University of Southern California Dissertations and Theses

Applications of all-optical wavelength shifting using semiconductor optical amplifiers for switching and routing functions in a dynamically reconfigurable wavelength-division-multiplexed fiber-op...

(USC Thesis Other)

Applications of all-optical wavelength shifting using semiconductor optical amplifiers for switching and routing functions in a dynamically reconfigurable wavelength-division-multiplexed fiber-op...

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell A Howell Information Company
300 North Zeeb Road, Ann Arbor MI 48106-1346 USA
313/761-4700 800/521-0600

APPLICATIONS OF ALL-OPTICAL WAVELENGTH SHIFTING
USING SEMICONDUCTOR OPTICAL AMPLIFIERS FOR
SWITCHING AND ROUTING FUNCTIONS IN A DYNAMICALLY
RECONFIGURABLE WAVELENGTH-DIVISION-MULTIPLEXED
FIBER-OPTIC COMMUNICATION NETWORK
by
Eugene Park
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Electrical Engineering)
May 1996
Copyright 1996 Eugene Park
UMI Number: 9636739
Copyright 1996 by
Park, Eugene
All rights reserved.
UMI Microform 9636739
Copyright 1996, by UMI Company. All rights reserved.
This microform edition is protected against unauthorized
copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, MI 48103
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES. CALIFORNIA 90007
This dissertation, written by
Eugene P ark
under the direction of hkf. Dissertation
Committee, and approved by all its members,
has been presented to and accepted by The
Graduate School, in partial fulfillment of re
quirements for the degree of
Date
DISSERTATION COMMITTEE
DOCTOR OF PHILOSOPHY
Dean of Graduate Studies
Chairperson
To my father, mother, and brother for their everlasting love and support.
Acknowledgements
I would like to thank my advisor and dissertation committee
chairman, Dr. Alan E. Willner, for providing me with opportunity,
guidance, and mentorship throughout my graduate school career. I
would also like to extend my great appreciation to the other dissertation
committee members, Dr. P. Daniel Dapkus and Dr. Murray Gershenzon.
In addition, I would like to thank Dr. Elsa Garmire and Dr. Antonio
Mendez for their support in the early portion of my graduate career.
I would like to acknowledge those people and organizations th at
have made invaluable contributions which have allowed me to undertake
the research projects presented in this dissertation. They are: N. Dutta,
A. Piccirilli, J. Zyskind, T. Tanbun-Ek, and T. Uno.
I would like to recognize the following funding agencies which have
supported the research presented in this dissertation. They are: The
National Science Foundation's All-Optical Networks Initiative,
Presidential Faculty Fellows Award, and Young Investigator's Award;
the Packard Foundation Fellowship; and the BMDO/AFOSR Focused
Research Initiative.
There are many others who must be acknowledged as people who
have been of utmost support. They are: My family--Dr. Harry S. Park,
Mrs. Kyung Soon Park, and Dale Park; and my colleagues— Dr. Shouhua
Huang, Dr. XingYu Zou, Dr. Syang-Myau Hwang, Jam es Leight, Jieh-
Chian Wu, Imran Hayee, Mustafa Cardakli, Bogdan Hoanca, Kai-Ming
Feng, Jong-Jin Yoo, and Jin-Xing Cai. Two colleagues which deserve
special recognition are Dr. David Norte and Dr. William Shieh, who have
both undertaken similar research endeavors and have provided
invaluable assistance and insightful discussions.
I would like to thank the many friends th at have provided me with
a tremendous amount of emotional support throughout or a t some point
in my graduate career. They are: Dr. Michael MacDougal, Dr. Anne
Blood, Dr. Sabeur Siala, Beckie Kravetz, Lynn Hill, Angie Massel, Dr.
Greg Mitchell, Dr. Steve and Lisa Stockman, Alexis Alexiades, Dr. Lin
Chase, Alessandra Cunha, Dr. Teresa George, Chris Haller, Lisa Hunter,
Dr. Michael and Katy Jupina, Jam es Knoebber, Andy Gee, Dr. Eliav
Haskal, Susan Hayashida, Dr. Steve Hummel, Dr. Steffen Koehler,
Tennye Kohatsu, Shana Marowitz, Natalie Masson, Sandra Myers, Dean
Nakasone, Ed Perkins, Dr. Cathy Priest, Dr. Dave Rutan, Adel Saidi,
Anne and Don Watson, and Chris Werner.
Last but not least, I'd like to acknowledge the various people who
have contributed in some way to the successes of my research and
academic endeavors. They are: Milly Montenegro, Hermine Fermanian,
Kim Reid, Margery Berti, Marylin Ikegami, Bill Johnson, Joshua Davis,
Christopher Brovm, Mayumi Thrasher, Natalie Neilson, Dr. Michael
Schneir, Dr. William Spitzer, Dr. Bill Steier, Dr. Vinod Ramakrishnan,
Anna Fong, Mona Gordon, Mary Froehlig, Chuck Anderson, and John
Quintus.
v
Table of Contents
List of Figures..........................................................................................................ix
Abstract.................................................................................................................xiii
Preface.................................................................................................................xv
Chapter 7. Introduction......................................................................................7
1.1 Progress in Optical Communications Research................................. 1
1.1.1 The "Highway" Analogy.............................................................. 2
1.1.2 Wavelength Division Multiplexed (WDM) Networks............. 4
1.1.3 WDM Network Enhancement Using All-Optical Wavelength
Shifting for Dynamic Reconfigurability.............................................. 7
1.2 Looking Ahead-Enabling Technologies for All-Optical Networks:
Demonstrations and Analysis................................................................... 11
1.2.1 Receiver Sensitivity and Power Penalty.................................12
1.3 Dissertation Organization................................................................... 13
1.4 References.............................................................................................. 15
Chapter 2. All-Optical Wavelength Shifting Using Semiconductor
Optical Amplifier Cross-Gain Compression..............................................77
2.1 Introduction........................................................................................... 17
2.1.1 Facet Reflectivity....................................................................... 19
2.1.2 Polarization Independence...................................................... 21
2.1.3 Gain Saturation..........................................................................22
2.2 All-Optical Wavelength Shifting........................................................ 26
2.2.1 All-Optical Wavelength Shifting Using SOA Cross-Gain
Compression......................................................................................... 26
2.2.2 Wavelength Requirements....................................................... 28
2.2.3 C hirp............................................................................................29
2.2.4 Multiple-Input Wavelength Shifting...................................... 30
2.3 Measured SOA Characteristics.......................................................... 30
2.4 Chapter Summary................................................................................ 32
2.5 References..............................................................................................33
Chapter 3. Other Wavelength-Shifting Methods.................................... 38
3.1 Cross-Phase Modulation......................................................................38
3.2 Four-Wave Mixing................................................................................ 39
3.3 Laser Carrier Depletion.......................................................................41
3.4 Non-linear Optical Loop Mirror.......................................................... 42
3.5 References..............................................................................................43
Chapter 4. Simultaneous Wavelength Shifting and Header
R eplacem ent..................................................................................................... 45
4.1 Motivation..............................................................................................45
4.2 Three-Level Probe Modulation........................................................... 47
4.3 Experimental Design............................................................................50
4.3.1 Electronic Processor...................................................................51
4.4 Results and Discussion........................................................................53
4.4.1 Sensitivities................................................................................58
4.5 Issues and H urdles...............................................................................60
4.6 Chapter Summary.................................................................................62
4.7 References..............................................................................................63
Chapter 5. Self-Routing of Wavelength Packets Using Subcarrier
Control Headers and All-Optical Wavelength Shifting........................ 65
5.1 Motivation..............................................................................................65
5.2 Subcarrier-Multiplexed Control......................................................... 68
5.2.1 Subcarrier Modulation Technique........................................... 68
5.3 Experimental Design............................................................................69
5.3.1 Electronic Processor...................................................................72
5.4 Results and Discussion........................................................................73
5.4.1 Sensitivities................................................................................75
5.5 Issues and H urdles.......................................... 76
5.6 Chapter Summary.................................................................................78
5.7 References.............................................................................................. 78
Chapter 6. Multiple-Wavelength-lnput All-Optical Wavelength-
Shifting with Subcarrier-Multiplexed Control for Self-Routing................ 80
6.1 Motivation..............................................................................................80
6.2 Experimental Design............................................................................81
6.3 Results and Discussion........................................................................83
6.3.1 Sensitivities................................................................................84
6.4 Chapter Summary................................................................................ 85
6.5 References..............................................................................................86
Chapter 7. All-Optical Wavelength Shifting of Baseband Data and
Subcarrier-Multiplexed Signals...................................................................... 87
7.1 Motivation..............................................................................................87
7.2 Computer Modeling.............................................................................. 89
7.3 Experimental Design............................................................................90
7.4 Subcarrier Modulation Depth............................................................. 93
7.5 Results and Discussion........................................................................93
7.6 Chapter Summary................................................................................ 96
7.7 References..............................................................................................96
Chapter 8. Conclusion.................................................................................... 98
Bibliography.......................................................................................................700
Appendix A .........................................................................................................709
Appendix B.........................................................................................................770
List of Figures
Figure 1.1. Optical-to-electrical and electrical-to-optical conversions
create a speed bottleneck..................................................................................2
Figure 1.2. Optical amplification eliminates speed bottlenecks...............3
Figure 1.3. Optical amplification enables WDM transmission, (a) Fixed
transm itters tunable receivers, (b) Tunable transm itters, fixed receivers.
.............................................................................................................................. 4
Figure 1.4. Loss spectra of typical optical fiber as a function wavelength
and Gain spectra of typical SOAs, PDFAs, and EDFAs..............................6
Figure 1.5. All-optical wavelength shifting allows data to change
wavelength-paths...............................................................................................7
Figure 1.6. Wavelength shifting enables the reuse of wavelengths within
a large network...................................................................................................9
Figure 1.7. Wavelength shifting enables packet wavelength-rerouting. 10
Figure 1.8. The gateway node utilizes a wavelength shifter which routes
data from the WAN to the intended LAN node...........................................11
Figure 1.9. Typical receiver sensitivity curve..............................................13
Figure 2.1. Typical SOA................................................................................. 18
Figure 2.2. Behavior as a function of injection current. After [21].........19
Figure 2.3. SOA with buried facets. After [24]......................................... 21
Figure 2.4. SOA with angled-stripe structure........................................... 21
Figure 2.5. Modulation of SOA gain due to intense input signal............. 25
Figure 2.6. All-optical wavelength shifting using SOA cross-gain
compression...................................................................................................... 26
Figure 2.7. Gain-compression due to strong pump.................................... 28
Figure 2.8. Measured SOA spectrum. Bias current = 80 mA, T = 22.5° C.
............................................................................................................................ 31
Figure 2.9. Experimentally measured gain vs. fiber input power at 1571
nm for different SOA bias currents...............................................................32
Figure 3.1. Interferometric wavelength shifters: (a) Mach-Zender, (b)
Michelson...........................................................................................................39
Figure 3.2. Four-wave mixing spectrum...................................................... 40
Figure 3.3. Carrier depletion in a DBR laser. After [8]........................... 41
Figure 4.1. The presented method replaces the old header of an incoming
packet with a new header and simultaneously shifts the entire packet
onto a new wavelength. The complement bars indicate wavelength-
shifted data.......................................................................................................46
Figure 4.2. (a) The "truth table" shows which probe level is required for
obtaining the desired bit change, (b) Illustration of a three-level probe
signal required to simultaneously perform bit-by-bit header
replacement, (0 100101 l)->(6oToTToo), and wavelength 49
Figure 4.3. Experimental setup and pre-filtered optical spectrum 51
Figure 4.4. Logic diagram using a 4-bit header example to illustrate
operation of the three-level circuit. = RF power combiner. Gating
signal is derived from the master clock.......................................................51
Figure 4.5. Scope traces (different vertical scales) of (a) Apump incoming
packet, (b) time-magnification of old header and portion of payload, (c)
three-level probe signal, (d) wavelength-shifted new header and portion
of payload, and (e) wavelength-shifted outgoing packet............................ 54
Figure 4.6. Three-level eye-diagrams of (a) RF power combiner output
and (b) probe laser prior to the SOA input...................................................55
Figure 4.7. Eye-diagram of header time-frame..........................................55
Figure 4.8. Wavelength-shifted signal (a) without header replacement
and (b) with header replacement...................................................................56
Figure 4.9. Eye-diagrams of wavelength-shifted signals using (a) optimal
and (b) non-optimal pump/probe powers. Same vertical scales..............57
Figure 4.10. Measured sensitivity curves for, from right to left, (a)
simultaneous header replacement and wavelength shifted data, (b)
wavelength shifted data without header replacement (same conditions
as previous case) (c) optimized wavelength shifted data, and (d) baseline
pump data.........................................................................................................58
Figure 5.1. Conceptual diagram of a wavelength-router..........................67
Figure 5.2. Implementation of subcarrier-multiplexed control for
baseband data self-routing............................................................................68
Figure 5.3. Diagram of experiment. The optical spectrum is a maximum-
held trace showing all wavelengths signals. The SOA gain peak is a t
1565 nm, and the bias is set to 180 mA......................................................69
Figure 5.4. Details of the Wavelength Router........................................... 71
Figure 5.5. Logic diagram of the electronic processor...............................7 2
Figure 5.6. Oscilloscope traces of the transm itted packets, recovered
control signals, probe-laser signals, and self-routed baseband
wavelength packets. Details of the guard time are also shown. Traces of
self-routed packets are taken using a trigger derived from the low-speed
clock....................................................................................................................74
Figure 5.7. Sensitivity curves for 1-Gh/s payload data which have been
routed from X-path A to X-paths 1 through 4. Eye-diagrams of packets
and a 50-Mb/s control signal are also shown. Intersymbol interference of
the control signal is attributed to the QPSK assembly............................7 5
Figure 6.1. Diagram of a Multiple-Channel Wavelength Shifter for a
WDM routing node...........................................................................................81
Figure 6.2. Packets on Xa (= 1542 nm) are either down-shifted to Xi (=
1535 nm) or up-shifted to Xb’ (= 1557 nm) using S O A a. Packets Xb (=
1557 nm) are either down-shifted to Xa’ (= 1542 nm) or up-shifted to X2
(= 1571 nm) using S O A b.................................................................................................. 8 2
Figure 6.3. Scope traces showing X-shifted self-routed packets. 83
Figure 6.4. Sensitivity curves for wavelength-routed packets.................84
Figure 7.1. (a) Computer modeled baseband data multiplexed with an
encoded subcarrier. Input signals are RF filtered to avoid crosstalk, (b)
All optical X-shifter and computer modeled ^.-shifted output.
Baseband+subcarrier traces are time-expanded........................................89
Figure 7.2. Experimental setup along with RF and optical spectra at
specific points in the setup. The SOA gain peak is 1565 nm, and the bias
is 140 mA. (OA=optical attenuator, HPF=high-pass filter, LPF=low-
pass filter).........................................................................................................91
Figure 7.3. Sensitivities @ 10'9 BER as a function of m measured (a)
before and (b) after the X-shifter. Each plot shows a crossover indicating
an optimal point for 10'9 BER operation of both the baseband and
subcarrier.......................................................................................................... 94
Figure 7.4. BER sensitivity curves for m=0.32.......................................... 95
Abstract
The deployment of an all-optical network infrastructure will
