OptiPerformer

Examples from Govind P. Agrawal‘s Book

Fiber-Optic Communication Systems

OptiSystem Project Simulator

Version 8.0
OptiPerformer
Examples from Govind P. Agrawal‘ Book
Fiber-Optic Communication Systems

OptiSystem Project Simulator

Copyright © 2010 Optiwave

All rights reserved.

All OptiPerformer documents, including this one, and the information contained therein, is copyright material.

No part of this document may be reproduced, stored in a retrieval system or transmitted in any form or by any means whatsoever,
including recording, photocopying, faxing, etc., without prior written approval of Optiwave.

Disclaimer
Optiwave makes no representation or warranty with respect to the adequacy of this documentation or the programs which it
describes for any particular purpose or with respect to its adequacy to produce any particular result. In no event shall Optiwave, its
employees, its contractors, or the authors of this documentation be liable for special, direct, indirect, or consequential damages,
losses, costs, charges, claims, demands, or claim for lost profits, fees, or expenses of any nature or kind.
Technical support

If you purchased Optiwave software from a distributor that is not listed here, please send technical
questions to your distributor.

Optiwave Canada/US
Tel (613) 224-4700 E-mail [email protected]
Fax (613) 224-4706 URL www.optiwave.com

Cybernet Systems Co., Ltd. Japan

Tel +81 (03) 5297-3342 E-mail [email protected]
Fax +81 (03) 5297-3646 URL www.cybernet.co.jp

Optiwave Europe Europe

Tel +33 (0) 494 08 27 97 E-mail [email protected]
Fax +33 (0) 494 33 65 76 URL www.optiwave.eu
Table of contents
Introduction ..................................................................................................... 1

Basic Concepts of Optical Communication Systems ................................. 5

Optical NRZ and RZ Modulation Formats ..................................................................7

Lightwave System Components...............................................................................11

Optical Fibers ................................................................................................ 15

Fiber Dispersion .......................................................................................................17

Chirped Gaussian Pulses.........................................................................................21
Attenuation Coefficient .............................................................................................29
Stimulated Light Scattering - Raman .......................................................................31
SPM and XPM..........................................................................................................33
Four-Wave Mixing ....................................................................................................35

Optical Transmitters ..................................................................................... 37

LED Spectral Distribution .........................................................................................39

LED Modulation Response ......................................................................................41
Light-Current Characteristics ...................................................................................45
Direct Modulation and External Modulation .............................................................47
Laser Intensity Noise................................................................................................49

Optical Receivers .......................................................................................... 51

Receiver Design.......................................................................................................53
Receiver Shot and Thermal Noise ...........................................................................57
Receiver PIN x APD.................................................................................................59
Receiver BER and Q Factor.....................................................................................61
Minimum Received Power .......................................................................................63
ER Induced Sensitivity Degradation.........................................................................65

Lightwave Systems ....................................................................................... 67

Power Budget...........................................................................................................69
Power Management for Ring Networks ...................................................................71
Dispersion Induced Degradation..............................................................................85
Multichannel Systems .................................................................................. 87

Fabry-Perot Filter .....................................................................................................89

Uniform FBG Filter ...................................................................................................91
AWG Demultiplexer..................................................................................................93
Star Coupler .............................................................................................................95
Optical Cross-Connects ...........................................................................................97
Channel Multiplexing................................................................................................99
Channel Demultiplexing by FWM...........................................................................101
Interchannel Crosstalk at ADM in A Ring Network.................................................103
Subcarrier Multiplexing...........................................................................................107

Optical Amplifiers ....................................................................................... 109

EDFA Basic Concepts............................................................................................111

Booster Amplifier....................................................................................................117
In-Line Amplifier .....................................................................................................119
Preamplifier ............................................................................................................121
SLA Cascaded in-line Amplifiers: Soliton Pattern Effect at 10 Gb/s in SMF ..........123
SLA Gain Recovery................................................................................................127
SLA Gain Saturation of Chirped and Super Gaussian Pulses ...............................141
SLA Gain Saturation of Gaussian Pulses ..............................................................149
Raman Gain and Bandwidth ..................................................................................159
Raman Gain Saturation..........................................................................................161
Broadband Flat Gain Raman Amplifier ..................................................................163
Pumping Requirements..........................................................................................165
Multichannel Amplification......................................................................................169
Optical Preamplifiers ..............................................................................................171
Power Boosters......................................................................................................173
Metro EDFA ...........................................................................................................175
Cascaded In-Line Amplifiers ..................................................................................177
EDFA Dynamic.......................................................................................................179

Dispersion Management ............................................................................ 181

Dispersion Compensation Techniques ..................................................................183

Precompensation Technique using SOA ...............................................................189
40 Gbps Transmission System ..............................................................................195
Dispersion Compensation with FBG ......................................................................203
Fiber Nonlinearities and Dispersion .......................................................................205
Negative Dispersion Fiber and Applications for Metro Networks ...........................211
Migrating to 10 Gbps Metro System and Dispersion Limited Transmission...........215
Soliton Systems .......................................................................................... 221

Fundamental and Higher-Order Solitons ...............................................................223

Interaction of A Pair of Fundamental Solitons........................................................227
Average Soliton Regime ........................................................................................229
Interaction of Multicolor Solitons ............................................................................231

Advanced Lightwave Systems .................................................................. 233

Intensity Modulation Formats (NRZ, RZ, AMI, DB) ................................................235

DPSK Modulation and Demodulation.....................................................................251
QPSK Modulation and Demodulation ....................................................................257
100 Gbps PM-QPSK ..............................................................................................263
OD8PSK Modulation and Demodulation................................................................265
4-QAM OFDM Modulation and Demodulation .......................................................269

Optical Signal Processing .......................................................................... 273

Nonlinear Optical Loop Mirror ................................................................................275

Parametric Amplifiers .............................................................................................277
Wavelength Conversion Based on FWM in SOA...................................................279
Wavelength Conversion Based on XGM in SOA ...................................................283
Wavelength Conversion Based on XPM in HNLF ..................................................287
Data-Format Conversion Based on XPM in HNLF.................................................291
SPM Based 2R Regenerator for RZ Signal............................................................293
XPM Based 3R Regenerator for RZ Signal............................................................295
Introduction
OptiPerformer is a software tool that enables technical sales and marketing teams of
component and system vendors to powerfully and cost-effectively demonstrate their
products. The demonstration is carried out using design scenarios that accurately
project performance characteristics of real systems built using vendor components or
subsystems. By using OptiPerformer, an application engineer can easily demonstrate
the benefits of a product, or propose different scenarios for the product, without
having to develop the full depth of R&D-level technical knowledge typical of proficient
OptiSystem users.

OptiPerformer’s capabilities are derived from Optiwave’s award-winning OptiSystem

product, an optical system design tool that enables users to plan, test, and simulate
almost every type of optical link in the physical layer of a broad spectrum of optical
networks. OptiSystem addresses the needs of research scientists and optical telecom
engineers, and offers the benefits of dramatic reductions of investment risk and time-
to-market, rapid, low-cost prototyping, and global insight into system performance.

OptiPerformer extends OptiSystem’s simulation power beyond an organization’s R&D

department, addressing the needs of the ‘customer-facing’ team. OptiPerformer
features key streamlined functionality and enhanced ease-of-use, enabling technical-
sales and marketing oriented users to convincingly demonstrate to their potential
customers the value of their specific component solution. Further to this, potential
customers can simulate and validate component and system vendors proposed
component and sub-system performance versus competing offers within the virtual
environment of OptiPerformer.

1
 Figure 0-1: OptiPerformer GUI

OptiPerformer can be distributed and used by both system and component vendors,
as well as their customers and prospects.

2
Main features
The main features of the OptiPerformer interface include:

Feature Description

Graphical user interface A comprehensive Graphical User Interface (GUI) controls the optical
component layout, component models, and presentation graphics.
The Performer Control window provides simple, easy-to-use button
controls for simulation control.

Mixed signal representation OptiSystem handles mixed signal formats for optical and electrical
signals in the Component Library. OptiSystem calculates the signals
using the appropriate algorithms related to the required simulation
accuracy and efficiency.

Quality and performance In order to predict the system performance, OptiSystem calculates
algorithms parameters such as BER and Q-Factor using numerical analysis or
semi-analytical techniques for systems limited by inter symbol
interference and noise.

Advanced visualization tools Advanced visualization tools produce OSA Spectra, Oscilloscope,
and Eye diagrams. Also included are WDM analysis tools listing
signal power, gain, noise figure, and OSNR.

State-of-the-art calculation The Calculation Scheduler controls the simulation by determining

data-flow the order of execution of component modules according to the
selected data flow model. The main data flow model that addresses
the simulation of the transmission layer is the Component Iteration
Data Flow (CIDF). The CIDF domain uses run-time scheduling,
supporting conditions, data-dependent iteration, and true recursion.

3
Notes:

4
Chapter 1 - Basic Concepts of Optical
Communication Systems
This chapter contains the following examples.

• Optical NRZ and RZ Modulation Formats

• Lightwave System Components

5
Notes:

6
 OPTICAL NRZ AND RZ MODULATION FORMATS

Optical NRZ and RZ Modulation Formats

Optical NRZ RZ formats.osp

There are two typical choices for the modulation format of an optical signal. These are shown in the
project file “Optical NRZ RZ formats.osp” and are known as the return-to-zero (RZ) and nonreturn-to-zero
(NRZ).

Figure 1-1: Layout - NRZ and RZ modulation formats

In the RZ format, each pulse that represents bit 1 is shorter than the bit slot, and its amplitude returns to
zero before the bit duration is over. In the NRZ format, the pulse remains on throughout the bit slot and
its amplitude does not drop to zero between two or more successive bits. As a result, pulse width varies
depending on the bit pattern, whereas it remains the same in the case of RZ format.

7
OPTICAL NRZ AND RZ MODULATION FORMATS

Figure 1-2: NRZ and RZ (Amplitude vs. Time)

An advantage of the NRZ format is that the bandwidth associated with the bit stream is smaller than that
of the RZ format by about a factor of 2 simply because on-off transitions occur fewer times.

Figure 1-3: Spectrum of NRZ and RZ signals

8
 OPTICAL NRZ AND RZ MODULATION FORMATS

Figure 1-4: Eye diagrams of (a) NRZ and (b) RZ signals

(a) (b)

9
OPTICAL NRZ AND RZ MODULATION FORMATS

Notes:

10
 LIGHTWAVE SYSTEM COMPONENTS

Lightwave System Components

Lightwave system components.osp

Project “Lightwave system components.osp” shows a generic block diagram of an optical communication
system. It consists of a communication channel with a transmitter in the front and a receiver at the back.

Figure 1-5: Layout - Lightwave system components

(a) Communication Channel

The role of the communication channel is to transport the optical signal from transmitter to receiver
without distorting it. Most lightwave communication systems use optical fibers as the communication
channel because fibers can transmit light with a relatively small amount of power loss.

Fiber loss is, of course, an important design issue, as it determines directly the repeater spacing of a long-
haul lightwave system. Another important design issue is fiber dispersion, which leads to broadening of
individual pulses inside the fiber.

In order to observe the effects of loss and dispersion in the optical signal the user can change the values
of fiber length and visualize the degradation of the signal at the receiver stage.

11
LIGHTWAVE SYSTEM COMPONENTS

Figure 1-6: Received eye diagram after 50 km of fibe

Figure 1-7: Received eye diagram after 100 km of fiber

12
 LIGHTWAVE SYSTEM COMPONENTS

Figure 1-8: Received eye diagram after 126 km of fiber

(b) Optical Transmitters

The role of the optical transmitter is to convert the electrical signal into optical form and to launch the
resulting optical signal into the optical fiber. It consists of an optical source, an electrical pulse generator
and an optical modulator.

The launched power is an important design parameter, as it indicates how much fiber loss can be
tolerated. It is often expressed in units of dBm with 1 mW as the reference level.

(c) Optical Receivers

An optical receiver converts the optical signal received at the output end of the optical fiber back into the
original electrical signal. It consists of a photodetector, a filter and a demodulator. Usually the received
signal is in the form of optical pulses representing 1 and 0 bits and it is converted directly into an electrical
current. Such a scheme is referred to as intensity modulation with direct detection (IM/DD).
Demodulation is done by a decision circuit that identifies bits as 1 or 0, depending on the amplitude of
the electrical current.

The performance of a digital lightwave system is characterized through the bit- error rate (BER). It is
customary to define the BER as the average probability with incorrect bit identification. Most lightwave
systems specify a BER of 10-9 as the operating requirement. For this project, a laser output power of 0
dBm with 126 km of fiber length will lead to an output signal with BER of 10-9.

13
LIGHTWAVE SYSTEM COMPONENTS

Figure 1-9: Received eye diagram and BER after 126 km of fiber

14
Chapter 2 - Optical Fibers
This chapter contains the following examples.

• Fiber Dispersion
• Chirped Gaussian Pulses
• Attenuation Coefficient
• Stimulated Light Scattering - Raman
• SPM and XPM
• Four-Wave Mixing

15
Notes:

16
 FIBER DISPERSION

Fiber Dispersion

Fiber dispersion.osp

In this example we demonstrate the influence of the group velocity dispersion in SMF at 1.55 μm using
externally modulated optical sources. The fiber dispersion is set to D = 16.75 ps/nm-km @ 1550 nm,
attenuation and fiber nonlinearity are not considered in this example. We consider the unchirped pulse
and neglect the influence of the initial pulse shape.

Externally modulated laser with a small spectral width is used as an optical source. A transmission bit
rate of 10 Gb/s is considered in the layout which is shown in Figure 2-1:

Figure 2-1: Layout - Externally modulated laser

2
For simulation, we consider duty cycle = 0.5, TFWHM = 50 ps, T0 = 30.03 ps, LD = T 0 ⁄ β 2 = 42.24 km,
Δω = 1/T0= 33.3 GHz, and σ ω ~ 2π × 10 MHz. Figure 2-2 shows the generated spectrum of the CW laser.
Since σ ω ⁄ Δ ω << 1, for the spread of the pulse width, we can expect the validity of the following formula :

σ
2 ⎛ β2 z⎞ 2
-⎟ = 1 + ⎛⎝ ------
z -⎞ 2
------ = 1 + ⎜ -------- (2-1)

σ0
2
⎝ 2σ 20⎠ L D⎠

where σ 0 is the RMS width of the input Gaussian pulse σ 0 = T 0 ⁄ ( 2 ) . Eq. 2-1 shows that to get 4 times
increase of the pulse width a distance of approximately 4LD ( ~ 169 km ) is needed.

17
FIBER DISPERSION

Figure 2-2: Spectrum of the CW laser

(Note that to get enough spectral resolution a very large time window was used with a sequence length
of 2048 bits and 64 samples per bit).

Obtained results after the simulation are presented in Figure 2-3, Figure 2-4, and Figure 2-5.

Figure 2-3: Pulse profile after the modulator (TFWHM = 50 ps)

18
 FIBER DISPERSION

Figure 2-4: Spectrum after the modulator

Figure 2-5: Pulse profile after propagation in 169 km of SMF (TFWHM = 197.7 ps)

As can be seen that the pulse width increases approximately 3.95 times (cf. Eq. 2-1).

Reference:

[1] G.P. Agrawal, Nonlinear Fiber optics, third edition, Academic Press, 2001.

19
FIBER DISPERSION

Notes:

20
 CHIRPED GAUSSIAN PULSES

Chirped Gaussian Pulses

Chirped Gaussian pulses.osp

Chirped Gaussian pulses - Dispersion compensation.osp

(a) Chirped Gaussian pulses

In this example we show the influence of the GVD on the spatial evolution of the pulse width of initially
chirped Gaussian pulses. Again, attenuation and fiber nonlinearity are not considered. The general
expression for the spatial evolution of pulse width in dispersive media is given by Eq. 2-2:

σ
2 ⎛ Cβ 2 z⎞ 2 2 ⎛ β2 z⎞ 2
------ = ⎜ 1 + -------------⎟ + ( 1 + V ω ) × ⎜ ---------⎟ (2-2)

σ0
2
⎝ 2σ 0 ⎠
2
⎝ 2σ 20⎠

Where: C is the initial chirp of the pulse

σ is the root-mean-square (RMS) width of the pulse

V ω = 2σ ω σ 0 = ( 2σ ω ) ⁄ ( Δω ) is the parameter describing the relative contribution of spectral width of the

source and of the pulse

σ 0 = T0 ⁄ ( 2 ) , where T0 is the initial pulse width

2πc
D = – ---------β 2
2
λ
Δω = 1 ⁄ T 0 is the spectral width of the unchirped Gaussian pulse

The typical value of D = 16.75 ps/nm-km at 1.55 μm for SMF leads to β 2 = -21.35 ps2/km. For the
considered optical source (source with small spectral width) the limiting transmission distances are
determined by the pulse width and the GVD. For such source, the dispersive action of the medium is
2
described by the dispersion length L D = T 0 ⁄ β 2 , which characterizes the pulse spreading in dispersive
medium. In this case, for the initially unchirped pulse, if V ω « 1 , Eq. 2-2 can be simplified into Eq. 2-3:

σ
2 ⎛ β2 z⎞ 2
------ = 1 + ⎜ ---------⎟ = 1 + ⎛⎝ ------
z -⎞ 2 (2-3)
σ0
2
⎝ 2σ 20⎠ L D⎠

21
CHIRPED GAUSSIAN PULSES

If for a Gaussian pulse we assume duty cycle = 0.5, then

B = 2.5 Gb/s -> TB = 400 ps -> TFWHM = 200 ps -> T0 = 120.12 ps -> LD = 675.86 km -> Δω = 8.3250
GHz -> σ ω ⁄ ( Δω ) « 1

B = 10 Gb/s -> TB = 100 ps -> TFWHM = 50 ps -> T0 = 30.03 ps -> LD = 42.24 km -> Δω = 33.3 GHz ->
σ ω ⁄ ( Δω ) « 1

B = 40 Gb/s -> TB = 25 ps -> TFWHM = 12.5 ps -> T0 = 7.51 ps -> LD = 2.64 km -> Δω = 133.16 GHz ->
σ ω ⁄ ( Δω ) « 1

Let us consider B= 40 Gb/s. In accordance with Eq. 2-3, at 4 x LD = 10.56 km the width of the initially
unchirped pulse will increase approximately 4 times. This can be demonstrated with the project Chirped
Gaussian pulses.osp which has the following layout:

Figure 2-6: Layout - Chirped Gaussian pulses

The obtained results for 40 Gb/s transmission, duty cycle = 0.5 and D = 16.75 ps/nm-km at 1.55 μm are
shown in Figure 2-7.

22
 CHIRPED GAUSSIAN PULSES

Figure 2-7: (a) Initial pulse shape (intensity and chirp) (b) Initial pulse spectrum (c) Pulse shape after
propagation in 10.56 km of SMF (intensity and chirp) (d) Pulse spectrum after propagation in 10.56 km of
SMF.

(a) (b)

(c) (d)

As expected (cf. Eq. 2-3), an increase of the initially unchirped pulse width by approximately 4 times can
be clearly seen. Note the appearance of the negative chirp of the pulse after transmission in 10.56 km
of fiber in anomalous group velocity dispersion region (Figure 2-7(c)). Because the GVD is the only
included effect in the calculation, no changes in the spectrum can be observed after the transmission.

Now, we consider the pulse evolution of the initially chirped pulse (chirp parameter C = -2) for a
transmission distance of 10.56 km. From Eq. 2-2 we see that when Cβ 2 > 0 the broadening of the pulse
is increased with comparison to unchirped initial pulse. Results are shown in Figure 2-8 and Figure 2-9.

23
CHIRPED GAUSSIAN PULSES

Figure 2-8: Initial chirped pulse shape (C = -2). Note the negative initial chirp shown in the figure.

Figure 2-9: Increase of pulse width after propagation in 10.56 km of SMF.

As we can see from Figure 2-9, after only 10.56 km of SMF, the pulse width increases approximately
11.52 times or much more as compared to the case of initially unchirped pulse.

However, if we substitute Cβ 2 < 0 in Eq. 2-2, the pulse compression could be expected. To demonstrate
this type of behavior, we consider an initial chirped pulse with C = 2 and a transmission distance of 1.056
km.

Results are shown in Figure 2-10 and Figure 2-11.

24
 CHIRPED GAUSSIAN PULSES

Figure 2-10: Initial chirped pulse shape (C = 2, TFWHM = 12.5 ps). Note the positive initial chirp shown in the
figure.

Figure 2-11: Decrease of pulse width (TFWHM = 5.6 ps) after propagation in 1.056 km of SMF.

As we can see, pulse width gets reduced approximately two times as compared to its initial value after
propagation in 1.056 km of SMF.

(b) Chirped Gaussian pulses - Dispersion compensation

Now, we will pay attention to the cancellation of chirp in single mode fiber (SMF) and dispersion
compensation fiber (DCF). The project has the following layout (Figure 2-12):

25
CHIRPED GAUSSIAN PULSES

Figure 2-12: Layout - Chirped Gaussian pulses - Dispersion compensation

We choose the following parameters for SMF and DCF:

DSMF = 16 ps/nm-km, DispersionSlopeSMF = 0.08 ps/nm2-km, α = 0.2 dB/km and LSMF = 10 km.

DDCF = -80 ps/nm-km, DispersionSlopeDCF = 0.08 ps/nm2-km, α = 0.5 dB/km and LDCF = 2 km.

The initial Gaussian pulse profile can be seen in Figure 2-13.

Figure 2-13: Initial Gaussian pulse profile

26
 CHIRPED GAUSSIAN PULSES

Figure 2-14: Pulse profile after propagation in 10 km of SMF

If the pulse is sent to 10 km of SMF, the pulse profile after propagation is given in Figure 2-14.Note the
negative chirp appeared after the propagation.

If the pulse is sent to 2 km of DCF, the pulse profile after the propagation is given in Figure 2-15, and
there is a positive chirp being developed.

Figure 2-15: Pulse profile after propagation in 2 km of DCF

Note that the obtained chirp after 2 km of DCF is with opposite sign as compared to after 10 km of SMF,
which would lead to complete compensation of dispersion if we put the DCF after the SMF. Simulation
result for this case is given in Figure 2-16. As can be seen, after dispersion compensation, no chirp is
observed and hence the complete recovery of the pulse is achieved.

27
CHIRPED GAUSSIAN PULSES

Figure 2-16: Pulse profile after propagation in 10 km of SMF and 2 km of DCF

References:

[1] G.P. Agrawal, Nonlinear Fiber optics, third edition, Academic Press, 2001.

[2] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC, 2010.

28
 ATTENUATION COEFFICIENT

Attenuation Coefficient

Attenuation coefficient.osp

Project file “Attenuation coefficient.osp” shows power loss in the fiber. In this example a basic 50 km fiber
link is designed. The system works in 1.3 μm window and the loss of fiber at this wavelength is 0.5 dB/km.
Fiber is spliced after every 5 km and has two connectors of 1 dB loss at both ends. Splice loss is 0.2 dB.
The channel loss in total is 50 x 0.5 dB + 2 x 1 dB + 9 x 0.2 dB = 28.8 dB. You can see the received signal
power by double clicking on power meter after you run the simulation with 24 dBm launched power, the
received power is -4.8 dBm.

The received power changes when the connector attenuation is changed. As an additional work, change
the attenuation constant of the fibers and repeat your calculation with this new parameter. Check the
validity of your calculation by running a new simulation with this new attenuation constant.

29
ATTENUATION COEFFICIENT

Notes:

30
 STIMULATED LIGHT SCATTERING - RAMAN

Stimulated Light Scattering - Raman

Raman spectrum - With broad band source signal.osp

Raman spectrum - With two pump signals.osp

Project file “Raman spectrum - With broad band source signal.osp” and “Raman spectrum - With two
pump signals.osp“ shows SRS in spectral domain for two different cases. In the first case, a signal at
1450 nm (pump) is multiplexed with the broad band source signal. You can see how the spectrum of the
signal from broadband source is affected. Fiber acts as a Raman amplifier such that longer wavelengths
are amplified by the shorter ones. In this example, some power is transferred from the pump to other
broad band spectrum components. As can be seen from the spectrum, the gain is a function of
wavelength. In the second case, we have used two pump signals and swept the separation between
them. You can see how the gain spectrum is changed depending on the separation of the pumps. Note
also that the power of the second pump increases as its wavelength increases, because it sees more and
more gain according to the gain spectrum.

31
STIMULATED LIGHT SCATTERING - RAMAN

Notes:

32
 SPM AND XPM

SPM and XPM

SPM.osp

XPM and GV mismatch.osp

(a) Self Phase Modulation (SPM)

Project file “SPM.osp” shows how SPM affects the spectral shape of a Gaussian pulse. In this example,
we enable only the SPM effect for our fiber model; all other effects are set to off. You can see the spectral
shape of the pulse depending on its power using the OSA at the fiber end. The nonlinear phase shift
depends on input pulse power and nonlinear parameter. Changing the effective area of the fiber changes
the nonlinear parameter. You can re-run the simulation with different effective areas and see how the
results are changed. Since dispersion effect is set to off, you will not see any change on the pulse
intensity profile. In fact, nonlinear phase shifts with time, resulting in frequency chirp, which in turn affects
the pulse shape through group velocity dispersion (GVD). You can see this by setting the GVD effect on
and re-run the simulation. In that case, you will see that pulse goes through a contraction phase when
the dispersion parameter D is positive ( β 2 < 0, anomalous-GVD). With correct parameters you can even
get a “soliton”. As an extra work, try to see what happens when the dispersion is negative (D < 0, β 2 > 0).

(b) Cross Phase Modulation (XPM) and GV mismatch

Project file “XPM and GV mismatch.osp” demonstrates that XPM induced distortions can be canceled
out, if the dispersion is sufficiently high. As in the case of SPM, there is no pulse broadening introduced
by XPM if there is no GVD to transform the spectral broadening to pulse broadening. On the other hand,
walk off between the channels because of GVD also tends to lessen the impact of XPM. In this example,
we consider two isolated pulses in different channels propagating in a lossless fiber, and the signal pulse
will sweep across the pump pulse when propagating in the fiber. When the signal pulse is overlapped
with the leading edge of the pump pulse, the signal pulse will receive a red shift due to rising intensity of
the interfering pulse. As the pulses walk through each other, and trailing edge of the pump pulse is
superimposed with the signal pulse, the signal pulse receives a blue shift. In fact, the net frequency shift
experienced by a particular part of the signal pulse is proportional to the integral of the derivative of the
intensity of the pump pulse. Therefore, when the pulses have completely walked through each other,
there is no residual effect.

33
SPM AND XPM

Figure 2-17: Pulse relative positions (a) input (b) after 12.5 km of fiber(c) after 25 km of fiber

Here we have considered two 50 ps pulses with an initial 300 ps center-to-center separation. The
channels are separated by 6 nm. Dispersion of the fiber is set to -4 ps/nm-km. The signal pulse has a
lower power than the pump pulse. As shown in Figure 2-17, the pulses are perfectly superimposed after
12.5 km of fiber, and are separated again after 25 km of fiber. Since power of the signal pulse is low, SPM
is negligible. However, XPM should be large because the power of pump pulse is high. The spectrum of
signal and pump at the input, after 12.5 km of fiber, and after 25 km of fiber are shown in Figure 2-18.
You can see that the spectrum broadening of the signal due to XPM is smaller than the spectrum
broadening of the pump pulse arising from SPM by looking at the spectrum at 12.5 km. Note that, pulses
are completely overlapped at this distance. You can also see the new spectrum components generated
by FWM. At 25 km, the pulses are separated and effects of XPM have completely vanished, while the
effect of SPM on the pump pulse remains. So, this example has demonstrated that dispersion can reduce
the effects of XPM.

Figure 2-18: Power spectrum (a) input (b) after 12.5 km of fiber and (c) after 25 km of fiber

Reference:

[1] Optical Fiber Telecommunications IIIa, ed. by I. Kaminov and T. Koch, chapter “Fiber Nonlinearities
and their Impact on Transmission Systems” by F. Forghieri, R. Tkach and A. Chraplyvy

34
 FOUR-WAVE MIXING

Four-Wave Mixing

FWM.osp

Project file “FWM.osp” demonstrates the FWM effect. When three optical fields with carrier frequencies
f 1 , f 2 , and f3 co-propagate inside the optical fiber simultaneously, nonlinear effect generate a fourth
field whose frequency f4 is related to other frequencies by a relation f 4 = f 1 ± f2 ± f 3 . In principle, several
frequencies corresponding to different plus and minus signs combinations are possible. In practice, the
frequency combinations of the form f 4 = f 1 + f 2 – f 3 are most troublesome. For an N channel system,
1 3 2
M = --- ( N – N ) mixing products are generated. In this example, when the 3 channels are equidistant,
2
some of the generated fields by FWM overlap with existing channels. Therefore, you will not be able to
see all the mixing products in the spectrum for this case. When the 3 channels are not equidistant, you
can see all the mixing products in the spectrum. You can also see that the above given formula is satisfied
and the number of generated mixing product is 9. You can change the separation between channels to
see how the spectrum is affected.

35
FOUR-WAVE MIXING

Notes:

36
Chapter 3 - Optical Transmitters
This section contains the following examples.

• LED Spectral Distribution

• LED Modulation Response
• Light-Current Characteristics
• Direct Modulation and External Modulation
• Laser Intensity Noise

37
Notes:

38
 LED SPECTRAL DISTRIBUTION

LED Spectral Distribution

LED spectral distribution.osp

The spectral distribution of light sources affects the performance of optical communication systems
through fiber dispersion. The spectral distribution is governed by the spectrum of spontaneous emission
and typically follows a Gaussian shape. Figure 3-1 shows the layout for “LED spectral distribution.osp”.
It generates the output spectrum of a typical 1300 nm LED with 50 nm of spectral width (Figure 3-2 and
Figure 3-3).

Figure 3-1: Layout - LED spectral distribution

Figure 3-2: Output spectrum of LED

39
LED SPECTRAL DISTRIBUTION

Figure 3-3: Output spectrum of LED with 1 nm resolution bandwidth

40
 LED MODULATION RESPONSE

LED Modulation Response

LED modulation response.osp

The modulation response of LEDs depends on carrier dynamics and is limited by the carrier lifetime. The
LED 3-dB modulation bandwidth is defined as the modulation frequency at which the LED transfer
function is reduced by 3 dB or by a factor of 2. Figure 3-4 shows the layout for “LED modulation
response.osp”.

Figure 3-4: Layout - LED modulation response

The project “LED modulation response.osp” shows the closure of the eye diagram when the
modulation bit rate increases from 100 MB/s to 400 MB/s (Figure 3-5, Figure 3-6, Figure 3-7, and
Figure 3-8). This phenomenon occurs because of the LED modulation response. Reducing the carrier
lifetime value will increase the LED modulation response.

Figure 3-5: LED modulation at 100 MB/s

41
LED MODULATION RESPONSE

Figure 3-6: LED modulation at 200 MB/s

Figure 3-7: LED mdulation at 300 MB/s

42
 LED MODULATION RESPONSE

Figure 3-8: LED modulation 400 MB/s

Figure 3-9 shows the electrical signal after the photodetector and before the electrical filter.

Figure 3-9: Electrical signal - LED modulation at 400 MB/s

43
LED MODULATION RESPONSE

Notes:

44
 LIGHT-CURRENT CHARACTERISTICS

Light-Current Characteristics

Semiconductor laser L-I curve.osp

The Light-Current (L-I) curve characterizes the emission properties of a semiconductor laser, as it
indicates not only the threshold level but also the current that needs to be applied to obtain a certain
amount of power. Project “Semiconductor laser L-I curve.osp” (Figure 3-10) generates the L-I curve of a
laser.

Figure 3-10: Layout - Semiconductor laser L-I curve

The L-I curve is generated after sweeping the parameter “Amplitude” of the Bias Generator component
from 0.01 to 100 mA. After running the simulation, results are given in Figure 3-11 (it can also be
accessed from the report page).

Figure 3-11: Laser L-I curve

45
LIGHT-CURRENT CHARACTERISTICS

Notes:

46
 DIRECT MODULATION AND EXTERNAL MODULATION

Direct Modulation and External Modulation

Semiconductor laser large signal modulation.osp

Direct modulation vs External modulation.osp

Amplitude modulation in semiconductor lasers is accompanied by phase modulation. A time-varying

phase is equivalent to transient changes in the mode frequency from its steady-state value. Such a pulse
is called chirped. The project “Semiconductor laser large signal modulation.osp” (Figure 3-12) generates
the amplitude and signal chirp for a directly modulated semiconductor laser. Figure 3-13 shows the
amplitude and signal chirp for the directly modulated semiconductor laser

Figure 3-12: Layout - Semiconductor laser large signal modulation

Figure 3-13: Amplitude and signal chirp for directly modulated semiconductor laser

47
DIRECT MODULATION AND EXTERNAL MODULATION

Figure 3-14: Output spectrum

48
 LASER INTENSITY NOISE

Laser Intensity Noise

Laser intensity noise.osp

Project “Laser intensity noise.osp” (Figure 3-15) shows the laser spectral under CW operation at
several power levels.

Figure 3-15: Layout - Laser intensity noise

The laser exhibits fluctuations in its intensity, phase and frequency even when the laser is biased at a
constant current with negligible current fluctuations (Figure 3-16).

Figure 3-16: Laser intensity noise (from report page)

49
LASER INTENSITY NOISE

Notes:

50
Chapter 4 - Optical Receivers
This section contains the following examples.

• Receiver Design
• Receiver Shot and Thermal Noise
• Receiver PIN x APD
• Receiver BER and Q Factor
• Minimum Received Power
• ER Induced Sensitivity Degradation

51
Notes:

52
 RECEIVER DESIGN

Receiver Design

Receiver design.osp

The design of an optical receiver depends largely on the modulation format used by the transmitter.
Project “Receiver design.osp” (Figure 4-1) shows a digital receiver design. Its components can be
arranged into three groups, the front end, the linear channel and the data-recovery section.

Figure 4-1: Layout - Receiver design

The front-end of a receiver consists of a photodiode with a built in preamplifier, the linear channel consists
of a high-gain amplifier and an low-pass filter and the data recovery is done by the BER Analyzer with a
built in clock recovery and decision circuit.

The linear channel has a low pass filter that shapes the pulse. The receiver noise is proportional to the
receiver bandwidth and can be reduced by using a low-pass filter whose bandwidth is smaller than the
bit rate. The pulse spreads beyond the time slot. Such spreading can interfere with the detection of
neighboring bits, the phenomenon referred to as intersymbol interference (ISI).

When the attenuation is 0 dB, the ideal eye diagram is shown as below: (Figure 4-2)

53
RECEIVER DESIGN

Figure 4-2: Received eye diagram (ideal)

By setting the attenuation to 32 dB, you can observe the degraded eye diagram (Figure 4-3).

Figure 4-3: Received eye diagram (degraded)

The decision circuit compares the output from the linear channel to a threshold, at a decision instant
determined by the clock recovery circuit, and decides whether the signal corresponds to bit 1 or bit 0.
The optimum threshold level is calculated for each decision instant in order to minimize the BER
(Figure 4-4).

54
 RECEIVER DESIGN

Figure 4-4: Received eye diagram and threshold vs. decision time instant

55
RECEIVER DESIGN

Notes:

56
 RECEIVER SHOT AND THERMAL NOISE

Receiver Shot and Thermal Noise

Receiver shot and thermal noise.osp

There are two fundamental noise mechanisms in a photodetector: shot noise and thermal noise. The
project “Receiver shot and thermal noise.osp” (Figure 4-5) shows the signal degraded by thermal and
shot noise in the PIN photodetector. The low-pass filter has cutoff frequency of the same value as the bit
rate.

Figure 4-5: Layout - Receiver shot and thermal noise

The upper system has the photodetector without thermal noise; the only noise generated at the output is
the shot noise. You can observe that the shot noise is signal amplitude dependent (Figure 4-6).

Figure 4-6: Receiver shot noise

57
RECEIVER SHOT AND THERMAL NOISE

The lower system has the photodetector without shot noise; the only noise generated at the output
is the thermal noise. You can observe the thermal noise is signal amplitude independent (Figure 4-7).

Figure 4-7: Receiver thermal noise

58
 RECEIVER PIN X APD

Receiver PIN x APD

Receiver PIN x APD.osp

Optical receivers with APD generally provide a higher SNR for the same incident optical power. The
improvement is due to the internal gain that increases the photocurrent by the multiplication factor M. The
project “Receiver PIN x APD.osp“ (Figure 4-8) shows a comparison with two systems using PIN and
APD, respectively.

Figure 4-8: Layout - Receiver PIN x APD

When the gain factor of the APD photodetector is set to 3.5, the eye diagram has a Q factor higher
(Figure 4-9) than the eye diagram generated using the PIN photodetector (Figure 4-10).

Figure 4-9: Q factor - APD

59
RECEIVER PIN X APD

Figure 4-10: Q factor - PIN

If we increase the gain factor then there is a limit where the shot noise will degrade the system
performance, therefore it’s important to find the optimum APD gain. If we run the same simulation varying
the value of the gain factor we can observe the evolution of the Q factor (Figure 4-11).

Figure 4-11: Q factor vs. APD gain

In this case, for a gain higher than 21.2, there is no advantage of using the APD (which is represented
by the black line), compared to using PIN (which is represented by the red line), since it will not improve
the receiver sensitivity.

60
 RECEIVER BER AND Q FACTOR

Receiver BER and Q Factor

Receiver BER - Q factor.osp

The performance criteria for digital receivers, is governed by the bit-error-rate (BER), defined as the
probability of incorrect identification of a bit by the decision circuit of the receiver.

Figure 4-12: Layout - Receiver BER - Q factor

The project “Receiver BER - Q factor.osp” generates the BER and Q factor at the data recovery stage
for different values of input power. In the project, by changing the attenuator attenuation, we can calculate
not only Q- Factor versus attenuation (Figure 4-13) and BER versus attenuation (Figure 4-14), but also
the BER versus Q factor (Figure 4-15).

Figure 4-13: Q factor vs. attenuation

61
RECEIVER BER AND Q FACTOR

Figure 4-14: BER vs. attenuation

Figure 4-15: BER vs. Q factor

62
 MINIMUM RECEIVED POWER

Minimum Received Power

Receiver – Minimum received power.osp

The project “Receiver – Minimum received power.osp” (Figure 4-16) shows the minimum optical power
that a receiver needs to operate reliably with a BER below a specific value. In this example we calculate
this input power by targeting a BER of 10-9 (which means a Q factor equals to 6) for a PIN photodetector
and an APD.

Figure 4-16: Layout - Receiver - Min. received power

This example shows that the receiver sensitivity by using the PIN is -34.65 dBm (Figure 4-17). When
using the APD with a gain equals to 3 the sensitivity increases to -38.25 dBm (Figure 4-18). Figure 4-19
and Figure 4-20, show the corresponding eye diagrams for the case when using PIN and APD receivers,
respectively.

Figure 4-17: Optical power when using PIN

63
MINIMUM RECEIVED POWER

Figure 4-18: Optical power when using APD

Figure 4-19: Eye diagram when using PIN

Figure 4-20: Eye diagram when using APD

64
 ER INDUCED SENSITIVITY DEGRADATION

ER Induced Sensitivity Degradation

Sensitivity degradation - ER.osp

A simple source of power penalty is related to the energy carried by 0 bits. Some power is emitted by
transmitters even in the off-state. In the project “Sensitivity degradation - ER.osp” (Figure 4-21), we have
an externally modulated laser where you can specify the extinction ratio (ER) at the modulator.

Figure 4-21: Layout - Sensitivity degradation - ER

In this project we vary the value of the ER and we calculate the Q factor at the receiver. The variation of
Q factor versus extinction ratio has been shown in Figure 4-22.

Figure 4-22: Q factor vs. extinctino ratio

65
ER INDUCED SENSITIVITY DEGRADATION

Notes:

66
Chapter 5 - Lightwave Systems
This section contains the following examples.

• Power Budget
• Power Management for Ring Networks
• Dispersion Induced Degradation

67
Notes:

68
 POWER BUDGET

Power Budget

System design - Power budget.osp

The purpose of power budget is to ensure that enough power will reach the receiver to maintain a reliable
performance during the entire system lifetime. The minimum average power required by the receiver is
the receiver sensitivity. The average launch power is generally specified for each transmitter with optical
powers expressed in dBm.

In order to estimate the maximum fiber length we should specify the output power of the transmitter and
the receiver sensitivity. We can also specify a system margin. The purpose of the system margin is to
allocate a certain amount of power to additional sources of power penalty that may develop during the
system lifetime.

The project “System design - Power budget.osp” shows (Figure 5-1) a system designed using a receiver
with -34.4 dBm sensitivity and an externally modulated transmitter with output power of -3.0 dBm.

Figure 5-1: Layout - System design - Power budget

Using a system margin of 0 dB the total loss allocated for the channel will be:

Channel Loss = Transmitter power - System Margin - Receiver Sensitivity

Therefore, Channel Loss = -3 + 34.4 = 31.4 dB

For a typical fiber with attenuation of 0.25 dB/km, the maximum distance the light can propagate in this
system would be 125.6 km. In this case, we are not including the dispersion effect which will limit the
system performance.

69
POWER BUDGET

Notes:

70
 POWER MANAGEMENT FOR RING NETWORKS

Power Management for Ring Networks

Optical power level management in metro networks - No amplifier.osp

Gain variation of EDFA depending on input power.osp

Optical power level management in metro networks - Lumped

amplification.osp

Optical power level management in metro networks - Unity gain.osp

Project files “Optical power level management in metro networks - No amplifier.osp”, “Gain variation of
EDFA depending on input power.osp”, “Optical power level management in metro networks - Lumped
amplification.osp”, and “Optical power level management in metro networks - Unity gain.osp” consider
the power budgeting for metro networks. They show alternative amplifier implementations in the network
and their effect to system performance. Details of calculations are also included in this example.

Case I: No amplifier option

Optical power level management in metro networks - No amplifier.osp

In this example, our goal is to design a basic ring network without amplification and discover the power
related issues.

Metro networks using ring topology are expected to have more dynamic traffic pattern compared to most
long-haul networks. In addition to this, they are expected to have optically transparent nodes with
minimum number of regenerators being used. Therefore, we need to consider all the optical power
variations. These variations may result from the wavelength dependent loss in fibers, wavelength
dependent gain in amplifiers, and channel to channel insertion loss variation in the
multiplexers/demultiplexers etc. These optical degradations accumulate along the links until the optical
termination. Due to dynamic nature of the network, only proper dynamic power level management can
reduce them. Let us consider first the loss elements.

Optical Add/Drop Multiplexers (OADMs) introduce insertion loss. It contains two parts: deterministic and
random. In fact, loss varies with the wavelength. The loss of fiber depends on fiber type and wavelength.
Variation of optical switch types and different path lengths in the network will result in a random loss,
which is difficult to predict.

There are several limiting factors in terms of power in the network. These are nonlinear effects, receiver
sensitivity, and losses in the fiber, OADMs, OXCs etc.

Typical loss values are given in Table 5-1.

71
POWER MANAGEMENT FOR RING NETWORKS

Table 5-1: Typical losses for components used in metro links

Symbol Description Loss value

Lmon Tap loss for monitoring 0.2 dB

Lconn Loss of a connector 0.25 dB

Lsplice Loss of a splice 0.15 dB

Lspan Loss of a fiber span 0.25 dB/km

Lmux Loss of a multiplexer 4 dB

Ldemux Loss of a demultiplexer 4 dB

Lfoadm Loss of a fixed OADM 2 dB

Lroadm Loss of a re-configurable OADM 10 dB

Total allowable loss depends on the transmitted power and receiver sensitivity. Receiver sensitivity is
also a function of bit rate, and in this example, the receiver has a sensitivity of -34.4 dBm. The allowable
loss is given by Lallowable = Ptransmitted – Psensitivity. Assuming two splices and six connectors total loss at
each node is given by Lnode = 2 x Lsplice + 6 x Lconn + Lfoadm.

Let us consider a typical WDM metro ring at 2.5 Gbps with 200 km in circumference with four intermediate
fixed OADMs as shown in Figure 5-2. In this network, node 1 and 4 communicate on channel 1 whereas
node 2 and 3 communicate on channel 2. Design project is given in “Optical power level management in
metro networks - No amplifier.osp” file.

Figure 5-2: A basic optical ring network with four nodes and two channels

Without any inline amplification, the allowable loss by using externally modulated source is Lallowable = 0
– (-34.4) = 34.4 dB (if we assume the transmitter output power to be 0 dBm). Node (including two splices
and six connectors) and span loss are Lnode = 2 x 0.15 + 6 x 0.25 + 2 = 3.8 dB and Lspan50km = 50 x 0.25
= 12.5 dB, respectively. Therefore, the total loss will be 12.5 x 4 + 3.8 x 4 = 65.2 dB.

Comparing the allowable loss and total loss in the network shows that it is not a good idea to make a
design without amplification. Of course, increasing power is another option, but nonlinear effects may

72
 POWER MANAGEMENT FOR RING NETWORKS

become important impairments at high powers. In this case, the required transmitter output power will be
65.2 – 34.4 = 30.8 dBm/channel. Note transmitter output power means the power after the modulator;
while CW laser power mentioned later on means the output power of the CW laser. Let us now discuss
these nonlinear effects [1,2].

Self-Phase Modulation

The effect of SPM is to chirp the pulse and broaden its spectrum. As a rough design guideline SPM
effects are negligible when P 0 < α ⁄ γ where γ = ( n 2 ω 0 ) ⁄ ( cAeff ) [W-1km-1] is the non-linearity coefficient,
P 0 is the peak power and α is the loss parameter. For the fiber we used γ = 1.3 W-1km-1. Therefore,
SPM effects can be negligible when the peak power is below 192.3 mW or 22.8 dBm.

Cross-Phase Modulation

When two or more pulses of different wavelengths propagate simultaneously inside fibers, their optical
phases are affected not only by SPM but also by XPM. Fiber dispersion converts phase variations into
amplitude fluctuations. XPM induced phase shift should not affect the system performance if the GVD
effects are negligible. As a rough estimate, the channel power is restricted with P th < α ⁄ ( γ ( 2N ch – 1 ) )
where N ch is the number of channels.

For a two channel system, limiting power is around 64.1 mW (18.1 dBm). Separation between channels
also affects the XPM. Increase in the separation will decrease the penalty which originates from XPM.

Four-Wave Mixing

FWM is the major source of nonlinear cross-talk for WDM communication systems. It can be understood
from the fact that beating between two signals generates harmonics at difference frequencies. If the
channels are equally spaced new frequencies coincide with the existing channel frequencies. This may
lead to nonlinear cross-talk between channels. When the channels are not equally spaced, most FWM
components fall in between the channels and add to overall noise. FWM efficiency depends on signal
power, channel spacing, and dispersion. If the GVD of the fiber is relatively high β 2 > 5 ps2/km, FWM
efficiency factor almost vanishes for a typical channel spacing of 50 GHz or more. If the channel is close
to zero dispersion wavelength of the fiber, considerably high power can be transferred to FWM
components. To reduce the effect of FWM to the system performance, one can use either uneven
channel spacing or use dispersion-management technique.

Stimulated Raman Scattering

Because of SRS, short wavelength channels act as pump for longer wavelength channels. In a WDM
system, the SRS will impair the system performance such that energy is transferred from the short-
wavelength to the long-wavelength channels. The Raman threshold for a single channel system is given
by Eq. 5-4:

16A eff
P th ≈ ---------------
- (5-4)
g R L eff

73
POWER MANAGEMENT FOR RING NETWORKS

where

– αL
L eff = 1-------------------
–e - (5-5)
α

and it is approximately equal to 1 ⁄ α for long fibers. In fact, the Raman threshold is a function of the
number of the channels and channel power. For a single channel system it is around 500 mW near 1.55
μm if g R = 1 x 10-13 m/W. For a 20 channel system P th exceeds 10 mW whereas it is around 1 mW for
a 70 channel system. For a single channel system, the SRS has little impact on system performance.

Stimulated Brillouin Scattering

It results from the interaction of light with acoustic wave in fiber and scatters power backward. Back
scattered light is downshifted in frequency from the original signal frequency. Threshold level depends
on effective core area and effective fiber length and is given below:

21A eff
P th ≈ ---------------
- (5-6)
g B L eff

Typical value for g B is around 5 x 10-11 m/W.The threshold value also depends on modulation format,
source linewidth and duration of pulse. Some values for threshold power are given as below:

(1) 9 dBm for CW light

(2) 12 dBm for externally modulated transmitter

(3) >18 dBm for externally modulated transmitter with source wavelength dither.

Generally speaking, SBS has little effect on system performance.

Figure 5-3 shows the Q factor for channel 1 at the receiver end versus the launched CW laser power,
when no amplifier is used. This figure shows that after about 20 dBm of launched CW laser power (as
predicted) nonlinearity becomes an issue and Q factor starts to decay.

74
 POWER MANAGEMENT FOR RING NETWORKS

Figure 5-3: Q factor versus launched CW laser power at node 4 when no amplifier is used

Figure 5-4 shows OSNR at the receiver with respect to launched CW laser power. Figure 5-5, Figure 5-
6, and Figure 5-7 show the eye diagram for the launched CW laser power of 13 dBm/channel, 20
dBm/channel, and 22 dBm/channel; respectively. From these figures we can conclude that, limiting factor
below 20 dBm region is mainly OSNR whereas above this value, it is nonlinearity. This example shows
that non-linearity may become an issue in no- amplifier ring networks.

Figure 5-4: OSNR versus launched CW laser power at node 4 when no amplifier is used

75
POWER MANAGEMENT FOR RING NETWORKS

Figure 5-5: Eye diagram at the receiver when the launched CW laser power is 13 dBm

Figure 5-6: Eye diagram at the receiver when the launched CW laser power is 20 dBm

76
 POWER MANAGEMENT FOR RING NETWORKS

Figure 5-7: Eye diagram at the receiver when the launched CW laser power is 22 dBm

Case II: Lumped amplification

Gain variation of EDFA depending on input power.osp

Optical power level management in metro networks - Lumped amplification.osp

Metro ring networks have more dynamic traffic pattern compared to most long-haul networks. In this
configuration we need to consider all the optical power variations resulting from the wavelength
dependent loss in the fiber, wavelength dependent gain in the amplifiers, channel to channel insertion
loss variation in the multiplexers/demultiplexers, etc. These optical degradations accumulate along the
links and may grow very fast depending on the path followed by the wavelength. Due to dynamic nature
of the network, only proper dynamic power level management can reduce them. Typical values for the
loss elements are given in the Table 5-1.

In this example we consider compensation of the loss in fibers and ADMs using amplifiers. To do so, let’s
first investigate the amplifier gain characteristics.

The gain of the optical amplifier changes with the input power. This is shown below (Figure 5-8) for two
cascaded EDFA amplifiers (each with a fiber length of 5 m, and pump power at 980 nm is 15 dBm), we
consider a 16.3 dB loss between the amplifiers to model the total span and node loss. Two channels (one
@ 193.1 THz of which the power is swept from -15 dBm to 10 dBm, another one @ 193.2 THz with a
constant power of 0 dBm) are used as the input signals. The gain of each channel versus the power of
channel 1 signal is shown in Figure 5-9.

77
POWER MANAGEMENT FOR RING NETWORKS

Figure 5-8: Layout - Gain variation of EDFA depending on input power

The project file is “Gain variation of EDFA depending on input power.osp”. When the span losses of
different spans in the ring are different, the input power of different amplifier will be different. Therefore,
depending on the path of the signal, each communication link between two nodes will experience
different amplification. Besides, the optical gain of an amplifier is also wavelength dependent. To
eliminate these effects, pump control, variable attenuator at the input or gain tilt compensating
amplifier/filter can be used.

Figure 5-9: Value and shape of optical gain depending on input power

Let us now consider a typical WDM metro ring at 2.5 Gbps with 200 km in circumference with four
intermediate fixed OADMs as shown in Figure 5-10. In this network, node 1 and 4 communicate on

78
 POWER MANAGEMENT FOR RING NETWORKS

channel 1 whereas node 2 and 3 communicate on channel 2. Design project is given in “Optical power
level management in metro networks - Lumped amplification.osp” file.

Figure 5-10: Ring netword layout with two amplifiers

Assuming a transmitter output power of -3 dBm (approximately a launched CW laser power of 0 dBm)
and receiver sensitivity of -34.4 dBm, the amount of amplification needed in the ring is:

Grequired = -34.4 + 3 x 0.25 x 50 + 3 x 3.8 – (-3) = 17.5 dB

This gain can be obtained by using two EDFAs each with a gain of 9 dB as shown in Figure 5-10. In this
configuration, two amplifiers are placed just before node 2 and node 4. However, this implementation of
lumped amplification prevents the flexibility of adding new nodes into the ring [2,3].

Measured signal power and OSNR at the receiver is given below (Figure 5-11) when two 9 dB amplifiers
are used. Figure 5-12 shows the eye diagram for this case, and the Q factor is 5.7.

Figure 5-11: Measured signal power and OSNR at the receiver

79
POWER MANAGEMENT FOR RING NETWORKS

Figure 5-12: Eye diagram at node 4 when two amplifiers with 9 dB gain each are inserted before node 4 and
node 2

Even though this configuration will do the job for a static configuration, it will not allow the addition of new
node or complex wavelength routing schemes where one signal may follow different paths depending on
the routing.

Case III: Unity gain approach

Optical power level management in metro networks_Unity gain.osp

In optical ring topology, we need to consider all the optical power variations resulting from the wavelength
dependent loss in fiber, wavelength dependent gain in amplifiers, and channel to channel insertion loss
variation in multiplexers/demultiplexers etc. These optical degradations accumulate along the links and
may grow very fast depending on the path followed by the wavelength. Due to dynamic nature of the
network, only proper dynamic power level management can reduce them. Typical values for the loss
elements are given in Table 5-1. We have also discussed the characteristics of optical amplifiers and
placement of amplifiers at certain points in network in Case II. In the two cases discussed above we have
seen that no amplifier is not a good choice for loss compensation since required high power may trigger
fiber nonlinearity. Even though lumped loss compensation seems to do the job in terms of good eye
diagram, it is not the best considering the dynamic nature of metropolitan networks.

In this example we will show an alternative approach, so called unity gain approach. In this approach,
each amplifier matches the loss of one node and span. The loss that need to be compensated will be the
node ( Lnode = 2 x 0.15 + 6 x 0.25 + 2 = 3.8 dB) and span (Lsapn50km = 50 x 0.25 = 12.5 dB) loss preceding
it. For this approach, an amplifier with modest gain of 16.3 dB at each node can be used. This approach
is only economically feasible with a low cost-compact optical amplifier yet the best one considering
dynamic routing in the network.

Let us now consider a typical WDM metro ring at 2.5 Gbps with 200 km in circumference with four
intermediate fixed OADMs as shown in Figure 5-13. In this network node 1 and 4 communicate on
channel 1 whereas node 2 and 3 communicate on channel 2. Loss of each span and node is
compensated at the node following the span. Gain of each amplifier is 16.3 dB. Design project is given

80
 POWER MANAGEMENT FOR RING NETWORKS

in “Optical power level management in metro networks - Unity gain.osp” file. This configuration is
scalable. Moreover, regardless of path followed by a certain wavelength, the power is always balanced
[1-4].

Figure 5-13: Ring network layout with amplifier at each node

Eye diagrams of channel 1 at node 4 are shown in Figure 5-14, when CW laser power is 0.6 dBm.
Figure 5-15, on the other hand, shows the eye diagram at node 4 when the CW laser power is -26 dBm.
Figure 5-16 shows the Q factor versus the power of the CW laser.

Figure 5-14: Eye diagram of channel 1 at node 4 when CW laser power is 0.6 dBm and bit rate is 2.5 Gbps

81
POWER MANAGEMENT FOR RING NETWORKS

Figure 5-15: Eye diagram of channel 1 at node 4 when CW laser power is -26 dBm and bit rate is 2.5 Gbps

Figure 5-16: Q factor versus CW laser power at node 4 when bit rate is 2.5 Gbps

References:

[1] S. Bigo et. al., “Investigation of cross-phase modulation limitation over various types of fiber
infrastructures”, IEEE Photon. Tech. Lett. 11, pp. 605, 1999.

[2] T.S. El-Bawab et. al., “Design considerations for transmission systems in optical metropolitan
networks”, Opt. Fib. Tech. 63, pp. 213, 2000

[3] R. Ramaswami and K.N. Sivarajan, Optical Networks: A Pactical Perspective, Morgan Kaufmann,
1998.

82
 POWER MANAGEMENT FOR RING NETWORKS

[4] S. Shiba et. al., “Optical power level management in metro networks”, NFOEC’01, 2001.

83
POWER MANAGEMENT FOR RING NETWORKS

Notes:

84
 DISPERSION INDUCED DEGRADATION

Dispersion Induced Degradation

Dispersion vs power.osp

Project file “Dispersion vs power.osp” shows that the dispersion broadening penalty of the system using
un-chirped NRZ input pulses is lower when the dispersion is positive. In this example, a 10 Gb/s NRZ
signal is transmitted through two nominally identical 150 km of fibers with opposite signs of dispersion.
For the first case, fiber has –2.27 ps/nm-km dispersion and for the second case it is +2.27 ps/nm-km.
Power of the CW laser is swept from 2 dBm to 20 dBm for these two cases. This simulation shows that
system penalty when a negative dispersion fiber is used is larger than when a positive dispersion fiber is
used. The penalty in the negative dispersion fiber arises from the pulse broadening due to SPM.
However, in the case of positive dispersion, nonlinear pulse compression interacts with dispersion,
resulting in negative power penalties. Up to about 18.3 dBm of input power, the Q factor increases with
the input power when the dispersion is positive (Figure 5-17).

Figure 5-17: Q factor versus input power for negative and positive dispersion fibers

Figure 5-18 compares negative and positive dispersion cases when the input power is 18 dBm.

85
DISPERSION INDUCED DEGRADATION

Figure 5-18: Eye diagrams (a )after 150 km of negative dispersion fiber (b) after 150 km of positive dispersion
fiber

Reference:

[1] Optical Fiber Telecommunications IIIa, Eds: I. Kaminov and T. Koch, chapter “Fiber Nonlinearities and
their Impact on Transmission Systems” by F. Forghieri, R. Tkach and A. Chraplyvy.

86
Chapter 6 - Multichannel Systems
This section contains the following examples.

• Fabry-Perot Filter
• Uniform FBG Filter
• AWG Demultiplexer
• Star Coupler
• Optical Cross-Connects
• Channel Multiplexing
• Channel Demultiplexing by FWM
• Interchannel Crosstalk at ADM in A Ring Network
• Subcarrier Multiplexing

87
Notes:

88
 FABRY-PEROT FILTER

Fabry-Perot Filter

Fabry-Perot filter.osp

The project “Fabry-Perot filter.osp“ (Figure 6-1) shows the transmission response of a Fabry- Perot filter
used as a tunable optical filter. The frequency spacing between two successive transmission peaks is
known as free spectral range (Figure 6-2).

Figure 6-1: Layout - Fabry-Perot filter

Figure 6-2: Fabry-Perot filter transmission response

89
FABRY-PEROT FILTER

Notes:

90
 UNIFORM FBG FILTER

Uniform FBG Filter

FBG filter transmission.osp

A separate class of optical filters makes use of the wavelength selectivity provided by a Bragg
grating. The project “FBG filter transmission.osp” (Figure 6-3) shows the transmission response of a
uniform FBG filter (Figure 6-4).

Figure 6-3: Layout - FGB filter transmission

Figure 6-4: FBG filter transmission response

91
UNIFORM FBG FILTER

Notes:

92
 AWG DEMULTIPLEXER

AWG Demultiplexer

AWG demultiplexer.osp

Fiber Bragg gratings can be used for making all fiber demultiplexers. The project “AWG
demultiplexer.osp” (Figure 6-5) shows an Array waveguide grating (AWG) demultiplexer.

Figure 6-5: Layout - AWG demultiplexer

The OSA at the AWG output shows the selected WDM channel. For the first channel, the AWG output
has been shown in Figure 6-6.

93
AWG DEMULTIPLEXER

Figure 6-6: AWG output

The AWG transmission response is shown in Figure 6-7.

Figure 6-7: AWG transmission response

94
 STAR COUPLER

Star Coupler

Star couplers.osp

The role of a star coupler is to combine the optical signals entering its multiple input ports and divide them
equally amongst its output ports. The project “Star couplers.osp” (Figure 6-8) shows a star coupler with
8 input ports, each port has a transmitter working at different wavelengths.

Figure 6-8: Layout - Star coupler

The output of the star coupler can be visualized by attaching an OSA at any of the output ports (Figure 6-
9).

95
STAR COUPLER

Figure 6-9: Output spectrum

96
 OPTICAL CROSS-CONNECTS

Optical Cross-Connects

OXC project.osp

The development of wide-area WDM networks requires wavelength routing that can reconfigure the
network while maintaining its transparent nature. Project “OXC project.osp” (Figure 6-10) shows an
optical cross connects with 2 input ports and 2 output ports, each port accommodates 4 wavelengths:

97
OPTICAL CROSS-CONNECTS

Figure 6-10: (a) Layout - OXC project (b) subsystem structure

(a)

(b)

98
 CHANNEL MULTIPLEXING

Channel Multiplexing

OTDM multiplexer.osp

In optical time-division multiplexing (OTDM) systems, several optical signals modulated at the bit rate B
using the same carrier frequency, are multiplexed optically to form a composite optical signal at a bit rate
N x B, where N is the number of multiplexed optical channels. This has been shown in the project “OTDM
multiplexer.osp” (Figure 6-11).

Figure 6-11: Layout - OTDM multiplexer

Each modulator generates short pulses with 10 Gb/s bit rate.

Figure 6-12: Pulses after modulator

99
CHANNEL MULTIPLEXING

The signals are combined in order to generate a 40 Gb/s signal at the multiplexer output (Figure 6-
13).

Figure 6-13: Pulses after multiplexer

The demultiplexer stage uses a modulator working with a 10 Gb/s clock. In this example the output signal
for the first channel is shown in Figure 6-14.

Figure 6-14: Pulses after the demultiplexer for the first channel

100
 CHANNEL DEMULTIPLEXING BY FWM

Channel Demultiplexing by FWM

DeMUX by FWM.osp

Project file “DeMUX by FWM.osp” generates a demultiplexing scheme that makes use of FWM in
nonlinear fiber. In this configuration, the optical TDM signal at 1550 nm is launched together with the
clock signal at a different frequency of 1552 nm into fiber. The clock signal acts as a pump for the FWM
process. A new pulse train is generated at a third frequency (1554 nm), which is an exact replica of the
channel that needs to be demultiplexed. An optical band-pass filter centered at 1554 nm is then used to
separate the demultiplexed channel from the optical TDM signal and clock signal.

Figure 6-15: Layout - DeMUX by FWM

101
CHANNEL DEMULTIPLEXING BY FWM

Notes:

102
 INTERCHANNEL CROSSTALK AT ADM IN A RING NETWORK

Interchannel Crosstalk at ADM in A Ring Network

Interchannel crosstalk at ADM in a ring network.osp

Project file “Interchannel crosstalk at ADM in a ring network.osp” analyzes the effect of linear interchannel
crosstalk in a ring network. The ring network in this example contains four nodes. One can see the
crosstalk effect by looking at the eye diagram as well as the optical spectrum. Crosstalk is originated from
the filters in ADMs.

Inter-channel cross-talk can arise from a variety of sources. One example of this is an optical filter or ADM
that selects one channel (drop) and passes others over the network. Another example is an optical switch
(switching different wavelengths) where cross-talk arises owing to imperfect isolation between the switch
ports. Other cross-talk types are amplifier induced, router induced, XPM induced, etc. [1]. A particular
signal can accumulate cross-talk from different elements and channels over the network. Cross-talk can
be reduced by using several techniques such as wavelength dilation and filter cascading. In this example,
we will investigate the effect of inter-channel cross-talk at ADM in a ring network. The project is given in
“Interchannel crosstalk at ADM in a ring network.osp”. This network contains 4 nodes and these nodes
communicate over two channels at 193 THz and 193.1 THz as shown in Figure 6-16. We have
considered a bit rate of 10 Gb/s. ADMs at each node is modeled by using WDM add and drop
components. WDM add and drop components are created by using 4th order Bessel filters. The ring is
ended with a ring control component that can circulate the signals in the ring for a given number of times.
The distance between the nodes is 12.5 km and we inserted an ideal amplifier just before node 2 to
compensate for the fiber loss.

Figure 6-16: A basic ring network with 4 nodes

103
INTERCHANNEL CROSSTALK AT ADM IN A RING NETWORK

Figure 6-17: (a) Layout of a ring network (b) structure of the subsystem “EM Tx/Rx - 16 dBm sensitivity at 10
Gbps, 4th order Bessel”

To show the effect of inter-channel cross-talk, we have swept the bandwidth of the filters in ADM of node
2 from 10 to 40 GHz. The eye diagrams and power spectra at node 2 for several bandwidths are given
in Figure 6-18. Of course this type of characteristic of the ADM at node 2 will also affect the received
signals at other nodes and other channel performances.

Figure 6-19 shows the eye diagram at node 3 which communicates on channel 1 at 193.1 THz.

104
 INTERCHANNEL CROSSTALK AT ADM IN A RING NETWORK

Figure 6-18: Eye diagrams and corresponding power spectra at node 2 when bandwidth of the filters in node
2 is (a) 40 GHz (b) 17.5 GHz and (c) 10 GHz

(a) (b) (c)

Figure 6-19: Eye diagram of the signal dropped at node 3 that communicates on channel 1 at 193.1 THz

References:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

[2] T. Gyselings et. al, “Crosstalk analysis of multi-wavelength optical cross connects”, J. Light. Tech. 17,
pp. 1273, 1999.

105
INTERCHANNEL CROSSTALK AT ADM IN A RING NETWORK

[3] R. Ramaswami and K.N. Sivarajan, Optical Networks: A practical Perspective, Morgan Kaufmann,
1998.

106
 SUBCARRIER MULTIPLEXING

Subcarrier Multiplexing

Subcarrier multiplexing.osp

Project file “Subcarrier multiplexing.osp” generates an analog SCM lightwave system designed with a
single optical carrier at 193.1 THz. Three sinusoidal signals (0.5+0.5*sin(ω0*t)) with frequencies of 56
MHz, 116 MHz, and 176 MHz are first combined in electronic domain and then used to modulate the
amplitude of an optical carrier. Light is sent to fiber and at the receiver end, subcarriers are separated by
filters.

Figure 6-20: Layout - Subcarrier multiplexing

107
SUBCARRIER MULTIPLEXING

Notes:

108
Chapter 7 - Optical Amplifiers
This section contains the following examples.

• EDFA Basic Concepts

• Booster Amplifier
• In-Line Amplifier
• Preamplifier
• SLA Cascaded in-line Amplifiers: Soliton Pattern Effect at 10 Gb/s in SMF
• SLA Gain Recovery
• SLA Gain Saturation of Chirped and Super Gaussian Pulses
• SLA Gain Saturation of Gaussian Pulses
• Raman Gain and Bandwidth
• Raman Gain Saturation
• Broadband Flat Gain Raman Amplifier
• Pumping Requirements
• Multichannel Amplification
• Optical Preamplifiers
• Power Boosters
• Metro EDFA
• Cascaded In-Line Amplifiers
• EDFA Dynamic

109
Notes:

110
 EDFA BASIC CONCEPTS

EDFA Basic Concepts

EDFA Basic Concepts - Gain Spectrum.osp

EDFA Basic Concepts - Gain Saturation.osp

EDFA Basic Concepts - Amplifier Noise.osp

The sample files in this example show the basic characterization of the Erbium doped fiber amplifier. We
have considered the three design versions namely, (a) Gain Spectrum (b) Gain Saturation and (c)
Amplifier noise. These enable the characterization of the gain, noise figure and output power under
unsaturated and saturated conditions. These examples refer to the topics: Gain Spectrum and
Bandwidth, Gain Saturation, and Amplifier Noise, respectively.

In these examples the amplifier performance is characterized by using sweeps for the input parameters
of the amplifier. In examples Gain Spectrum and Amplifier Noise, the signal wavelength is swept; while
in example Gain Saturation, we sweep the input power for the cases: small signal input power (-30 dBm)
and saturated input signal power (0 dBm).

For each design version we have plotted three graphs showing the Output Signal Power, Gain, and Noise
Figure as a function of the sweep parameter. In the project, some components settings can be modified
and the simulations can be repeated in order to analyze the amplifier performance.

The input parameters which are critical in the numerical solution of coupled rate and propagating
equations are displayed in Erbium doped fiber component. The gain and noise figure are calculated
using the power results. Figure 7-1 shows the cross-section file used in the projects.

111
EDFA BASIC CONCEPTS

Figure 7-1: Cross-section (absorption curve in black and emission curve in red) file used in the Erbium
doped fiber component

Figure 7-2: Layout - EDFA basic concepts

The basic layout used in the calculations for each version is shown in Figure 7-2.

Version - Gain Spectrum

In this first version named Gain Spectrum the signal wavelength of the CW laser signal is swept from
1525 nm to 1600 nm which enables evaluating the spectral gain and the bandwidth of the amplifier.

The spectral gain calculated for -20 dBm of input signal power is shown in Figure 7-3.

112
 EDFA BASIC CONCEPTS

Figure 7-3: Gain versus signal wavelength (given in [nm] unit) with signal input power equals to -20 dBm

An additional graph that can be checked in the Gain Spectrum version is the output power versus
signal wavelength as shown in Figure 7-4.

Figure 7-4: Signal output power versus signal wavelength (given in [nm] unit) calculated with a signal input
power equals to -20 dBm

Version - Gain Saturation

The version Gain Saturation enables evaluating the amplifier performance as a function of the signal
input power. The signal input power is swept from small signal to large signal regime, i.e. - 40 dBm to 0
dBm. The amplifier performance in terms of gain, output power and noise figure are shown in Figure 7-
5, Figure 7-6 and Figure 7-7, respectively.

113
EDFA BASIC CONCEPTS

Figure 7-5: Gain vs. signal input power

Figure 7-6: Amplified signal power vs. input signal power

Figure 7-7: Noise figure vs. input signal power

Version - Amplifier Noise

114
 EDFA BASIC CONCEPTS

The signal wavelength is the parameter swept in this version, which enables calculating the spectral
noise of the amplifier. The total noise power versus signal wavelength is shown in Figure 7-8. Signal and
pump power can be changed for further evaluation of the amplifier performance.

Figure 7-8: Amplifier noise characterization

Figure 7-9 shows the ASE spectrum stored after calculating each iteration of the signal wavelength.

Figure 7-9: Noise bins as a function of wavelengths used in the calculations. Each curve in this graph is
resulted from different iterations.

115
EDFA BASIC CONCEPTS

Notes:

116
 BOOSTER AMPLIFIER

Booster Amplifier

Booster Amplifier.osp

The project file “Booster amplifier.osp” shows the characterization of a booster/power amplifier setup in
a single heavily doped erbium-doped fiber, bidirectionally pumped by two 980 nm-pump lasers. In this
case, large signal input power is considered because large signal input power contributes to obtain high
output power, which is a requirement to booster/power amplifiers. As a consequence, moderate gain will
be observed in this case. The low noise figure requirement for EDFAs is not so tight here, where an
increase in NF can be tolerated. It is important to mention that typical configurations of booster amplifiers
include multiple Er-doped fiber stages.

Figure 7-10 shows the layout of the booster amplifier. Bidirectional pump was used to produce the typical
pump scheme observed in booster amplifiers. The optical spectrum analyzer connected at the output
port 1 shows the amplified signal obtained after running this project file. The component dual port WDM
analyzer gives the calculated results performed to all the propagating signals and pump.

The output power, gain and noise figures calculated as a function of signal input power are shown in
Figure 7-11, Figure 7-12, and Figure 7-13, respectively. It is possible to observe the gain being
compressed as a function of the increasing signal input power as shown in Figure 7-12.

The noise figure curve is calculated by sweeping the signal input power to the booster amplifier assuming
identical input parameters. There is a region in the curve where NF is minimized as a function of the
signal input power.

Figure 7-10: Layout - Booster amplifier

117
BOOSTER AMPLIFIER

Figure 7-11: Signal output power vs. signal input power calculated for the booster amplifier

Figure 7-12: Gain vs. signal input power calculated for the booster amplifier

Figure 7-13: Noise figure vs. signal input power calculated for the booster amplifier

The power of both pump lasers (co- and counter-propagating) as well as the Er-doped fiber length
can be modified.

118
 IN-LINE AMPLIFIER

In-Line Amplifier

In-line Amplifier.osp

In-line Amplifier - Co & Counter-propagating Pump.osp

The project file “In-line amplifier.osp” shows the characterization of an in-line amplifier setup in a single
erbium-doped fiber stage, pumped by one 980 nm- pump laser.

In this case, small signal input power is considered to obtain high gain and small noise figure values. The
signal input power is swept from -40 dBm to -20 dBm that enables checking the gain, noise figure,
output power and OSNR as a function of signal input power.

After sweeping iterations over the signal input power, the amplifier performance can be checked in the
graphs displayed on the report page, as shown in Figure 7-14.

Figure 7-14: Graphs presenting the inline amplifier performance setup in a co-propagating pump scheme

119
IN-LINE AMPLIFIER

Comparison of two different pump schemes that consider co- and counter-propagating pump are
available in “In-line Amplifier - Co & Counter-propagating Pump.osp” file, of which the layout is shown in
Figure 7-15. The performance of the inline amplifier with co- and counter-propagating pump schemes
can be seen in the graphs of gain, noise figure and output power displayed in the report page.

Figure 7-15: Layout - In-line amplifier with co- & counter-propagating pump schemes

The amplifier performance in case of co- and counter-propagating pump schemes can be compared
through graphs available. Pump wavelength equal to 980 nm was considered in both cases. The
gain versus pump power is shown in Figure 7-16 considering the co- and counter-pump schemes that
enable evaluating the most efficient pump scheme for the EDFAs.

Figure 7-16: Comparing gain performance of an EDFA setup in co- and counter-propagating pump schemes

120
 PREAMPLIFIER

Preamplifier

Preamplifier.osp

The project file “Preamplifier.osp” shows the characterization of an in-line amplifier setup in a single
erbium-doped fiber stage, pumped by one pump laser in a co-pump scheme.

In this case, small signal input power is considered to obtain high gain and small noise figure values. The
erbium doped fiber length is swept from 6 m to 15 m that enables checking the gain, noise figure, output
power and OSNR as a function of fiber length. The pump wavelengths considered in the layout are 980
nm and 1480 nm.

In the two different versions, signal input power is -35 dBm at 1550 nm, respectively. The performance
of the preamplifier for the pump wavelength can be seen in the graphs of gain, noise figure and output
power displayed in the report page.

The basic layout used in this preamplifier example is shown in Figure 7-17.

Figure 7-17: Layout - Preamplifier

The gain for the preamplifier is shown in Figure 7-18. Different pump power, signal wavelength, and fiber
parameters can be considered to perform additional simulations.

121
PREAMPLIFIER

Figure 7-18: Gain vs. erbium doped fiber length considering 980 nm (black curve) and 1480 nm (red curve)
as pump wavelength

122
 SLA CASCADED IN-LINE AMPLIFIERS: SOLITON PATTERN EFFECT AT 10 GB/S IN SMF

SLA Cascaded in-line Amplifiers: Soliton Pattern

Effect at 10 Gb/s in SMF

SLA cascaded in-line amplifiers soliton pattern effect 10 Gb/s in

SMF.osp

One of the possibilities to upgrade existing network of already installed standard single mode optical
fibers operating at 1.3 μm is the use of SOA. The advantages of using SOA are the low dispersion of the
SMF at this carrier wavelength and attractive features of semiconductor optical amplifiers. However,
there are two major negative factors in the utilization of SOA as in-line optical amplifiers: (a) gain-
saturation effects, which lead to non-equal amplification of pulses in the pattern (so called pattern effect)
and (b) the chirp that pulse acquires after amplification.

The aim of this project is to demonstrate the pattern effect at 10 Gb/s transmission over a 500
km optical link consisting of SMF and in-line SOAs [1]. We consider the parameters similar to those in [1].

Figure 7-19: Layout - SLA cascaded in-line amplifiers - soliton pattern effect at 10 Gb/s in SMF

We fix B= 10 Gb/s -> TB = 100 ps. The sequence length is 16 bit. The carrier wavelength of the pulse
is λ ~ 1300 nm. The TFWHM = 20 ps -> T0 = 0.567 x TFWHM = 11.34 ps. The input peak power is 21.7
mW.

We will consider SMF with a length o f 50 km and loss of 0.4 dB/km. The parameter k2 (= - λ2D)/(2πc))
~ -1.5 (ps2/km) -> D ~ 1.67 (ps/nm.km) -> LD (= T02/|k2|) ~ 85 km. After each fiber the signal will be
amplified with SOA, therefore LA ~ 50 km. Note that the condition LA < LD is satisfied.

123
SLA CASCADED IN-LINE AMPLIFIERS: SOLITON PATTERN EFFECT AT 10 GB/S IN SMF

The Kerr nonlinearity coefficient γ equals to n2ω0/cAeff. For the fixed values of nonlinear refractive index
n2 = 2.6 x 10-20 m2/W, ω0/c = 2π/λ = 2π/1.3 x 10-6 m-1, and Aeff = 62.8 μm2, γ will be 2 W-1km-1.

Figure 7-20, Figure 7-21, Figure 7-22, and Figure 7-23 show the initial pattern of pulses, and the same
pattern of pulses after 200, 350 and 500 km of SMF with periodic amplification at every 50 km using SOA.

Figure 7-20: Initial pattern of pulses

Figure 7-21: Pulses after 200 km of SMF with periodic amplification using SOA at every 50 km

124
 SLA CASCADED IN-LINE AMPLIFIERS: SOLITON PATTERN EFFECT AT 10 GB/S IN SMF

Figure 7-22: Pulses after 350 km of SMF with periodic amplification using SOA at every 50 km

Figure 7-23: Pulses after 500 km of SMF with periodic amplification using SOA at every 50 km

Wha we can see form these figures is the pattern effect, which leads to reduction in the gain of the pulses
after previous pulses.

References:

[1] M. Settembre, F. Matera, V. Hagele, I. Gabitov, A. W. Mattheus, and S. Turitsyn, “Cascaded optical
communication systems with in-line semiconductor optical amplifiers”, Journal of Lightwave Technology,
vol.15, pp. 962-967, 1997.

[2] F. Matera and M. Settembre, “Study of 1.3 μm transmission systems on standard step-index fibers
with semiconductor optical amplifiers”, Optics communications, vol. 133, pp.463-470, 1997.

[3] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, Inc., 2010.

125
SLA CASCADED IN-LINE AMPLIFIERS: SOLITON PATTERN EFFECT AT 10 GB/S IN SMF

Notes:

126
 SLA GAIN RECOVERY

SLA Gain Recovery

SLA gain recovery.osp

When the pulse width becomes comparable to the carrier lifetime, the saturated gain has time to recover
during the pulse. The recovery effect influences the shape and spectrum of the amplified pulse.

The aim of this project is to study this influence. Three subtopics will be considered:

a) Partial gain recovery;

b) Complete gain recovery

c) Appearance of gain-saturation induced self-phase modulation.

SOA parameters considered are : carrier lifetime τC ~ 1.4 ns, G = 0.22 and Esat = 5 pJ. Table 7-1 shows
the physical parameters of SOA. The unsaturated single pass amplifier gain ~ 30 dB and Inner loss =
0 are used to make the comparison with [1].

Table 7-1: Semiconductor optical amplifier parameters (physical)

Figure 7-24 shows the layout of the project.

127
SLA GAIN RECOVERY

Figure 7-24: Layout - SLA gain recovery

(a) Partial gain recovery

As the carrier lifetime of SOA is: τC ~ 1.4 ns, we will consider here pulses with widths smaller than 1.4
ns. Keeping the ratio Ein / Esat = 0.1, we will investigate the influence of the increase of the pulse
duration which will approach the carrier lifetime but will remain smaller than it.

The bit rate B = 1 Gbps implies TB = 1ns.

We will consider three pulses (pulse 1,2 and 3) with widths of 0.25 ns, 0.5 ns and 1 ns, respectively. In
all the three cases, the ratio Ein / Esat = 0.1 will be kept constant.

The parameters for pulse 1 are: duty cycle = 0.25, TFWHM ~ 0.25 ns -> T0 ~ 0.14 ns, and Pin = 3.5 mW.
Therefore, T0 /τC ~ 0.1.

Figure 7-25, Figure 7-26, Figure 7-27, and Figure 7-28 show the shape and spectrum of the initial and
amplified pulse 1.

Figure 7-25: Shape of initial pulse 1

128
 SLA GAIN RECOVERY

Figure 7-26: Spectrum of initial pulse 1

Figure 7-27: Shape of amplified pulse 1

Figure 7-28: Spectrum of amplified pulse 1

129
SLA GAIN RECOVERY

The parameters for pulse 2 are: duty cycle = 0.5, TFWHM ~ 0.5 ns -> T0 ~ 0.28 ns, and Pin = 1.75 mW.
Therefore, T0 /τC ~ 0.2.

Figure 7-29, Figure 7-30, Figure 7-31, and Figure 7-32 show the shape and spectrum of the initial and
amplified pulse 2.

Figure 7-29: Shape of initial pulse 2

Figure 7-30: Spectrum of initial pulse 2

130
 SLA GAIN RECOVERY

Figure 7-31: Shape of amplified pulse 2

Figure 7-32: Spectrum of amplified pulse 2

The parameters for pulse 3 are: duty cycle = 1, TFWHM ~ 1 ns -> T0 ~ 0.567 ns, and Pin = 0.875 mW.
Therefore, T0 /τC ~ 0.4.

Figure 7-33, Figure 7-34, Figure 7-35, and Figure 7-36 show the shape and spectrum of the initial and
amplified pulse 3.

131
SLA GAIN RECOVERY

Figure 7-33: Shape of initial pulse 3

Figure 7-34: Spectrum of initial pulse 3

Figure 7-35: Shape of amplified pulse 3

132
 SLA GAIN RECOVERY

Figure 7-36: Spectrum of amplified pulse 3

Following properties of the three amplified pulses considered above can be observed. First, we see that
by increasing the pulse width and keeping the ratio Ein / Esat = 0.1, the pulse shape becomes broader
and less asymmetrical than the input pulse. At the same time increasing the pulse width, the spectrum
becomes less asymmetrical and the spectral shift reduces. Note the behavior of the phase that is shown
in the figures of the shapes of amplified pulses. As the phase of amplified pulse gives the information for
the gain we can see that for such pulses the gain doesn’t have enough time to recover completely.

(b) Complete gain recovery ( the quasi CW operation )

As the carrier lifetime of SOA is: τC ~ 1.4 ns, we will consider pulses with widths larger than 1.4 ns. The
bit rate B = 100 Mbps implies TB = 10 ns.

We will consider three pulses (pulse 4, 5 and 6) with widths of 2.5 ns, 5 ns and 10 ns, respectively. In
all the three cases the ratio Ein / Esat = 0.1 will be kept constant.

The parameters for pulse 4 are: duty cycle = 0.25, TFWHM ~ 2.5 ns, T0 ~ 1.42 ns, and Pin = 0.15 mW.
Therefore, T0 /τC ~ 1.8 and Pin / Psat = 0.04. Figure 7-37, Figure 7-38, Figure 7-39, and Figure 7-40 show
the shape and spectrum of the initial and amplified pulse 4.

Figure 7-37: Shape of initial pulse 4

133
SLA GAIN RECOVERY

Figure 7-38: Spectrum of initial pulse 4

Figure 7-39: Shape of amplified pulse 4

Figure 7-40: Spectrum of amplified pulse 4

The parameters for pulse 5 are: duty cycle = 0.5, TFWHM ~ 5 ns -> T0 ~ 2.83 ns, and Pin = 0.28 mW.
Therefore, T0 /τC ~ 3.6 and Pin / Psat = 0.08.

134
 SLA GAIN RECOVERY

Figure 7-41, Figure 7-42, Figure 7-43, and Figure 7-44 show the shape and spectrum of the initial and
amplified pulse 5.

Figure 7-41: Shape of initial pulse 5

Figure 7-42: Spectrum of initial pulse 5

135
SLA GAIN RECOVERY

Figure 7-43: Shape of amplified pulse 5

Figure 7-44: Spectrum of amplified pulse 5

The parameters for pulse 6 are: duty cycle = 1, TFWHM ~ 10 ns -> T0 ~ 5.67 ns, and Pin = 0.736 mW.
Therefore, T0 /τC ~ 7.1 and Pin / Psat = 0.2.

Figure 7-45, Figure 7-46, Figure 7-47, and Figure 7-48 show the shape and spectrum of the initial and
amplified pulse 6.

136
 SLA GAIN RECOVERY

Figure 7-45: Shape of initial pulse 6

Figure 7-46: Spectrum of initial pulse 6

Figure 7-47: Shape of amplified pulse 6

137
SLA GAIN RECOVERY

Figure 7-48: Spectrum of amplified pulse 6

We see from the properties of the last three amplified pulses that increasing the pulse width well above
the carrier lifetime leads to much more symmetrical shapes of the amplified pulses. They become
also much broader than the shapes of the corresponding initial pulses. At the same time increasing the
pulse width the spectrum of the amplified pulses become less asymmetrical and the red shift
continuously reduce. The phase is shown in the figures of the shapes of amplified pulses. We see now
that for such pulses the gain has enough time to recover completely.

(c) Appearance of gain-saturation induced self- phase modulation

Now we will compare the obtained results for the pulse 6 (for which Pin / Psat = 0.2) with that of Pin
/ Psat = 0.4. To make this we will increase the initial power 10 times. The compared Results for the
shapes and spectrum of the amplified pulses are shown in Figure 7-49 and Figure 7-50, respectively.

Figure 7-49: Comparision of results for the shapes of amplified pulses

138
 SLA GAIN RECOVERY

Figure 7-50: Comparision of results for the spectrum of the amplified pulses

Obtained spectrum for the complete gain recovery example exhibits the shape that is very similar to the
case of pulse propagation in a medium with Kerr nonlinearity [2]. These spectrum show the concept of
gain saturation induced self-phase modulation [1]. Note that such kinds of spectrum are possible only in
the case of pulses longer than the carrier lifetime.

References:

[1] G.P. Agrawal and N.A. Olsson, “Self-phase modulation and spectral broadening of optical pulses in
semiconductor laser amplifiers”, IEEE Journal of Quantum electronics, vol. 25, pp.2297-2306, 1989.

[2] G.P. Agrawal, Nonlinear Fiber optics, third edition, Academic Press, 2001.

139
SLA GAIN RECOVERY

Notes:

140
 SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

SLA Gain Saturation of Chirped and Super Gaussian

Pulses

SLA gain saturation of chirped and super-gaussian pulses.osp

In this project we will study the effect of gain saturation induced self-phase modulation for the
amplification of optical pulses. However, we will concentrate on the pulses with different shape and
initial frequency shift. The chirped Gaussian input pulses are the pulses that are usually produced from
directly modulated semiconductor laser.

We will consider the pulses with pulse width much shorter than the carrier lifetime. The carrier
wavelength of the Gaussian pulse is 1.55 μm.

In the first part of this project we will consider chirped Gaussian pulses, and in the second part, the
super Gaussian pulses.

Saturation of amplification of chirped Gaussian pulse

For the default values of physical parameters of SOA, we obtain the following values of the
parameters:

Carrier lifetime τC ~ 1.4 ns

Saturation energy Esat ~ 3.7 pJ

Amplification factor G0 = 29 dB.

Note that as a typical value of the linewidth enhancement factor, α = 5 is chosen.

Gaussian pulse with the parameters: Energy E0 ~ 0.73 pJ, TFWHM = 14 ps -> P ~ 50 mW and the Chirp
linewidth enhancement factor of 5 is considered.

To get the required carrier wavelength and power the parameters of the chirped and super optical
Gaussian pulse generators have been fixed. To get chirped and super Gaussian pulse with desired
TFWHM = 14 ps, the global parameters have been chosen as shown in Table 7-2.

141
SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

Table 7-2: Global parameters

As a result, we get, T0 /τC ~ 0.006 and E0 / Esat ~ 0.2. According to the first ratio the input pulse width
is much smaller than the carrier lifetime. The meaning of the second one is that pulse energy is
comparable with the SOA saturation energy.

The project in which the amplification of the Gaussian pulse with SOA will be analyzed is shown in
Figure 7-51.

Figure 7-51: Layout - SLA gain saturation of chirped and super Gaussian pulses

The shape and spectrum of the initial chirped Gaussian pulse are shown in Figure 7-52 and Figure 7-53.
Note that the initial negative chirp is also shown in Figure 7-52.

142
 SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

Figure 7-52: Shape of the initial chirped Gaussian pulse

Figure 7-53: Spectrum of the initial chirped Gaussian pulse

Note that the frequency in Figure 7-52 decreases with time across the pulse, and this is usually referred
to as negative chirp.

Figure 7-54 and Figure 7-55 show the shape and spectrum of the amplified chirped pulse.

143
SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

Figure 7-54: Shape of the amplified chirped pulse

Figure 7-55: Spectrum of the amplified chirped pulse

Figure 7-54 reveals already expected sharpening of the leading edge. Note the different form of the
output pulse spectrum and the reduced red shift in comparison with the amplified Gaussian pulse.

In Figure 7-56, phase of the pulse after amplification is shown together with its shape. In Figure 7-57, the
deformation of the initial negative chirp after amplification can be easily seen.

144
 SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

Figure 7-56: Shape and phase of the pulse after amplification

Figure 7-57: Deformation of the initial negative chirp after amplification

For negative initial chirp, the gain saturation induced red shift reduces. For the opposite sign of initial
chirp the spectrum shifts even more to the red side than for the case of zero initial chirp.

Saturation of amplification of super-Gaussian pulse

The additional pulse parameter which is considered for the super Gaussian pulse is m = 3 (Note that the
pulse now will be without initial chirp).

The shape and spectrum of the initial super Gaussian pulse are shown in Figure 7-58 and Figure 7-59.

145
SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

Figure 7-58: Shape of the initial super Gaussian pulse

Figure 7-59: Spectrum of the initial super Gaussian pulse

Figure 7-60 and Figure 7-61 show the shape and spectrum of the amplified super Gaussian pulse.

Figure 7-60: Shape of the amplified super Gaussian pulse

146
 SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

Figure 7-61: Spectrum of the amplified super Gaussian pulse

The output pulse has a long tail on the trailing edge and appears to be narrower than the input pulse.
The situation is very different from the case of amplification of Gaussian pulse where the TFWHM of the
output pulse was larger than the input one.

The spectrum of the output spectrum has a multi-peak structure and very well expressed red shift.
Comparison of this spectrum with the one of the amplified Gaussian pulse shows a clear difference.
Therefore the form of the spectrum and amount of the red shift strongly depends on the initial pulse
shape.

In Figure 7-62, phase of the pulse after amplification is shown together with its shape. In Figure 7-63, the
formation of the chirp after amplification can be easily seen.

Figure 7-62: Shape and phase of the pulse after amplification

147
SLA GAIN SATURATION OF CHIRPED AND SUPER GAUSSIAN PULSES

Figure 7-63: Formation of the chirp after amplification

In this project we described the basic properties of the chirped and super Gaussian pulse amplification
with SOA component in OptiSystem software. We have shown that the shape and spectrum distortions
with the amplification of SOA strongly depends on the shape and initial frequency modulation of the
pulse. These properties can be very important in the case when pulses produced by directly modulated
semiconductor lasers are amplified.

The obtained results are in complete agreement with results published in [1] and [2]:

References:

[1] G.P. Agrawal and N.A. Olsson, “Self-phase modulation and spectral broadening of optical pulses in
semiconductor laser amplifiers”, IEEE Journal of Quantum electronics, vol. 25, pp.2297-2306, 1989.

[2] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

148
 SLA GAIN SATURATION OF GAUSSIAN PULSES

SLA Gain Saturation of Gaussian Pulses

SLA gain saturation of Gaussian pulses.osp

SLA gain saturation of Gaussian pulses - Power.osp

SLA gain saturation of Gaussian pulses - Conf.osp

SLA gain saturation of Gaussian pulses - Loss.osp

Amplification of ultra-short optical pulses in SOA produces considerable spectral broadening and
distortion due to the non-linear phenomenon of self-phase modulation. The physical mechanism behind
SPM is gain saturation, which leads to intensity dependent changes of refractive index in response to
variations in carrier density. Signal-gain saturation in SOA is caused by a reduction of the population
inversion in the active layer due to an increase in stimulated emission. Gain saturation characteristics are
especially important in optical repeaters and multi-channel amplifiers which require high-power
operation.

In this project the effect of gain saturation induced self-phase modulation on the amplification of Gaussian
pulse will be considered. We will consider the pulses with pulse width much shorter than the carrier
lifetime. The carrier wavelength of the Gaussian pulse is 1.55 μm.

For these default values of physical parameters of SOA we obtain the following values of parameters:

The carrier lifetime τC ~ 1.4 ns

The saturation energy Esat ~ 3.7 pJ

The amplification factor G0 = 29 dB.

As a typical value of the linewidth enhancement factor, α = 5 is chosen. Note that we have chosen
the internal losses to be zero. Such choice allows a comparison to be done with existing analytical
solution (see the referred paper at the end of this section). We will also analyze the influence of the
internal losses.

Gaussian pulse with following parameters is considered: Energy E0 ~ 0.73 pJ, TFWHM = 14 ps, and
P ~ 50 mW. Note the large value of the pulse initial power. It allows in a best way to demonstrate the
basic qualitative features of gain saturation induced properties of the pulse amplification. The
influence of the value of the pulse initial power will be considered later in this section.

The global parameters have been chosen to get a Gaussian pulse with desired TFWHM = 14 ps. As a
result we get T0 /τC ~ 0.006 and E0 / Esat ~ 0.2.

According to the first ratio the input pulse width is much smaller than the carrier lifetime. The meaning of
the second one is that pulse energy is comparable with the SOA saturation energy.

149
SLA GAIN SATURATION OF GAUSSIAN PULSES

The project layout for amplification of the Gaussian pulse with SOA is shown in Figure 7-64.

Figure 7-64: Layout - SLA gain saturation of Gaussian pulses

To illustrate the appearance of the gain saturation induced effects first we will show the amplification of
very weak pulse. The shape and spectrum of the weak Gaussian pulse are shown in Figure 7-65 and
Figure 7-66.

Figure 7-65: Shape of the weak Gaussian pulse

150
 SLA GAIN SATURATION OF GAUSSIAN PULSES

Figure 7-66: Spectrum of the weak Gaussian pulse

Figure 7-67 and Figure 7-68 show the shape and spectrum of the amplified weak pulse.

Figure 7-67: Shape of the amplified weak pulse

Figure 7-68: Spectrum of the amplified weak pulse

151
SLA GAIN SATURATION OF GAUSSIAN PULSES

Comparison of the above figures shows that the pulse is amplified in unsaturated region. No qualitative
changes in the spectrum of the amplified pulse could be observed.

Figure 7-69 and Figure 7-70 show the shape and spectrum of the initial Gaussian pulse, which will
be amplified in saturation region.

Figure 7-69: Shape of the initial Gaussian pulse

Figure 7-70: Spectrum of the initial Gaussian pulse

Figure 7-71 and Figure 7-72 show the shape and spectrum of the amplified pulse.

152
 SLA GAIN SATURATION OF GAUSSIAN PULSES

Figure 7-71: Shape of the amplified pulse

Figure 7-72: Spectrum of the amplified pulse

Figure 7-71 reveals that the amplified pulse becomes asymmetric so that its leading part is sharper
compared with the trailing edge. Note that sharpening of the leading edge is a common feature of all
amplifiers.

The pulse spectrum of the amplified pulse, however, reveals features that are particular for the SOA.

The spectrum develops a multi-peak structure and the dominant peaks shift toward longer (red)
wavelengths. This red shift increases with the increase of the amplification factor. The red shift observed
from this figure is ~ 0.001 μm (~ 120 GHz).

In Figure 7-73, the phase of the pulse after amplification is shown together with its shape. The important
characteristic that has to be mentioned is that the phase follows the time evolution of the gain.
We see that for such short pulse, gain has no time to recover (remember T0 /τC ~ 0.006).

153
SLA GAIN SATURATION OF GAUSSIAN PULSES

Figure 7-73: Phase and shape of the pulse after amplification

In Figure 7-74, the gain saturation induced chirp across the pulse is shown together with its shape. Note
that the initial Gaussian pulse was unchirped.

Figure 7-74: Gain saturation induced chirp across the pulse

The gain saturation induced chirp is negative across the entire pulse, therefore the instantaneous
frequency is down shifted (the red shift) from the incident frequency. Nevertheless because of the fact
that induced chirp increases towards the trailing part of the pulse, such chirp is known as positive one.
From the theory of propagation of phase modulated pulses in the medium with group velocity dispersion
[3], it follows that appearance of positive chirp can be used for pulse compression in the medium with
anomalous group velocity dispersion. Let us now consistently analyze the influence of the initial power,
optical confinement factor and inner losses on the observed properties. Note that in model used in [1,2],
the inner losses have been neglected.

First we will see how the shape and spectrum of the amplified Gaussian pulse (for the same pulse
and SOA parameters as above) change as a function of the initial pulse power. The values of the initial
power considered are P = 5, 10, 20, 30, and 50 mW. In the first case we consider E0 / Esat ~ 0.02.

To get these results, mode sweep can be used.

154
 SLA GAIN SATURATION OF GAUSSIAN PULSES

In Figure 7-75 and Figure 7-76, the shape and spectrum of the amplified Gaussian pulse are shown for
different initial powers.

Figure 7-75: Shape of the amplified Gaussian pulse for different initial powers

Figure 7-76: Spectrum of the amplified Gaussian pulse for different initial powers

As we could expect increasing the ratio E0 / Esat (for fixed pulse width) leads to increase in the number
of the peaks in the spectrum along with the increase of the red shift.

In Figure 7-77 and Figure 7-78, the shape and spectrum of the amplified Gaussian pulse (for the same
pulse and SOA parameters as above) are shown as a function of the optical confinement factor. The
values of optical confinement factor considered are Γ = 0.1, 0.2, 0.3, 0.4 and 0.5.

155
SLA GAIN SATURATION OF GAUSSIAN PULSES

Figure 7-77: Shape of the amplified Gaussian pulse (for the same pulse and SOA parameters as above) as a
function of the optical confinement factor

Figure 7-78: Spectrum of the amplified Gaussian pulse (for the same pulse and SOA parameters as above)
as a function of the optical confinement factor

As we see the optical confinement factor influences the pulse spectrum and shape.

Figure 7-79 and Figure 7-80 show the shape and spectrum of the amplified Gaussian pulse (for the same
pulse and SOA parameters as above) as a function of the internal losses. The values of internal losses
considered are αint = 0, 1000, 2000, 3000 and 4000 (1/m).

156
 SLA GAIN SATURATION OF GAUSSIAN PULSES

Figure 7-79: Shape of the amplified Gaussian pulse (for the same pulse and SOA parameters as above) as a
function of the internal losses

Figure 7-80: Spectrum of the amplified Gaussian pulse (for the same pulse and SOA parameters as above)
as a function of the internal losses

As we can see, in contrast to the optical confinement factor, the internal losses do not influence the pulse
spectrum and shape.

References:

[1] G.P. Agrawal and N.A. Olsson, “Self_phase modulation and spectral broadening of optical pulses in
semiconductor laser amplifiers”, IEEE Journal of Quantum electronics, vol. 25, pp.2297-2306, 1989.

[2] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

[3] G.P. Agrawal, Nonlinear Fiber Optics, third edition, Academic Press, 2001.

157
SLA GAIN SATURATION OF GAUSSIAN PULSES

Notes:

158
 RAMAN GAIN AND BANDWIDTH

Raman Gain and Bandwidth

Forward-Backward Raman amplification.osp

Project file “Forward-Backward Raman amplification.osp” demonstrates forward and backward Raman
amplification in a 16-channel WDM system. This project file includes two configurations in the same
project: one with forward pump and one with backward pump. 16 channels in 191 ~ 195.5 THz band,
separated with 3 GHz are multiplexed and sent to Raman amplifier. The frequency difference between
stokes and pump beam (Stokes shift) is important in SRS process. The Raman gain coefficient scales
linearly with pump frequency. Maximum gain is obtained when the Stokes shift is about 13.2 THz. This
example shows that, a relatively flat and high gain can be obtained when the frequency of the pump
signal is about 206.75 THz, in which case the Stokes shift is about 13.3 THz for the center channel. In
fact, signal frequency can differ from pump frequency up to 20 THz. You can experiment how the value
of Stokes shift affects the gain by changing the frequency of the pump signal. Overall optical gain is a
function of signal frequency and pump intensity (pump power divided by effective area of the fiber). To
test this, you can change the effective area of the fiber or the pump signal power.

159
RAMAN GAIN AND BANDWIDTH

Notes:

160
 RAMAN GAIN SATURATION

Raman Gain Saturation

Forward-Backward Raman gain saturation.osp

Project file “Forward-Backward Raman gain saturation.osp” demonstrates the gain saturation due to
pump depletion in forward and backward pumped Raman amplifier in a 4-channel WDM system. In
Raman amplifier, since the pump supplies energy for signal amplification, it begins to deplete as the
signal power increases. A decrease in the pump power reduces the optical gain. This reduction in the
gain is referred to as gain saturation. In this example, we see how the gain saturation differs depending
on forward or backward pump configuration. You can see this by clicking on results table for both cases
and comparing them.

161
RAMAN GAIN SATURATION

Notes:

162
 BROADBAND FLAT GAIN RAMAN AMPLIFIER

Broadband Flat Gain Raman Amplifier

Broadband flat gain Raman amplifier.osp

The characteristics of the stimulated Raman scattering effect allow for an interesting new technique for
gain flattening of broadband Raman amplifiers. Project file “Broadband flat gain Raman amplifier.osp”
demonstrates that multi-pump can be used to flatten the Raman amplifier gain in a broad band. In this
example, several pumps are arranged in 90 nm broad spectral range and they are used to pump in 100
km o f SMF-28 in backward direction. This gain flattening technique requires only a suitable multiple-
pump configuration and does not rely on any passive gain equalizing optics. The key feature of this
technique is the proper choice of number of pumps, pump wavelengths, and pump powers. You can
experiment by altering the power distribution and wavelengths of the pumps.

163
BROADBAND FLAT GAIN RAMAN AMPLIFIER

Notes:

164
 PUMPING REQUIREMENTS

Pumping Requirements

Pumping requirements - Co-propagating.osp

Pumping requirements - Counter-propagating.osp

Pumping requirements - Bidirectional.osp

The amplifier performance characterized by the signal output power, gain and noise figure depends on
the pump wavelength. The project file “Pumping requirements - Co-propagating.osp” (Figure 7-81)
shows two different configurations, “Pump @ 980 nm” and “Pump @ 1480 nm”. The 980 nm and 1480
nm pump wavelengths are the most important ones to be used in EDFAs.

Figure 7-81: Layout - Pumping requirements - Co-propagating

High amplifier gains in the range of 30 ~ 35 dB can be obtained in this project file by sweeping the pump
power from 10 mW to 200 mW. The signal input power considered in this case is -20 dBm. Results
obtained with these versions are shown in Figure 7-82.

EDFAs can be designed to operate in such a way that the pump and signal beams propagate in
opposite directions, a configuration referred to as backward pumping to distinguish it from forward-
pumping configuration in which both beams propagate along the same direction. Backward or
counter-propagating pump scheme is discussed in the project “Pumping requirements - Counter-

165
PUMPING REQUIREMENTS

propagating.osp”. The pump power is swept from 10 to 150 mW with the signal input power kept to be
-20 dBm.

Figure 7-82: Gain vs. pump power considering a co-propagating pump schemes @ 980 nm (black) and @
1480 nm(red) pump wavelength

The layout considering backward or counter-propagating pump scheme is shown in Figure 7-83.
Figure 7-84 shows the gain versus pump power when 980 nm and 1480 nm are considered as pump
wavelength. The influence of the selected pump scheme on the amplifier performance can be seen
by considering unsaturated or saturated regime by just changing the signal input power range.

Figure 7-83: Layout of pumping requirements - counter-propagation

166
 PUMPING REQUIREMENTS

Figure 7-84: Gain vs. pump power considering a counter-propagating pump schemes @ 980 nm (black) and
@ 1480 nm (red) pump wavelength

The project “Pumping requirements - Bidirectional.osp” considers the bidirectional pumping

configuration, where the amplifier is pumped in both directions simultaneously by using semiconductor
pump lasers located at the two fiber ends. The advantage of this pump scheme can be checked by setting
small-signal or large-signal input and observing the amplifier performance.

167
PUMPING REQUIREMENTS

Notes:

168
 MULTICHANNEL AMPLIFICATION

Multichannel Amplification

Multichannel amplification.osp

EDFA performance is evaluated considering multiple signal input. The amplifier is setup in a bidirectional
pump scheme, as shown in Figure 7-85.

Figure 7-85: Layout - Multichannel amplification

In the project, the amplifier is setup in a bidirectional pump scheme that enables extending the analysis
to co-propagating or counter-propagating pump schemes.

169
MULTICHANNEL AMPLIFICATION

Notes:

170
 OPTICAL PREAMPLIFIERS

Optical Preamplifiers

Preamplifier system.osp

The performance of a preamplifier positioned at the receiver input is evaluated in the project
“Preamplifier system.osp” (Figure 7-86). Characteristics of noise, gain and the improvement in the
sensitivity of the photodetector/receiver can be analyzed.

Figure 7-86: Layout - Preamplifier system

A modulated signal is coupled to the EDFA preamplifier and the corresponding characteristics are
displayed on Dual Port WDM Analyzer by double clicking on this visualizer.

The direct-detected signal on the receiver is pre- and post- filtered and the BER is evaluated through
the BER Analyzer. The eye diagram is shown in Figure 7-87.

The amplifier performance will affect the eye diagram observed at the BER Analyzer. Input conditions,
such as signal wavelength, signal and pump power, and fiber length specifications will affect the amplifier
performance when being modified.

171
OPTICAL PREAMPLIFIERS

Figure 7-87: Eye diagram

172
 POWER BOOSTERS

Power Boosters

Booster system.osp

Booster amplifiers are usually positioned at the transmitter output or for amplifying the multiple
outputs of nodes in broadcast networks.The project “Booster system.osp” (Figure 7-88) presents the
characteristics of a booster amplifier positioned at the transmitter output.

Two different types of modulation are considered in this project: direct and external modulation. In
the first case, the electrical signal is directly modulated on the signal laser. The other option uses an
electroabsorption modulator to modulate the signal source laser. Both schemes can be observed by
coupling the modulated signal to the EDFA and evaluating the obtained results.

Figure 7-89 and Figure 7-90 show the eye diagram obtained after detecting the signal using direct
modulation and external modulation, respectively.

Figure 7-88: Layout - Booster system

173
POWER BOOSTERS

Figure 7-89: Eye diagram observed after detecting the signal using direct modulation

Figure 7-90: Eye diagram observed after detecting the signal using external modulation

174
 METRO EDFA

Metro EDFA

Metro EDFA.osp

This example “Metro EDFA.osp” illustrates EDFAs included in a four nodes network setup in a ring
topology where Node 1 communicates with Node 3 at 193.2 THz, and Node 2 communicates with
Node 4 at 193.1 THz. Add-drop components enable the selective signal input/output in each node. Ideal
amplifiers are included in these simulations. Subsystems like this can be found in metropolitan area
networks. Figure 7-91 shows the layout used to evaluate the amplifier performance in a metropolitan
network.

The component Ring Control determines the number of loops considered in the two nodes connected in
a ring topology. The signal output and bit error rate are visualized in the Optical Spectrum Analyzer and
BER Analyzer for different loops where signal index 0 refers to the first turn in the loop and 1 is related
to the second turn in the loop.

The signal going through Node 1 to Node 3 on channel 1 at 193.2 THz as well as the communication
between Node 2 and Node 4 at 193.1 THz can be observed using different visualizers inserted in the
project file.

Figure 7-91: Layout - Metro EDFA

The amplified signal observed at the Node 3 output is shown in Figure 7-92. The eye diagram is shown
in Figure 7-93.

175
METRO EDFA

Figure 7-92: Spectrum observed at OSA referring to the turn 1 in the loop

Figure 7-93: Eye diagram observed at the BER analyzer referring to the turn 1 in the loop

176
 CASCADED IN-LINE AMPLIFIERS

Cascaded In-Line Amplifiers

In-line cascade.osp

In the example “ In-line cascade.osp”, ideal amplifiers are cascaded to simulate the propagation of
signal along the system. A signal laser source is coupled to the cascaded amplifiers and the amplifier
performance can be evaluated.

The effective noise figure for this cascaded chain amplifier can be calculated after running the
simulation. One can change the input signal power as well as gain and noise figure for each amplifier
in the system. Figure 7-94 shows three ideal cascaded amplifiers at the output of a signal laser source.
The Dual Port WDM Analyzer inserted in the layout enables checking the amplifier performance. The
effective noise figure for a cascaded chain can be compared considering different amplifier gains and
noise figures.

Figure 7-94: Layout - Cascaded in-line amplifiers

177
CASCADED IN-LINE AMPLIFIERS

Notes:

178
 EDFA DYNAMIC

EDFA Dynamic

EDFA dynamic.osp

In the example “EDFA dynamic.osp” (Figure 7-95) we simulate the function response of a C-band EDFA
to a step excitation at a signal wavelength. The step function has a period of 1.0 ms, longer enough to
enable the amplifier to achieve the steady-state solution. The pump power is constant over the time
and is counter-propagating with relation to the signal. Power transients which occur in the amplifier can
be studied as a function of the pump power and the signal power.

Figure 7-95: Layout - EDFA dynamic

The Optical Time Domain Visualizer enables the observation of power decay at the amplifier output.

The signal input coupled to the amplifier that enables dynamic analysis is shown in Figure 7-96, while
the signal output is shown in Figure 7-97.

179
EDFA DYNAMIC

Figure 7-96: Input signal observed at the output of the amplitude modulator

Figure 7-97: Power versus time oberved at the amplifier output

180
Chapter 8 - Dispersion Management
This section contains the following examples.

• Dispersion Compensation Techniques

• Precompensation Technique using SOA
• 40 Gbps Transmission System
• Dispersion Compensation with FBG
• Fiber Nonlinearities and Dispersion
• Negative Dispersion Fiber and Applications for Metro Networks
• Migrating to 10 Gbps Metro System and Dispersion Limited Transmission

181
Notes:

182
 DISPERSION COMPENSATION TECHNIQUES

Dispersion Compensation Techniques

Dispersion compensation techniques.osp

Project file “Dispersion compensation techniques.osp” (Figure 8-1) compares three different dispersion
compensation techniques, namely pre-, post-, and symmetrical dispersion compensations by using
DCFs. These examples consider just one channel. The bit rates used for the examples are 2.5 Gb/s and
10 Gb/s, respectively.

Although there are several forms of dispersion, the main form of dispersion we are concerned with is
chromatic dispersion or group velocity dispersion (GVD). Pulse broadening effect of chromatic dispersion
causes the signals in the adjacent bit periods to overlap. This is called inter-symbol interference (ISI).
Broadening is a function of distance as well as dispersion parameter D. Dispersion parameter is
given in ps/nm-km and changes from fiber to fiber. D is also a function of wavelength and is usually about
17 ps/nm-km in the 1.55 μm wavelength range for a standard single mode fiber (SMF). It is about 3.3
ps/nm-km in the same window for a dispersion-shifted fiber (DSF). Nonzero dispersion shifted fiber
(NZDSF) has a chromatic dispersion between 1 and 6 ps/nm-km for (+) NZDSF and –1 and –6 ps/nm-
km for (-) NZDSF.

For externally modulated sources, transmission distance limited by chromatic dispersion is given by [1]:

2πc -
L < ------------------------- (8-1)
2 2
16 D λ B

For D = 16 ps/nm-km and B = 2.5 Gb/s, L ~ 500 km whereas it drops to 30 km at 10 Gb/s bit rate.

Several techniques such as dispersion compensating fiber or fiber Bragg grating can be used to
compensate the accumulated dispersion in the fiber. In this example we will show three different
schemes, namely pre-, post-, and symmetrical compensation, to compensate the fiber dispersion. We
will then show how the amount of accumulated dispersion from the dispersion compensator affects the
performance.

Pre-, post-, and symmetrical compensation configurations are shown in Figure 8-1. In our simulations we
have used optical amplifiers after each fiber to compensate for the fiber loss. Dispersion parameter of
SMF is 16 ps/nm-km and it is 120 km long. Therefore, total accumulated dispersion is 16 x 120 = 1920
ps/nm. This amount of dispersion can be compensated by using a 24 km long DCF with – 80 ps/km-nm
dispersion. In post-compensation case, DCF is placed after SMF whereas in pre-compensation case it is
inserted before SMF. In symmetrical compensation case, fiber placement follows the sequence of DCF,
SMF, SMF, DCF.

183
DISPERSION COMPENSATION TECHNIQUES

Figure 8-1: Layout - Dispersion compensation techniques

Simulation results are shown in Figure 8-2 and Figure 8-3. Figure 8-2 shows Q factor of received signal
versus CW laser power for these three schemes at 2.5 and 10 Gb/s bit rates, respectively. These figures
are obtained by sweeping the power from -10 dBm to 10 dBm and using the software’s report page. From
these figures, we can conclude that the best performance is obtained by using symmetrical dispersion
compensation. The worst case is dispersion pre-compensation. Post- compensation gives a result as
good as symmetrical compensation [2,3]. This can also be seen from the eye diagrams given in Figure 8-
3.

184
 DISPERSION COMPENSATION TECHNIQUES

Figure 8-2: Q factor vs. CW laser power at 2.5 Gbps and 10 Gbps for pre-, post-, and symmetrical dispersion
compensations

B = 2.5 Gbps

B = 10 Gbps

185
DISPERSION COMPENSATION TECHNIQUES

Figure 8-3: System performance for pre-, post-, and symmetrical dispersion compensations.

Pre- Post- Symmetrical

B = 2.5 Gbps

P = -10 dBm

B = 2.5 Gbps

P = 10 dBm

B = 10 Gbps

P = -10 dBm

B = 10 Gbps

P = 10 dBm

186
 DISPERSION COMPENSATION TECHNIQUES

References:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

[2] R. Ramaswami and K.N. Sivarajan, Optical Networks: A practical Perspective, Morgan Kaufmann,
1998.

[3] M.I. Hayee and A.E. Willner, “Pre- and post-compensation of dispersion and nonlinearities in 10-Gb/s
WDM systems”, IEEE Poton. Tech. Lett. 9, pp. 1271, 1997.

187
DISPERSION COMPENSATION TECHNIQUES

Notes:

188
 PRECOMPENSATION TECHNIQUE USING SOA

Precompensation Technique using SOA

Amplifier induced chirp SLA compression.osp

The gain saturation induced self-phase modulation in SOA affects in the formation of chirp of the
pulse in the process of amplification. From the theory of propagation of phase modulated optical signals
in dispersive medium [1], we know that depending on the signs of the chirp of signal and the chirp of the
dispersive medium, different scenarios are possible. For equal signs of chirps the dispersion effect is
amplified. On other hand, in the case of different signs of the chirps the effect of pulse dispersion can be
reduced which effectively leads to pulse compression.

If we would like to use the red frequency shift created in the process of pulse amplification in SOA for
compression, we have to propagate this pulse in a dispersive medium that creates a blue frequency shift.
Such medium could be standard single mode fiber used for wavelengths larger than the zero dispersion
wavelength or in other words in the region of anomalous group velocity dispersion. This idea has
been proposed and realized in [2]. In this project, using SOA component we will try to interpret results
obtained in [2].

Following parameters of SOA have been used in [2]: Unsaturated single pass gain G0 = 30 dB; linewidth
enhancement factor α = 5; saturation energy Esat ~ 6 pJ. Ratio Ein / Esat has been varied over the range
from 0.01 to 0.2. Typical values for the carrier lifetime are given as 200 ~ 300 ps. Pulses for which the
condition T0 / Tsat << 1 is valid have been considered.

Following simulation parameters are used: bit rate B = 40 Gb/s -> TB = 25 ps. We also consider pulses
width TFWHM = 12.5 ps -> T0 ~ 7.5 ps. The default values of the SOA component, i.e., the carrier lifetime
and the saturation energy are τC ~ 1.4 ns and Esat ~ 5 pJ, respectively.

To satisfy the conditions T0 / Tsat << 1 and Ein / Esat ~ 0.01 to 0.2 we can use the Gaussian pulse with T0
~ 7.5 ps for which T0/Tsat ~ 0.0055 and initial peak pulse powers as 4 mW, 40 mW, and 80 mW (which
correspond to Ein / Esat = 0.01, 0.1 and 0.2 respectively).

Dispersion length LD in standard single mode fiber (D ~ 16 ps /nm-km) for pulse with T0 ~ 7.5 ps will be
LD ~ 2.76 km.

To satisfy the condition LC / LD ~ 0.3 as given in [1], where LC is the length of the fiber used for
compression, the value of LC ~ 0.828 km has been used. The layout for the project is shown in Figure 8-4.

189
PRECOMPENSATION TECHNIQUE USING SOA

Figure 8-4: Layout - Amplifier induced chirp SLA compression

Initial pulse shape and spectrum with TFWHM = 12.5 ps and initial pulse peak power = 4 mW have been
shown in Figure 8-5 and Figure 8-6, respectively.

Figure 8-5: Initial shape of pulse

Figure 8-6: Initial spectrum of pulse

Figure 8-7 and Figure 8-8 show the shape (with corresponding positive chirp) and spectrum after
amplification with SOA (G0 = 30 dB and linewidth enhancement factor α = 5).

190
 PRECOMPENSATION TECHNIQUE USING SOA

Figure 8-7: Shape after amplification with SOA

Figure 8-8: Spectrum after amplification with SOA

Figure 8-9 and Figure 8-10 show, the shape and spectrum of the amplified pulse after passing through
0.828 km of standard single mode fiber (SMF).

191
PRECOMPENSATION TECHNIQUE USING SOA

Figure 8-9: Shape of the amplified pulse after propagating through SMF

Figure 8-10: Spectrum of the amplified pulse after propagating through SMF

The comparison of the width of the pulse after fiber shows a compression factor of about 5 times with
peak power enhanced by a factor of several hundreds. Two important physical facts should be mentioned
in the Figure 8-9. First, there is a broad pedestal on the leading side of the pulse that leads to increased
dispersion of this part of the pulse. The second point is connected with the shift of the pulse towards later
time. This fact is due to the red shift of central frequency of the amplified pulse that in anomalous group
velocity dispersion region leads to time delay of the pulse.

As is mentioned in [2], the compression factor depends on several input parameters, such as the pulse
energy, the pulse width, the amplifier gain, and the compressor length. For non optimal parameters,
another type of phenomena can appear, for example, pulse breaking as demonstrated in Figure 8-11 and
Figure 8-12, where the shapes and spectrum of pulses with initial peak powers of 4, 40 and 80 mW are
presented.

192
 PRECOMPENSATION TECHNIQUE USING SOA

Figure 8-11: Shape of the pulse for non-optimal parameters

Figure 8-12: Spectrum of the pulse for non-optimal parameters

As we can see from Figure 8-11, larger input peak power can lead to pulse breaking phenomena with
connected complicated spectrum structures.

We have demonstrated pulse compression of weak pulses (peak energy smaller than pJ) based on the
gain saturation induced self-phase modulation effect in SOA.

References:

[1] G.P. Agrawal, Nonlinear fiber optics, third edition, Academic press, 2001.

[2] G.P. Agrawal and N.A. Olsson, “Amplification and compression of weak picosecond optical pulses by
using semiconductor laser amplifier”, Optics Letters, vol.14, pp. 500-502, 1989.

193
PRECOMPENSATION TECHNIQUE USING SOA

Notes:

194
 40 GBPS TRANSMISSION SYSTEM

40 Gbps Transmission System

40 Gbps - RZ and NRZ.osp

40 Gbps - Amplifier noise and SPM for RZ.osp

40 Gbps – Post- and pre- dispersion compensation for RZ.osp

The fundamental limitation to high speed communication systems over the standard single mode fiber at
1.55 μm is the linear chromatic dispersion. Typical value of β2 = -20 ps2/km at 1.55 μm for SMF leads to
D = 16 ps/nm-km. For bit rate B = 40 Gb/s, the slot duration will be TB = 25 ps. For duty cycle = 0.5 ->
TFWHM = 12.5 ps -> T0 = 7.5 ps, the corresponding dispersion length will be approximately LD ~ 2.8 km.

The aims of this project are as follows:

a) to compare RZ versus NRZ modulation format transmission in SMF at 40 Gb/s taking into account:
group velocity dispersion, self-phase modulation due to the Kerr nonlinearity, linear losses and periodical
amplification with ASE noise. The large group velocity dispersion is compensated with post- dispersion
compensation scheme.

b) to analyze influence of accumulated amplifier noise and self-phase modulation for RZ modulation
format transmission in SMF at 40 Gb/s.

c) to compare post- and pre- dispersion compensation schemes for RZ modulation format transmission
in SMF at 40 Gb/s.

195
40 GBPS TRANSMISSION SYSTEM

Figure 8-13: Layout - RZ and NRZ modulation formats

Global parameters used for this project are: Bit rate = 40 Gb/s, Sequence length = 256 Bits and Samples
per bit = 64.

RZ generator has following properties: rectangle shape is Gaussian; duty cycle= 0.5 bit; rise time= 0.15
bit; fall time = 0.25 bit.

As an optical source, an externally modulated CW Laser with carrier wavelength = 1550 nm and linewidth
= 10 MHz is used.

The standard single mode optical fiber used has dispersion coefficient D = 16 ps/nm-km, the slope
( ∂D ) ⁄ ( ∂λ ) = 0.08 ps/nm2-km, the nonlinear-index coefficient n2 = 3 x 10-20 m2/W, effective mode area
Aeff = 80 μm2, linear losses α = 0.2 dB/km and fiber length LSMF = 50 km.

After each segment of SMF an amplifier compensates the linear losses. Dispersion compensation fiber
has dispersion coefficient D = -80 ps/nm-km, the slope ( ∂D ) ⁄ ( ∂λ ) = 0.08 ps/nm2-km, the nonlinear-index
coefficient n2 = 3 x 10-20 m2/W, effective mode area Aeff = 23 μm2, linear losses α = 0.5 dB/km and fiber
length LDCF = 10 km. After each segment of DCF an amplifier compensates the linear losses.

Properties of Bessel optical filter used are: carrier wavelength = 1550 nm and bandwidth = 4 x Bit rate.
The cutoff frequency of the low pass Bessel electrical filter used is 0.75 x Bit rate.

Ideal periodical amplification is performed with the help of ideal EDFA component of OptiSystem that also
takes ASE noise into account. The amplifier used after SMF has a gain of 10 dB and noise figure of 4 dB
and the amplifier used after DCF has a gain of 5 dB with the same noise figure value.

Part I: Comparison of RZ and NRZ formats

196
 40 GBPS TRANSMISSION SYSTEM

In the project “40 Gbps – RZ and NRZ.osp” (Figure 8-13) the transmission over a distance of 1000 km
SMF ( 20 loops x 50 km) with periodical amplification after every 50 km of fiber has been considered. For
the RZ modulation format, the result for the dependence of the max Q on the transmitter power (after the
MZM modulator) is shown in Figure 8-14. Figure 8-15 shows the eye diagram for the optimal point when
transmitter power ~ 0.3 mW (cf. Figure 8-14).

Figure 8-14: Max Q vs. transmitter power (RZ modulation)

Figure 8-15: Eye diagram when transmitter power ~ 0.3 mW

For the NRZ modulation format, the same transmission link and receiver structure is used, and the result
of the dependence of the max Q vs. transmitter power is shown in Figure 8-16. Figure 8-17 shows the
eye diagram for the optimal point when input power ~ 0.17 mW (cf. Figure 8-16).

197
40 GBPS TRANSMISSION SYSTEM

Figure 8-16: Max Q vs. transmitter power (NRZ modulation)

Figure 8-17: Eye diagram when transmitter power ~ 0.17 mW

Comparing positions of the maximums of both curves we can clearly see the shift toward larger input
powers for the RZ format. Obtained results show that the RZ modulation format with duty cycle = 0.5 is
superior compared to conventional NRZ modulation format.

Part II: Influence of accumulated amplifier noise and self-phase modulation on RZ signal
transmission

In this project we have considered two different situations. In the first case, the self-phase modulation
was assumed to be zero, and in another case the noise figure was not taken into account. Obtained
results are presented in Figure 8-18.

198
 40 GBPS TRANSMISSION SYSTEM

Figure 8-18: Max Q vs. input CW laser power (dBm) for two separate cases (Case I: SPM = 0; Case II: Noise
figure = 0)

As we see, at low power levels, the performance is mainly hampered by the accumulated amplifier noise.
At high input power levels, the transmission distance is significantly reduced due to the self-phase
modulation. As we see, just these effects determine the well expressed maximum in the Q curve in
Figure 8-14. The optimum Q happens when CW laser power is 0.55 dBm, which corresponds to a power
of ~ 0.3 mW after the modulator, matches with Figure 8-14.

Part III: Post- and pre- dispersion compensation for RZ modulation format

In the project “40 Gbps – Post and pre dispersion compensation for RZ.osp” (Figure 8-19), two schemes
of dispersion compensation are studied, i.e., post- and pre- compensation.

199
40 GBPS TRANSMISSION SYSTEM

Figure 8-19: Layout - 40 Gbps - Post and pre dispersion compensation for RZ

The corresponding results for the RZ signal transmission in SMF at 40 Gb/s are shown in Figure 8-20.

Figure 8-20: RZ modulation format transmission at 40 Gb/s over 250 km in SMF. Max Q vs. CW laser power

200
 40 GBPS TRANSMISSION SYSTEM

Results indicate that post-compensation scheme is superior compared to pre- compensation in

dispersion compensated system at 40 Gbps in SMF.

We have thus shown that for upgrading the existing standard fiber network at 1.55 μm to 40 Gb/s:

a) Dispersion compensation is necessary

b) RZ is superior compared to conventional NRZ modulation format

c) At low power levels, the performance is mainly hampered by the accumulated amplification noise and
at higher input power levels, the transmission distance is significantly reduced by self-phase modulation

d) Post compensation scheme is superior compared to pre-compensation scheme.

References:

[1] D. Breuer and K. Petermann, "Comparison of NRZ and RZ- modulation format for 40 Gbit/s TDM
standard-fiber system", IEEE Photon. Technol. Lett., vol. 9, pp.398-400, 1997.

[2] M.I. Hayee and A.E. Willner, "NRZ Versus RZ in 10-40 Gb/s Dispersion – Managed WDM
Transmission systems", IEEE Photon. Technol. Lett., vol. 11, pp.991-993, 1999.

[3] C.M. Weinert, R. Ludvig, W. Papier, H.G. Weber, D. Breuer, K. Petermann and F. Kuppers, "40 Gbit/s
Comparison and 4 x 40 Gbit/s TDM/WDM Standard Fiber Transmission", Journal of Lightwave
Technology, vol. 17, pp.2276-2284, 1999.

201
40 GBPS TRANSMISSION SYSTEM

Notes:

202
 DISPERSION COMPENSATION WITH FBG

Dispersion Compensation with FBG

Dispersion compensation – Post with FBG.osp

Project “Dispersion compensation - Post with FBG.osp” (Figure 8-21) shows the dispersion
compensation by using fiber Bragg grating. It shows how the amount of compensating dispersion affects
the system performance.

In the project we prefer post-compensation scheme because of its simplicity comparing to symmetrical
compensation scheme. Its performance is also as good as symmetrical compensation scheme.

Figure 8-21: Layout - Dispersion compensation - Post with FBG

In this case, the total accumulated dispersion of the SMF is 16 x 100 = 1600 ps/nm. We swept the total
dispersion of FBG from 0 to -2000 ps/nm. In this simulation, we just want to see the dispersion-limited
performance of the system.

Therefore we keep the CW laser power at 0 dBm. This is needed to avoid triggering of any fiber
nonlinearity. Effects of residual dispersion to nonlinear effects will be considered in other examples.
Figure 8-22 shows the eye diagrams at the receiver end for several total compensation dispersion
values. This simulation shows that, in the linear regime (low input signal power), compensate fiber
dispersion completely gives the best result. Over compensation degrades the system performance.

203
DISPERSION COMPENSATION WITH FBG

Figure 8-22: Eye diagrams at the receiver end when the FBG dispersion is (a) 0 (b) 1600 and (c) 2000 ps/nm

(a) (b) (c)

204
 FIBER NONLINEARITIES AND DISPERSION

Fiber Nonlinearities and Dispersion

Fiber nonlinearities and dispersion – Single channel.osp

Fiber nonlinearities and dispersion – Multi channel equal spacing.osp

Fiber nonlinearities and dispersion – Multi channel unequal

spacing.osp

As long as the optical power within an optical fiber is small, the fiber can be treated as a linear medium.
However, when the power level is high, we have to consider the impact of nonlinear effects. They can be
classified into two categories. Some of them occur in multi channel WDM systems where interaction of
signals at different wavelengths is possible. These categories and nonlinear effects in single channel and
multi-channel systems are shown in Table 8-1.

Table 8-1: Nonlinear effects in the fiber

Single channel Multi channel

Refractive index related Self phase modulation (SPM) Cross phase modulation (XPM)

Four wave mixing (FWM)

Scattering related Stimulated scattering (SBS) Stimulated raman scattering (SRS)

SPM and XPM affect the phase of signals and cause spectral broadening, which in turn, leads to increase
in dispersion penalties. SBS and SRS provide gains to some channels at the expense of depleting power
from other channels. The nonlinear interaction depends on transmission length and effective area of the
fiber. Since loss in the fiber decreases the signal power, we use an effective instead of physical length.
In practice, SPM can be a significant consideration in designing 10 Gbps systems, and restricts maximum
power per channel below a few dBm. XPM becomes an important consideration when the channel
spacing is of the order of a few tens of GHz. FWM efficiency depends on signal power and dispersion,
as well as channel separation in WDM systems. If the channel is close to zero dispersion wavelength of
the fiber, considerably high power can be transferred to FWM components. Using unequal channel
spacing can reduce the effect of FWM.

Dispersion plays a key role in reducing the effects of nonlinearities. However, dispersion itself causes
inter-symbol interference. Fortunately, the good thing is that we can engineer systems with zero total
dispersion but with a certain amount of local dispersion along the link.

Fiber nonlinearities and dispersion – Single channel

In this section we will consider effects of dispersion compensation to system performance in high power
regime where nonlinearities are active. Let us first consider a dispersion post-compensated system with
single channel. The project is given in “Fiber nonlinearities and dispersion - Single channel.osp” file. In

205
FIBER NONLINEARITIES AND DISPERSION

this file, we have two different versions of the transmission link. In the first version, system residual
dispersion is 0, whereas it is 800 ps/nm in the second version. Transmission link contains 5 spans and
bit rate is 10 Gbps. Dispersion of the 100 km SMF is 16 ps/nm-km and its effective area is 72 micron-
square. Dispersion of the 20 km DCF is – 80 ps/nm-km for the first version, and –72 ps/nm-km for the
second one. Effective area of DCF is 30 micron-square. SMF and DCF losses are compensated by an
EDFA with 35 dB gain. Noise figure of the EDFA is set to zero, because the purpose of this example
is to deal with nonlinear and dispersion effects. If we only consider the SMF and also include the effect
of amplification (5 times), we can estimate the threshold power for SPM [1,2]. If we do so, we find that
after about 10 dBm average power, SPM becomes a limiting effect. However, we need to investigate if
that is the case then how the residual system dispersion affects it. Figure 8-23 shows the eye diagrams
for two different residual dispersion values and three different signal input powers.

Figure 8-23: Eye diagrams of the received signal for several signal input powers when system residual
dispersion is (a) 0 and (b) 800 ps/nm. Inset on the left shows the dispersion map for each case.

0 dBm 10 dBm 13 dBm

This simulation shows that the effect of SPM can be reduced by not completely compensating the
dispersion, but leaving some residual dispersion in the system. Notice how the increase of power results
in eye closure in zero residual dispersion case.

Fiber nonlinearities and dispersion – Multi channel equal spacing

Let us now consider a multi-channel system with 4 channels. The first channel is at 193.1 THz and they
are separated with 100 GHz. SMF and DCF parameters are same as above. Even though the dispersion
of fiber depends on wavelength, we set both fibers’ dispersion to constant values. This will not affect the
basic idea that we want to show here. We again have two versions of the design. The project is given in
“Fiber nonlinearities and dispersion - Multi channel equal spacing.osp” file. In this file, we have two
different versions of the transmission link. In the first version, system residual dispersion is 0, whereas it

206
 FIBER NONLINEARITIES AND DISPERSION

is 800 ps/nm in the second one. In Figure 8-24, simulation results are given for these two cases with
several channel powers.

Figure 8-24: Eye diagrams of the received signal for several signal input powers when system residual
dispersion is (a) 0 and (b) 800 ps/nm. Insets on the left show the dispersion map as well as the input power
spectrum for each case.

0 dBm 10 dBm 13 dBm

0 dBm 10 dBm 13 dBm

For a four-channel system, the threshold power for nonlinear effects is about 10 dBm per channel.
Simulation results justify this finding. In this simulation, both SPM and XPM affect the system

207
FIBER NONLINEARITIES AND DISPERSION

performance. This simulation shows that nonlinear effects can be reduced by local dispersion. Better
performance is obtained with nonzero residual dispersion [3].

Fiber nonlinearities and dispersion - Multi channel unequal spacing

We will now show another example with 4 unequally spaced channels. Channels are at 193.1, 193.4,
193.65, and 193.85 THz. Bit rate is 10 Gb/s. In this case, we used wavelength dependent dispersion
profile. Dispersion of the SMF is 16 ps/nm-km and DCF is -80 ps/nm-km @ 1550 nm. Dispersion slope
of both SMF and DCF are set to 0.08 ps/nm2-km. Project is given in “Fiber nonlinearities and dispersion
- Multi channel unequal spacing.osp”.

In the output spectrum, note the FWM generated spectrum components. Since they do not overlap with
channels, FWM effect is decreased. Figure 8-25 shows the eye diagrams for each channel.

Figure 8-25: Eye diagrams of channels as obtained from the simulation. The first row shows the input signal
spectrum; the second row shows the output spectrum as well as the eye diagrams for the four channels
when channel input power is 0 dBm/ch; the third row is the same as the second row except that channel
power is 10 dBm/ch.

References:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

[2] R. Ramaswami and K.N. Sivarajan, Optical Networks: A practical Perspective, Morgan Kaufmann,
1998.

208
 FIBER NONLINEARITIES AND DISPERSION

[3] G. Bellotti et. al., “Dependence of self-phase modulation impairments on residual dispersion in 10 Gb/s
based terrestrial transmission using standard fiber”, IEEE Poton Tech. Lett. 11, pp. 824, 1999.

[4] H. Taga, “Long distance transmission experiments using the WDM technology”, J. Lightwave Tech.
14, pp. 1287, 1996.

209
FIBER NONLINEARITIES AND DISPERSION

Notes:

210
 NEGATIVE DISPERSION FIBER AND APPLICATIONS FOR METRO NETWORKS

Negative Dispersion Fiber and Applications for Metro

Networks

Negative dispersion fiber and applications for metro networks.osp

Project file “Negative dispersion fiber and applications for metro networks.osp” compares two types of
fibers namely Corning’s MetroCor and SMF-28 Fibers. MetroCor Fiber has a negative dispersion
parameter whereas SMF-28 has a positive dispersion parameter in the EDFA band where we run our
simulations. We also consider effects of source chirp on the network performance. For metro
applications, directly modulated sources are preferred because of their low cost, but they result in more
chirp, and more penalty due to dispersion when classic positive dispersion SMF is used. One option to
overcome this effect is to use a fiber with negative dispersion. This example shows that, with the same
amount of accumulated dispersion, MetroCor fiber performs better when a directly modulated source is
used.

In the project, directly modulated laser sources are used with bit rate equals to 2.5 Gb/s. Three channels
at 191.9 THz, 192.3 THz, and 192.5 THz are multiplexed and propagated in a fiber loop, which consists
of either MetroCor or SMF-28 fiber followed by an amplifier. Noise from the EDFA is neglected in the
simulation. The chirp of the pulse at the source is shown in Figure 8-26.

Figure 8-26: Pulse shape and chirp at the output of the directly modulated laser

211
NEGATIVE DISPERSION FIBER AND APPLICATIONS FOR METRO NETWORKS

It is known that the linewidth of the source affects the broadening factor [3]. It is given as

σ
2 ⎛ Cβ 2 z⎞ 2 2 ⎛ β2 z⎞
2
-⎟ + ( 1 + ( 2σ ω σ 0 ) ) ⎜ --------
------ = ⎜ 1 + ------------ -⎟ (8-2)

σ0
2
⎝ 2σ 0 ⎠
2
⎝ 2σ 20⎠

Where σ ω is the RMS width of the Gaussian source spectrum. Directly modulated sources have broader
spectrums and this contributes more in the broadening of the pulse. It can even dominates the
2
broadening originated from the group velocity dispersion (GVD) itself when ( 2σ ω σ 0 ) » 1 .

Figure 8-27 shows eye diagrams of each channel after propagate in 300km MetroCor where total
accumulated dispersion is about 1680 ps/nm and 100 km SMF-28 where the total accumulated
dispersion is about 1675 ps/nm.

Figure 8-27: Eye diagrams of each channel after propagation in MetroCor and SMF-28, respectively.

Ch1 Ch2 Ch3

MetroCor

SMF-28

Even though the accumulated dispersion is the same for both tested links, we have clearer eye in the
case of MetroCor fiber link. This is achieved from the interplay between the negative chirp of the directly
modulated laser and the opposite chirp introduced by the negative dispersion parameter for the case of
MetroCor fiber. Note that in the transmission band, the dispersion parameter of MetroCor is negative
whereas that of SMF is positive.

212
 NEGATIVE DISPERSION FIBER AND APPLICATIONS FOR METRO NETWORKS

These results show that with the same amount of accumulated dispersion, MetroCor fiber performs better
when a directly modulated source is used. Moreover, we can conclude that the use of negative dispersion
fiber increases transmission distance.

References:

[1] C. Kennedy et. al., “The performance of negative dispersion fiber at 10 Gbps and significance of
externally and directly modulated lasers”, NFOEC’01, 2001.

[2] D. Culverhouse et. al., “Corning MetroCore Fiber and its Application in Metropolitan Networks”,
Corning’s White Paper WP5078.

[3] G.P. Agrawal, Nonlinear Fiber Optics, Second Edition, Academic Press, 1995.

213
NEGATIVE DISPERSION FIBER AND APPLICATIONS FOR METRO NETWORKS

Notes:

214
 MIGRATING TO 10 GBPS METRO SYSTEM AND DISPERSION LIMITED TRANSMISSION

Migrating to 10 Gbps Metro System and Dispersion

Limited Transmission

Migrating to 10 Gbps - No dispersion compensation.osp

Migrating to 10 Gbps - Dispersion compensation at each node.osp

Migrating to 10 Gbps - Lumped dispersion compensation.osp

In project files “Migrating to 10 Gbps – No dispersion compensation.osp”, “Migrating to 10 Gbps -

Dispersion compensation at each node.osp”, and “Migrating to 10 Gbps - Lumped dispersion
compensation.osp”, we investigate the issues at high data bit rate, particularly 10 Gb/s (OC-192) systems
in metro area ring networks. For high bit rates, in addition to loss compensation, we need to consider the
chromatic dispersion compensation. Two possibilities, namely “lumped” and “at each node” dispersion
compensation schemes are shown and compared for metro applications.

For this case, the receiver sensitivity is –16 dBm which reduces the allowable loss to Lallowable = 3 – (-
16) = 13 dB. Assuming again a power of 3 dBm per channel, the amount of amplification needed in the
ring is Grequired = - ( 3 - 4 x 0.25 x 50 - 4 x 3.8 - ( -16) - 3 ) = 49.2 dB. We can again use unity gain segment
approach for power compensation. This calculation shows that migration is easily done in terms of power
budgeting since the total loss is not changed, even though the allowable loss is changed.

Let us now look at the criteria for dispersion and try to see if it is a real issue for metro networks.

In the case of directly modulated DFB lasers, the GVD limited transmission distance is calculated by [1,2]

1 -
L < -------------------- (8-3)
4B D σ λ

Where σ λ is root-mean-square spectruml width, with a typical value around 0.15 nm, for D = 16 ps/nm-
km and 2.5 Gbps, L ~ 42 km.

215
MIGRATING TO 10 GBPS METRO SYSTEM AND DISPERSION LIMITED TRANSMISSION

For an externally modulated source, the GVD limited transmission distance is

2πc -
L < ------------------------- (8-4)
2 2
16 D λ B

Again for D = 16 ps/nm-km and 2.5 Gbps, L ~ 500 km. therefore group velocity dispersion (GVD) is not
a limiting effect in the above example for a 2.5 Gb/s communication network.

No dispersion compensation

At high bit rates, GVD may become a limiting parameter for the transmission distance. When the bit rate
is increased to 10 Gb/s, GVD limited transmission distance drops to ~ 30 km for an EML and ~ 10 km for
DML. Figure 8-29 shows the eye diagram when the bit rate is 10 Gb/s and transmitter (DML) power is 3
dBm/channel. In this example, we have used unity gain segment approach, which means the loss in each
fiber span and node is compensated immediately before the following node. This project is given in
“Migrating to 10 Gbs - No dispersion compensation.osp” file.

Figure 8-28: Layout - Migrating to 10 Gbps - No dispersion compensation

216
 MIGRATING TO 10 GBPS METRO SYSTEM AND DISPERSION LIMITED TRANSMISSION

Figure 8-29: Eye diagram of channel 1 at node 4 when the transmitter power is 3 dBm and bit rate is 10 Gb/s

Therefore, at this rate one has to consider dispersion management. Total dispersion per 50 km span is
around 16 x 50 = 800 ps/nm. This will require 10 km of dispersion compensating fiber (DCF) with D = -
80 ps/nm-km for each fiber span. We can again follow several avenues to compensate the accumulated
dispersion.

“Per span” dispersion compensation

For example, DCF can be used after each span or total accumulated dispersion can be compensated at
a certain point. Again for the scalability of the network and considering the dynamic structure of the metro
traffic, it seems to be the best choice to distribute the dispersion compensation to every node. This will
also make sure that dispersion experienced by dynamically routed signals will be compensated properly.
For this case, you also need to compensate the power loss of DCFs. For that purpose 2.5 dB more gain
is added to each amplifier. The simulation project is given in “Migrating to 10 Gbps - Dispersion
compensation at each node.osp” file. Figure 8-30 shows the network layout when unity gain approach
and per node dispersion compensation is used.

Figure 8-30: Ring network layout when unity gain approach and per span dispersion compensation is used

217
MIGRATING TO 10 GBPS METRO SYSTEM AND DISPERSION LIMITED TRANSMISSION

Figure 8-31: Layout - Dispersion compensation at each node

Figure 8-32 shows the eye diagram at the receiving end when per span dispersion compensation is used.

Figure 8-32: Eye diagram at node 4 when per span dispersion compensation is used

“Lumped” dispersion compensation

Now consider the “lumped” dispersion compensation case. In this case, one DCF and extra amplifier is
inserted just before node 2 to compensate the total dispersion and extra loss of DCF. This project is given
in “Migrating to 10 Gbps - Lumped dispersion compensation.osp” file. Figure 8-34 shows the eye diagram
at node 4 for this configuration. As can be seen from the eye diagram, this configuration definitely
improves the network performance for the considered channel comparing with the uncompensated case,
but worse than the case with distributed compensation. Moreover, “lumped” dispersion compensation is
not suitable for metro networks in which the wavelength routing path cannot be estimated.

218
 MIGRATING TO 10 GBPS METRO SYSTEM AND DISPERSION LIMITED TRANSMISSION

Figure 8-33: Layout - Lumped dispersion compensation

Figure 8-34: Eye diagram at node 4 when “lumped” dispersion compensation is used

References:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

[2] R. Ramaswami and K. N. Sivarajan, Optical Networks: A practical Perspective, Morgan Kaufmann,
1998.

219
MIGRATING TO 10 GBPS METRO SYSTEM AND DISPERSION LIMITED TRANSMISSION

Notes:

220
Chapter 9 - Soliton Systems
This section contains the following examples.

• Fundamental and Higher-Order Solitons

• Interaction of A Pair of Fundamental Solitons
• Average Soliton Regime
• Interaction of Multicolor Solitons

221
Notes:

222
 FUNDAMENTAL AND HIGHER-ORDER SOLITONS

Fundamental and Higher-Order Solitons

Fundamental and higher-order solitons.osp

The aim of this example is to present numerical results for fundamental and higher-order solitons in SMF
at 1.55 μm.

The basic parameter N is defined by N2 = γP0LD, where γ = 2πn2/(λAeff) is nonlinear coefficient, LD =

T02/|β2| is the dispersion length, P0 is the peak power, n2 is the nonlinear refractive index, Aeff is the
effective core area, T0 = TFWHM / 1.763, and β 2 = ( ∂ β ) ⁄ ( ∂ω ) at ω0 where β is the propagation constant.
2 2

The basic parameter N represents a dimensinonless combination of the pulse and fiber parameter.

The optical pulse which corresponds to N = 1 is called fundamental soliton. Pulses with N = 2, and 3 are
called second and third-order solitons. The layout for generation of the solitons is shown in Figure 9-1.

Figure 9-1: Layout - Soliton generation

As an optical source, an optical sech pulse generator has been chosen. A 40 Gb/s transmission at 1.55
μm in SMF (β2 = -20 ps2/km or D = 16 ps/nm-km) is considered. We consider TB = 25 ps, TFWHM = 12.5
ps ( duty cycle = 0.5) -> LD = 2.51 km and Aeff = 80 μm2, n2 = 2.6e-20 m2/W -> γ = 1.3 W-1km-1 -> the
peak power of the fundamental soliton P0 ~ 0.302 W. The power required for obtaining the second- and
third-order solitons are 1.208 W and 2.718 W, respectively.

223
FUNDAMENTAL AND HIGHER-ORDER SOLITONS

Distance at which the three types of solitons were propagated is equal to so-called soliton period (Z0 =
π/2*LD). In our case Z0 = 3.9482 km. The evaluation of fundamental, second order and third order solitons
is shown in Figure 9-2, Figure 9-3, and Figure 9-4.

Figure 9-2: Evolution of the fundamental soliton over one soliton period

Figure 9-3: Evolution of the second-order soliton over one soliton period

224
 FUNDAMENTAL AND HIGHER-ORDER SOLITONS

Figure 9-4: Evolution of the third order soliton over one soliton period

References:

[1] G.P. Agrawal, Nonlinear Fiber optics, third edition, Academic Press, 2001.

[2] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

225
FUNDAMENTAL AND HIGHER-ORDER SOLITONS

Notes:

226
 INTERACTION OF A PAIR OF FUNDAMENTAL SOLITONS

Interaction of A Pair of Fundamental Solitons

Interaction of a pair of fundamental solitons.osp

Project file “Interaction of a pair of fundamental solitons.osp” demonstrates the interaction of two adjacent
solitons. If the solitons are not well separated in time, depending on their phases they will either
periodically collide or repeal from each other. In this example, two in phase soliton pulses are launched
into a nonlinear dispersive fiber. They periodically collide inside the fiber. The collision period depends
on the input time shift between the pulses. As a comparision, the propagation of two π-out-of-phase
soliton pulses in a fiber is also shown.

Figure 9-5: Layout - Interaction of a pair of fundamental solitons

227
INTERACTION OF A PAIR OF FUNDAMENTAL SOLITONS

Notes:

228
 AVERAGE SOLITON REGIME

Average Soliton Regime

Average soliton regime.osp

The aim of this example is to demonstrate the average soliton regime at 10 Gb/s transmission over a 500
km optical link consisted of SMF. The corresponding layout has been shown below in Figure 9-6.

Figure 9-6: Layout - Average soliton regime

We fixed B = 10 Gb/s -> TB = 100 ps. The sequence length is 16 bit. The carrier wavelength of the pulse
is ~ 1300 nm. TFWHM = 20 ps -> T0 = 0.567 x TFWHM ~ 11.34 ps. The input pulse peak power is 27.1 mW.

We have considered SMF with a length of 50 km and loss of 0.4 dB/km. D = 1.67 ps/nm-km -> LD = 85
km. (The effects of group delay and third order dispersion are not taken into account). After every 50 km
of fiber, the signal will be amplified using an EDFA, therefore, LA = 50 km. Note tha the condition LA < LD
is satisfied.

The value of Kerr nonlinearity coefficient γ = n2ω0/(cAeff) can be calculated with n2 = 2.6 x 10-20 m2/W
and Aeff = 62.8 μm2. The value comes out to be 2 W-1km-1.

The linear loss for 50 km SMF is 20 dB. The loss is periodically compensated with an ideal EDFA with
20 dB gain.

Soliton peak power for this fiber is 5.8 mW. The pulse peak power of the average soliton is 27.1 mW. To
demonstrate the importance of the input pulse peak power of the average soliton, we will consider soliton
propagation in 500 km of SMF with two different input pulse peak power values: 5.8 mW (usual soliton
peak power) and 27.1mW (modified soliton peak power) which takes into account the periodical

229
AVERAGE SOLITON REGIME

amplification, respectively. Each case includes sweeps on the number of loops: 0, 4, 7 and 10 which
represent the propagation distances of 0, 200, 350 and 500 km in SMF.

Figure 9-7 shows the initial pattern of pulses and the pattern of pulses after 200, 350 and 500 km of SMF,
respectively. The periodic amplification has been done with EDFA at every 50 km of SMF with 27.1 mW
as the peak power.

Figure 9-7: (a) Initial pulse pattern (b) Pulse pattern after 200 km of SMF (c) Pulse pattern after 350 km of SMF
(d) Pulse pattern after 500 km of SMF

As can be seen, the pattern of pulses is well-preserved. The average soliton concept is valid for these
pulse, fiber, and amplifier parameters.

Figure 9-8 shows the initial pattern of pulses and the pattern of pulses after 200, 350 and 500 km of
SMF, respectively. The periodic amplification has been done with EDFA at every 50 km of SMF with 5.8
mW as the peak power.

Figure 9-8: (a) Initial pulse pattern (b) Pulse pattern after 200 km of SMF (c) Pulse pattern after 350 km of SMF
(d) Pulse pattern after 500 km of SMF

Because of the improper pulse peak power used, the pulses do not preserve their original forms. As a
result, pulses broaden and complicated structures appear.

References:

[1] G.P. Agrawal, Applications of nonlinear fiber optics, Academic Press, 2001.

[2] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., 2010.

230
 INTERACTION OF MULTICOLOR SOLITONS

Interaction of Multicolor Solitons

Interaction of multicolor solitons.osp

Project file “Interaction of multicolor solitons.osp” demonstrates a 2 channel soliton communication

system. The first case shows the interaction of two isolated pulses carried on two different carrier
frequencies. In this case, power of each channel must be adjusted according to the value of dispersion
at that specific wavelength to satisfy the soliton condition for each soliton pulse from different channels.
In the second case, each channel contains two soliton pulses; they interact among themselves and with
the pulses in other channels.

Figure 9-9: Layout - Interaction of multicolor solitons

231
INTERACTION OF MULTICOLOR SOLITONS

Notes:

232
Chapter 10 - Advanced Lightwave
Systems
This section contains the following examples.

• Intensity Modulation Formats (NRZ, RZ, AMI, DB)

• DPSK Modulation and Demodulation
• QPSK Modulation and Demodulation
• 100 Gbps PM-QPSK
• OD8PSK Modulation and Demodulation
• 4-QAM OFDM Modulation and Demodulation

233
Notes:

234
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Intensity Modulation Formats (NRZ, RZ, AMI, DB)

NRZ (NonChirped, Chirped, Alternate-Chirp, VSB).osp

RZ (50%RZ, 33%RZ, 67%CSRZ, Chirped, Alternate-Chirp, VSB).osp

DB.osp

AMI.osp
For a dual-drive Mach-Zehnder modulator (MZM), the power transfer function is shown in Figure 10-1.

Figure 10-1: Power transfer function of a dual-dirve MZM (from Ref. [1])

The optical field transfer function TE(V1,V2) of the MZM is [1]:

V 1 + V 2 ψ⎞
j ⎛ π ------------------
- + ----
⎝ 2V π 2⎠ V1 – V
T E ( V 1, V 2 ) = e cos π ------------------2- – ψ
---- (10-1)
2V π 2

where V1 and V2 are the voltages applied on the two MZM arms, and ψ is an additional, temporally
constant phase shift in one of the arms, referred to as the modulator bias.

As can be seen, when V1 = V2, we can achieve a constant power transmission with some phase
modulation added to the signal; when V1 = -V2, the MZM won’t add any phase modulation to the input
signal, but the intensity of input signal is modulated.

235
INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

For all the projects we discussed in the example Intensity Modulation (NRZ, RZ, AMI, DB), the Bit Rate
is set to a constant value of 10 Gbps, so readers can compare the spectrum of different intensity
modulation formats directly.

(a) NRZ (NonChirped, Chirped, Alternate-Chirp, VSB) modulation format

Figure 10-2 shows the layouts for generating NonChirped, Chirped, Alternate-Chirp, and VSB NRZ
signals, separately.

236
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-2: Layout - NonChirped, Chirped, Alternate-Chirp, and VSB NRZ signals

The nonchirped NRZ signal is generated by biasing the dual-drive MZM modulator at the quadrature
point, which is 50% transmission, and driving from minimum to maximum transmission with a voltage
swing of Vpi [1]. For the chirped NRZ signal, it is generated by sending the nonchirped NRZ signal to an

237
INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

additional phase modulator, which is driven by a sine wave generator working at the same bit rate as the
NRZ signal. When it comes to the alternate-chirp NRZ signal, a dual-drive MZM modulator (acting as a
phase modulator by setting equal voltage on the two arms) driven by a sine wave generator at half the
bit rate is used after the generation of nonchirped NRZ signal. VSB NRZ signal is generated by applying
an optical filter (centered with an offset from the optical carrier frequency) to the nonchirped NRZ signal
to suppress major parts of one sideband as well as perform some filter action on the other [1].

The output signal profile (both intensity and chirp) as well as the spectrum for NonChirped, Chirped,
Alternate-Chirp, and VSB NRZ signals can be seen in Figure 10-3.

Figure 10-3: Output signal profile (intensity and chirp) and spectrum of NRZ signals

NonChirped

Chirped

238
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-3: Output pulse profile (intensity and chirp) and spectrum of NRZ signals

Alternate-Chirp

VSB

(b) RZ (50%RZ, 33%RZ, 67%CSRZ, Chirped, Alternate-Chirp, VSB) modulation format

The layouts for generating 50%RZ, 33%RZ, 67%CSRZ, Chirped, Alternate-Chirp, and VSB RZ signals
are shown in Figure 10-4, separately.

239
INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-4: Layout - 50%RZ, 33%RZ, 67%CSRZ, and Chirped RZ signals

240
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-4: Layout - Chirped, Alternate-Chirp, and VSB RZ signals

RZ signal is usually obtained by carving a NRZ signal using an additional modulator. If this additional
modulator is sinusoidally driven at the data rate between minimum and maximum transmission, 50%RZ
(RZ pulse with a FWHM of 50% of the bit duration) signal can be generated; if this additional modulator
is sinusoidally driven at half the data rate between its transmission minima, then 33%RZ signal is

241
INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

generated; if this additional modulator is sinusoidally driven at half the data rate between its transmission
maxima, then 67%CSRZ (phase inversion between adjacent bits is caused because the field transfer
function of the MZM changes its sign when crossing the transmission minimum) signal can be obtained
[1].

For the chirped RZ signal, one way is to use a third MZM to introduce the additional phase modulation;
alternatively, chirped RZ signal can be generated using only one MZM, as shown in Figure 10-4. Two
methods of generating alternate-chirp RZ signal are also shown in Figure 10-4. More details about the
principles of the Chirped RZ generation - Using a single modulator, Alternate-chirp RZ generation -
Method1, and Alternate-Chirp RZ generation - Method 2 can be found in [2].

Similar to the generation of VSB NRZ signal, the VSB-CSRZ signal can be obtained by applying a filter
on the CSRZ signal.

The output signal profile (intensity and chirp) and spectrum for the 50%RZ, 33%RZ, 67%CSRZ, Chirped,
Alternate-Chirp, and VSB RZ signals are shown in Figure 10-5:

242
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-5: Output signal profile (intensity and chirp) and spectrum of RZ signals

50% RZ

33% RZ

67% CSRZ

243
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-5: Output signal profile (intensity and chirp) and spectrum of RZ signals

Chirped RZ with
three modulators

Chirped RZ with
a single modulator

Alternate-Chirp
RZ - Method 1

244
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-5: Output pulse profile (intensity and chirp) and spectrum of RZ signals

Alternate-Chirp
RZ - Method 2

VSB-CSRZ

(c) Duobinary modulation format

Two methods of generating a duobinary signal are shown in Figure 10-6, with details of principles
explained in [1].

245
INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-6: Layout - Duobinary signals

The corresponding output signal profile (intensity and phase) as well as spectrum are shown in
Figure 10-7.

246
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-7: Output signal profile (intensity and phase) and spectrum of Duobinary signals

Input
bit sequence

Output
Method 1

Output
Method 2

247
INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

(d) Alternate-mark inversion (AMI) modulation format

Two methods of generating an AMI signal are shown in Figure 10-8, with details of principles explained
in [1].

Figure 10-8: Layout - AMI signals

The corresponding output signal profile (intensity and phase) as well as spectrum are shown in
Figure 10-9.

248
 INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

Figure 10-9: Output signal profile (intensity and phase) and spectrum of AMI signals

Input
bit sequence

Output
Method 1

Output
Method 2

249
INTENSITY MODULATION FORMATS (NRZ, RZ, AMI, DB)

References:

[1] P. J. Winzer, R.-J. Essiambre, “Advanced optical modulation formats“, Proceedings of the IEEE, Vol.
94, No. 5, May 2006

[2] P. J. Winzer, C. Dorrer, R.-J. Essiambre, and I. Kang, “Chirped retrun-to-zero modulation by
imbalanced pulse carver driving signals“, IEEE Photonics Technology Letter, Vol. 16, No. 5, May 2004

250
 DPSK MODULATION AND DEMODULATION

DPSK Modulation and Demodulation

NRZ-DPSK.osp

33% RZ-DPSK.osp

50% RZ-DPSK.osp

67% RZ-DPSK.osp
In contrast with the intensity modulation formats, information of the differential binary PSK (DPSK) is
encoded on the binary phase change between adjacent bits: a 1 bit is encoded onto a pi phase change,
whereas a 0 bit is represented by the absence of a phase change [1]. The layouts for generating NRZ-
DPSK, 33% RZ-DPSK, 50% RZ-DPSK, and 67% RZ-DPSK signals are shown in Figure 10-10. Again,
all the simulation projects presented in this example have a bit rate equals to 10 Gbps.

Figure 10-10: Layout - NRZ-DPSK and 33% RZ-DPSK signals

NRZ-DPSK

33% RZ-DPSK

251
DPSK MODULATION AND DEMODULATION

Figure 10-10: Layout - 50% and 67% RZ-DPSK signals

50% RZ-DPSK

67% RZ-DPSK

For the NRZ-DPSK signal generation, the dual-driven MZM is biased at the maximum transmission point
and driven with voltage V1 = V2, so only the phase of the input light is being modulated. To get a RZ-
DPSK signal, a second modulator is used to carve the NRZ-DPSK signal, in a way similar to the
generation of RZ signal from NRZ signal. For the demodulation, the delay demodulation scheme is being
employed.

The corresponding output signal profile (intensity and phase) as well as the spectrum are shown in
Figure 10-11:

252
 DPSK MODULATION AND DEMODULATION

Figure 10-11: Output signal profile (intensity and phase) and spectrum of DPSK signals

Input
bit sequence

NRZ-DPSK

33% RZ-DPSK

253
DPSK MODULATION AND DEMODULATION

Figure 10-11: Output signal profile (intensity and phase) and spectrum of DPSK signals

Input
bit sequence

50% RZ-DPSK

67% RZ-DPSK

254
 DPSK MODULATION AND DEMODULATION

Reference:

[1] P. J. Winzer, R.-J. Essiambre, “Advanced optical modulation formats“, Proceedings of the IEEE, Vol.
94, No. 5, May 2006.

255
DPSK MODULATION AND DEMODULATION

Notes:

256
 QPSK MODULATION AND DEMODULATION

QPSK Modulation and Demodulation

NRZ-QPSK.osp

33% RZ-DQPSK-Method 1.osp

33% RZ-DQPSK-Method 2.osp

Compared with DPSK signal, the carrier phase of QPSK signal is allowed to take four different values,
thus leading to a transmission rate of 2 bits per symbol and an increased spectral efficiency. For the
projects presented in this example, the bit rate is set to 10 Gbps, and we can clearly see that the bandwith
of QPSK signals is half as that of BPSK signals.

(a) NRZ-QPSK modulation format

The layout for the generation of NRZ QPSK signal is shown in Figure 10-12.

Figure 10-12: Layout - NRZ-QPSK signal

The inner structures for the subsystems QPSK Modulator and Coherent Receiver are shown in
Figure 10-13.

257
QPSK MODULATION AND DEMODULATION

Figure 10-13: Subsystems - QPSK Modulator and Coherent Receiver

The corresponding NRZ-QPSK signal profile (intensity and phase), spectrum, as well as the constellation
diagram (after the coherent receiver) are shown in Figure 10-14.

258
 QPSK MODULATION AND DEMODULATION

Figure 10-14: (a) NRZ-QPSK signal profile (intensity and phase) (b) NRZ-QPSK signal spectrum and (c) the
constellation eye diagram after the coherent receiver.

(a) (b)

(c)

(b) RZ-DQPSK modulation format

Two methods for generating 33% RZ-DQPSK signals are shown in Figure 10-15 and Figure 10-16
separately. For the one shown in Figure 10-15 [1], the signal generated after the first MZM is a NRZ-
DPSK signal, a second MZM (acting as the phase modulator) is then used to generate the NRZ-DQPSK
signal. The third MZM, sinusoidally driven at Bit Rate / 4, is used for pulse carving. For the one shown in
Figure 10-16 [2], the first MZM is used for RZ pulse carving, and the next two MZMs in parallel are used
for RZ-DQPSK signal generation.

For the demodulation, the delay demodulation scheme is being employed.

259
QPSK MODULATION AND DEMODULATION

Figure 10-15: Layout - 33% RZ-DQPSK signal generation - Method 1

260
 QPSK MODULATION AND DEMODULATION

Figure 10-16: Layout - 33% RZ-DQPSK signal generation - Method 2

The corresponding RZ-DQPSK signal profile (intensity and phase) as well as the spectrum are shown in
Figure 10-17.

261
QPSK MODULATION AND DEMODULATION

Figure 10-17: Output signal profile (intensity and phase) as well as spectrum of 33% RZ-DQPSK signals

Method 1

Method 2

References:

[1] C. Wree, J. Leibrich, and W. Rosenkranz, “RZ-DQPSK format with high spectral efficiency and high
robustness towards fiber nonlinearities,” in Proc. ECOC’02, 2002, paper 9.6.6.

[2] P. J. Winzer, R.-J. Essiambre, “Advanced optical modulation formats“, Proceedings of the IEEE, Vol.
94, No. 5, May 2006.

262
 100 GBPS PM-QPSK

100 Gbps PM-QPSK

A design example of a 100 Gbps polarization-multiplexed quadrature phase-shift-keying (PM-QPSK)

system is shown in Figure 10-18.

Figure 10-18: Layout - 100 Gbps PM-QPSK system

The ideal constellation diagram and the degraded constellation diagram (when both CW laser and LO
linewidth are set to 1 MHz, fiber dispersion = 5ps/nm-km) are shown in Figure 10-19. The user can play
with different parametes in this project to see how they affect the received constellation diagram.

263
100 GBPS PM-QPSK

Figure 10-19: Ideal and degraded constellation diagrams

Ideal

Degraded

264
 OD8PSK MODULATION AND DEMODULATION

OD8PSK Modulation and Demodulation

OD8PSK.osp
The layout of the OD8PSK modulation format is shown in Figure 10-20, with the inner structures of the
subsystems Optical Encoder and Optical Demodulators and Receivers shown in Figure 10-21 and
Figure 10-22, separately.

Figure 10-20: Layout - OD8PSK signal

265
OD8PSK MODULATION AND DEMODULATION

Figure 10-21: Subsystem - Optical Encoder

266
 OD8PSK MODULATION AND DEMODULATION

Figure 10-22: Subsystem - Optical Demodulators and Receivers

The electrical encoder, optical encoder, and the optical demodulators and receivers are implemented
according to [1]. The generated OD8PSK signal profile (intensity and phase) as well as spectrum are
shown in Figure 10-23.

267
OD8PSK MODULATION AND DEMODULATION

Figure 10-23: Output signal profile (intensity and phase) as well as spectrum of OD8PSK signals

Reference:

[1] C. Kim and G. Li, “Direct-detection optical differential 8-level phase-shift keying (OD8PSK) for
spectrally efficient transmission,“ Optics Express, Vol. 12, No. 15, July 2004

268
 4-QAM OFDM MODULATION AND DEMODULATION

4-QAM OFDM Modulation and Demodulation

Direct-detection optical OFDM.osp

Coherent-detection optical OFDM.osp

(a) Direct-detection optical OFDM design

The layout of a direct-detection optical OFDM system is shown in Figure 10-24.

Figure 10-24: Layout - Direct-detection optical OFDM system

The generated RF OFDM signal spectrum after the quadrature modulator is shown in Figure 10-25.

269
4-QAM OFDM MODULATION AND DEMODULATION

Figure 10-25: RF OFDM signal spectrum after the quadrature modulator

The optical OFDM signal spectrum after the MZM modulator, filtering, and fiber propagation are shown
in Figure 10-26.

Figure 10-26: Optical OFDM signal spectrum after (a) MZM modulator (b) filtering and (c) fiber propagation

(a) (b) (c)

The received constellation diagram after quadrature demodulation and OFDM demodulation is shown in
Figure 10-27.

270
 4-QAM OFDM MODULATION AND DEMODULATION

Figure 10-27: Received constellation diagram after demodulation

(b) Coherent-detection optical OFDM design

The layout of a coherent-detection optical OFDM system is shown in Figure 10-28.

Figure 10-28: Layout - Coherent-detection optical OFDM system

The generated optical OFDM signal spectrum before and after fiber propagation are shown in Figure 10-
29.

271
4-QAM OFDM MODULATION AND DEMODULATION

Figure 10-29: Optical OFDM signal spectrum before and after the propagation in fiber

Before After

The received constellation diagram after demodulation is shown in Figure 10-30.

Figure 10-30: Received constellation diagram after demodulation

272
Chapter 11 - Optical Signal Processing
This section contains the following examples.

• Nonlinear Optical Loop Mirror

• Parametric Amplifiers
• Wavelength Conversion Based on FWM in SOA
• Wavelength Conversion Based on XGM in SOA
• Wavelength Conversion Based on XPM in HNLF
• Data-Format Conversion Based on XPM in HNLF
• SPM Based 2R Regenerator for RZ Signal
• XPM Based 3R Regenerator for RZ Signal

273
Notes:

274
 NONLINEAR OPTICAL LOOP MIRROR

Nonlinear Optical Loop Mirror

Nonlinear optical loop mirror.osp

A nonlinear optical-loop mirror (NOLM) is a Sagnac interferometer in which different nonlinear phase
shifts are acquired by the counter-propagating waves inside the fiber loop [1]. The layout of the NOLM
example is shown in Figure 11-1:

Figure 11-1: Layout - Nonlinear optical loop mirror

The incoming light is splitted into two counterpropagating parts by the coupler. At the output, whether the
light is transferred or reflected is determined by the relative phase difference of the two
counterpropagating waves. The transmitted power as a function of the incident power is shown in the
report page, as in Figure 11-2:

275
NONLINEAR OPTICAL LOOP MIRROR

Figure 11-2: Transmitted power vs the incident power

Reference:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., New York,
2010.

276
 PARAMETRIC AMPLIFIERS

Parametric Amplifiers

Parametric amplifiers.osp

Project file “Parametric amplifiers.osp” shows an example of a single-pump parametric amplifier. The
pump wavelength is fixed at 1542.4 nm, while the wavelength of probe light is swept from 1450 nm to
1650 nm, so the parametric gain for different pump-signal detunings can be characterized. The gain
experienced by the probe light is calculated using the Dual Port WDM Analyzer component. The layout
and obtained parametric gain spectrum are shown in Figure 11-3 and Figure 11-4, respectively.

Figure 11-3: Layout - Parametric amplifiers

277
PARAMETRIC AMPLIFIERS

Figure 11-4: Parametric gain spectrum

Reference:

[1] M.-C. Ho, K. Uesaka, M. E. Marhic, Y. Akasaka and L. G. Kazovsky, “200-nm-Bandwidth fiber optical
amplifier combining parametric and raman gain,” IEEE Journal of Lightwave Technology, Vol. 19, No 7,
pp. 977-981, July 2001.

278
 WAVELENGTH CONVERSION BASED ON FWM IN SOA

Wavelength Conversion Based on FWM in SOA

Wavelength conversion based on FWM in SOA.osp

A wavelength converter based on FWM in SOA operating at 2.5 Gb/s is shown as an example in the
project file “Wavelength conversion based on FWM in SOA.osp”. The layout for the wavelength
conversion scheme is shown in Figure 11-5:

Figure 11-5: Layout - Wavelength conversion based on FWM in SOA

The pump laser has a frequency of 193.1 THz and a power of 10 dBm; while the NRZ signal has a
frequency of 193 THz and power of 5 dBm. The pump and signal light are combined using a 3 dB coupler
and then copropagate in a SOA, where the FWM effect occurs. After the SOA, an optical filter centered
at (2 x PumpFrequency - SignalFrequency) is used to select the FWM generated new frequency
component. The optical spectrum after the SOA is shown in Figure 11-6, where we can clearly see the
FWM generated new frequency components.

279
WAVELENGTH CONVERSION BASED ON FWM IN SOA

Figure 11-6: Spectrum after the SOA

The optical spectrum as well as time domain signal profile after the first optical filter are shown in
Figure 11-7.

Figure 11-7: Spectrum and time domain signal profile after the first optical filter

After amplifying and filtering out-of-band noise, the eye diagram obtained after the wavelength
conversion process is shown in Figure 11-8.

280
 WAVELENGTH CONVERSION BASED ON FWM IN SOA

Figure 11-8: Eye diagram of the wavelength converted output

References:

[1] J. M. H. Elmirghani and H. T. Mouftah, “All-optical wavelength conversion: Technologies and

applications in DWDM networks,” IEEE Communications Magazine, March 2000.

[2] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., New York,
2010.

281
WAVELENGTH CONVERSION BASED ON FWM IN SOA

Notes:

282
 WAVELENGTH CONVERSION BASED ON XGM IN SOA

Wavelength Conversion Based on XGM in SOA

Wavelength conversion based on XGM in SOA -

CounterPropagation.osp

Wavelength conversion based on XGM in SOA - CoPropagation.osp

Two schemes of wavelength conversion based on XGM in SOA, namely, CounterPropagation and
CoPropagation, are shown in this example. The bit rate for both scheme is set to 2.5 Gb/s.

Figure 11-9: Layout - Wavelength conversion based on XGM in SOA

CounterPropagation

CoPropagation

The original signal profile, wavelength-converted signal profile, as well as the eye diagram of the
wavelength converted output,are shown in Figure 11-10.

283
WAVELENGTH CONVERSION BASED ON XGM IN SOA

Figure 11-10: Comparison of the signal profiles before and after wavelength conversion

CounterPropagation CoPropagation

Input signal

Wavelength
converted
output

Output
eye diagram

284
 WAVELENGTH CONVERSION BASED ON XGM IN SOA

Reference:

[1] J. M. H. Elmirghani and H. T. Mouftah, “All-optical wavelength conversion: Technologies and

applications in DWDM networks,” IEEE Communications Magazine, March 2000.

[2] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., New York,
2010.

285
WAVELENGTH CONVERSION BASED ON XGM IN SOA

Notes:

286
 WAVELENGTH CONVERSION BASED ON XPM IN HNLF

Wavelength Conversion Based on XPM in HNLF

Wavelength conversion based on XPM in HNLF.osp

In the project file “Wavelength conversion based on XPM in HNLF.osp”, the nonlinear phenomenon of
XPM is employed to achieve the wavelength conversion. The layout is shown in Figure 11-11:

Figure 11-11: Layout - Wavelength conversion based on XPM in HNLF

The wavelength conversion scheme is demonstrated at 10 Gb/s. The signal (@ 1545 nm), which is a
sequence of gaussian pulses, is amplifed and then copropagate with a CW probe (@ 1554 nm) in a
HNLF. The CW probe will experience XPM induced spectral broadening, and the optical spectrum before
and after the HNLF are shown in Figure 11-12.

287
WAVELENGTH CONVERSION BASED ON XPM IN HNLF

Figure 11-12: Spectrum before and after the HNLF

Before After

After the HNLF, a BPF, which is 0.8 nm offset from the center wavelength of the CW laser, is used to filter
out the original frequency components at 1554 nm and keep the spectral components that have been
shifted by XPM. The original signal profile and the wavelength-converted output after the BPF are shown
in Figure 11-13.

Figure 11-13: Comparison of the signal profile before and after wavelength conversion

Before After

The eye diagram of the wavelength-converted output is shown in Figure 11-14.

288
 WAVELENGTH CONVERSION BASED ON XPM IN HNLF

Figure 11-14: Eye diagram of the wavelength converted output

If the BPF is aligned exactly to the CW laser wavelength, then a polarity-inversed output can be obtained.
The reader can play with the filter center wavelength as well as the bandwidth to see how the output will
be affected.

Reference:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., New York,
2010.

289
WAVELENGTH CONVERSION BASED ON XPM IN HNLF

Notes:

290
 DATA-FORMAT CONVERSION BASED ON XPM IN HNLF

Data-Format Conversion Based on XPM in HNLF

NRZ to RZ format conversion based on XPM in HNLF.osp

RZ to NRZ format conversion based on XPM in HNLF.osp

The project file “NRZ to RZ format conversion based on XPM in HNLF.osp” shows an example of 10 Gb/s
all-optical NRZ to RZ format conversion using single HNLF assisted by optical bandpass filter. A 10 Gb/s
NRZ signal at 1554 nm (acting as the probe) is combined with a synchronized RZ clock (acting as the
pump) at 1545 nm and then launched into the HNLF. The pulsewidth of the RZ clock is ~ 5 ps. Due to
the high power of the RZ clock, the NRZ signal will experience XPM induced spectral broadening. An
subsequent optical BPF is then used to extract the sideband spectrum with an offset to the original NRZ
wavelength. Since the spectral shift occurs only during the RZ pulse duration, a RZ pulse train which has
the same data information as the NRZ signal can thus be obtained after the offset filtering. More details
about the principle of this simulation can be seen in [1].

The project file “RZ to NRZ format conversion based on XPM in HNLF.osp” shows the example of
spectral filtering from a cross-phase modulated signal for RZ to NRZ format conversion. The principle is
similar to the XPM based wavelength conversion and is detailed in [2].

Reference:

[1] J. Dong, X. Zhang, J. Xu, D. Huang, S. Fu, and P. Shum, “40Gb/s all-optical NRZ to RZ format
conversion using single SOA assisted by optical bandpass filter,” Opt. Express 15, 2907-2914 (2007).

[2] S. H. Lee, K. Chow, and C. Shu, “Spectral filtering from a cross-phase modulated signal for RZ to NRZ
format and wavelength conversion,” Opt. Express 13, 1710-1715 (2005).

291
DATA-FORMAT CONVERSION BASED ON XPM IN HNLF

Notes:

292
 SPM BASED 2R REGENERATOR FOR RZ SIGNAL

SPM Based 2R Regenerator for RZ Signal

SPM based 2R regenerator for RZ signal.osp

Project file “SPM based 2R regenerator for RZ signal.osp” shows the improvement of eye diagram after
passing the signal throught the SPM based 2R regenerator. The principle of SPM based 2R regenerator
is already explained in [1-2] and the regeneration property comes from the different power transfer
function for “0“s and “1“s.

At the same receiver power level of -29.3 dBm, the recovered eye diagrams without/with 2R regeneration
are shown in Figure 11-15. We can clearly see the improvement with 2R regeneration.

Figure 11-15: Eye diagrams without and with 2R regeneration

Reference:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., New York,
2010.

[2] M. Rochette, L. Fu, V. Ta’eed, D. J. Moss, and B. J. Eggleton, “2R optical regeneration: An all-optical
solution for BER improvement,” IEEE Journal of Selected Topics in Quantum Electronics, Vol. 12, No. 4,
July/August 2006.

293
SPM BASED 2R REGENERATOR FOR RZ SIGNAL

Notes:

294
 XPM BASED 3R REGENERATOR FOR RZ SIGNAL

XPM Based 3R Regenerator for RZ Signal

XPM based 3R regenerator for RZ signal.osp

Compared with 2R regenerator, a 3R regenerator also performs the retiming function and reduces the
timing jitter of a bit stream. Project file “XPM based 3R regenerator for RZ signal.osp” shows an example
of 3R regenerator based on XPM in HNLF [1-2].

Time jitter and white light source are added to the input signal to degrade the input signal quality. At the
same receiver power level of -26.9 dBm, the recovered eye diagrams without/with 3R regeneration are
shown in Figure 11-16. We can clearly see the improvement with 3R regeneration.

Figure 11-16: Eye diagrams without and with 3R regeneration

References:

[1] G.P. Agrawal, Fiber-Optic Communication Systems, 4th ed., John Wiley & Sons, INC., New York,
2010.

[2] M. Rochette, J. L. Blows, and B. J. Eggleton, “3R optical regeneration: An all-optical solution with BER
improvement,” Optics Express, Vol. 14, No. 14, July 2006.

295
XPM BASED 3R REGENERATOR FOR RZ SIGNAL

Notes:

296
Optiwave
7 Capella Court
Ottawa, Ontario, K2E 7X1, Canada

Tel.: 1.613.224.4700
Fax: 1.613.224.4706

E-mail: [email protected]
URL: www.optiwave.com