close

Вход

Забыли?

вход по аккаунту

?

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

код для вставкиСкачать
EXPERIMENTAL DEMONSTRATION OF TECHNIQUES TO IMPROVE
SYSTEM PERFORMANCE IN NON-STATIC MICROWAVE FREQUENCY
ANALOG AND DIGITAL SIGNAL TRANSMISSION OVER FIBER-OPTIC
COMMUNICATION SYSTEMS
by
Asaf Sahin
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)
August 2003
Copyright 2003
Asaf Sahin
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI N um ber: 3116779
INFORMATION TO USERS
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 bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send 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.
®
UMI
UMI Microform 3116779
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90089-1695
This dissertation, written by
8.
SA U T l\i _____________________________
under the direction o f h t 5
dissertation committee, and
approved by a ll its members, has been presented to and
accepted by the D irector o f Graduate and Professional
Programs, in partial fulfillment o f the requirements fo r the
degree o f
DOCTOR OF PHILOSOPHY
D irector
Date
7
Dissertation Committee
llo ii f p r
DANIEL
A u g u s t 12 f 2 0 0 3
LLflRDi
iZ£CM
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dedication
To my mother and brother for their trust and support.
And to the memory of my dear father.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgements
I would like to thank my advisor and dissertation committee chairman Prof. Alan E.
Willner, for his invaluable attention, guidance, and support in my graduate studies.
Furthermore, I wish to thank Prof. John D. O’Brien, Prof. Alexander Sawchuk, Prof.
Daniel Rich, and Prof. Robert Gagliardi for serving on my dissertation and
qualifying committee.
I would like to thank my colleagues and friends from OCLab. My whole endeavor
would have been impossible without their knowledge and help.
First the ones, who have already made it: Dr. Mustafa C. Cardakli, Dr. Steve A.
Havstad, Dr. Olaf H. Adamczyk, Dr. Reza Khosravani, Dr. Sangeon Lee, Dr. Yong
Xie.
My generation, crop of 2003: Dr. Zhongqi Pan, Yon-Won Song, Deniz Gurkan.
And my professors at Middle East Technical University (METU), who have taught
and inspired me to not to loose hope.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table o f Contents
Dedication
_____________________________________________________ ii
Acknowledgements_______________________________________________ iii
List of Figures
___________________
vii
Abstract
xii
Chapter 1
1
Introduction and background_________________________________________ 1
1.1
Non-static fiber-optic communication links
__________________ 3
1.2
Subcarrier multiplexing in optical communication systems___________ 5
1.3
System issues in analog and digital fiber-optic links_____________
1.3.1 Subcarrier Based All Optical Switching___________________
8
9
1.3.2 RF fading____________________________________________ 10
1.3.3 Nonideal fiber characteristics_____________________________ 12
1.3.4 Chromatic dispersion___________________________________ 12
1.3.5 Nonlinearities
Chapter 2
15
__________________ _________________________________ 18
Inline Dispersion Slope Monitoring of Many WDM Channels Using
Dispersion-Induced RF Clock Regeneration_________________ 18
2.1 Introduction________________________________
18
2.2 Concept for Dispersion Slope Monitoring Using Dispersion-Induced
RF Clock Regeneration
__________________________________ 20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.3 Experimental Setup_________________________________________ 22
2.4 Results__________________________________________________ 24
2.5 Conclusion
Chapter 3
25
_________________________________________________
26
Distance-independent RF fading compensation using a tunable nonlinearlychirped FBG in a phase diversity configuration______________ 26
3.1 Introduction_________
26
3.2 Concept of RF fading compensation using phase diversity__________ 28
3.3 Experimental setup of the compensation module
3.3.1
____________ 31
Nonlinearly-chirped fiber Bragg grating_____________________ 34
3.4 Experimental results________________________________________ 40
3.5 Conclusion
45
Chapter 4
46
Wavelength Conversion of Subcarrier Channels using Difference Frequency
Generation in a PPLN Waveguide^________________________ 46
4.1
Introduction______________________________________________46
4.2
Experimental Setup_______________________
4.3
Experimental Results
4.4
Conclusion
47
____________________________________ 50
54
Chapter 5
55
Statistics of PMD-induced power fading for double sideband and single
sideband subcarrier-multiplexed signals_______________________ 55
5.1 Introduction______________________________________________ 55
5.2 Experimental Setup
______________________________________ 56
5.3 Experimental Results_______________________________________ 58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
5.4
61
Conclusion
Chapter 6
62
Dispersion Division Multiplexing for In-Band Subcarrier-Header-Based AllOptical PacketSwitching_________________________________62
6.1 Introduction_______________________________________________62
6.2 Concept^______________
64
6.3 Experimental Setup
________
65
6.4 Experimental Results___________
68
6.4.1 Receiver RF Spectra______________________________________ 68
6.4.2 Data and Header Bit Error Rate Performance__________________ 69
6.5 Conclusion
Chapter 7
71
72
Bias-Induced Diversity-Detection (BIDD) Technique for Robust Transmission
of Subcarrier-Multiplexed Channels_______________________ 72
7.1 Introduction________________________________________________ 72
7.1
Concept_____________________
74
7.2 Experimental Results_______________________________________ 77
7.2.1
Data Rate Performance_______________
7.2.2
Chromatic Dispersion Induced RF Power Fading_____________ 78
7.2.3
Polarization Mode Dispersion induced Power Fading__________ 80
7.2.4
RF Intermodulation Terms______________
7.3
Conclusion
Conclusion
________________
77
82
83
84
Bibliography ____________________________________________________ 86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
Fig. 1.2:
(a) Simple point-to-point link, (b) Dynamic reconfigurable
network employing switching nodes to route the signal to its
destination.
___
Fig. 1.3:
Schematic of a typical subcarrier multiplexed lightwave system.
Fig. 1.4:
Electronic header processing at a switching node requires the
entire data stream to be detected and processed at the bit rate.______ 9
Fig. 1.5:
Optical network model that uses subcarrier header and/or labels.___ 10
Fig. 1.6:
The received RF power spectrum after transmitting a double­
sideband subcarrier multiplexed signal through single mode
fiber
Fig. 1.7:
11
Chromatic dispersion coefficient D versus wavelength of
conventional single mode fiber and dispersion shifted fiber. ______ 14
Fig. 2.1: The concept of residual dispersion slope in fiber optic links_______ 19
Fig. 2.2: Dispersion induces the regeneration of the clock frequency
component, which increases in power with increasing
dispersion. _____________________________________________ 21
Fig. 2.3:
(a) The experimental setup for monitoring the residual
dispersion slope evolution versus the transmission distance, (b)
Details of the recirculating fiber loop, (c) The dispersion slope
monitor. (TOM: Tunable optical filter, PD: Photodetector,
NBPF: clock frequency narrow band electrical filter)____________ 22
Fig. 2.4: (a) Regenerated RF clock power vs. dispersion, (b) Measured
dispersion value evolution for two DWDM channels along the
Vll
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dispersion compensated link, (c) Dispersion slope accumulation
obtained from Fig. 3(b)._____________________
24
Fig. 3.1:
Conceptual diagram of the phase diversity technique for
distance independent RF fading compensation._________________ 29
Fig. 3.2:
RF Fading compensation module detail._______
Fig. 3.3:
Experimental setup for distant-independent RF fading
________________________________________ 33
compensation
Fig. 3.4:
Conceptual diagram of a chirped fiber Bragg grating, where
different wavelengths reflect at different positions in the grating.
32
35
Fig. 3.5:
Relative time delay versus wavelength curves for (a) Linearly
chirped FBG and (b) Nonlinearly chirped FBG.________________ 36
Fig. 3.6:
A nonlinearly-chirped FBG induces a time delay for the two
optical sidebands relative to the optical carrier of a subcarrier
multiplexed signal._______________________________________ 37
Fig. 3.7:
Reflected spectrum of the nonlinearly-chirped FBG for different
applied stretching voltages to the piezoelectric transducer.________ 39
Fig. 3.8:
Time delay versus wavelength curves of the nonlinearly-chirped
FBG for different applied stretching voltages to the
piezoelectric transducer.___________________________________ 39
Fig. 3.9:
Normalized RF power versus distance for uncompensated and
compensated transmission for (a) 8 GHz, (b) 10 GHz and (c) 12
GHz subcarrier frequency. The dashed line shows the
theoretical case for RF power fading. ______________________ _41
vui
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 3.10: Required grating control voltage versus subcarrier frequency to
achieve RF fading compensation.___________________________ 42
Fig. 3.11: (a) RF power spectrum of the transmitted subcarrier multiplexed
amplitude-shift-keyed data signal, (b) Recovered bit stream
after envelope detection____________
43
Fig. 3.12: Measured BER versus received optical power for different
transmission distances of 0 km, 27.7 km and 52.4 km (data rate
= 155 Mbit/s, subcarrier frequency = 8 GHz). The insert depicts
an eye diagram of the recovered signal.__________________
44
Fig. 4.1:
Conceptual diagram of (a) wavelength conversion of subcarrier
channels, (b) DFG in a PPLN waveguide using a c(2):c(2)
process______________
48
Fig. 4.2:
Experimental setup.
Fig. 4.3:
Optical spectrum after wavelength conversion in the PPLN
waveguide._____________________________________________ 50
Fig. 4.4:
Linearity of DFG: wavelength converted signal power vs. signal
power, measured after the PPLN waveguide___________________ 51
Fig. 4.5:
RF spectra before and after the wavelength conversion.__________ 52
Fig. 4.6:
BER curves of subcarrier multiplexed channels for before and
after the wavelength conversion. 6 _ _ _________
Fig. 4.7:
________________________________ 49
53
Received optical power sensitivities for 10-9 BER vs.
wavelength spacing between the input and output data signal._____ 53
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 5.1:
First-order PMD induces a differential group delay in an optical
sideband of a SCM signal, which leads to a phase difference in
the corresponding received subcarrier signals in the
photodetector, possibly causing serious power fading.___________ 55
Fig. 5.2:
Experimental setup. (TL: tunable laser, 900: phase shifter, PC:
polarization controller, OF: optical filter, Rx: receiver)__________ 57
Fig. 5.3:
PMD induced power fading vs. DGD curves for double
sideband (DSB-SCM) and single sideband (SSB-SCM) intensity
modulation of a 7 GHz subcarrier with and without dynamic
first-order PMD compensation, (a) Measurement of 350
independent polarization samples, and (b) Simulation of 10000
independent polarization samples for each modulation format.
The solid line corresponds to the theoretical fading penalty for
equal polarization coupling into the PSP for first-order PMD
(i.e., DGD)._____________________________________________ 59
Fig. 5.4:
Measured bit error rate vs. received optical power for 155
Mbit/s DSB-SCM-BPSK and SSB-SCM-BPSK signals at 7
GHz with and without first-order compensation. The measured
DGD for both modulation formats is ~40 ps. The inserts show
an error-free recovered eye diagram after first-order
compensation.___________________________________________ 61
Fig. 6.1:
Conceptual diagram for shadow subcarrier multiplexed header
transmission and detection for header based optical switching.
64
Fig. 6.2:
Experimental setup for shadow subcarrier multiplexed header
transmission and detection for switching information.___________ 65
Fig. 6.3:
(a-c) Data receiver RF spectrum plots for the shadow SCM
only, data channel only, and combined cases, (d-f) Shadow
SCM Recovery Module RF spectrum plots for the shadow SCM
only, data channel only, and combined cases.__________________ 68
x
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Fig. 6.4:
Bit-error-rate curve plots for the effect of the shadow SCMheader channel on the 9.85 Gb/s data channel, (a) Plots for the
cases when the SCM-header channel is (i) not modulated, (ii)
PSK modulated (iii) ASK modulated, (b) Plots for the cases
when the SCM-header frequency is off from the set NL-FBG
dispersion value RF fading frequency (7.7 GHz), (c) (i)
recovered SCM header, (ii) output of the threshold detector, (iii)
10 Gb/s data stream before the switch, (iv) output of the switch
port 0, (v) output of the switch port 1.________________________ 70
Fig. 7.1:
Bias-Induced Diversity Detection (BIDD) Concept: Generation
of first and second harmonics in the RF domain by beat terms of
lower and upper SCM sidebands and optical carrier (c) Total
RF power of the different bias method._________
75
Fig. 7.2:
(a) The 1st harmonic RF power fades due to CD while the 2nd
harmonic survives, (b) The 1st and 2nd harmonics both fade due
to PMD. BIDD method results in a robust optical system, (c)
Total RF power considering both CD and PMD________________ 76
Fig. 7.3:
Sensitivity of quadrature bias, minimum bias, and BIDD
technique as a function of bit ra te ___________________________ 77
Fig. 7.4:
RF power fading as a function of fiber transmission distance______ 79
Fig. 7.5: RF power fading histograms for quadrature, minimum, and BIDD
bias for 8-GHz subcarrier _______________
Fig. 7.6:
80
RF Power fading distribution curves for different bias
techniques and 6, 7 and 8-GHz subcarriers____________________ 81
Fig. 7.7: RF Spectra for quadrature, minimum, and BIDD bias for two
subcarriers at 7 and 8-GHz. __________________
82
xi
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
The overall performance of non-static analog and digital optical communication
systems and networks may be degraded for various reasons. For instance, chromatic
dispersion of standard single mode fiber causes pulse distortion for digital
transmission systems and RF power fading in analog fiber-optic links. Polarization
mode dispersion also affects the microwave signals in a manner akin to that of
multipath fading in wireless transmission. These effects can result in unacceptable
power penalties and even complete loss of signal.
As data speeds continue to increase, latency at switching nodes due to O-E-O
conversion is becoming a major bottleneck in optical networks. Moreover, it may
become impractical and overly expensive to use electronics at each node of a
network to detect the data, process the header information, and retransmit at rates of
40 Gb/s and higher. Subcarrier multiplexed, and low bit-rate header or label signals
may prove to be the required mean to overcome these obstacles.
To combat the mentioned performance degrading effects, all-optical techniques are
highly desirable to enable high-speed on-the-fly processing, which would be
essential for future high throughput and dynamically reconfigurable optical
networks.
xii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This paper will present the following experimental demonstrations to enhance the
system performance of such optical networks: (1) doubling the usable spectral
bandwidth and number of channels in subcarrier-modulated data transmission over
optical fiber; (2) Dynamic dispersion slope monitoring for accurate and continuous
dispersion and dispersion slope compensation; (3) Distance-independent RF fading
compensation using a tunable nonlinearly-chirped fiber Bragg grating; (4)
Wavelength conversion of subcarrier channels using difference frequency generation
in a PPLN waveguide; (5) Statistics of PMD-induced power fading for double
sideband and single sideband subcarrier-multiplexed signals; (6) Dispersion Division
Multiplexing for In-Band Subcarrier-Header-Based All-Optical Packet Switching;
(7) Bias-Induced Diversity-Detection (BIDD) Technique for Robust Transmission of
Subcarrier-Multiplexed Channels.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 1
Introduction and background
Optical fiber communication systems thrive on the very high bandwidth capacity and
low-loss characteristics of single mode fiber. Fig. shows the low-loss regions of
single mode silica fiber centered around 1.3 pm and 1.55 pm. The available optical
information bandwidth in the 1.55 pm transmission window is approximately 25
THz, which is predominantly used in telecommunication systems.