enable ultra-high-capacity, ultra-high-bandwidth, data-format and data-
rate independent links between users. In all-optical networks, there are
no optical-to-electrical (O/E) and electrical-to-optical (El/O) conversions
in the data path, eliminating speed bottlenecks. These networks will
utilize wavelength-division multiplexing (WDM) in which many high
speed data stream s are simultaneously transm itted on separate
wavelengths. All-optical networks will also be dynamically
reconfigurable, in which the network-path configuration will change in
response to data-routing requirements, traffic load, and broken links.
This dissertation presents experimental demonstrations of novel
switching and routing functions which are applicable to dynamically
reconfigurable all-optical WDM networks. These networks which use
wavelengths as routing paths may require high-speed all-optical
wavelength shifting in which high-speed data is transferred from one
wavelength to another without O/E and E/O conversions. The following
novel applications and experimental results will be presented: (1) "On-
the-fly" simultaneous packet header replacement and wavelength
shifting; (2) self-routing of wavelength packets using wavelength shifting
and quadrature-phase-shift keyed (QPSK) subcarrier-multiplexed
control; (3) multi-channel-input wavelength shifting; and (4) wavelength
shifting of combined baseband and subcarrier signals. In all
demonstrations, the method of all-optical wavelength shifting uses
cross-gain compression of semiconductor optical amplifiers. Data rates
of 1 Gb/s and 622 Mb/s are used, and 10'9 bit-error rates are achieved in
all cases.
xiv
Preface
The Optical Communications Laboratory at the University of
Southern California was formed in 1992, under the direction of Dr. Alan
E. Willner. The members of this laboratory represent a cross-section of
disciplines including electrophysics, communication systems, networks,
and physics. The main focus of research is wavelength-division
multiplexed optical communication systems. The work presented in this
dissertation proudly represents some of the first experimental research
performed in this laboratory.
xv
Chapter 1
Iniroduction
7 . 7 Progress in Optical Communications Research
Optical communications has proven to be a highly efficient means
of transporting information. With the invention of the semiconductor
laser and optical fiber, communicating via optical fiber has replaced
copper wire as the preferred method of long-distance transmission. The
envisioned deployment of an "all-optical" network infrastructure will
enable ultra-high-capacity, ultra-high-bandwidth, data-format and data-
rate independent finks between users. This infrastructure would be
greatly beneficial to many entities including the telephone/cable TV
companies, internet service providers, and the general public. A
multitude of voice conversations, high-definition real-time video, medical
images, world-wide web images, and supercomputer data would all be
able to be transm itted simultaneously and all-optically across the
country (and eventually the world). This chapter begins with a "highway
analogy" to illustrate the general progress of optical communications
research, beginning with first-generation single-wavelength systems to
current research in multi-wavelength systems.
1.1.1 The "Highway" Analogy
Figure 1.1 illustrates the first generation of optical
communication systems using the analogy between data transmission
on an optical fiber and automobiles commuting on a highway. This
analogy is in reference to the popularized term, "information
superhighway."
Figure 1.1. Optical-to-electrical and electrical-to-optical conversions create a
speed bottleneck.
First-generation systems utilized either light emitting diodes
(LEDs) or single wavelength lasers to transm it data. To overcome fiber
losses after kilometers of transmission, periodic electrical regeneration
was required. This involved optical-to-electrical (O/E) conversion,
electrical amplification, and electrical-to-optical (E/O) conversion; these
L aser
T ransm itter
W avelength
* 1
L aser
Transm itter
W avelength
^ D etector
Electrical Electrical Electrical-to-
C onversion Amplification/ Optical
Electrical
R egeneration C onversion^
TRAFFIC JAMMED
2
conversions create a speed bottleneck. The automobiles in Figure 1.1
represent data "packets," which are utilized in a type of data-
transm ission architecture called packet-switching, and the speed
bottleneck is represented by the toll booth.
Laser
Transmitter
Xi Signal
o
Wavelength
*1
- ►
Optical
Amplifier,
O A .1 Signal
TRAFFIC CLEAR
Figure 1.2. Optical amplification eliminates speed bottlenecks.
Extensive research into methods of optical amplification has
resulted in the invention of the erbium-doped fiber amplifier [1] (EDFA)
and the semiconductor optical amplifier (SOA) [2]-[3]. Both the EDFA
and the SOA amplify an input optical signal completely in the optical
domain, th a t is, there is no O/E and E/O conversions. Hence, an "all-
optical" data path is created. The result is an elimination of the speed
bottleneck. This is illustrated in Figure 1.2 by the automobiles now
traveling in a carpool lane.
1.1.2 W avelength D ivision M ultiplexed (WDM) N etw orks
A consequence of the invention of SOAs and EDFAs is th at efforts
in multi-wavelength transmission research have exploded! A prime
Multiplexer
(b)
Tx,A,i
Tx,Aj
~ r ^ r
Tx, X k |
"nT
Tunable
L asers
Multiplexer
n
Opt*
Amplifier
T ransm itted
W avelengths
n
Opt»
Amplifier
^ 1 . fy ,
Transm itted
W avelengths
TRAFFIC CLEAR
Splitter/
Demultiplexer
Tunable
O ptical
Filter
r-H
~r f
D etector
Splitter/
Demultiplexer
Fixed
Optical
Filter
Rx 1
Rx2 .
B - M
Rx3 i
t—r
D etector
Figure 1.3. Optical amplification enables WDM transmission, (a) Fixed
transm itters, tunable receivers, (b) Tunable transmitters, fixed receivers.
4
motivation is th at the SOAs and EDFAs are able to amplify a range of
wavelengths simultaneously. Simultaneous transmission of
independent data streams on different wavelength channels is called
w avelen gth -d ivision m ultiplexing (WDM). This is illustrated in
Figure 1.3 as different color cars traveling in their own lanes and at their
own speeds. Figure 1.3 shows an example of three wavelengths, however,
there has been demonstrations using 100 wavelengths [4]! In a very
recent capacity demonstration, transmission of 1.1 terabits/second was
demonstrated [5]. That's over one trillion bits of information in one
second! The method used 55 wavelengths, each transm itting at 20
gigabits/second.
SOAs have approximately a 50-nm bandwidth and can amplify
WDM channels in either the 1300-nm or the 1550-nm low-loss window of
optical fiber (See Figure 1.4). EDFAs have approximately a 25-nm
uniform bandwidth and can amplify WDM channels in the 1550-nm low-
loss window. Because many of the systems currently in place operate in
the 1300-nm regime, there is a great research effort into using
praseodymium-doped fiber amplifiers (PDFAs) for use within this
regime [6]. However, these amplifiers have not reached the current state
of m aturity as EDFAs.
5
Fiber
Loss
Spectra
_i
~ 0.2 dB/km
1550 1300
Optical
Amplifier
Gain
Spectra
▲
EDFAs
PDFAs
1550 1300
Wavelength (nm)
Figure 1.4. Loss spectra of typical optical fiber as a function wavelength and
Gain spectra of typical SOAs, PDFAs, and EDFAs.
WDM can be utilized in a point-to-point architecture, as shown in
Figure 1.3. Each user has a signature wavelength. In a fixed-
transm itter scheme (Figure 1.3a), the receiver would tune it's bandpass
optical filter to detect the wavelength of interest. In a fixed-receiver
scheme (Figure 1.3b), the transm itter would time it's laser wavelength to
correspond with the intended receiver's fixed filter; a means to avoiding
contention (two transm itters intending to simultaneously communicate
with the same receiver) must be provided. The point-to-point
architecture suffers from physical and cost limitations due to the fact
6
th at as the number of users in the network increases, the number of
required lasers increases.
1.1.3 WDM N etw ork E nhancem ent U sing A ll-O ptical W avelength
S h iftin g for D ynam ic R econfigurability
Another way of viewing WDM is to realize that wavelengths can
be used as routing paths [7]-[14]. In addition to the optical fiber as the
physical path medium, each wavelength which is transm itted within the
fiber can act as a data path.
o
M
ALL-OPTICAL
WAVELENGTH
SHIFTER
A2
o
TRAFFIC CLEAR
Figure 1.5. All-optical wavelength shifting allows data to change wavelength-
paths.
Returning to the highway analogy, suppose th at the lane you are
driving in becomes congested, and you desire to switch to a less
congested lane. Because each lane is restricted to a certain color, to
7
switch lanes you would have to switch the color of the car you are riding
in. That would mean stopping your car, getting out, and then getting into
another car which has the color corresponding to the less-congested lane.
Imagine the bottleneck you would create with such an action! Now
suppose th at there was a tunnel on the highway (as shown in Figure 1.5)
inside which a group of graduate students would spray-paint your car the
desired color as you drove by so you could get into the less-congested lane
without slowing down at all. This situation analogously describes a
process called "all-optical w avelength shifting" (or "wavelength
conversion"). Data packets are transferred to another wavelength-path,
without O/E and El/O conversions, allowing the packets to reach their
destination in a more efficient manner by avoiding congested
wavelength-paths.
The concept of wavelength shifting is the focal point of this
dissertation. This dissertation will present experimental
demonstrations of switching and routing functions which are enhanced or
enabled by using all-optical wavelength shifting. The following are three
examples of how wavelength shifting can increase network functionality.
8
EXAMPLE 1. Wavelength Reuse.
W avelength
Shifter
NETWORK NODE
Figure 1.6. Wavelength shifting enables the reuse of wavelengths within a large
network.
Wavelength shifting enables different parts of a WDM network to
reuse the same wavelengths decreasing the total number of required
network-wide wavelengths and still provide full connectivity between all
nodes. Figure 1.6 shows two local-area networks using the wavelengths
and ta. The shaded node in Ring A wishes to transm it to the shaded
node in Ring B using A ,i. However, Xi is already being used in Ring B.
Wavelength shifting allows the data to be transferred to an unused
wavelength X .2 prior to entering Ring B.
9
EXAMPLE 2. Routing in a Packet-Switched Network.
Wavelength-
Router
Xi Path
Incoming Packet on A a
X 2 Path
Header
Processoi
Header Payload
Header Payload
All-Optical
Wavelength
Shifter
Outgoing Packet on A .2
Figure 1.7. Wavelength shifting enables packet wavelength-rerouting.
Figure 1.7 shows a wavelength router which is routing a packet
from a wavelength-path X a to Xz within a packet-switched network. In
such a network, packets are transm itted with the routing information
contained in the packet header. Routing nodes process the packet
header to determine how to direct the outgoing packet. Rerouting of a
packet may be necessary due to traffic congestion or to avoid contention.
EXAMPLE 3. Gateway Between a Wide-Area and Local-Area Network.
Figure 1.8 shows a gateway node which connects a wide-area
network (WAN) to a local-area network (LAN). The gateway node routes
incoming data from the WAN on wavelength Xa to one of the four LAN
intended destination.
10
From O
W A N-----
G atew ay
N od e
LAN
LAN NODE
Figure 1.8. The gateway node utilizes a wavelength shifter which routes data
from the WAN to the intended LAN node.
7.2 Looking Ahead-Enabling Technologies for All-Optical
Networks: Demonstrations and Analysis
An ultimate goal of optical communications research is to create a
network with the following features [8]-[15]: (1) All-optical-there should
be no O/E conversions between end users, although the network may
include electronics to control the data flow. (2) Transparency-the
network should be capable of handling a wide range of bit-rates and
data-formats. (3) Dynamically reconfigurable-the networks should be
able to alter its configuration and routing paths (spatial and
wavelength) based on routing requirements, traffic load, and link
outages.
11
Before a dynamically reconfigurable WDM network can be fully
realized, there are a number of technological challenges th a t m ust be
overcome. Much of the challenge is to demonstrate novel ideas that will
serve as functional building blocks for an all-optical network. This
dissertation presents a number of demonstrations, using novel concepts
and proven technologies, th at may be implemented in a dynamically
reconfigurable WDM network.
1.2.1 R eceiver S en sitivity and Pow er P enalty
A standard analysis criteria m ust be applied to any concept to
prove whether or not the concept can actually be implemented in a
realistic system. For communication networks, a figure of m erit is the
measured receiver sensitivity of a system. In general, if a 10'9 bit-error
rate (BER) (that is, one error in one billion transm itted bits, from
transm itter to receiver, or one error in a one second of 1-Gbit/s
transmission) can be achieved after data transmission through a
network or network component, it can be argued th at such a network or
network component could be a viable technology. The word "sensitivity"
refers to how effective a receiver can distinguish a mark ("l"-bit) from a
space ("0"-bit) in the midst of quantum (shot) noise and thermal noise.
12
The shot noise is due to the statistically random nature of light. The
thermal noise is due to dark current flowing through a detector and is a
function of the detector's termination resistance.
Another figure-of-merit called the "power penalty" is used to
indicate the amount of increased optical power a receiver requires to
achieve 10'9 BER compared to a baseline. The power penalty indicates
the amount of system degradation. Both these figure-of-merits are
illustrated in Figure 1.9.
R eceiver S en sitivity
(0
D C
0
IU
1
P o w e r
P e n a lty
R eceived P ow er (dBm)
Figure 1.9. Typical receiver sensitivity curve.
7.3 Dissertation Organization
This dissertation is organized in the following manner. Chapter 2
discusses a method of all-optical wavelength shifting using cross-gain
13
compression of semiconductor optical amplifiers. Chapter 3 discusses
other methods of all-optical wavelength shifting. Chapter 4 presents an
experimental demonstration of simultaneous all-optical wavelength
shifting and packet header replacement. Chapter 5 presents an
experimental demonstration of a wavelength-router which directs self-
routing wavelength packets using all-optical wavelength shifting and
subcarrier-multiplexed control. Chapter 6 presents an experimental
demonstration of a two-channel wavelength shifter which perform self
routing and wavelength interchange. Chapter 7 presents computer and
experimental demonstrations of all-optical wavelength shifting of both
baseband data and subcarrier control. The dissertation concludes with
Chapter 8.
14
1.4 References
[1] E. Desurvire, Erbium-Doped Fiber Amplifiers-Principles and
Applications. John Wiley & Sons Inc., 1994, New York.
[2] N. A. Olsson, "Semiconductor optical amplifiers," Proc. of the
IEEE, vol. 80, no. 3, pp. 375-382, Mar. 1992.
[3] S. Kobayashi and T. Kimura, "Semiconductor optical amplifers,"
IEEE Spectrum, pp. 26-31, May 1984.
[4] H. Toba, K. Oda, K. Nakanishi, N. Shibata, K. Nosu, N. Takato,
and M. Fukuda, "A 100-channel optical FDM transmission/
distribution at 622 Mb/s over 50 km," IEEE/O SA J. Lightwave
Technol., vol. 8, no. 9, pp. 1396-1401, 1990.
[5] H. Onaka, H. Miyata, G. Ishikawa, K. Otsuka, H. Ooi, Y. Kai, S.
Kinoshita, M. Seino, H. Nishimoto, and T. Chikama, "1.1 Tb/s
WDM transmission over a 150 km 1.3 pm zero-dispersion single
mode fiber," Opt. Fiber Commun. Conf., OFC ‘96, San Jose, CA,
postdeadline paper PD19.
[6] T. J. Whitley, "A review of recent system demonstrations
incorporating 1.3-pm praseodymium-doped fluoride fiber
amplifiers," IEEE/O SA J. Lightwave Technol., vol. 13, no. 5, pp.
744-760, 1990.
[7] K. Sato, S. Okamoto, and H. Hadama, "Network performance and
integrity enhancement with optical path layer technologies,"
IEEE J. Select. Areas Commun., vol. 12, no. 1, pp. 159-170, Jan.
1994.
[8] A. A. M. Saleh, "Transparent optical networks for the next-
generation information infrastructure," Opt. Fiber Commun. Conf.,
OFC ‘95, paper ThEl, pp. 241, San Diego, CA.
15
[9] J. E. Berthold, "Paving the way for multiwavelength networking
on a local-to-national scale," Opt. Fiber Commun. Conf., OFC ‘95,
paper ThE2, pp. 241-242, San Diego, CA.
[10] A. M. Hill, A. J. N. Houghton, "Optical networking in the European
ACTS Programme," Opt. Fiber Commun. Conf., OFC ‘96, paper
T h ll, pp. 238-239, San Jose, CA.
[11] B. Hui, "Progress in optical networking and technology in the
U.S.A.," Opt. Fiber Commun. Conf., OFC ‘ 96, paper ThI2, pp. 239-
240, San Jose, CA.
[12] A. A. M. Saleh, "Overview of MONET, multiwavelength optical
networking, program," Opt. Fiber Commun. Conf., OFC ‘96, paper
ThI2, pp. 240, San Jose, CA.
[13] S. B. Alexander, R. S. Bondurant, D. Byrne, V. W. S. Chan, S. G.
Finn, R. Gallager, B. S. Glance, H. A. Haus, P. Humblet, R. Jain, I.
P. Kaminow, M. J. Karol, R. S. Kennedy, A. Kirby, H. Q. Le, A. A.
M. Saleh, B. A. Schofield, J. H. Shapiro, N. K Shankaranarayanan,
R. E. Thomas, R. C. Williamson, and R. W. Wilson, "A
precompetetive consortium on wide-band all-optical networks,"
IEEE/O SA J. Lightwave Technol., vol. 11, no. 5/6, pp. 714-735,
May/June, 1993.
[14] C. A. Brackett, A. S. Acampora, J. Sweitzer, G. Tangonan, M. T.
Smith, W. Lennon, K C. Wang, and R. H. Hobbs, "A scalable
multiwavelength multihop optical network: A proposal for
research on all-optical networks," IEEE/O SA J. Lightwave
Technol., vol. 11, no. 5/6, pp. 736-753, May/June, 1993.
[15] P. E. Green, Jr., L. A. Coldren, "All-optical packet-switched
metropolitan-area network proposal," IEEE/O SA J. Lightwave
Technol., vol. 11, no. 5/6, pp. 754-763, May/June, 1993.
Chapter 2
All-Optical Wavelength Shifting
Using Semiconductor Optical Amplifier
__________ Cross-Gain Compression
Chapter Highlights
• Physical description of semiconductor optical amplifiers.
• Cross-gain compression for all-optical wavelength shifting.
• Measured semiconductor optical amplifier characteristics.
2.1 Introduction
This chapter provides an overview of semiconductor optical
amplifiers (SOAs), also called traveling wave amplifiers (TWAs).
Besides in-line amplification, SOAs have been utilized for preamplifiers
[1], high-speed optical gating switches [2], modulators [3]-[4], and even
multi-function detector/transmitters [5]-[6]. This chapter will focus on
pertinent characteristics which relate to using SOAs for cross-gain
compression wavelength shifting [7]-[19].
A semiconductor optical amplifier is essentially a diode laser with
very low reflective facets. Actually, the same device can operate as a
light-emitting diode (LED), SOA, or a laser, depending on the facet
17
reflectivity and internal gain. A typical SOA device is shown in Figure
2.1. An input optical signal coupled into the active region encounters a
single-pass gain.
Injection Current
Contact
Active Region
Waveguide
Contact
Figure 2.1 Typical SOA.
Many desirable features of SOAs include: (1) wide gain-
bandwidth, typically around 50 nm (6 THz)~they can be used for
multiwavelength (WDM) amplification; (2) compactness-they can be
monohthically integrated with other semiconductor devices such as
lasers and modulators [20]; (3) preservation of phase information-they
can be used in coherent optical communications.
There are a number of physical param eters which affect the
performance of an SOA. They include facet reflectivity, polarization
dependency, and gain saturation.
18
2.1.1 F acet R eflectiv ity
As previously mentioned, an SOA is essentially a laser with low
reflective facets. One can classify the semiconductor diode device based
on it's facet reflectivity and internal gain.
SOA Saturation
LASER
(Partially reflective facets)
0
Amplification Lasing Injection Current
Threshold Threshold
Figure 2.2. Behavior as a function of injection current. After [21].
