Itu-T G.663
Itu-T G.663
Itu-T G.663
)45 4
'
TELECOMMUNICATION
STANDARDIZATION SECTOR
OF ITU
(10/96)
G.100G.199
).4%2.!4)/.!, !.!,/'5% #!22)%2 3934%GENERAL CHARACTERISTICS COMMON TO ALL ANALOGUE CARRIERTRANSMISSION SYSTEMS
G.200G.299
G.300G.399
G.400G.449
G.450G.499
G.600G.609
G.610G.619
G.620G.629
Submarine cables
G.630G.649
G.650G.659
' '
G.700G.799
DIGITAL NETWORKS
G.800G.899
G.900G.999
Summary
This Recommendation covers application related aspects of OFA devices and sub-systems, primarily
used in digital systems and is valid only for systems operating in the 1550 nm region. Applications
include both single-channel and multi-channel systems used in point-to-point and point-to-multipoint
configurations for use in long-distance networks and optical access networks. The purpose of this
Recommendation is to identify which aspects should be considered for each application and to
specify appropriate parameter values and ranges for each type of OFA device.
Source
ITU-T Recommendation G.663 was prepared by ITU-T Study Group 15 (1993-1996) and was
approved by WTSC (Geneva, 9-18 October 1996).
FOREWORD
ITU (International Telecommunication Union) is the United Nations Specialized Agency in the field of
telecommunications. The ITU Telecommunication Standardization Sector (ITU-T) is a permanent organ of
the ITU. The ITU-T is responsible for studying technical, operating and tariff questions and issuing
Recommendations on them with a view to standardizing telecommunications on a worldwide basis.
The World Telecommunication Standardization Conference (WTSC), which meets every four years,
establishes the topics for study by the ITU-T Study Groups which, in their turn, produce Recommendations
on these topics.
The approval of Recommendations by the Members of the ITU-T is covered by the procedure laid down in
WTSC Resolution No. 1.
In some areas of information technology which fall within ITU-Ts purview, the necessary standards are
prepared on a collaborative basis with ISO and IEC.
NOTE
In this Recommendation, the expression "Administration" is used for conciseness to indicate both a
telecommunication administration and a recognized operating agency.
ITU 1997
All rights reserved. No part of this publication may be reproduced or utilized in any form or by any means,
electronic or mechanical, including photocopying and microfilm, without permission in writing from the ITU.
ii
Recommendation G.663
(10/96)
CONTENTS
Page
1
Scope...........................................................................................................................
References...................................................................................................................
Terminology................................................................................................................
Abbreviations..............................................................................................................
Applications ................................................................................................................
5.1
5.2
Pre-amplifier ...............................................................................................................
5.3
Line amplifier..............................................................................................................
5.4
5.5
5.6
5.7
5.8
Point-to-point applications..........................................................................................
5.9
7.1
7.1.1
Point-to-point.................................................................................................
7.1.2
Point-to-multipoint ........................................................................................
Pre-amplifier ...............................................................................................................
7.2.1
Point-to-point.................................................................................................
Line amplifier..............................................................................................................
7.3.1
Point-to-point.................................................................................................
7.3.2
Point-to-multipoint ........................................................................................
Parameter values and ranges for OFA related aspects of OFA sub-systems..............
8.1
8.2
10
Appendix I Parameter values and ranges for OFA devices in single-channel point-to-point
systems........................................................................................................................
I.1
7.2
7.3
Recommendation G.663
(10/96)
iii
Page
I.2
Pre-amplifier ...............................................................................................................
II.1
Purpose........................................................................................................................
II.2
10
II.3
Optical non-linearities.................................................................................................
10
II.3.1
11
II.3.2
Four-wave mixing..........................................................................................
12
II.3.3
Modulation instability....................................................................................
15
II.3.4
16
II.3.5
Soliton formation...........................................................................................
17
II.3.6
18
II.3.7
18
II.3.8
19
19
II.4.1
19
II.4.2
22
II.4.3
23
24
II.5.1
Chromatic dispersion.....................................................................................
24
25
II.6.1
25
II.6.2
Self-filtering effect.........................................................................................
26
II.6.3
27
29
31
IV.1
OFA sub-systems........................................................................................................
31
IV.2
OFA devices................................................................................................................
31
II.4
II.5
II.6
iv
Recommendation G.663
(10/96)
Recommendation G.663
APPLICATION RELATED ASPECTS OF OPTICAL FIBRE
AMPLIFIER DEVICES AND SUB-SYSTEMS1
(Geneva, 1996)
1
Scope
This Recommendation covers application related aspects of Optical Fibre Amplifiers (OFA) devices
and sub-systems, primarily used in digital systems and is valid only for systems operating in the
1550 nm region. Generic characteristics of OFA devices and sub-systems are described in
Recommendation G.662. Applications include both single-channel and multi-channel systems used
in point-to-point and point-to-multipoint configurations for use in long-distance networks and optical
access networks.
The purpose of this Recommendation is to identify which aspects should be considered for each
application and to specify common parameter values and ranges for each type of OFA device
(booster amplifier, pre-amplifier, and line amplifier). This Recommendation pertains to the
development of new equipment including OFA devices and includes guidelines to apply OFA
devices and OFA sub-systems to existing equipment. As an objective, for systems specified as
transversely compatible, the degree of compatibility should not be changed by the use of OFA
devices.
Important topics contained in this Recommendation include transmission aspects, maintenance
aspects, and optical safety.
2
References
The following ITU-T Recommendations and other references contain provisions which, through
reference in this text, constitute provisions of this Recommendation. At the time of publication, the
editions indicated were valid. All Recommendations and other references are subject to revision; all
users of this Recommendation are therefore encouraged to investigate the possibility of applying the
most recent edition of the Recommendations and other references listed below. A list of the currently
valid ITU-T Recommendations is regularly published.
ITU-T Recommendation G.650 (1997), Definition and test methods for the relevant
parameters of single-mode fibres.
ITU-T Recommendation G.655 (1996), Characteristics of a non-zero dispersion singlemode optical fibre cable.
____________________
1
(10/96)
ITU-T Recommendation G.661 (1996), Definition and test methods for the relevant generic
parameters of optical fibre amplifiers.
ITU-T Recommendation G.681 (1996), Functional characteristics of interoffice and longhaul line systems using optical amplifiers, including optical multiplexers.
ITU-T Recommendation G.955 (1996), Digital line systems based on the 1544 kbit/s and the
2048 kbit/s hierarchy on optical fibre cables.
ITU-T Recommendation G.957 (1995), Optical interfaces for equipments and systems
relating to the synchronous digital hierarchy.
ITU-T Recommendation G.958 (1994), Digital line systems based on the synchronous
digital hierarchy for use on optical fibre cables.
IEC 825-2:1993, Safety of laser products Part 2: Safety of optical fibre communication
systems.
Terminology
For the purposes of this Recommendation, the definitions given in Recommendations G.661 and
G.662 apply.
4
Abbreviations
Amplitude Modulation
ASE
BA
BER
Bit-Error Ratio
CD
Chromatic Dispersion
CMC
Coherent Multi-channel
CW
Continuous Wave
DA
Dispersion Accommodation
EDFA
FDM
Frequency-Division Multiplexing
FWHM
FWM
Four-Wave Mixing
Recommendation G.663
(10/96)
LA
Line Amplifier
LOS
Loss of Signal
MI
Modulation Instability
OAM
OAN
OAR
OAT
OFA
OFDM
ORL
PA
Pre-Amplifier
PM
Phase Modulation
PDH
PDL
PHB
PMD
PSK
SBS
SDH
SMF
Single-Mode Fibre
SNR
Signal-to-Noise Ratio
SPM
SRS
TRE
WDM
XPM
Applications
Application of OFAs in optical transmission systems offers a number of advantages. Chief among
these advantages is the ability to realize very significant unrepeatered system lengths, as well as very
long unregenerated system lengths. Deployment of OFAs is likely to permit the retirement of many
existing conventional regenerator sites and, in the case of new routes, to render unnecessary the
construction of many new sites. OFAs also enable serious consideration of new optical system
architectures for application in terrestrial and submarine long haul, and access networks.
Two examples of this are wavelength division multiplexing and point-to-multipoint applications,
approaches heretofore generally considered prohibitively complex and expensive. OFAs also offer
potential advantages with respect to network upgrade options due to their independence from
modulation format and bit rate.
However, the application of OFAs also brings to light some new and potentially serious system
impairments, which result from the high power levels produced by the OFAs and the long distances
between regeneration. These transmission effects include optical fibre non-linearities, polarization
effects and effects due to the amplification characteristics of the OFA itself. Chromatic dispersion
Recommendation G.663
(10/96)
also becomes increasingly significant for the long unregenerated systems enabled by the OFA. In
addition to determining chromatic dispersion limitations, the dispersion characteristics of the fibre
influence the severity of the impairment produced by several of the dominant non-linear effects. As a
result, dispersion management has emerged as a critical consideration in system design. In the
following subclauses, the OFA applications are described and important considerations for each
application are identified.
If the characteristics are longitudinally compatible, the OFA devices specified in this
Recommendation can also be used to increase the distance in G.955 Plesiochronous Digital
Hierarchy (PDH) optical line systems (which are not transversely compatible). To ensure transverse
compatibility between OFA devices and Synchronous Digital Hierarchy (SDH) equipments and to
satisfy the requirements of G.690-Series Recommendations, either G.957 equipment in accordance
with 6.1/G.957 and 6.22/G.957 or a transponder(s) may be used.
5.1
The power (booster) amplifier (BA) is a high saturation-power OFA device to be used directly after
the optical transmitter to increase its signal power level. The BA does not need stringent
requirements for noise and optical filtering. The Operation, Administration and Maintenance (OAM)
functions for the BA may or may not be shared with the optical transmitter.
The application of BAs (often in conjunction with pre-amplifiers) is very attractive, especially in
those cases where intermediate locations with active equipment are either undesirable or
inaccessible, as in submarine systems. In any case, fewer intermediate locations implies easier
maintenance for the network operator. The most direct and simple means to increase the available
power budget is to use either a BA directly after the regular transmitter or an optically amplified
transmitter (see 5.4).
Because of the relatively high level of output power, the undesirable Amplified Spontaneous
Emission (ASE) noise, inherently present due to the statistical process of photon generation inside
the OFA, is usually negligible. However, application of BAs may result in fibre non-linearity induced
system penalties due to the high optical power levels produced by BAs and the long interactive
lengths provided by the optical path.
5.2
Pre-amplifier
The Pre-Amplifier (PA) is a very low noise OFA device to be used directly before an optical receiver
to improve its sensitivity. The requisite low level of ASE noise may be achieved through the use of
narrow-band optical filters. In this case, automatic tuning of the centre wavelength of the preamplifier filter to the transmitter wavelength would be advantageous, since it would permit the
relaxation of requirements on both the initial transmitter wavelength tolerance and its long-term
stability. As noted previously, the use of PAs (usually in conjunction with BAs) is a straightforward
means to realize significant increases in available power budget. The OAM function for PAs may or
may not be shared with the optical receiver. In STM-64 or higher level systems, the use of a PA-only
configuration can be useful (e.g. to avoid potential problems due to fibre non-linearities).
5.3
Line amplifier
The Line Amplifier (LA) is a low-noise OFA device to be used between passive fibre sections to
increase the regeneration lengths or in correspondence of a multipoint connection to compensate for
branching losses in the optical access network As noted previously, line amplifiers might replace
some or all conventional regenerators in long-haul fibre sections. It can be envisioned that more than
one conventional regenerator can be replaced by a single LA, with the evident advantage of reduced
equipment in transmission links. Furthermore, a situation can be envisaged where both line
Recommendation G.663
(10/96)
amplifiers for compensation of signal attenuation, and conventional regenerators for compensation of
signal distortion, appear in long-distance networks.
A separate communication channel must exist in systems using LAs, which would allow alarming,
supervision and control of the installed remote LAs. Such a supervisory channel should not place
restrictions on the selection of pump laser wavelength or operating window. Because each LA must
be able to insert its own status and alarm information, the supervisory channel must be recovered,
regenerated (with new information inserted) and retransmitted at each LA.
Theoretically, ultra-long (thousands of kilometres) transmission distances can be realized by
periodically inserting line amplifiers in the optical path. However, in the case where many OFAs are
cascaded, deteriorated system performance can occur due to noise accumulation, spectral dependency
of total gain, effects of polarization and chromatic dispersion and non-linear effects. Laboratory tests
have shown that the overall system behaviour in the case of many cascaded line amplifiers is much
more complex than that for the case of a few cascaded line amplifiers. In particular, the total gain of
a chain of line amplifiers in series is generally peaked around a specific wavelength, depending on
the specific amplifier configuration, giving considerable reduction of the usable OFA operating
wavelength range. Therefore, design of this type of system will be very much different from the
situation with only a few cascaded line amplifiers.
5.4
The Optically Amplified Transmitter (OAT) is an OFA sub-system in which a power amplifier is
integrated with the laser transmitter, resulting in a high-power transmitter. The connection between
the transmitter and the OFA is proprietary and shall not be specified. The application considerations
of OATs are generally the same as those for BAs (see 5.1).
5.5
Single-channel applications
BAs, PAs, LAs, OATs and OARs may all find application in single-channel systems. In the case of
single-channel transmission, noise accumulation can be reduced by using low noise OFAs in
combination with adequate band pass optical filtering. The dispersion limitations can be normally
minimized by operating close to the fibre zero dispersion wavelength or by using suitable Dispersion
Accommodation (DA) techniques. Furthermore, care must be taken to keep non-linear effects, like
stimulated Brillouin scattering (see II.3.1) and self phase modulation (see II.3.4), under control.
The main optical path interface parameter values for point-to-point, single-channel, long-haul
Synchronous Digital Hierarchy (SDH) systems utilizing OFAs are described in G.690-Series
Recommendations. The reference configurations and functional characteristics of these systems are
described in Recommendation G.681.
5.7
Multi-channel applications
OFAs will also find application in multi-channel systems. In addition to the transmission
impairments found in single-channel systems, multi-channel systems may also suffer degraded
performance due to certain non-linear effects. These include Four-Wave Mixing (FWM), Cross
Phase Modulation (XPM) and potentially Stimulated Raman Scattering (SRS). As a result, special
Recommendation G.663
(10/96)
precautions must be taken when designing multi-channel systems to avoid or alleviate these
impairments. In multi-channel systems using a series of LAs to provide long unregenerated lengths,
the effects of cascaded amplifiers, especially the reduction of the usable gain spectrum, must also be
considered.
The optical interface parameter values for point-to-point, multi-channel, long-haul SDH systems
utilizing OFAs are described in G.690-Series Recommendations. The reference configurations and
functional characteristics of these systems are described in Recommendation G.681.
5.8
Point-to-point applications
Schemes of insertion of OFA devices and sub-systems in point-to-point applications are indicated in
Recommendation G.662 (see in particular, Figures 1/G.662, 2/G.662 and 3/G.662).
5.9
Point-to-multipoint applications
OFAs can be used in Optically Amplified Networks (OANs) to increase the optical power budget or
to allow for higher splitting ratios in point-to-multipoint networks. Therefore, generally speaking,
both power and line amplifiers will be used in these types of networks. The use of OFAs in ring
networks just before branching devices, in order to compensate for splitting/branching losses, is an
example of a potential application.
NOTE One of the first applications of the OFAs may be for the distribution of video signals. At the
beginning, these systems will probably be analogue but they will gradually change into digital systems. In the
case of analogue video systems, additional requirements compared to the ones of the OFAs for digital
applications, are necessary in order to maintain adequate carrier-to-noise ratio and avoid signal distortion. If
AM modulated lasers are used, for instance, the OFA gain must be very flat, in order to avoid frequency to
intensity modulation conversions, leading to increased signal distortions. This effect can be minimized when
using Continuous Wave (CW) operated lasers in combination with external modulators. In this way
polarization effects could be minimized too. Also, special care must be taken to keep OFAs output power at
acceptable levels in order to reduce non-linear effects like SBS.
Certain considerations on Operation, Administration and Maintenance (OAM) aspects of OFAs are
given in Appendix IV.
7
In this clause, parameter values of OFA devices (BA, PA, LA) are provided, according to the lists
given in clauses 7/G.662, 8/G.662 and 9/G.662, with the aim of ensuring, as far as possible, optical
transverse compatibility amongst OFA sub-systems and OFA devices in point-to-point system
configurations.
7.1
7.1.1
Point-to-point
The output parameters for power amplifiers are system-specific and are specified in the relevant
system Recommendations (G.690-Series). Some example input parameter values for power
amplifiers can be found in Appendix I. The output parameters for power amplifiers, as described by
the interfaces in the relevant system Recommendations, are assumed to be applicable for PDH
systems as well.
Recommendation G.663
(10/96)
7.1.1.1
Single-channel systems
Some example input parameter values for power amplifiers can be found in Appendix I. Output
parameter values are specified in G.690-Series Recommendations.
7.1.1.2
Multi-channel systems
Input parameter values for power amplifiers to be used in digital multi-channel systems are for
further study. Output parameter values are specified in G.690-Series Recommendations.
7.1.2
Point-to-multipoint
Because applications in point-to-multipoint architectures requiring power amplifiers are more likely
to be analogue in nature, the associated parameter values are for further study.
7.2
Pre-amplifier
7.2.1
Point-to-point
The input parameters for pre-amplifiers are system-specific and are specified in the relevant system
G.690-Series Recommendations. Some example output parameter values for pre-amplifiers can be
found in Appendix I. The input parameters for pre-amplifiers, as described by the interfaces in the
relevant system Recommendations, are assumed to be applicable for PDH systems as well.
7.2.1.1
Single-channel systems
Some example output parameter values for pre-amplifiers can be found in Appendix I. Input
parameter values are specified in G.690-Series Recommendations.
7.2.1.2
Multi-channel systems
Output parameter values for pre-amplifiers to be used in digital multi-channel systems are for further
study. Input parameter values are specified in G.690-Series Recommendations.
7.3
Line amplifier
7.3.1
Point-to-point
7.3.1.1
Single-channel systems
Specific parameter values for line amplifiers in point-to-point single-channel systems are specified in
G.690-Series Recommendations.
7.3.1.2
Multi-channel systems
Specific parameter values for line amplifiers in point-to-point multi-channel systems are specified in
G.690-Series Recommendations.
7.3.2
Point-to-multipoint
Because applications in point-to-multipoint architectures requiring line amplifiers are more likely to
be analogue in nature, the associated parameter values are for further study.
Recommendation G.663
(10/96)
Parameter values and ranges for OFA related aspects of OFA sub-systems
8.1
Parameter values for optically amplified transmitters for single-channel and multi-channel systems
are specified in G.690-Series Recommendations.
8.2
Parameter values for optically amplified receivers for single-channel and multi-channel systems are
specified in G.690-Series Recommendations.
9
Environmental conditions
Various classes of environmental conditions are specified in IEC 721-3-0 (1984) and Amendment 1
(1987). Further details can be found in IEC 721-3-1 (1987), Amendment 1 (1991) and Amendment 2
(1993) for storage; IEC 721-3-2 (1985), Amendment 1 (1991) and Amendment 2 (1993) for
transportation; IEC 721-3-3 (1994) for stationary use, weather protected; and IEC 721-3-4 (1995) for
stationary use, non-weather protected.
10
Under certain condition (fibre break, open connectors), the optical output power of an OFA may
become accessible and may be hazardous to the human eye. Under IEC 825-1, the inherently safe
Class 1 power limits are 8.8 mW (+9.4 dBm) for the 1310 nm band and 10 mW (+10 dBm) for the
1550 nm band, while the limited safe Class 3 A levels (safe unless viewing aids are used) are
24 mW (+13.8 dBm) for the 1310 nm band and 50 mW (+17 dBm) for the 1550 nm band
respectively. Special precautions and requirements for installation and use of optical systems
(including amplifiers) and a description of viewing aids are given in IEC 825-2.
Due to the high level of optical power involved, mainly in booster amplifiers and OATs, optical
safety and optical surge generation issues (i.e. protection of personnel and equipment) are of primary
relevance in line systems using OFA devices and sub-systems.
An appropriate procedure for shut-down and restart of OFAs is given in Recommendation G.681.
APPENDIX I
Parameter values and ranges for OFA devices in
single-channel point-to-point systems
In this appendix, example parameter values for OFA devices (BA and PA) are shown for
single-channel point-to-point systems. These parameter values are according to the lists given in
clauses 7/G.662, 8/G.662 and 9/G.662 and were developed to ensure as far as possible compatibility
with parameter values specified in Recommendation G.957.
As indicated in clause 7, optical interface specifications for the main optical path are provided by
G.690-Series Recommendations.
I.1
The parameter values given in Table I.1 are examples of a minimum list of relevant input parameters
(as given in Recommendation G.662) for power amplifiers in point-to-point single-channel systems.
Recommendation G.663
(10/96)
Only the input parameters have been included in the table, since the output parameters are systemspecific and are specified in G.690-Series Recommendations.
Table I.1/G.663 Input parameters for power amplifiers in single-channel systems
Parameter
Unit
Value
dBm
6/+3
dBm
20
dB
27
dBm
15
dB
27
nm
1530-1565
a)
I.2
a)
The measured value of 15 dBm (max) takes into account all of the
contributions coming from the power distribution around the peak
value down to 30 dB below the peak value itself.
Pre-amplifier
The parameter values given in Table I.2 are examples of a minimum list of relevant output
parameters (as given in Recommendation G.662) for pre-amplifiers in point-to-point single-channel
systems. Only the output parameters have been included in the table, since the input parameters are
system-specific and are specified in G.690-Series Recommendations.
Table I.2/G.663 Output parameters for pre-amplifier in single-channel systems
Parameter
Unit
Value
dBm
16/9
dB
20
Noise figure
dB
under study
dB
27
dBm
nm
1530-1565
APPENDIX II
Transmission-related aspects
II.1
Purpose
OFAs permit the consideration of new optical transmission system architectures by providing very
high optical power levels and allowing much longer distances between regenerators. As a result, new
transmission effects and limitations have emerged which must now be considered. These include
optical non-linearities such as stimulated Brillouin scattering, four-wave mixing and self phase
modulation and dispersion properties such as polarization mode dispersion. In addition,
Recommendation G.663
(10/96)
characteristics of the OFA can also contribute to system impairments. Many approaches for solving
transmission-related problems have been investigated. However, the technologies differ in their
performance and application.
In order to design economical and reliable OFA-supported systems following the ITU-T
Recommendations on optical interface parameters, it is important that both operators and equipment
vendors share a common understanding regarding OFA-related problems and their remedy. To that
end, this subclause seeks to do the following:
Identify the limitations due to the mentioned effects in the various types of relevant
transmission systems (e.g. digital/analogue, coherent, wavelength division multiplexing,
etc.).
Indicate possible remedies to these impairments, together with the corresponding field of
effectiveness.
II.2
The factors influencing the OFA transmission applications can be divided into four general
categories: optical non-linearities, polarization properties, dispersion properties, and other
OFA-related properties. The factors dealt with in each category are given below and their effects are
described in the following subclauses:
a)
Optical non-linearities
1) stimulated Brillouin scattering;
2) four-wave mixing;
3) modulation instability;
4) self phase modulation;
5) soliton formation;
6) cross phase modulation;
7) stimulated Raman scattering.
b)
Polarization properties
1) polarization mode dispersion;
2) polarization dependent loss;
3) polarization hole burning.
c)
Dispersion properties
d)
chromatic dispersion.
II.3
Optical non-linearities
Non-linear interactions between the signal and the silica fibre transmission medium begin to appear
as optical signal powers are increased to achieve longer span lengths at high bit rates. Consequently,
10
Recommendation G.663
(10/96)
non-linear fibre behaviour has emerged as an important consideration both in high capacity systems
and in long unregenerated routes. These non-linearities can be generally categorized as either
scattering effects (stimulated Brillouin scattering and stimulated Raman scattering) or effects related
to the Kerr effect, that is, the intensity dependence of the refractive index (self phase modulation,
cross phase modulation, modulation instability, soliton formation and four-wave mixing). A variety
of parameters influence the severity of these non-linear effects, including fibre dispersion
characteristics, the effective core area of the fibre, the number and spacing of channels in multiple
channel systems, overall unregenerated system length, as well as signal intensity and source
linewidth.
II.3.1
II.3.1.1
In an intensity-modulated system using a source with a narrow linewidth, significant optical power is
transferred from the forward-propagating signal to a backward-propagating signal when the
Stimulated Brillouin Scattering (SBS) threshold is exceeded. In SBS, the forward-propagating light
is scattered from acoustic phonons. However, only the backward-propagating scattered light is
guided by the single-mode fibre. The scattered light is downshifted or Brillouin-shifted by
approximately 11 GHz at 1550 nm.
25
15
15
10
15
25
10
35
5
10
15
20
Of the non-linear effects described here, SBS has the lowest threshold power. While studies have
shown that the SBS threshold can vary between fibre types and even among individual fibres, it is
typically on the order of 5 to 10 mW for externally modulated, narrow linewidth sources, but may be
20 to 30 mW for directly modulated lasers. The SBS threshold for a system deployed on
Recommendation G.653 fibre is slightly lower than that for a system using Recommendation G.652
fibre, due to the smaller effective area of Recommendation G.653 fibre. This is generally true for all
of the non-linear effects. The SBS threshold is sensitive to the source linewidth and power level. It is
independent of the number of channels.
15
25
T1523360-96
II.3.1.2
SBS effectively limits the amount of light that can be transmitted through a fibre path. Figure II.3-1
shows this effect for a narrow-band source, where all of the signal power falls within the Brillouin
Recommendation G.663
(10/96)
11
bandwidth (see Figure II.3-1). The transmitted power becomes saturated and the backscattered power
rapidly increases. The input power level to the fibre at which this rapid increase occurs is defined as
the SBS threshold. In the general case, the SBS threshold is expressed as:
Pth = 21
KAeff
gLeff
p + B
B
where g denotes the Brillouin gain coefficient, and Aeff is the effective core area. K is a constant
determined by the degree of freedom of the polarization state (in Recommendation G.652 fibres,
K=2). B and p represent the Brillouin bandwidth and a linewidth of a pump light, respectively.
Leff denotes the effective length defined as:
Leff =
1 exp( L)
Pth = 21
II.3.1.3
KAeff
gLeff
Stimulated Brillouin scattering impairments will not arise in systems where the source linewidth
significantly exceeds the Brillouin bandwidth or where the signal power is below the threshold
power (calculated from expressions in II.3.1.2).
NOTE Further information concerning SBS can be found in the Appendix III/G.650.
f123,213
f113
f132,312
f112
f223
f231,321
f221
f1
f2
f332
f3
f331
Frequency
T1523370-96
II.3.2
Four-wave mixing
12
Recommendation G.663
(10/96)
phase modulation creates two sidebands at frequencies given by this difference. In the case of three
signals, more and stronger mixing products are produced (see Figure II.3-2) which will fall directly
on adjacent signal channels when the channel spacings are equal in frequency. Two optical waves
propagating along a fibre produce FWM with high efficiency if the phase matching condition is
achieved between sidebands and initial signals. For relatively low optical powers that means:
2 2 D2t L / c << 1
where is the frequency spacing among the wavelengths, D is the chromatic dispersion of the
fibre, t is the central wavelength, L is the fibre length and c is the speed of light in vacuum (losses
can be taken into account by slightly modifying the equation above). The efficiency of FWM is also
sensitive to overall optical power in the fibre. Indeed, for two waves with optical power P1 and P2,
the maximum parametric gain coefficient for the sidebands, gmax, can be estimated to be:
g max 2 g ( P1 + P2 )
2.0 nm
Channel spacing
1.5 nm
100
4-Channel capacity (Gbit/s)
1.0 nm
10
4
T1523380-96
II.3.2.2
The generation of FWM sidebands can result in significant depletion of the signal power.
Furthermore, when the mixing products fall directly on signal channels, they cause parametric
interference which manifests as amplitude gain or loss in the signal pulse, depending on the phase
interaction of the signal and sideband.
Parametric loss causes closure of the eye pattern at the receiver output, thereby degrading Bit-Error
Rate (BER) performance. Frequency spacing and chromatic dispersion reduce the efficiency of the
FWM process by destroying the phase matching between the interacting waves. Systems deployed
Recommendation G.663
(10/96)
13
over Recommendation G.652 fibre experience less FWM impairment compared to systems deployed
over Recommendation G.653 fibre. Conversely, the placement of a signal channel directly at or near
the dispersion zero can result in a very significant buildup of FWM products over a relatively short
fibre length (i.e. 10s of km). FWM is also sensitive to channel separation.
Four-wave mixing can create a serious system impairment in multiple wavelength systems on
Recommendation G.653 fibre, since the signal channels experience only a small value of chromatic
dispersion. The impact of dispersion on achievable system capacity for a four-channel system over
three amplifier spans is shown in Figure II.3-3. This illustrates what can happen at high power levels
when conditions promote the generation of mixing products. The capacity limitation is based on a
worst-case calculation of the mixing products generated by the FWM process with four 8 dBm signal
channels centred around the dispersion value shown. This system develops intolerable levels of
distortion due to FWM as the dispersion experienced by the signal channels approaches zero.
In single-channel systems, the FWM interaction can occur between OFA ASE noise and the
transmission channel, as well as between the main mode and the side modes of the optical
transmitter. Phase noise is added to the signal carrier due to the accumulated ASE via the fibre Kerr
effect, thus broadening the signal spectral tails.
II.3.2.3
As previously noted, dispersion, such as that in Recommendation G.655 fibres, may be used to
suppress the generation of the FWM sidebands. Uneven channel spacing may also be incorporated to
mitigate the severity of the FWM impairment. Reduction of the input power levels in
Recommendation G.653 fibre systems could permit multiple channel operation, but might
compromise the economic advantages of optical amplification.
In order to adequately suppress the generation of mixing products, use of a fibre [covered by either
existing or new Recommendations under study] with a minimum permitted (i.e. non-zero) dispersion
within the region of the OFA amplification band has been proposed. Alternating spans of such nonzero dispersion fibre with opposite dispersion characteristics has also been considered as a potential
option, since the resultant cable would maintain a net chromatic dispersion of approximately zero.
However, this alternative may present difficulties in the areas of installation, operations and
maintenance by introducing a second fibre type into the outside plant environment. Similar
approaches using long spans of fibre with small finite dispersion and short lengths of opposite and
higher dispersion fibre (to provide compensation) have also been demonstrated. In particular, in links
with periodical amplification, a short piece of compensating fibre can be located inside the box in
which the optical amplifier is located.
Uneven channel spacing and larger channel spacing have been proposed as means to mitigate the
effects of non-linearities and allow deployment of dense WDM systems on Recommendation G.653
fibre. Uneven channel spacing ensures that mixing products generated by three or more channels do
not fall directly on other channel wavelengths. However, the transfer of power from the signals into
the mixing products (i.e. signal depletion) remains unaffected by making the channel spacing uneven
and may still cause significant eye closure. Increased channel spacing also reduces the effects of
four-wave mixing. Use of these mitigation techniques may be constrained by the effects of gain
narrowing due to the concatenation of optical amplifiers, which reduces the width of the usable
amplification spectrum.
14
Recommendation G.663
(10/96)
1315
1317
1319
1321
1323
1325
T1523390-96
Wavelength (nm)
II.3.3
II.3.3.1
Modulation instability
Description of the effects
Modulation Instability (MI) breaks a CW signal or a pulse into a modulated structure. It can be
observed in the anomalous dispersion regime (i.e. above the zero-dispersion wavelength), where a
quasi-monochromatic signal spontaneously tends to generate two symmetric spectral sidebands, as
shown in Figure II.3-4. Frequency separation and gain of the sidebands are determined by the
intensity of the wave and by dispersion and non-linear coefficients of the fibre. The maximum
conversion efficiency occurs at a frequency separation given by:
max
8 2 cn P
2 o
= 3
A eff D( )
1/ 2
where n2 is the silica non-linear coefficient, A eff is the effective area of the fibre, P o is the
launched power, D( ) is the chromatic dispersion coefficient and is the operation wavelength;
sidebands located at max from the carrier experience a gain per unit length:
g max = 4Po / Aeff . Fibre loss can be taken into account by slightly modifying the equations
above. Dependence of MI gain on the frequency deviation with respect to the signal is given in
Figure II.3-5 in presence of fibre loss and for various values of fibre dispersion.
The MI can be viewed as particular case of FWM where two photons of the intense incoming signal
are converted into two photons at two different frequencies.
Recommendation G.663
(10/96)
15
15
10
5
T1523400-96
0
400
0
200
Frequency detuning (GHz)
200
400
Figure II.3-5/G.663 Calculated MI gain versus frequency detuning, from 30 km long fibres
(0.24 dB/km loss) with five different dispersion coefficient values,
for +16 dBm CW signal launched power
II.3.3.2
Modulation instability may decrease the Signal-to-Noise Ratio (SNR) due to generation of sidebands
either spontaneously or seeded by the amplifier spontaneous emission. As the maximum degradation
of the signal is expected for high values of gmax and for close to the bandwidth of the signal, MI
may be critical when using very powerful boosters in dispersion-shifted fibre links with directly
modulated lasers. On long-distance unrepeatered systems, MI can be observed at lower power levels
and may cause excess amplification of the spontaneous emission noise of the cascaded OFAs. This
Kerr-effect-induced broadening at signal spectral tails can cause signal-carrier depletion and the tails
may be attenuated by the narrow-band ASE filters or by the self-filtering effect in very long systems.
II.3.3.3
The effect of MI can be minimized either by decreasing the power level or by operating at
wavelengths below the zero-dispersion wavelength of the link. Dispersion managing may also be
useful to reduce MI sidebands formation. Otherwise, the received signal should be electrically
filtered in order to lower the level of the spurious amplified noise. External modulation of the lasers,
giving narrower spectra, may decrease considerably the impact of MI.
II.3.4
II.3.4.1
Because of the Kerr effect, the temporal variation of the optical intensity of the signal induces a
modulation of its own phase. This effect is called Self Phase Modulation (SPM). In single
wavelength systems, Self Phase Modulation will gradually broaden the signal spectrum when
changes in optical intensity result in changes in phase (see Figure II.3-6). Once spectral broadening is
introduced by SPM, the signal experiences a greater temporal broadening as it propagates along the
length of the fibre, due to the effects of chromatic dispersion, in the normal dispersion region of the
fibre (i.e. below the zero-dispersion wavelength). Conversely, in the anomalous dispersion region,
the chromatic dispersion and SPM can compensate each other, giving less temporal broadening. The
soliton propagation is based on this phenomenon (see II.3.5).
16
Recommendation G.663
(10/96)
l(t)
Kerr effect
n = n0+ n2I
t
dl/dt
dn/dt
d/dt = 2 f
t
f(t)
T1523410-96
II.3.4.2
Generally, the effects of SPM are significant only in systems with high cumulative dispersion or very
long systems. Systems which are dispersion-limited may not tolerate the effects due to SPM. In
multiple-channel systems with very closely spaced channels, the spectral broadening induced by
SPM may also create interference between adjacent channels.
SPM on low-chirp intensity-modulated signals leads to pulse compression on Recommendation
G.652 fibres and on Recommendation G.655 fibres with anomalous dispersion as a function of
transmitter power. The pulse compression counteracts the chromatic dispersion and offers some
dispersion accommodation. However, limits of maximum dispersion and related transmission length
exist.
II.3.4.3
The use of Recommendation G.653 fibre and the placement of the signal channel near the dispersion
zero will reduce the impact of SPM. For systems less than approximately 1000 km, SPM may be
controlled through the implementation of dispersion compensation at appropriate intervals along the
length of a Recommendation G.652 fibre system. The effects of SPM may mitigated by either
minimizing the input optical power or operating at wavelengths above the zero-dispersion
wavelength of Recommendation G.655 fibre.
II.3.5
II.3.5.1
Soliton formation
Description of the effects
In the anomalous dispersion region of the fibre (i.e. above the zero-dispersion wavelength), the
interplay between Kerr non-linearity and chromatic dispersion may give origin to particular
propagation regimes, called soliton pulses, which propagate for long distances periodically
reproducing their time shape and frequency spectrum. Other kind of signals can spontaneously
evolve into solitons. Although the fundamental soliton is very robust and can be used to achieve
long-distance transmission, the higher order solitons undergo very complex evolution, involving
Recommendation G.663
(10/96)
17
pulse reshaping, spectral modifications, and also pulse breaking. These effects may be detrimental to
system operation.
II.3.5.2
Effects due to soliton formation may be relevant both in Recommendations G.652, G.653 and G.655
fibre systems. While fundamental soliton formation can be useful, other solitons generally give rise
to a strong degradation of the transmitted signal. Thus, higher order soliton formation sets a limit to
the maximum power that can be launched into the fibre.
II.3.5.3
Soliton formation can be avoided by operating at wavelengths below the zero-dispersion wavelength
of the link. However, in this regime, soliton transmission is not supported and both dispersion and
non-linearity contribute to pulse broadening. Signal degradation can be minimized by proper
management of the dispersion along the link.
II.3.6
II.3.6.1
In multiple wavelengths systems, Cross Phase Modulation (XPM) will gradually broaden the signal
spectrum when changes in optical intensity result in changes in phase due to interactions between
adjacent channels. The amount of spectral broadening introduced by XPM is related to the channel
separation, since the dispersion-induced differential group velocities will cause the interacting pulses
to separate as they propagate down the fibre. Once spectral broadening is introduced by XPM, the
signal experiences a greater temporal broadening as it propagates along the length of the fibre due to
the effects of chromatic dispersion.
II.3.6.2
Impairments from XPM are more significant in Recommendation G.652 fibre systems, relative to
Recommendations G.653 and G.655 fibre systems. The broadening due to XPM may result in
interference between adjacent channels in multiple-channel systems.
II.3.6.3
XPM can be controlled through appropriate selection of channel spacing. Studies have shown that
only adjacent channels contribute significantly to XPM-induced signal distortion in multiple-channel
systems. The Signal-to-Noise Ratio (SNR) of the centre channel of a three-channel system will
approach that of a single-channel system as channel separation is increased. As a result, the effect of
XPM can be rendered negligible with adequate spacing between the signal channels. Channel
separations of 100 GHz were shown to be sufficient to reduce XPM effects in a simulation of a
system with 5 mW of power/channel. Dispersion penalties due to XPM may also be controlled by the
implementation of dispersion compensation at appropriate intervals along the length of the system.
II.3.7
II.3.7.1
Stimulated Raman scattering is a broadband effect which involves the interaction of light and the
vibrational modes of silica molecules. SRS causes a signal wavelength to behave as a Raman pump
for longer wavelengths, either other signal channels or spontaneously scattered Raman-shifted light.
In any case, the shorter wavelength signal is attenuated by this process, which amplifies the longer
wavelength signal.
18
Recommendation G.663
(10/96)
II.3.7.2
Stimulated Raman Scattering (SRS) can occur in both single- and multiple-channel systems. Signal
powers on the order of 1W or more are needed to experience impairment from this phenomenon with
only a single channel without line amplifiers. However, shorter wavelength signals in multiplechannel systems with widely spaced channels can suffer degraded signal-to-noise performance when
a portion of their power is transferred to longer wavelength channels through SRS. This results in
total system capacity limitations based on the total number of channels, channel spacing, average
input power and overall system length. In particular, the threshold for the observation of a 1 dB
penalty in a multi-channel system due to Raman gain in dispersion-unshifted fibre can be estimated
to be:
Ptot Leff < 40mW nm Mm
where Ptot is the combined power of all of the channels, is the optical spectrum over which the
channels are distributed and Leff is the effective length [in units of 106 metres (Mm)]. The SRS
threshold for a system deployed on Recommendation G.653 fibre is slightly lower than that for a
system using Recommendation G.652 fibre, due to the smaller effective area of Recommendation
G.653 fibre. SRS does not practically degrade single-channel systems; conversely it may limit the
capability of WDM systems.
II.3.7.3
In single-channel systems, filters can be used to remove the unwanted spectrum. However, no
practical techniques to eliminate the effects of SRS in multiple-channel systems have been reported.
The effects of SRS may also be mitigated by reducing the input optical power. However, SRS does
not appear to present a practical limitation to the deployment of currently contemplated WDM
systems.
II.3.8
A summary of the non-linear effects described in previous subclauses is given in Table II.1.
II.4
Polarization properties
II.4.1
II.4.1.1
It is well known that the fundamental mode of a circularly symmetric dielectric waveguide is doubly
degenerate. In a real optical fibre this degeneracy is split by birefringence. The birefringence may be
introduced deliberately, as in polarization-maintaining fibre for example, or it may be an unwanted
by-product of fibre manufacture or cable manufacture. In this case the birefringence is introduced in
a random way by, for example, geometrical or stress-induced perturbations.
The propagation constants, i ( ) , of the two orthogonal modes can be expanded in a Taylor series
around the centre frequency, o ;
i ( o ) = i ( o ) +
=o
( o ) +
1 2i
2 2
( o )2 +L
=o
2i
i
is related to the group velocity vg, and
2
related to the dispersion of the group velocity (or chromatic dispersion, D) etc.
Recommendation G.663
(10/96)
is
19
With the development of dispersion-shifted fibres and the deployment of systems operating near the
dispersion-zero wavelength, the contribution to the dispersion from the second order term, or
chromatic dispersion, reduces and the first order term can now become significant. For the case of
birefringent fibres, this first order term leads to a group delay called polarization dispersion. This
polarization dispersion introduces a differential group delay between orthogonal states of
polarization. Although the effect of Polarization Mode Dispersion (PMD) is to change randomly the
polarization state of a pulse propagating in a fibre, it is possible to define a pair of orthogonal states,
or principal states, which, when launched at the input to fibre, will emerge unchanged at the output.
These states are, to first order, independent of wavelength. (In some situations, however, this
approximation falls apart and the principal states can show a wavelength dependence, leading to a
further system degradation through a coupling to chromatic dispersion.)
Table II.1/G.663 Non-linear optical effects in glass fibres
Non-linear
optical effect
Cause
Self Phase
Modulation
(SPM) and Cross
Phase Modulation
(XPM)
Optical Kerr
effect: intensitydependent
refractive index
Characteristics
Phase shift
Impact
Spectral
broadening
increases effect of
dispersion
Power/dispersion
limited high
bit-rate
transmission
Pulse compression
positive
dispersion
Pulse propagation
(Solitons)
Limitations in PSK
systems by AM/PM
conversion
Raman Lines
f-n f (Stokes)
f = 12 THz
= 70 nm (1310 nm)
= 102 nm (1550 nm)
Signal power
Raman amplification
depletion
in a WDM system
with critical channel
spacing,
Stimulated
Brillouin
Scattering (SBS)
Interaction:
photon-acoustic
phonons
Brillouin lines
in backward direction
f f
f = 13.2 GHz (1310 nm)
f = 11.1 GHz (1550 nm)
Pc> ~5 mW
(for narrow
linewidth optical
source)
Pc increases with
signal line width
Signal instability
Optical loss in fibre
Optical crosstalk in
bidirectional
coherent multichannel systems
Four-photon
mixing or FourWave Mixing
(FWM)
Multi-photon
interaction
Mixing products
generated
f4 = f1 f2 f3
Optical crosstalk in
WDM systems
Signal power
depletion
20
Recommendation G.663
(10/96)
As alluded to in the first subclause above, the birefringence introduced to the fibre is caused by local
random and asymmetric mechanisms such as stress, bending and twisting. These random
birefringence mechanisms redefine the local birefringence axes along the length of the fibre, thus
causing random coupling between the polarization modes along the length of the fibre. The cabling
process also introduces a certain amount of random birefringence and random mode coupling. The
fibre length between such changes is usually referred to as the coupling length, which for a fibre is
usually quoted as the ensemble average of all of the local coupling lengths. (In a Hi-bi fibre the
concept of the beat length could be loosely regarded as a similar concept to the coupling length of
fibres.) Furthermore, changes in local environmental conditions, such as temperature for example,
cause fluctuations in the local birefringence axes, thus causing random polarization coupling. As a
result of the randomly changing polarization coupling, the magnitude of the differential group delay
becomes a statistically varying function. It can be shown that distribution of differential group delays
is described by a Maxwellian distribution function, defined by:
4 2
exp 2
P( ) = 2
2
3
32 2
where is the differential group delay between the two principal states, and is the mean
differential group delay. As a consequence of the statistical nature of polarization mode dispersion,
the magnitude of the dispersion increases with the square root of the fibre, or cable, length, for
lengths much longer than the coupling length. Polarization mode dispersion is usually quoted in units
of ps or ps/km. The unit of ps is usually reserved for single optical elements which have a fixed
dispersion (e.g. a coupler or isolator) or short fibre sections which do not exhibit mode coupling.
II.4.1.2
In a digital transmission system, the principal effect of polarization mode dispersion is to cause
intersymbol interference. As an approximate rule of thumb, a 1-dB penalty occurs for a total
dispersion equal to 04 T , where T is the bit period. This is the accepted value for the maximum
tolerable system power penalty. Although still unresolved, current studies indicate that optical fibres
and cables will be specified according to the mean level of polarization mode dispersion, a view
reflected by studies of single- and multi-channel systems where the mean level will also be specified.
This corresponds to a mean differential group delay equal to one-tenth of a bit period, 01 T .
Computer simulations have predicted that if the polarization mode dispersion has a Maxwellian
distribution with a mean value not exceeding 01 T , there is less than a 10 9 probability that the
system power penalty will exceed 1 dB.
Furthermore, in long-haul amplifier systems employing polarization scramblers (devices which
deliberately modulate the polarization state of a signal laser so that it appears to be unpolarized), the
polarization mode dispersion causes an increase in the degree of polarization of the signal. This
degrades system performance through interactions with polarization-dependent loss and polarization
hole burning (see following subclauses). In an analogue system, the interaction of polarization mode
dispersion with laser chirp leads to a second order distortion proportional to the modulation
frequency. A further second order penalty, independent of modulation frequency, is incurred when
additional polarization-dependent loss is present in the system.
It has also been shown, and mentioned briefly above, that a second order effect can cause a coupling
between polarization mode dispersion and chromatic dispersion. This is caused by the wavelength
dependence of the differential group delay. This leads to a statistical contribution to the chromatic
dispersion. This is an area which is not well understood and is under study.
Recommendation G.663
(10/96)
21
II.4.1.3
Given that the problem arises from induced birefringence, much of the effort in reducing the effects
of polarization mode dispersion have been concerned with minimizing the birefringence introduced
by fibre or cable manufacture. Care is taken to optimize fibre production to ensure concentricity of
the fibre core. Optical cables are manufactured using materials and processes which minimize the
residual strain in the cable structure across the fibre core. Elaborate cable structures can also be used
which introduce a circular component to the induced birefringence. By careful design, such an effect
can counteract linear birefringence to produce a cable with a resultant zero polarization mode
dispersion. Typically, the mean polarization mode dispersion of fibres and cables lie in the range:
0 < < 0.5
ps/km
Another method makes use of the concept of principal states which was introduced earlier. In this
scheme, a polarization controller is inserted at the input and output ends of the system. A
polarization beam splitter follows the output polarization controller and is used to generate an error
signal. The output polarizer searches for the error signal and the input polarizer is adjusted to
minimize this error signal. At the point of no error signal, the input polarization state is one of the
principal states for the system. Using such a technique up to 1-bit period of delay has been
compensated for in a 5 Gbit/s system. A similar technique has been applied to coherent
Frequency-Division Multiplexing (FDM) systems.
II.4.2
II.4.2.1
Polarization Dependent Loss (PDL) arises from dichroism of the passive optical components such as
isolators, couplers etc. in the signal path. When the signal passes through the dichroic element, the
component of its electric field parallel to the lossy axis is attenuated. As in the case of polarization
mode dispersion, the axes which define the polarization dependent loss are oriented randomly with
respect to each other.
II.4.2.2
To examine the effect of polarization dependent loss, let us examine a possible system configuration.
In amplified systems, one mode of amplifier control is to operate at constant signal power. Both the
signal and noise are affected by polarization dependent losses. However, because the noise is
unpolarized, the signal and noise are affected differently. The noise can be resolved into a component
parallel to the signal and a component orthogonal to the signal. It can be shown that the combined
effect of PDL and optical amplification is always to increase the component of the noise orthogonal
to the signal. Furthermore, the magnitude of the orthogonal noise component changes with time as
the signal polarization changes due to polarization mode dispersion. This leads to a reduction in the
signal-to-noise ratio and the Q-value at the receiver. Furthermore, the fluctuations time lead to fading
of the signal-to-noise and Q-value at the receiver, both of which lead to an impairment in system
performance.
In analogue systems, the polarization dependent loss can interact with laser chirp and polarization
mode dispersion to reduce the system performance in terms of composite second order distortion. As
expected, this impairment is time varying and leads to fluctuations in the system composite second
order with time.
II.4.2.3
As for the case of polarization mode dispersion, it is important that the polarization dependent loss of
the optical components are minimized. However, it should be noted that the impact of polarization
dependent loss on the system performance increases as the number of amplifiers increases. In long-
22
Recommendation G.663
(10/96)
haul submarine systems, for example, the requirements are extremely tight, because the number of
amplifiers can be several hundred. In a short-haul terrestrial system, where only a few amplifiers are
concatenated, the impact of polarization dependent loss on system performance is still under study.
Polarization modulation, or scrambling, has been shown to improve system performance by reducing
the fluctuations and improving the average Q. This technique is described more fully in II.4.3.3.
II.4.3
II.4.3.1
Polarization Hole Burning (PHB) results from an anisotropic saturation created by a polarized
saturating signal launched into the erbium fibre. This results in a selective depopulation of excited
states aligned with polarized field. Consequently, the available gain in the orthogonal direction is
higher. Although the erbium ions are distributed randomly within the glass matrix, on a microscopic
level the dipole associated with the erbium ion is anisotropic. The polarization hole burning effect is
maximum where the linearly polarized saturating signal is aligned with the major axis of the dipole
and is reduced where the polarization state of the saturating signal is elliptical or circular. Both the
signal laser and the pump laser contribute to the total effect, the total differential gain being the
vector sum of the two contributions. The degree of hole burning is proportional to the degree of
polarization of the saturating signal. For an unpolarized saturating signal there is no hole burning. In
principle, this is similar to the case of a circularly polarized signal.
II.4.3.2
The polarization hole burning impacts the system performance by causing the noise build-up along
the amplifier chain to be greater than that which would be predicted from simple linear theory. That
is, the signal-to-noise is reduced by polarization hole burning and, as for the cases of polarization
mode dispersion and polarization dependent loss, the measured Q fluctuates in time. As there are two
contributions to polarization hole burning, there are two ways in which the system performance is
affected. The total effect is proportional to the gain saturation, increasing with an increased degree of
saturation.
First of all, let us consider the effect of the polarized pump laser. It can be considered for the
purposes of this discussion that the pump polarization is fixed and invariant. The pump causes a
differential gain in the direction orthogonal to its polarization axis. Noise aligned orthogonally to the
pump experiences a higher gain than noise aligned with the pump. However, the polarization axes of
the pump lasers in each amplifier along a chain are uncorrelated with each other. Therefore, the
cumulative effect is similar to a random walk, and the pump-induced polarization hole burning can
be considered as a contribution to the polarization dependent loss of the amplifier. Thus, averaged
over a number of amplifiers, the noise build-up should be linear as expected from simple theory.
The signal laser induced polarization hole burning is slightly different. As the signal laser propagates
along the system, the noise polarized along parallel to the signal laser will see the same gain as the
signal. However, noise polarized orthogonal to the signal laser will always experience a higher gain
because it will always be orthogonal to the signal polarization axis. Therefore, the total noise will
increase in a non-linear way along the chain of amplifiers.
The total differential gain due to polarization hole burning varies as the polarization state of the
signal changes (due to polarization mode dispersion) along the amplifier chain. It varies because the
signal hole burning effect is correlated with the pump effect. As the relative polarization states of the
signal and pump lasers change, the magnitude of the differential gain changes. Therefore, although
the total noise increases non-linearly along the chain, it does so in such a way that the total noise
fluctuates in time. Consequently, as explained above, the signal-to-noise is reduced and fluctuates in
time. The system Q is, therefore, reduced and fluctuates in time.
Recommendation G.663
(10/96)
23
II.4.3.3
There exist several ways to reduce the effect of polarization hole burning. Operating the amplifiers in
the small signal regime is one potential solution, but this is not always possible and, in many cases,
not desirable. In practice, the simplest solution is to use a depolarized signal. A depolarized signal
can be created in many ways, but is most commonly generated by polarization scrambling. Using a
phase modulator, the polarization state is varied between two orthogonal states in time. The signal
then appears to be depolarized.
It has been shown that it is optimal to impose the polarization modulation at twice the bit rate. This is
because polarization dependent loss in the amplifier converts the polarization modulation to
amplitude modulation. By polarization modulating at twice the bit rate, the amplitude fluctuations
are at a rate higher than the detector bandwidth and so are not seen by the receiver. Using such
techniques, the performance of very long haul systems have been improved to the point where the
predicted performance is met with a high degree of confidence. Polarization modulation is now a
standard implementation in transoceanic amplified systems.
However, in long amplified systems, polarization mode dispersion causes a re-polarization of the
signal, thus allowing polarization hole burning to again degrade the system performance. Such an
effect illustrates the complex nature of the interaction of polarization phenomena in amplified links.
II.5
Dispersion properties
II.5.1
Chromatic dispersion
II.5.1.1
Chromatic dispersion is predominantly a transmission capacity limitation which results from the
transmission source spectral characteristics and the chromatic dispersion of the optical fibre. Optical
amplifiers inherently do not significantly change the chromatic dispersion, unless a form of
dispersion compensation function is incorporated into an amplified sub-system.
Optical amplifiers produce light around the signal transmission wavelength, known as Amplified
Spontaneous Emission (ASE). Usually the ASE component is not modulated and so it is not detected
synchronously with the signal. In many optical amplifier implementations, the ASE is blocked by a
filter within the amplifier, signal path or receiver. Unless the ASE is modulated along with the
transmission signal, its presence results in optical noise as described more fully in II.6.1. The broader
spectrum of the ASE does not alter the interaction of the signal spectral characteristics with the
optical fibre chromatic dispersion.
Generally, optical amplifiers do not significantly alter the overall chromatic dispersion by their
addition to a system. There is a small amount of chromatic dispersion added by the rare-Earth doped
fibre used as the active gain medium in an OFA, but those fibre lengths are on the order of tens of
metres to a few hundred metres. The chromatic dispersion of the rare-Earth doped fibres is not
significantly different from that found in Recommendations G.652, G.653 and G.655 fibres. For
systems which are tens to hundreds of kilometres in length, this dispersion contribution is considered
negligible.
II.5.1.2
Generally, optical amplifiers do not alter transmission limitations due to chromatic dispersion.
Certain non-linear effects may interact with the fibre chromatic dispersion due to the high output
amplified power levels. These effects and methods to minimize induced limitations are described
more fully in II.3.
24
Recommendation G.663
(10/96)
II.5.1.3
The presence of an optical amplifier does not effect the chromatic dispersion in a system and thus, no
specific methods to minimize these effects are required. However, OFAs do enable consideration of
long unregenerated systems, where system penalties due to chromatic dispersion may become
significant.
In some optically amplified sub-systems, a passive dispersion compensation function may be
combined with the optical amplifier to result in an amplified sub-system which adds a finite amount
of chromatic dispersion to the system, with a sign opposite that of the system fibre. This results in
lower overall system chromatic dispersion. This function may be co-located with an OFA in order to
overcome the losses associated with the passive dispersion compensation function. The changes to
the chromatic dispersion in the transmission system are due to the presence of the passive dispersion
compensation function (more fully described in G.690-Series Recommendations), and not the optical
amplifier.
NOTE Additional techniques to minimize the limitations induced by chromatic dispersion are described in
G.690-Series Recommendations.
II.6
II.6.1
Noise accumulation
II.6.1.1
In transmission systems with cascaded OFAs, the ASE noise generated at an OFA repeats a cycle of
attenuation and amplification in the same way as the signal light. Since the incoming ASE noise is
amplified at each OFA and added to the ASE noise generated at that OFA, the total ASE noise power
increases nearly proportionally with the number of OFAs, and the signal power accordingly
decreases. The noise power can exceed the signal power.
The ASE noise spectral profile also evolves along the system length. When ASE noise from the first
OFA is input to a second OFA, the gain profile of the second OFA changes due to the ASE noise
power via the gain saturation effect. Similarly, the effective gain profile of the third OFA is then
modified by the output-power spectrum of the second OFA. Such an effect is transmitted all the way
down to the last OFA. The ASE noise accumulates even if narrow-band filters are used at each OFA
because the noise exists over frequency ranges that include the signal frequency.
II.6.1.2
ASE noise accumulation affects the SNR of the system because the degradation in the receivedsignal SNR is due predominantly to ASE-related beat noise. Such beat noises increase linearly with
the number of OFAs. Thus, the error rate worsens with increasing numbers of OFAs. In addition,
noise accumulates exponentially with the magnitude of the amplifier gain.
As a result of the gain spectrum of the OFA, the ASE noise spectrum after many OFAs tends to have
a peak at a wavelength due to the self-filtering effect, which will be described in II.6.2. In particular,
if a closed all optical ring network architecture is considered, the ASE noise accumulates as if an
infinite number of OFAs were cascaded. Although the accumulation of ASE noise in filtered systems
is considerably reduced by the filters, in-band ASE still increases with the number of cascaded
OFAs. Thus, the SNR degrades as the number of OFAs increases.
II.6.1.3
The ASE noise accumulation can be reduced by decreasing the OFA spacing (while maintaining the
total gain equal to the total loss of the transmission path), since ASE noise accumulates
Recommendation G.663
(10/96)
25
exponentially with the magnitude of the amplifier gain. One of the following filtering techniques can
further reduce the unwanted effect of ASE noise: use of ASE noise filters, or use of the self-filtering
effect (self-filtering method).
The self-filtering method is applicable for systems with several tens of OFAs or more. In this
method, the signal wavelength is aligned with the self-filtering wavelength so that the ASE noise
received at the detector is reduced, just as if a narrow-band filter had been used. This is most
effective when used with shortened OFA spans and low-gain OFAs to reduce the initial ASE noise.
The self-filtering method is not applicable if an all-optical WDM closed ring network is considered.
In fact, the resulting peak in the OFAs overall gain spectrum may strongly affect the system
performance. In this case, the accumulation of the ASE noise can be minimized with the ASE-filter
method, which is obtained by filtering the WDM channels not dedicated to the network node before
switching them out of the node.
For systems with fewer OFAs, the self-filtering method is less effective than the ASE-filter method.
The ASE-filter method allows flexibility in the choice of the signal wavelength and provides other
advantages (see II.6.2). Care must be taken in the selection of the filter characteristics, since the
cascaded-filter passband is narrower than the passband of a single filter (unless it has a rectangular
spectral passband). Conventional filters with a full width at half maximum (FWHM) on the order of
3 nm could be used in long-distance single-channel systems.
II.6.2
II.6.2.1
Self-filtering effect
Description of the effect
As a result of the ASE noise accumulation in non-filtered systems, the characteristic profile of the
ASE spectrum (or the overall gain spectrum) tends to have a peak. The peak spectral linewidth
narrows with increasing numbers of OFAs until it finally saturates after some number of OFAs. This
may result in only 2~3 nm wide spectral linewidth after several tens of OFAs. This effect is called
self-filtering.
The self-filtering effect is determined by the spectral shape of the emission and absorption
cross-sections and by the degree of inversion of OFAs. The self-filtering wavelength may change
with changes in the host glass composition, input optical power, or inter-amplifier loss and their
dependence on wavelength, pump wavelength and the length of the doped fibre. The self-filtering
effect can generally be considered desirable in single-channel systems (as noted in II.6.1.3), but
undesirable in multi-channel systems.
II.6.2.2
For systems with a limited number of OFAs, the spectral width of the self-filtering gain peak remains
broad and does not reduce ASE noise accumulation, even with the signal wavelength adjusted to the
peak wavelength. In systems with many OFAs and a well-developed self-filtering gain peak, the SNR
can be high but may degrade if the signal wavelength shifts from the self-filtering wavelength. This
shift can occur after system reconfiguration or repair because of changes in inter-amplifier loss.
Multi-channel systems with cascaded OFAs can suffer from power variation among channels that
exponentially increases with the number of OFAs (the number of OFAs is the exponent). For
example, the power spread for a five-channel system of ~3 dB after the first erbium doped fibre
amplifier (EDFA) increased to ~15 dB after the sixth EDFA. For multi-channel systems, changes in
the total number of channels result in gain-spectral changes which perturb other channels. Saturationinduced spectrally-dependent gain also generates gain variation among the channels.
26
Recommendation G.663
(10/96)
Therefore, the loss-budget margin must be carefully designed to accommodate such EDFA-gain
spectral changes both in filtered and non-filtered systems. The use of ASE filters eliminates the
problem as described in II.6.2.3.
II.6.2.3
This subclause discusses the use of both OFA self-filtering effects and ASE noise filters to improve
system performance.
Using the self-filtering method to improve SNR is most effective when the optimum dispersion
wavelength, self-filtering wavelength, and signal wavelength coincide. This method does not require
the use of ASE noise filters, which can bring degradation associated with polarization-dependent
losses in the filter. This is particularly true in transoceanic submarine systems. On the other hand,
reliance on the self-filtering effect complicates system design, reconfiguration and repair because the
signal wavelength must always meet the changeable self-filtering wavelength. The spectral
characteristics of both OFA gains and inter-amplifier losses should be as uniform as possible.
Otherwise, the self-filtering gain peak may not become sufficiently narrow, thus degrading the SNR
improvement and making it more difficult to prevent ASE noise-induced saturation in long-haul
systems.
To avoid such disadvantages, the ASE-filter method can be used to reduce ASE accumulation with
the filter passband adjusted to the signal frequency. Narrow-band filters with FWHM < 1 nm are
commercially available. The ASE noise accumulation is then minimized and the system can be freed
from restrictions such as complexities of system design, reconfiguration and repair, and the
requirements of short OFA span and uniform performance of each OFA.
In multi-channel systems, the inter-channel power spread due to the self-filtering effect can be
avoided by amplifying each channel in a physically separated OFA. This method, however, demands
a costly demultiplexer, separate OFAs and a multiplexer. An alternative is to provide optical channel
power equalization at each network node, even if this method demands additional control electronics
and a more critical power budget through the network. An additional method is to use a less saturated
or less strongly inverted OFA, because this makes the attenuation less wavelength-dependent and
accordingly, reduces inter-channel power spread. In a strongly inverted EDFA cascade, however, the
ASE grows with the number of EDFAs at the first gain peak of ~1530 nm and needs to be eliminated
with a short-wavelength-eliminating filter. In addition, the pre-emphasis method minimizes SNR
differences for all channels by adjusting the transmitter optical powers for each channel based on the
received signal information from the end terminal.
II.6.3
II.6.3.1
When optical signal input power rapidly increases in sufficiently pumped EDFAs, optical surge
occurs because of the slow gain dynamics of EDFAs. The optical surge is likely to occur particularly
from output-power-controlled EDFAs. This is because the pump power for such EDFAs increases
with decrease in the optical input power and optical surge tends to occur when the input power
suddenly recovers. Figure II.6-1 illustrates optical surge generation.
When reconnecting a conventional connector, an optical-power rise time (10-90%) was ~0.3 ms,
which is short enough to cause an optical surge. If the EDFA is already optically saturated, however,
additional rapid increase in optical input power may not cause significant optical surges.
Recommendation G.663
(10/96)
27
Input
power
Input
Output
Gain
Erbium-doped
active fibre
Unsaturated
gain
Time
Operating
gain
Time
Optical
surge
Output
power
Constant pump
power
Time
T1523420-96
II.6.3.2
In EDFA cascades, under particular conditions, the optical-surge peak power can range up to several
watts, possibly damaging O/E converters and/or electrical pre-amps and, optical components
including optical-connector end faces. Optical surge is also harmful to human eyes. The threshold for
damage for a contaminated optical-connector end face has been observed to be at optical power
levels of about 20 dBm.
II.6.3.3
to reduce or shut down the pump power to EDFAs when the loss of input signal is detected;
2)
to recover the pump power level to EDFAs when the recovery of the input signal is detected.
In this respect, the reactivation of the pump laser power level up to its nominal value should
happen gradually and a related minimum reactivation time should be defined.
Figure II.6-2 shows an example of configuration that achieves surge prevention, where a loss-ofsignal detection unit controls the pump laser on/off status. In case of a loss of signal just for a time
sufficiently shorter than the gain dynamics of the EDFA, e.g. < 1 ms, the pump laser power needs to
be kept unchanged.
Erbium-doped
active fibre
Output
Input
Loss-of-signal
detection unit
Pump
laser
T1523430-96
28
Recommendation G.663
(10/96)
APPENDIX III
Pre-amplifier parameters
Definition of the relevant characteristic parameters of the pre-amplifier which allow, by
standard formulae, to obtain the minimum input power which ensures a BER = 1012 at the
Terminal Receiver Equipment (TRE)
On the basis of theoretical considerations and experimental tests related to optical amplifier noise,
the relevant PA characteristic parameters which allow to obtain the minimum input power which
ensures a BER = 1012 at the TRE are:
The optimum amplifier (without internal reflection noise) noise is dominated by two principal
components in addition to the contribution of the shot noise of the signal and spontaneous emission
noise. These two principal components are the signal-spontaneous emission beat noise and the
spontaneous emission-spontaneous emission beat noise.
NOTE The output noise terms can be subdivided into five categories:
a)
b)
c)
d)
e)
Reflection noise.
Each of the noise contribution can be expressed in the form of a partial noise figure "F", where the total noise
figure is the sum of all individual contributions. The total noise figure (in linear, not in logarithmic units) is:
Ftotal = Fshot-sig + Fshot-ASE + Fsig-sp + Fspsp + Frefl
This equation can be used both for evaluating the noise figure from optical measurements, as well as for
estimating the influence of such parameters as gain, reflections, source line width and baseband frequency. It
represents a complete noise figure model of an OFA.
In these conditions it is possible to obtain, by standard formulae, the following expression for total
noise figure:
hvn sp B sp sp G 1
NF = NFs sp 1 +
2 P in
G
where:
n2
G 1
NFs sp = 2n sp
and n sp =
G
n2 n1
h is the signal photon energy, Pin is the optical power at PA input, n2 is the population density of the
excited level, n1 is the population density of the base level and G and Bsp-sp have been defined before.
It is clear that NF reduces, by reducing the Bsp-sp value, but, for compatibility with existing
Recommendation G.663
(10/96)
29
Recommendation G.957, it has to be limited at values of the order of 1 nm (120 GHz). The SNR at
the TRE is given by the following formula:
m 2 P in
SNR =
hv B NF +
N Rx
Pin G 2
ExtRatio
to the bit rate) and N Rx is the TRE equivalent noise. From the previous two formulae it is possible
to obtain the minimum input power of the PA for BER = 10 12 on the TRE. It is clear that the
optimum PA must have a sufficiently high gain in order to make negligible the noise contribution of
the TRE compared to the pre-amplifier one. The SNR value for BER better than 1012 must be
maintained in all the input power range. The PA will be characterized, from the noise point of view,
by defining small signal gain (G), signal-spontaneous emission contribution to the noise figure
(NFs-sp)and spontaneous emission-spontaneous emission beat noise equivalent bandwidth (Bsp-sp). It is
important to verify the total noise figure of the PA in order to be sure that no contributions from
internal reflection noise are affecting the PA.
Example
Calculation of the SNR ratio for a receiver equipped with a PA able to ensure a BER = 1012 at
2.5 Gbit/s SDH-TRE
On the basis of the previous considerations and the previously defined parameters, the SNR at the
output of the PA can be calculated from the following formula:
m 2 Pin
SNR( PA) =
hv B NF
In the above formula it has not considered the shot noise contribution because it is negligible
compared to the beating noise components, having defined a small signal gain of 20 dB. This
formula assumes the worst case approximation where the noise on the zeros is considered equal to
that on the ones (zeros do not present a signal-spontaneous emission beat noise) and where the
output noise is considered Gaussian.
From the above-mentioned formula, it is easy to obtain the sensitivity of the PA.
The overall sensitivity of the composite receiver (PA + TRE) is obtained adding to the divisor of the
N Rx
previous formula the term
, in which N Rx is the TRE equivalent noise.
Pin G 2
SNR( PA+TRE ) =
30
Recommendation G.663
N Rx
Pin G 2
(10/96)
m2 Pin
hv B NF +
N Rx
Pin G 2
Taking into account the compatibility in accordance with Recommendation G.957 for a 2.5 Gbit/s
TRE with a minimum sensitivity of 26 dBm for BER = 1012 and the characteristic parameters of
the PA previously considered:
24 dB
20 dB
5.5 dB
1 nm (0.8 1.2)
3 dB
8.2 dB
2.5
=
GHz
min. value
2
2.5
15
. GHz
average value =
2
GHz
maximum value = 2.5
It is easy to establish that for values of the receiver signal of 16 dBm the contribution of the TRE
noise can be considered negligible. With the parameter values previously considered, a minimum
input power of 36 dBm for BER = 10 12 at the TRE is obtained. This value can be considered
inclusive of the margin due to the worst case hypothesis introduced in the calculations. Consequently
to the maximum gain value, the output signal from the PA will be 12 dBm, allowing the sensitivity
of the PA to be considered as that of the composite receiver (PA + TRE). Considering for the PA the
minimum gain operating conditions (small-signal gain, G = 20 dB), the average optical power at the
output of the PA will be 16 dBm allowing again the sensitivity of the PA to be considered as that of
the composite receiver (PA + TRE).
By using the previous values and formulae, it is also possible to compute that a pre-amplifier output
SNR ratio better than 13 dB is achievable with a pre-amplifier input power level down to 43 dBm,
when a configuration without line amplifiers is used.
APPENDIX IV
OAM aspects of OFAs
Possible generic OAM aspects for OFA sub-systems and devices are given below.
IV.1
OFA sub-systems
Because OATs and OARs from a maintenance perspective are not different from conventional
transmitters and receivers, the associated Operation, Administration and Maintenance (OAM)
aspects are identical to those of transmitters and receivers specified in Recommendation G.783.
IV.2
OFA devices
The following OAM conditions can be presented as information to the involved management system:
OFA Loss of Signal (LOS), indicating possible loss of input signal (transmission defect);
OFA failure, indicating that an optical amplifier can no longer perform its basic function of
optical amplification (equipment defect);
Recommendation G.663
(10/96)
31
This distinction is made to provide the user with information on the equipment and/or cable plant.
Furthermore, OFAs can be provided with an "enable/disable output" control signal in order to be able
to limit optical power levels possibly needed to satisfy optical safety requirements.
NOTE The use of further sources of equipment failure like "OFA degradation" or "Loss of optical
supervisory channel" (in case of the presence of line amplifiers) is for further study.
32
Recommendation G.663
(10/96)
Series B
Series C
Series D
Series E
Overall network operation, telephone service, service operation and human factors
Series F
Series G
Series H
Series I
Series J
Series K
Series L
Construction, installation and protection of cables and other elements of outside plant
Series M
Series N
Series O
Series P
Series Q
Series R
Telegraph transmission
Series S
Series T
Series U
Telegraph switching
Series V
Series X
Series Z
Programming languages