Wavelength
division multiplexing (WDM) is the prime candidate to take advantage of the
immense available bandwidth of optical fiber, by spreading the transmitted
information in the wavelength domain [1], [2].
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
WDM = w avelength-division m ultiplexing
■25
1
co
3 THz
<S 0.4
ro
_Q
CD
!§
10
02
25 THz
o
WDM c h a n n e ls
1.2
1.3
1.4
1.5
1 .6
5
1.7
W a v e le n g th , f j m
Fig. 1.1 : Attenuation spectrum of single mode silica fiber. The insert shows a
typical gain spectrum for an erbium-doped fiber amplifier (EDFA).
WDM technology uses multiple optical carriers at different wavelengths multiplexed
onto the same fiber, each modulated with independent high speed data signals. The
combined throughput of several Tbit/s has been reported for such a WDM system
[3], A second very import milestone was the invention of the erbium-doped fiber
amplifier (EDFA) [4], [5], which enables lightwave transmission over trans-oceanic
distances [6],
Without EDFA's, electronic regeneration of the optical signal is
essential to overcome the loss along the transmission path. Electronic regeneration
includes optical to electrical conversion of the optical signal, followed by retiming,
reshaping and retransmission of the original information. An EDFA consists of a
short length of erbium doped fiber coupled to a pump laser operating at a wavelength
of 980 nm or 1480 nm.
The pump laser excites the erbium ions to a higher,
2
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
metastable energy state.
An optical signal, located in the 1.55 pm wavelength
region, passing through the erbium doped fiber will induce stimulated emission,
which causes the excited ions to fall back to their ground state and to generate a new
photon at the same phase and wavelength as the traversing signal.
Another
advantage of EDFA’s is the capability of amplifying multiple WDM channels
simultaneously, independent of the bit rate and data format of each signal.
1.1 Non-static fiber-optic communication links
The simplest transmission network architecture is the point-to-point link, where for
instance, point A is directly connected via a designated transmission medium with
point B and point C (see Fig. 1.1a). In an optical communication system, optical
fiber serves as the transmission medium carrying one or, in case of a WDM system,
several optical data channels. Such a simple link seems static at first, but aging and
environmental variations can alter the characteristics of the employed devices, such
as signal lasers, EDFA's and passive devices. Even small changes can result in
degradation of the signals and severely limit the overall performance of the link.
Passive compensation schemes to combat the link dynamics might not function
properly, therefore tunable techniques need to be employed, to warrant the signal
integrity and performance of the link.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Switching/Routing Node
(a)
(b)
Fig. 1.1: (a) Simple point-to-point link, (b) Dynamic reconfigurable network
employing switching nodes to route the signal to its destination.
A typical point-to-point link may interconnect local, metropolitan or wide area
networks (LAN, MAN or WAN), with a user base varying from a few tens to several
hundred-thousands. The demand for the available bandwidth may also increase;
requiring an expansion in information capacity, by boosting the data rate or assigning
additional wavelength channels. This request for more capacity should be met in a
dynamic way to allocate more bandwidth only when needed. To better utilize the
available bandwidth and installed hardware resources, a more complex network
topology is necessary. Such a topology should by dynamic and reconfigurable to
satisfy the demands and requirements of future communication systems [7]. Fig.
1.1b depicts a dynamic reconfigurable network, where switching and routing nodes
determine the data path between two points (circuit-switched or packet-switched
networks). The length and the characteristics of the data path can vary significantly,
depending on how the signal is routed to its final destination. Those variations can
4
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
potentially be more extreme than in a point-to-point link, having much larger
consequences on the channel performance. For instance, when a signal going from
point A to point B switches to another path, the accumulated chromatic dispersion
may be totally different because either the transmission distance changed or the new
path employs fiber with dissimilar amounts of chromatic dispersion. The techniques
to reduce and compensate the channel the degrading effects in such switched
networks have to be actively tunable to be able to react to the dynamically changing
environment.
1.2 Subcarrier multiplexing in optical communication systems
Wavelength division multiplexed optical communication systems utilize intensity
modulation of the light source and direct detection at the photodiode (IM-DD) to
transmit and receive information. This information can consist of digital and/or
analog signals. In a pure digital system, multiple lower speed data streams are time
division multiplexed (TDM) and the resulting baseband transmitted at the high data
rate (i.e. 10 Gbit/s). A disadvantage of TDM is that each receiver has to operate at
the high data rate, even if the user at that receiver is only interested in only a few of
the lower speed data streams. This problem is especially evident in a personal
communication access network, which serves as a fiber-optic backbone for wireless
applications [8]. In such a network the number of users is very large and may
dynamically vary over time, and also the data rates are low (<1 Gbit/s) compared to a
5
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
digital system. The technique of subcarrier multiplexing (SCM) can be used to
better utilize the available modulation bandwidth [9], [10], [11].
A SCM fiber-optic communication link employs several modulated microwave
subcarriers transmitted over single mode fiber.
A diagram of a typical SCM
lightwave system is shown in Fig. 1.2. The analog or digital data is electrically upconverted onto the microwave subcarrier frequency using a microwave mixer.
Several modulated subcarriers are combined together and this composite signal can
either intensity modulate a semiconductor laser or externally modulate a cw laser
source utilizing a Mach-Zehnder modulator. After transmission through single mode
fiber, the signal is directly detected with a photodiode and pre-amplified. If only a
single channel needs to be recovered, a tunable oscillator and microwave mixer can
simultaneously select the SCM channel and down-convert it to baseband. Note that
there is usually at least another conversion step to an intermediate frequency
involved, but it is omitted for simplicity.
Also, several SCM channels can be
demodulated at the same receiver by using multiple down-conversion mixers driven
with the appropriate subcarrier frequency. In addition, the up- and down-conversion
is not limited to the RF domain.
It can also be accomplished with all-optical
techniques, by using external modulators as optical mixers.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Analog or
Digital Data
<
5Hx)
VCO
f
Combiner
Recovered
Signal
SMF
Laser
Photo
Detector
,
.. .
Low Noise
Amplifier
Mixer
Fig. 1.2 : Schematic of a typical subcarrier multiplexed lightwave system.
Typical applications for fiber-optic SCM systems include cable television (CATV)
and video on demand, antenna remoting and the fiber backbone in a wireless
network, and multiuser, interactive LANs.
Furthermore, SCM signals can be
combined with baseband data, if the lowest subcarrier frequency is located beyond
the baseband data. It is useful for separating baseband payload data and low-speed
control signals with routing and timing information [12], [13]. This enables the
independent detection and processing of the SCM control channels from the
baseband data.
Employing SCM in fiber-optic links has various key advantages, such as utilizing the
large available bandwidth of lasers, modulators and detectors, without being limited
by the bandwidth of TDM electronic circuitry. Dividing the broad power spectrum
into a group of narrowband channels allows the receiver to operate at much lower
7
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
data rate, which increases the receiver sensitivity and lessens the high-speed
requirements in the processing electronics [14]. On the other hand SCM has several
disadvantages, the most important being the issue of source nonlinearity.
This
especially poses a problem when many subcarriers are transmitted from a single
source. Secondly, in digital systems SCM requires a larger bandwidth per channel
than TDM.
Furthermore, despite operating at lower channel data rates, SCM
systems perform at much higher frequencies in the millimeter and microwave
regions. It is basically a tradeoff between the possibility of accessing more of the
laser and detector bandwidth with SCM systems and occupying a larger bandwidth
portion to transmit the same amount of information.
1.3 System issues in analog and digital fiber-optic links
Digital as well as analog signals are been used extensively in fiber-optic transmission
links. Certain effects in optical communication systems, such as chromatic
dispersion can severely degrade the signal integrity and ability for error free recovery
of the transmitted data. Furthermore, network specific problems such as output-port
contention in switching and routing nodes can reduce system performance and limit
the throughput in dynamically reconfigurable optical networks. The following part
will address particular issues in analog and digital optical communication systems
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.3.1 Subcarrier Based All Optical Switching
As data speeds continue to increase, latency at switching nodes due to O-E-O
conversion is becoming a major bottleneck in optical networks. Moreover, it may
become impractical and overly expensive to use electronics at each node of a
network to detect the data, process the header information, and retransmit at rates of
40 Gb/s and higher (Fig. 1.4).
Fiber
IllATAlllHljilA T A i^ l
Q delay
H: Header
>
> Optical
t Switch J
IdataBB
DATA
Electronic
Header
Recognition
Fig. 1.3 : Electronic header processing at a switching node requires the entire data
stream to be detected and processed at the bit rate.
Electronic processing poses the additional problems of being modulation-format and
bit-rate dependent. Therefore, there is great interest in finding a way to allow a flow
of data to traverse the network with little or no header processing until it reaches the
network edge. Adding a subcarrier to each WDM channel that contains the routing
information for that channel is one of the more promising means for enabling low
speed header processing without having to detect the high speed data (Fig. 1.5). The
subcarrier may contain simple Mb/s data that is easily received and detected from a
tapped off data stream at a networking node. The subcarrier information is
independent of the bit-rate and data format, so it is easy to upgrade the data formats
9
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
used in the network without having to ehange the header processing modules. The
data can be RZ, NRZ, solitons, etc.
User G roup
Si m pi e L a b e l S u b c a r r i e r s u s e d
for fast routing th r o u g h the core
O ptical Access Node
C o re O ptics
o \(
oxe
xc
(A » M
O p tic a l Access Node
O ptical
.Terminal
n d I Fser s
Detail ed S u b c a r r i e r s used for
r o uting at the n e tw o r k edge
M e tr o N etw o r k
Fig. 1.4: Optical network model that uses subcarrier header and/or labels.
1.3.2 RF fading
An increasing number of applications require transmission of analog or digital
subcarrier-multiplexed (SCM) RF channels over fiber. The millimeter- and micro­
wave frequency bands offer adequate bandwidth for future high-capacity broadband
wireless and local area networks (LAN). One application would be the efficient
distribution of millimeter-wave signals from a central office to remote antennas via a
fiber-optic link. However, fiber induced distortions can ultimately limit the
performance of such a SCM lightwave system [12], especially fiber's chromatic
dispersion can critically degrade a channels signals integrity [13]. In case of a
transmitted double-sideband (DSB) signal, this frequency-dependent fiber dispersion
produces a deleterious time delay between the two transmitted sidebands in the
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
optical domain, causing serious RF power fading that is a function of subcarrier
frequency, fiber distance, and accumulated dispersion [14].
Fig. 1.5: The received RF power spectrum after transmitting a double-sideband
subcarrier multiplexed signal through single mode fiber
Fig. 1.6 shows a received RF power spectrum versus the transmitted distance for a
DSB-SCM signal. It can be seen, that the RF power vanishes periodically over the
transmission distance, due to chromatic dispersion.
Subcarrier
RF Power
Single Mode Fiber
Distance
Fig. 1.5: The received RF power spectrum after transmitting a double-sideband
subcarrier multiplexed signal through single mode fiber
Unfortunately, many potential applications involve reconfigurable optical paths,
where RF fading dynamically changes with transmission distance. For robust RFbased optical systems, distance-independent RF fading compensation is highly
desirable. An experimental demonstration for such a distance-independent
compensation scheme will be presented in chapter 4 [32].
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1.3.3 Nonideal fiber characteristics
Silica fiber exhibits several nonideal characteristics, which can limit the performance
of optical communication systems. Besides the attenuation spectrum of single mode
fiber (shown in Fig. 1.1), chromatic dispersion and the fiber’s nonlinear behavior can
cause severe restrictions in lightwave transmission systems.
1.3.4 Chromatic dispersion
Fiber’s chromatic dispersion emerged as a major limiting factor in optical
communication systems, after the advent of the EDFA successfully neutralized fiber
loss. Chromatic dispersion, or group velocity dispersion, is caused by the frequency
dependence of the refractive index n(co) of the dielectric waveguide, i.e. silica fiber.
This leads to a variation in the group velocity of the spectral components of the
optical signal with each frequency traveling at a slightly different speed through the
fiber, therefore causing pulse broadening of the lightwave signal. Consequently, this
signal distortion can severely impair the system performance.
The effects of fiber dispersion can be expressed by expanding the mode-propagation
constant (3 in a Taylor series around the center frequency w0 [16]:
( 1. 1)
fi(<o) = n(a>)~ = /30 +/3l (6>-6>0) + ^-j32(6)-co0) 2 + .......
c
2
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
where
CO = (0Q
The pulse envelope travels at the group velocity Vg= l / 0 i , while the coefficient 02 is
responsible for pulse broadening. The wavelength for which 02 = 0 is often referred
to as the zero-dispersion wavelength Xo . Conventional single mode fiber (SMF) has
a zero-dispersion wavelength Xq= 1.3 pm. A more commonly used system parameter
for chromatic dispersion is the group velocity dispersion coefficient D and it can be
expressed as:
( 1.2)
dX vg
dX
dX d a
£
with c being the speed of light. For SMF, the dispersion coefficient D is about +17
ps/(nm*km) in the 1.55 pm wavelength region. In general, the dispersion parameters
can be changed by tailoring the waveguide profile of the fiber. For instance, such a
specialty fiber with a smaller mode field parameter, known as dispersion shifted fiber
(DSF), has a zero dispersion wavelength A,o« 1.5 pm and D varies between -2.5 and
+2.5 ps/(nm*km) around A,<>. Fig. 1.7 depicts the dispersion parameter D as a function
of wavelength for conventional SMF and DSF.
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
20
C onventional
- - 17
Dispersion Shifted
1.1
W avelength (pm)
1.2 /1 .3
Slope = 0.08 ps/km-nm2
S
-10
-20
Fig. 1.6 : Chromatic dispersion coefficient D versus wavelength of conventional
single mode fiber and dispersion shifted fiber.
The dispersion of optical fiber can vary significantly for different wavelength
channels over the EDFA gain bandwidth of ~3 THz This wavelength dependence of
chromatic dispersion is known as the dispersion slope (dD/dX) or second-order
dispersion. A typical value for the dispersion slope for SMF and DSF is around 0.08
ps/(nm2*km).
As mentioned before, dispersion induces pulse broadening in a modulated optical
signal, which in turn can cause inter-symbol interference (ISI). This interference
imposes a limitation on the maximum transmission distance without the need of
regenerating the original optical signal. The maximum dispersion limited distance
can be approximated by determining the transmission distance after which a pulse
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
has been broadened by one bit time. For an intensity modulated signal, carrying non
return to zero (NRZ) data, the estimated dispersion limited distance Ld is given by
[17]:
ft
^
T
1
c
L° ~ BDAZ = B 2DA2
where B is the data rate, AX is the spectral width and X is the wavelength of the
optical signal. The second term for Ld in equation (1.3) is valid for single mode
lasers with a relatively large spectral width AX compared to the modulation data rate
B. The dispersion limited distance Ld is approximately 70 km for transmitting 10
Gbit/s NRZ data over SMF at an optical carrier wavelength of 1.55 pm, therefore
requiring a dispersion compensation scheme for long distance optical systems. Such
a compensation scheme should be tunable [18], to effectively correct for different
amounts of accumulated dispersion occurring in dynamically reconfigurable optical
networks.
1.3.5 Nonlinearities
The use of dispersion shifted fiber can minimize the consequences of chromatic
dispersion on an optical signal along a transmission link, but various nonlinear
effects can also accumulate and cause severe limitations. Under conditions of high
optical power and long interaction length, the nonlinear behavior of silica fiber can
produce degrading effects such as attenuation, distortion and cross channel
interference. This can place constraints on the channel spacing, the maximum
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
allowed optical power of each channel and may also limit the maximum data rate in
WDM systems. To complete the discussion of fiber characteristics, a qualitative
overview of nonlinearities in fiber is given. Four basic nonlinearities exist in fiber
[19]: (1) Stimulated Raman scattering (SRS), (2) Stimulated Brillouin scattering
(SBS), (3) Self- and cross-phase modulation (SPM and XPM) and (4) Four wave
mixing (FWM).
The first two nonlinearities are caused by stimulated scattering effects within the
transmission medium, which manifests itself in intensity dependent gain or loss. SRS
transfers a small fraction of optical power from one channel at a shorter wavelength
to longer wavelength channels. The channels interact with each other through a
vibrational wave as they propagate in the forward direction through the fiber. The
shorter wavelength channels experience loss, while the longer wavelength channels
gain power. SRS can impose limitations on the maximum allowed number of
channels in long-haul systems. SBS can also cause a frequency shift of an optical
signal, due to interaction with an acoustical wave. A backwards traveling optical
wave is generated, which causes optical power degradation in the affected channel.
SBS is mostly critical for lightwaves with very narrow spectral widths (<20 MHz).
The last two nonlinear effects derive in general from the nonlinearity of the index of
refraction in silica fiber, which is dependent on the intensity of the optical signal.
SPM is caused by fluctuations in the power of an optical signal present in the fiber
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
and results in variations of the phase of the same signal. This self modulation can
temporally disperse the optical pulse and may lead to spectral broadening.
Furthermore, high values of dispersion can exaggerate the nonlinear effect induced
by SPM. XPM, on the other hand, is an interaction, via the nonlinear refractive
index, between the intensity of one optical signal and the phase of the other signals
propagating on different wavelengths in a multiple channel WDM system. XPM can
cause asymmetric spectral broadening and, combined with SPM and dispersion,
affect the pulse shape in the time domain. FWM is the result of the nonlinear
waveguide medium causing two propagating optical waves to beat with each other.
This generates new tones as sidebands at the difference frequencies, which can
interfere with other channels located at those frequencies. Note, that the FWM
products occur only for closely spaced channels located at the zero dispersion
wavelength Xo of the fiber.
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2
Inline Dispersion Slope Monitoring of Many WDM
Channels Using Dispersion-Induced RF Clock
Regeneration
2.1 Introduction
The two trends leading to the continued growth in capacity for optical systems are
higher channel bit rates and more parallel wavelength-division-multiplexed (WDM)
channels. For 310-Gbit/s channel rates in any WDM system for links >150 km,
periodic compensation of transmission-fiber-induced chromatic dispersion is
required. Such compensation is typically accomplished by using a lumped element
that produces negative dispersion, and the most common deployed type is dispersion
compensating fiber (DCF).
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
£
c
Increasing Residual
e0)
a
CO
Q
jo
3
E
3
O
o
<
Decreasing Residual
Dispersion
Fig. 2.1 : The concept of residual dispersion slope in fiber optic links
Unfortunately, all deployed fiber types produce a wavelength-dependent dispersion
value such that each WDM channel will accumulate a different amount of dispersion
as in Fig. 2.1. Moreover, the wavelength-dependent slope of DCF's negative
dispersion does not inversely match the wavelength-dependent slope of transmission
fiber's positive dispersion [36,37]. Add to this scenario the fact that many fiber links
are not composed of only one type of transmission fiber end-to-end, and the
challenge to know the residual dispersion slope value becomes even greater.
In general, the dispersion slope among the WDM channels must be compensated
periodically along the link, and this compensation is dependent on data rate, fiber
type, and overall system wavelength range [38]. An important constraint in this
problem is that the WDM dispersion slope may change over time due to: (i) fiber
plant replacement and upgrade, (ii) reconfigurable optical networking [39], (iii) fiber
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
aging, and (iv) temperature changes [40], especially for 40-Gbit/s systems.
Therefore, dispersion-slope compensation, either in-line or at the receiver, must
include some form of dynamic monitoring in order to ensure precise dispersion
compensation.
2.2 Concept for Dispersion Slope Monitoring Using DispersionInduced RF Clock Regeneration
We demonstrate a scheme for dynamic dispersion slope monitoring of many WDM
channels [35]. Specifically, we use the power of the RF clock that is regenerated at
the detection of an NRZ signal due to the residual chromatic dispersion of the fiber
link [41]. The RF regenerated clocks of t wo widely seperated WDM channels are
monitored in real time, producing an accurate measurement of the residual WDM
dispersion slope.
We measured the dispersion slope for a 3.5-nm WDM system covering an 800-km
link at intervals of 100 km. The deviation of the experimental results from the actual
fiber slope values is <5%, which corresponds to a negligible system power penalty
even at 40 Gbit/s. This technique can be easily modified to accommodate RZ signals
for which the RF clock degrades with an increase in accumulated dispersion.
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Back to Back
-76.17dBm
9.96 GHz
20 km SMF
-50.50dBm
9.96 GHz
40 km SMF
-40.50dBm
9.96 GHz
55 km SMF
-38.67dBm
9.96 GHz
RF Requency (GHz)
Fig. 2.2: Dispersion induces the regeneration of the clock frequency component,
which increases in power with increasing dispersion.
When an NRZ format data is modulated on an optical channel and transmitted over
back-to-back or zero residual dispersion link, the RF spectrum of the detected signal
has a negligible clock frequency component. As the residual dispersion of the link
increases, we see that the mentioned clock component regenerates, at a positive
nonlinear rate dependent on the net dispersion value (Fig. 3.2). From the power of
the regenerated clock component, we can determine the residual dispersion for that
optical channel.
In order to effectively calculate the net or residual dispersion slope that the WDM
channels experience, we need two or more dispersion vs. wavelength points. To this
end, we utilize the calculation of dispersion value from the regenerated clock
component method. We need to monitor only two WDM channels. Our proposed
monitoring module dynamically measures the dispersion induced regenerated clock
components of two WDM channels, and calculates the dispersion slope value for the
WDM system.
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.3 Experimental Setup
Our experimental setup is depicted in Fig. 2.3(a). Two CW laser lights, one at
1555.00 nm and the other at 1558.53 ran, are modulated with 9.85 Gbit/s 215-1
pseudo-random bit sequences. The optical channels are then transmitted through a
dispersion-managed recirculating fiber loop.
LD
MUX
LD
Dispersion
Slope Monitor
Recirculating |
Fiber Loop !
EOM
9.85-Gb/s
PRBS
Loop Control
& Trigger
(a)
20 km
SMF
AOM
-MkzJVH-gTOF
Loop
Control
I
RF
Power
NBPF
$,aV
etere
AOM
Micro
Processor
84 km
EDFA
SMF
Loop
Trigger
P a in
'
Flattener
EDFA
(b)
*
*6 km
DCF
EDFA
»
Dispersion
Slope
(c)
Fig. 2.3: (a) The experimental setup for monitoring the residual dispersion slope
evolution versus the transmission distance, (b) Details of the recirculating fiber
loop, (c) The dispersion slope monitor. (TOM: Tunable optical filter, PD:
Photodetector, NBPF: clock frequency narrow band electrical filter)
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In order to demonstrate the dynamic dispersion slope monitoring, we utilize a
circulating dispersion-managed fiber loop (Fig. 2.3b). The dispersion slope
experienced by the WDM channels builds up at every consecutive circulation of the
loop, The loop consists of 84 km of SMF, 16 km of DCF, and a series of controlled
gain EDFAs to maintain the consistency of the optical power within the loop after
each round trip. Acousto-optical modulators (AOM) are used as switches to direct
the incoming WDM packets to-and-from the circulating loop. Each loop circulation
takes 500 us to complete, and each loop packet has a duty ratio of 4%. The fiber loop
has been adjusted to circulate the incoming WDM channels up to 8 cycles, which
enables us to observe the effect of dispersion-managed transmission, as long as 800
km, on WDM channels.
The details of the dispersion slope monitor are presented in Fig. 2.3(c). The RF
power of the regenerated clock component is measured for two WDM channels. Due
to the characteristics of the circuits used for the regenerated clock RF power
measurement and the photodiode used to monitor the optical power, the measured
clock amplitude needs to be normalized to the square root of the optical power. The
normalized clock frequency RF power provides us with the accumulated dispersion
value of that particular WDM channel. These values are processed in the
microprocessor and the residual dispersion slope of the link is calculated.
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.4 Results
The relationship between the normalized clock amplitude and the chromatic
dispersion has been investigated for calibrating the dispersion slope monitor module
microprocessor (Fig. 2.4(a)). Mathematical interpolation of the measured clock RF
power with respect to this data provides us with a measure of the individual WDM
channel net dispersion value for each additional loop circulation (Fig. 2.4(b)). In Fig.
2.4(c) we can see the evolution of the residual dispersion slope versus traveled fiber
distance. As predicted, the dispersion difference—hence the residual dispersion
slope—is positively dependent on the fiber link distance.
520
-45
500
■ 1555.00 nm
▼ 1558.54 nm
480 :
® -50
460
440 :
<o 420
.55
400
C>
380
300
400
500
600
700
800
900
Fiber Transmission Distance (km)
Dispersion (ps/nm)
Fig. 2.4: (a) Regenerated RF clock
power vs. dispersion, (b) Measured
— 12
dispersion value evolution for two
DWDM channels along the
dispersion compensated link, (c)
300
(C)
400
500
600
700
800
900
Dispersion slope accumulation
Fiber Transmission Distance (km)
obtained from Fig. 3(b).
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Previously the residual dispersion slope of the loop has been measured to be ~2.3
ps/nm2/loop. From Fig. 2.4(c), the residual dispersion slope for a single loop
circulation is calculated as -2.33 ps/nm2/loop, which is rather close to the stand­
alone measurement. Due to sensitivity limitations, data points for first two
circulations are omitted. The resolution of the system (0.5 ps/nm2/100 km) can be
improved further by deploying better electrical circuits. This method for monitoring
the residual dispersion slope can be used veiy efficiently for controlling dispersion
slope compensation subsystems.
2.5 Conclusion
We demonstrate an inline dispersion slope monitoring scheme for dynamically
measuring the residual dispersion of a dispersion-managed fiber optic link by taking
advantage of chromatic dispersion induced RF clock regeneration. From the
simultaneous monitoring of two WDM channels’ regenerated clock powers, we
deduce the residual dispersions for the channels and the residual dispersion slope of
the fiber link.
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 3
Distance-independent RF fading compensation
using a tunable nonlinearly-chirped FBG in a phase
diversity configuration
3.1 Introduction
An increasing number of applications require transmission of analog or digital
subcarrier-multiplexed (SCM) RF channels over fiber. However, transmitting
traditional double-sideband (DSB) signals is problematic due to chromatic
dispersion. This frequency-dependent fiber dispersion produces a deleterious time
delay between the two transmitted sidebands in the optical domain, causing serious
RF power fading that is a function of subcarrier frequency, fiber distance, and
accumulated dispersion [25]. Unfortunately, many potential applications involve
reconfigurable optical paths, where RF fading dynamically changes with
transmission distance. For robust RF-based optical systems, distance-independent RF
fading compensation is highly desirable.
Several approaches have been proposed to compensate for RF power fading in
conventional DSB systems, including minimum transmission biasing [31], adjustable
modulator chirp, and linearly-chirped fiber Bragg gratings [31], but they are all
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
dependent on transmission distance and must be actively tuned to accommodate
different path lengths. We have previously reported the use of a nonlinearly-chirped
fiber Bragg grating to provide tunable compensation for dispersion-induced RF
power degradation in variable-length multiple-channel SCM transmission links [30].
This grating has the ability to uniquely provide a tunable dispersion to incoming
signals since the time delay as a function of wavelength has a nonlinear shape.
However, this technique also suffers from being transmission-length dependent.
Single sideband (SSB) transmission, which is immune to the problem of dispersioninduced power fading, has also been proposed [45]. This technique can be
complicated to implement, and additionally, in a multiple-channel SCM system each
channel must have its own SSB-generating circuitry.
We use a nonlinearly-chirped fiber Bragg grating (FBG) in a phase diversity
configuration to achieve distance-independent RF power fading compensation for
DSB subcarrier-multiplexed systems [32]. In this technique, the incoming signal is
split into two components, and a □ phase shift is induced in the sidebands of one arm
relative to the other by reflection off the stretched nonlinearly-chirped FBG. At the
output of this phase-diversity configuration, orthogonally-polarized components of
the signals in the two arms are combined to reduce coherent crosstalk effects in the
fiber. We demonstrate distance-independent compensation of RF power fading from
0 to 150 km for 8, 10, and 12 GHz subcarrier frequencies, with received RF power in
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
all cases flat to within 1 dB. Error-free transmission is also achieved for 0,27, and 52
km transmission of a 155-Mbit/s SCM/amplitude-shift-keyed channel.
3.2 Concept of RF fading compensation using phase diversity
As mentioned before, fiber's chromatic dispersion can cause severe RF power fading
as a function of subcarrier frequency, transmission distance and accumulated
dispersion. Without compensation, the received subcarrier power is a function of
transmission distance, as shown in Fig. 1.6, with periodic power drop-outs:
(3-1 )P e ljRF ^ c o s
f 9\ +<Pi] = cos2 ncLD
I
2
( Jf RF
J
2"
= cos2
k LDA2fx F
opt j
c
where <pl and <p2 are the phases of the sidebands relative to the carrier, c is the speed
of light, L is the transmission distance, D is the dispersion, fRF is the subcarrier
frequency, and fopt is the optical carrier frequency for the transmission wavelength
X. The periodic length AL at which the drop-outs occur, can be expressed by:
( 0 .2 )
AL =
D Z flp
The distance Li where the first power drop-out occurs is given by:
(3-3)
I, = ----- % -r
2D t f e
To give an example, a 10 GHz subcarrier transmitted on an optical carrier at a
wavelength of 1.55 pm over standard single mode fiber with a chromatic dispersion
of +17 ps/nm*km experiences complete extinction after a transmission distance of
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-36 km. This demonstrates the severity of the RF fading problem, since fiber spans
of >30 km are highly desirable in optical communication systems.
The concept behind our distance-independent RF fading compensation technique is
illustrated in Fig. 3.1. In our compensation module, the nonlinearly-chirped fiber
Bragg grating arm induces an additional relative phase shift of □ between the optical
sidebands, resulting in a received electrical power that is again a function of distance,
with periodic power drop-outs.
Compensation Module
RF Power
Distance
SMF
RF Power
Distance
RF Power
Constant RF Power
Distance
Fig. 3.1: Conceptual diagram of the phase diversity technique for distance
independent RF fading compensation.
However, the different phases in the two arms cause the power drop-outs through
one arm to occur at different distances than the drop-outs in the other arm. Since the
incident optical signals are orthogonally-polarized, they do not interact in the
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
photodetector. After photodetection, these two optical signals result in a pair of
currents, one from the arm containing the grating (Ig):
Ig oc c, cos
(3.4)
j cos (2rfRFt + Gg)
and one from the other arm (I0):
(3.5)
I 0 OC
c2cos -'1+
V 2
cos{l7rfRFt + 0o)
where cl and c2 are constants representing the optical powers from the two arms
incident on the photodetector, and 0g and 0Oare the phase (or path length) differences
between the two arms, respectively. The average electrical power of the
superposition of the two currents can be calculated by integrating over one period T
of the subcarrier frequency:
(3-6)
T/2
00
-T /2
T/2
■J
CjCOS — 1 2+
j c o s ^ / ^ r + 0g )
-T /2
+ c2COS
^2
JCOS^ ^ + <9°)
dt
where T = ITrf is the period of the subcarrier frequency. By setting the optical
powers in the arms equal and introducing an additional optical delay between the two
arms o f the m odule such that 0g —0O= 7i/2 equation (3.6) can be written as:
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(3.7)
\
J
(
\
~P
sin — + — sin (2 ^r^jp /)+ co s — + ^ 2 c o s ( 2 ^ r /^ ) dt
Solving the integral in equation (3.7) leads to the total received subcarrier power
proportional to:
(3.8)
= constant
Equation (3.8) shows that the total received RF power of the combined signal is
constant, and therefore independent of the transmission distance. Note that without
equalizing the optical powers of both arms and more important, without offsetting
the optical path length in the two arms to achieve a 7t/2 phase differential of the two
received currents, the total combined electrical RF power would fluctuate depending
on the transmission distance.
3.3 Experimental setup of the compensation module
Details of the implementation of the RF fading compensation module are shown in
Fig. 3.2. The subcarrier signals are transmitted through a certain distance of single
mode fiber (SMF), ranging from 0 to 150 km, before entering the RF compensation
module.
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
RF Fading Compensation Module
Subcarrier
Circulator
SMF
Coupler J
Piezoelectric
Stretcher
Nonlinearly-chirped
FiberBragg Grating
Fig. 3.2 : RF Fading compensation module detail.
The lengths of the upper and lower arms are phase matched, and an additional D/2
phase shift is applied to one arm to warrant constant received RF power.
The splitting ratio at the input to the module is adjusted to compensate for different
losses in the two arms. The grating provides the required differential phase shift to
the optical modulation sidebands in that arm. It is attached to a piezoelectric
transducer (PZT), which stretches the grating to allow tuning to a different subcarrier
frequency. The signal bounces of the grating and leaves the circulator out of the third
port. A polarization controller and a polarization beam splitter allow the signals from
the two arms to recombine with minimum coherent crosstalk. The optical path
lengths of the two arms are matched at the required grating voltage, and then the
lengths are offset by 1/4 of the period of the subcarrier frequency fRF, which
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
corresponds to an additional phase shift of D/2. With this adjustment, the received
subcarrier power is independent of the transmission distance.
Note that the RF fading compensation module is effective at any location between
the transmitter and receiver. In addition, the module can compensate for different
transmitter chirps, since the effect of chirp manifests itself in the received subcarrier
power as a simple distance change.
The experimental setup to conduct bit error rate measurements is drawn in Fig. 4.3.
A 155 Mbit/s pseudo random bit stream (PRBS) can be multiplexed onto a 8 GHz
subcarrier, to generate an electrical double sideband signal (DSB). This subcarrier
multiplexed amplitude-shift-keyed (SCM-ASK) signal is transmitted in the optical
domain through a distance of SMF, before arriving at the RF fading compensation
module at the receiver.
8 GHz Subcarrier
155 Mbit/s
PRBS Data
SMF
Laser
EO-MOD
RF Fading
Com pensationl I
M odule
J
► Rx
----- >
Envelope
Detector
Fig. 33 : Experimental setup for distant-independent RF fading compensation
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
After compensation and optical to electrical conversion, the original data signal is
recovered using envelope detection. Note that the subcarrier is added to the electrical
DSB signal by shifting the high level of the PRBS data to 0 V, therefore it enables
envelope detection at the receiver.
3.3.1 Nonlinearly-chirped fiber Bragg grating
3.3.1.1 Operating principle
An optical fiber Bragg (FBG) grating is a periodic perturbation in the refractive
index along the fiber which is formed by exposure of the to an intense optical
interference pattern [45]. After the first demonstration of a permanent grating in an
optical fiber [46], fiber gratings have emerged as an important device in a variety of
optical fiber applications, due to their inherent fiber compatibility, low loss, low cost
and polarization insensitivity. Applications of FBGs include optical filtering,
wavelength
division multiplexing
and demultiplexing,
fiber-optic
sensors,
wavelength stabilization in lasers, dispersion compensation and gain equalization
[26].
The periodic refractive index perturbation in a FBG basically acts as a filter. Small
amounts of the incident optical field can be reflected at the index changes and those
partial reflected lightwaves can constructively interfere to generate a strong backreflected wave. For uniform FBGs this strong interaction occurs at specific
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
wavelength known as the Bragg wavelength Xb and depends on the refractive index
of the waveguide and the grating period. On the other hand, a chirped grating, where
the grating period is not constant but monotonically increasing (or decreasing), can
reflect different wavelengths at different positions in the grating, as shown in Fig.
3.4. The different time delays of the spectral components of the incident signal,
where the short wavelength components travel further into the grating before being
reflected, result in pulse compression for the reflected signal.
Chirped FBG
Xj
Incident
t<-
A
R eflected
a
X<2
) ) ) ) ) )
^3
p :
j
External
M echanical
Stretcher
Fig. 3.4: Conceptual diagram of a chirped fiber Bragg grating, where different
wavelengths reflect at different positions in the grating.
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The relative time delay versus wavelength curves for a linearly-chirped FBG is
plotted in Fig. 4.5a. When the grating is uniformly stretched with an external
mechanical stretcher, i.e. a piezoelectric transducer, the induced time delay curves
shifts towards longer wavelength. But the slope of the time delay curves at a certain
wavelength A,o, which represents dispersion, does not vary by stretching the grating.
To achieve a wavelength dependency of the induced dispersion of a chirped FBG,
Linearly-chirped
Relative *
Time
D elay (ps)
N onlinearly-chirped
Relative i
Time i k stretch
D elay (ps)) \
stretch
W avelength (nm)
W avelength (nm)
(a)
(b)
Fig. 3.5: Relative time delay versus wavelength curves for (a) Linearly chirped
FBG and (b) Nonlinearly chirped FBG.
the grating pitch has to follow a nonlinear function. The time delay versus
wavelength curves of such a nonlinearly-chirped FBG are depicted in Fig. 35.b. It
can clearly be seen, that the dispersion changes for a constant wavelength as as the
grating is stretched towards longer wavelengths.
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Relative
Time
D el ay (ps)
At
At
W a v e l e n g t h ( nm )
R F
Fig. 3.6: A nonlinearly-chirped FBG induces a time delay for the two optical
sidebands relative to the optical carrier of a subcarrier multiplexed signal.
If the two optical sidebands of a subcarrier multiplexed signal reflect of a
nonlinearly-chirped FBG, they experience a different time delay relative to the
optical carrier. Fig. 3.6 shows a time delay versus wavelength curve for a
nonlinearly-chirped FBG with two sidebands of a SCM signal located at Xo - fRF and
Xo + fRp, respectively.
To achieve a 180° phase difference between the two sidebands, the combined delay
At* has to add as follows:
(0.9)
Ax , = Ax x + At 2 = - i —
J RF
where An and At2 are the time delays of the sidebands relative to the carrier after
reflecting of the grating, and fRF is the subcarrier frequency.
The dispersion value to provide the □ phase shift between the optical sidebands can
be approximated by:
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(0.10)
D«
AArp
= -----= — J—
2f ftpAXgp ^fnp A0
with AXrf being the separation of the optical sidebands in the wavelength domain,
and Xo the transmission wavelength. This leads to a necessary amount of ~500 ps/nm
of dispersion for a 8 GHz subcarrier signal to achieve a 180° phase difference
between the two sidebands in the 1.55 pm wavelength region.
Using the nonlinearly-chirped FBG is essential for the phase diversity technique,
since operating at different points along the grating enables to achieve different
dispersion values so that different subcarrier frequencies can be compensated by
applying the □ phase shift between the optical sidebands. This would not have been
possible by using a linearly-chirped FBG.
3.3.1.2 Measured grating characteristics
The grating used in the setup of the RF fading compensation module is a
nonlinearly-chirped FBG [48]. The grating is mounted onto a piezoelectric
transducer to provide the mechanically stretching ability. The measurements are
conducted with the grating connected to the second port of the 3-port circulator. Fig.
3.7 shows the reflected spectrum of the of the nonlinearly-chirped FBG for different
applied stretching voltages to the PZT. The bandwidth of the nonlinearly-chirped
FBG is -1 dB and the reflected band can be shifted ~0.8 nm towards longer
wavelengths, by applying 500 V to the PZT.
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
-35
300 V
-40
500 V
ov
1)
ts
-60
-65
1550
1551
1552
1553
1554
Wavelength (nm)
Fig. 3.7: Reflected spectrum of the nonlinearly-chirped FBG for different
applied stretching voltages to the piezoelectric transducer.
Fig. 3.8 depicts the shift in the time delay versus wavelength curves for different
applied control voltages of 100 V, 300 V and 500 V to the PZT.
500
400
13
G
<D
1
300 y
300
200
500 V
100 V
100
0
1551
1552
1553
Wavelength (nm)
Fig. 3.8: Time delay versus w avelength curves o f the nonlinearly-chirped FBG
for different applied stretching voltages to the piezoelectric transducer.
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The slope of the time delay curve, which represents dispersion, increases from -300
ps/nm to -900 ps/nm for a fixed wavelength as the control voltage is increased. Note
that the curve remains its form as it shifts across the wavelength regime. The
operation point on the nonlinearly-chirped FBG can be chosen by tuning the PZT
voltage to achieve the necessary time delay slope required for the 180° relative phase
shift between the sidebands. Virtually, there is no upper boundary on the subcarrier
frequency, which can be compensated for, only the lower frequency may be limited
by the maximum available amount of dispersion induced by the grating.
3.4 Experimental results
Unmodulated RF subcarriers at frequencies of 8 GHz, 10 GHz and 12 GHz are
transmitted over a range of distances up to 150 km to illustrate the performance of
the RF fading compensation module from Fig. 3.1. The received RF power versus
the transmission distance for the uncompensated and the compensated case is shown
in Fig. 4.9 together with the theoretically expected RF fading curves for the three
subcarrier frequencies. The theoretical curves for the received RF power are
calculated using equation (3.1) on page 28 and the appropriate dispersion values for
the employed fiber. The periodic RF power drop outs occur at transmission distances
of -65 km for 8 GHz, -36 km and -108 km for 10 GHz, and -26 km, -77 km and
-129 km for 12 GHz. This shows that even relative short spans of fiber (<50 km) can
impose a severe power penalty in subcarrier multiplexed systems.
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Using the compensation module, the received RF power is flat to within ~1 dB for
the three subcarrier frequencies. The 3 dB power penalty compared to the best case
value for uncompensated transmission for the same received optical power is
inherent to the phase diversity technique. It can be explained by the equal splitting
C om p e n s a t e d
Ih
O
&
4
- 10
T3
'§
C4
M e as ur e d
-20
T3
|
I
I
-30
Theoretical
-40
0
8 GHz
50
100
150
D i st anc e ( k m )
(a)
C om p e n s a t e d
C om p e n s a t e d
—, 4
4
1
I
-10
-10
M easured
M easured f /
- 20
104 ' 2°
-30
19
21
Theoretical
-40
50
J 10 GHz
100
D i st anc e ( k m )
(b)
150
n
-3 0
19.
-40
h e o r et i c a U
50
100
1 50
D i s t a n c e (km )
(C)
Fig. 3.9: Normalized RF power versus distance for uncompensated and
compensated transmission for (a) 8 GHz, (b) 10 GHz and (c) 12 GHz subcarrier
frequency. The dashed line shows the theoretical case for RF power fading,
ratio in the compensation module, which yields the two currents in the photodetector
from equation (3.4) and (3.5). Even if one of the current is at its minimum the other
is at its maximum, because of the 90° phase difference between the two signals. This
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
leads to 50 % less received power than the best uncompensated case, but it is
constant for any transmission distance.
Fig. 3.10 depicts the required control voltage for the PZT to stretch the grating
versus the subcarrier frequency to obtain the 180° phase shift between the optical
sidebands, thus achieving RF fading compensation. It displays the necessary
monotonous tuning characteristics to accomplish reliable compensation.
500
450
400
00
M
0
350
1
300
>
o
U
250
200
7
8
9
10
11
12
13
Subcarrier Frequency (GHz)
Fig. 3.10: Required grating control voltage versus subcarrier frequency to
achieve RF fading compensation.
BER measurements were taken with 155 Mbit/s pseudorandom data, which was
electrically up-converted to 8 GHz. The SCM-ASK data signal is transmitted and
envelope detected at the receiver. Fig. 3.11 a shows the RF power spectrum of the
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
SCM-ASK signal before transmission and Fig. 3.11 b displays recovered data bits
after envelope detection at the receiver.
The BER versus received optical power curves for fiber spans of 0 km, 27.7 km and
52. km are plotted in Fig. 3.12. Error free transmission is achieved for the three
transmission distances, where 0 km corresponds to maximum RF power fading in the
grating arm of the module, while 52.4 km corresponds to almost maximum RF
power fading in the non-grating arm.
RF Power Spectrum
7.5
8
Recovered Data
8.5
0
10
20
30
Frequency (GHz)
Time (ns)
(a)
(b)
40
50
Fig. 3.11: (a) RF power spectrum of the transmitted subcarrier multiplexed
amplitude-shift-keyed data signal, (b) Recovered bit stream after envelope
detection
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2
3
4
5
6
0km
27.7 km
52.4 km
7
8
Baseline
9
10
-13 -12
-11
-10
-9
-8
-7
Received Optical Power (dBm)
Fig. 3.12: Measured BER versus received optical power for different
transmission distances of 0 km, 27.7 km and 52.4 km (data rate = 155 Mbit/s,
subcarrier frequency = 8 GHz). The insert depicts an eye diagram of the
recovered signal.
The BER performance does not depend on whether the received electrical subcarrier
power passes through the grating arm or the non-grating arm, therefore resulting in
distance independent RF fading compensation. Most of the -2.5 dB optical power
penalty relative to the back-to-back BER measurement comes from the 3 dB
electrical power penalty inherent in the phase diversity technique. The asymmetric
duty cycle of the recovered eye diagram shown in the insert of Fig. 4. 12, was caused
by the nonideal characteristics of the employed envelope detector, which was not
optimal suited to recover digital amplitude modulated signals.
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5 Conclusion
We demonstrated the use of a nonlinearly-chirped fiber Bragg grating in a phase
diversity configuration to
achieve distance-independent RF power fading
compensation for double sideband subcarrier multiplexed fiber-optic communication
systems. RF power fading compensation for distances ranging from 0 to 150 km and
for subcarrier frequencies of 8, 10, and 12 GHz was shown. The received RF power
was flat to within ~1 dB in all cases. Furthermore, BER measurements of transmitted
155 Mbit/s PRBS data, subcanier multiplexed to 8 GHz, proved error free recovery
of the received signal with an optical power penalty of ~3 dB, which is inherent to
our technique.