Figure 2.2 shows the internal gain as a function of the injection current.
When current is applied the electrons and holes combine and
19
spontaneously emit photons, and the device behaves like an LED. When
the gain increases above zero, the device behaves like an optical
amplifier (SOA) and input light is amplified via stimulated emission. If
there is a feedback mechanism, such as partial reflecting facets, the
device begins to lase a t Fabry-Perot modes determined by the cavity
length and the active medium index of refraction. In this situation, the
device can be used as a Fabry-Perot amplifier, also known as a
semiconductor laser amplifier (SLA), in which amplification of an input
signal is restricted to the Fabry-Perot modes of the SLA. If the facets
have low reflectivities, the Fabry-Perot modes are suppressed and the
device continues to behave like an SOA; the gain eventually saturates.
Low reflectivities (< 10'4 ) are achieved by using very effective anti
reflection coatings such as SiO* [22] and Si3N4 [23]. Other techniques
using angled-stripe and buried-facet structures are effective in further
reducing reflections back into the active region.
In SOAs with buried facets [24]-[25], shown in Figure 2.3, there is
a small window region (-15 - 55 pm) separating the end of the active
region waveguide and the end-facet. This window allows light to diverge
at the waveguide output before reflecting off the end-facet. This
decreases the reflected power back into the active region.
C a p L ay er
o >
Si-lnP Si-ln P
Active R eg io n
InP S u b s tra te
W indow reg io n s
Figure 2.3. SOA with buried facets. After [24]
.J
T o p View
Figure 2.4. SOA with angled-stripe structure.
In SOAs with an angled stripe structure [26], shown in Figure 2.4,
the waveguide is actually tilted from being perpendicular to the end-
facets. As with the buried facet scheme, the reflections back into the
active region are minimized.
2.1.2 P olarization Independence
One major goal for fabricating SOAs is to make them polarization
independent to eliminate signal polarization dependencies of the optical
network. There are two main parameters th at affect the amount of
21
polarization dependence. First, a rectangular waveguide structure
creates a polarization-dependent gain because the confinement factor is
different for transverse electric (TE) and transverse magnetic (TM)
fields. Second, the modal refractive indices of the TE and TM fields
encounter different reflectivities off the end-facet resulting in different
reflectivity minimas [24]. Solutions to these problems include using
buried facets, as described in Section 2.1.1, tapered regions, or a more
square-shaped waveguide cross-section to equalize the confinement
factors [27]-[29].
2.1.3 Gain Saturation
It would be beneficial if SOAs could amplify any input optical
signal, regardless of the power. However, as the power is increased,
there is a point where the SOA gain saturates; as the power is increased
further, the gain begins to compress, or decrease. The point at which the
amplified output power has decreased by half (3 dB) compared to the
small-signal gain is called the 3-dB gain saturation power. Saturation
occurs when the rate of recombination within the active region increases
due to the increase in injected photons and a point is reached when the
amount of electron-hole pairs contributing the optical gain decreases.
22
An effect of the increase in photon density is band-filling which causes a
red-shift of the gain peak, th at it, the peak moves toward higher
wavelengths.
An SOA has the property of homogeneously broadened gain [30],
and the SOA can be modeled as a two-level system; the equations for
small-signal and large-signal gain can be derived using this model. The
modeled gain is [31]:
g(co)=----------------- (2.1)
1 H ta-O ofT i+ P JP "
where go is the gain peak parameter determined by the amplifier
pumping level, co is the optical frequency of the input signal, c o o is the
atomic transition frequency, Ti is the dipole relaxation time, Pin is the
input power, and P sat is the saturation power. For small signal inputs,
Pin « P sat and (2.1) simplifies to
g ( c o ) = ----------------— — -— — ( 2 .2 )
5 i+(ffl-fi>0)2r2 2
For large signal inputs where Pin - P sat and considering signals
which equal the gain peak co = cm (to simplify the discussion), (2.1)
becomes
g(Q})=---- ^ ---- (2.3).
1+P IP
1 ^ 1 m l * s a t
23
For an amplifier of length L, we find that the amplifier gain G is
f ^ i \
G = ^ - = exp(sL)=G0exp
G - 1 GP; .
G P satJ
(2.4)
which m ust be solved numerically. Go is the unsaturated amplifier gain.
Finally, the 3-dB gain saturation power is
(2.5).
G0 2
For a mathematical model which takes into account more accurately the
gain far from the peak, a cubic fit (matching experimental data) rather
that the Lorentzian of (2.1) can be used [15].
If the SOA input fight is intensity modulated, the gain becomes
time-dependent. Thus, the gain can be calculated for a time-dependent
input power Pin using (2.4). We see that as Pin increases, the gain begins
to decrease. Now if Pin is on-off modulated, the gain G(t) will respond
inversely,
G(t) = G 0exp
' G(f)-1 G(t)Pin (t)^
(2.6).
G(r) P s
This is shown in Figure 2.5. In the presented analysis, it is assumed
that the modulation speed is much less than the SOA gain-recovery time
T re e (carrier lifetime within the active region) which is the speed-limiting
24
factor. The gain-recovery occurs during the pump's high-to-low transition
in which the probe goes from low to high.
c
‘cd
0
<
O
0
Input Signal Power
4 * 4
S ’ 5b
• * > 4
Figure 2.5. Modulation of SOA gain due to intense input signal.
The gain-recovery time is given by
(2-7)
— + a S
*nr
where T m is the nonradiative recombination time, primarily due to Auger
recombination, a is the stimulated emission rate constant, and S is the
amplifier internal photon density.
There have been demonstrated attem pts to decrease this gain-
recovery time, for example, by using a "holding beam" (CW light) [32]
25
(increase S), or creating carrier storage regions near the active region for
faster carrier diffusion (decrease Tnr) [33].
2.2 All-Optica I Wavelength Shifting
There are generally three methods of all-optical wavelength
shifting which utilize SOAs. They are cross-gain compression, cross
phase modulation using a Mach-Zender interferometer, and four-wave
mixing. The work presented in this dissertation utilizes the cross-gain
compression method. Other all-optical methods including SOA and non-
SOA methods will be presented in Chapter 3.
2.2.1 A ll-O ptical W avelength Shifting U sing SOA Cross-Gain
C om pression
Apump Data
Optical
Bandpass
Filter
A,probe
SOA
Gain(t)
CW Aprobe
Figure 2.6. All-optical wavelength shifting using SOA cross-gain compression.
26
As discussed in the previous section, a strong input signal
inversely modulates the SOA gain. Because the gain is homogeneously
broadened, a strong optical "pump" (A ,pump) input compresses the entire
SOA gain, and any wavelengths within the SOA gain bandwidth will
also experience this compression. This is called cross-gain
com pression. It is also called cross-gain saturation or cross-gain
modulation. This is the basis of all-optical SOA cross-gain compression
wavelength shifting [7]-[18]. If a weak CW "probe" laser (A ,probe) is also
coupled into the SOA, it experiences the inversely modulated gain.
Hence, a complemented replica of the X pump signal is transferred onto
Aprobe- Also, because the gain is homogeneous, multicasting (in which
one signal is broadcasted to many output wavelengths) is possible by
using multiple probe lasers.
The maximum speed of the wavelength shifting process is limited
by the SOA gain-recovery time, as mentioned in the previous section. By
increasing the probe power (increasing S in Equation 2.1), data rates of
up to 20 Gb/s have been achieved [18]. Another method is to increase the
length of the SOA cavity. Recently, 40 Gb/s wavelength shifting has been
achieved using two cascaded SOAs [34].
27
It should be noted that contrary to the case where the SOA is used
to amplify signals, and thus requires a large P sat, in SOA cross-gain
compression wavelength shifting the amplifier should have a low P sa t so
th at a very strong pump signal is not required.
The combination of the SOA, the probe laser(s), coupling optics
and fiber-couplers will occasionally be referred to in this dissertation as
a "wavelength shifter."
2.2.2 W avelength R equirem ents
There has been a number of theoretical analysis of wavelength-
shifting using cross-gain compression [14]-[15]. To account for the gain-
peak shift upon compression, it is advantageous for the pump
c
a
(3
<
O
< />
Wavelength
Figure 2.7. Gain-compression due to strong pump.
28
wavelength to be located on the longer wavelength side of the SOA gain
peak while the probe(s) should be placed at the shorter wavelength side
of the gain peak. The result is a greater contrast ratio for wavelength
shifting to shorter wavelengths than longer wavelengths, as illustrated
in Figure 2.7.
However, there exists another factor th at indicates th at the probe
should not be placed too far on the shorter-wavelength side because the
spontaneous emission factor N s P, which contributes to noise, increases
towards shorter wavelengths [35].
2.2.3 Chirp
One disadvantage of cross-gain compression is th at chirp, an
alteration of a signal's spectral content, is introduced to the wavelength-
shifted signal. The chirp in an SOA (as well as in directly modulated
lasers) is due to the dynamic changes in carrier concentration which
affects the refractive index and, consequently, the phase of an intensity-
modulated pulse [4],[36]. A chirped wavelength-shifted signal could
suffer significant dispersion if propagated along tens of kilometers of
standard fiber, resulting in a severe system penalty.
29
2.2.4 M ultiple-Input W avelength S h iftin g
Although the homogeneous property of the SOA gain allows
multiple ou tp u t channels (multicasting) to be possible, as discussed in
Section 2.1.3, multiple in p u t channels (WDM channels) are not possible.
If more than one intense pump signal is coupled into the SOA, severe
cross-talk will occur between the two signals and both will contribute to
the gain modulation [37]. This limitation is addressed in Chapter 6,
where a method of using multiple SOAs to perform multiple-channel
wavelength shifting will be presented.
2.3 Measured SOA Characteristics
This section describes the SOAs used in the experimental
demonstrations presented in this dissertation. Bulk InGaAsP SOA
chips mounted on ceramic substrates which were then bonded onto
copper studs were provided to us courtesy of AT&T Bell Laboratories.
These studs were then placed in custom designed mounts which included
Peltier heat pumps for temperature stabilization. Commercially
available pigtailed coupling optics (E-Tek LOFI/LIFIs) were used to
couple light in and out of the SOA. Coupling losses were measured to be
30
~4 dB by comparing the detected light a t the fiber output with th at using
a wide-area detector directly from the SOA facet.
A typical SOA spectrum is shown in Figure 2.8. The maximum
amplified spontaneous emission (ASE) ripple was < 1 dB (@ 120 mA).
Reflectivity R was estimated by measuring the maxima and minima of
the ASE ripple [22] and was found to be -10"*. The difference between
TE and TM gain was approximately < 2 dB.
P M !BtB9:52 SEP 24, 1393_____________________________
LED Test
■ean (Fillffl) = 15EG.4B n. FM H It = 3H.75 n.
aean (3dB) = 1565.25 na 3 dB width = 33.25 ne
peak xavaln = 1567.25 ne to ta l poxer = -IB .19 dBe
aigaa________ = 1 4 .7B ne pk dene (lnal= -2 5 .B2 dBe
RL -23.14 dBa H K R t l U U L 1565.2 na
SENS
H Q 0
- 4 9
1 «4D /(
dBa
Til _i
■23.0! dB a
4
f VOrl ------ j
CENT
i c r r
:r m
a n_
ELENjj v t
1 3 Da >v n i
-
I'LNI E H l 5 6 6 . B n a SPR N 2 B 0 .0 n a
RB 2 na U B IBB kHz ST 98 aoec
Figure 2.8. Measured SOA spectrum. Bias current = 80 mA, T = 22.5° C.
31
2 0 - i
m
u. 5 -
6
Bias Current
o ■
o m ..
o
| 0 -
■ 140 mA
A 120 mA
O 8 0 mA
o
5 ii 1111 i ii 1 1 ii i 111 i 'H i '■ 111 ii i 1111 i i |
-30 -25 -20 -15 -10 -5 0 5
Fiber Input (dBm)
Figure 2.9. Experimentally measured gain vs. fiber input power at 1571 nm for
different SOA bias currents.
Figure 2.9 shows the fiber-to-fiber gain for different drive currents,
as the optical input power into the input fiber is increased. The 3-dB
saturation power was +1 dBm, measured at the fiber output.
2.4 Chapter Summary
In this chapter, SOAs were discussed. Physical attributes such as
facet reflectivity, polarization dependency, and most importantly, gain
saturation were discussed. Gain saturation is the mechanism which
enables all-optical wavelength shifting using SOA cross-gain
compression. Some measured characteristics of a typical SOA used in
the experiments contained in this dissertation were presented.
32
2.5 References
[1] H. Nakano, S. Tsuji, S. Sasaki, K. Uomi, and K Yamashita, "10-
Gb/s 4-channel wavelength division multiplexing fiber
t r a n s m i s s i o n using semiconductor optical amplifier modules,"
IEEE/O SA J. Lightwave Technol., vol. 11, no. 4, pp. 612-618, April
1993.
[2] J. Spring and R. S. Tucker, "Photonic 2x2 packet switch with input
buffers," Electron. Lett., vol. 29, no. 3, pp. 284-285, Feb. 1993.
[3] J. Riishoj, T. N. Nielsen, U. Gliese, K. E. Stubkjaer, "3-Gbit/s 2-
level to 1.5 Gsymbol/s 4-level converter GaAs IC for 2 electrode
semiconductor optical amplifier modulators," Electron. Lett., vol.
29, no. 14, pp. 1173-1174, July 1993.
[4] L. Gillner, "Modulation properties of a near travelling-wave
semiconductor laser amplifier," IEE Proc. Pt. J, vol. 139, no. 5, pp.
331-338, Oct. 1992.
[5] K. T. Koai, R. Olshansky, and P. M. Hill, "Dual-function
semiconductor laser amplifier in a broad-band subcarrier
multiplexed system," IEEE Photon. Technol. Lett., vol. 2, pp. 926-
928, 1990.
[6] K. T. Koai and R. Olshansky, "Simultaneous optical amplification,
detection, and transmission using in-line semiconductor laser
amplifiers," IEEE Photon. Technol. Lett., vol. 4, no. 5, pp. 441-443,
May 1992.
[7] T. Durhuus, C. Joergensen, etal., "Optical wavelength conversion
and gating a t 4 Gbit/s using semiconductor optical amplifiers,"
OFC/IOOC '93 TuH6.
[8] B. Glance, J. M. Wiesenfeld, U. Koren, A. H. Gnauck, H. M. Presby,
and A. Jourdan, "High performance optical wavelength shifter,"
Electron. Lett., vol. 28, no. 18, pp. 1714-1715, Aug 1992.
33
[9] C. Joergensen, T. Durhuus, et al., "4 Gb/s Optical wavelength
conversion using semiconductor optical amplifiers, " IEEE Photon.
Technol. Lett., vol. 5, no. 6, pp. 657-660, June 1993.
[10] C. Joergensen, T. Durhuus, et al., "Optically controlled
semiconductor optical amplifier gates," ECOC 1992, WeB8.2, pp.
377-380.
[11] T. Mukai, K. Inoue, and T. Saitoh, "Signal gain saturation in two-
channel common amplification using a 1.5mm InGaAsP traveling-
wave laser amplifier," Electron. Lett., vol. 29, no. 8, pp. 396-397,
April 1987.
[12] J. S. Perino, J. M. Wiesenfeld, and B. Glance, "Fiber transmission
of lOGbit/s signals following wavelength conversion using a
traveling-wave semiconductor optical amplifier," Electron. Lett.,
vol. 29, no. 25, pp. 256-258.
[13] R. Schnabel, W. Peiper, et al., "Wavelength conversion and
switching of high speed data signals using semiconductor laser
amplifiers," Electron. Lett., vol. 29, no. 23, pp. 2047-2048.
[14] I. Valiente, J. C. Simon, and M. Le Ligne, "Theoretical analysis of
semiconductor optical amplifier wavelength shifter," Electron.
Lett., vol. 29, no. 5, pp. 502-503, March 1993.
[15] A. E. Willner and W. Shieh, “Optimal spectral and power
param eters for all-optical wavelength shifting: single stage,
fanout, and cascadability,” IEEE /O SA J. Lightwave Technol.,
Special issue on Optical Amplifiers, vol. 13, no. 771-781, May
1995.
[16] J. M. Wiesenfeld and B. Glance, "Cascadability and fanout of
semiconductor optical amplifier wavelength shifter," IEEE
Photon. Technol. Lett., vol. 4, no. 10, pp. 1168-1171, Oct. 1992.
34
[17] J. M. Wiesenfeld and B. Glance, J. S. Perino, and A. H. Gnauck,
"Wavelength conversion a t 10 Gb/s using a semiconductor optical
amplifier," IEEE Photon. Technol. Lett., vol. 5, no. 11, pp. 1300-
1303, Nov. 1993.
[18] J. M. Wiesenfeld, J. S. Perino, A. H. Gnauck, and B. Glance, "Bit
Error Rate Performance For Wavelength Conversion At 20
Gbit/s," Electron. Lett., vol. 30, no. 9, pp. 720-721, April 1994.
[19] S. Dubovitsky, P. D. Dapkus, A. Mathur, W. H. Steier,
"Wavelength conversion in a quantum-well polarization-
insensitive amplifier," IEEE Photon. Technol. Lett., vol. 6, no. 7,
pp. 804-807, July 1994.
[20] M-A. Newkirk, B. I. Miller, U. Koren, M. G. Young, M. Chien, R M.
Jopson, G. Raybon, C. A. Burrus, and H. M. Presby, "1.55-pm
multiquantum-well semiconductor optical amplifier with low gain
ripple and high coupling efficiency for photonic circuit integration,"
Electron. Lett., vol. 29, no. 5, pp. 443-444, Mar. 1993.
[21] S. Kobayashi and T. Kimura, "Semiconductor optical amplifers,"
IEEE Spectrum, pp. 26-31, May 1984.
[22] M. J. O'Mahony, "Semiconductor laser optical amplifiers for use in
future fiber systems," IEEE/O SA J. Lightwave Technol., vol. 6, no.
4, pp. 531-544, April 1988.
[23] J. C. Simon, "GalnAsP semiconductor laser amplifiers for single
mode fiber communications," IEEE/O SA J. Lightwave Technol.,
vol. LT-5, no. 9, pp. 1286-1295, Sept. 1987.
[24] N. A. Olsson, R. F. Kazarinov, W. A. Nordland, C. H. Henry, M. G.
Oberg, H. G. White, P. A Garbinski, and A. Savage, "Polarisation-
independent optical amplifier with buried facets," Electron. Lett.,
vol. 25, no. 16, pp. 1048-1049, Aug. 1989.
35
[25] I. Cha, M. Kitamura, H. Honmou, and I. Mito, "1.5 pm band
travelling-wave semiconductor optical amplifer with window facet
structure," Electron. Lett., vol. 25, no. 18, pp. 1241-1242, Aug.
1989.
[26] C. E. Zah, J. S. Osinski, C. Caneau, S. G. Menocal, L. A. Reith, J.
Salzman, F. K. Shokoohi, and T. P. Lee, Electron. Lett., vol. 23, pp.
990, 1987.
[27] L. Eskildsen, B. Mikkelsen, T. Durhuus, C. G. Joergensen, K E.
Stubkjaer, P. Doussiere, P. Garabedian, F. Leblond, J.-L.
Lafragette, and B. Femier, "Polarization-insensitive
semiconductor optical preamplifier at 1.55 pm," Electron. Lett.,
vol. 28, no. 21, pp. 2019-2021, Oct. 1992.
[28] S. Kitamura, K. Komatsu, and M. Kitamura, "Polarization-
insensitive semiconductor optical amplifier array grown by
selective MOVPE," IEEE Photon. Technol. Lett., vol. 6, no. 2, pp.
173-175, Feb. 1994.
[29] P. Doussiere, P. Garabedian, C. Graver, D. Bonnevie, T. Fillion, E.
Derouin, M. Monnot, J.G. Provost, D. Leclerc, and M. Klenk, "1.55-
pm Polarization-independent semiconductor optical amplifier
with 25-dB fiber to fiber gain," IEEE Photon. Technol. Lett., vol. 6,
no. 2, pp. 170-172, Feb. 94.
[30] T. Mukai, K. Inoue, and T. Saitoh, "Homogeneous gain saturation
in 1.5 pm InGaAsP traveling-wave semiconductor laser
amplifiers," Appl. Phys. Lett., vol. 51, no. 6, pp. 381-383, Aug.
1987.
[31] G. P. Agrawal, Fiber-Optic Communication Systems. New York:
John Wiley & Sons, Inc., 1992.
[32] R. J. Manning, D. A. O. Davies, J. K. Lucek, "Recovery rates in
semiconductor-laser amplifiers - optical and electrical bias
dependencies," Electron. Lett., vol. 30, no. 15, pp. 1233-1235, July
1994.
36
[33] G. Eisenstein, R. S. Tucker, J. M. Wiesenfeld, P. B. Hansen, G.
Raybon, B. O. Johnson, T. J. Bridges, F. G. Storz, and C. A. Burrus,
"Gain recovery time of traveling-wave semiconductor optical
amplifiers," Appl. Phys. Lett., vol. 54, no. 5, pp. 454-456, Jan.
1989.
[34] S. L. Danielsen, C. Joergensen, M. Vaa, and K. E. Stubkjaer, "Bit
error rate assessm ent of a 40 Gb/s all-optical polarisation
independent wavelength convertor," Opt. Fiber Commun. Conf.,
OFC ‘96, San Jose, CA, postdeadline paper PD12.
[35] M. G. Oberg and N. A. Olsson, "Wavelength dependence of noise
figure of a travelling-wave GalnAsP/InP laser amplifier," Electron.
Lett., vol. 24, no. 2, pp. 99-100, Jan. 1988.
[36] J. S. Perino and J. M. Wiesenfeld, "Linewidth enhancement factor
and chirp for high bit rate semiconductor optical amplifier
wavelength converter," Conf. on Lasers and Electro-Optics (CLEO),
1994, paper CThEl, pp. 298-299.
[37] K. Inoue, "Crosstalk and its power penalty in multichannel
transm ission due to gain saturation in a semiconductor laser
amplifier," IEEE/O SA J. Lightwave Technol., vol. 7, no. 7, pp.
1118-1124, July 1989.
37
Chapter 3
Other Wavelength-Shifting Methods
Chapter Highlights
• Overview of all-optical wavelength-shifting methods other than
SOA cross-gain compression
3. 7 Cross-Phase Modulation
Cross-phase modulation has origins identical to cross-gain
compression. Recall from Section 2.1.3 th at a strong pump signal
induces a refractive index change within the gain region of an SOA. This
results in a phase modulation of an introduced CW probe, similar to
cross-gain compression. By using a Mach-Zender (Figure 3.1a) or a
Michelson interferometer (Figure 3.1b) arrangement, an input pump
signal can be all-optically wavelength shifted to a probe wavelength via
constructive and destructive interference. The advantages of this
method are chirp elimination, contrast ratio enhancement, and no data
inversion. Recently, a cascade of 10 monolithically integrated Michelson
interferometer wavelength shifters (in a re-circulating loop) was
demonstrated, operating at 10 Gb/s and propagating a total of 500 km
(a)
CW Xprobe
Wavelength-shifted
Output, Aprobe
SOAs
ASK, A,pump
(b)
CW Xprobe — ►
*
ASK, Apump
Wavelength-shifted
Output, Aprobe
SOAs
Figure 3.1. Interferometric wavelength shifters: (a) Mach-Zender, (b) Michelson.
3.2 Four-Wave Mixing
This section discusses methods of all-optical wavelength shifting
using four-wave mixing. In four-wave mixing, optical fields from two
beams coherently interact within a non-linear medium to essentially
create a periodic index grating off of which scattering will occur. The
result is newly generated sideband wavelengths, as shown in Figure 3.2.
This phenomena has been demonstrated in SOAs [2]-[5], optical fibers
[6]-[7], and passive waveguides.
39
M(t) X2
Down-Shifted Pump Probe Up-Shifted
Signal
X3(0 = 2Xl(t)~
\ “ t /
- X 2 \ *Xd (t) =
1 1
Signal
\4(t) = 2X2-XK0
Wavelength
ri = 3 & - 2 I + R ( & X ) [dB]
Figure 3.2. Four-wave mixing spectrum.
In Figure 3.2, rj is the conversion efficiency defined as the ratio (in dB) of
the shifted signal power to the input probe power, G is the saturated
optical gain (in dB), IP is the input optical pump (in dB), and R(A X ) is the
relative efficiency which is function of physical parameters.
Unlike SOA cross-gain compression wavelength shifting, four-
wave mixing allows data-format transparency and a greater bandwidth.
Data format is preserved because the shifted signal is a phase-conjugate
version of the original. A wide modulation bandwidth is due to the fact
that unlike SOA cross-gain compression which is limited by the gain-
recovery (carrier lifetime), four-wave mixing is limited by intraband
relaxation times (< 1 ps). The major disadvantage is th a t the conversion
process is not efficient due to the reliance on intraband dynamics for a
weak non-linear gain effect. The efficiency depends on carrier density
40
modulation, dynamic carrier heating, and spectral hole burning. Also,
the efficiency drops as the wavelength spacing between the two input
beams increases. Another disadvantage of such a method is the
sensitivity to polarization.
In fiber four-wave mixing, phase matching conditions m ust be
met. This is possible by operating one wavelength at the zero-dispersion
point in silica fiber (1.3 pm) and the second wavelength slightly offset. A
conversion efficiency of -24 dB over 7.6 nm has been demonstrated [6].
3.3 Laser Carrier Depletion
The carrier-depletion method relies on the optical depletion of the
carrier density in the gain region of a distributed Bragg reflector (DBR)
laser (shown in Figure 3.3) [8] or a distributed feedback (DFB) [9]-[10]
laser which is biased above threshold.
Gain Phase Bragg
—> • U T J I T
laser
Figure 3.3. Carrier depletion in a DBR laser. After [8].
41
The speed limitation is the relaxation frequency of the laser. The laser,
emitting on A ja se r is biased above threshold while a intensity modulated
signal on A P u m P is coupled into the laser. The pump signal induces
stim ulated recombinations which deplete the carriers thus perturbing
the gain and refractive index of the gain medium. The result is an
amplitude (inverted) and frequency modulation (inverted and non
inverted) of the input laser at A ja se r.
3.4 Non-linear Optical Loop Mirror
The last method discussed in this chapter is all-optical
wavelength shifting using a non-linear optical loop mirror (NOLM) [11]-
[12]. The NOLM utilizes a fiber Sagnac interferometer and a non-linear
element (SOA or dispersion shifted fiber) within a closed loop. Shifting
of data from 1.5 pm to 1.3 pm has been demonstrated using walk-off
compensation techniques [11].
42
3.5 References
[1] B. Mikkelsen, R. J. S. Pedersen, M. Nissov, H. N. Poulsen, C.
Joergensen, S. L. Danielsen, K E. Stubkjaer, M. Gustavsson, W.
van Berio, and M. Janson, "Transmission through 10-all-optical
interferometric wavelength converter spans at 10 Gb/s," Opt. Fiber
Commun. Conf., OFC ‘ 96, San Jose, CA, postdeadline paper PD13.
[2] G. Grosskopf, R. Ludwig, and H. G. Weber, "140 Mbit/sDPSK
transmission using an all-optical frequency convertor with a 4000
GHz conversion range," Electron. Lett., vol. 24, no. 17, pp. 1106-
1107, Aug 1988.
[3] J. Zhou, N. Park, K. J. Vahala, M. A. Newkirk, and B. I. Miller,
"Broad-band wavelength conversion with amplification by 4-wave-
mixing in semiconductor traveling-wave amplifiers," Electron.
Lett., vol. 30, no. 11, pp. 859-860, May 1994.
[4] J. Zhou, N. Park, J. W. Dawson, K. J. Vahala, M. A. Newkirk, and
B. I. Miller, "Efficiency of broad-band four-wave mixing
wavelength conversion using semiconductor traveling-wave
amplifiers," IEEE Photon. Technol. Lett., vol. 6, no. 1, pp. 50-52,
Jan 1994.
[5] P. P. Iannone, P. R. Prucnal, G. Raybon, U. Koren, and C. A.
Burrus, "Broadband wavelength shifter for picosecond optical
pulses a t 1.5mm," Electron. Lett., vol. 29, no. 17, pp. 1518-1519,.
[6] K Inoue and H. Toba, "Wavelength conversion experiment using
fiber four-wave mixing," IEEE Photon. Technol. Lett., vol. 4, no. 1,
pp. 69-72, Jan. 1992.
[7] T. Morioka, S. Kawanishi, and M. Saruwatari, "Tunable error-free
optical frequency conversion of a 4ps optical short pulse over
25nm by four-wave mixing in a polarisation-maintaining optical
fiber," Electron. Lett., vol. 30, no. 11, pp. 884-885, May 1994.
43
[8] T. Durhuus, R. J. S. Pedersen, et al., "Optical wavelength
conversion over 18nm at 2.5 Gh/s by DBR-laser," IEEE Photon.
Technol. Lett., vol. 5, no. 1, pp. 86-88, Jan. 1993.
[9] K. Inoue, M. Yoshino, and F. Kano, "Polarisation insensitive
wavelength conversion using a light injected DFB-LD with a loop
configuration," Electron. Lett., vol. 30, no. 5, pp. 438-439, Mar.
1994.
[10] S. B. Mikkelsen, R. J. S. Pedersen, T. Durhuus, C. Braagaard, C.
Joergensen and K. E. Stubkjaer, "Wavelength conversion of high
speed data signals," Electron. Lett., vol. 29, no. 19, pp. 1716-1718,
Sept. 1993.
[11] K.A. Rauschenbach, K. L. Hall, J. C. Livas, and G. Raybon, "All-
optical pulse width and wavelength conversion at 10 Gb/s using a
nonlinear optical loop mirror," IEEE Photon. Technol. Lett., vol. 6,
no. 9, pp. 1130-1132, Sept. 1995.
[12] D. Mahgerefteh and M. W. Chbat, "All-optical 1.5 pm to 1.3 pm
wavelength conversion in a walk-off compensating nonlinear
optical loop mirror," IEEE Photon. Technol. Lett., vol. 7, no. 5, pp.
497-499, May 1995.
44
Chapter 4
Simultaneous Wavelength Shifting
_________ and Header Replacem ent
Chapter Highlights
• Novel experimental demonstration of sim ultaneous all-optical
wavelength-shifting and on-the-fly header replacement.
• This function may be necessary in future dynamically
reconfigurable WDM packet-switching architectures.
4.7 Motivation
The combination of wavelength shifting and header replacement
will potentially be im portant in future high-capacity, packet-switched,
wavelength-division-multiplexed (WDM) networks [1]. As stated in
Chapter 1, wavelength shifting is important because it allows for (i) the
system-wide re-use of wavelengths in a WDM network with more users
than available wavelengths and (ii) the reconfiguration of wavelength-
paths in a dynamic network when changing traffic patterns require data-
packet rerouting at a switching node [2]-[3]. Header replacement [4]-[6]
is also im portant because it facilitates the proper routing of d ata
packets in a dynamic network to their final destinations; the packet
45
header, which contains destination, routing, and priority designation,
may require replacement with a new updated header. The need for both
wavelength shifting a n d header replacement may arise at WDM
switching nodes where it is necessary to reroute a packet onto a new
wavelength-path and, consequently, update the packet’s header. Figure
4.1 illustrates an incoming packet on wavelength Apump being shifted to
^probe and its header being replaced simultaneously.
Incoming Packet on X .p u m p
Flat
Outgoing Packet on X prob«
Sgl "blew" Headerfp Payload i
-.........
Wavelength
Shifter
"New" Header Generator
ew Header] on A .p r0be
Figure 4.1. The presented method replaces the old header of an incoming packet
with a new header and simultaneously shifts the entire packet onto a new
wavelength. The complement bars indicate wavelength-shifted data.
A straightforward method to implement these functions is to
detect the packet, electronically change the header, and then retransm it
the entire packet onto the new wavelength. However, this method
creates (i) a speed bottleneck because the entire packet would undergo
optical-to-electrical conversion, electrical processing, and electrical-to-
optical conversion, and (ii) moderate power consumption because the
46
entire packet would have to be retransm itted. To m aintain optical
transparency and avoid opto-electronic bottlenecks, both header
replacement and wavelength shifting would ideally be implemented all-
optically.
In this chapter, an experimental demonstration of all-optical
sim u ltan eou s header replacement and wavelength shifting is presented
[7]. Both functions are implemented by using cross-gain compression in
an SOA and by using th ree-level m odulation of the probe laser only
when the header bits require changing. "On-the-fly" header replacement
of a 1 Gb/s packet and simultaneous shifting of the packet's wavelength
by 19 nm is presented. A sensitivity of -27 dBm at 10'9 BER is
measured using a 416-bit non-retum-to-zero (NRZ) modulated packet.
The presented approach has the following advantages: (i) the process is
all-optical; (ii) both header and payload are at the same bit rate and
wavelength; (iii) no guard bits are required before or after the header; (iv)
the process requires low modulation-induced power consumption.
4 .2 Three-Level Probe Modulation
As explained in Chapter 2, in wavelength shifting using SOA
cross-gain compression, the probe laser is operated under CW
47
conditions. To add the function of header replacement, the probe is
simply modulated at the appropriate time, with the appropriate
modulation. The result is the wavelength-shifted bits being altered as it
appears at the SOA output. However, the type of probe modulation is
not an NRZ signal but rather a three-level signal. The probe is
modulated only when the header bits are to be changed, effectively
replacing the header bit-by-bit and on-the-fly. A three-level signal is
required to achieve bit-change or no-bit-change cases.
The method is described as follows. When the entire header or an
individual header bit requires no changes, wavelength shifting is
performed by setting the probe input power to a "middle" level. When an
individual header bit requires changing, the probe input power is either
decreased to a "low" level, forcing a low probe output power, or increased
to a "high" level, forcing a high probe output power. Because wavelength
shifting using SOA cross-gain compression inverts the input bits, the
following notations will be adopted to avoid confusion: (1) wavelength-
shifted bits will be indicated by complement bars; (2) at the SOA input,
a pump signal bit will be indicated as either a "0"-bit corresponding to a
low optical power level or a "l"-bit corresponding to a high optical power
level; (3) at the SOA output, a probe wavelength-shifted bit will be
48
indicated as either a "0"-bit corresponding to a high optical power level
or a "T"-bit corresponding to a low optical power level. The method is
summarized in the "truth table" of Figure 4.2a, illustrated in Figure
4.2b, and restated as follows:
(a) (b)
Old Header on Xpum p
“TRUTH TABLE"
(ii) 0
m 1
(iv) 1
r w °r° ° n ° i ' 'i n * » « <
— 'T'-bit
bit
Modulation on kprob•
r
( O
I
New Header on Xprobe
0 |T |0 | 1 1 10 Q j Payload
♦ ♦ ♦ ♦ ♦ “
Altered Bits
*-t
— “O'-bit
1—‘ T'-bit
Figure 4.2. (a) The "truth table" shows which probe level is required for
obtaining the desired bit change, (b) Illustration of a three-level probe signal
required to simultaneously perform bit-by-bit header replacement, (0 10 010 1
1) -> (0 0 1 0 1 1 0 0), and wavelength.
(i ) No b it change. "0"-bit =± "0"-bit: The probe input power is set a t
the m iddle level. This is the conventional cross-gain compression
wavelength-shifting case. The probe middle level is also
maintained outside the header time-frame, allowing the payload to
be wavelength-shifted.
49
(ii) Change "0"-bit ^ Decreasing the probe input power to the
low level forces the wavelength-shifted bit to become a "I"-bit,
regardless of what the pump input bit is.
(iii) C hange "lw -bit "O '-b it: Increasing the probe input power to
the h igh level forces the wavelength-shifted bit to become a "0"-bit.
However, because of excessive gain compression, the probe power
level at the SOA output will depend on the combination of the pump
"l"-bit power and the probe high-level power.
(iv) No b it changft. "V'-bit -> ”1'-bit: The probe is set at the m iddle
level, as in case (i).
Note th at since the probe is modulated only when header bits require
changing, the modulation-induced power consumption is low.
4.3 Experimental Design
Figure 4.3 shows the experimental setup. A pattern generator
was programmed with a 1-Gb/s 416-bit NRZ packet (closely matching an
ATM packet length of 53 bytes), which included an 8-bit flag, an 8-bit
header, and payload bits. Further details of the packet structure can be
found in Appendix A. This pattern directly modulated a 1571-nm DFB
pump laser. This optical pump signal was first tapped and detected by
a 2.4 Gb/s receiver (BT&D PDC2201 GaAs PINFET) and then optically
delayed to allow for electronic processing.
D ateP ath— Optical Delay
.Q o c i
1 Gb/s
Pattern
Generator
Pump
Laser
E C L
Circuit ^
New Header
Generator
Electronic Processor
Isolator
Ftcvr -|
Error
Detector
Probe
Laser
PC: Polarization Controller
X: Power Combiner
Optical Path
Electrical Path
Optical
Filter
Probe
«— 19 n m - >
Hump
.......... __
r ’ "” 1552
_ _ 1 _ _ 1 _ _
.1 n m 1571.1 n m
■ » -----« _ _ 1 _ _ l— l___
5 nm /div
Figure 4.3. Experimental setup and pre-filtered optical spectrum.
4.3.1 E lectron ic P rocessor
Packat w/ Old Haadar
Thraa-Laval Signal
Gating Signal
Figure 4.4. Logic diagram using a 4-bit header example to illustrate operation
of the three-level circuit. "X" = RF power combiner. Gating signal is derived
from the master clock.
51
To create a three-level signal, an electronic processor, (shown in
Figure 4.3), which consisted of an emitter-coupled-logic circuit (>1 GHz
Sony SPECL integrated circuit chips), a power combiner, and a new-
header generator was used. This processor compared bit-by-bit the old
header, which was extracted from the tapped signal, with the new header
to create a three-level electrical signal. Details of the logic circuit is
shown in Figure 4.4. Using an inverted new-header allowed for a simpler
circuit design. (For details on how the new header was generated, see
Appendix A). Manual adjustm ent of the middle level is performed by
using an variable RF attenuator prior to the power combiner. The clock
from the pattern generator was used to gate the ECL circuit on only
during the header time period.
The output of the processor directly modulated a 1552-nm DFB
probe laser. The optical probe signal was combined with the delayed
pump signal and both were then coupled into the SOA. The SOA
characteristics were as follows: 1565-nm gain peak, 140-mA drive
current, 15-dB fiber-to-fiber gain, and a 3-dB gain compression at +1
dBm (1571 nm, measured a t the fiber output). The 1552-nm probe was
optically bandpass-filtered (using a 1-nm FWHM filter) and detected by
a second 2.4-Gb/s receiver (BT&D PDC2201 GaAs PINFET). Optimal
52
conditions for m a x i m u m receiver sensitivity were experimentally found
by adjusting the average pump and probe powers, the variable RF
attenuator (for probe middle-level control), and the SOA current.
4.4 Results and Discussion
Results of the experimental demonstration are shown in Figure
4.5 through Figure 4.10. Figure 4.5 shows oscilloscope traces of the
incoming packet, the old 8-bit packet header ( 0 1 0 0 1 0 1 1 ) , the three-
level probe signal, and the wavelength-shifted signal with the new
header ( 0 0 1 0 1 100). Five packet-header bits were changed all-
optically (altered bits are indicated in Figure 4.5d), the payload bits
remained unchanged, and the entire packet was wavelength shifted by
19 nm from Apump to Aprobe- No guard bits were required either before
or after the packet header, and the packet header was at the same bit-
rate and wavelength as the payload. The pump and probe powers were
as follows: P PumP = -14 dBm, P Pmbe = -14 dBm (average powers at the SOA
input facet).
The overshoot of header " 1" bits, as seen in Figure 4.5d and
Figure 4.5e, was caused by the probe low level corresponding to no light.
The overshoot is expected to decrease and the overall contrast ratio to
53
Incoming Packet, 1571 nm
(a)
(b)
(c)
(d)
(e)
f.
ir( a ««
I' f t I i w llh 'ii 'i w w V w
• • •
Flag P a y lo a d
N o l g h l l e v e l ♦
Old Header:
8 n s
r-bt
O'-bit
01001011
— high
middle
— "0 -b it
— "V-bit
Overshoot
N o W t l e v e l " " * ■
N o - S g h l l e w e l ~~\
NewHeader: .° 9 1 0 1 1 2 P i
i i -----i i - i j — Altered Bits
i s • O v e r s h o o t P a y lo a d
■ o
c
3
•o
ii
cn
mmk
3
3
■ o
o
O ’
< D
I I
o n
in
io
3
3
Outgoing Packet, 1552 nm
Figure 4.5. Scope traces (different vertical scales) of (a) Xpump incom ing
packet, (b) time-magnification of old header and portion of payload, (c) three-
level probe signal, (d) wavelength-shifted new header and portion of payload, and
(e) wavelength-shifted outgoing packet.
54
increase if full independent control of the probe's three levels was
possible. The lack of this independent control is mainly due to the rise
time variations between the two logic signals prior to the power
combiner as evidenced in the three-level eye-diagrams of Figure 4.6.
Level non-uniformity is seen in the probe optical signal Figure 4.6b.
Also seen in Figure 4.6 is ringing following the header time-frame.
8 ns ■
(a)
Figure 4.6. Three-level eye-diagrams of (a) RF power combiner output and (b)
probe laser prior to the SOA input.
Header (8 ns)
Ringing
Unaffected Bit
Figure 4.7. Eye-diagram of header time-frame.
55
Figure 4.7 shows the eye-diagram of the wavelength-shifted
header time-frame. Intersymbol interference is attributed to the level
non-uniformity of the three-level probe signal. Also shown is the
unaffected wavelength-shifted bit prior to the header as well as the bit
immediately following the header. This following bit is affected by the
ringing caused by the three-level circuit as previously mentioned.
(a)
Figure 4.8. Wavelength-shifted signal (a) without header replacement and (b)
with header replacement.
Figure 4.8 shows what occurs when the gating signal of the ECL
circuit is turned off, th a t is, the header is not changed. The limiting
factor in the demonstrated method is the "1" to "0" bit-change, as
mentioned in Section 2.2.4, because if the pump and probe powers were
too large, the wavelength-shifted high level would decrease, creating an
ambiguous logic level. Consequently, the resulting pump and probe
56
powers required for adding the function of header replacement (non-
optimal) is different for the case when the pump and probe powers are
optimized for only the function of wavelength shifting (optimal). To
investigate the difference between the optimal and non-optimal cases,
the pump and probe powers were adjusted to provide the best
performance of wavelength shifting using the existing experimental set
up. These optimal powers were P ^ p~ -5 dBm and P ° pp r' o b e ~ -5 dBm,
measured a t the SOA input facet.
Figure 4.9. Eye-diagrams of wavelength-shifted signals using (a) optimal and (b)
non-optimal pump/probe powers. Same vertical scales.
Figure 4.9 shows the eye-diagrams (using a DC-coupled New
Focus NF1611 1-GHz InGaAs PIN receiver) of a wavelength shifted
signal using (a) optimal and (b) non-optimal pump and probe powers.
1 ns
(a) (b)
57
Clearly, the contrast ratio for the optimal case is superior. Hence, a
trade-off exists when adding the function of header replacement to
wavelength shifting; we sacrifice in the degradation of the contrast ratio.
4.4.1 S en sitiv ities
Figure 4.10 shows receiver sensitivity curves for (a) simultaneous
header replacement and wavelength shifted data in which a -27 dBm
B
(0
cc
o
I-
IU
4 -«
■ 4
■ 6
■ 8
-3.5 dB
11
-28 -26 -34 -32 -30
Received Power (dBm)
Figure 4.10. Measured sensitivity curves for, from right to left, (a)
simultaneous header replacement and wavelength shifted data, (b) wavelength
shifted data without header replacement (same conditions as previous case) (c)
optimized wavelength shifted data, and (d) baseline pump data.
58
average received probe power was measured for a 1 0 9 BER; (b)
wavelength-shifted data without header replacement (same conditions
as the previous case); (c) optimal wavelength shifting, and (d) baseline
pump data (no header replacement, wavelength shifting, or optical
amplification). For the simultaneous header-replacement and
wavelength-shifting case, a power penalty of -3.5 dB was incurred. In
Figure 4.5d, we see th at the contrast ratio has decreased by
approximately the same amount which indicates that the incurred power
penalty is mainly due to contrast ratio degradation. As previously
mentioned, this power penalty was partially due to the fact th at the "1"
— » "0" bit-change case prevented the application of an intense pump in
order to achieve a high contrast ratio [8]. If either the pump "l"-bit or
the probe high level was increased, the excessive gain compression would
consequently decrease the wavelength-shifted "0"-bit level, resulting in
an ambiguous logic state. As a comparison, a ~1.75-dB power penalty
was incurred using optimal pump and probe powers. Therefore, the
additional ~1.75-dB power penalty (for a total of -3.5 dB) was due to the
weak pump requirement necessary for the added header replacement
function. This corresponds to the contrast ratio degradation shown in
Figure 4.9. Finally, given the pump and probe powers used to achieve
59
simultaneous header replacement and wavelength shifting, the process
of c h a n g in g the header using the three-level probe modulation resulted
in only <0.5 dB additional power penalty.
4.5 Issues an d Hurdles
The intention of the presented experiment was to demonstrate the
functions of simultaneous header replacement and wavelength shifting
using SOA cross-gain compression. All synchronization was performed
manually using RF and optical delays. If the method described in this
chapter was to be implemented in a network, issues of data-inversion,
synchronization, and pump/probe power management would arise. As
mentioned in Chapter 2, th e issue of data inversion is always an issue in
SOA cross-gain compression wavelength shifting. One method of
addressing this issue may be to incorporate a "polarity code" within the
header that informs the receiver as to the polarity of the signal. This
method would however increase the overhead length and processing time
of the packet. There h a s been a demonstration of non-inverting
wavelength shifting using two SOAs in which the method also enhanced
the contrast ratio [9].
60
In th e presented demonstration, the only clock-sensitive aspect
was that of the gating signal. Using a slightly more complex electronic
logic as well as a clock recovery circuit, it may be possible to make the
presented method bit-rate independent up to bandwidth of the
electronics. Careful synchronization of the processing delays and the
fiber-buffer delay would be necessary. In the demonstration, the
processing delay of the ECL circuit was < 2 ns, essentially equaling the
sum of all logic gate-delays.
It should be noted that a three-level probe signal could have been
created by simply summing the inverted new header and the old header
(using electrical-amplitude addition or optical-intensity addition), under
the condition that both amplitudes are equal prior to summing. This
would result in a three-level probe signal symmetric in amplitude about
the middle level. However, it was found th at an asymmetric probe
signal, made possible using the electronic processor, provided optimal
results.
Another issue is that of cascadability. The presented method
introduces a degradation in contrast ratio due to non-optimal pump and
probe powers. It is expected th at this contrast ratio degradation will be
the limiting factor of the functionality in a network where multiple
61
header replacement operations are necessary. There have been
demonstrations of contrast ratio enhancement of SOA cross-gain
compression wavelength shifting [9], [10] th at may be applied to the
presented method.
4.6 Chapter Summary
In this chapter, an experimental demonstration of simultaneous
header replacement and 19-nm wavelength shifting was presented. The
packet data rate was 1 Gh/s and the method utilized cross-gain
compression in an SOA. The method requires modulating the probe
laser with a three-level signal. The measured sensitivity was -27 dBm
at 109 BER using a 416-bit packet.
62
4 .7 References
[1] C. A. Brackett, A. S. Acampora, et al., "A scaleable
multiwavelength multihop optical network: A proposal for
research on all-optical networks," IEEE/O SA J. Lightwave
Technol., vol. 11, no. 5/6, pp. 736-753, May/June 1993.
[2] K. Sato, S. Okamoto, and H. Hadama, "Network performance and
integrity enhancement w ith optical path layer technologies,"
IEEE J. Select. Areas Commun., vol. 12, no. 1, pp. 159-170, Jan.
1994.
[3] K. C. Lee and V. O. K. Li, "A wavelength-convertible optical
network," IEEE/O SA J. Lightwave Technol., vol. 11, no. 5/6, pp.
962-970, May/June 1993.
[4] T. Miki, "Optical transport networks," Proc. IEEE., vol. 81, no. 11,
pp. 1594-1609, Nov. 1993.
[5] D. J. Blumenthal, R. J. Feuerstein, and J. R. Sauer, "First
demonstration of multihop all-optical packet switching," IEEE
Photon. Technol. Lett., vol. 6 , no. 3, pp. 457-460, March 1994.
[6 ] J. Spring, R. M. Fortenberry, and R. S. Tucker, "Photonic header
replacement for packet switching," Electron. Lett., vol. 29, no. 17,
pp. 1523-1525, 1993.
[7] E. Park and A E. Willner, "Simultaneous all-optical packet-
header replacement and wavelength shifting for a dynamically-
reconfigurable WDM network," OFC '95, TuQ4, San Diego, CA,
Feb. 1995; and IEEE Photon. Technol. Lett., vol. 7, no. 7, pp. 810-
812, July 1995.
[8 ] A. E. Willner and W. Shieh, "Optimal spectral and power
param eters for all-optical wavelength-shifting: single stage,
fanout, and cascadability," OFC '94, ThC4; and IEEE/O SA J.
Lightwave Technol., vol. 13, no. 5, pp. 771-781, May 1995.
63
[9] D. Norte and A. E. Willner, “Simultaneous probe and pump
extinction ratio enhancement demonstration in all-optical
noninverted wavelength shifting,” Opt. Fiber Commun. Conf., OFC
‘96, San Jose, CA, paper WM7, pp. 188-190.
[10] J. C. Simon, L. Lablonde, I. Valiente, L. Billes, and P. Lamouler,
"Two-stage wavelength converter with improved extinction ratio,"
Opt. Fiber Commun. Conf., OFC ‘95, San Diego, CA, post-deadline
paper PD15.
64
Chapter 5
Self-Routing of Wavelength Packets
Using Subcarrier Control Headers and
_______ All-Optical Wavelength Shifting
Chapter Highlights
• Novel experimental demonstration of a wavelength router to
enable the self-routing of wavelength-packets
5.7 Motivation
If packet switching is utilized in a dynamically reconfigurable
WDM network, the issue of implementing self-routing control of packets
may become apparent. In the previous chapter, the header of the packet
was located prior to the data payload. The header also operated at the
same bit-rate as the payload. This arrangement is disadvantageous a t
routing nodes which do not require the detection and processing of the
payload, but m ust however operate a t the same bit-rate as the payload.
As the bit-rate of the packet increases, processing speeds m ust also
increase. One method to alleviate this problem is to transm it the
header at a lower bit-rate than the payload. However, this will increase
the total packet size, thus decreasing the overall transm ission network
throughput.
65
There exists a routing control method which utilizes a microwave
subcarrier, multiplexed with the baseband payload data, on which a low-
speed control header is encoded [l]-[5]. This method of "tagging" the
payload data is advantageous for many reasons: (1) The subcarrier
information can be ind ep en d en tly detected and processed from the
payload data due to the subcarrier being spectrally out-of-band from the
baseband payload data. Consequently, the control signal can be
transm itted simultaneously with the payload and at a lower bit-rate.
Routing nodes can then utilize inexpensive electronics and off-the-shelf
microwave components. (2) This subcarrier control method also enables
the resolution of multiwavelength packet contention if each wavelength
is assigned a corresponding subcarrier frequency [3]-[5]; all of the
subcarriers can then be received by using a single detector. (3) Finally,
recovery of the header requires only a small portion of optical energy.
In this chapter, a novel experimental demonstration of a
wavelength (A ,) router is presented. This A-router, illustrated in Figure
5.1, directs self-routing optical packets, containing subcarrier-
multiplexed control information, from an input A-path to a designated
destination A-paths based on the header information. The A-router
incorporates 1 ) an all-optical wavelength shifter, based on SOA cross-
66
Incoming Packets on A ,a W avelenath-
B fiU ifiC
Outgoing Packets
Header 2
Payload 2
_HeaderN
Payload N
Header 1
Payload 1
All-Optical
wavelength
X,
Heater
Payload
Figure 5.1. Conceptual diagram of a wavelength-router.
gain compression, and 2 ) a low-speed electronic processor which
determines the destination X,-path based on the routing information
contained in the subcarrier header. Each packet is comprised of 1-Gb/s
NRZ baseband payload data and a 50-Mb/s header, quadrature-phase-
shift-key (QPSK)-encoded onto a 1.5-GHz subcarrier. The payload and
subcarrier header are transm itted concurrently. Experimental results
indicate 109 BER operation of routing packets to each of the four X-
paths. The maximum X,-shift is 23 nm, the maximum receiver
sensitivity differential among the four X— shifted packets is ~ 2 dB, and
the inter-packet guard time is 20 ns. The ^-router has the following
features: 1 ) the packets are routed onto different wavelengths rather
than fiber paths, 2 ) the packet transfer from input wavelength to output
wavelength is all-optical, 3) the packets are self-routed, and 4) the
67
subcarriers are processed by using commercially available microwave
components.
5.2 Subcarrier-Multiplexed Control
Figure 5.2 illustrates the implementation of subcarrier control. A
subcarrier is modulated w ith the header control information and is then
combined with the baseband data. Also shown in Figure 5.2 is the RF
spectrum of the packets which illustrates no RF spectral overlap of the
subcarrier and the baseband. This can be accomplished by low-pass
filtering both the control and the payload signals [4], [6 ].
5.2.1 S ubcarrier M odulation T echnique
High -Speed
Baseband
■ m n r i ,
J » .N
Recovered
Header
j\n,
Figure 5.2. Implementation of subcarrier-multiplexed control for baseband data
self-routing.
baseband
Low-Speed
Header
J i n
Gen. ^ ------
, . Hi- spectrum V
Subcarrier
r
Wavelengt
Router
Basebai
Dale
Sm
+r
SC Rx &
Recovery
Header
p RF Spectrum
Subcarrier
'SC 9
68
In our demonstration, QPSK modulation format was chosen
because it is a bandwidth efficient transmission technique compared to
ASK or FSK. Two independent phase modulations, one in-phase and one
quadrature-phase, are possible allowing two independent bit-streams to
be simultaneously transm itted. QPSK modulation enables the control
signal to be divided into two parallel bit-streams, thus reducing by half
either the header transmission time or the header bit-rate, given th at
parallel processing electronics is utilized at the routing node.
5.3 Experimental Design
Pscbol.fiom rator
Payload
Pattern
Generatoi
1 G b /s
Control
Pattern
Benaratoi
5 0 M b/s
1 t
1.5 GHz
jl I Control
n P I L I Subcarriei
Baseband
I L U
wflvatonqth
B&ut er,
'_ 1 .5 G H z L O
‘ ( ~ ) 1
Tunable
Filter
Figure 5.3. Diagram of experiment. The optical spectrum is a maximum-held
trace showing all wavelengths signals. The SOA gain peak is at 1565 nm, and
the bias is set to 180 mA.
Figure 5.3 shows the experimental setup. The packet generator is
described as follows. The baseband payload data is formed by
69
programming a pattern generator with a 1-Gb/s, 480-bit long, NRZ bit
stream comprised of multiple 2 7 -l pseudo-random sequences and 2 0
guard bits. The control signal is formed by programming a second
generator with a 50-Mh/s, 16-bit long, NRZ bit-stream which contains an
8 -bit flag, to signal the beginning of a packet, followed by an 8 -bit
header. Details of the transm itted packet can be found in Appendix B.
This signal is then used to drive two inputs of a QPSK modulator (one
modulates the in-phase, the other modulates the quadrature-phase) to
realize transm ission of parallel control bit-streams. Within the
modulator, the two inputs are mixed with a 1.5-GHz local oscillator to
form the QPSK-encoded subcarrier. The baseband payload data is
repeated for every packet while the control signal cyclically changes
among four different headers. Also, the pattern generators are
synchronized such th at the beginning of the control signal coincides in
time with the beginning of the baseband payload. A power combiner
multiplexes the subcarrier with the baseband payload to form the self-
routing packet. This combined signal is then used to directly modulate a
1571-nm ( X , a ) DFB pump laser. A 990-MHz and a 35-MHz low-pass
filter directly following the payload and control pattern generators,
respectively, are used to minimize crosstalk as previously mentioned.
90/10 _
,a p - O
X a Filter
soaH B B -
To Rx
EDF/
CM
XI Tx
CM
*9
,2 T x
QPSK
Demod
Clock
LO
Figure 5.4. Details of the Wavelength Router.
The wavelength router (Figure 5.4) is described as follows.: A
90/10 optical coupler taps 10% of the input A a signal for detection by a
1.7-GHz receiver. A high-pass RF filter passes only the subcarrier to a
QPSK demodulator, in which the same 1.5-GHz local oscillator used for
the transm itter is also used to mix with the subcarrier. One of two
demodulator mixed outputs is low-pass filtered to recover the control
signal which is then fed into an electronic routing processor, described in
the following section. The processor determines which one of four probe
lasers (on wavelengths Ai =1548 nm, A 2 =1552 nm, A 3 =1557 nm, or A 4
=1562 nm) within the A-shifter should be switched "ON". The input A a
signal is amplified using an EDFA (to ensure sufficient SOA gain
compression for proper A-shifting) and delayed (to allow for -320 ns
processing time). A 1571-nm bandpass filter after the EDFA limits the
71
amount of amplified spontaneous emission (ASE) entering the following
SOA. An angle-tuned 1 -nm bandpass optical filter selects which probe
wavelength should be passed to a 1-GHz receiver (BT&D PDC2201
GaAs PINFET). For optimal operation of the processor and sufficient
contrast ratio for ^.-shifting, an experimentally determined subcarrier
modulation depth [4] of -0.16 is used at the transm itter.
5.3.1 E lectron ic P rocessor
Q 7 B O
DIP
SW ITCH
COMP B-BH
SR
MASTER
CLOCK IN
CLOCK
DISTRIBUTOR
- 8 - bit
d ela y
Q 0
A7
CLK
INPUT
q g iltin a Cod* Pro*— ln?
Q7
Q 3
D2
S-BH 04
SR
02
D1
Q1
DO
Q O
Q O
4 3
CLK
Monitoring
Figure 5.5. Logic diagram of the electronic processor.
The processor determines which one of four probe lasers should be
switched "ON" based on two bits contained in the subcarrier header.
72
Details of the header structure can be found in Appendix B. The
electronic processor is comprised of flag detection and header processing
electronics (>l-GHz Motorola ECLinPS integrated circuit chips). It was
overdesigned, in term s of function and speed, for future use. The
processor operation is as follows. The recovered header is input into the
processor and split with one branch delayed by approximately eight bit-
times. Both signals are then converted from serial to parallel. The flag
recognition circuit compares the incoming bit-stream with a preset flag
code, which in our demonstration was ( 0 1 1 1 1 1 1 0 ) . The reason for
the 8 -bit delay was to synchronize the flag detection signal w ith routing
code signal such th a t both signals enter a D-flip/flop simultaneously.
The AND/NAND gates create a decoder. The binary routing code at the
decoder input instructs which one of four D-flip/flop outputs to turn on.
The control signal pattern generator provides the 50-MHz clock source
for the processor.
5.4 Results and Discussion
Figure 5.6 shows oscilloscope traces illustrating successful A -
routing of packets to each destination A-path. Packets are cyclically
routed from A a to A * , A 4, A 2, and A 3, with no packets being dropped. The
73
Figure 5.6 inset is an oscilloscope trace of overlapping packet-switching
transitions (showing jitter of the 50-MHz control signal generator clock)
and surrounding X-shifted data bits separated by a 20 ns (~4% of the
packet time) guard time. A guard time during packet-switching
transitions is necessary to eliminate bit errors when a low-speed circuit
is used to packet-switch high-speed data. In addition, because cross
gain compression A.-shifting inverts the data, guard bits are set to "1 's"
Transmitted Packets on
Packet length = 480 ns
I Pkt1 I Pkt 2 I Pkt 3 I Pkt 4 I
D ata
1 Gb/s , J j' ^ ! i J , „ !
C ontrol
5 0 Mb/s _ _
Routing . Q O 11........10........ 01...
bits ^ |
D .
R ecovered
Control Signals
Routing Scheme
Pkt 1 : A a ->*.1
Pkt 2: A A — t M
Pkt 3: A A -»te
Pkt 4: AA-»A3
n
Guard Time
Packet Packet
IN SE T
Wavelength
Router
Probe L aser Signals
Self-Routing Wavelength Packets
M J ( | __________| |
Pkt 1 h k iu
* ■
u i m 3m
u
Pkt 2
Figure 5.6. Oscilloscope traces of the transmitted packets, recovered control
signals, probe-laser signals, and self-routed baseband wavelength packets.
Details of the guard time are also shown. Traces of self-routed packets are taken
using a trigger derived from the low-speed clock.
74
instead of "0's." This effectively compresses the SOA gain during the
packet-switching transitions (Figure 5.6 inset), forcing the output to the
"low" state.
5.4.1. S en sitiv ities
Figure 5.7 shows sensitivity curves for the ^.-routed baseband
payload data and their corresponding 10'9 BER eye-diagrams. All peak
10"2 n 1 Gb/s Baseband Payload Data Eye-Diagrams
C 1 0
U J
m
1 0
1 0
1 0
V
-6
O x i
C D X .2
-8
OX3
A X 4
-1 1
O
O
50 Mb/s Recovered
Subcarrier
Control Signal
OoU 0
%
| i i i | i i i | i i " i | i 1 i | i i 1 | i i i |
-3 0 -2 8 -2 6 -2 4 -2 2 -2 0 -1 8 -1 6
R e c e iv e d P ow er (dB m )
Figure 5.7. Sensitivity curves for 1-Gb/s payload data which have been routed
from A.-path A to X-paths 1 through 4. Eye-diagrams of packets and a 50-M b/s
control signal are also shown. Intersymbol interference of the control signal is
attributed to the QPSK assembly.
75
probe powers are set to -15.5 dBm, except for a less efficient A * probe
laser which is set to -19.5 dBm. The A a pump power is set at -1.0 dBm
to provide optimal performance for packets routed to Ai. To measure the
BER of each A-path, the BER unit was programmed to recognize the
inverted packet payload followed by three packet lengths of "0's". The
sensitivity curves and baseband eye-diagrams reflect the expected result
of an increasing contrast ratio of the A-shifted signals as the probe
wavelength decreases [7], given the wavelengths used in the
demonstration. The maximum sensitivity differential among the A-
routed packets is less than 2 dB.
Figure 5.7 also shows an eye-diagram of the recovered error-free
control signal at the processor input. The intersymbol interference is
attributed to operating limitations of the QPSK modulator/demodulator.
5.5 Issues and Hurdles
A number of issues m ust be considered when using the
demonstrated A-router in a network node. To handle contention
resolution, the A-router can be used in conjunction with demonstrated
contention resolution schemes [4]-[5] which utilize subcarrier-multiplex
headers. The issue of data-inversion of SOA cross-gain compression X -
shifters can be addressed by possibly encoding a polarity code on one of
the two QPSK phase-modulations, instructing the receiving node to
electronically invert the detected baseband data. An alternative method
is to use two SOAs which, in addition to maintaining data polarity, can
simultaneously enhance the X-shifter output contrast ratio [8 ]. In the
presented demonstration, the electronic processor utilized the
transmitter-source clock and local oscillator. In a network node, the X -
router would require subcarrier recovery before decoding the header;
processing errors may then occur if the recovered subcarrier frequency
does not match or drifts from the transm itted subcarrier frequency. To
avoid carrier recovery circuitry, differential-QPSK encoding/decoding can
be employed [9]. Also, in a network node, a dock recovery circuit as
demonstrated in [3] would be required to derive the clock from the
received control signal.
The presented demonstration does not address the effect of cross
gain compression X-shifting on the subcarrier, which is im portant for
cascaded X-routers. In Chapter 7, the issue of X-shifting the baseband
an d subcarrier will be addressed. With the proper operating conditions
77
it is possible to also detect a X-shifted subcarrier header. Also, we
expect the measured sensitivity of the X-shifted payload to degrade as
the subcarrier modulation index is increased.
5.6 Chapter Summary
The combined usage of an all-optical wavelength shifter, based on
SOA cross gain compression, and subcarrier multiplexing to direct self
routing packets was experimentally demonstrated.
5.7 References
[1] A. Budman, E. Eichen, J. Schlafer, R. Olshansky, and F.
McAleavey, “Multigigabit optical packet switch for self-routing
networks w ith subcarrier addressing,” Opt. Fiber Commun. Conf.,
OFC‘92, San Jose, CA, 1992, paper Tu04, pp. 90-91.
[2] S. F. Su, A. R. Bugos, V. Lanzisera, and R. Olshansky,
"Demonstration of a multiple-access WDM network with
subcarrier-multiplexed control channels," IEEE Photon. Technol.
Lett., vol. 6 , no. 3, pp. 461-463, March 1994.
[3] M. Cerisola, T. K. Fong, R. T. Hofmeister, L. G. Kazovsky, C. L. Lu,
P. Poggiolini, D. J. M. Sahido IX, "Subcarrier multiplexing of
packet headers in a WDM optical network and a novel ultrafast
header dock-recoveiy technique," Opt. Fiber Commun. Conf., OFC
‘95, San Diego, CA, paper THI4, pp. 273-274.
[4] C.-L. Lu, D. J. M. Sabido IX, P. Poggiolini, R. T. Hofmeister, and L.
G. Kazovsky, “CORD— A WDMA optical network: Subcarrier-based
signaling and control scheme,” IEEE Photon. Technol. Lett., vol. 7,
no. 5, pp. 555-557, May 1995.
78
[5] W. Shieh and A. E. Willner, "Demonstration of output-port
contention in a 2x2 WDM switching node based on all-optical
wavelength shifting and subcarrier-multiplexed routing-control
headers," Opt, Fiber Commun. Conf., OFC '96, San Jose, CA, post
deadline paper PD36.
[6 ] P. M. Hill and R. Olshansky, “Bandwidth efficient transmission of
4 Gh/s on two microwave QPSK subcarriers over a 48 km optical
link,” IEEE Photon. Technol. Lett., vol. 2, no. 7, pp. 510-512, July
1990.
[7] A. E. W illner and W. Shieh, "Optimal spectral and power
param eters for all-optical wavelength-shifting: single stage,
fanout, and cascadability," IEEE/O SA J. Lightwave Technol.,
Special Issue on Optical Amplifiers, vol. 13, no. 5, pp 771-781,
May 1995.
[8 ] D. Norte and A. E. Willner, “Simultaneous probe and pump
extinction ratio enhancement demonstration in all-optical
noninverted wavelength shifting,” Opt. Fiber Commun. Conf., OFC
‘96, San Jose, CA, paper WM7, pp. 188-190.
[9] R. Gross and R. Olshansky, "Heterodyne video transmission with
differentially encoded quadrature phase shift keying," IEEE/O SA
J. Lightwave Technol., vol. 10, no. 5, pp. 679-685, May 1992.
79
Chapter 6
Multiple-Wovelength-lnput All-Optical
Wavelength-Shifting with Subcarrier-
_____ Multiplexed Control for Self-Routing
Chapter Highlights
• Novel experimental demonstration of a two-input wavelength-
router
• Wavelength-router incorporates wavelength-interchange and
subcarrier control for self-routing.
6.7 Motivation
The method of SOA cross-gain compression ^.-shifting is incapable
of shifting more than one input wavelength a t a time, thereby severely
limiting network functionality. In this chapter, a method for all-optical
wavelength shifting of multiple-input-wavelengths is presented (Figure
6 .1). The technique involves spatial separation of incoming wavelengths
and then shifting these multiple signals in parallel. Subcarrier-
multiplexed routing control [l]-[2 ] and wavelength-interchange are
incorporated to increase functionality. Error-free A,-shifting of 1-Gb/s
data is measured for all cases.
A A A 1
Incoming
WDM Channels
A B A2
o
Multi-
Channel
Wavelength
Shifter
o
Outgoing ArShifted
WDM Channels
Figure 6.1. Diagram of a Multiple-Channel Wavelength Shifter for a WDM
routing node.
6.2 Experimental Design
Two incoming wavelengths ( A ,a and X B ) are spatially separated by
splitting and filtering, and each shifted in parallel by a separate SOA;
additionally, subcarrier-multiplexed routing control is used and
wavelength interchange is incorporated. Each channel contains packets
comprised of480-bit long, 1-Gb/s, NRZ baseband data multiplexed with
a subcarrier (fA or fB ) which uniquely identifies the channel wavelength.
The subcarriers are QPSK -modulated w ith 16-bit long 50-Mh/s control
headers. See Appendix B for packet structure details. The combined
transm itted WDM signal enters the multiple-channel X-shifter and is
tapped and detected using a 1.7 GHz subcarrier receiver (BT&D
PDC2201 GaAs PINFET). The headers on subcarriers fl and f2 are
recovered and used to instruct an electronic routing processor to turn
"ON" one of two probe lasers. The processors are identical to th at which
was described in Chapter 5 and will not be repeated here. Based on the
header information, packets are X-shifted to the other channel's
wavelength (i.e., X A -> XB) (i.e., w avelength interchange) or to an
entirely different wavelength (i.e., X A — »Xl); similarly for the X B channel.
The pumps and the selected probe signals are then coupled into their
respective SOAs in a counter-propagating fashion in order to avoid the
pump signals from appearing at the output. An angle-timed 1 -nm
bandpass optical filter is used to select which probe wavelength should
be passed to a 1.7-GHz baseband receiver (BT&D PDC2201 GaAs
PINFET) at the output.
B ase-
Band
Data
Gen.
1 Gb/s
Sub-
Carrier
Control
Gen.
SO Mb/s
MUXa
/ a =
1.4 GHzl
MUXb
/ b =
1.2 GHzl
A A /I*
EDFA
Header
Recovery
& Proc.
X1 \ j
■ g A
SC
Fix
— t
Header
Recovery
& Proc.
■ i m K /
- l ^ J A
SDAl
_ B _ \J ~ *
A .
X1, A .B ’
X .2 . X A ’
S O A a
XA=XA’ Xb =Xb ’
L ♦
\ 2
t
S O A B
— —
“I---1
1530
I 1 I I I
1540 1550 1560 1570
I ^ V
1580nm
Figure 6.2. Packets on XA (= 1542 nm) are either down-shifted to X, (= 1535 nm)
or up-shifted to XB* (= 1557 nm) using SOAa. Packets XB (= 1557 nm) are either
down-shifted to X A’ (= 1542 nm) or up-shifted to X2 (= 1571 nm) using SOAB .
82
6.3 Results and Discussion
A 10'9BER is achieved for all A-shifting cases for both techniques.
Figure 6.3 shows scope traces of self-routed ^.-shifted packets. The
optical powers used are (measured at the SOA input fibers): P a= -3.5
dBm, PB = -4.6 dBm, P i = -1 dBm, PB' = - 8 dBm, PA' = -7 dBm, and P2 =
-22 dBm.
A1 < — AA
1A 2A 3A 4A 5A 6A.
i , 1 ','i
AA-> AB’
Mis A v
Figure 6.3. Scope traces showing A-shifted self-routed packets.
Probe power for A ,i is large in order to compensate for poor coupling into
one facet of SOAa. Probe power ta is much smaller than probe fa'
because the wavelength is near the gain peak of SOAb and thus
83
experiences more gain. Although pump-probe counter-propagation was
used, the presence of the pump is observed at the output due to
reflections. For the interchange case, optimized filtering methods can
avoid the pump from interfering with the matching probe (with the same
wavelength) from the other SOA after optical combining.
6.3.1 S en sitiv ities
-3-,
□ +
-4.
□ +
- 6 -
flC
L U
ffi
- 8 .
□
-40 -35 -30 -25 -20
Received Power (dBm)
Figure 6.4. Sensitivity curves for wavelength-routed packets.
84
Figure 6.4 shows the sensitivity curves. Due to unavailability of
certain components, BER measurements were obtained separately for
each SOA; th at is, the two SOA outputs were not optically combined
prior to detection. As expected, the contrast ratio is better for
downshifting (X A -» Xl and X B -» X A ') than for upshifting (XA — > X B 1 and
X B - » X 2 ) as evidenced by the curves in Figure 6.4.
6.4 Chapter Summary
Error-free operation of two-X-input all-optical X-shifters were
demonstrated. Wavelengths were demultiplexed and each signal was
wavelength shifted using separate SOAs with offset gain-peaks. Self
routing of packets, using subcarrier control, and wavelength interchange
were also incorporated. Multiple-input X-shifters will have the
potential of increasing the functionality of reconfigurable WDM
networks.
85
6.5 References
[1] A. Budman, E. Eichen, J. Schlafer, R. Olshansky, and F.
McAleavey, “Multigigabit optical packet switch for self-routing
networks with subcarrier addressing,” Opt. Fiber Commun. Conf.,
OFC ‘92, San Jose, CA, paper Tu04, pp. 90-91,1992.
[2] E. Park and A.E. Willner, "Self-routing of wavelength packets
using an all-optical wavelength shifter and QPSK subcarrier
routing control headers," IEEE Photon. Technol. Lett., July 1996.
86
Chapter 7
All-Optical Wavelength Shifting
of Baseband Data and
________ Subcarrier-Multiplexed Signals
Chapter Highlights
• Successful experimental demonstration of all-optical wavelength
shifting of baseband and subcarrier-multiplexed data.
• Computer modeling
7.7 Motivation
Chapters 5 and 6 have shown how im portant the combination of
all-optical wavelength shifting and subcarrier-multiplexing is for
enabling packet self-routing in a dynamically reconfigurable WDM
network. The method of subcarrier multiplexing can be generalized to
include simultaneous transmission of multiple independent analog and
digital data stream s on a single optical carrier wavelength. In the
wavelength-router presented in Chapter 5, the baseband packets were
successfully routed via the wavelength-shifter. However, the subcarrier
information was not transferred along with the payload. This instance
may occur when the router acts as a gateway between a wide-area
network and a local area network (e.g., Figure 1.8) where it may not be
necessary for the routing information to be passed on locally. There may
be situations where the subcarrier information m ust rem ain intact even
after the packet has been wavelength-shifted. An example is in future
WDM networks in which packets may traverse multiple wavelength (A ,)
paths [1], involving multiple X-shifts via cascaded X-routing nodes. It is
therefore critical to determine the limitations of transm itting baseband
digital data with a multiplexed subcarrier within such a network.
In this chapter, experimental results are presented indicating
th at it is possible to all-optically wavelength-shift both the baseband
and the subcarrier when using SOA cross-gain compression. To achieve
simultaneous 109 BER X-shifting of both the baseband data and the
demodulated control signal, the subcarrier modulation depth m ust be
increased from the baseline (no X-shifting). Because the subcarrier
survives the X-shifting process, it is available for processing at a
subsequent network node.
88
7.2 Computer Modeling
P ir ff) R F S P ic tm m
Subcarrier
High-Speed
Low-Speed Control
Control BPSK-Encoded
onto Subcarrier
fsc
Subcameron Baseband T s '
^ To X-Shitter
Input
Baseband + Subcarrier
P in®
f Wavelenoth Shifter
fcpump
s o A ^ J a L
A ,p robe
filter j
^CW X peobi
^.-Shifter Output Diminished Subcarrier
Amplitude
r\
Baseband + Subcarrier
(b)
lUfW"
Hecovered Subcarrier Controf
Figure 7.1. (a) Computer modeled baseband data multiplexed with an encoded
subcarrier. Input signals are RF filtered to avoid crosstalk, (b) All optical X-
shifter and computer modeled X-shifted output. Baseband+subcarrier traces are
time-expanded.
Figure 7.1 shows a computer-modeled representation of baseband
data and a subcarrier-multiplexed control signal, before and after X-
shifting. A requirement for multiplexing a subcarrier with baseband
data is th a t there is essentially no RF spectral overlap between the two
signals (Figure 7.1a). This is accomplished by low-pass filtering both
signals to suppress their sidelobes [2], [3]. Figure 7.1b shows the SOA
89
cross-gain compression X-shifter, used in our demonstration. Using a
basic SOA saturated-gain expression (Equation 2.6) [4],
we numerically calculate the SOA gain-compression response G(t) due to
the input pump signal Pm(t), where Go is the unsaturated gain and Ps is
the saturation power. For simplicity of illustration, the modeled
subcarrier is encoded using the binary-phase-shift-keyed (BPSK) format.
Figure 7.1b shows that the compression causes a non-uniform amplitude
transfer of the subcarrier; the subcarrier oscillation amplitude on the
baseband "l's" is diminished because the SOA gain is strongly
compressed in this situation. Despite this effect, we anticipate the
control information to be ftdly recoverable from the X-shifted subcarrier
if a phase modulation format is used. In our experiment, the QPSK
modulation format is used to encode the control signal onto the
subcarrier.
7.3 Experimental Design
Figure 7.2 shows the experimental setup. A 622-Mb/s baseband
data pattern generator is programmed with repeating 22 3 -l pseudo-
(7.1)
90
0 \ 1 2 GHz 1 \ 2 GHz Jo GHz
Control
Pat. Gen.
50 Mb/a
Data
Pat Gen.
622 Mh/«
1552 nm Probe
'O p tic a l S p e c t r u m
QPSK
Mod QPSK HlPFJ-^
Demoo 5 0 jj
^ Term.
I m n j -
Figure 7.2 Experimental setup along with RF and optical spectra at specific
points in the setup. The SOA gain peak is 1565 nm, and the bias is 140 mA.
(OA=optical attenuator, HPF=high-pass filter, LPF=low-pass filter).
random NRZ baseband sequences. A 50-Mb/s control pattern generator,
programmed with repeating 5-byte NRZ signals, drives the in-phase and
quadrature-phase inputs of a commercially available QPSK modulator.
A 5-byte control pattern is used because it corresponds to an ATM
packet header length and because of the unavailability of a second
pattern generator which could provide a longer sequence. We simulate
two independent stream s of control information by delaying one QPSK
input with respect to the other. A 990-MHz signal generator (HP 8656B)
is used as the QPSK mod/demodulator local oscillator. To minimize
spectral overlap between the subcarrier and baseband data, a 35-MHz
low-pass filter and a 570-MHz low-pass filter are used after the control
91
and data generators, respectively. The filtered signals are then
combined and used to directly modulate a 1571-nm DFB pump laser.
The pump signal is fed into the X-shifter, which is comprised of a 1552-
nm DFB probe CW laser, an SOA w ith a 1565-nm gain peak, and a 1-nm
bandpass optical filter centered around 1552 nm . A 1.7-GHz receiver
(BT&D PDC2201 GaAs PINFET) detects the X-shifter output. The
optical powers measured a t the SOA input fiber are -5.7 dBm for the
pump and -7.5 dBm for the probe. The received RF signal is amplified
and split into two output paths, one for baseband detection, the other for
control signal detection. For baseband detection, a 770-MHz low-pass
filter is used to only pass the X-shifted baseband signal and block the
subcarrier. For control detection, a band-pass filter (comprised of a high-
pass/low-pass filter cascade) is used to only pass the subcarrier to a
QPSK demodulator. A 67-MHz low-pass filter is used a t one of two
demodulator outputs to recover the 50-Mh/s control signal, the other
output is 50-Q-terminated. In our demonstration, we detect only one of
the two phase modulations. Because X-shifting inverts not only the
baseband data but also the subcarrier (i.e., introduces a 180° phase
shift), the demodulated control signal is also inverted.
92
7.4 Subcarrier Modulation Depth
The subcarrier modulation depth m, defined as the ratio of optical
subcarrier modulation amplitude to the optical subcarrier-plus-
baseband modulation amplitude, is used to describe the amount of
power distributed to the subcarrier and baseband. ra=0 if only the
baseband is transm itted and m=l if only the subcarrier is transm itted
[2]. The subcarrier modulation depth, measured at the pump output, is
varied by attenuating the subcarrier RF amplitude while adjusting the
baseband RF amplitude using the pattern generator controls. The total
subcarrier-plus-baseband amplitude and the pump laser bias current
are held constant.
7.5 Results an d Discussion
Figure 7.3 shows experimental results of the received power (@
109 BER) for both the baseband data and demodulated control signal as
a function of m. When measuring the BER for the baseband data, the
output ports of the QPSK demodulator are properly terminated.
Similarly, when measuring the BER for the demodulated control signed,
the payload path is properly terminated. A cross-over point is observed
in both the baseline (Figure 7.3a) and the ^.-shifted case (Figure 7.3b)
93
-20 -
•
Before X-Shlfter
E*
ID -25
2 ,
C C
IU -30
ID
o>
* -35
* * *
Optimal Point
▲
▲
• 6
Baseband
NRZ Data
Subcarrier
Control
&
• i i i
) 0.2
' 1 1
0.4
i i i i i
0.6
i i i i i i i
0.8 1
% -1 0 :
O L-Shifter
o
a- -15-
XI
a>
> -20
a >
2 a
A
A
Baseband
NRZ Data
a > -25
A C
-30
A o
Optimal Point
--- r— r— . —,--- ■ --- ,--- 7
0 o
---1 --- 1 --- j -r-
Subcarrier
Control
- i V | . . I |
0 0.2 0.4 0.6 0.8 1
Subcarrier Modulation Depth, m
Figure 7.3. Sensitivities @ 10'9 BER as a function of m measured (a) before and
(b) after the A.-shifter. Each plot shows a crossover indicating an optimal point
for 10'9 BER operation of both the baseband and subcarrier.
indicating values of m which provide optimal and simultaneous 10'9
BER detection of the baseband d ata and demodulated control signal.
The optimal value of m is -0.20 for the baseline case (no X-shifter) and
-0.32 for the X-shifted case. This indicates that if the combined
baseband and subcarrier multiplexed signals are to be X-shifted and
detected, the subcarrier modulation depth a t the transm itter m ust be
94
increased; this adjustm ent effectively compensates for the SOA gain-
compression effect on the subcarrier amplitude.
10 2-
oc 10 4
U J
m
10 6
10 8
1 0 ' 1 1 -
Optimized For Wavelength Shifting (m = 0.321
-35
Before ^.-Shifter:
▲ Baseband NRZ Data
• Subcarrier Control
After A,-Shifter:
A Baseband NRZ Data
O Subcarrier Control
A ° 0
a a o
- A
I 11 i i » I i » i
-30 -25 -20
Received Power (dBm)
i 1 i i
-15
Figure 7.4. BER sensitivity curves for m =0.32.
Figure 7.4 shows receiver sensitivities measured before and after
the X-shifter, with the subcarrier modulation depth set at ra=0.32. The
curves of the X-shifting case do not exactly coincide at 10'9 BER. This
can be attributed to non-ideal experimental conditions. The reduced
sensitivity at low BERs, reflected in the slope change in Figure 7.4, is
attributed to 1) crosstalk between the subcarrier and baseband signals
introduced by the X-shifter, and 2) the non-optimal bandwidth of the
95
high-pass/low-pass filter cascade used for bandpass-filtering the
subcarrier.
7.6 Chapter Summary
A demonstration of baseband and multiplexed-subcarrier all-
optical wavelength shifting was presented. 19-nm all-optical
wavelength shifting of 622-Mb/s baseband data multiplexed with a
microwave subcarrier which is QPSK-encoded with a low-speed control
signal was presented. Experimental results were shown indicating th a t
if the combined baseband data and subcarrier multiplexed control
signals are to be ^-shifted, a larger subcarrier modulation depth at the
transm itter is required compared to the case with no X-shifting.
7.7 References
[1] K. Sato, S. Okamoto, and H. Hadama, "Network performance and
integrity enhancement with optical path layer technologies,"
IEE E J. Select. Areas Commun., vol. 12, no. 1, pp. 159-170, Jan.
1994.
[2] C.-L. Lu, D. J. M. Sabido IX, P. Poggiolini, R. T. Hofineister, and L.
G. Kazovsky, “CORD--A WDMA optical network: Subcarrier-based
signaling and control scheme,” IEEE Photon. Technol. Lett., vol. 7,
no. 5, pp. 555-557, May 1995.
96
[3] P . M. Hill and R. Olshansky, “Bandwidth efficient transmission of
4 Gb/s on two microwave QPSK subcarriers over a 48 km optical
link,” IEEE Photon. Technol. Lett., vol. 2, no. 7, pp. 510-512, July
1990.
[4] G. P. Agrawal, Fiber-Optic Communication Systems. New York:
John Wiley & Sons, Inc., 1992.
97
Chapter 8
Conclusion
This dissertation has presented concepts and experimental
demonstrations of all-optical wavelength shifting using semiconductor
optical amplifiers which are applicable to dynamically reconfigurable
WDM networks. The presented demonstrations were as follows. (1)
Simultaneous header replacement and all-optical wavelength shifting—
by modulating the wavelength-shifter probe laser with a three-level
signal, header bits were replaced on-the-fly and bit-by-bit; the result was
the additional function of header replacement was incorporated with
wavelength shifting. (2) Self-routing of wavelength packets using
subcarrier control and all-optical wavelength shifting-two concepts were
combined to construct a wavelength router which directed packets
containing baseband data information tagged with a subcarrier control
signal. (3) Multiple-input wavelength shifting incorporating self-routing
and wavelength-interchange-input data packets were wavelength-
demultiplexed and each wavelength signal was individually wavelength
shifted using separate SOAs. (4) All-optical wavelength shifting of
baseband and subcarrier signals-it was shown th at along with the
98
baseband data, the subcarrier control signal is recoverable after SOA
cross-gain compression wavelength shifting.
All demonstrations utilized cross-gain compression of
semiconductor optical amplifiers to perform all-optical wavelength
shifting. BERs of 10'9 were achieved in all demonstrations. The
demonstrated concepts can be utilized in network nodes within a
dynamically reconfigurable all-optical WDM network to enhance or
enable routing/switching functions.
99
Bibliography
G. P. Agrawal, Fiber-Optic Communication Systems. New York: John
Wiley & Sons, Inc., 1992.
S. B. Alexander, R. S. Bondurant, D. Byrne, V. W. S. Chan, S. G. Finn, R.
Gallager, B. S. Glance, H. A. Haus, P. Humblet, R. Jain, I. P. Kaminow,
M. J. Karol, R. S. Kennedy, A. Kirby, H. Q. Le, A. A. M. Saleh, B. A.
Schofield, J. H. Shapiro, N. K Shankaranarayanan, R. E. Thomas, R. C.
Williamson, and R. W. Wilson, "A precompetetive consortium on wide
band all-optical networks," IEEE / OSA J. Lightwave Technol., vol. 11, no.
5/6, pp. 714-735, May/June, 1993.
J. E. Berthold, "Paving the way for multiwavelength networking on a
local-to-national scale," Opt. Fiber Commun. Conf., OFC ‘ 95, paper
ThE2, pp. 241-242, San Diego, CA.
D. J. Blumenthal, R. J. Feuerstein, and J. R. Sauer, "First
demonstration of multihop all-optical packet switching," IEEE Photon.
Technol. Lett., vol. 6, no. 3, pp. 457-460, Mar. 1994.
C. A. Brackett, A. S. Acampora, J. Sweitzer, G. Tangonan, M. T. Smith,
W. Lennon, K. C. Wang, and R. H. Hobbs, "A scalable multiwavelength
multihop optical network: A proposal for research on all-optical
networks," IEEE/O SA J. Lightwave Technol., vol. 11, no. 5/6, pp. 736-
753, May/June, 1993.
A. Budman, E. Eichen, J. Schlafer, R. Olshansky, and F. McAleavey,
“Multigigabit optical packet switch for self-routing networks with
subcarrier addressing,” Opt. Fiber Commun. Conf., OFC ‘92, San Jose,
CA, paper Tu04, pp. 90-91,1992.
M. Cerisola, T. K. Fong, R. T. Hofineister, L. G. Kazovsky, C. L. Lu, P.
Poggiolini, D. J. M. Sabido IX, "Subcarrier multiplexing of packet
headers in a WDM optical network and a novel ultrafast header clock-
recovery technique," Opt. Fiber Commun. Conf., OFC ‘95, San Diego, CA,
paper THI4, pp. 273-274.
I. Cha, M. Kitamura, H. Honmou, and I. Mito, "1.5 pm band travelling-
wave semiconductor optical amplifer with window facet structure,"
Electron. Lett., vol. 25, no. 18, pp. 1241-1242, Aug. 1989.
100
S. L. Danielsen, C. Joergensen, M. Vaa, and K. E. Stubkjaer, "Bit error
rate assessment of a 40 Gb/s all-optical polarisation independent
wavelength convertor," Opt. Fiber Commun. Conf., OFC ‘96, San Jose,
CA, postdeadline paper PD12.
E. Desurvire, Erbium-Doped Fiber Amplifiers-Principles and
Applications, John Wiley & Sons Inc., 1994, New York.
P. Doussiere, P. Garabedian, C. Graver, D. Bonnevie, T. Fillion, E.
Derouin, M. Monnot, J.G. Provost, D. Lederc, and M. Klenk, "1.55-pm
Polarization-independent semiconductor optical amplifier with 25-dB
fiber to fiber gain," IEEE Photon. Technol. Lett., vol. 6, no. 2, pp. 170-172,
Feb. 94.
S. Dubovitsky, P. D. Dapkus, A. Mathur, W. H. Steier, "Wavelength
conversion in a quantum-well polarization-insensitive amplifier," IEEE
Photon. Technol. Lett., vol. 6, no. 7, pp. 804-807, July 1994.
T. Durhuus, C. Joergensen, et al., "Optical wavelength conversion and
gating at 4 Gbit/s using semiconductor optical amplifiers," OFC/IOOC
'93 TuH6.
T. Durhuus, R. J. S. Pedersen, et al., "Optical wavelength conversion over
18nm at 2.5 Gb/s by DBR-laser," IEEE Photon. Technol. Lett., vol. 5, no.
1, pp. 86-88, Jan. 1993.
G. Eisenstein, R. S. Tucker, J. M. Wiesenfeld, P. B. Hansen, G. Raybon,
B. O. Johnson, T. J. Bridges, F. G. Storz, and C. A Burrus, "Gain recovery
time of traveling-wave semiconductor optical amplifiers," Appl. Phys.
Lett., vol. 54, no. 5, pp. 454-456, Jan. 1989.
L. Eskildsen, B. Mikkelsen, T. Durhuus, C. G. Joergensen, K E.
Stubkjaer, P. Doussiere, P. Garabedian, F. Leblond, J.-L. Lafragette,
and B. Femier, "Polarization-insensitive semiconductor optical
preamplifier at 1.55 pm," Electron. Lett., vol. 28, no. 21, pp. 2019-2021,
Oct. 1992.
L. Gillner, "Modulation properties of a near travelling-wave
semiconductor laser amplifier," IEE Proc. Pt. J, vol. 139, no. 5, pp. 331-
338, Oct. 1992.
101
B. Glance, J. M. Wiesenfeld, U. Koren, A. H. Gnauck, H. M. Presby, and
A. Jourdan, "High performance optical wavelength shifter," Electron.
Lett., vol. 28, no. 18, pp. 1714-1715, Aug. 1992.
P. E. Green, Jr., L. A. Coldren, "All-optical packet-switched
metropolitan-area network proposal," IEEE/O SA J. Lightwave Technol.,
vol. 11, no. 5/6, pp. 754-763, May/June, 1993.
R. Gross and R. Olshansky, "Heterodyne video transmission with
differentially encoded quadrature phase shift keying," IEEE/O SA J.
Lightwave Technol., vol. 10, no. 5, pp. 679-685, May 1992.
G. Grosskopf, R. Ludwig, and H. G. Weber, "140 Mbit/sDPSK
transmission using an all-optical frequency convertor with a 4000 GHz
conversion range," Electron. Lett., vol. 24, no. 17, pp. 1106-1107, Aug.
1988.
A. M. Hill, A. J. N. Houghton, "Optical networking in the European
ACTS Programme," Opt. Fiber Commun. Conf., OFC ‘96, paper T hll, pp.
238-239, San Jose, CA.
P. M. Hill and R. Olshansky, “Bandwidth efficient transmission of 4 Gh/s
on two microwave QPSK subcarriers over a 48 km optical link,” IEEE
Photon. Technol. Lett., vol. 2, no. 7, pp. 510-512, July 1990.
B. Hui, "Progress in optical networking and technology in the U.S.A.,"
Opt. Fiber Commun. Conf., OFC ‘ 96, paper ThI2, pp. 239-240, San Jose,
CA.
P. P. Iannone, P. R. Prucnal, G. Raybon, U. Koren, and C. A. Burras,
"Broadband wavelength shifter for picosecond optical pulses at 1.5mm,"
Electron. Lett., vol. 29, no. 17, pp. 1518-1519.
K. Inoue, "Crosstalk and its power penalty in multichannel transmission
due to gain saturation in a semiconductor laser amplifier," J. Lightwave
Technol., vol. 7, no. 7, pp. 1118-1124, July 1989.
K Inoue and H. Toba, "Wavelength conversion experiment using fiber
four-wave mixing," IEEE Photon. Technol. Lett., vol. 4, no. 1, pp. 69-72,
Jan. 1992.
102
K. Inoue, M. Yoshino, and F. Kano, "Polarisation insensitive wavelength
conversion using a light injected DFB-LD with a loop configuration,"
Electron. Lett., vol. 30, no. 5, pp. 438-439, Mar. 1994.
C. Joergensen, T. Durhuus, et al., "4 Gb/s Optical wavelength conversion
using semiconductor optical amplifiers, " IEEE Photon. Technol. Lett.,
vol. 5, no. 6, pp. 657-660, June 1993.
C. Joergensen, T. Durhuus, et al., "Optically controlled semiconductor
optical amplifier gates," ECOC 1992, WeB8.2, pp. 377-380.
K. T. Koai, R. Olshansky, and P. M. Hill, "Dual-function semiconductor
laser amplifier in a broad-band subcarrier multiplexed system," IEEE
Photon. Technol. Lett., vol. 2, pp. 926-928,1990.
K T. Koai and R. Olshansky, "Simultaneous optical amplification,
detection, and transmission using in-line semiconductor laser
amplifiers," IEEE Photon. Technol. Lett., vol. 4, no. 5, pp. 441-443, May
1992.
S. Kobayashi and T. Kimura, "Semiconductor optical amplifers," IEEE
Spectrum, pp. 26-31, May 1984.
S. Kitamura, K Komatsu, and M. Kitamura, "Polarization-insensitive
semiconductor optical amplifier array grown by selective MOVPE,"
IEEE Photon. Technol. Lett., vol. 6, no. 2, pp. 173-175, Feb. 1994.
K C. Lee and V. O. K Li, "A wavelength-convertible optical network,"
IEEE/O SA J. Lightwave Technol., vol. 11, no. 5/6, pp. 962-970, May/June
1993.
C.-L. Lu, D. J. M. Sabido IX, P. Poggiolini, R. T. Hofmeister, and L. G.
Kazovsky, “CORD--A WDMA optical network: Subcarrier-based
signaling and control scheme,” IEEE Photon. Technol. Lett., vol. 7, no. 5,
pp. 555-557, May 1995.
D. Mahgerefteh and M. W. Chbat, "All-optical 1.5 pm to 1.3 pm
wavelength conversion in a walk-off compensating nonlinear optical loop
mirror," IEEE Photon. Technol. Lett., vol. 7, no. 5, pp. 497-499, May
1995.
103
R. J. Manning, D. A. 0. Davies, J. K. Lucek, "Recovery rates in
semiconductor-laser amplifiers - optical and electrical bias
dependencies," Electron. Lett., vol. 30, no. 15, pp. 1233-1235, July 1994.
S. B. Mikkelsen, R. J. S. Pedersen, T. Durhuus, C. Braagaard, C.
Joergensen and K.E.Stubkjaer, "Wavelength conversion of high speed
data signals," Electron. Lett., vol. 29, no. 19, pp. 1716-1718, Sept. 1993.
B. Mikkelsen, R. J. S. Pedersen, M. Nissov, H. N. Poulsen, C. Joergensen,
S. L. Danielsen, K. E. Stubkjaer, M. Gustavsson, W. van Berio, and M.
Janson, "Transmission through 10-all-optical interferometric
wavelength converter spans at 10 Gb/s," Opt. Fiber Commun. Conf., OFC
‘96, San Jose, CA, postdeadline paper PD13.
T. Miki, "Optical transport networks," Proc. IEEE., vol. 81, no. 11, pp.
1594-1609, Nov. 1993.
T. Morioka, S. Kawanishi, and M. Saruwatari, "Tunable error-free
optical frequency conversion of a 4ps optical short pulse over 25nm by
four-wave mixing in a polarisation-maintaining optical fiber," Electron.
Lett., vol. 30, no. 11, pp. 884-885, May 1994.
T. Mukai, K. Inoue, and T. Saitoh, "Signal gain saturation in two-channel
common amplification using a 1.5mm InGaAsP traveling-wave laser
amplifier," Electron. Lett., vol. 29, no. 8, pp. 396-397, Apr. 1987.
T. Mukai, K. Inoue, and T. Saitoh, "Homogeneous gain saturation in 1.5
pm InGaAsP traveling-wave semiconductor laser amplifiers," Appl.
Phys. Lett., vol. 51, no. 6, pp. 381-383, Aug. 1987.
H. Nakano, S. Tsuji, S. Sasaki, K Uomi, and K Yamashita, "10-Gb/s 4-
channel wavelength division multiplexing fiber transm ission using
semiconductor optical amplifier modules," IEEE/O SA J. Lightwave
Technol., vol. 11, no. 4, pp. 612-618, Apr. 1993.
M-A. Newkirk, B. I. Miller, U. Koren, M. G. Young, M. Chien, R M.
Jopson, G. Raybon, C. A. Burrus, and H. M. Presby, "1.55-pm
multiquantum-well semiconductor optical amplifier with low gain ripple
and high coupling efficiency for photonic circuit integration," Electron.
Lett., vol. 29, no. 5, pp. 443-444, Mar. 1993.
104
D. Norte and A. E. Willner, “Simultaneous probe and pump extinction
ratio enhancement demonstration in all-optical noninverted wavelength
shifting,” Opt. Fiber Commun. Conf., OFC ‘96, San Jose, CA, paper WM7,
pp. 188-190.
D. Norte and A.E. Willner, "Experimental demonstrations of all-optical
conversions between the RZ and NRZ data formats incorporating
noninverting wavelength shifting leading to format transparency," to
appear in IEEE Photon. Technol. Lett.
M. G. Oberg and N. A. Olsson, "Wavelength dependence of noise figure of
a travelling-wave GalnAsP/InP laser amplifier," Electron. Lett., vol. 24,
no. 2, pp. 99-100, Jan. 1988.
N. A Olsson, "Semiconductor optical amplifiers," Proc. of the IEEE, vol.
80 no. 3, pp. 375-382, Mar. 1992.
N. A. Olsson, R. F. Kazarinov, W. A Nordland, C. H. Henry, M. G. Oberg,
H. G. White, P. A Garbinski, and A. Savage, "Polarisation-independent
optical amplifier with buried facets," Electron. Lett., vol. 25, no. 16, pp.
1048-1049, Aug. 1989.
H. Onaka, H. Miyata, G. Ishikawa, K. Otsuka, H. Ooi, Y. Kai, S.
Kinoshita, M. Seino, H. Nishimoto, and T. Chikama, "1.1 Tb/s WDM
transm ission over a 150 km 1.3 pm zero-dispersion single-mode fiber,"
Opt. Fiber Commun. Conf., OFC ‘96, San Jose, CA, postdeadline paper
PD19.
M. J. O'Mahony, "Semiconductor laser optical amplifiers for use in future
fiber systems," IEEE/O SA J. Lightwave Technol., vol. 6, no. 4, pp. 531-
544, Apr. 1988.
E. Park and A. E. Willner, "Simultaneous all-optical packet-header
replacement and wavelength shifting for a dynamically-reconfigurable
WDM network," OFC '95, TuQ4, San Diego, CA, Feb. 1995; and IEEE
Photon. Technol. Lett., vol. 7, no. 7, pp. 810-812, July 1995.
E. Park and A.E. Willner, "Self-Routing of wavelength packets using an
all-optical wavelength shifter and QPSK subcarrier routing control
headers," IEEE Photon. Technol. Lett., July 1996.
105
J. S. Perino, J. M. Wiesenfeld, and B. Glance, "Fiber transmission of
lOGbit/s signals following wavelength conversion using a traveling-wave
semiconductor optical amplifier," Electron. Lett., vol. 29, no. 25, pp. 256-
258.
J. S. Perino and J. M. Wiesenfeld, "Linewidth enhancement factor and
chirp for high bit rate semiconductor optical amplifier wavelength
converter," Conf. on Lasers and Electro-Optics (CLEO), 1994, paper
CThEl, pp. 298-299.
K. A. Rauschenbach, K L. Hall, J. C. Livas, and G. Raybon, "All-optical
pulse width and wavelength conversion at 10 Gb/s using a nonlinear
optical loop mirror," IEEE Photon. Technol. Lett., vol. 6, no. 9, pp. 1130-
1132, Sept. 1995.
J. Riishoj, T. N. Nielsen, U. Gliese, K. E. Stubkjaer, "3-Gbit/s 2-level to
1.5 Gsymbol/s 4-level converter GaAs IC for 2 electrode semiconductor
optical amplifier modulators," Electron. Lett., vol. 29, no. 14, pp. 1173-
1174, July 1993.
A. A. M. Saleh, "Transparent optical networks for the next-generation
information infrastructure," Opt. Fiber Commun. Conf., OFC ‘95, paper
ThEl, pp. 241, San Diego, CA.
A. A. M. Saleh, "Overview of MONET, multiwavelength optical
networking, program," Opt. Fiber Commun. Conf., OFC ‘96, paper ThI2,
pp. 240, San Jose, CA.
K. Sato, S. Okamoto, and H. Hadama, "Network performance and
integrity enhancement with optical path layer technologies," IEEE J.
Select. Areas Commun., vol. 12, no. 1, pp. 159-170, Jan. 1994.
R. Schnabel, W. Peiper, et al., "Wavelength conversion and switching of
high speed data signals using semiconductor laser amplifiers," Electron.
Lett., vol. 29, no. 23, pp. 2047-2048.
W. Shieh and A. E. Willner, "Demonstration of output-port contention in
a 2x2 WDM switching node based on all-optical wavelength shifting and
subcarrier-multiplexed routing-control headers," Opt. Fiber Commun.
Conf., OFC '96, San Jose, CA, post-deadline paper PD36.
106
J. C. Simon, L. Lablonde, I. Valiente, L. Billes, and P. Lamouler, "Two-
stage wavelength converter with improved extinction ratio," Opt. Fiber
Commun. Conf., OFC ‘95, San Diego, CA, post-deadline paper PD15.
J. C. Simon, "GalnAsP semiconductor laser amplifiers for single-mode
fiber communications," IEEE/O SA J. Lightwave Technol., vol. LT-5, no.
9, pp. 1286-1295, Sept. 1987.
J. Spring, R. M. Fortenberry, and R. S. Tucker, "Photonic header
replacement for packet switching," Electron. Lett., vol. 29, no. 17, pp.
1523-1525, 1993.
J. Spring and R. S. Tucker, "Photonic 2x2 packet switch with input
buffers," Electron. Lett., vol. 29, no. 3, pp. 284-285, Feb. 1993.
S. F. Su, A. R. Bugos, V. Lanzisera, and R. Olshansky, "Demonstration of
a multiple-access WDM network with subcarrier-multiplexed control
channels," IEEE Photon. Technol. Lett., vol. 6, no. 3, pp. 461-463, Mar.
1994.
H. Toba, K. Oda, K. Nakanishi, N. Shibata, K. Nosu, N. Takato, and M.
Fukuda, "A 100-channel optical FDM transmission/ distribution at 622
Mb/s over 50 km," IEEE/O SA J. Lightwave Technol., vol. 8, no. 9, pp.
1396-1401, 1990.
I. Valiente, J. C. Simon, and M. Le Ligne, "Theoretical analysis of
semiconductor optical amplifier wavelength shifter," Electron. Lett., vol.
29, no. 5, pp. 502-503, Mar. 1993.
T. J. Whitley, "A review of recent system demonstrations incorporating
1.3-|jm praseodymium-doped fluoride fiber amplifiers," IEEE/O SA J.
Lightwave Technol., vol. 13, no. 5, pp. 744-760,1990.
J. M. Wiesenfeld and B. Glance, "Cascadability and fanout of
semiconductor optical amplifier wavelength shifter," IEEE Photon.
Technol. Lett., vol. 4, no. 10, pp. 1168-1171, Oct. 1992.
J. M. Wiesenfeld, B. Glance, J. S. Perino, and A. H. Gnauck, "Wavelength
conversion a t 10 Gb/s using a semiconductor optical amplifier," IEEE
Photon. Technol. Lett., vol. 5, no. 11, pp. 1300-1303, Nov. 1993.
107
J. M. Wiesenfeld, J. S. Perino, A. H. Gnauck, and B. Glance, "Bit error
rate performance for wavelength conversion at 20 Gbit/s," Electron. Lett.,
vol. 30, no. 9, pp. 720-721, Apr. 1994.
A. E. Willner and W. Shieh, "Optimal spectral and power param eters
for all-optical wavelength-shifting: single stage, fanout, and
cascadability," OFC '94, ThC4; and IEEE/O SA J. Lightwave Technol.,
vol. 13, no. 5, pp. 771-781, May 1995.
C. E. Zah, J. S. Osinski, C. Caneau, S. G. Menocal, L. A. Reith, J.
Salzman, F. K. Shokoohi, and T. P. Lee, Electron. Lett., vol. 23, pp. 990,
1987.
J. Zhou, N. Park, J. W. Dawson, K J. Vahala, M. A. Newkirk, and B. I.
Miller, "Efficiency of broad-band four-wave mixing wavelength
conversion using semiconductor traveling-wave amplifiers," IEEE
Photon. Technol. Lett., vol. 6, no. 1, pp. 50-52, Jan. 1994.
108
Appendix A
New Header Generation for
Simultaneous Wavelength Shifting and
Header Replacem ent (See Chapter 4)
Due to limited resources, the new header was derived from the
packet pattern generator and gated using the logic circuit shown in
Figure 4.4. The inverted new header is embedded in the transm itted
packet as shown below.
Transmitted Packet (from pattern generator non-lnverted output)
, Payload ,
} f ln n i- m p m irmmn(finiiiiiiinif|nr^|-
Rag ^ d e r 27-1 PRBS Padding Bits ^ 2 7-1PRBS
-208 ns Delay-
New Header Generation
Delayed Inverted Packet w/ embedded New Header
(from pattern generator Inverted output)
} . . . . . . { ■ jn n riTTTniiTTiff n r
Embedded New Header
New
Header
Gating Signal (derived from pattern generator clock)
8 ns I
iiiiiflnnnT m m ifj i
4 ----------------------------------------------- 416 ns-
Three-Level Output to Probe Laaer
{ I — d - . . — |
I Three- I I I Three-
Level
Signal
109
Appendix B
Self-Routing Packet Structure (See Ch. 5 8(6)
P ack et stru ctu re d eta ils for th e dem onstration in C hapter 5:
128 128 128
Btoj 27-1 PRBS J 27-1 PRBS J 27-1 PRBS j
t i l l
-------------------- J-------------------- 1 ------------------- L — S y
Data Pattern Generator Output, 1 Gb/s
460 Bits (-57 Bytes)
Baseband Packet #1 Baseband Packet #2 Baseband Packet #3 Baseband Packet #4
•
480 ns 480 ns 480 ns 480 ns
Control Pattern Generator, 50 Mb/s
01111110 11000111 01111110 11011111 01111110 11010111
Header #1 Header#4 Header#2
01111110 11001111
Header#3
320 ns 320 ns 320 ns 320 ns
/
\
r y x X-Path X
8-Bit Rag j 8-Bit Header
0 0 1 1540 nm
0 1 2 1542
1 0 3 1548
1 1 4 1552
^ J
0 1 1 1 1 1 1 0 i t O iyl 1 1
16 Bits
P acket stru ctu re d eta ils for th e dem onstration in C hapter 6:
128 128 128
2^-1 PRBS 27-1 PRBS 27-1 PRBS |* s G uard
Bits
456 Bits (57 Bytes)
Data Pattern Generator Output, 1 Gb/s
Baseband Packet #2 Baseband Packet #4 Baseband Packet #1 Baseband Packet #3
480 ns 480 ns 480 ns 480 ns
Control Pattern Generator, 50 Mb/s
01111110 11000111 01111110 11010111
Header#1
320 ns
Header «2
320 ns
110

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Experimental demonstration of optical router and signal processing functions in dynamically reconfigurable wavelength-division-multiplexed fiber -optic networks

PDF

Experimental demonstrations of optical techniques to implement networking functions in dynamically reconfigurable wavelength-division-multiplexed fiber -optic communication networks

PDF

Experimental demonstrations of all -optical networking functions for WDM optical networks

PDF

A monolithic multichannel receiver array for gigabit/s parallel synchronous optical digital data links

$A new signal flow matrix fraction description technique for computer-aided frequency domain analysis of linear active networks$

PDF

A new signal flow matrix fraction description technique for computer-aided frequency domain analysis of linear active networks

PDF

Dispersive and nonlinear effects in high-speed reconfigurable WDM optical fiber communication systems

PDF

Experimental demonstration of techniques to improve system performance in non-static microwave frequency analog and digital signal transmission over fiber -optic communication systems

PDF

Experimental demonstration of techniques to improve system performance in fiber -optic communication systems using subcarrier -multiplexed and digital baseband signals

PDF

Indium Gallium Arsenic Phosphide-Based Optoelectronics Grown By Gas Source Molecular Beam Epitaxy

PDF

Compact VLSI array processors design for multimedia applications

PDF

A precision receiver for CDMA communication using neural-based array processors

PDF

Dispersive properties in photonic crystals applications to beam-steering and wavelength demultiplexing

PDF

All-optical devices based on carrier nonlinearities for optical filtering and spectral equalization

PDF

Generation, filtering, and application of subcarriers in optical communication systems

PDF

Gyrator-based synthesis of active inductances and their applications in radio -frequency integrated circuits

PDF

Experimental demonstrations of wavelength routing functions for an all-optical wavelength-division-multiplexed fiber-optic communication network switching node

PDF

A study of growth mechanisms and electrical and optical properties of epitaxial AlₓGa₁₋ₓN layers grown by atmospheric pressure metalorganic chemical vapor deposition

PDF

Quasiepitaxial Growth Of Archetype Organic Molecules: Modeling And Observation

PDF

High performance components of free -space optical and fiber -optic communications systems

PDF

Chromatic and polarization mode dispersion monitoring for equalization in optical fiber communication systems

Asset Metadata

Creator Park, Eugene (author)

Core Title Applications of all-optical wavelength shifting using semiconductor optical amplifiers for switching and routing functions in a dynamically reconfigurable wavelength-division-multiplexed fiber-op...

Contributor Digitized by ProQuest (provenance)

Degree Doctor of Philosophy

Degree Program Electrical Engineering

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag engineering, electronics and electrical,OAI-PMH Harvest

Language English

Advisor Willner, Alan E. (committee chair), Dapkus, Daniel (committee member), Gershenzon, Murray (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c17-101064

Unique identifier UC11354534

Identifier 9636739.pdf (filename),usctheses-c17-101064 (legacy record id)

Legacy Identifier 9636739.pdf

Dmrecord 101064

Document Type Dissertation

Rights Park, Eugene

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA