗. ⎜2. 𝐵𝐵 ⎟⎟
⎟ ⎟
⎟
𝑘𝑘=0
⎝ ⎝ ⎝ ⎠⎠
⎠
where 𝑓𝑓𝑐𝑐 is the filter centre cut-off frequency parameter defined by the -3dB point, 𝑓𝑓 is the
frequency of the signal being filtered and 𝐵𝐵 represents the bandwidth parameter [30].
2.6.3.2 Rectangular Filter

It is a low pass optical filter with a linear phase response; it is able to remove all frequencies
above a given bandwidth and leaves the low frequencies alone. This property enables the filter
to eliminate unwanted frequency harmonics above a given cut off frequency, without
affecting lower frequencies. The real time filters of the Optisystem platform can only
approximate an ideal rectangular filter since it has an infinite delay and is non-causal [29],
which means that the output at any time interval is dependent on past, present and
anticipated values of the input from the future [41]. The transfer function 𝐻𝐻 (𝑓𝑓 ) is given by
[30]:
𝐵𝐵 𝐵𝐵
𝐻𝐻 (𝑓𝑓 ) = � 𝛼𝛼, 𝑓𝑓𝑐𝑐 − < 𝑓𝑓 < 𝑓𝑓𝑐𝑐 + (2.7)
2 2 �
𝑑𝑑, 𝑜𝑜𝑜𝑜ℎ𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
where 𝛼𝛼 is the parameter Insertion loss, 𝑑𝑑 is the parameter Depth, 𝑓𝑓𝑐𝑐 is the filter center
frequency defined by the parameter Frequency, 𝐵𝐵 is the parameter Bandwidth, and 𝑓𝑓 is the
frequency.
31
2.6.4 Optical Link
The optical link consists of three components:
• The fiber optic transmission medium component – transports the composite beam of
light from the MUX to the DEMUX in a point-to-point link.
• The optical amplification component – amplifies the gain of the composite light signal.
• The dispersion compensation technique – reduces the effects of linear and non-linear
losses.
The main components and theoretical design method of the fiber optic link is discussed in the
following sections.
2.6.4.1 Fiber optic transmission medium

There are two type of fibers namely, multi-mode fiber (MMF) and single mode fiber (SMF).
MMF has a large core diameter either 50 microns or 62.5 microns and operates at wavelengths
of either 850nm or 1300nm [9]. The large core size of MMF makes them susceptible to modal
dispersion, has a relatively high attenuation and a low bandwidth, which limits its transmission
of light to only short distances. SMF has a very small core size of 9 microns and operates at
wavelengths of 1310nm and 1550nm [9]. The small core size of SMF limits the transmission of
light to only one mode of propagation, which eliminates modal dispersion and increases the
transmission capabilities for long haul communication [9]. The multiplexed optical signal in
the DWDM system is launched over SMF due to its higher performance with respect to
bandwidth and attenuation [9]. DWDM systems in this dissertation operates in the C-band
(1530nm to 1565nm) since it has the least attenuation and has the highest availability of
optical amplifiers operating at the centre wavelength of 1550nm. The International
Telecommunication Union (ITU) distinguishes SMF into three main categories, namely the
G.652 for conventional SMF, G.653 for dispersion shifted fiber (DSF) and the G.655 for non-
zero dispersion shifted fiber (NZDSF) [22]. However, within the C band, conventional SMF (ITU
G.652 compliant [44]) undergo high dispersion, which severely limits its transmission distance.
Both DSF and NZDSF minimises the effects of dispersion and operates at the 1550nm
wavelength. Even though DSF (ITU G.653 compliant [45]) had a zero-dispersion point at the
1550nm region, there are unwanted non linearities (such as FWM) in this fiber near the zero-
dispersion point for which there is no effective compensation. For this reason, the ITU G.655D
compliant [10], NZDSF is the preferred choice since it has a low positive chromatic dispersion
32
(CD) property (~4 ps/nm.km at 1550 nm) that is sensibly selected to be small enough to enable
high-speed transmission over long distances, and at the same time be large enough to
suppress FWM. In addition, to further suppress the unwanted effects of dispersion over the
link, a dispersion compensating fiber (DCF) is used. Dispersion compensating fiber (DCF) has a
negative dispersion property and a negative dispersion slope that compensates for the
positive dispersion and slope of installed fibers. The length of the DCF is designed in such a
way that it cancels out the dispersion gathered in the SMF [3]. A combination of the NZDSF
and a DCF is used as the preferred choice to compensate losses that arise due to CD and FWM
for the fiber optic transmission medium.
2.6.4.2 Optical Amplification

Optical amplifiers (OAs) magnify and regenerate light signals directly without any need for an
optical to electrical conversion. In doing so, OAs lessen the effects of dispersion and
attenuation for longer transmission distances, which makes it essential for any long-haul
optical network. There are currently three types of OAs available, namely semiconductor
optical amplifiers (SOAs), erbium doped fiber amplifiers (EDFAs), and raman optical amplifiers
(ROA). The focus of this dissertation is on the EDFA, since it is the best known, cost-effective
and most frequently used optical amplifier suited for DWDM systems. This is due to the
operational frequency of the erbium ion (used in EDFAs) being within the C band, which is a
low loss optical window of silica based fiber [49]. EDFA uses an internal fiber (doped with
erbium ions) as a gain medium to boost an optical signal. The input signal is multiplexed with
a pump laser from the EDFA before it is sent into the doped fibers. The interaction of the
multiplexed signal with the doped erbium ions in the fiber amplifies the signal. Subsequently,
this process releases natural emissions within the EDFA that adds noise to the signal, which
determines the noise figure [3]. The EDFA used in this dissertation has a gain of 21 dB and a
noise figure of 6 dB.
There are two types of compensation techniques used in this dissertation, namely the post
compensation and the post symmetric compensation. They both make use of EDFAs to
compensate the losses of the fiber. For the post compensation technique, a single EDFA is
used as a pre-amplifier to boost the signal pulses at the receiver’s side, as can be seen in Figure
3.3. For the post symmetric compensation technique shown in Figure 3.4, three EDFAs are
placed strategically in the optical link. The first is placed before the NZDSF, this acts as a post-
33
amplifier, which boosts the signal at the transmitter’s side. The second is placed in between
the NZDSF and DCF, which acts as an in-line amplifier that recovers the signal before it is
degraded significantly. And finally, the third amplifier is placed after DCF, which acts as pre-
amplifier that performs the same function as the amplifier used in the post compensation
technique [11].
2.6.5 Optical Receiver

The optical receiver component used in this dissertation consists of a built-in selector to
choose between a P-N junction type (PIN) photodetector and an avalanche photodetector
(APD), one Bessel filter and a 3R regenerator. This can be seen in Figure 2.6.
Photodetectors receive demultiplexed optical signals and interpret the information signals
found within the received signals [5]. They can do this function because they are
optoelectronic devices that convert the received optical signal into an electrical signal for the
purpose of decoding the optical signal to the actual information signal [5]. For such
applications, photo-detectors have high sensitivity, high responsivity and minimum noise [5].
There are two types of photodetectors that can be used in Optisystem namely, PIN type and
the APD photodetector [29]. The fundamental concept of a PIN photodiode is built on the P-
N junction which has the (p) type material connected to a lightly doped (n) type material with
a wide depletion region; this allows longer wavelengths of light to pass through. This depletion
region is called the intrinsic region (i) and subsequently it makes a low resistance contact with
less noise on a signal passing through. It has a heavily doped (n+) layer near the intrinsic
region; and thus, forms the P-I-N assembly [20]. However, an avalanche photodetector (APD)
requires a higher operating voltage and this is due to its output being non-linear. This in turn
causes APDs to produce a higher level of noise than a PIN diode [21].
The 3R regenerator component is used to recover the binary data from the electrical signal,
regenerate the original bit sequence and modulate the electrical signal to be used for bit error
rate (BER) analysis. It performs the 3R functions such as re-shaping, re-timing and re-
amplification of the electrical signal [29].
The component properties of the optical receiver used in this dissertation allows the user to
select the internal component parameters such as the -3dB cut-off frequency of the filter (set
as 0.75 x bit rate), order of the low pass Bessel filter function (set as = 4), insertion loss of the
34
filter (set as 3dB), user-defined centre frequency of the internal filter (set as 1550nm),
reference bit rate to use for the decision instant calculation (which in this dissertation is 25
Gb/s) and the receiver sensitivity parameter ( set as -18 dBm) [29]. The received signal of the
optical receiver is analysed using eye diagram analysers.
Figure 2.6: Internal setup of Optical Receiver
2.7 Theoretical Concepts for the Design of the Optical Transmission Link
The theoretical design method of the optical transmission link involves obtaining the length
of dispersion compensation fiber (DCF), the initial span loss before amplification, the
minimum launch power of transmitter, power budget, power margin, the total span loss after
amplification, and the received signal power. The theoretical formulas for each of the above
parameters is summarised below with relevance to a 100 km transmission distance. The same
parameters for the optical link used in [1] and is used in the design of the optical link used for
the 48 channel DWDM networks.
When designing DWDM networks for long-haul communications, a DCF needs to be used. A
DCF with a negative dispersion component must be added to a DWDM system for the purpose
of compensating the positive dispersion component of the non-zero dispersion shifted fiber
(NZDSF), which subsequently decreases the chromatic dispersion (CD) of the system. The
dispersion coefficient of the DCF and NZDSF is assumed as -147 ps/nm.km and 4 ps/nm.km,
respectively [1].
The length of the DCF is calculated using the expression in (2.8):
35
(𝐷𝐷𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 )(𝐿𝐿𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 ) = −(𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 )(𝐿𝐿𝐷𝐷𝐷𝐷𝐷𝐷 ) (2.8)
where 𝐿𝐿𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 , 𝐷𝐷𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝐿𝐿𝐷𝐷𝐷𝐷𝐷𝐷 and 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 are the respective lengths and dispersion coefficients
of corresponding fibers.
The initial span loss (𝑃𝑃𝑆𝑆𝑆𝑆 ) for both NRZ and RZ before amplification is calculated by adding the
various linear and nonlinear losses, which is shown by the following expression in [1] :
𝑃𝑃𝑆𝑆𝑆𝑆 = 𝛼𝛼𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 + 𝛼𝛼𝐷𝐷𝐷𝐷𝐷𝐷 + 𝛼𝛼𝐶𝐶 (2.9)
where 𝛼𝛼𝐷𝐷𝐷𝐷𝐷𝐷 , 𝛼𝛼𝐷𝐷𝐷𝐷𝐷𝐷 and 𝛼𝛼𝐶𝐶 are the losses of the NZDSF, DCF, connectors, and safe margin (SM),
respectively. To compensate for unpredictable losses due to non-linear effects and
temperature fluctuations when employing the DCF, a safe margin of 6 dB is used.
If PT is the power of the WDM transmitter and in Optisystem, the NRZ WDM transmitter has
an internal optical attenuation (𝛼𝛼 𝑇𝑇 𝑁𝑁𝑁𝑁𝑁𝑁 ) and the RZ WDM system has an optical attenuation
(𝛼𝛼 𝑇𝑇 𝑅𝑅𝑅𝑅 ); then the minimum launch power (PTMIN) that is transmitted into the WDM MUX is the
difference between transmitter power (PT) and its internal optical attenuation (either 𝛼𝛼 𝑇𝑇 𝑁𝑁𝑁𝑁𝑁𝑁
or 𝛼𝛼 𝑇𝑇 𝑅𝑅𝑅𝑅 ).
To ensure that the 100km fiber span has sufficient power for correct operation, the span's
power budget (PB) is calculated. This is the maximum amount of power that can be transmitted
over a span. From a design perspective, the power budget (PB) is the difference between the
minimum transmitter launch power (PTMIN) and the minimum receiver sensitivity (PRMIN) [1].
The receiver’s sensitivity level (PRMIN) indicates the minimum signal power required by the
photodetector to detect the signal with the least errors. Based on the parameters of the
photodetector, the receiver sensitivity is -18 dBm, therefore, the signal power received at the
detector for the NRZ and RZ should be above the receiver sensitivity to prevent receiver
saturation and ensure system viability [1].
The next calculation involves the power margin (PM), which represents the amount of power
available to operate the receiver or it can be referred to as the maximum allowable power
loss over the span. PM is the difference between power budget (PB) and the total span losses
(𝑃𝑃𝑃𝑃𝑃𝑃). The PM has to be greater than zero for the power budget to be adequate to operate the
receiver without further amplification [1].
36
Once amplifiers are added to the span, the total span loss after amplification (𝑃𝑃𝑆𝑆 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 ) is
calculated by obtaining the difference between the total gain after amplification (𝑁𝑁. 𝐺𝐺𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 )
and the initial span loss (𝑃𝑃𝑆𝑆𝑆𝑆 ). The gain of the amplifier is 𝐺𝐺𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 and the number of amplifiers
used in the network is 𝑁𝑁𝑁𝑁.
To calculate the received signal power (𝑃𝑃𝑅𝑅 ) at the photodetector after amplification, the
following expression in [1] is used:
𝑃𝑃𝑅𝑅 (𝑑𝑑𝑑𝑑𝑑𝑑) = 𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 − 𝑃𝑃𝑃𝑃𝑃𝑃 + (𝑁𝑁𝑁𝑁 . 𝐺𝐺𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 ) − 𝛼𝛼𝑀𝑀𝑀𝑀𝑀𝑀 − 𝛼𝛼𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 (2.10)
where 𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 , 𝑃𝑃𝑃𝑃𝑃𝑃 , 𝛼𝛼𝑀𝑀𝑀𝑀𝑀𝑀 , 𝛼𝛼𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 , 𝑁𝑁𝑁𝑁 and 𝐺𝐺𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 are minimum transmitter launch power,
total initial span loss, losses of multiplexer, demultiplexer, number of EDFAs used and gain of
EDFA, respectively. This equation assumes that all amplifiers has the same gain. The set
parameters values listed in Table 2 of [1] are used to find the power received at the
photodetector for the 100km transmission link.
2.8 Measuring system performance

The performance of the signal is evaluated using three measurements, namely the optical
signal to noise ratio (OSNR), Quality (Q) factor and Bit error rate (BER).
2.8.1 Optical Signal to Noise Ratio (OSNR)

𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 measures the signal quality of an optical transmission system, which is the digital
equivalent to the analogue signal to noise ratio (SNR) of an electrical transmission system. The
addition of components such as an amplifier in a long-haul transmission system causes
performance degradation in the OSNR. The final OSNR at the receiver is used from [1] and is
shown below:
𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 (𝑑𝑑𝑑𝑑) = 𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑑𝑑𝑑𝑑 − 𝑁𝑁𝑁𝑁 𝑑𝑑𝑑𝑑 − 𝑃𝑃𝑆𝑆𝑆𝑆 − 10 log[ℎ. 𝑣𝑣] − 10 log[𝛻𝛻𝛻𝛻 ] − 10 log[𝑁𝑁𝑁𝑁] (2.11)
where 𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 is transmitter launch power, 𝑁𝑁𝑁𝑁 is noise figure, 𝑃𝑃𝑆𝑆𝑆𝑆 is total loss for single span
(100km), ℎ (= 6.6260 × 10−34 𝐽𝐽/𝑠𝑠) is planks constant, 𝑣𝑣 (= 193.4 𝑇𝑇ℎ𝑧𝑧) is optical signal
frequency, 𝛻𝛻𝛻𝛻 (= 12.5 𝐺𝐺ℎ𝑧𝑧) is optical bandwidth of 0.1nm measuring 𝑁𝑁𝑁𝑁 and 𝑁𝑁𝑁𝑁 is number
of amplifier stages.
37
OSNR can also be calculated as the difference between the Signal power (𝑃𝑃𝑃𝑃) and the noise
power (𝑃𝑃𝑃𝑃) received at the optical receiver as shown in (2.12), which is essentially found at
the output of the demultiplexer.
𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂 (𝑑𝑑𝑑𝑑 ) = 10 log�𝑃𝑃𝑃𝑃�𝑃𝑃𝑃𝑃� = 𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃(𝑑𝑑𝑑𝑑 ) − 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃𝑃(𝑑𝑑𝑑𝑑) (2.12)
2.8.2 Quality (Q) factor

The Quality of the signal (QoS) in the eye pattern of a digital signal is represented by the
Quality (Q) factor. Q-factor is the ratio of the signal difference between the mean value of the
signals for 1 level state and 0 level state to the sum of the noise value at those signal levels.
The Q factor expression shown below is taken from [2]:
𝜇𝜇1 − 𝜇𝜇0
𝑄𝑄 = (2.13)
|𝜎𝜎1 + 𝜎𝜎0|
where 𝜇𝜇0, 𝜇𝜇1, 𝜎𝜎0 and 𝜎𝜎1 denotes respectively, the average signal values for the 1 and 0 levels,
and the standard deviations of the sampled noise values for signal at 1 and 0 levels in the eye
diagram [10]. The wider the eye opening (eye height), the greater is the difference between
signal to noise values of at the 1 level and 0 level states, which results in better BER
performance.
2.8.3 Bit Error Rate (BER)

Bit error rate (BER) is the percentage that shows the number of error bits received in relation
to the total number of bits received. Normally, the total number of bits transmitted is larger
than the BER of the information signal. The BER may be improved by optimising the signal
strength. This can be done by applying compensation techniques or line coding modulation
that is able to enhance the signal, or by choosing filter and other active parameters that is able
to minimise bit errors [39].
Based on the Gaussian approximation for the noise distribution in the received signal, the
relationship between Q-factor and 𝐵𝐵𝐵𝐵𝐵𝐵 for this model is given in [10] and [2]:
1 𝑄𝑄 1 𝑄𝑄 2
−
𝐵𝐵𝐵𝐵𝐵𝐵 = 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 � � ≈ 𝑒𝑒 2 (2.14)
2 √2 𝑄𝑄√2𝜋𝜋
38
where 𝐵𝐵𝐵𝐵𝐵𝐵 is the bit error rate, 𝑄𝑄 is the linear quality factor and ′ 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 ′ is the complementary
error function, which is taken from [2] and given by the expression:
∞ 2
2 −𝑡𝑡 2
𝑒𝑒 −𝑥𝑥
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 (𝑥𝑥 ) = � 𝑒𝑒 𝑑𝑑𝑑𝑑 ≈ (2.15)
√𝜋𝜋 𝑥𝑥 𝑥𝑥 √𝜋𝜋
𝑄𝑄
The variable 𝑥𝑥 of (2.15) will be replaced by when substituting (2.15) in (2.14).
√2
The Forward Error Communication (FEC) threshold sets the minimum acceptable BER value
for an optical link to be > 10−9 . This provides a minimum Q factor of 6 for an error free optical
communication. This can be obtained in terms of Q-factor using the following expressions in
[38]:
1
𝑄𝑄 (𝑑𝑑𝑑𝑑 ) = 20. log(22. 𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 −1(2. 𝐵𝐵𝐵𝐵𝐵𝐵)) (2.16)
And if Q in linear format is known then Q (dB) can be calculated using (2.17) :
𝑄𝑄 (𝑑𝑑𝑑𝑑 ) = 20. log (𝑄𝑄𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙 ) (2.17)
So, throughout this paper Q-factor of 6 and BER of 10−9 are considered as the reference
criteria and the performances of various parameters are evaluated using these two criteria.
2.8.4 Overall Signal Performance and Bandwidth Efficiency

The total average Q factor (𝐴𝐴𝐴𝐴𝐴𝐴. 𝑄𝑄) of a DWDM System for a total number of 𝑁𝑁 channels is
equal to a ratio between the sum of the Q factors for the number of recognisable channels
(∑𝑁𝑁
𝑘𝑘=1 (𝑄𝑄𝑘𝑘 )) in the system (where Q > 0 and BER < 1) and the total number of transmitted
channels (𝑁𝑁) in the system. This gives an indication of the overall signal performance of that
system, which is shown in (2.18):
∑𝑁𝑁
𝑘𝑘=1 (𝑄𝑄𝑘𝑘 )
𝐴𝐴𝐴𝐴𝐴𝐴. 𝑄𝑄(𝑁𝑁)𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝑆𝑆𝑆𝑆𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 = (2.18)
𝑁𝑁
The Bandwidth or spectral efficiency shows a percentage of the total Bandwidth that is utilised
in the designed spectral frequency band, which has recognisable signals at the receiver (where
Q > 0 and BER < 1). Each frequency in the spectral band is sent over its own channel. Therefore,
39
Bandwidth efficiency is a percentage of the ratio between the total number of channels that
can produce a recognisable signal at the optical receiver and the total number of transmitted
channels (𝑁𝑁) in that DWDM system. This is shown in (2.19):
𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇𝑇 𝑁𝑁𝑁𝑁. 𝑜𝑜𝑜𝑜 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟 𝑐𝑐ℎ𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎𝑎 𝑎𝑎𝑎𝑎 𝑡𝑡ℎ𝑒𝑒 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟

𝐵𝐵𝐵𝐵 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 (%) = 𝑋𝑋 100 (2.19)
𝑁𝑁
2.9 Summary
In this chapter, the literature reviews substantiated DWDM as the best fiber optic technology
to meet the current bandwidth requirements. The performance limitations of DWDM was
explored and it was established that CD and FWM were the most critical limiting factors that
deteriorates a DWDM signal. The various design components of the DWDM system, their
properties, functionalities, limitations and benefits were discussed with the focus being on the
optimisation of a 48 channel DWDM system. The theoretical design method of designing and
measuring the performance of the optical transmission link was also enumerated.
40
DESIGN OF A 48 X 25Gb/s DWDM SYSTEM
3.1 Introduction
In this chapter, the design methodology and parameters of the 48-channel DWDM system are
discussed using references to the system components and theoretical concepts explained in
Chapter 2. The design is performed using the Optisystem v7.0 simulation platform. The set
and varied parameters used to simulate, model and optimise the existing 48 channel DWDM
network is enumerated.
The design process of the DWDM network systems in this dissertation mainly consists of three
main design blocks, namely the transmitter block, the optical transmission link and the
receiver block. The set parameters of all components in each block are calculated using the
theoretical concepts of the literature review that was discussed in Chapter 2 and are exhibited
in this chapter.
The varied parameters that are chosen to create a comparative model between the
compensation techniques and modulation types used in the design is also enumerated in this
chapter. These are used to optimise the signal performance of the 48 channel DWDM network
for various narrow channel spacings. The block diagrams of the designed network models,
compensation techniques and modulation types are also illustrated in this chapter.
3.2 Design Methodology

The numerous 48 channel DWDM networks that are designed in this dissertation have a bit
rate of 25 Gb/s for a transmission distance of 100 km and centre wavelength of 1550 nm. The
performance analysis of these networks is done by varying the channel spacing from 100 GHZ,
50 GHz, 25 GHz, 12.5 GHz and 6.25 GHz. The 1550 nm wavelength is a low attenuation window
in the C band and this is chosen according to the ITU standard DWDM wavelength grid since
it has a huge bandwidth and the operational wavelength of many optical components are in
this band [26].
Furthermore, various design parameters such as the modulation type, extinction ratio of the
transmitter, filter type and bandwidth of the multiplexer, filter type and bandwidth of the
41
demultiplexer and the dispersion compensation techniques of the optical link are varied and
compared to discover the system with the most optimal signal performance. The system
performance is based on the results of Q factor, minimum BER and eye height of the received
signal, which is represented by eye diagrams. The Optisystem v7.0 is used as the design
simulation platform to analyse the 48-channel DWDM networks as explained in section 2.5 of
chapter 2. Eye diagram analysers are connected to each optical receiver to evaluate the
system performance. The impact of the varied design parameters on the effects of non-
linearities on the system performance of the 48 channel DWDM network at various channel
spacings is also observed and evaluated. The study then determines the best design
parameters for the most optimised DWDM network with the highest bandwidth efficiency
across all 48 channels for each channel spacing.
Figure 3.1 shows the block diagram of the 48 channel DWDM network. This shows the
different components that make up the network, the design parameters used and the two
types of dispersion compensation techniques used for the optical link. Figure 3.2 shows an
example of the 48 Channel DWDM Network Design using the NRZ modulation scheme for a
100 GHz channel spacing. The same design setup is used for the RZ modulation scheme and
for all other channel spacings of 50 GHz, 25 GHz, 12.5 GHz and 6.25 GHz.
The design of the two types of dispersion compensation techniques of the optical link, namely
the Post Compensated Optical Link Design with 1 EDFA and the Post Symmetric Optical Link
Design with 3 EDFAs were shown in Figure 3.3 and Figure 3.4 respectively. The design of both
dispersion compensation techniques uses the post compensation design technique, that has
the NZDSF placed before the DCF, as the foundation. The post compensated optical link with
one EDFA (placed after the DCF) is shown in Figure 3.3. The post symmetric compensated
optical link with three EDFAs shown in Figure 3.4 is designed to have a total of three
symmetrically distributed EDFA optical amplifiers placed before and after the NZDSF and the
DCF. The post symmetric compensation technique is used to further compensate for the signal
degradation that occurs due to non-linear effects when decreasing channel spacings.
In order to improve the system performance such that the least amount of BER is achieved for
each of the channel spacings for the NRZ and RZ modulation schemes, the MUX and DEMUX
bandwidths are varied. The methodology used to vary the bandwidth of the MUX and DEMUX
42
is by comparing the system performance of the initially used bandwidth of MUX and DEMUX
of 80 GHz with a varied bandwidth parameter that is half of the used channel spacing until an
acceptable system performance is achieved. If signal quality is not improved by varying the
MUX and DEMUX bandwidths, then the extinction ratio is decreased from 30 dB to 5 dB and
the Bessel filter in the MUX and DEMUX is replaced with a rectangular filter.
If the signal quality could not be improved with the NRZ and RZ modulation schemes, further
improvement of the signal quality is done using an advanced MDRZ modulation scheme.
Figure 3.5 shows the design of a 48 channel DWDM network design using the advanced MDRZ
modulation scheme. In this network, a 48 output CW laser array is connected to 48 MDRZ
subsystems. The MDRZ subsystem allows the use of the MDRZ modulation method. Figure 2.5
shows the internal design setup of the MDRZ subsystem. Each of these MDRZ subsystems are
connected to a 48 input WDM MUX with equally spaced channels. See Figure 3.5. The single
output port of the MUX is then connected to a post symmetric optical link with 3 EDFAs. The
single output of the optical link is connected to the single input of a WDM DEMUX that has 48
output ports. A 2nd order Bessel filter is used for both the MUX and DEMUX.
3.3 Design Parameters

The design parameters are divided into two groups, namely set parameters and varied
parameter. Set design parameters are the common constant parameters used for all
configurations of the 48 channel DWDM network. The varied design parameters are the
various parameter configurations that are changed to optimise signal performance and
bandwidth efficiency for all 48 channels of the designed DWDM networks.
3.3.1 Set Design Parameters

The main steps of designing this network, consists of a transmitter block, optical transmission
link, and receiver block, respectively.
3.3.1.1 Design of the Transmitter block

The transmitter block consists of the WDM transmitter, which has the modulation type used
by the transmitter and the WDM MUX.
For the design of NRZ and RZ DWDM networks, a 48-output port WDM transmitter with a bit
rate of 25 Gb/s and a transmitting power of 5 dBm is chosen. The transmitter type chosen for
43
the Optisystem platform is an externally Modulated Laser (EML). C-band wavelengths with a
centre frequency of 1550nm are assigned to the lasers to modulate the data. The WDM
transmitter has a default extinction ratio of 30 dB and an internal modulation selector that
allows the option to switch between NRZ and RZ modulation schemes. The WDM transmitter
of NRZ and RZ are connected to a 48 input WDM MUX with equally spaced channels.
For the WDM transmitter, the following set values are chosen:
• Power (PT = 5 dB), internal attenuation for NRZ (𝛼𝛼 𝑇𝑇 𝑁𝑁𝑁𝑁𝑁𝑁 = 4.27 dB) and internal
attenuation for RZ (𝛼𝛼 𝑇𝑇 𝑅𝑅𝑅𝑅 = 9.49 dB).
• Minimum launch power, (PTMIN) for NRZ = 0.73 dB and (PTMIN) for RZ = - 4.49 dB
For the MDRZ DWDM networks, a 48 output CW laser array is connected to 48 MDRZ
subsystems. The MDRZ subsystem allows the use of the MDRZ modulation method. Each
channel of the CW laser array is connected to its own MDRZ Subsystem. The internal design
setup and theoretical explanation of the MDRZ system is shown in section 2.5.2.1 of chapter
2. The output of the MDRZ subsystem produces a bit rate of 25 Gb/s and a modulated MDRZ
signal with a total insertion loss of 10 dB (5dB insertion loss for each MZM). Refer to Figure
2.5 to see the internal design setup of the MDRZ subsystem. Each of the outputs from the
MDRZ subsystem is connected to its own input port on the 48 port WDM MUX.
The 48-channel multiplexer has an insertion loss of 5 dB and has an internal selector to change
the filter type of the MUX. As a default, it uses a 2nd order Bessel filter but it has the option to
change to a rectangular filter. The multiplexer combines the 48 channels into a single output
port which is then connected to an optical link.
Each channel is launched into the optical fiber with a relatively low power (PTMIN) to minimise
the nonlinear effects as discussed in section 2.6 of chapter 2.
3.3.1.2 Design of the Optical Transmission Link

The communications link is designed using a non-zero dispersion shifted fiber (NZDSF) as the
transmission medium due to its low attenuation, low dispersion coefficient and subsequently
its ability to minimise FWM at the operating wavelength of 1550 nm.
44
The chromatic dispersion coefficient of the NZDSF with 1550nm wavelength has to be in the
𝑝𝑝𝑝𝑝 𝑝𝑝𝑝𝑝
range between 𝐷𝐷𝑚𝑚𝑚𝑚𝑚𝑚 (𝜆𝜆 ) (= 2.8 /𝑘𝑘𝑘𝑘) and 𝐷𝐷𝑚𝑚𝑚𝑚𝑚𝑚 (𝜆𝜆 ) (= 6.2 𝑛𝑛𝑛𝑛 /𝑘𝑘𝑘𝑘) which is calculated
𝑛𝑛𝑛𝑛
using (2.1) from section 2.3.1 of chapter 2. Therefore, a mid-value of 4 ps/nm/km is chosen
for the dispersion coefficient of NZDSF. Given that the dispersion coefficient of -147 ps/nm/km
is chosen for the DCF, the length of the DCF is calculated using (2.8) from section 2.6 of chapter
2.
The optical link has the following common design components, which can also be seen in
Figure 3.1:
• NZDSF length of 100km, attenuation of 0.19 dB/km, dispersion coefficient of 4

ps/nm/km, a PMD coefficient of 0.04 ps/sqrt(km) and an effective area of 72 um2.
• DCF length of 2.7 km, attenuation of 0.24 dB/km, dispersion coefficient of -147
ps/nm/km, a PMD coefficient of 0.04 ps/ sqrt(km) and an effective area of 27 um2.
• EDFA with gain of 21 dB and a noise figure of 6 dB
• Connectors with insertion loss of 0.75 dB
• Post compensation technique uses 1 EDFA and 6 connectors
• Post symmetric compensation technique uses 3 EDFAs and 7 connectors
• Initial span loss for both NRZ and RZ before amplification is calculated using (2.9):
(𝑃𝑃𝑆𝑆𝑆𝑆 = 24.15 𝑑𝑑𝑑𝑑).
• Span loss after amplification for NRZ and RZ is calculated using the following expression
in (3.1):
𝑃𝑃𝑃𝑃𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 = (𝑃𝑃𝑃𝑃𝑃𝑃 + 𝑛𝑛 . 𝛼𝛼𝐶𝐶 ) − (𝑁𝑁𝑁𝑁 . 𝐺𝐺𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 ) (3.1)
where 𝑃𝑃𝑠𝑠𝑠𝑠 , 𝑛𝑛, 𝛼𝛼𝐶𝐶 , 𝑁𝑁𝑁𝑁 and 𝐺𝐺𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 represents the total initial span loss before
amplification, number of extra connectors added after amplification, loss of
connectors, number of EDFAs used and gain of EDFA, respectively. For 1 EDFA no extra
connectors were added after amplification. However, when 3 EDFAs were added,
there is one extra connector added. Therefore, 𝑃𝑃𝑃𝑃 𝑓𝑓𝑓𝑓𝑓𝑓 1 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 = 3.15 𝑑𝑑𝑑𝑑 and
𝑃𝑃𝑃𝑃 𝑓𝑓𝑓𝑓𝑓𝑓 3 𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 = −38.1 𝑑𝑑𝑑𝑑. Thus, the span loss for 1 EDFA is much greater than the
span loss for 3 EDFAs.
45
3.3.1.3 Design of the Receiver Block
The receiver is composed of two parts: a DEMUX with 48-output channels that has an insertion
loss of 5 dB, and 48 photodiodes. For simplicity, 25 optical receivers are used evenly to
represent the 48 channel outputs i.e. optical receivers were placed in Channel 1, 2, 4, 6..., and
so forth till channel 48 as can be seen in Figure 3.1 and Figure 3.2.
One photodiode is connected for each output channel of the DEMUX. Each Photodiode is
connected to its own eye-diagram analysers. The PIN photodiode is chosen as the
photodetector, because of its lower cost and minimal sensitivity to the noise level in
comparison to avalanche photodiodes (APDs). These types of photodiodes need to be reverse
biased to operate; they provide better performances for wide bandwidth and high dynamic
range applications. For each channel of the DEMUX, a PIN photodiode with an insertion loss
of 3 dB and a dark current of 10 nA is used to change the received signals into their electrical
equivalents. An eye-diagram analyser has been used to analyse the detected signals of each
channel. The receiver also operates at a centre frequency of 1550nm and has a power
sensitivity of -18 dBm.
The received signal power at the photodetector is calculated using (2.10):

• For 1 EDFA: (𝑃𝑃𝑅𝑅 𝑓𝑓𝑓𝑓𝑓𝑓 𝑁𝑁𝑁𝑁𝑁𝑁 = −12.42 𝑑𝑑𝑑𝑑𝑑𝑑), (𝑃𝑃𝑅𝑅 𝑓𝑓𝑓𝑓𝑓𝑓 𝑅𝑅𝑅𝑅 = −17.64 𝑑𝑑𝑑𝑑𝑑𝑑)
• For 3 EDFA: (𝑃𝑃𝑅𝑅 𝑓𝑓𝑓𝑓𝑓𝑓 𝑁𝑁𝑁𝑁𝑁𝑁 = 28.83 𝑑𝑑𝑑𝑑𝑑𝑑), (𝑃𝑃𝑅𝑅 𝑓𝑓𝑓𝑓𝑓𝑓 𝑅𝑅𝑅𝑅 = 23.61 𝑑𝑑𝑑𝑑𝑑𝑑)
Therefore, it is evident that for both 1 EDFA and 3 EDFA networks, the signal could be detected
using both NRZ and RZ modulation formats since all power received is greater than the
receiver’s power sensitivity. But it could be noticed that the power received for 3 EDFAs is far
greater than for 1 EDFA.
3.3.2 Varied design parameters

This section shows all the various combination of parameters used to optimise the 48 channel
DWDM network at channel spacing of 100GHz, 50 GHz, 25 GHz, 12.5 GHz, and 6.25 GHz.
3.3.2.1 100 GHz channel spacing

• Modulation Type of either NRZ or RZ for the following MUX/DEMUX Bandwidth
Configuration with ER = 30 dB using Bessel Filter:
46
a) MUX Bandwidth of 80 GHz and DEMUX Bandwidth of 80 GHz with 1 EDFA and
3 EDFAs
b) MUX Bandwidth of 80 GHz and DEMUX Bandwidth of 50 GHz with 3 EDFAs
c) MUX Bandwidth of 50 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs

• Modulation Type of NRZ and RZ for the following MUX/DEMUX Bandwidth
3 EDFA
d) MUX Bandwidth of 50 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs

3 EDFAs
d) MUX Bandwidth of 12.5 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs
e) MUX Bandwidth of 25 GHz and DEMUX Bandwidth of 25 GHz with 3 EDFAs
• Modulation Type of MDRZ for the following MUX/DEMUX Bandwidth Configuration,
with ER = 5dB using Rectangular Filter:
f) MUX Bandwidth of 80 GHz and DEMUX Bandwidth of 25 GHz with 3 EDFAs
g) MUX Bandwidth of 25 GHz and DEMUX Bandwidth of 25 GHz with 3 EDFAs
3.3.2.4 12.5 GHz channel spacing

• WDM Transmitter Modulation Type of NRZ and RZ for the following MUX/DEMUX
Bandwidth Configuration with ER = 30 dB using Bessel Filter:
47
3 EDFAs
b) MUX Bandwidth of 80 GHz and DEMUX Bandwidth of 12.5 GHz with 3 EDFAs
c) MUX Bandwidth of 80 GHz and DEMUX Bandwidth of 6.25 GHz with 3 EDFAs
e) MUX Bandwidth of 6.25 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs
f) MUX Bandwidth of 3.125 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs
• Modulation Type of MDRZ for the following MUX/DEMUX Bandwidth Configuration
with ER = 30dB using Bessel Filter:
g) MUX Bandwidth of 3.125 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs
3.3.2.5 6.25 GHz channel spacing
Configuration with ER = 30dB using Bessel Filter:
3 EDFAs
b) MUX Bandwidth of 80 GHz and DEMUX Bandwidth of 6.25 GHz with 3 EDFAs
c) MUX Bandwidth of 6.25 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs
• Modulation Type of MDRZ for the following MUX/DEMUX Bandwidth Configuration
with ER = 30dB using Bessel Filter:
e) MUX Bandwidth of 3.125 GHz and DEMUX Bandwidth of 80 GHz with 3 EDFAs
48
Figure 3.1: Block Diagram Design of 48 Channel DWDM system with bit rate of 25 Gb/s and various other design parameters
49
Figure 3.2: 48 Channel DWDM Network Design using either NRZ or RZ Modulation
50
Figure 3.3: Post Compensated Optical Link Design with 1 EDFA
Figure 3.4: Post Symmetric Optical Link Design with 3 EDFAs
51
‘
Figure 3.5: 48 Channel DWDM Network Design using MDRZ Modulation

52
3.4 Summary
In this chapter, the simulation and design process of the 48-channel DWDM system design
was shown using the Optisystem simulation platform. The system basically consisted of 48
closely spaced DWDM channels. Each channel was able to transmit at a 25 Gb/s data rate for
a 100 km distance at various narrow channel spacings of 100 GHz, 50 GHz, 25 GHz, 12.5 GHz
and 6,25 GHz.
The WDM transmitter and WDM MUX used for optimisation were completely matched in
frequency. This matched network was used to produce equally spaced channels as opposed
to the unmatched active components (WDM transmitter and WDM MUX) network used in [2].
This was done to reduce the non-linear effects of FWM
Various design parameters were used to model and optimise the existing 48 channel fiber
optic network such that an acceptable system performance could be achieved for future
growth in end-user requirements. The optimisation of the 48-channel network was done using
various parameters in different combinations. Three of the line encoding techniques (NRZ, RZ
and MDRZ) were used in numerous DWDM networks with different filter types in the MUX
and DEMUX. The two types of filter types were the Bessel and rectangular filter. The BW of
the MUX/DEMUX filter and the ER of the transmitter were also varied to obtain an optimised
network that had the most acceptable signal performance with the least amount of non-
linearities for each of the above-mentioned channel spacings. A comparative model of two
dispersion compensation techniques was also done for each channel spacing to evaluate a
reduction in linear effects (CD) and non-linear effects (FWM). The two compensation
techniques used in the design process were the 1 EDFA post compensation technique and the
3 EDFA post symmetric compensation techniques.
53
SIMULATION RESULTS FOR THE 48 CHANNEL DWDM NETWORKS
4.1 Introduction
This chapter shows the simulation results of the 48 channel DWDM system designed in
Chapter 3. The Optisystem v7.0 platform is used to simulate the 48-channel DWDM systems.
For simplification, the results of 25 optical receivers are analysed evenly to represent the 48
channel outputs i.e. the analysis of the received signals from optical receivers placed in
Channel 1, 2, 4, 6..., and so forth till channel 48 was done.
Eye diagram results of each channel is simulated by varying parameters such as channel
spacings, modulation formats, filter types and BW parameters of the filters in the
MUX/DEMUX for each network combination. The eye diagram results represent the BER eye
height and Q factor measurements of the received signal for each channel in the DWDM
system. For simplification and analysis of each network, the BER and Q factor measurements
are tabulated and compiled in an excel spread sheet. Subsequently, this tabulated data is then
graphically represented in the form of a Q factor vs BER comparative charts. These charts are
an equivalent representation of the numerous tables of data that is recorded for each varied
parameter combination of the 48 channel DWDM network shown in section 3.3.2 of chapter
3. The comparative chart shows the inversely proportional relationship between Q factor and
BER for each combination of the varied parameters. The BER vs Q factor charts are used to
verify the effect that each modified parameter (used in this design) has on the overall signal
performance and bandwidth efficiency of the entire system. The blue bar pillars on the chart
represent the quality factor on a linear scale for each channel within that network and the
orange line showed the BER on a logarithmic scale for that corresponding channel in relation
to the other channels within that network.
The eye diagrams results shown in this chapter illustrates each of the channels that has the
highest signal performance (highest Q factor, lowest BER and its corresponding eye height) of
the received signal for channel spacings of 100GHz NRZ, 100GHz RZ, 50GHz NRZ, 50GHz RZ,
25GHz NRZ, 25GHz RZ, 12.5GHz NRZ, 12.5GHz RZ, 6.25GHz NRZ and 6.25GHz RZ respectively.
54
4.2 The 48 Channel x 25 Gbps at 100 GHz Spacing using NRZ
Figure 4.1: Eye Diagram of 100 GHz Channel Spacing Using NRZ, 1 EDFA, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of a 48x25Gbps@100Ghz NRZ

with 1 EDFA, MUX BW = 80 GHz and DEMUX BW = 80 GHz
20
1.00E-06
15 1.00E-17
Q-Factor
Min BER
1.00E-28
10
1.00E-39
5
1.00E-50
0 1.00E-61
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Channel Q Factor Min BER
Figure 4.2: Q Factor vs BER Graph of 100 GHz Spacing Using NRZ, 1 EDFA, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Figure 4.1 and Figure 4.2 represented the signal data for a DWDM network with 100 GHz
channel spacing using NRZ and 1 EDFA. The MUX BW = 80 GHz and DEMUX BW = 80 GHz.
Figure 4.2 showed the Q Factor vs BER Graph of this network. It can be seen from this chart
that the highest quality factor was shown in channel 1, which can be seen to have the lowest
BER as well. Figure 4.1 showed the eye diagram of channel 1 with its Q factor, BER and eye
height. The BW efficiency was 100% because all transmitted channels were received.
55
Figure 4.3: Eye Diagram of 100 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80
GHz and DEMUX BW = 80 GHz

with Symmetric 3 EDFAs
20
1.00E-11
15 1.00E-25
Q-Factor
Min BER
10 1.00E-39
1.00E-53
5
1.00E-67
0 1.00E-81
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.4: Q factor vs BER Graph of 100 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80
that the highest quality factor was shown in channel 1, which had the lowest BER as well.
Figure 4.3 showed the eye diagram of channel 1 with its Q factor, BER and eye height. The BW
efficiency was 100% because all transmitted channels were received.
56

(MUX BW 80 GHz - DEMUX BW 50 GHz)
60 1
1E-59
40
Q-Factor
Min BER
1E-118
1E-177
20
1E-236
0 1E-295
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.6: Q Factor vs BER of 100 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
and DEMUX BW = 50 GHz
The Min BER is truncated at channel 20 and 22 because they had the highest BER and the
lowest Q in comparison to all other channels. BER for all other channels was almost = 0 and
negligible. Figure 4.5 showed the eye diagram of channel 38 with its Q factor, BER and eye
height. The BW efficiency was 100% because all transmitted channels were received.
57

25
1.00E-11
20
1.00E-23
15 1.00E-35
Q-Factor
Min BER
1.00E-47
10 1.00E-59
1.00E-71
5
1.00E-83
0 1.00E-95
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.7 showed the eye diagram of channel 44 with its Q factor, BER and eye height. The
BW efficiency was 100% because all transmitted channels were received.
58
4.3 The 48 Channel x 25 Gbps at 100 GHz Spacing using RZ
Figure 4.9: Eye Diagram of 100 GHz Channel Spacing Using RZ, 1 EDFA, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of a 48x25Gbps@100GHz RZ

with 1 EDFA
18
16 1.00E-05
14 1.00E-12
12 1.00E-19
Q-Factor
Min BER
10 1.00E-26
8 1.00E-33
6 1.00E-40
4 1.00E-47
2 1.00E-54
0 1.00E-61
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.10: Q Factor vs BER of 100 GHz Channel Spacing Using RZ, 1 EDFA, MUX BW = 80 GHz and DEMUX BW = 80 GHz
channel spacing using RZ and 1 EDFA. The MUX BW = 80 GHz and DEMUX BW = 80 GHz.
Figure 4.9 showed the eye diagram of channel 1 with its Q factor, BER and eye height. The BW
efficiency was 100% because all transmitted channels were received.
59
Figure 4.11: Eye Diagram of 100 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80

30 1.00E+00
25 1.00E-28
20 1.00E-56
Q-Factor
Min BER
15 1.00E-84
10 1.00E-112
5 1.00E-140
0 1.00E-168
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.12: Q factor vs BER of 100 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
60

100
1.00E-248
80
1.00E-257
Min BER
Q-Factor
60 1.00E-266
40 1.00E-275
1.00E-284
20
1.00E-293
0 1.00E-302
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.14: Q Factor vs BER of 100 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
80 GHz and DEMUX BW = 50 GHz
that the highest quality factor was shown in channel 1, which had the lowest BER as well. All
Q factors for all channels in this system are so high that the BER for this system = 0, BER for all
channels in this system is negligible. Figure 4.13 showed the eye diagram of channel 1 with its
Q factor, BER and eye height. The BW efficiency was 100% because all transmitted channels
were received.
61
Q-Factor vs Min BER Chart of a 48x25Gbps@100GHz RZ (MUX

BW 50 GHz - DEMUX BW 80 GHz)
70
1E-16
60
1E-33
50
1E-50
Min BER
Q-Factor
40 1E-67
30 1E-84
1E-101
20
1E-118
10 1E-135
0 1E-152
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.16: Q Factor vs BER of 100 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 50 GHz
62
Figure 4.17: Eye Diagram of 50 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of a 48x25Gbps@50GHz NRZ

with 1 EDFA
3.5 1
3
2.5
0.1
Q-Factor
Min BER
2
1.5
0.01
1
0.5
0 0.001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.18: Q factor vs BER of 50 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
63

3.5 1
3
2.5 0.1
Q-Factor
Min BER
2
0.01
1.5
1 0.001
0.5
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
64

(MUX BW 80GHz - DEMUX BW 50GHz)
12
1.00E-02
10
1.00E-06
8
Q-Factor
Min BER
6 1.00E-10
4 1.00E-14
2 1.00E-18
0 1.00E-22
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
65
Q-Factor vs Min BER Chart of a 48x25Gbps@50GHz NRZ (MUX

BW 80GHz - DEMUX BW 25GHz)
12
1.00E-02
10
1.00E-05
8 1.00E-08
Q-Factor
Min BER
6 1.00E-11
1.00E-14
4
1.00E-17
2 1.00E-20
0 1.00E-23
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
66

4
(MUX BW 50 GHz - DEMUX BW 80 GHz) 1
3 0.1
Q-Factor
2 0.01 Min BER
1 0.001
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.26: Q Factor vs BER of 50 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
67
Figure 4.27: Eye Diagram of 50 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz

with 1 EDFAs
4 1
0.1
3
Q-Factor
0.01
2 Mi BER
0.001
1
0.0001
0 0.00001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.28: Q factor vs BER of 50 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
68
Q-Factor vs Min BER Chart of a 48x25Gbps@50GHz RZ with

Symmetric 3 EDFAs
3.5 1
3
2.5
0.1
Q-Factor
Mi BER
1.5
0.01
1
0.5
0 0.001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.30: Q factor vs BER of 50 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 80 GHz
69
Q-Factor vs Min BER Chart of a 48 x 25 Gbps@50 GHz RZ

(MUX BW 80GHz - DEMUX BW 50GHz)
7 1.00E+00
6
1.00E-02
5
Q-Factor
1.00E-04
Mi BER
4
3 1.00E-06
2
1.00E-08
1
0 1.00E-10
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.32: Q Factor vs BER of 50 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz and
DEMUX BW = 50 GHz
70
Q-Factor vs Min BER Chart of a 48 x 25 Gbps@50 GHz RZ (MUX

BW 80GHz - DEMUX BW 25GHz)
30
1.00E-11
25 1.00E-29
20 1.00E-47
Q-Factor
1.00E-65
Mi BER
15 1.00E-83
10 1.00E-101
1.00E-119
5
1.00E-137
0 1.00E-155
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.34: Q factor vs BER Diagram of 50 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80
Figure 4.33 and Figure 4.34 represent the signal data for a DWDM network with 50 GHz
71
Q-Factor vs Min BER Chart of a 48 x 25 Gbps@50 GHz RZ

(MUX BW 50GHz - DEMUX BW 80 GHz)
6 1
5 0.1
0.01
4
Q-Factor
0.001
Mi BER
3
0.0001
2
0.00001
1 0.000001
0 0.0000001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
DEMUX BW = 80 GHz
72
Figure 4.37: Eye Diagram of 25 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min Ber Chart of a 48x25Gbps@25GHz NRZ

4 with 1 EDFAs 1
3 0.1
Q-Factor
Min BER
2 0.01
1 0.001
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.38: Q factor vs BER of 25 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
BW efficiency was 93.75% because only 45 out of the 48 channels were received. Three
channels did not have a recognisable signal since the BER = 1 and Q = 0.
73
Q-Factor vs Min Ber Chart of a 48x25Gbps@25GHz NRZ with

4 Symmetric 3 EDFAs 1
3 0.1
Q-Factor
Min BER
2 0.01
1 0.001
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BW efficiency was 81.25% since only 39 out of 48 channels were recognisable. Nine channels
had BER = 1 and Q = 0.
74
Q-Factor vs Min Ber Chart of a 48x25Gbps@25GHz NRZ (MUX

BW 80GHz - DEMUX BW 25 GHz)
6
1.00E-01
4
Q-Factor
Min BER
1.00E-03
2 1.00E-05
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BW efficiency was 89.58% because only 43 out of 48 channels was recognisable. Five of the
channels had BER = 1 and Q = 0.
75
GHz and DEMUX BW = 12.5 GHz

(MUX BW 80GHz - DEMUX BW 12.5 GHz)
3 1
2.5
2 0.1
Q-Factor
Min BER
1.5
1 0.01
0.5
0 0.001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
and DEMUX BW = 12.5 GHz
channel spacing using NRZ and 3 EDFA. The MUX BW = 80 GHz and DEMUX BW = 12.5 GHz.
Figure 4.43 shows the eye diagram of channel 46 with its Q factor, BER and eye height. The
BW efficiency was 97.91% because 47 out of 48 channels were recognisable. One channel had
BER = 1 and Q = 0.
76

3.5 1
3
2.5 0.1
Q-Factor
Min BER
2
0.01
1.5
1 0.001
0.5
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BW efficiency was 93.75% because 45 out of 48 channels were transmitted. Three channels
had a BER = 1 and Q = 0.
77
Figure 4.47: Eye Diagram of 25 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 12.5

BW 12.5 GHz - DEMUX BW 80 GHz)
3.5 1
3
2.5 0.1
Q-Factor
Min BER
2
0.01
1.5
1 0.001
0.5
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.48: Q Factor vs BER of 25 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 12.5 GHz
channel spacing using NRZ and 3 EDFA. The MUX BW = 12.5 GHz and DEMUX BW = 80 GHz.
BW efficiency was 66.66% because 32 out of 48 channels were recognisable. Sixteen channels
had BER =1 and Q = 0.
78
Figure 4.49 Eye Diagram of 25 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 25

4 1
0.1
3
Q-Factor
0.01
Min BER
2
0.001
1
0.0001
0 0.00001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BW efficiency was 93.75% because 45 out 48 channels were recognisable. Three channels had
a BER = 1 and Q = 0.
79
Figure 4.51: Eye Diagram of 25 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB,
Rectangular Filter, MUX BW = 80 GHz and DEMUX BW = 25 GHz

(MUX BW 80 GHz - DEMUX BW 25 GHz), ER = 5dB &
Rectangular Filter
6 1
4 0.01
Q-Factor
Min BER
2 0.0001
0 0.000001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.52: Q Factor vs BER of 25 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB, Rectangular
Filter, MUX BW = 80 GHz and DEMUX BW = 25 GHz
Figure 4.51 and Figure 4.52 represented the signal data for a 25 GHz Spacing Using NRZ, Post
Symmetric Dispersion with 3 EDFAs, ER = 5dB and a Rectangular Filter. The MUX BW = 80 GHz
and DEMUX BW = 25 GHz.
BW efficiency was 91.66% because 44 out of 48 channels had a recognisable signal. Four
channels had BER =1 and Q = 0.
80
Figure 4.53: Eye Diagram of 25 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB,

BW 25 GHz - DEMUX BW 25 GHz), ER = 5dB & Rectangular
Filter
8
1.00E-01
6 1.00E-03
Q-Factor
Min BER
4 1.00E-05
1.00E-07
2
1.00E-09
0 1.00E-11
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.54: Q Factor vs BER of 25 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB, Rectangular
Figure 4.53 and Figure 4.54 represented the signal data for a 25 GHz Spacing Using NRZ, Post
Symmetric Dispersion with 3 EDFAs, ER = 5dB and a Rectangular Filter. The MUX BW = 25 GHz
and DEMUX BW = 25 GHz.
BW efficiency was 100% since all transmitted channels were received.
81
Figure 4.55: Eye Diagram of 25 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of 48x25Gbps@25GHz RZ with 1

4 EDFAs 1
3 0.1
Q-Factor
Min BER
2 0.01
1 0.001
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.56: Q Factor vs BER of 25 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Figure 4.55 showed the eye diagram of channel 16 showing its Q factor, BER and eye height.
The BW efficiency was 79.16% because 38 out of 48 channels were recognisable. Ten channels
82
Q-Factor vs Min BER Chart of 48x25Gbps@25GHz RZ

4 1
3 0.1
Q-Factor
2 0.01 Min BER
1 0.001
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
DEMUX BW = 80 GHz
BW efficiency was 83.33% because 40 out of 48 channels were recognisable. Eight channels
83

(MUX BW 80GHz - DEMUX BW 25 GHz)
8
0.1
6
0.001
Q-Factor
Min BER
4
0.00001
2 0.0000001
0 1E-09
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
DEMUX BW = 25 GHz
BW efficiency was 91.66% because 44 out of 48 channels were recognisable. Four channels
84

(MUX BW 80GHz - DEMUX BW 12.5 GHz)
6 1.00E+00
5 1.00E-01
1.00E-02
4
Q-Factor
Min BER
1.00E-03
3
1.00E-04
2
1.00E-05
1 1.00E-06
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
DEMUX BW = 12.5 GHz
channel spacing using RZ and 3 EDFA. The MUX BW = 80 GHz and DEMUX BW = 12.5 GHz.
BW efficiency was 85.41% because 41 out of 48 channels were recognisable. Seven channels
85
Figure 4.63 Eye Diagram of 25 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 25 GHz

3.5 1.00E+00
3
2.5 1.00E-01
Q-Factor
Min BER
2
1.00E-02
1.5
1 1.00E-03
0.5
0 1.00E-04
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
DEMUX BW = 80 GHz
BW efficiency was 89.58% because 43 out of 48 channels were recognisable. Five channels
86
Figure 4.65: Eye Diagram of 25 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 12.5

(MUX BW 12.5 GHz - DEMUX BW 80 GHz)
3.5 1.00E+00
3
2.5 1.00E-01
Q-Factor
Min BER
2
1.00E-02
1.5
1 1.00E-03
0.5
0 1.00E-04
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.66: Q Factor vs BER of 25 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 12.5 GHz
channel spacing using RZ and 3 EDFA. The MUX BW = 12.5 GHz and DEMUX BW = 80 GHz.
BW efficiency was 79.16% because 38 out of 48 channels were recognisable. Ten channels had
BER =1 and Q = 0.
87
Figure 4.67:Eye Diagram of 25 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 25

6 1.00E+00
5 1.00E-01
1.00E-02
4
Q-Factor
Min BER
1.00E-03
3
1.00E-04
2
1.00E-05
1 1.00E-06
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.68: Q factor vs BER of 25 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 25
BER =1 and Q = 0.
88
Figure 4.69: Eye Diagram of 25 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB,
Q-Factor vs Min Ber Chart of a 48x25Gbps@25GHz RZ

4
Rectangular Filter 1
0.1
Q-Factor
Min BER
2
0.01
0 0.001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.70: Q Factor vs BER of 25 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB, Rectangular
Spacing using RZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB and a Rectangular Filter.
The MUX BW = 80 GHz and DEMUX BW = 25 GHz.
BER =1 and Q = 0.
89
Figure 4.71: Eye Diagram of 25 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB,
Q-Factor vs Min Ber Chart of a 48x25Gbps@25GHz RZ

Rectangular Filter
15
1.00E-09
10
Q-Factor
Min BER
1.00E-20
5
1.00E-31
0 1.00E-42
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.72: Q factor vs BER of 25 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB, Rectangular
Spacing using RZ, Post Symmetric Dispersion with 3 EDFAs, ER = 5dB and a Rectangular Filter.
The MUX BW = 25 GHz and DEMUX BW = 25 GHz.
BER =1 and Q = 0
90
4.8 The 48 Channel x 25 Gbps at 12.5 GHz Spacing using NRZ
Figure 4.73: Eye Diagram of 12.5 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of [email protected] NRZ with 1

EDFA
4 1.00E+00
3.5
3 1.00E-01
2.5
Q-Factor
Min BER
2 1.00E-02
1.5
1 1.00E-03
0.5
0 1.00E-04
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.74: Q factor vs BER of 12.5 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Figure 4.73 and Figure 4.74 represented the signal data for a DWDM network with 12.5 GHz
BW efficiency was 81.25% because 39 out of 48 channels were recognisable. Nine channels
91
Figure 4.75: Eye Diagram of 12.5 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80
Q-Factor vs Min BER Chart of [email protected] NRZ

4
with Symmetric 3 EDFAs 1.00E+00
3 1.00E-01
Q-Factor
Min BER
2 1.00E-02
1 1.00E-03
0 1.00E-04
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.76: Q factor vs BER of 12.5 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
BW efficiency was 75% because 36 out of 48 channels were recognisable. Twelve channels
had BER =1 and Q = 0
92
Q-Factor vs Min BER Chart of [email protected] NRZ (MUX

BW 80 GHz - DEMUX BW 12.5 GHz)
5 1
4 0.1
0.01
Q-Factor
Min BER
0.001
2
0.0001
1 0.00001
0 0.000001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.78: Q Factor vs BER of 12.5 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
BW efficiency was 72.91% because 35 out of 48 channels were recognisable. Thirteen channels
93

(MUX BW 80 GHz - DEMUX BW 6.25 GHz)
5 1
4 0.1
3 0.01
Q-Factor
Min BER
2 0.001
1 0.0001
0 0.00001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
94
Figure 4.81: Eye Diagram of 12.5 GHz Channel Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
12.5 GHz and DEMUX BW = 80 GHz

4 1
0.1
3
Q-Factor
0.01
Min BER
2
0.001
1
0.0001
0 0.00001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.82: Q Factor vs BER of 12.5 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 12.5
95

4 1
3
0.1
Q-Factor
Min BER
2
0.01
1
0 0.001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.84: Q factor vs BER of 12.5 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 6.25 GHz
96

5 1
4 0.1
3 0.01
Q-Factor
Min BER
2 0.001
1 0.0001
0 0.00001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.86: Q factor vs BER of 12.5 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 3.125
97
4.9 The 48 Channel x 25 Gbps at 12.5 GHz Spacing using RZ
Figure 4.87: Eye Diagram of 12.5 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of [email protected] RZ

6
with 1 EDFAs
1.00E-01
4
Q-Factor
Min BER
1.00E-03
2 1.00E-05
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.88: Q Factor vs BER of 12.5 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
BW efficiency was 66.66% because 35 out of 48 channels were recognisable. Sixteen channels
98
Figure 4.89: Eye Diagram of 12.5 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80

6
1.00E-01
4
Q-Factor
Min BER
1.00E-03
2 1.00E-05
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.90: Q Factor vs BER of 12.5 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
99
Q-Factor vs Min BER Chart of [email protected] RZ (MUX

5 1.00E+00
4 1.00E-01
1.00E-02
Q-Factor
Min BER
3
1.00E-03
2
1.00E-04
1 1.00E-05
0 1.00E-06
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BW efficiency was 70.83% because 34 out of 48 channels were recognisable. Fourteen
channels had BER =1 and Q = 0
100

6 1.00E+00
5 1.00E-01
1.00E-02
4
Q-Factor
Min BER
1.00E-03
3
1.00E-04
2
1.00E-05
1 1.00E-06
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BER =1 and Q = 0.
101
Figure 4.95: Eye Diagram of 12.5 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 12.5

4 1
3.5
3 0.1
2.5
Q-Factor
Min BER
2 0.01
1.5
1 0.001
0.5
0 0.0001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.96: Q factor vs BER of 12.5 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 12.5 GHz
BW efficiency was 77.08% because 37 out of 48 channels were recognisable. Eleven channels
102
Figure 4.97: Eye Diagram of 12.5 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 6.25

3.5 1
3
2.5
0.1
Q-Factor
Min BER
2
1.5
0.01
1
0.5
0 0.001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.98: Q Factor vs BER of 12.5 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 6.25 GHz
Figure 4.97 and Figure 4.98 represent the signal data for a DWDM network with 12.5 GHz
Figure 4.98 shows the Q Factor vs BER Graph of this network. It can be seen from this chart
Figure 4.97 shows the eye diagram of channel 30 with its Q factor, BER and eye height. The
BW efficiency was 68.75% because 33 out of 48 channels were recognisable. Fifteen channels
103
Figure 4.99: Eye Diagram of 12.5 GHz Channel Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =

5 1
4 0.1
3 0.01
Q-Factor
Min BER
2 0.001
1 0.0001
0 0.00001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.100: Q Factor vs BER Diagram of 12.5 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
BW efficiency was 100% since all channels transmitted were received.
104
4.10 The 48 Channel x 25 Gbps at 12.5 GHz Spacing using MDRZ
Figure 4.101: Eye Diagram of 12.5 GHz Channel Spacing Using MDRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
Q-Factor vs Min BER Chart of [email protected] MDRZ

14
1.00E-02
12
1.00E-08
10
1.00E-14
Q-Factor
8
1.00E-20 Min BER
6
4 1.00E-26
2 1.00E-32
0 1.00E-38
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.102: Q Factor vs BER of 12.5 GHz Spacing Using MDRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
channel spacing using MDRZ and 3 EDFA. The MUX BW = 3.125 GHz and DEMUX BW = 80 GHz.
105
4.11 The 48 Channel x 25 Gbps at 6.25 GHz Spacing using NRZ
Figure 4.103: Eye Diagram of 6.25 GHz Channel Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of [email protected] NRZ with 1

EDFAs
5 1.00E+00
4 1.00E-01
Q-Factor
Min BER
3 1.00E-02
2 1.00E-03
1 1.00E-04
0 1.00E-05
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.104: Q factor vs BER of 6.25 GHz Spacing Using NRZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
BER =1 and Q = 0.
106

5 1.00E+00
4 1.00E-01
Q-Factor
Min BER
3 1.00E-02
2 1.00E-03
1 1.00E-04
0 1.00E-05
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.106: Q factor vs BER of 6.25 GHz Spacing Using NRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
BW efficiency was 89.52% because 43 out of 48 channels were recognisable. Five channels
107
80 GHz and DEMUX BW = 6.25 GHz

BW = 80 GHz & DEMUX BW = 6.25 GHz)
5 1
4 0.1
0.01
Q-Factor
Min BER
3
0.001
2
0.0001
1 0.00001
0 0.000001
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
108

4 1.00E+00
1.00E-01
3
Q-Factor
Min BER
1.00E-02
2
1.00E-03
1
1.00E-04
0 1.00E-05
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BW efficiency was 95.83% because 46 out of 48 channels were recognisable. Two channels
109

5 1.00E+00
1.00E-01
4
1.00E-02
3
Q-Factor
Min BER
1.00E-03
2 1.00E-04
1.00E-05
1
1.00E-06
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
BW efficiency was 100% since all transmitted channel were received.
110
4.12 The 48 Channel x 25 Gbps at 6.25 GHz Spacing using RZ
Figure 4.113: Eye Diagram of 6.25 GHz Channel Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
Q-Factor vs Min BER Chart of [email protected] RZ with 1

EDFA
5 1.00E+00
4 1.00E-01
3 1.00E-02
Q-Factor
Min BER
2 1.00E-03
1 1.00E-04
0 1.00E-05
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.114: Q Factor vs BER of 6.25 GHz Spacing Using RZ, 1 EDFAs, MUX BW = 80 GHz and DEMUX BW = 80 GHz
BW efficiency was 70.83% because 34 out of 48 channels were recognisable. Fourteen
channels had BER =1 and Q = 0.
111
Q-Factor vs Min BER Chart of [email protected] RZ with

Symmetric 3 EDFAs
5 1.00E+00
4 1.00E-01
1.00E-02
3
Q-Factor
Min BER
1.00E-03
2
1.00E-04
1 1.00E-05
0 1.00E-06
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.116: Q factor vs BER of 6.25 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
BER =1 and Q = 0.
112

(MUX BW 80 GHz - DEMUX BW 6.25 GHz)
10 1.00E+00
8 1.00E-03
Q-Factor
Min BER
6 1.00E-06
4 1.00E-09
2 1.00E-12
0 1.00E-15
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.118: Q factor vs BER of 6.25 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 80 GHz
113

4 1.00E+00
1.00E-01
3
Q-Factor
1.00E-02
Min BER
2
1.00E-03
1
1.00E-04
0 1.00E-05
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.120: Q Factor vs BER of 6.25 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 6.25 GHz
BW efficiency was 87.5% because 42 out of 48 channels were recognisable. Six channels had
BER =1 and Q = 0.
114

5 1.00E+00
1.00E-01
4
1.00E-02
3
Q-Factor
Min BER
1.00E-03
2 1.00E-04
1.00E-05
1
1.00E-06
0 1.00E-07
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.122: Q factor vs BER of 6.25 GHz Spacing Using RZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW = 3.125
115
4.13 The 48 Channel x 25 Gbps at 6.25 GHz Spacing using MDRZ
Figure 4.123: Eye Diagram of 6.25 GHz Channel Spacing Using MDRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
Q-Factor vs Min BER Chart of [email protected] MDRZ

15
1.00E-04
10 1.00E-10
Q-Factor
1.00E-16 Min BER

5
1.00E-22
0 1.00E-28
1 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48
Figure 4.124: Q Factor vs BER of 6.25 GHz Spacing Using MDRZ, Post Symmetric Dispersion with 3 EDFAs, MUX BW =
channel spacing using MDRZ and 3 EDFA. The MUX BW = 3.125 GHz and DEMUX BW = 80 GHz.
116
4.14 Summary
The chapter analysed eye-diagram results of each channel in the DWDM system. The eye
diagrams were a representation of the simulated received signals in each combination case of
the varied parameters. The eye diagram data showed the Q factor, minimum BER and eye
height of the received signal of all 48 channels for every channel spacing, modulation format
and design parameter combinations of the designed networks.
Since BER was inversely proportional to the Q factor, the channels with the highest Q factor
had the lowest BER and vice versa was also true. The channels with the highest Q was used
for analysis for all networks to compare the maximum signal performance of that network to
the other networks. A signal with a BER = 1 indicated that the bit errors received at the optical
receiver were too high and the signal was not recognisable. Therefore, the Q factor will be =
0 for such signals. This is clearly shown on the comparative charts. The BW efficiency of each
network was calculated using (2.19) shown in section 2.8.4 of chapter 2. This indicated the
number of channels that would provide a recognisable signal at the optical receiver. For such
signals BER < 1 and Q > 0. BW efficiency was also an indication of how much of the bandwidth
of the 48 channels were utilised to send a signal that could be recognised by the receiver.
However, a recognisable signal at the optical receiver was not always an acceptable signal that
would be able to transmit information with the least amount of errors.
The minimum BER required for an acceptable signal must be > 10-9 for the optical receiver to
detect the information signal with the least amount of bit errors. The Q factors for such signals
would be > 6.
The eye height of an eye diagram illustrated the size of the vertical eye opening of a received
signal. The more noise there is in a received signal, the more closure there will be in the eye
opening. The eye height is therefore an indication of the amount of noise in the received
signal. Furthermore, the OSNR value of high-speed data signals are directly impacted by the
amount of eye closure. This means that a higher eye height of a received signal will have a
higher OSNR and vice versa.
117
DISCUSSION OF RESULTS FOR THE 48 CHANNEL DWDM NETWORKS
5.1 Introduction
This chapter analysed and discussed all the results obtained for the designed 48 channel
DWDM network models from Chapter 4. The tabulated OSNR results of the mismatched active
component design from [2] was compared with the tabulated OSNR results of the matched
active component design for each channel spacing of the 48-channel network used in this
dissertation. In both cases, the OSNR was used to compare and evaluate the signal
performance of the 1 EDFA post compensation technique with that of the 3 EDFA post
symmetric compensation technique. See Table 5.1
Furthermore, in this chapter, the results of all the eye diagrams with the highest Q factor and
lowest BER illustrated in chapter 4 was tabulated for signal performance comparison of each
channel spacings and parameter combinations. The section analysed and compared the
effectiveness of using the NRZ, RZ and MDRZ modulation formats at various channel spacings
using the 3 EDFA Post Symmetric Compensation Technique. This was done to find the most
optimised design parameter for maximum signal performance of a matched 48 Channel
DWDM system.
118
5.2 Comparison of OSNR for Post Compensation Technique (with 1 EDFA) vs
Post Symmetric Compensation Technique (with 3 EDFAs)
The signal power received at the optical receiver was calculated using (2.10) shown in section
2.7 of chapter 2. The calculated signal power was found to be equal to the maximum signal
power measured in Optisystem. The noise power was measured placing a WDM analyser at
the output of the WDM DEMUX. The OSNR was calculated using (2.12) found in section 2.8.1
of chapter 2.
The mismatched frequencies between the 48 channel WDM transmitter and WDM MUX for
networks used in [2] increased the chromatic distortion (CD), which then significantly
decreased signal power and its OSNR. However, this was done to decrease the non-linear
effects of FWM because FWM effects were inversely dependant on CD. There was a definite
and clear improvement in OSNR when using matched components than as opposed to using
mismatched components for all channel spacings as shown in Table 5.1. This was because
matched active components used frequency channels that were aligned to each other and the
BW of each channel did not overlap with each other.
From section 3.3.1.2 of chapter 3, it had been established that the span loss for 1 EDFA is much
greater than the span loss for 3 EDFAs and power received at the photodetector for 3 EDFAs
was far greater than for 1 EDFA. The received signal for 3 EDFA was more easily detected and
had a higher signal quality than with 1 EDFA.
Furthermore, Table 5.1 shows the maximum signal power and OSNR of the post Symmetric 3
EDFA technique, which was always higher than the 1 EDFA post compensation technique for
all channel spacings in any type of network. The addition of the 3 symmetric EDFAs in all
networks compensated for the increased attenuation and dispersion introduced when the bit
rate and number of channels were increased as shown from the design done in [1]. It can also
be seen that the NRZ format always displayed a higher signal to noise ratio than the RZ. This
was because NRZ was more robust to impacts of chromatic dispersion as compared to RZ.
Linear and non-linear losses were mostly supressed at 100GHz channel spacing for NRZ. NRZ
had higher signal power and lower noise power than RZ due to NRZ eliminating more effects
of CD.
119
Table 5.1: OSNR Comparison of Post Compensation Technique with 1 EDFA vs Post Symmetric compensation Technique
with 3 EDFAs [2]
Channel Optical Mod. Max Max Max

Spacing Link Type Signal Power Noise Power OSNR
(GHz) (dBm) (dBm) (dB)
1 EDFA
(WDM Tx & NRZ -43.83 -38.28 0
100 WDM MUX
mismatched) RZ -48.24 -38.28 0
Symmetric
Post Comp. NRZ -2.98 -16.13 13.64
3 EDFAs
(WDM Tx &
WDM MUX RZ -8.43 -16.13 9.11
mismatched)
1 EDFA NRZ -12.38 -38.28 26.01

(Matched)
RZ -17.70 -38.28 20.66
Symmetric
NRZ 28.87 -16.13 45.11
Post Comp.
3 EDFAs
(Matched) RZ 23.54 -16.13 39.76
1 EDFA
(WDM Tx & NRZ -43.22 -38.28 24.39
50 WDM MUX
mismatched) RZ -46.64 -38.28 19.01
Symmetric
Post Comp. NRZ -2.41 -16.13 43.55
3 EDFAs
(WDM Tx &
WDM MUX RZ -7.53 -16.13 38.17
mismatched)
1 EDFA NRZ -12.41 -38.28 26.08

(Matched)
RZ -17.81 -38.28 20.58
Symmetric
NRZ 28.87 -16.13 45.23
Post Comp.
3 EDFAs
RZ 23.53 -16.13 39.77
(Matched)
1 EDFA
(WDM Tx & NRZ -25.67 -38.28 24.81
25 WDM MUX
mismatched) RZ -31.80 -38.28 18.67
Symmetric
Post Comp. NRZ 15.58 -16.13 43.92
120
3 EDFAs
(WDM Tx &
WDM MUX RZ 10.03 -16.13 38.36
mismatched)
1 EDFA NRZ -12.36 -38.28 26.15

(Matched)
RZ -19.64 -38.28 19.90
Symmetric
NRZ 28.93 -16.13 45.34
Post Comp.
3 EDFAs
RZ 22.69 -16.13 39.18
(Matched)
1 EDFA
NRZ -19.03 -38.28 23.67
(WDM Tx &
12.5 WDM MUX
mismatched) RZ -23.89 -38.28 18.62
Symmetric
Post Comp. NRZ 22.22 -16.13 42.78
3 EDFAs
(WDM Tx &
WDM MUX RZ 17.91 -16.13 37.73
mismatched)
1 EDFA NRZ -11.53 -38.28 27.35

(Matched)
RZ -17.33 -38.28 22.12
Symmetric
NRZ 29.60 -16.13 46.62
Post Comp.
3 EDFAs
RZ 23.98 -16.13 41.49
(Matched)
1 EDFA
(WDM Tx & NRZ -13.42 -38.28 24.86
6.25 WDM MUX
mismatched) RZ -18.58 -38.28 19.69
Symmetric
Post Comp. NRZ 27.38 -16.13 45.51
3 EDFAs
(WDM Tx &
WDM MUX RZ 22.12 -16.13 38.25
mismatched)
1 EDFA NRZ -7.77 -38.28 31.55

(Matched)
RZ -14.20 -38.28 24.94
Symmetric
NRZ 33.69 -16.13 50.49
Post Comp.
3 EDFAs
RZ 27.25 -16.13 44.05
(Matched)
121
5.3 Signal performance comparison of NRZ, RZ and MDRZ at various channel
spacings for maximum optimisation of a 48 Channel DWDM system
The section showed the tabulated results of all eye diagrams with the highest Q factor
illustrated in chapter 4 for signal performance comparison of each channel spacings and
parameter combinations. Table 5.2, Table 5.3, Table 5.4, Table 5.5 and Table 5.6 showed the
tabulated results of the 100 GHz, 50 GHz, 25 GHz, 12.5 GHz and 6.25 GHz channel spacing
respectively.
The average Q-factor of a network was an indication of the overall signal performance of the
system and was calculated using (2.18) shown in section 2.8.4 of chapter 2. The overall signal
performance of a system was deemed acceptable and optimised only if the average Q factor
of the system was > 6. For this to be true, almost all channels had to perform with a Q factor
that was either ≈ or ≥ 6, which was the minimum Q factor needed to obtain a BER of ≥ 10−9.
This was needed to produce an acceptable signal that can be detected at the optical receiver
with the least amount of bit errors. The BW efficiency of each network was an indication of
the number of channels that were received by the optical receiver that had some recognisable
signal. For this to happen, BER must be < 1 and Q factor must be > 0. The BW efficiency was
calculated using (2.19) shown in section 2.8.4 of chapter 2.
Table 5.2: Results of 100GHz channel spacing with various design parameters and modulation types
Avg. Q-Factor
Channel Various of all BW
Q-Factor
Channel
Highest
Mod Min. Eye

Spacing Design 48 Channels Eff.
Type BER Height
(GHz) Parameters (Overall (%)
Performance)
Symmetric
Post Comp. NRZ 10.97 48 20.34 2.76E-92 8.25E-04 97.9
3 EDFAs
(WDM Tx &
100 GHz WDM MUX RZ 4.77 48 24.20 8.53E-130 6.66E-04 100
Mismatched)
MUX BW
1 EDFA NRZ 12.77 1 16.47 2.67E-61 8.82E-05 100
= 80 GHz
(Matched)
DEMUX BW RZ 11.76 1 16.50 1.63E-61 5.67E-05 100
= 80 GHz Symmetric
NRZ 13.88 1 18.99 9.76E-81 1.21 100
Post Comp.
3 EDFAs
RZ 18.83 1 27.59 5.82E-168 0.77 100
(Matched)
122
MUX BW
NRZ 40.68 38 54.32 0 1.35 100
100 GHz = 80 GHz
DEMUX BW
RZ 53.37 1 82.60 0 0.78 100
= 50 GHz
Symmetric
MUX BW
Post Comp. NRZ 13.52 44 20.66 3.30E-95 1.27 100
= 50 GHz
3 EDFAs
DEMUX BW
(Matched) RZ 24.67 1 59.89 0 0.72 100
= 80 GHz
Table 5.2 showed the summarised results of the eye diagram for the 100GHz channel spacing
with various design parameters and modulation types. A post symmetric compensation
technique with 3 EDFAs using MUX BW = 80 GHz and DEMUX BW = 50 GHz was the most
optimised network with highest signal performance. This was underlined and highlighted in
italics in Table 5.2. For this network configuration, the RZ (Avg. Q= 53.37) performed better
than NRZ (Avg. Q = 40.68). The BER for NRZ and RZ is equal to 0.
Table 5.3: Results of 50GHz channel spacing with various design parameters and modulation types
Avg. Q-Factor
Q-Factor
Channel
Highest
Mod Min. Eye

Type BER Height
Performance)
Symmetric
Post Comp. NRZ 1.04 1 4.15 1.61E-05 6.16E-02 66.6
3 EDFAs
(WDM Tx &
WDM MUX RZ 1.33 18 3.09 9.66E-04 1.22E-05 75
mismatched)
50 GHz
NRZ 2.08 48 3.08 1.00E-03 2.91E-06 100
MUX BW 1 EDFA
= 80 GHz (Matched)
DEMUX BW RZ 2.22421 36 3.79 7.47E-05 2.35E-05 100
= 80 GHz
Symmetric NRZ 2.06 48 3.14 8.29E-04 6.19E-02 100

Post Comp.
3 EDFAs
(Matched) RZ 2.22573 48 2.8 2.07E-03 -0.026 100
123
MUX BW
NRZ 7.03 32 9.58 4.47E-22 1.00 100
= 80 GHz
DEMUX BW
RZ 3.56 38 6.31 1.10E-10 0.41 100
50 GHz = 50 GHz
MUX BW
NRZ 8.72 1 9.93 1.06E-23 0.90 100
= 80 GHz
Symmetric
DEMUX BW 1.41E-
Post Comp. RZ 10.02 48 26.53 0.48 100
= 25 GHz 155
3 EDFAs
(Matched) MUX BW
NRZ 2.23 42 3.71 1.01E-04 0.34 100
= 50 GHz
DEMUX BW
RZ 2.34 42 5.11 1.58E-07 0.38 100
= 80 GHz
Table 5.3 showed the summarised results of the eye diagram for the 50 GHz channel spacing
with various design parameters and modulation types (NRZ and RZ). A post symmetric
compensation technique with 3 EDFAs using MUX BW = 80 GHz and DEMUX BW = 25 GHz was
the most optimised network with the highest signal performance. This was underlined and
highlighted in italics in Table 5.3. For this network configuration, RZ (Avg. Q= 10.02) performed
better than NRZ (Avg. Q = 8.72). The BER of RZ was lower than NRZ.
Table 5.4: Results of 25 GHz channel spacing with various design parameters and modulation types
Avg. Q-Factor
Q-Factor
Channel
Highest
Mod. Min. Eye

Type BER Height
Performance)
Symmetric
Post Comp. NRZ 2.27 4 4.34 6.23E-06 3.34E-01 91.6
3 EDFAs
(WDM Tx &
25 GHz WDM MUX RZ 2.50 4 5.00 2.36E-07 2.77E-01 97.9
mismatched)
MUX BW
= 80 GHz 1 EDFA NRZ 2.16 32 3.11 9.08E-04 1.46E-05 93.75
DEMUX BW (Matched)
RZ 1.69 16 3.55 1.49E-04 2.06E-05 79.16
= 80 GHz
Symmetric
NRZ 1.63 4 3.15 7.76E-04 0.24 81.25
Post Comp.
3 EDFAs
RZ 1.86 28 3.66 1.02E-04 0.32 83.33
(Matched)
124
MUX BW
NRZ 2.41 1 4.93 3.68E-07 0.51 89.58
= 80 GHz
DEMUX BW
RZ 2.24 26 5.90 1.52E-09 0.24 91.66
= 25 GHz
MUX BW
NRZ 2.28 46 2.67 3.60E-03 -0.14 97.91
= 80 GHz
DEMUX BW
RZ 2.26 16 5.03 2.40E-07 0.29 85.41
25 GHz = 12.5 GHz
MUX BW
NRZ 2.21 16 3.21 6.39E-04 0.39 93.75
= 25 GHz
Symmetric
DEMUX BW
Post Comp. RZ 1.91 44 3.15 7.97E-04 3.00E-02 89.58
= 80 GHz
3 EDFAs
(Matched) MUX BW
NRZ 0.98 16 3.29 4.95E-04 0.34 66.66
= 12.5 GHz
DEMUX BW
RZ 1.51 10 3.11 9.22E-04 3.99E-02 79.16
= 80 GHz
MUX BW
NRZ 2.09 44 3.73 7.37E-05 0.33 93.75
= 25 GHz
DEMUX BW
RZ 3.13 1 4.84 6.03E-07 0.13 97.91
= 25 GHz
25 GHz MUX BW NRZ 2.56 48 4.56 2.39E-06 0.28 91.66
= 80 GHz
ER = 5dB & DEMUX BW -1.99E-
RZ 1.39 8 2.90 1.81E-03 79.16
Rectangular = 25 GHz 02
Filter
NRZ 5.83 8 6.51 3.54E-11 0.45 100
MUX BW
Symmetric = 25 GHz
Post Comp. DEMUX BW
3 EDFAs RZ 9.59 32 13.55 3.54E-42 0.30 97.91
= 25 GHz
(Matched)
Table 5.4 showed the summarised results of the eye diagram for the 25 GHz channel spacing
with various design parameters and modulation types (NRZ and RZ). A post symmetric
compensation technique with 3 EDFAs, an ER =5dB and a rectangular filter using MUX BW =
25 GHz and DEMUX BW = 25 GHz was the most optimised network with highest signal
performance. This was underlined and highlighted in italics in Table 5.4. For this network
configuration, RZ (Avg. Q= 9.59) performed better than NRZ (Avg. Q = 5.83). BER of RZ is lower
than NRZ. Further improvement of this network can be done in the future for the average Q
of NRZ to be also greater than 6 by using MDRZ or by decreasing the BW of the filter, but this
was not the focus of this dissertation.
125
Table 5.5: Results of 12.5 GHz channel spacing with various design parameters and modulation types
Avg. Q-Factor
Q-Factor
Channel
Highest
Mod Min. Eye
Type BER Height
Performance)
Symmetric
Post Comp. NRZ 2.56 1 3.27 1.91E-04 6.20E-02 95.8
3 EDFAs
(WDM Tx &
12.5 GHz WDM MUX RZ 2.77 1 3.89 4.09E-05 1.33E-01 97.9
mismatched)
MUX BW
= 80 GHz 1 EDFA NRZ 1.66 4 3.43 2.90E-04 1.55E-04 81.2
DEMUX BW (Matched)
RZ 1.17 10 4.87 5.46E-07 1.29E-04 66.6
= 80 GHz
Symmetric
NRZ 1.51 2 3.65 1.07E-04 2.73 75
Post Comp.
3 EDFAs
RZ 1.65 26 4.98 3.17E-07 1.83 72.9
(Matched)
MUX BW
NRZ 1.57 22 4.43 4.62E-06 1.09 72.9
= 80 GHz
DEMUX BW
RZ 1.48 18 4.64 1.67E-06 0.32 70.8
= 12.5 GHz
MUX BW
NRZ 1.46 18 4.19 1.30E-05 0.59 72.9
= 80 GHz
DEMUX BW
RZ 2.26 16 4.84 6.26E-07 0.25 79.1
= 6.25 GHz
12.5 GHz
MUX BW
NRZ 1.41 12 3.80 7.07E-05 2.94 75
= 12.5 GHz
Symmetric DEMUX BW
Post Comp. RZ 1.64 48 3.59 1.59E-04 0.48 77.0
= 80 GHz
3 EDFAs MUX BW
(Matched) NRZ 1.92 1 3.05 1.11E-03 0.15 85.4
= 6.25 GHz
DEMUX BW
RZ 1.07 30 3.01 1.27E-03 1.72E-02 68.7
= 80 GHz
MUX BW NRZ 3.06 24 4.00 2.54E-05 2.59 100
= 3.125 GHz
DEMUX BW RZ 3.24 34 4.12 1.55E-05 0.84 100
= 80 GHz MDRZ 8.66 14 12.77 7.49E-38 0.43 100
Table 5.5 showed the summarised results of the eye diagrams for the 12.5 GHz channel
spacing with various design parameters and modulation types. A post symmetric
126
compensation technique with 3 EDFAs and a MDRZ line coding modulation scheme using MUX
BW = 3.125 GHz and DEMUX BW = 80 GHz was the most optimised network with highest signal
configuration, MDRZ (Avg. Q= 8.66) performed better than NRZ (Avg. Q = 3.06) and RZ (Avg.
Q = 3.24). The BER of MDRZ is lower than that of NRZ and RZ.
Table 5.6: Results of 6.25 GHz channel spacing with various design parameters and modulation types
Avg. Q-Factor
Q-Factor
Channel
Highest
Mod Min. Eye
Type BER Height
Performance)
Symmetric
NRZ 3.329 26 3.77 7.82E-05 7.35E-03 100
Post Comp.
3 EDFAs
(WDM Tx &
6.25 GHz RZ 1.868 26 3.65 1.27E-04 5.83E-04 79.1
WDM MUX
mismatched)
MUX BW
= 80 GHz 1 EDFA NRZ 1.79 34 3.85 5.86E-05 1.02E-03 79.1
DEMUX BW (Matched)
RZ 1.51 8 4.13 1.62E-05 1.70E-04 70.8
= 80 GHz
Symmetric
NRZ 2.41 2 4.06 2.37E-05 7.49 89.5
Post Comp.
3 EDFAs
RZ 2.14 40 4.41 4.94E-06 2.73 79.1
(Matched)
MUX BW
NRZ 1.48 46 4.18 7.26E-06 0.94 75
= 80 GHz
DEMUX BW
RZ 2.67 46 7.86 1.86E-15 0.75 85.4
= 6.25 GHz
6.25 GHz
MUX BW
NRZ 3.03 18 3.57 1.73E-04 7.28 95.8
= 6.25 GHz
Symmetric DEMUX BW
RZ 2.49 38 3.61 5.55E-05 1.94 87.5
Post Comp. = 80 GHz
3 EDFAs
(Matched) MUX BW NRZ 4.04 32 4.63 6.26E-07 14.25 100
= 3.125 GHz
DEMUX BW RZ 4.19 36 4.70 4.56E-07 4.26 100
= 80 GHz MDRZ 8.93 38 10.7 7.17E-28 1.45 100
Table 5.6 showed the summarised results of the eye diagrams for the 6.25 GHz channel
spacing with various design parameters and modulation types. A post symmetric
127
compensation technique with 3 EDFAs and a MDRZ line coding modulation scheme using MUX
BW = 3.125 GHz and DEMUX BW = 80 GHz was the most optimised network with highest signal
configuration, MDRZ (Avg. Q= 8.93) performed better than NRZ (Avg. Q = 4.04) and RZ (Avg.
Q = 4.19). The BER of MDRZ was lower than that of NRZ and RZ.
5.3.1 Comparison of Post Compensation Technique with 1 EDFA vs Post Symmetric

Compensation Technique with 3 EDFAs
The common parameters used for all networks compared in this section was that they all had
Bessel Filters in the MUX and DEMUX, all MUX’s had a Bandwidth of 80 GHz, DEMUX’s had a
Bandwidth of 80 GHz, and ER was 30 dB.
• For the 100 GHz channel spacing, using MUX BW = 80 GHZ and DEMUX BW = 80 GHZ
(See Table 5.2 for results):
 For the 48 Channel network with mismatched components, the Avg. Q-Factor of
NRZ = 10.97 and RZ = 4.77, which were both lower than that of the 48-channel
network with matched components (Avg. Q of NRZ = 13.88 and RZ = 18.83). This
means that overall signal performance of the matched network was better since it
can lower the effect of chromatic distortion.
 The highest Q factor when using 1 EDFA for RZ (= 16.50) was higher than NRZ (=
16.47). The min. BER when using 1 EDFA of RZ (= 1.63E-61) was lower than NRZ (=
2.67E-61). Similarly, the highest Q factor when using 3 EDFAs of RZ (= 27.59) was
higher than that of NRZ (= 18.99). The min. BER when using 3 EDFAs of RZ (= 5.82E-
168) was lower than that of NRZ (= 9.76E-81). The RZ modulation scheme when
using 3 EDFAs had the lowest BER. The min BER of the post symmetric
compensation technique with 3 EDFAs produced less bit errors than the post
compensation technique of using 1 EDFA for both NRZ and RZ. Furthermore from
Figure 5.1, it was noticed that the highest Q and Average Q of NRZ and RZ was
higher in the post symmetric compensation (3 EDFAs) than the post compensation
(1 EDFA) technique. The RZ modulation scheme when using 3 EDFAs had the best
overall signal performance since the Avg. Q of RZ was also the best overall.
128
 For the 48-channel network with matched components, when using 1 EDFA, the
Avg. Q factor of NRZ modulation scheme (= 12.77) was higher than that of RZ (=
11.76). Similarly, when using 3 EDFAs, the Avg. Q-Factor of 48 Channels of the RZ
modulation scheme (= 18.83) was higher than that of NRZ (= 13.88). The signal
performance of the post symmetric compensation technique with 3 EDFAs was
higher than the post compensation technique of using 1 EDFA for both NRZ and RZ.
Furthermore, it can be established that the RZ modulation scheme, when using 3
EDFAs, had the highest signal performance.
Figure 5.1: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 100GHz spacing
• For the 50 GHz channel spacing, using MUX BW = 80 GHZ and DEMUX BW =80 GHz
NRZ = 1.04 and RZ = 1.33, which were both lower than that of the 48-channel
means that overall signal performance of the matched network was better since it
can lower the effect of chromatic distortion.
However, further improvements were done on the matched network for it have an
acceptable Q > 6 and BER value > 10-9. This was done in section 5.3.2.
129
 The highest Q factor, when using 1 EDFA, of RZ (= 3.79) was higher than NRZ (=
3.08). Therefore, when using 1 EDFA, the min. BER of RZ (= 7.47E-05) was lower
than NRZ (= 1.00E-03). The RZ modulation scheme when using 1 EDFA had the
highest Q factor and lowest BER than the NRZ. This was because a decrease in
channel spacings increases the effects of non-linear losses and RZ was more robust
to the effects of non-linear losses than NRZ.
 The highest Q factor, when using 3 EDFAs, of RZ (= 2.86) was lower than that of
NRZ (= 3.14). Therefore, when using 3 EDFAs, the min. BER of RZ (= 2.07E-03) was
higher than that of NRZ (= 8.29E-04). The NRZ modulation scheme had the highest
Q factor and lowest BER when using 3 EDFAs than the RZ. This was because, when
the symmetric EDFAs were added, the effects of non-linear losses were minimised
significantly. This meant that dispersion was the only linear loss left in the network
and NRZ was more robust to impacts of chromatic dispersion than RZ. However,
these were just an indication of the individual signal performance of the channel
with the highest Q in both compensation techniques. The best overall performance
would be determined after considering the average Q of the system as well.
 For the matched 48-Channel network, when using 1 EDFA, the Avg. Q-Factor of the
NRZ modulation scheme (= 2.224) was higher than that of RZ (= 2.08). Similarly,
when using 3 EDFAs, the Avg. Q-Factor of the RZ modulation scheme (= 2.225) was
higher than that of NRZ (= 2.06). Therefore, the signal performance of the post
symmetric compensation technique with 3 EDFAs using the RZ modulation
technique was slightly higher than that of the post compensation technique with 1
EDFA. However, when considering the fact that the RZ modulation scheme using 3
EDFAs only had a slightly higher overall signal performance (avg. Q) than NRZ, as
could be seen from Figure 5.2, and NRZ showed a higher Q factor when compared
to RZ when using 3 EDFAs; the NRZ modulation scheme was believed to showcase
the best overall signal performance for the post symmetric compensation
technique.
130
• For the 25 GHz channel spacing, using MUX BW = 80 GHZ and DEMUX BW =80 GHz
NRZ = 2.27 and RZ = 2.50, which were both higher than that of the 48-channel
means that overall signal performance of the mismatched network was better
since at narrow channel spacings of 25 GHz, it increases the effect of chromatic
distortion due to an overlap in channel bandwidths. An increase in chromatic
dispersion in narrow channel spacings greatly reduces the effects of non-linear
losses such as FWM. Therefore, further improvements on the design parameters
of the matched network had to be done to produce a far greater signal
performance with an acceptable Q and BER. This was done in section 5.3.2.
3.11). The min. BER, when using 1 EDFA, of RZ (= 1.49E-04) was lower than NRZ (=
9.08E-04). And, the highest Q factor, when using 3 EDFAs, of RZ (= 3.66) was higher
than that of NRZ = 3.15. The min. BER, when using 3 EDFAs, of RZ (= 1.02E-04) was
lower than that of NRZ (= 7.76E-04). The RZ modulation scheme when using 3
EDFAs had the highest quality factor and the lowest BER. The highest quality factor
131
of the post symmetric compensation technique with 3 EDFAs was higher than that
of the post compensation technique of using 1 EDFA for RZ.
 The Avg. Q-Factor of the matched 48 Channels, when using 1 EDFA, for the NRZ
modulation scheme (= 2.16) was higher than that of RZ (= 1.69). Similarly, the Avg.
Q-Factor of 48 Channels, when using 3 EDFAs, of the RZ modulation scheme (=
1.86) was higher than that of NRZ (= 1.63). The NRZ modulation scheme had the
highest performance for system when using 1 EDFA and RZ had the highest
performance for a system when using 3 EDFAs, which can be seen in Figure 5.3.
However, since the RZ modulation scheme had the highest quality factor and the
lowest BER, the best overall signal performance was achieved using the RZ with the
post symmetric compensation technique.
132
• For the 12.5 GHz channel spacing, using MUX BW = 80 GHZ and DEMUX BW =80 GHz
NRZ = 2.56 and RZ = 2.77, which were both higher than that of the 48-channel
network with matched components (Avg. Q factor of NRZ = 1.51 and RZ = 1.65).
This means that overall signal performance of the mismatched network was better
since at narrow channel spacings of 12.5 GHz, it increases the effect of chromatic
distortion due to an overlap in channel bandwidths. An increase in chromatic
dispersion in narrow channel spacings greatly reduces the effects of non-linear
losses such as FWM. Therefore, further improvements on the design parameters
of the matched network had to be done to produce a far greater signal
performance with an acceptable Q and BER. This was done in section 5.3.2 and
section 0.
 The highest Q factor, when using 1 EDFA, of RZ (= 4.87) was higher than NRZ (= 3.43).
The min. BER, when using 1 EDFA, of RZ (= 5.46E-07) was lower than NRZ (= 2.90E-04).
When using 3 EDFAs, the highest Q factor of RZ (= 4.98) was higher than that of NRZ (=
3.65). The min. BER, when using 3 EDFAs, of RZ (= 3.17E-07) was lower than that of
NRZ (= 1.07E-04). The RZ modulation scheme when using 3 EDFAs had the highest
quality factor and the lowest BER. The highest quality factor of the post symmetric
compensation technique with 3 EDFAs was higher than that of the post compensation
technique of using 1 EDFA for RZ.
 For the 48 Channel matched network, when using 1 EDFA, Avg. Q-Factor of the NRZ
modulation scheme (= 1.66) was higher than that of RZ (= 1.17). Similarly, when using
3 EDFAs, the Avg. Q factor of the RZ modulation scheme (= 1.65) was higher than that
of NRZ (= 1.51). The NRZ modulation had the highest signal performance when using 1
EDFA and RZ had the highest signal performance when using 3 EDFAs, which can be
seen in Figure 5.4. However, since the RZ modulation scheme had the highest quality
factor and the lowest BER, the best overall signal performance was achieved using the
RZ format with the post symmetric compensation technique.
133
Figure 5.4: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 12.5 GHz spacing
• For the 6.25 GHz channel spacing, using MUX BW = 80 GHZ and DEMUX BW = 80 GHz
had NRZ = 3.329, which was higher than that of the NRZ (= 2.41) of the 48-channel
network with matched components. However, the Avg. Q-Factor of RZ (=1.868) of
the mismatched components was lower than that of the RZ (=2.14) of the 48-
channel network with matched components. This means that for very narrow
channel spacings of 6.25 GHZ, the overall signal performance of the NRZ
mismatched network was better due to NRZ being more robust to impacts of
chromatic dispersion as compared to RZ. At very narrow channel spacings for a
mismatched network, there was a massive overlap in channel bandwidths and
increased effect of chromatic distortion. An increase in chromatic dispersion in
narrow channel spacings greatly reduced the effects of non-linear losses such as
FWM. Therefore, further improvements on the design parameters of the matched
network had to be done to produce a far greater signal performance with an
acceptable Q and BER. This was done in section 5.3.2 and section 5.3.3.
3.85). However, when using 3 EDFAs, the highest Q factor of RZ (= 4.41) was higher
134
than that of NRZ (= 4.06). The RZ modulation scheme when using the post
symmetric compensation technique with 3 EDFAs had the highest quality factor
than that of the post compensation technique of using 1 EDFA for both NRZ and
RZ.
 For the 48 Channels matched network, when using 1 EDFA, the Avg. Q-Factor of
the NRZ modulation scheme (= 1.79) was higher than that of RZ (= 1.51). However,
when using 3 EDFAs, the Avg. Q-Factor of the RZ modulation scheme (= 2.14) was
lower than that of NRZ (= 2.41). Both the NRZ and RZ modulation schemes had avg.
Q factors that showed the highest signal performance in the post symmetric
compensation technique with 3 EDFAs than the post compensation technique of
using 1 EDFAs for both NRZ and RZ, as can be seen in Figure 5.5.
Figure 5.5: Comparison of Signal Performance of 1 EDFA vs 3 EDFAs for 6.25 GHz spacing
135
5.3.2 Optimisation of DWDM System by varying different parameters for RZ and
NRZ Modulation formats
The common parameters used for all networks in this section was the 2nd order Bessel Filters
in the MUX and DEMUX, a Post Symmetric Compensation Technique with 3 EDFAs in the
optical link and an ER of 30 dB at the WDM transmitter.
• For the optimisation of the 100 GHz channel spacing, see Table 5.2 and Figure 5.6:
 The Avg. Q-Factor of 48 Channels (Bandwidth Efficiency) of the following networks
were compared:
 MUX BW = 80 GHz and DEMUX BW = 80 GHz,
 MUX BW = 80 GHz and DEMUX BW = 50 GHz.
 There was a definite improvement when decreasing the DEMUX BW from 80 GHZ
to 50 GHz and not much improvement when decreasing the MUX BW from 80 GHZ
to 50 GHz.
 When MUX BW = 80 GHz and DEMUX BW = 80 GHz was used, the Avg. Q-
Factor of 48 Channels of the RZ modulation scheme (= 18.83) was higher
than that of NRZ (= 13.88).
 When MUX BW = 50 GHz and DEMUX BW = 80 GHz was used, the Avg. Q-
Factor of 48 Channels of the RZ modulation scheme (= 24.67) was higher
than that of NRZ (= 13.52).
 Similarly, when using MUX BW = 80 GHz and DEMUX BW = 50 GHz, the Avg.
Q-Factor of 48 Channels of the RZ modulation scheme (= 53.37) was higher
than that of NRZ (= 40.68). Therefore, the RZ modulation scheme when
using MUX BW = 80 GHz and DEMUX BW = 50 GHz, had the highest signal
performance.
 There was definite improvement in the quality factor of NRZ and RZ modulation
scheme when decreasing the DEMUX BW from 80 GHz to 50 GHz, and finally to 25
GHz. However, there was not much improvement in quality factor when decreasing
MUX BW from 80 GHz to 50 GHz.
 The highest Q factor of RZ (= 27.59), when using MUX BW = 80 GHz and
DEMUX BW = 80 GHz, is higher than the Q of NRZ (= 18.99).
136
 Similarly, the highest Q factor of RZ (= 59.89), when using MUX BW = 50
GHz and DEMUX BW = 80 GHz, is higher than the Q of NRZ (= 20.66).
 The RZ modulation scheme had the highest Q factor (=82.6), when using
MUX BW = 80 GHz and DEMUX BW = 50 GHz, which was higher than the Q
of NRZ (= 54.32) for that same bandwidth configuration. This bandwidth
configuration produced the best QoS.
 The min. BER of both RZ and NRZ (= 0), when using MUX BW = 80 GHz and DEMUX
BW = 50 GHz. However, there is a min. BER of NRZ (= 3.30E-95) and RZ (= 0) when
using MUX BW = 50 GHz and DEMUX BW = 80. The min BER for both NRZ and RZ,
when using MUX BW = 80 GHz and DEMUX BW = 50 GHz, was the lowest compared
to all other bandwidth configurations.
 For the 100 GHz channel spacing, with the NRZ and RZ modulation scheme, the
signal quality was already of an acceptable level; however, when the MUX and
DEMUX bandwidth was reduced 50 GHz, the system performance was significantly
improved.
 Nevertheless, the RZ modulation scheme, when using MUX BW = 80 GHz and
DEMUX BW = 50 GHz, had the highest bandwidth efficiency (=100 %) and the best
overall signal performance for a channel spacing of 100 GHZ with a Q > 6 and a BER
= 0 is greater than 10-9. This is illustrated in Figure 5.6.
Figure 5.6: Comparison of Signal Performance of 100GHz spacing for Various MUX/DEMUX BW Configurations
137
 The Avg. Q-Factor for 48 Channels of the following networks were compared:
 MUX BW = 80 GHz and DEMUX BW = 25 GHz
 There was definite improvement in bandwidth efficiency and signal performance
in NRZ and RZ modulation scheme when decreasing the DEMUX BW from 80 GHz
to 50 GHz, and finally to 25 GHz. However, there was not much improvement in
bandwidth efficiency when decreasing MUX BW from 80 GHz to 50 GHz.
 When using MUX BW = 80 GHz and DEMUX BW = 80 GHz, the Avg. Q-Factor
of NRZ (= 2.06) was lower than that of RZ (= 2.22).
of NRZ (= 7.03) was higher than that of RZ (= 3.56).
 Finally, when using MUX BW = 80 GHz and DEMUX BW = 25 GHz, the Avg.
Q-Factor of NRZ (= 8.72) was lower than that of RZ (= 10.02).
 There was definite improvement in the quality factor in NRZ and RZ modulation
scheme when decreasing the DEMUX BW from 80 GHz to 50 GHz, and finally to 25
GHz. However, there was not much improvement in the quality factor when
decreasing MUX BW from 80 GHz to 50 GHz.
 When using MUX BW = 80 GHz and DEMUX BW = 80 GHz, the highest Q
factor of NRZ (= 3.14) was higher than the Q of RZ (= 2.86).
 Similarly, when using MUX BW = 50 GHz and DEMUX BW = 80 GHz, the
highest Q factor of RZ (= 5.11) was higher than the Q of NRZ (= 3.71).
factor of NRZ (= 9.58) was higher than the Q of RZ (= 6.31).
 Finally, when using MUX BW = 80 GHz and DEMUX BW = 25 GHz, the highest
Q factor of RZ (= 26.53) was higher than the Q of NRZ (= 9.93).
 The min. BER of both RZ and NRZ (= 0) when using MUX BW = 80 GHz and DEMUX
BW = 50 GHz. However, when using MUX BW = 50 GHz and DEMUX BW = 80 GHz,
138
the min. BER of NRZ = 3.30E-95 and RZ = 0. The min BER for both NRZ and RZ, when
using MUX BW = 80 GHz and DEMUX BW = 50 GHz, was comparatively lower than
when using MUX BW = 50 GHz and DEMUX BW = 80 GHz.
 For the 50 GHz channel spacing, the DEMUX bandwidth had to be reduced to 25
GHz to achieve an acceptable system performance. BW Efficiency was always
maintained at 100%.
 Therefore, the RZ modulation scheme, when using MUX BW = 80 GHz and DEMUX
BW = 25 GHz, had the best overall signal performance for a channel spacing of 50
GHZ with the highest Q factor being > 6 and lowest BER > 10-9. This is illustrated in
Figure 5.7.
139
 The Avg. Q-Factor for 48 Channels of the following networks were first compared:
 MUX BW = 80 GHz and DEMUX BW = 12.5 GHz,
 MUX BW = 12.5 GHz and DEMUX BW = 80 GHz,
 The trial to improve the bandwidth efficiency and signal performance in NRZ and
RZ modulation scheme was done by decreasing the MUX and DEMUX BW from 80
GHz to 12.5 GHz.
of NRZ (= 2.41) was higher than that of RZ (= 2.24). This bandwidth
configuration showed a slight improvement in signal performance however
it was still not anywhere near the acceptable value.
 When using MUX BW = 80 GHz and DEMUX BW = 12.5 GHz, the Avg. Q-
Factor of NRZ (= 2.28) was higher than that of RZ (= 2.26).
 When using MUX BW = 12.5 GHz and DEMUX BW = 80 GHz, the Avg. Q-
Factor of NRZ (= 0.98) was lower than that of RZ (= 1.51).
 Finally, when using MUX BW = 25 GHz and DEMUX BW = 25 GHz, the Avg.
Q-Factor of NRZ (= 2.09) was lower than that of RZ (= 3.13).
 There was definite improvement in the quality factor and BER in NRZ and RZ
modulation scheme when decreasing the DEMUX BW from 80 GHz to 25 GHz, and
finally the MUX BW to also 25 GHz. However, there was no improvement when
decreasing to 12.5 GHz.
factor of NRZ (= 4.93) was lower than the Q of RZ (= 5.9). The BER of NRZ
(3.68E-07) is higher than that of RZ (1.52E-09).
140
 Similarly, when using MUX BW = 25 GHz and DEMUX BW = 25 GHz, the
highest Q factor of RZ (= 4.84) was higher than the Q of NRZ (= 3.73). The
BER of RZ (6.03E-07 is lower than the BER of NRZ (7.37E-05).
 In order to try improve the system performance further, the extinction ratio of the
WDM transmitter was decreased and a rectangular filter was used in the MUX and
DEMUX instead of the Bessel filter. There was a definite improvement in
bandwidth efficiency and signal performance in NRZ and RZ modulation scheme
when decreasing the extinction ratio from 30 dB to 5dB and changing the filter type
from a Bessel filter to a rectangular filter, as can be seen in Figure 5.9.
 When using MUX BW = 80 GHz and DEMUX BW = 25 GHz while using a 5dB
ER with rectangular filer, the Avg. Q-Factor of NRZ (= 2.56) is higher than
that of RZ (= 1.39). The highest Q for NRZ (= 4.56), was higher than the
highest Q for RZ (= 2.9). However, all these values were not acceptable since
Q < 6.
 However, when using MUX BW = 25 GHz and DEMUX BW = 25 GHz while
using a 5dB ER with rectangular filer, the Avg. Q-Factor of NRZ (= 5.83) was
141
lower than that of RZ (= 9.59). The highest Q factor for NRZ (=6.51) was
lower than that of RZ (=13.55). The BER of RZ (=3.54E-42) was lower than
that of NRZ (=3.54E-11). Both signals were acceptable signals since Q > 6
and BER > 10-9.
 It is therefore established that the RZ modulation scheme, when using MUX BW =
25 GHz and DEMUX BW = 25 GHz, had the best bandwidth efficiency (=100%) and
the best signal performance for a channel spacing of 25 GHZ when using a 5dB
extinction ratio in combination with a rectangular filter.
• For the optimisation of the 12.5 GHz channel spacing, see Table 5.5 and Figure 5.10:
 MUX BW = 3.125 GHz and DEMUX BW = 80 GHz.
142
GHz to 3.125 GHz.
 Finally, when using MUX BW = 3.125 GHz and DEMUX BW = 80 GHz, the
Avg. Q-Factor of NRZ (= 3.06) was lower than that of RZ (= 3.24). This
bandwidth configuration showed a slight improvement in signal
performance however it was still not anywhere near the acceptable value.
 There was definite improvement in the quality factor and BER in NRZ and RZ
modulation scheme when decreasing the MUX BW from 80 GHz to 3.125 GHz while
maintaining the DEMUX BW at 80 GHz. However, there was no improvement when
decreasing DEMUX BW to 6.25 GHz.
 When using MUX BW = 3.125 GHz and DEMUX BW = 80 GHz, the highest Q
factor of NRZ (= 4) was lower than the highest Q of RZ (= 4.12). The BER of
NRZ (2.54E-05) is higher than that of RZ (1.55E-05)
 In order to try improve the system performance further, an advanced modulation

method (MDRZ) was used. The optimisation of the 12.5 GHz channel spacing by
using the MDRZ modulation format was done in section 5.3.3.
143
Figure 5.10: Comparison of Signal Performance of 12.5GHz spacing for Various MUX/DEMUX BW Configurations
• For the optimisation of the 6.25 GHz channel spacing, see Table 5.6 and Figure 5.11:
 MUX BW = 3.125 GHz and DEMUX BW = 80 GHz.
GHz to 3.125 GHz.
 Finally, when using MUX BW = 3.125 GHz and DEMUX BW = 80 GHz, the
Avg. Q-Factor of NRZ (= 4.04) was lower than that of RZ (= 4.19). This
144
showed a slight improvement in signal performance however it was still not
acceptable.
 There was definite improvement in the overall performance (Avg. Q) of the system
when decreasing the MUX BW from 80 GHz to 3.125 GHz while maintaining the
DEMUX BW at 80 GHz.
 When using MUX BW = 3.125 GHz and DEMUX BW = 80 GHz, the highest Q
factor of NRZ (= 4.63) was lower than the highest Q of RZ (= 4.7). The BER
of NRZ (6.26E-07) is higher than that of RZ (4.56E-07)
 In order to try improve the system performance further, an advanced modulation

method (MDRZ) was used. The optimisation of the 6.25 GHz channel spacing by
using the MDRZ modulation format was done in section 5.3.3.
Figure 5.11: Comparison of Signal Performance of 6.25GHz spacing for Various MUX/DEMUX BW Configurations
145
5.3.3 Optimisation of 12.5 GHz and 6.25 GHz Channel spacing using MDRZ
The optimisation of both the 12.5 GHz spacing and the 6.25 GHz spacing was done using the
advanced modulation format, MDRZ. The performance of both the above channel spacing was
done using the filter bandwidth parameter of MUX BW = 3.125 GHz and DEMUX BW = 80 GHz.
This was because, this parameter showed the best performance in comparison to the other
filter bandwidth configurations discussed in section 5.3.2.
For the 12.5 GHz spacing: MDRZ had the highest Average Q (= 8.66) and the highest Q (=
12.77), which were higher when compared to the average Q of NRZ (= 3.06), highest Q of NRZ
(= 4), average Q of RZ (= 3.24) and the highest Q of RZ (= 4.12). The BER of the MDRZ signal (=
7.49E-38) was lower than the BERs of the NRZ (= 2.54E-05) and RZ (= 1.55E-05). The
comparison is shown in Figure 5.12.
For the 6.25 GHz spacing: MDRZ had the highest Average Q (= 8.93) and the highest Q (=
10.79), which were higher when compared to the average Q of NRZ (= 4.04), highest Q of NRZ
(= 4.63), average Q of RZ (= 4.19) and the highest Q of RZ (= 4.7). The BER of the MDRZ signal
was 7.17E-28, which was lower the NRZ and RZ BERs of 6.26E-07 and 4.52E-07 respectively.
The comparison is shown in Figure 5.13.
The MDRZ format provided an acceptable signal that had a Q > 6 and BER > 10-9 for both 12.5
GHz and 6.25 GHz channel spacing. Therefore, it was established that the MDRZ modulation
scheme had the best bandwidth efficiency (=100%) and the best signal performance when
using MUX BW = 3.125 GHz and DEMUX BW = 80 GHz for both channel spacings of 12.5 GHZ
and 6.25 GHZ. This was because the MDRZ format had a much narrower optical bandwidth
over the NRZ and RZ formats. This led to a greater chromatic dispersion tolerance, higher fiber
non-linearity tolerance and thereby achieved an acceptable signal transmission over longer
distances.
146
Figure 5.12: Comparison of Signal Performance of 12.5GHz spacing using various Line Coding Modulation Formats
Figure 5.13: Comparison of Signal Performance of 6.25GHz spacing using various Line Coding Modulation Formats
147
5.4 Summary
The simulation, design and comparative analysis of the 48-channel DWDM system with a 25
GB/s bit rate was performed using the Optisystem simulation platform. The optimised 48
channel DWDM network models produced point to point communication link with minimal
bit errors for channel spacings of 100 GHz, 50 GHz, 25 GHZ and 12.5 GHz and 6.25 GHz. In this
chapter, the analysis and discussion of the various design parameters used to optimise the
network was done to validate the best configuration that provided an acceptable system
performance for future growth in end-user requirements.
Networks with matched active components (WDM transmitter and WDM MUX) had equally
spaced frequency channels that were aligned to each other and had a higher OSNR than
mismatched networks. This was because mismatched networks with misaligned frequency
channels increased the levels of linear losses (CD) and this level increased further for narrower
channel spacings. Matched networks minimised the effects of chromatic dispersion for wide
channel spacings. However, unmodified mismatched networks had a better signal
performance at smaller channel spacings because of their increased effects of CD. This
minimised the effects of non-linear losses such as FWM.
The maximum signal power and OSNR of the 3 EDFA-post Symmetric compensation technique
was always higher than the 1 EDFA-post compensation technique for all channel spacings in
any type of network. The addition of the 3 symmetric EDFAs in all networks compensated for
the increased attenuation and dispersion introduced when the bit rate was increased, number
of channels were increased and the channel spacings were decreased. The 3 EDFA-post
symmetric compensation technique produced less bit errors than the 1 EDFA-post
compensation technique since it minimised the effects of CD and FWM to some extent.
However, this could only be achieved when used in combination with the NRZ and RZ
modulation formats.
NRZ was more robust to impacts of CD as compared to RZ. Therefore, it had a greater signal
performance for wider channel spacing that had a high amount of CD and a low amount of
FWM harmonic components. RZ was more robust to the effects of non-linear losses than NRZ.
148
This meant that the RZ modulation scheme produced a higher overall signal performance at
smaller channel spacings that had a larger amount of FWM harmonic components.
However, the capabilities of NRZ and RZ were limited; because they minimised the effects of
CD and FWM respectively but they compromised on signal quality and BER at lower channel
spacings. For this purpose, the signal performance of the 48 channel DWDM network was
optimised for each channel spacing using various design parameters that produced networks
that had the least amount of bit errors for the optical receiver to detect and decode any signal
in a channel successfully.
The MDRZ modulation format implemented FEC at very narrow bandwidths and at very
narrow channel spacings. This led to a greater CD tolerance, higher fiber non-linearity
tolerance and achieved an acceptable signal transmission over longer distances with the least
amount of errors.
The decreased extinction ratio of the transmitter in combination with a rectangular filter
increased signal power and dispersion, which subsequently decreased the effects of FWM.
Simultaneously, the rectangular filter eliminated low power frequency harmonics of FWM
without affecting the lower harmonics and signal propagation over the length of fiber in a
DWDM system, which also caused dispersion.
Varying the filter bandwidth configurations of the MUX and DEMUX decreased the finite
passband of a 2nd order Bessel filter, which meant that all FWM harmonic products outside
this band is eliminated or the power of the harmonic components was decreased significantly.
The signal performance was evaluated using eye diagrams that analysed the Q factor and BER.
For each of the optimised parameters below, the signal detected by the receiver had a Q factor
> 6 and the BER > 10-9. The optimised parameter designs and configurations for each channel
spacing that had the highest signal performance and bandwidth efficiency were as follows:
• For 100 GHz: The RZ format with MUX BW = 80 GHz and DEMUX BW = 25 GHz produced
an avg. Q = 53.37, highest Q= 82.60 and BER = 0.
149
• For 50 GHz: The RZ format with MUX BW = 80 GHz and DEMUX BW = 25 GHz produced
an avg. Q = 10.02, highest Q = 26.53 and BER = 1.41 x 10-155.
• For 25 GHz: The RZ format with MUX BW = 25 GHz and DEMUX BW = 25 GHz; used
together with a 5dB extinction ratio in combination with a rectangular filter produced
an avg. Q = 9.59, highest Q = 13.55 and BER = 3.54 x 10-42.
• For 12.5 GHz: The MDRZ format with MUX BW = 3.125 GHz and DEMUX BW = 80 GHz
produced an avg. Q = 8., highest Q = 12.77, and BER = 7.49 x 10-38.
• 6.25 GHz: The MDRZ format with MX BW = 3.125 GHz and DEMUX BW = 80 GHz
produced an avg. Q = 8.93, highest Q = 10.79, and BER = 7.17 x 10-28.
150
CONCLUSION AND RECOMMENDATIONS
6.1 Introduction
This dissertation focused on achieving an acceptable signal performance for a 48 channel

DWDM network model with a 25 Gb/s bit rate and a 100 km transmission distance for channel
spacing of 100 GHz, 50 GHz, 25 GHz, 12.5 GHz and 6.25 GHz. This chapter concludes all the
summaries and lessons learnt from Chapter 1 to Chapter 5. It also gives possible
recommendations of how to decrease non-linear effects to further improve the current
DWDM network design. Furthermore, an overview of the current maximum bandwidth
potential of the designed DWDM network model is given with an oversight into future
applications of the design.
6.2 Chapter Conclusions
In Chapter 1, a brief overview of some of the challenges faced in an optical communication

networks due to future bandwidth requirements was enumerated. The motivation to this
dissertation established that the usage of DWDM technology was an attractive strategy to
meet the challenges of current optical communication systems and thereby increase the
capacity of future optical networks. The objectives of this dissertation focused essentially on
creating a pilot network that was be able to test the limitations of the untapped channel
spacing of 6.25 GHz to further expand the channel capacity of the ITU frequency grid.
In Chapter 2, the literature review explored the performance limitations of DWDM technology
and it was established that CD and FWM were the most critical limiting factors that
deteriorates a DWDM signal. The various design components of the DWDM system, their
properties, functionalities, limitations and benefits were discussed with the focus being on the
optimisation of a 48 channel DWDM system. The theoretical design method of designing and
measuring the performance of the optical transmission link was also enumerated.
In Chapter 3, the simulation and design process of the 48-channel DWDM system design was
shown using the Optisystem simulation platform. The various design parameters used to
model and optimise the existing 48 channel fiber optic network was calculated and
151
enumerated such that an acceptable system performance was achieved for future growth in
end-user requirements. The optimisation of the 48-channel network was done using various
parameters in different combinations. Three of the line encoding techniques (NRZ, RZ and
MDRZ) were used in numerous DWDM networks with different filter types (Bessel and
rectangular) in the MUX and DEMUX to minimise the effects of linear and non-linear losses.
The BW of the filter used in the MUX/DEMUX and the extinction ratio of the transmitter were
also varied to obtain an optimised network that had the most acceptable signal performance
with the least amount of non-linearities for each of the channel spacings. Furthermore, a
comparative model of two dispersion compensation techniques (1 EDFA post compensation
and the 3 EDFA post symmetric compensation) was also done for each channel spacing to
evaluate a reduction in linear effects (CD) and non-linear effects (FWM).
In chapter 4, the eye-diagram results of each channel in the DWDM system was illustrated and
analysed. The results showed that since BER was inversely proportional to the Q factor, the
channels with the highest Q factor had the lowest BER and vice versa was also true. The
channels with the highest Q was used for analysis for all networks to compare the maximum
signal performance of that network to the other networks. The BW efficiency of each network
was calculated to have an indication of the overall signal performance for each type of
network. The minimum BER required for an acceptable signal was greater than 10-9 for the
optical receiver to detect the information signal with the least amount of bit errors. The Q
factors for such signals was greater than 6.
In Chapter 5, the analysis and discussion of the various design parameters, used to optimise
the DWDM network models, validated the best system configuration that could provide an
acceptable signal quality for future growth in end-user requirements. It was established that
networks that have matched active components with equally spaced frequency channels
minimise the effects of CD at wide channel spacings and therefore have a higher OSNR than
mismatched networks. The maximum signal power and OSNR of the 3 EDFA-post symmetric
compensation technique is always higher than the 1 EDFA-post compensation technique for
all channel spacings in any type of network. NRZ is more robust to the effects of linear losses
(CD) as compared to RZ and has a greater signal performance for wider channel spacing. RZ is
more robust to the effects of non-linear losses (FWM) than NRZ and produces a higher overall
signal performance at smaller channel spacings. The MDRZ modulation format implements
152
FEC at very narrow bandwidths and at very narrow channel spacings. This leads to a greater
CD tolerance and a higher fiber non-linearity for MDRZ than for NRZ and RZ. The usage of a
rectangular filter, a decrease in the ER of the transmitter and a decrease in the BW of the
MUX/DEMUX filter are the parameter variances that substantially decreased the effects of
FWM. The optimised parameter designs and configurations for each channel spacing has the
highest signal performance and BW efficiency for the RZ modulation format at 100 GHz, 50
GHz and 25 GHz channel spacing. Similarly, the MDRZ has the highest signal performance and
BW efficiency for channel spacings of 12.5 GHz and 6.25 GHz.
6.3 Recommendations
FWM could be further reduced in future applications without varying the bandwidth filter.
• One possible way is to use irregular channel spacings in the design to decrease the
possibility of FWM forming multiple harmonic products [3], [6].
• The second possibility is to either lower the signal power in the fiber or design the
non-zero dispersion shifted fiber to have a larger cross-sectional area. This reduces the
signal power density and should avoid FWM cross-talk interference [3], [6].
• Lastly, the use of polarisation-multiplexed DWDM channels to polarise DWDM
channels is another possibility that may suppress FWM harmonics and reduce FWM
component power and cross talk [3], [6]. This is because each channel’s polarisation
state would be in an orthogonal state to the adjacent channel’s polarisation state.
6.4 Current Potential of Designed DWDM Network Model and Future Work
The maximum bit rate capacity of the DWDM network model designed in this dissertation is
1200 Gb/s or 1.2 Tb/s. This is because each channel in the 48 channel DWDM system has a bit
rate of 25 Gb/s. Furthermore, the total channel bandwidth utilised by the designed model in
the optical spectrum is only 300 GHz (2.4 nm) because each channel of the 48 channel DWDM
system used a channel spacing of 6.25 GHz (0.05 nm). This meant that for the 6.25 GHz
channels spacing, the 48 channel DWDM system only uses 2.4 nm of wavelength from the C
Band’s (1530nm to 1565nm) maximum range of 35 nm [3], [6]. Therefore, for the 6.25 GHz
channel spacing in the C band, the current 25 Gb/s bit rate per channel designed model has
the potential of providing good signal quality performance for a maximum transmission
capability of 700 DWDM channels with a total bit rate capacity of 17.5 Tb/s over a single fiber
153
core. This DWDM model should be currently sufficient to manage the demands for high data
rate applications and provide reasonable flexibility for broadband communication.
This flexibility should also open doors for the application of DWDM in possible radio-over-fiber
(RoF) systems, which is the future of broadband wireless communication systems such as
mobile communications and hotspots in suburban areas [51], [52]. The use of DWDM in RoF
systems is advantageous because it simplifies the network topology by allowing the potential
to allocate different wavelengths to individual base stations without changing the existing
fiber infrastructure. This provides simpler network management and enables easier network
or service upgrades [51].
154
REFERENCES
1. J.Z. Mohan and O. A. Sokoya, "Comparison of NRZ and RZ in a 32x5Gbps and a 32x10Gbps
DWDM Networks for 100km and 500km," Proceedings of The Southern Africa
Telecommunication Networks and Applications Conference (SATNAC), Hermanus, 2018,
pp. 142-148
2. J.Z. Mohan and O. A. Sokoya, "Performance Analysis of a 48 x 25Gb/s DWDM system at
Various Channel Spacings with NRZ and RZ," Proceedings of The Southern Africa
Telecommunication Networks and Applications Conference (SATNAC), Ballito, 2019, pp.
172-177
3. Introduction to DWDM Technology, San Jose: Cisco Systems, Inc., 2001.
4. C. A. Coombs and C. F. Coombs Jr., Communications Network Test and Measurement
Handbook, New York: McGraw-Hill, 1998.
5. L. Goleniewski and K. W. Jarrett, Telecommunications Essentials: Traditional Transmission
Media for Networking and Telecommunications, 2nd Ed. Stoughton: Addison-Wesley
Professional, 2006, pp. 41-82.
6. J. Laude, DWDM Fundamentals, Components, and Applications, Boston: Artech House
optoelectronics library, 2002.
7. O. Babatunde and S. Mbarouk, “A review of Plesiochronous Digital Hierarchy (PDH) and
Synchronous Digital Hierarchy (SDH),” International Journal of Scientific Research
Engineering & Technology (IJSRET), Vol. 3, No. 3, 2014.
8. Recommendation ITU-T G.702 (1988), Digital Hierarchy Bit Rates.
9. J. Laferrière, G. Lietaert, R. Taws and S. Wolszczak, Reference Guide to Fiber Optic Testing,
2nd Ed. JDS Uniphase Corporation, Vol. 1, 2011
10. Recommendation ITU-T G.655 A.8 (2009), Characteristics of a non-zero dispersion-shifted
single-mode optical fibre and cable.
11. Prabir Kumar Sethy, Kamal Pradhan and Dr. Madhumita Panda, “Dense Wavelength
Division Multiplexing (DWDM): A Review,” International Journal of Innovative Research in
Advanced Engineering (IJIRAE), Vol. 1, No. 8, 2014.
12. Duha S. Ahmed and Raad S. Fyath, “Effect of Four-Wave Mixing on the Transmission
Performance of O-Band Multichannel PAM-Based Unamplified Datacenter
155
Interconnects,” International Journal of Networks and Communications, Vol. 8, No. 3, pp
63-80, 2018.
13. S. J. Subash and J. S. Babu, “Application Specific DWDM System Optimization for High
Speed Transmission", International Journal of Engineering Research & Technology (IJERT),
vol. 3, no. 4, pp. 2278-0181, April 2014.
14. K. S. Bharath and N. Jayaraj, “Network Planning and Engineering for Fiber Optic Transport
Systems”, International Journal of Innovative Research in Computer and Communication
Engineering (IJIRCCE), vol. 2, no. 5, pp. 2320-9801, May 2014.
15. M. Borthakur, “A Survey of DWDM Networks, its Development and Future Scope in
Telecommunication Domain”, International Journal of Innovative Research in Computer
and Communication Engineering (IJIRCCE), vol. 3, no. 8, pp. 2320-9798, August 2015.
16. D. Kaur and V. Sharma, “Review of DWDM Technology in Optical Communication”,
American International Journal of Research in Science, Technology, Engineering &
Mathematics (AIJRSTEM), vol. 7, no. 1, pp. 80-84, June-August 2014.
17. R. Antil, Pinki and S. Beniwal, “An Overview of DWDM Technology & Network”,
International Journal of Scientific & Technology Research (IJSTR), vol. 1, no. 11, pp. 2277-
8616, December 2012.
18. S. M. Sajjan V. Seshasai, and G. Sadashivappa, “DWDM Link Design and Power Budget
Calculation”, International Journal of Advanced Research in Electrical, Electronics and
Instrumentation Engineering (IJAREEIE), vol. 4, no. 4, pp. 2278 – 8875, April 2015.
19. T. Saxena, and H. Saxena, “Optical Power Debugging In DWDM System having Fixed Gain
Amplifiers”, International Journal of Research in Engineering and Technology (IJRET), vol.
4, no. 2, pp. 2321-7308, Feb 2015.
20. B. D. Moses and G. B. Lakshmy, “Performance Evaluation of Optical Amplifiers in Optical
Communication”, International Journal of Engineering Research & Technology (IJERT),
vol. 3, no. 4, pp. 2278-0181, April 2014.
21. M Kaur, and H. Sarangal, “Simulative Investigation of 32x10, 32x20 and 32x40 Gb/s
DWDM Systems with Dispersion Compensating Fibers”, International Journal of Signal
Processing, Image Processing and Pattern Recognition (IJSIP), vol. 8, no. 8, pp.127-134,
2015.
156
22. D. Pal and S. Revathi, “Analysis of 32-channel 40 Gbit/s DWDM in Point-to-Point
topology”, International Journal on Recent and Innovation Trends in Computing and
Communication (IJRITCC), vol. 4, no. 3, pp. 2321-8169, March 2016.
23. H. L. Cho and W. Kim, “Simulation and optimization of the performance of 32 channel
DWDM system”, International Journal of Photonics and Optical Technology (IJPOT), vol. 1,
no. 1, pp. 1-3, December 2015.
24. N. Chaudhary, A. Kumar, P. Upadhaye and G. Kaur, “Evaluating the 32*45Gbps
Transmission over SMF link Based on DWDM/EDFA Architecture”, International Journal
of Emerging Technology and Advanced Engineering (IJETAE), vol. 4, no. 1, pp. 2250-2459,
February 2014.
25. V. Senthamizhselvan, R. Ramachandran and R. Rajasekar, “Performance Analysis of
DWDM Based Fiber Optic Communication with Different Modulation Schemes and
Dispersion Compensation Fiber”, International Journal of Research in Engineering and
Technology (IJRET), vol. 3, no. 3, pp. 2319-1163, March 2014.
26. M. Singh and R. Bhagat, “DWDM based Optical Fiber Communication Link using Different
Modulation Formats and Dispersion Compensating Fiber”, An International Journal of
Engineering Sciences (IJOES), vol. 17, pp. 2229-6913, January 2016.
27. C. kaur , Sonia and H. Singh, “Comparative Analysis of DWDM System Using OADM
(Optical Add/Drop Multiplexer) At Different Data Rate, Distance and Channel Spacing”,
International Journal of Engineering Development and Research (IJEDR), vol. 4, no. 2, pp.
2321-9939, 2016.
28. S.Rajalakshmi, N. Baid and V. Charan, “Characteristic Analysis of Dense WDM for Long
Haul Optical Networks”, International Journal of Advanced Scientific and Technical
Research (IJST), vol. 3, no. 4, pp. 2249-9954, May-June 2014.
29. J. Choudhary, L. S. Garia and R. S. Shahi, “Comparative analysis of DWDM system using
different modulation and dispersion compensation techniques at different bit rates”,
International Journal of Advanced Research in Computer and Communication Engineering
(IJARCCE), vol. 3, no. 5, pp. 2319-5940, May 2014.
30. OptiSys_Design Component Library, Ottawa: Optiwave Corporation, Vol. 1, 2001.
31. T. G. B. Mason, D. E. Kumar and G. Giaretta, “Laser driver circuit for externally modulated
lasers”, United States Patent 7194012B2, Mar. 20, 2007.
157
32. Vitex LLC, Difference between DML and EML lasers, Vitex LLC, Englewood Cliffs, NJ.
Accessed on: Oct. 30, 2017.[Online]. Available: http://vitextech.com/difference-dml-eml-
lasers/
33. Recommendation ITU-T G.959.1 (02/2012), Optical transport network physical layer
interfaces.
34. P. Chen, S. Feng, L. and S. Li, “Performance of Optical Network with Lower Bit Error Rate
(BER)”, The 2012 2nd International Conference on Circuits, System and Simulation (ICCSS
2012), Singapore, IPCSIT vol. 46, 2012.
35. B. Chomycz, Planning Fiber Optic Networks, The McGraw-Hill Companies Inc.,2009.
36. B. H. Marcus and P. H. Siegel, “On Codes with Spectral Nulls at Rational Submultiples of
the Symbol Frequency”, IEEE Transactions on Information Theory, Vol. IT-33, No. 4, 1987.
37. V. S. Latha, “Performance Evaluation of Different Line Codes, Indian Journal of Computer
Science and Engineering (IJCSE)”, Vol. 2 No. 4, 2011.
38. L. Sharan, A. G. Shanbhag and V.K. Chaubey, “Design and simulation of modified duobinary
modulated 40 Gbps 32 channel DWDM optical link for improved non-linear performance”,
Cogent Engineering (2016), Vol. 3, No. 1256562, 2016.
39. S. M. J. Alam, M. R. Alam, G. Hu, and Z. Mehrab, “Bit Error Rate Optimization in Fiber Optic
Communications”, International Journal of Machine Learning and Computing, Vol. 1, No.
5, 2011.
40. R. Singh, M. Ahlawat and D. Sharma, “A Review on Radio over Fiber communication
System”, International Journal of All Research Education and Scientific Methods
(IJARESM), Vol. 5, No. 4, 2017.
41. A. Goyanka, K. Gupta and T. Mukhopadhyay, Performance Analysis of Electrical Filters in
Dense Wavelength Division Multiplexing Based Passive Optical Networks at Various Data
Rates, VIT University, Vellore, Tamil Nadu. Accessed on: June 13, 2018. [Online].
Available:
https://www.academia.edu/6383312/Performance_Analysis_of_Electrical_Filters_in_De
nse_Wavelength_Division_Multiplexing_based_Passive_Optical_Networks_at_various
42. G. Fizza, A. Hussain, and S. H. Jokhio, “Experimental Analysis and Reduction of FWM Using
Optical Rectangle Filter for WDM”, Mehran University Research Journal of Engineering &
Technology, Vol. 35, No. 1, 2016.
158
43. N. M. Nawawi et al., “Dispersion compensation dense wavelength division multiplexing
(DC DWDM) for nonlinearity analysis at various propagation distance and input power”,
2015 International Conference on Computer, Communications, and Control Technology
(I4CT), Kuching, pp. 346-349, 2015.
44. Recommendation ITU-T G.652 (11/2016), Characteristics of a single-mode optical fibre
and cable.
45. Recommendation ITU-T G.653 (07/2010), Characteristics of a dispersion-shifted, single-
mode optical fibre and cable
46. A. C. Singer, N. R. Shanbhag, and H. M. Bae, Electronic Dispersion Compensation, IEEE
Signal Processing Magazine, University of Illinois, 2008
47. W. Ahmad, M. U. Ali, V. Laxmi and A. S. Syed, “Simulation and Characterization of PIN
Photodiode for Photonic Applications”, Asian Journal of Nanoscience and Materials, vol.
1, no. 3, pp.122-134, 2018.
48. Electronics 360 IEEE Globalspec, PIN vs. APD: Different Sensitivity, Different Applications,
Electronics 360 IEEE Globalspec, Industrial Electronics, Nov. 21, 2017. Accessed on: June
13, 2018. [Online]. Available: https://electronics360.globalspec.com/article/10397/pin-
vs-apd-different-sensitivity-different-applications
49. A. A. Khadir, B. F. Dhahir and X. Fu, “Achieving Optical Fiber Communication Experiments
by Optisystem”, International Journal of Computer Science and Mobile Computing
(IJCSMC), Vol. 3, No. 6, pp.42 – 53, 2014
50. Recommendation ITU-T G.694.1 (02/2012), Spectral grids for WDM applications: DWDM
frequency grid
51. S. Singh, N. Gupta, R. P. Shukla and A. Sharma, “Simulation of full duplex data transmission
in ROF system using Optisystem”, International Journal of Electronics and Computer
Science Engineering (IJECSE), Vol. 1, No. 3, 2012
52. E. Dadrasnia and F. R. M. Adikan, “DWDM Effects of Single Model Optical Fiber in Radio
over Fiber System”, 2nd International Conference on Electronic Computer Technology
(ICECT), 2010
159

Dut Masters Degree in Engineering Dessertation - Final 04 Oct 2020

Uploaded by

Copyright:

Available Formats

Dut Masters Degree in Engineering Dessertation - Final 04 Oct 2020

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Dut Masters Degree in Engineering Dessertation - Final 04 Oct 2020

Uploaded by

Copyright:

Available Formats

Comparative Analysis, Design and Optimisation of a

48 Channel DWDM System using Various Design

Submitted in the fulfilment for the requirements of the degree of

Jerrin Zachariah Mohan

2. This document is my own work.

3. I have appropriately referenced the work of other people I have used.

Jerrin Mohan Date

Student No.: 20919951

Approved for Final Submission by:

Supervisor: Dr. O. A. Sokoya Date

In this dissertation, a comparative analysis is performed on the simulative design of a 48-

PLAGIARISM DECLARATION ................................................................................................. i

ACKNOWLEDGEMENTS ...................................................................................................... iii

LIST OF FIGURES .............................................................................................................. viii

LIST OF TABLES ...............................................................................................................xviii

LIST OF ACRONYMS ......................................................................................................... xix

GENERAL INTRODUCTION TO THE RESEARCH IN DENSE WAVE DIVISION

1.2 Research Motivation .............................................................................................. 1

1.3 Research Objectives ............................................................................................... 2

1.4 Research Methodology .......................................................................................... 2

1.5 Publications ........................................................................................................... 3

1.6 Original Research Contributions ............................................................................. 4

1.7 Dissertation Overview ............................................................................................ 5

2.2 Evolution of Fundamental Concepts in Fiber Optical Communication ...................... 7

2.3 Advantages of DWDM .......................................................................................... 10

2.4 Performance limitation factors affecting DWDM systems...................................... 11

2.4.1 Linear effects ................................................................................................ 11

2.4.2 Non-linear effects.......................................................................................... 12

2.6 DWDM System Design Components ..................................................................... 22

2.6.1 WDM Transmitter ......................................................................................... 22

2.6.1.1 Pseudo-Random Bit Sequence Generator (PRBS) .................................... 23

2.6.2.1 Advanced Line coding Modulation format: MDRZ ................................... 28

2.6.3.1 Bessel Filter ........................................................................................... 30

2.6.4.1 Fiber optic transmission medium ........................................................... 32

2.8 Measuring system performance ........................................................................... 37

2.8.1 Optical Signal to Noise Ratio (OSNR) .............................................................. 37

2.8.2 Quality (Q) factor .......................................................................................... 38

2.8.3 Bit Error Rate (BER) ....................................................................................... 38

2.8.4 Overall Signal Performance and Bandwidth Efficiency .................................... 39

2.9 Summary ............................................................................................................. 40

DESIGN OF A 48 X 25Gb/s DWDM SYSTEM...................................................... 41

3.2 Design Methodology ............................................................................................ 41

3.3 Design Parameters ............................................................................................... 43

3.3.1.1 Design of the Transmitter block.............................................................. 43

3.3.2.1 100 GHz channel spacing.......................................................................... 46

SIMULATION RESULTS FOR THE 48 CHANNEL DWDM NETWORKS ................... 54

4.3 The 48 Channel x 25 Gbps at 100 GHz Spacing using RZ ......................................... 59

4.4 The 48 Channel x 25 Gbps at 50 GHz Spacing using NRZ ........................................ 63

4.5 The 48 Channel x 25 Gbps at 50 GHz Spacing using RZ ........................................... 68

4.6 The 48 Channel x 25 Gbps at 25 GHz Spacing using NRZ ........................................ 73

4.7 The 48 Channel x 25 Gbps at 25 GHz Spacing using RZ ........................................... 82

4.9 The 48 Channel x 25 Gbps at 12.5 GHz Spacing using RZ ........................................ 98

4.14 Summary ........................................................................................................... 117

DISCUSSION OF RESULTS FOR THE 48 CHANNEL DWDM NETWORKS ............. 118

5.1 Introduction....................................................................................................... 118

5.3.1 Comparison of Post Compensation Technique with 1 EDFA vs Post Symmetric

5.4 Summary ........................................................................................................... 148

CONCLUSION AND RECOMMENDATIONS...................................................... 151