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 4
Wavelength Conversion of Subcarrier Channels
using Difference Frequency Generation in a PPLN
Waveguide
4.1 Introduction
For many applications, it is quite advantageous to transmit several analog or digital
subcarrier-multiplexed (SCM) RF channels over an optical fiber link or network.
These applications include: cable TV, wireless network interfaces, microwave
photonic systems, and control information for optical networking and optical packet
switching [21,53]. Moreover, subcarrier modulation is important for data grooming,
bandwidth allocation flexibility, and access networks.
Next generation networks may have routing and switching capabilities in the
wavelength-division-multiplexing (WDM) layer for flexible bandwidth allocation.
Reconfiguration can be achieved either by a tunable transmitter or a tunable receiver.
In either case, wavelength conversion is needed in order to increase the efficient use
of the limited wavelength pool as well as resolve wavelength contentions [54].
It is only natural to envision a future platform that combines subcarrier multiplexing
with wavelength conversion for high-throughput network performance. In such a
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
network,
wavelength
conversion
should
support
signals
of
arbitrary
formats/protocols in order to accommodate various user applications and facilitate
network interoperability. All-optical wavelength shifting of subcarrier signals can be
accomplished by several methods, including: cross-gain modulation (XGM), cross­
phase modulation (XPM), four-wave-mixing (FWM), and difference-frequency
generation (DFG) [55-58]. However, only FWM and DFG offer complete format
transparency. Moreover, all methods except DFG generally use a semiconductor
optical amplifier (SOA) as their wavelength-shifting medium, which generates at
least one of the following disadvantages, depending on the specific method used: (i)
distortions due to nonlinear and population-dependent memory characteristics, (ii)
crosstalk, (iii) limited conversion speed, (iv) limited wavelength range, (v) limited
conversion efficiency, (vi) additive noise, (vii) spectral distortion, and (viii) limited
dynamic range. Alternatively, DFG achieves memoryless, linear, transparent
wavelength shifting with extremely low crosstalk and quantum limited additive noise
over a wide wavelength range. Wavelength shifting of subcarrier channels has been
demonstrated using DFG in an AlGaAs device [59].
4.2 Experimental Setup
In this paper, we demonstrate and characterize a transparent all-optical wavelength
conversion process for subcarrier-multiplexed channels in which a memoryless
c(2):c(2) DFG process uses 1550-nm pumping in a periodically poled lithium niobate
(PPLN) waveguide [52]. We achieve penalty-free all-optical wavelength conversion
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
of two 55-Mbit/s subcarrier channels, and the process shows a >30-dB linear
dynamic range for crosstalk-free, transparent operation. We note that the PPLN
structure has a lower insertion loss than the AlGaAs device in [59].
(a)
All-Optical
W avelength Converter
V.
(b)
Fig. 4.1: Conceptual diagram of (a) wavelength conversion of subcarrier
channels, (b) DFG in a PPLN waveguide using a c(2):c(2) process
Figure 4.1a illustrates the desired wavelength conversion function. Data is imposed
on modulated sidebands around an original carrier wavelength 11. After wavelength
conversion, these same sidebands are located around a new carrier wavelength 12.
Figure lb shows a conceptual diagram of the operation of the c(2):c(2) process used
to perform this function [62]. The first c(2) process involves the CW pump (lpump)
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
undergoing second harmonic generation (SHG) to produce a local oscillator at
lpump/2. This then mixes with the modulated signal lsignal(t) through DFG to form
a wavelength-shifted copy of the signal at a wavelength of ~(21pump - lsignal(t)).
Both parametric processes are instantaneous, permitting modulation bandwidths in
excess of several THz. The conversion efficiency is symmetric in the forward and
backward directions. Since the DFG conversion efficiency is not proportional to the
signal power, the process is linear over a large dynamic range. Moreover, there is no
crosstalk sideband at (21signal(t)-lpump) as there is in FWM.
EDFA
EOM
^-signal
H>
■K
m u
BP filter
@
2 ^pump"^signal
^Dunro EDFA
BPSK
BPSK
Modulator Modulator
( |)
RF,
WDM PPLN
4-
BPSK
Demodulator *•
t
Data
Out
RF| or RF2
d>
Q RFi
Pattern
Generator
Fig. 4.2: Experimental setup.
Our experimental setup is shown in Fig. 4.2. A tunable laser is set at 1555-nm and
used as a CW signal. A 55-Mb/s on-off-keyed bit stream is binary-phase-shift-key
(BPSK) modulated onto an RF1 carrier at 650 MHz, and a second 55-Mb/s channel
is modulated onto an RF2 carrier at 1050 MHz. The subcarrier channels are fed into
a 2-GHz-bandwidth LiNb03 external modulator. The EDFA is placed after the
external modulator and is used to control the signal power launched into the PPLN
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
waveguide. A tunable narrow-linewidth extemal-cavity laser at 1550 nm is used as
the pump source for initiating the DFG process. The 1550-nm pump source is
amplified to +22 dBm and fed into a wavelength-selective coupler. This coupler
combines the pump and signal and also filters out the amplified-spontaneousemission noise of the high power EDFA. At the output of the PPLN waveguide, the
wavelength-converted signal at 1545 nm is optically filtered and received. The
receiver is connected to a BPSK demodulator to recover the original bit stream. The
BPSK demodulator is driven by either a 650-MHz or a 1050-MHz RF carrier to
select between the subcarrier channels.
4.3 Experimental Results
W avelength
converted
Signal h i
1545 mil
1550 nm
1555 nm
W avelength (nm)
Fig. 4.3: Optical spectrum after wavelength conversion in the PPLN waveguide.
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The optical spectrum at the PPLN waveguide output is shown in Fig. 4.3. The power
difference between the 1555-nm original signal and the 1545-nm wavelength-shifted
signal at the output of the PPLN waveguide is the conversion efficiency. For a 1550nm input pump power of ~ 100 mW, a conversion efficiency of -21 dB is observed.
One reason for the low conversion efficiency in this particular device is anomalous
loss in the fiber pigtails. Typical fiber-to-fiber insertion loss is 3.5 dB [60], while the
device used in this work showed a much higher loss of ~ 7 dB. It should also be
noted that recent demonstration of a novel buried waveguide design and fabrication
technique have led to a threefold increase in the internal conversion efficiency [61].
An optimized device using this technique would allow for 0 dB wavelength
conversion with only 75 mW pump power.
5
10
t; S
1S3 8w
o g
J3 ?
& i£
s 5
1.1
15
-20
-25
-30
-35
-40
15
-10
-5
0
5
10
15
20
Signal Power (dBm)
Fig. 4.4: Linearity of DFG: wavelength converted signal power vs. signal power,
measured after the PPLN waveguide
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 4.4 shows the variation of the wavelength-shifted output signal power as a
function of the input signal power, for which the pump power is kept constant.
Wavelength-conversion efficiency of -21 dB is maintained regardless of the input
signal power level, demonstrating the large linear dynamic range of the DFG process
as expected from theory. To further characterize the linearity of the DFG process, we
have measured the RF spectra of the subcarrier channels before and after the
wavelength-conversion process.
a
B efore X co nversion
S3
vSUt
A fter X co n v ersio n
%
<2
'^vwyAvy/flfV
H I
RF i
RF2
R F F req u e n cy
Fig. 4.5: RF spectra before and after the wavelength conversion.
As shown in Fig. 4.5, the spectra are virtually identical, indicating transparent
wavelength conversion. In contrast, a nonlinear process would show harmonics and
spectral distortion.
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
ORF i Before X shift
A R F 2 Before X shift
RFi After X shift
RF2 After X shift
P4
UJ
PQ
'■
w'
bO
O
-34
-33
-32
-31
-30
Optical Power (dBm)
Fig. 4.6: BER curves of subcarrier multiplexed channels for before and after the
wavelength conversion. 6
Figure 4.6 shows the bit-error-rate (BER) curves of the data signals before and after
wavelength conversion for both subcarrier channels. No significant power penalty is
observed. To measure the effect of the wavelength spacing between the input signal
and the pump, the signal wavelength is varied across a broad range.
^-2 1 -
O RFi Before X conversion
□ RF2 Before X conversion
• RFi After X conversion
■ RF2 After X conversion
-28
£>-29
I -30
CO
S -31
co
-20
-10
0
10
20
Wavelength Conversion Distance (nm)
Fig. 4.7: Received optical power sensitivities for 10-9 BER vs. wavelength
spacing between the input and output data signal.
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In Fig. 4.7, receiver sensitivities for 10-9 BER in terms of optical power are plotted
as a function of the wavelength spacing between the input and output signals (i.e.,
twice the wavelength difference between the input signal and the pump). For both
up- and down-conversion, we observe up to 20-nm wavelength shifts with similar
performance, limited only by the EDFA bandwidth and filtering characteristics of
our equipment.
4.4 Conclusion
We demonstrate and characterize a transparent all-optical wavelength conversion
process for subcarrier-multiplexed channels. Our memoryless c(2):c(2) differenceffequency-generation process uses 1550-nm pumping in a periodically poled lithium
niobate (PPLN) waveguide. We achieve penalty-free all-optical wavelength
conversion of two 55-Mbit/s subcarrier channels. The process shows a >30-dB linear
dynamic range for crosstalk-free, transparent operation.
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 5
Statistics of PMD-induced power fading for double
sideband and single sideband subcarrier-multiplexed
signals
5.1 Introduction
Polarization mode dispersion (PMD), due to the random birefringence of single
mode optical fiber, is one of the critical challenges in next-generation optical
communication systems. A key feature of PMD is its statistical behavior, since the
relative orientation between the state-of-polarization (SOP) of the input signal and
the principal-states-of-polarization (PSPs) of the fiber varies randomly with time.
Slow Polarization
ii
Subcarrier
Transmitter
Fast Polarization
A Phase
Received
Subcarrier
Power
Square Law
Detector
DGD
Fig. 5.1: First-order PMD induces a differential group delay in an optical
sideband of a SCM signal, which leads to a phase difference in the corresponding
photo-received subcarrier signals, possibly causing serious power fading.
Moreover, the differential group delay (DGD) between the fast and slow PSP, i.e.
first-order PMD, is a random process with a Maxwellian probability distribution.
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
These characteristics of PMD induce a stochastic and dynamically-changing
degradation of high speed digital baseband channels (310 Gbit/s).
Transmission of analog and digital subcarrier-multiplexed (SCM) signals over fiber
will also be severely affected by PMD [63], although its statistical impact on SCM
signals has not yet been investigated. Therefore, it is imperative to examine the
fading characteristics for SCM signals using a realistic PMD source that closely
approximates the statistical nature of PMD.
The DGD between the fast and slow PSP of an optical sideband in a SCM signal
causes a phase difference in the corresponding received subcarrier signals in the
photodetector, as shown in Fig. 5.1. Superposition of the photo-currents may lead to
serious power fading of the recovered subcarrier signal due to destructive
interference that is a function of subcarrier frequency, accumulated DGD, and optical
power splitting ratio between the PSP’s [64]. Furthermore, higher-order PMD can
cause additional distortion and degradation of the transmitted signal [65,66].
5.2 Experimental Setup
PMD-induced power fading is similar to the fading that occurs in conventional DSB
systems due to chromatic dispersion. Although it has been shown that single
sideband (SSB) transmission is relatively immune to chromatic dispersion, it is
unclear whether SSB intensity modulation is also beneficial in reducing PMDinduced fading.
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Attenuator
riT"' -a > <
Analyzer
15-section PMD Emulator
Snbcarrier
lst-order PMD Compensator
Fig. 5.2: Experimental setup. (TL: tunable laser, 900: phase shifter, PC:
polarization controller, OF: optical filter, Rx: receiver)
We experimentally investigate the statistics of PMD-induced power fading as a
function of DGD for DSB-SCM and SSB-SCM signals using a PMD emulator with
an average DGD of -40 ps [62]. We find a similar susceptibility to PMD-induced
power fading for both modulation formats in the absence chromatic dispersion. A
significant improvement in the worst case power fading penalty (-20 dB) is achieved
by using a single section of polarization maintaining (PM) fiber in a dynamic firstorder PMD compensator. Furthermore, the results of numerical Monte Carlo
simulations support the measured data.
Figure 5.2 shows the experimental setup used to compare DSB-SCM and SSB-SCM
modulation formats under high PMD conditions. Besides conventional DSB intensity
modulation using an external single electrode Mach-Zehnder (MZ) modulator, SSB
modulation is achieved by employing a dual electrode MZ modulator and driving the
second input with a 90° phase shifted copy of the subcarrier signal input to the first
electrode [68]. The PMD emulator contains 15 sections of polarization-maintaining
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(PM) fiber, with 9 polarization controllers (PCs) distributed between the sections to
realize different polarization coupling and therefore closely emulate the Maxwellian
distribution of DGD (measured average DGD -40 ps) [68]. The input signal to the
PMD emulator can be selected between the modulated signal and the PMD analyzer
output, where a tunable laser is used to determine the actual DGD value of the
emulator. The compensated and uncompensated received subcarrier power is
measured for the same emulator state and same adjusted optical power by selecting
the corresponding optical path. The dynamic first-order PMD compensator consists
of an electronically controlled PC followed by a 24 m long section of PM fiber
(DGD -42 ps). Some of the light is tapped off after the PM fiber and detected to
generate a feedback signal by mixing the received subcarrier signal with itself. The
PC maximizes the feedback signal, which is proportional to the received RF power,
by optimizing the polarization coupling into the PM fiber. Note, that there is minimal
chromatic dispersion in the setup.
5.3 Experimental Results
We measured the received subcarrier power and the corresponding DGD values for
350 independent polarization samples by randomly changing the polarization
coupling inside the PMD emulator for DSB-SCM and SSB-SCM intensity
modulation using a 7 GHz subcarrier. Figure 5.3 (a) shows the RF power fading
penalty versus DGD for DSB and SSB formats with and without compensation. The
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
solid line in each figure plots the theoretical fading penalty for a 7 GHz subcarrier
for equal polarization coupling into the PSP for first-order PMD (i.e., DGD) only.
DSB-SCM and SSB-SCM modulation exhibit a similar sensitivity to PMD-induced
power fading. Higher-order PMD can lead to a significant fading penalty (>20 dB),
as indicated as by some points outside the theoretical fading curve. Unlike for the
No Compensation
lst-order Compensation
5 30
DSB
0
DSB
SSB
20 40 60 80 100
0
20 40 60 80 100
DGD(ps)
0
SSB
20 40 60 80 100
DGD(ps)
0
20 40 60 80 100 120
DGD(ps)
DGD(ps)
(a)
No Compensation
lst-order Compensation
o 30
DSB
0
SSB
20 40 60 80 100
DGD (ps)
0
DSB
20 40 60 80 100
DGD (ps)
0
SSB
20 40 60 80 100
DGD (ps)
0
20 40 60 80 100 12C
DGD (ps)
(b)
Fig. 5.3: PMD induced power fading vs. DGD curves for double sideband (DSBSCM) and single sideband (SSB-SCM) intensity modulation of a 7 GHz subcarrier
with and without dynamic first-order PMD compensation, (a) Measurement of 350
independent polarization samples, and (b) Simulation of 10000 independent
polarization samples for each modulation format. The solid line corresponds to the
theoretical fading penalty for equal polarization coupling into the PSP for firstorder PMD (i.e., DGD).
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
59
case of chromatic dispersion, SSB modulation does not avoid a fading penalty since
the PMD effects apply directly to the optical sideband, causing power fading when a
single optical sideband is present. The worst case fading penalty can be reduced by
-20 dB for both formats by using dynamic first-order compensation.
Monte Carlo simulations using Maxwellian PMD statistics were performed to
support our experimental data. Figure 6.3(b) shows the simulation results for RF
power fading versus DGD for 10000 independent samples. The simulations exhibit a
qualitatively comparable performance to the measured data for both formats with and
without compensation. The larger maximum fading penalty compared to the
measurements may come from the limited number of samples in the experiment. 1%
of the samples (100 worst cases) exhibit >14 dB of fading penalty for both formats
without compensation. After compensation the 1% power variance is ~4 dB,
respectively.
Figure 5.4 shows the measured bit error rate (BER) versus the received optical power
for a 155 Mbit/s DSB-SCM and SSB-SCM binaiy-phase-shift-keyed (BPSK) signal
modulated onto a 7 GHz subcarrier. The baseline is measured without the PMD
emulator. The DGD value of the PMD emulator is set to ~40 ps to measure both
modulation formats with and without compensation. The power penalty without
compensation is >1.5 dB for both formats, compared to <0.5 dB with first-order
compensation.
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Baseline
Received Optical Power (dBm)
Baseline
Received Optical Power (dBm)
Fig. 5.4 : Measured bit error rate vs. received optical power for 155 Mbit/s DSBSCM-BPSK and SSB-SCM-BPSK signals at 7 GHz with and without first-order
compensation. The measured DGD for both modulation formats is ~40 ps. The
inserts show an error-free recovered eye diagram after first-order compensation.
5.4 Conclusion
We experimentally and numerically compare the statistics of power fading for DSB
and SSB subcarrier-multiplexed signals under high PMD conditions (average DGD
~40 ps) with and without dynamic first-order PMD compensation. We find that both
SSB-SCM and DSB-SCM signals exhibit similar sensitivity to PMD-induced power
fading. First-order compensation reduces the worst case fading penalty by ~20 dB
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 6
Dispersion Division Multiplexing for In-Band
Subcarrier-Header-Based All-Optical Packet
Switching
6.1 Introduction
Packet-switched all-optical networks need to process the header to efficiently route
packets to the appropriate destination [69]. One of the most important requirements
during the address information extraction for routing purpose is that the header
should be able to be processed rapidly and on-the-fly. Subcarrier multiplexed (SCM)
header transmission has been proposed as a method to overcome this problem [7012}.
When the header is subcarrier multiplexed it can be processed at a rate much lower
than the data packet bit-rate. This eliminates the need for data-rate-fast digital
circuits in cross-connects. Yet the major disadvantages in the standard SCM header
transmission are (i) the subcarrier frequency is much higher than the data bit rate,
hence higher frequency modulators and receivers are required, (ii) the optical domain
bandwidth utilization is inefficient, and (iii) the proposed systems are not easily
applicable to commercial transmission systems.
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In this paper we propose a method to transmit SCM-header at a subcarrier frequency
lower than the data-rate, specifically at 7.7 GHz for 10 Gb/s data-rate, using
dispersion division multiplexed subcarrier transmission [20]. It is important to note
that this is not possible using standard SCM transmission, since the subcarrier placed
at such a low frequency can’t be filtered out efficiently and this would deteriorate the
detected data performance greatly. This technique permits efficient optical
bandwidth utilization and results in little power penalty for the received data, since
the SCM header channel is invisible with respect to the data channel detector. There
isn’t any need for modifying the data receiver architecture, such as adding very
narrow band optical receivers or low pass electrical filters. Also to our knowledge
this is the first demonstration for SCM-header transmission and switching at 10 Gb/s
data-rate.
63
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.2 Concept
The conceptual diagram for in-band frequency SCM header based switching is
depicted in Fig. 6.1. The data channels and SCM-header channels are modulated on
the same CW laser signal at different modulators. Then chromatic dispersion is
applied to the SCM-header modulated. The effect of dispersion on the SCM-header
channel is apparent only after the signal is detected at a photodetector. Applying
dispersion to a subcarrier channel introduces a time delay, or phase difference,
between the upper and lower modulation sidebands.
When the upper and lower sidebands have a 180° phase difference, the sidebands
D e te c to r
.Data
Data
Switch
10 Gb/s Data
RF Spectrum
Laser)
Switch Control !
LII3(ICISIUU
SCM-Header
-M b/s at f, GHz “ H j R F Fading)
Data + S hadow
SC M Header
1 Dispersion i
(RF Fading !I Recovery) I
Detection
& Down
Conversion
A SCM
Shadowed
SCM Header
R F Spectrum
# 1 Spectrum
Fig. 6.1: Conceptual diagram for shadow subcarrier multiplexed header
transmission and detection for header based optical switching,
cancel each other completely at the photodetector. This event is specifically called
RF fading [28]. This faded, or “shadowed” SCM signal is combined with the data
modulated signal. The “shadow” SCM channel is invisible for the data
photodetector, and has no effect on the data bit-error-rate performance.
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In order to recover the shadow SCM-header signal we apply the same amount of
dispersion that is used to fade it at the transmission side. This dispersion removes the
phase difference between the sidebands of the SCM channel, and by doing so it
compensates for the RF fading. Now the header information can be down-converted
from the subcarrier frequency and can be used to control the switch in the optical
cross-connect.
6.3 Experimental Setup
Transmitter
10 G b/s Data
Receiver — ^
GHz
RF Spectrum
Da
jn
e
DDM-SClVb
DM-SCMi
Header
I I ' *I
> [ S w itc h
rrm
Data + DDM
SCM H eader
Ts,p
— ^
—
Switch/ e Control
Detection
& Down
Conversion
Header
MHz
SCM
Header
SC M -H eader
M b/s at %G Hz
Switch
Fig. 6.2: Experimental setup for shadow subcarrier multiplexed header
transmission and detection for switching information.
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Our experimental setup is shown in Fig. 6.2. Continuous-wave (CW) laser power is
split using a polarization-beam-splitter (PBS). On one CW signal we modulate a 9.85
Gb/s PRBS (2A23-1). On the other CW signal, we modulate the subcarrier
multiplexed (SCM) 55 Mb/s binary-phase-shift-keyed (BPSK) header information,
with a subcarrier frequency of 7.7 GHz. Standard single-drive electro-optic
modulators and commercially-available nonlinearly chirped fiber Bragg gratings
(NL-FBG) are used to introduce the dispersion required for RF fading. Both NLFBGs have a center wavelength of 1546.7 nm with 0.55 nm bandwidth, and the
dispersion value range is [-600, -1600 ps/nm]. The amount of dispersion necessary to
fade the subcarrier frequency for the "shadow" SCM-header is -1080 ps/nm for 7.7
GHz. The dispersion is applied by reflecting the SCM modulated laser signal from
the tuned NL-FBG. The 9.85 Gb/s modulated and the “shadow” SCM-header signals
are then combined using a polarization-beam-combiner (PBC). We use polarization
beam combination in order to prevent coherent crosstalk between the two signals.
The channels are adjusted to have the same optical power after the combination.
The combination signal is transmitted over <1 km fiber and inputted to the lithiumniobate switch. Before the switch some portion of the signal is tapped off for the
SCM-header recovery module. The recovery NL-FBG in the SCM-header recovery
module is tuned to the same dispersion value as the “shadowing” NL-FBG. This
compensates for the RF fading of the shadow SCM-header channel. The
photodetected signal contains both the data channel and the recovered SCM-header
channel. Coherent demodulation is used to obtain the header information. The header
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
information is then sent to a threshold detector. The output of the threshold detector
is directly used to drive the 1X2 optical switch. This control signal is used to route
the incoming signals to the appropriate output ports. The switch output ports 0 and 1
are detected to demonstrate the switching.
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6.4 Experimental Results
6.4.1 Receiver RF Spectra
-20
-40
:
-25
I f -30
-45
00
■3 -38
« -50
|
O
O -45
CL
U. -55
oc
m
-601
i i /fa it A
-40
a.
a. -so
of
W
-55
-60
4
5
6
7
8
9
10
4
5
6
7
8
9
10
RF f r e q u e n c y (GHz)
RF f r e q u e n c y (GHz)
(d)
(a)
-10
-15
-20
-25
"E -2 0
(D
2. '35
2 . -25
<5 -40
§
0.
II.
0£
S -30
-45
O -35
-50
LL -40
-55
-45
-60
-50
RF f r e q u e n c y (GHz )
10
R F f r e q u e n c y ( G H z)
-10
-15
-20
-25
"g -30
ID
E" -20
00
2. ‘3S
2 -25
m -40
O -45
O
Q.
IL -50
-35
U. -40
00
-55
-60
10
RF f r e q u e n c y (GHz)
-45
-50
10
RF f r e q u e n c y (GHz)
Fig. 6.3: (a-c) Data receiver RF spectrum plots for the shadow SCM only, data
channel only, and combined cases, (d-f) Shadow SCM Recovery Module RF
spectrum plots for the shadow SCM only, data channel only, and combined cases.
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The electrical power spectra of the data receiver side and shadow SCM-header
recovery module are plotted in Fig. 6.3 (a-f). Fig. 6.3(a-c) show the data receiver RF
spectrum plots for the cases when (a) only the shadow SCM-header channel is on,
(b) only the data channel is on, and (c) both channels are on. Note that the shadow
SCM channel is not visible on the power spectrum (shadowed) when both channels
are on. Whereas Fig. 6.3(d-f) show the shadow SCM-header recovery module RF
spectrum plots for the same cases, respectively. In Fig. 6.3(d) it is seen that the
SCM-header channel is regenerated by more than 20 dB, when compared to Fig.
6.3(a). In Fig. 6.3(f) the SCM channel is visible and -17 dB above the noise floor
induced by the data channel.
6.4.2 Data and Header Bit Error Rate Performance
In Fig. 6.4(b) the effect of using a subcarrier frequency different from the fading
frequency is examined. The modulated SCM channel will be less faded as the
difference between the SCM frequency and the RF fading frequency for the used
NL-FBG dispersion value increases, so as expected the subcarrier frequency change
increases the power penalty. The offset in the subcarrier frequencies can also be
visualized as offset in the NL-FBG dispersion value. For example, the dispersion
difference between the dispersion values required for RF fading 7 GHz and 7.7 GHz
is -200 ps/nm, so we can assume that a residual dispersion of -200 ps/nm would
introduce another 0.5 dB power penalty.
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3
4
>
<a
*•>
5
as
as
O'
OH
u.
§
Ul
*>
• 7.7 GHz
■ No Modulation
♦ PSK SCM
• ASK SCM
in
-12
-11.5
-11
■ 7.0 GHZ
♦ 6.5 GHZ
▲ 6.0 GHZ
00 8
9
-10.5
-10
-9.5
Optical Power (dBm)
(a)
10
12
-11.5 -11
-10.5 -10
-9.5
-9
Optical Power (dBm)
(b)
(0
(ii)
(iii)
(iv)
m
i | iii
(V)
Fig. 6.4: Bit-error-rate curve plots for the effect of the shadow SCM-header
channel on the 9.85 Gb/s data channel, (a) Plots for the cases when the SCMheader channel is (i) not modulated, (ii) PSK modulated (iii) ASK modulated, (b)
Plots for the cases when the SCM-header frequency is off from the set NL-FBG
dispersion value RF fading frequency (7.7 GHz), (c) (i) recovered SCM header,
(ii) output of the threshold detector, (iii) 10 Gb/s data stream before the switch,
(iv) output of the switch port 0, (v) output of the switch port 1.
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Time domain waveforms are displayed in Fig. 6.4. (c) for the SCM-header based
switching operation, (i) and (ii) are the recovered header and the corresponding
threshold detector output, (iii) is a plot of the data stream before the switch, (iv) and
(v) are the outputs of the switch ports 0 and 1. It is seen that the switch directs the
data stream to the port 1 on for logic “1” output of the threshold detector and to the
port 0 for the logic “0” output.
6.5 Conclusion
We experimentally demonstrated the viable placement of a 55 Mb/s bit rate header
on a 7.7 GHz subcarrier for in band use with lOGb/s data packets [20]. The resultant
system is demonstrated to be spectrally efficient. This allowed easy recovery of the
header information without the need for faster than data rate photodetectors, and it
has the advantage of header being processed with simple, low-speed electronics. We
demonstrated that when such a subcarrier header channel is Dispersion Division
Multiplexed (DDM) with a 10 Gb/s data channel it has negligible effect on the data
channel performance. And at a switching node demonstration, we recover and
downconvert the DDM subcarrier header to obtain the routing information. This
experiment demonstrates the possible advantage of subcarriers for inline header
processing of very high data rate traffic channels.
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 7
Bias-Induced Diversity-Detection (BIDD) Technique
for Robust Transmission of Subcarrier-Multiplexed
Channels
7.1 Introduction
Data transmission using subcarrier modulation (SCM) has many applications,
including: CATV, microwave photonics, wireless remoting, and control signaling in
optical networks. In general, SCM channels tend to suffer from RF fading, in which
the power of the signal fades as a function of the subcarrier frequency and the
transmission distance.
Such fading can be caused by fiber chromatic dispersion
(CD) in the case of double sideband (DSB) transmission, for which each sideband
travels at a different speed down the fiber and will periodically become out-of-phase
relative to the carrier itself [28]. Moreover, fading can also occur due to polarizationmode-dispersion (PMD) in both single and double sideband transmission[75]. PMD
occurs when the speed of light is different between the two polarization axes, and the
light from the two axes will periodically be out-of-phase and cancel each other
causing fading. In general, it would be highly desirable to be able to transmit SCM
channels that do not experience RF fading and that are robust to chromatic dispersion
and PMD.
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Previous reports of minimizing CD-induced RF fading have included: (i) dispersion
compensating fiber Bragg gratings [34], (ii) optical single sideband (SSB)
modulation[77], (iii) optical single sideband detection[80], (iv) and minimum bias
operation of the modulator [4]. In minimum bias the detected RF signal is at the
second harmonic frequency which is robust to CD. However this method has not
been demonstrated for SCM data transmission until now. Several schemes that
require active polarization tracking and PMD compensation at the receiver end have
been studied[81], however PMD-robust transmission has yet to be demonstrated.
We demonstrate a technique that induces SCM data diversity during the detection
process without using redundant subcarriers or optical channels, or any extra optical
bandwidth. By adjusting the DC bias of the modulator, we suppress the optical
carrier to a certain level that generates first and second harmonics of the SCM signal
at equal RF powers at the photodetecter. In our technique, we employ the total RF
power of both the photodetected first and second harmonic SCM signals, thus
achieve diversity detection of the SCM data even though we have modulated and
transmitted only a single subcarrier signal. It should be noted that we achieve
diversity at the photodetector, without any loss, and the system has 3dB better
sensitivity than quadrature bias since the optical power is transferred to the
subcarriers. We show that our technique is robust to chromatic and polarization
mode dispersion. Under worst case CD for a given RF signal, our technique is robust
within 6 dB. And against the PMD our technique displays ~8 dB RF power
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
improvement for the 5% probability tail compared to quadrature bias for different
subcarrier frequencies.
7.1 Concept
Fig.7.1 shows the concept of Bias-Induced Diversity Detection (BIDD). After optical
modulation, the subcarrier is copied to two sidebands around the optical carrier. For
quadrature bias operation, the optical carrier is stronger than the two sidebands,
making it the dominant term during the photodetection process thus we only detect
the first RF harmonic of the subcarrier. In the case of minimum transmission bias, in
which the carrier is suppressed, upper and lower subcarrier sidebands become the
dominant terms, and at the receiver they yield a single beating term at the second RF
harmonic frequency. For the BIDD bias operation, the optical carrier is reduced to a
level that generates first and second RF harmonics of the subcarrier signal with equal
RF powers at the photodetector.
The second RF harmonic of the subcanier signal is completely robust to dispersion
since it is generated by the single beating term of the optical domain upper and lower
subcarrier sidebands. The first harmonic of the subcarrier signal is the summation of
the beating terms between optical domain upper and lower subcarrier sidebands with
the optical carrier. Fig.7.2(a) shows that by adding these two subcarriers we can
achieve a dispersion-robust SCM system with at most 3 dB power fading. Fig.7.2(b)
shows the RF fading due to PMD for the first and second RF harmonics, which are
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the terms obtained for quadrature and minimum transmission bias, respectively. We
will show that by using first and second RF harmonic power, BIDD fades at most 6
dB due to PMD. Fig 7.2. (c) shows the calculated close form formulae for the total
RF power for different bias techniques where pis the modulation depth.
Modulator Transfer Function
lias
Laser
(Xc)
DC Bias
SCM PMD & CD
PD
* Modulator
Quadrature Bias
Minimum Bias
>own Convert
(fs,2 fs)
BIDD Bias
Optical
Spectra
RF
Spectra
LA
fs
t
Q—
21s
fs
2fs
Fig. 7.1: Bias-Induced Diversity Detection (BIDD) Concept: Generation of first
and second harmonics in the RF domain by beat terms of lower and upper SCM
sidebands and optical carrier (c) Total RF power of the different bias method.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
RP Fading due to Chromatic Dispersion
1st Harmonic, f
(a)
t
2B(I Harmonic, 2fs
+
Y
X
3 dB
r
Dispersion (ps/nm)
Dispersion (ps/nn^
Quadrature Bias
Minimum Bias
Dispersion (p s/n n ^
BIDD Bias
RF Facing due to 1st order PMD (worst case)
Jkl" Harmonic, f,
**.
I
(b) ft.
^
.♦*’
**« •a*
»»
Harmonic, 2fs
\ S\ f\
*•* •**
••
ABIDD Total RF Power
f\ f
(\Aa A /\^ /\A
\l \l M \l
— I ----------- 1--------►
+
1
1
1
---------------: —
1
b.
DGD (ps)
DGD (ps)
Quadrature Bias
Minimum Bias
=
6 dB
DGD (ps)
BIDD Bias
j ________Total RFJPower____________
BIDD
(c) Minimum
Quadrature
0 . 5 J o ( a x ) 2 J i ( a x ) 2 c o s 2 ( x D L C f , f2/fc2) C o s 2 (^fDGD)
+0.25J,(a;r)4C os2 (2^-fDGD)
.25Ji(a2T)4 C o s2 (2;rfD G D )
Jo(«;r)2Ji(a?r)2 cos(^rZ)ZCf,f2/fc2) 2 C o s2 (;rfD G D )
Fig. 7.2: (a) The 1st harmonic RF power fades due to CD while the 2nd harmonic
survives, (b) The 1st and 2nd harmonics both fade due to PMD. BIDD method
results in a robust optical system, (c) Total RF power considering both CD and
PMD
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.2 Experimental Results
7.2.1 Data Rate Performance
In order to compare the BIDD technique with conventional SCM transmission in
terms of modulation bandwidth we modulate an 8-GHz RF tone with OOK 215-1
PRBS data, at data bit rates varying from 55 Mb/s to 2.5 Gb/s (Fig.7.3). Due to
reduced bias value hence optical carrier power, minimum transmission and BIDD
bias operations have 3-dB better sensitivity than the quadrature bias case throughout
the different data bit rates. Our BIDD technique displays no distortion or limitation
due to the increasing bit rate.
-18
-20
Quadrature
Minimum
BIDD
ffl-22
-o
-28
-30
0
500
1000
1500
2000 2500
Bit Rate (Mb/s)
Fig. 7.3: Sensitivity of quadrature bias, minimum bias, and BIDD technique as a
function of bit rate
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.2.2 Chromatic Dispersion Induced RF Power Fading
In order to demonstrate the effect of chromatic dispersion on the SCM system
performance, we transmit a 55 Mb/s PRBS 210-1 data SCM signal over 40 to 120 km
of single mode fiber (SMF) with an average chromatic dispersion value of 16.5
ps/nm/km. We use three different subcarrier frequencies of 6, 7, and 8-GHz and
measure the photodetector power sensitivity at a bit error rate (BER) of lxlO'10 for
different bias operations of quadrature, minimum transmission and BIDD. Figure-4
shows the RF fading for each subcarrier frequency with respect to the fiber length for
different bias operations. The solid line in each figure is the simulation result.
Minimum transmission bias case has the best performance since it has a maximum
power penalty of 3 dB at different subcarrier frequencies. This is because the RF
power at the second harmonic is the result of a single beating term in the case of
minimum transmission bias, which in return makes it robust to the dispersion.
Maximum RF fading is limited to 6-dB using the BIDD technique. The 3-dB
difference between the BIDD and minimum transmission bias is because of the first
RF harmonic in the BIDD technique which can fade completely due to chromatic
dispersion. When compared to the quadrature bias operation, in which there can be
as much as 40 dB RF fading, both minimum and BIDD bias techniques are robust
with respect to chromatic dispersion.
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
.M. 1j l
:
■
6 GHz
- * - Quadrature
’T i i m i j muin
v
. i.... i .... i.. i »11
40
80
60
100
120
Fiber Length (km)
7 GHz
40
60
\
80
100
120
Fiber Length (km)
8 GHz
- Quadrature
Minimum
BIDD
100
120
Fiber Length (km)
Fig. 7.4: RF power fading as a function of fiber transmission distance
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.2.3 Polarization Mode Dispersion induced Power Fading
We test the PMD effect on the received subcarrier RF powers using a three-section
variable-DGD PMD emulator. The emulator consists of three variable DGD
elements separated by two fiber-squeezer-based polarization controllers. The DGD
values of the sections were changed according to Maxwellian distribution and the
polarization rotation between the sections was set to be uniformly random for each
PMD data point. The output DGD of this emulator has the same statistical
distribution as a real fiber[6].
8 GHz Subcarrier, Average DGD = 35 ps
350
300
350
Q uadrature
Bias
300
350
M inim um
Bias
250
250
250
200
200
200
150
150
5% T ail
100
5% T ail
B IDD
300
150
5% Tail
100
100
30
RF Power Fading
Fig. 7.5: RF power fading histograms for quadrature, minimum, and BIDD bias for
8-GHz subcarrier
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
6 GHz Subcarrier
Minimum '
— —Q uadrature \ ~
Minimum
— —Q uadrature
Minimum
—Q uadrature
BIDD
BIDD
BIDD
99,9 99 95 80 5030 10
8 GHz Subcarrier
7 GHz Subcarrier
1
.1
99.9 99 95 80
5030 10
1
.1
99.9 99 95 80 5030 10
1
.1
Probability Percent (%)
Fig. 7.6: RF Power fading distribution curves for different bias techniques and 6, 7
and 8-GHz subcarriers
The average DGD of the PMD emulator is 35-ps, and we use subcarriers of 6, 7, and
8-GHz. For the different bias operations of quadrature, minimum, and BIDD we
measure the RF power for 1000 independent polarization samples. Fig. 7.5 shows the
RF power fading distribution due to PMD for the different bias techniques for an 8GHz subcarrier. The average RF-fading for quadrature, minimum, and BIDD bias
operations are measured as 9.2, 4.2, and 2.2-dB respectively. Fig. 7.6 shows the RF
power fading distribution curves for 6,7, and 8-GHz subcarriers. As it can be seen,
the 5% tail for RF fading, which is 26-28 dB in the minimum and 19-22 dB in
quadrature bias cases is reduced to 12 dB in the BIDD technique. The 5% tail means
that 95% of the samples have lower RF-fading. The results conclude that BIDD bias
technique is very robust to PMD effects compared to other bias techniques.
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.2.4 RF Intermodulation Terms
BIDD
Minimum
Quadrature
hi -30
43
40
£
O
P i -50
fSj fs-
2fs, 2fs,
-30
-30
40
40
A-3
"5 #
S
-60
-60
-70 " T
7
Z
9
■ ■
11 13
7 7 -7 7 +
15
Frequency (GHz)
17
2fsj 2fs,
fSj fs2
-70 L
,‘f ’ 7
9
--— wl
11 13
15
Frequency (GHz)
■►-70
17
J JU-vrH
' q **1■
IF1'
\
fs,+fs2
f
7
9
11 13
15
17
Frequency (GHz)
Fig. 7.7 : RF Spectra for quadrature, minimum, and BIDD bias for two subcarriers at
7 and 8-GHz.
Fig.7.7 shows the received RF spectrum for a two-SCM system with subcarrier
frequencies at 7 and 8-GHz carrying 55 Mb/s OOK PRBS binary data for different
bias techniques with 100 percent modulation depth. In BIDD the inter-modulation
terms of 2fsi-fS2 and 2fS2-fsi increase by 3 dB when compared to quadrature bias. The
minimum transmission and BIDD technique introduce an additional inter-modulation
at fsi+fS2, which is due to beating between pairs of fsi and fS2 which are at the
opposite sides of the optical carrier. It is important to note that this extra term does
not affect the subcarrier frequency spacing or data bandwidth since it is located in
the middle of the second SCM harmonics which are separated from each other twice
the frequency difference of the first SCM harmonics, at 14 and 16-GHz compared to
7 and 8-GHz.
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
7.3 Conclusion
We demonstrate SCM transmission and detection that is robust to RF power fading
due to chromatic dispersion and PMD [82]. We achieve signal diversity at the
photodetector, without any additional optical or electrical channels. We detect both
the first and second harmonic at the receiver and reduce the CD fading from 40 to
less than 6 dB, and we show ~8-dB improvement for the 5% PMD probability tail.
This is also the first demonstration for SCM data detection from the second harmonic
subcarriers.
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Conclusion
In this dissertation, fiber effects such as chromatic and polarization mode dispersions
have been investigated and methods for compensation have been demonstrated. A
new and novel multiplexing technique, dispersion division multiplexing has been
introduced. This technique has the potential to double the number o f subcarrier
channels at exactly the same wavelength and subcarrier frequencies. Dispersion
induced NRZ data clock tone regeneration has been used in order to facilitate inline
and real time dispersion slope monitoring o f a fiber optic link. Also effect o f PPLN
based wavelength shifting on subcarrier signals have been investigated.
. As proposed in the related proposal, we also investigated the potential o f placing a
shadowed DDM header subcarrier within the digital baseband data spectrum. We
experimentally demonstrated the viable placement o f a 55 Mb/s bit rate header on a
7.7 GHz subcarrier for in band use with lOGb/s data packets. The resultant system is
demonstrated to be spectrally efficient. This allowed easy recovery o f the header
information without the need for faster than data rate photodetectors, and it has the
advantage o f header being processed with simple, low-speed electronics. We
demonstrated that when such a subcarrier header channel is Dispersion Division
Multiplexed (DDM) with a 10 Gb/s data channel it has negligible effect on the data
channel performance. And at a switching node demonstration, we recover and
downconvert the DDM subcarrier header to obtain the routing information. This
84
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
experiment demonstrates the possible advantage of subcarriers for inline header
processing of very high data rate traffic channels.
In addition to Chromatic Dispersion (CD) Polarization Mode Dispersion (PMD) may
have severe implications on millimeter-wave subcarrier multiplexed fiber-optic
transmission systems. Those optical systems may serve as fiber backbones for high­
speed wireless applications, such as wireless local area networks. Mobile and
wireless applications are becoming more and more important in today’s dynamic
information society. Hence it may be feasible to provide an optical solution to both
of these problems at the same time. To this extend, we studied and experimentally
demonstrated a novel new technique, Bias Induced Diversity Detection BIDD). By
altering the DC modulation bias voltage it is possible to generate an additional
second harmonic frequency counterpart to the original subcarrier that is being
transmitted. The resultant optical signal has the same optical bandwidth, and we do
not need extra electrical signal generators. From the principle of diversity, we also
detect and downconvert the second harmonic component. This process proved to
robust to CD and PMD respectively within optical 1.5 dB and 3 dB power budgets
respectively.
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bibliography
[77] Adamczyk O. H. et al, Microwave Theory and Techniques,Vol.49, pp. 19621967,2001
[62] Adamczyk O.H., Sahin A.B., Yu Q., Lee S., and Willner A.E., “Statistics of
PMD-induced power fading for double sideband and single sideband subcarriermultiplexed signals,” Optical Fiber Communication Conference and Exhibit, 2001.
OFC 2001, Volume: 1,2001 Page(s): M05
[16] Agrawal G. P., "Nonlinear fiber optics," 2nd Ed., New York: Academic Press,
1995.
[2]
Bergano N. S. and Davidson C. R., "Wavelength division multiplexing in long-
haul transmission systems," J. Lightwave Technol., vol. 14, pp. 1299-1308, June
1996.
[1]
Brackett C. A., "Dense wavelength division multiplexing networks: Principles
and applications," IEEE J. Select. Areas in Commun., vol. 8, pp. 948-964, Aug.
1990.
[70] Budman, E. Eichen, J. Schlafer, et al., OFC’92, paper Tu04.
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[57] Capmany J., Peral E., and Pastor D., “Formula for Two-Carrier Intermodulation
Distortion in Wavelength Converted Subcarrier Multiplexed Signals via Cross Gain
Modulation,” IEEE Photon. Tech. Lett, pp. 278-280, March 2000.
[52] Cardakli M.C., Sahin A.B., Adamczyk O.A., Willner A.E., Parameswaran
K.R., and Fejer M .M ., “All-optical wavelength shifting of subcarrier channels using
difference-frequency-generation in a PPLN,” Lasers and Electro-Optics, 2001.
CLEO '01. Page(s): 382 -383
[69] Chang G. K., Ellinas G., Gamelin J., et al., Journal of Lightwave Technology.,
vol. 14, pp. 1320-1340, 1996.
[60] Chou M. H., Brener I., Fejer M. M., Chaban E. E., and Christman S. B., “1.5mm-band wavelength conversion based on cascaded second-order nonlinearity in
LiNb03 waveguides,” IEEE Photon. Technol. Lett., 11, pp. 653-655 (1999).
[19] Chraplyvy R., "Limitations on lightwave communications imposed by optical
fiber nonlinearities," J. Lightwave Technol., vol. 8, pp. 1548-1557, Oct. 1990.
[66] Ciprut P., B Gisin., Gisin N., Passy R., Von der Weid J. P., Prieto F., and
Zimmer C. W., "Second-order polarization mode dispersion: Impact on analog and
digital transmissions," J. Light. Tech., vol. 16, pp. 757-771, (1998).
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[80] Corral J. L. et al. Microwave Theory and Techniques,Vol.49,pp.l9681976,2001
[55] Danielsen S.L., Hansen P.B., and Stubkjaer K.E., “Wavelength Conversion in
Optical Packet Switching,” IEEE/OS A J. Lightwave Tech., pp. 2095-2108, Dec.
1998.
[8]
Darcie T. E., "Subcarrier multiplexing for lightwave networks and video
distribution systems," IEEE J. Select. Areas in Commun., vol. 8, pp. 1240-1248, Sep.
1990.
[21] Darcie T. E., Dixon M. E., Kasper B. L., et al., “Lightwave system using
microwave subcarrier multiplexing,” Electron. Lett., vol. 22, pp. 612—613, 1986.
[5]
Desurvire E., Simpson J. R., and Becker P. C., "High-gain erbium-doped fiber
amplifier," Optics Letters, vol. 12, pp. 888-890, Nov. 1987.
[18] Feng K.-M., Cai J.-X., Grubsky V., Starodubov D. S., Hayee M. I., Lee. S.,
Jiang X., Willner A. E., and Feinberg J., “Dynamic dispersion compensation in a 10Gb/s optical system using a novel voltage tuned nonlinearly chirped fiber Bragg
grating,” IEEE Photon. Technol. Lett., vol. 11, pp. 373-375, Mar. 1999.
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[51] Feng K.-M., Grubsky V., Starodubov D. S., Cai J.-X., Willner A. E., and
Feinberg J., “Tunable nonlinearly-chirped fiber Bragg grating for use as a dispersion
compensator with a voltage-controlled dispersion,” in Tech. Dig. Conf. on Optical
Fiber Communication OFC’98, paper TuM3, pp. 72-74, San Jose, CA, Feb 22-27,
1998.
[56] Freeman P.N., Dutta N.K., and J. Lopata, “Semiconductor Optical Amplifier
for Wavelength Conversion in Subcarrier Multiplexed Systems,” IEEE Photon.
Tech. Lett., pp. 46-48, Jan. 1997.
[31] Fuster J. M., Marti J., Polo V., Ramos F., and Corral J. L., “Mitigation of
dispersion-induced power penalty in millimetre-wave fibre optic links,” Electronics
Letters, vol. 31, pp. 1869-1870, Sep. 1998.
[22] Georges J. B., Cutrer D. M. and Solgaard O., “10-Gb/s subcarrier multiplexed
transmission over 490 km of ordinary single-mode fiber without dispersion
compensation,” IEEE Photon. Technol. Lett., vol. 9, pp. 1658-1660,1997.
[40] Ghosh G., Endo M., Iwasaki T., "Temperature-dependent Sellmeier coefficients
and chromatic dispersions for some optical fiber glasses," J. Lightwave Technol.,
vol. 12, no. 8, Aug. 1994, pp. 1338-1342.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[50] Giles C. R., "Lightwave applications of fiber Bragg gratings," J. Lightwave
Technol., vol. 15, pp. 1391-1404, Aug. 1997.
[14] Gliese U., Norskov S., and Nielsen T. N., "Chromatic dispersion in fiber-optic
microwave and millimeter-wave links," IEEE Trans. Microwave Theory Tech., vol.
44, pp. 1716-1724, Oct. 1996.
[30] Gnauck H., Korotky S. K., Veselka J. J., et al., “Dispersion penalty reduction
using an optical modulator with adjustable chirp,” IEEE Photon. Technol. Lett., vol.
3, pp. 916-918,1991.
[29] Hakki B.W., “Dispersion of microwave-modulated optical signals,” Journal of
Lightwave Technol., vol. 11, pp. 474-480, 1993.
[32] Havstad S. A., Sahin A. B., Adamc2yk O. A., et al., “Distance-independent
microwave and millimeter-wave power fading compensation using a phase diversity
configuration,“ IEEE Photon. Technol. Lett., vol. 12, pp. 1052-1054,2000.
[36] Hayee M. I. and Willner A. E., “Pre- and post-compensation of dispersion and
nonlinearities in 10-Gb/s WDM systems,” IEEE Photon. Technol. Lett., vol. 9, Sept.
1997, pp. 1271-1273.
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[24] Heim P. J., McClay C. P., “Frequency division multiplexed microwave and
baseband digital optical fiber link for phased array antennas,” IEEE Trans.
Microwave Theory Tech., vol. 38 no. 5, pp. 494—500,1990.
[48] Hill K. O. and Meltz G., "Fiber Bragg grating technology: Fundamentals and
overview," J. Lightwave Technol., vol. 15, pp. 1263-1276, Aug. 1997.
[49] Hill K. O., Fujji Y., Johnson D. C., and Kawasaki B. S., “Photo-sensitivity in
optical fiber waveguides: Application to reflection filter fabrication,” Appl. Phys.
Lett., vol. 32 pp. 647-649,1978.
[64] Hofstetter R., Schmuck H., and Heidemann R., "Dispersion effects in optical
millimeter-wave
systems using
self-heterodyne method
for transport and
generation," IEEE Trans. Microwave Theory Tech., vol. 43, pp. 2263-2269, (1995).
[12] Ih C. S. and Gu W., "Fiber induced distortions in a subcarrier multiplexed
lightwave system," IEEE J. Select. Areas in Commun., vol. 7, pp. 1296-1303, Sep.
1990.
[65] Jopson R. M., Nelson L. E., Kogelnik H., and Foschini G. J., "Polarization
mode dispersion beyond first order," in Tech. Dig. IEEE LEOS 12th Annual Meeting
LEOS'99, pp. 149-150,(1999).
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[3]
Kawanashi S., Takara H., Uchiyama K., Shake I. and Mori K., "3 Tbit/s (160
Gbit/s x 19 ch) OTDM-WDM transmission experiment, " Conf. on Optical Fiber
Communication OFC’99, paper PD-1, San Diego, CA, Feb. 1999.
[17] Kazovsky L. G., Benedetto S., and Willner A. E., "Optical fiber communication
systems," Boston: Artech House, 1996.
[81] Lee J. H. et al, “High performance, 64-chip, 160 Gchip/s fiber grating based
OCDMA receiver incorporating a nonlinear optical loop mirror,” OFC-2002, ,ThA4,
2002.
[6]
Li T., "The impact of optical amplifiers on long-distance lightwave
telecommunications," Proc. IEEE, vol. 81, pp. 1568-1579, Nov. 1993.
[68] Lima T., Khosravani R., Ebrahimi P., Ibragimov E., Willner A. E., and Menyuk
C. R., "Polarization mode dispersion emulator," in Tech. Dig. Conf. Optical Fiber
Comm. OFC'2000, paper ThB4, pp. 31-33, (2000).
[38]
Madani F. M. and Kikuchi K., "Design theory of long-distance WDM
dispersion-managed transmission system," J. Lightwave Technol., vol. 17, no. 8,
Aug. 1999, pp. 1326-1335.
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[82] Marra C. et al, IEEE Photonics Technology Letters,Vol. 14, pp. 1345-1347,2002
[45] Marti J., Fuster J. M., and Laming R. I., “Experimental reduction of chromatic
dispersion effects in lightwave microwave/millimetre-wave transmissions using
tapered linearly chirped fibre gratings,” Electronics Letters, vol. 33, pp. -1170-1171,
June 1997.
[4]
Mears R. J., Reekie L., Jauncey I. M., and Payne D. N., "Low-noise erbium-
doped fibre amplifier operating at 1.54 mm," Electronics Letters, vol. 23, pp. 10261028, Sep. 1987.
[27] Narasimha A., X. J. Meng, M. C. Wu, et al., “A tandem single sideband fiber­
optic system using a dual-electrode Mach-Zehnder modulator,” CLEO 2000, pp. 9596,2000.
[9]
Olhansky R., Lanzisera V. A., and Hill P. M., "Subcarrier multiplexed lightwave
systems for broad-band distribution," J. Lightwave Technol., vol. 7, pp. 1329-1342,
Sep. 1989.
[41] Pan Z., Yu Q., Xie Y., Havstad S.A., Willner A.E., Starodubov D.S., and
Feinberg J., "Chromatic dispersion monitoring and automated compensation for
NRZ and RZ data using clock regeneration.and fading without adding signaling,"
Optical Comm. Conf, vol. 3, pp. WH5 1-3,2001.
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[61] Parameswaran K.R., Route R. K., Kurz J. R., Roussev R.V, Fejer M.M., and
Fujimura M., “Highly Efficient SHG in Buried Waveguides Formed using Annealed
and Reverse Proton Exchange in PPLN,” IEEE LEOS Annual Meeting, paper ThL3,
2001 .
[11] Park E. and Willner A. E., "Self-routing of wavelength packets using an alloptical wavelength shifter and QPSK subcarrier routing control headers," IEEE
Photon. Technol. Lett., vol. 8, pp. 938-940, July 1996.
[58] Park E. and Willner A.E., “All-Optical Wavelength Shifting of Combined
Baseband and Subcarrier Signals Using SOA Cross-Gain Compression, ” Conf. on
Lasers and Electro-Optics, CLEO 96, pp. 516-517,1996.
[25] Park E., Willner A. E., “Self-routing of wavelength packets using an all-optical
wavelength shifter and QPSK subcarrier routing control headers,” IEEE Photon.
Technol. Lett., vol. 8, no. 7, pp. 938-940,1996.
[13] Park J., Elrefaie A. F., and Lau K. Y., "Fiber chromatic dispersion effects on
multichannel digital millimeter-wave transmission," IEEE Photon. Technol. Lett.,
vol. 8, pp. 1716-1718, Dec. 1996.
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[63] Poole C. D. and Darcie T. E., "Distortion related to polarization-modedispersion in analog lightwave systems," J. Light. Tech., vol. 11, pp. 1749-1759,
(1993).
[53] Popa E., "Military and Aerospace Applications of Lihtwave Technology," IEEE
MTT-S Int., vol. 2, pp. 893-896, 1988.
[33] Ramos F., Marti J., “Comparison of optical single-sideband modulation and
chirped fiber gratings as dispersion mitigating techniques in optical millimeter-wave
multichannel systems,” IEEE Photon. Technol. Lett., vol. 11, pp. 1479-1482,1999.
[73]
Sahin A. B., Adamczyk O. H., Willner A. E., “Dispersion Division
Multiplexing for subcarrier multiplexed data channels,” OFC’2001, paper WCC4.
[84] Sahin A.B., Saghari P., Yan L.-S., Gurkan D., and Willner A. E., “Bias Induced
Diversity Detection (BIDD) for dispersion robust subcarrier transmission,” OFC 03,
paper FD6.
[20] Sahin A.B., Willner A.E., “Dispersion Division Multiplexing for In-Band
Subcarrier-Header-Based
All-Optical
Packet
Switching,”
Optical
Fiber
Communication Conference and Exhibit, 2002. OFC 2002, Page(s): WOl
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[35] Sahin A.B., Yan L.-S., Yu Qian, Hauer M., Pan Z., Willner A. E., “Dynamic
Dispersion Slope Monitoring of Many WDM Channels Using Dispersion-Induced
RF Clock Regeneration,” European Conference on Optical Communication, ECOC
2001. Pages: 308-310.
[28]
Schmuck H., ’’Comparison of optical millimetre-wave system concepts with
regard to chromatic dispersion,” Electron. Lett., vol. 31, pp. 1848-1849, 1995.
[39]
Shah V., Kim K.S., Morreale J.P., Chandrasekhar S., and Nichols V., “10 Gb/s
transmission across 4 optical cross-connect nodes and 200 km fiber, “ European
Conf. on Optical Comm., vol. 1, pp. 601-602,1998.
[72] Shell M., Vaughn M. D., Wang A., et al., IEEE Photonics Technology Letters,
vol. 8, no. 10, October 1996.
[54]
Shieh W. and Willner A. E., "A Wavelength-Routing Node by Using
Multifunctional Semiconductor Optical Amplifiers and Multiple-Pilot-Tone-Coded
subcarrier Control Headers," -IEEE/LEOS 1996 Summer Topical Meetings, pp. 1415,1996.
[79]
Smith G. H. et al, Microwave Theory and Techniques,Vol.45,pp.l410-
1415,1997
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[26]
Smith G.H., Novak, D., and Ahmed, Z. “Technique for optical SSB generation
to overcome dispersion penalties in fibre-radio systems,” Electron. Lett., vol. 33, pp.
74-75, 1997.
[10]
Su S. F., Bugos A. R., Lanzisera V., and Olshansky R., "Demonstration of a
multiple-access WDM network with subcarrier-multiplexed control channels," IEEE
Photon. Technol. Lett., vol. 6, pp. 461-463, Mar. 1994.
[46]
Sun H., Cardakli M. C., Cai J. X., Feng K. M., Long H., Hayee M. I., and
Willner A. E., “Tunable compensation of dispersion-induced RF power degradation
in multiple-channel SCM transmission by nonlinearly-chirped FBGs,” in Tech. Dig.
Conf. on Lasers and Electro-Optics CLEO’99, paper CWK2, pp. 316-317, Baltimore
MD, May 23-28,1999.
[34]
Sun H., Cardakli M.C., Feng K.-M., et al., “Tunable RF-power-fading
compensation of multiple-channel double-sideband SCM transmission using a
nonlinearly chirped FBG,” IEEE Photon. Technol. Lett., vol. 12 no. 5, pp. 546-548,
2000 .
[7] Tonguz O. K. and Jung H., "Personal Communications access networks using
subcarrier multiplexed optical links," J. Lightwave Technol., vol. 14, pp. 1400-1409,
June 1996.
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
[23] Wood T. H., Feldman R. D., and Austin R. F., “Demonstration of a costeffective broadband passive optical network system,” IEEE Photon. Technol. Lett.,
vol. 6, pp. 575-577,1994.
[37] Xie Y., Lee S., Pan Z., Cai J.-X., Willner A.E, Grubsky V.,- Starodubov D.S.,
Salik E., and Feinberg J., "Tunable compensation of the dispersion slope mismatch
in dispersion-managed systems using a sampled nonlinearly chirped FBG," IEEE
Photon. Technol. Lett., vol. 12, Oct. 2000, pp. 1417-1419.
[59] Yoo S.J.B., Chang G.K., Xin W., and Koza M.A., “Simultaneous MultiChannel Conversion of Analog and Digital Signals by Polarization Independent
Difference-Frequency-Generation,” Optical Fiber Comm. Conf., paper FB 5,1999.
[83] Yu C., Yu Q., Pan Z., Sahin A.B., Willner A.E., “Optical compensation of
PMD-induced power fading for single sideband subcarrier-multiplexed systems,”
OFC-2002, WQ5 ,2002
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Документ
Категория
Без категории
Просмотров
0
Размер файла
4 173 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа