ch4_ofts
DESCRIPTION
Optical fiberTRANSCRIPT
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 1
Chapter 4
Wavelength Division Multiplexed Systems
2011/2012
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 2
Objectives
• Introduce WDM systems
• Provide knowledge on the main types of WDM systems
• Provide knowledge on the main limitations and design of point-to-point WDM systems
• Provide knowledge on the architecture, elements and main
impairments of WDM networks
Bibliography
Chapters 5, 7 and 13 of “Optical Networks”
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 3
λλλλ1, λλλλ2, ... ,λλλλNλλλλ1, λλλλ2, ... ,λλλλN
1 optical fibre
AmplificationSection
Basic Elements and Requirements of a (point-to-point) WDM System
Tunable optical sources
with reduced linewidth
WavelengthWavelengthCombinerCombiner
WavelengthWavelengthSplitterSplitter
Challenge:Low insertion loss
Challenge:High selectivity
in the optical domain(to separate channels with low crosstalk)
Individual OpticalReceivers
(One per wavelength channel, color blind)
= PIN + electrical receiver
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 4
WDM Multiplexing / Demultiplexing
• Methods– Selective: uses AWG (loss independent of number of wavelengths)– Non-selective: uses combination of optical filters (loss dependent of
number of wavelengths)
• Multiplexers / Demultiplexers: passive (reciprocal) devices– Technologies:
• Gratings• Fibre Bragg gratings• Arrayed waveguide gratings (AWGs)
• Techniques for high count multiplexing– Multi-stage (per wavelength band)– Interleaving
These technologies and techniques have been addressed in chapter 2
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 5
Tunable Lasers• Why tunable? (... and not fixed-wavelength?)
– More expensive
– With fixed-wavelength lasers, a 100-channel WDM system needs 100 different laser types � inventory and sparing issues from manufacturers, system providers to network operators
– Tunable lasers are also one of the key enablers of reconfigurable
optical networks:
• they provide the flexibility to choose the transmit wavelength at the source of a lightpath; we need as many tunable lasers in a network node as the number of lighpaths
• The tuning time required for such applications is on the order of milliseconds because the wavelength selection happens only at the times where the lightpath is set up, or when it needs to be rerouted in the event of a failure
[ON] section 3.5.3
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 6
Tunable Lasers (2)
• Tuning mechanisms:
– Injection current into a semiconductor laser– Temperature tuning– Mechanical tuning
[ON] section 3.5.3
• Ideal tunable laser
– can tune rapidly over a wide continuous tuning range of over 100 nm.
– should be stable over its lifetime and easily controllable and manufacturable.
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 7
Optical Channel Density
• WDM provides a number of uniformly spaced frequency-slots for optical channels
– The system operator may not populate all of them with signals– The spacing of these frequency slots determines the potential optical
channel density
Any channel spacing can be chosen but in practice most WDM systems fall into two specific categories based on industry standards.
Adoption of standards is important to guarantee the “communication” between different manufacturers’ equipments and reduce manufacturing costs
ch ,maxch
ch
Maximum number of channels (channel capacity)in a given bandwidth, :
: channel spacing (assumed the same over the whole bandwidth)
WDM
WDM
BBN
νν
=∆
∆
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 8
Categories of WDM: CWDM and ...• CWDM (Coarse Wavelength Division Multiplexing)
– Based on channel spacing of 20 nm across a range from 1260 to 1620 nm set by ITU standard G.694.2 � 18 channels (centered at 1250 + i×20 nm) across the low-loss window of dry fibres (bands O, E, S, C and L)
– For use in metro networks where in many cases data rates to 2.5 Gbit/s are used and transmission distances are at most tens of km � optical amplifiers are not required.
– Avoids high costs associated with precise wavelength control (uncooled lasers and mux/demux with weaker selectivity can be used)
CWDM spacingis uniform in wavelength but not in
frequency units
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 9
Categories of WDM: ..... and DWDM
• DWDM (Dense Wavelength Division Multiplexing)
– Uses channel spacing of 200 GHz or less
– Normal center-frequency spacings are 200, 100, 50 or 25 GHz based on a standard grid developed by ITU developed for the EDFA band
– Most DWDM systems operate in the EDFA band (around 1550 nm): used for long-haul, high-capacity transmission
DWDM spacingis uniform in
frequency but not in wavelength
units
Difference in system capacity between CWDM and DWDM is dramatic !!!
... while 40 100-GHz DWDM channelsfit into the EDFA C-band (1530-1565 nm),
two CWDM channels do not quite fit into the same band
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 10
DWDM Frequency Grid by ITU
Wavelength, nm
Fibr
e lo
ss p
aram
eter
,a, d
B/k
m
ITU-T G.692: different frequency spacingsbetween adjacent channels; grid anchored
at 193.1 THz (1552.52 nm); only C and L bands
Secondwindow
~1270-1350nm
Thirdwindow
~1480-1600nm
Alternativechannel spacings
• 25 GHz (~0.2 nm)• 50 GHz (~0.4 nm)• 100 GHz (~0.4 nm)• 200 GHz (~1.6 nm)
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 11
Wavelength, nm
EDFA
Gai
n,dB
Third Window Bands
C Band(Conventional)
~ 1530-1565nmBandwidth ~35nm
L Band(Long λλλλ)
~ 1565-1625nmBandwith: ~60nm
Total available bandwidth~ 90 nm
⇓⇓⇓⇓
~ 112 channels(spaced by 100 GHz,
about 0.8 nm)
Signals are split between a pair of parallel amplifiers (one for each band) with a 5 nm gap between
C and L bands. L band EDFA amplification can be achieved with longer (100 m or more) doped fibre.
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 12
Top-capacity Commercial (P2P) DWDM Systems
160 λλλλs × 10 Gb/s
80 λλλλs × 40 Gb/s
1.6 Tb/s
3.2 Tb/sTransXpress
InfinityNokia Siemens
Networks
160 λλλλs × 40 Gb/s6.4 Tb/sOPTera Long Haul 5000Nortel
64 λλλλs × 40 Gb/s
128 λλλλs × 10 Gb/s
2.56 Tb/s
1.28 Tb/sLambdaXtremeLucent
160 λλλλs × 10 Gb/s
640 λλλλs × 2.5 Gb/s1.6 Tb/sCoreStreamCiena
Number of wavelengthsCapacityEquipment
DesignationSystem Vendor
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 13
Channel Spacing Requirements• What is the required channel spacing?
– Ideally, the WDM demultiplexer should transmit all light (no loss) at the center wavelength of the optical channel, and block adjacent channels completely.
– In practice, the demux has some attenuation at the center of the channel (typically, 3 to 5 dB), and adjacent channels are attenuated by 20 to 40 dB.
– Actual transmission of the demux depends on the technology used and spreads out more than over the adjacent channels � origin of crosstalk (XTalk) between optical channels.
εi =pc,i/p0
Signal
Interchannel Crosstalk
p0
pc,i
ν
Dem
ux tr
ansm
ittan
ce (d
B)
ν1 ν2 νS νNch
pc,2
pc,1
νi ...... ......
( )( )
dem
20 dem S
Demux with transfer function,
Signal power at demux output: : input power per channel (same for all channels)
H
p H pp
ν
ν≈
( ) 2, dem
, 0
10 0 ,
Power from XTalk channel : Normalised XTalk power from channel :
Difference in dB between signal and XTalk levels(channel suppression, dB): 10log
c i i
i c i
i c i
i p H pi p p
P P
νε
ε
≈
=
− = −
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 14
Demux Bandwidth Requirements• The crosstalk can be reduced by using demuxes with bandwidth narrower
than the optical channel bandwidth (channel spacing) � reduces the amount of leakage power of other optical channels detected by the PIN
• ... but the bandwidth should be wide enough to transmit the signal and cope with drift of source wavelength
� required bandwidth is due to signal bandwidth (2Rb,ch) and
drift of nominal emission wavelength (2∆νch/5)
Rule of thumb for minimum -3 dB Mux and Demux bandwidths:
B-3 dB = 2Rb,ch + 2∆νch/5
Rb,ch = channel bit rate∆νch = channel spacing
� ITU-T specifies a maximum drift of ±∆νch/5 for a channel spacing ≥ 200 GHz(higher drift causes increase of crosstalk and signal loss)
This means
B-3 dB < ∆νch
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 15
Crosstalk (XTalk) Modeling
( ) ( ) ( )[ ] ( ) ( )[ ]
( )( )
0 S S S 0
S
Electric field at the PIN input (just one channel interferer):
= 2 cos 2 2 cos 2
={0, 1}, depending on whether a 0 or 1 is being sent in the desired channel; = {0, 1},
i i i i
i
E t p d t t t p d t t t
d td t
πν φ ε πν φ⋅ + + ⋅ +
( ) ( )S
S
depending on whether a 0 or 1 is being sent in the XTalk channel; and is the optical frequency of the signal and XTalk carriers; and are the random phases of the signal and XTalk channels
(
i
it tν ν
φ φit is assumed that all channels have an infinite extinction ratio)
( ) ( )
( ) ( )
S ,1 0
S ,0 0
Worst case XTalk (lower power level for 1 and higher power level for 0)
For bit 1, 1, and 0 (worst case):
For bit 0, 0, and 1 (worst case):
i i
i i i
d t d t p p
d t d t p pε
= = =
= = =
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )0 S 0 0 S S S
Power incident on PIN within the receiver bandwidth (proportional to the squared electric field)
= 2 cos 2i i i i i i ip t p d t p d t p d t d t t t tε ε π ν ν φ φ+ + ⋅ − + −
( ) ( ) ( )
S
0 S 0
For interchannel XTalk, , and the electrical receiver cuts-off the last term:
=
i
i i ip t p d t p d t
ν ν
ε
≠
+
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 16
Power Penalty due to Interchannel Crosstalk
Ideal situation (no interchannel XTalk)
,1 0 ,0
0
2
no XT
, 0
2
12
i i
i
i
p p p
Q k p k p
Qpk
= = ⇒
= = ⇒
=
Real situation (with interchannel XTalk):
( ) ( )
( )
,1 0 ,0 0
0 0
2
2w/ XT
,
2 1
1 12 1
i i i
i i i
i
i
p p p p
Q k p p k p
Qpk
ε
ε ε
ε
= = ⇒
= − = − ⇒
= −
Interchannel XTalk power penalty, in dB
( )w/ XTinterXT 10 10
no XT
10log 20log 1ii
i
pP
pε
∆ = = − −
Q factor dependence on power levels of bits 1 and 0 (for dominance of signal-ASE beat noise)
( ) ( ),1 ,0,1 ,0
,1 ,0
n i ii i
r i r i
k p pQ k p p
k p k p−
= = −+
Definition of power penalty, in dB
real w/ XTinterXT 10 10
ideal no XT
10log 10logi i
i i
p pP
p p
∆ = =
Exercise: derive expression for the
power penalty in case of an unamplified system
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 17
Example of Application
( ) ( )1 2
220 20interXT interXT10
When the two channels have the same suppression,
1 10 2 Channel suppression (dB)= 20log 1 10 3P P
ε ε ε
ε −∆ −∆
= =
= − ⇒ − − +
interXT 1 dB 0.0118 Channel suppression 19.27 dBiP ε⇓
∆ ≤ ⇒ ≤ ⇒ ≥
Calculation of the required adjacent channel suppression (dB) for an interchannel XTalk penalty not exceeding 1 dB.
1. Assuming just one adjacent channel
( ) ( )220 20interXT interXT101 10 Channel suppression (dB)= 20log 1 10P P
iε −∆ −∆= − ⇒ − −
( )1
1interXT 10
In case of interfering channels, should be replaced by :
20log 1
Nii i
Ni i
N
P
ε ε
ε=
=
∑
∆ = − − ∑
interXT 1 dB 0.0059 Channel suppression 22.27 dBP ε⇓
∆ ≤ ⇒ ≤ ⇒ ≥
2. Assuming the main XTalk comes from the two adjacent channels (neglecting XTalk coming from the other channels)
Required suppression
increases with the number of
interfering channels
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 18
Capacity Limitations in DWDM systems
Maximum count of channels:
Nch,max = pmax / pch
pmax = maximum power imposed by,e. g., EDFAs
pch = average power per channel(necessary to fulfil the required
system margin)
… imposed by the maximumpower of the WDM signal
… imposed by the maximumbandwidth of the WDM signal
BWDM,max = maximum bandwidth imposed
by, e. g., the EDFAs’ gain
∆νch = channel spacing
Maximum count of channels:
Nch,max = BWDM,max / ∆νch
Rb,ch = channel bit rateMaximum bit rate of the WDM signal
Rb,WDM,max = Nch,max · Rb,ch
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 19
Example of Application
Maximum count of channels:
Nch,max = pmax / pch = 100 / 10–6/10 = 398.1
Calculation of the maximum bit rate of a WDM signal in a 25 GHz channel spacing and 2.5 Gbit/s per channel link
... that uses EDFAs with uniform gain in the 1530-1560 nm band and maximum output power of 20 dBm
… assuming that the average signal power per channel required at the PIN input to guarantee the target system margin is -6 dBm
EDFAs’ bandwidth, in Hz:
BWDM,max = c / (λ0)2 (∆λ)max = 3768 GHz
(λ0 = 1545 nm ; (∆λ)max = 30 nm)
Maximum bit rate of the WDM signal
Rb,WDM,max = Nch,max · Rb,ch = 150 × 2.5 = 375 Gbit/s
Maximum count of channels:
Nch,max = BWDM,max / ∆νch = 3768 / 25 = 150.7
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 20
Gain Equalization in EDFA Chains
• The preferred solution today is to add an optical filter within the amplifier with a carefully designed passband to compensate for the gain spectrum of the amplifier so as to obtain a flat spectrum at its output.
• Both dielectric thin-film filters and long-period fiber gratings are good candidates for this purpose.
No equalisation
Pre-emphasis
Equalizing filter within each amplifier
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 21
Dependence of Q-factor on Launch Power Level
• Q-factor variations with launched power in long-haul systems.
• Q factor increases initially with launched power, reaches a peak value, and then decreases with a further increase in power because of the onset of the nonlinear effects.
• The reduction in multi-channel systems is more pronounced than in single channel systems due to the contribution of interchannel nonlinear fibre effects.
• XPM is usually the most important multi-channel nonlinear impairment at 10 Gbit/s per channel: just a few (~6) channels contribute to this limitation.
Q fa
ctor
Section input power
Single-channel
Multi-channel
Reduction of maximum input power �Reduction of maximum nonlinear phase shift
due to interchannel nonlinear fibre effects(XPM, FWM and SRS)
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 22
Design of WDM Links
The similar steps to single channel systems design are followed but with a reduction of the maximum input power per section
Remarks
1) Transmission penalties due to multi-channel fibre nonlinear effects (XPM, FWM and SRS) should be taken into account.
2) FEC has no impact on the penalty due to each multi-channel fibre nonlinear effect.
System margin of the link
, ,FEC ,SPM,max ,XPM,max ,FWM,maxdBs R R i i i iM OSNR OSNR P P P= − − ∆ − − ∆ − ∆ −… …
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 23
WDM Networks
• These networks provide circuit-switched end-to-end optical channels, or lightpaths, between network nodes to their users, or clients.
• A lightpath consists of an optical channel, or wavelength, between two network nodes that is routed through multiple intermediate nodes. Intermediate nodes may switch and convert wavelengths.
• These networks may thus be thought of as wavelength-routing networks.
• Lightpaths are set up and taken down as dictated by the users of the network.
• Noteworthy features of these networks:
– Wavelength reuse– Wavelength conversion – Transparency– Circuit switching – Survivability
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 24
Architecture of WDM NetworksWDM Network Elements
• Optical Line Terminals (OLTs)
• Optical Add-Drop Multiplexers (OADMs),
• Optical Crossconnects (OXCs)
• Optical line amplifiers -deployed along the fibre link at periodic locations to amplify the light signal
• OLTs, OADMs, and OXCs may themselves incorporate optical amplifiers
• OLTs are widely deployed, and OADMs are deployed to a lesser extent. OXCs are just beginning to be deployed
• The architecture supports a variety of topologies, including ring and mesh topologies.• The users (or clients) of this network are connected to the OLTs, OADMs, or OXCs.• The network supports a variety of client types, such as IP routers, ATM switches, and SONET terminals and ADMs.
Lightpaths
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 25
OLT
• Transponder adaptation functions:– Modification of the signal wavelength
(in order to get an ITU wavelength)– Addition of overhead for control and
management functions– Addition of FEC– BER monitoring
O/E/O
O/E/O
Non ITU λ
Non ITU λ
Router IP
ADM SDH
ADM SDH
Laser
λOSC
ITU λ1
MUX EDFAITU λ2
ITU λ3
Adaptation functions
Optical Line Terminal
Transponder
λ1, λ2, λ3, λOSC
Only the multiplexing function is illustrated. The demultiplexing is also performed in the OLT (for
the opposite direction of communication)
• OLTs multiplex multiple wavelengths into a single fiber and also demultiplex a composite WDM signal into individual wavelengths.
• OLTs are used at either end of a point-to-point link.
Addition of optical supervisory channel, λOSC
= 1510 or 1620 nm
• Transponder aspects:– Can be fixed-wavelength or tunable.– typically are the bulk of the cost, footprint,
and power consumption in an OLT. – reducing the number of transponders helps
minimize both the cost and the size of the equipment deployed.
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 26
OADMs• OADMs are used at locations where some fraction of the wavelengths need to
be terminated locally and others need to be routed to other destinations.
• They are typically deployed in linear or ring topologies.
• Several architectures (parallel, series) using different technologies (AWG, FBG) have been proposed
• Static and reconfigurable OADMs are available– In reconfigurable OADMs, the wavelengths that are dropped and added in the
OADM can be changed– Reconfigurable OADMs allow the lightpaths can be set and removed as needed– Reconfigurable OADMs can be implemented using optical switches or tuned FBGs.
λNch
MUXDMUXλ1, λ2, ..., λNch
Optical switch (electrically controled by the network management)
Transponders
λ2
λ1
Example of a reconfigurable OADM using a
parallel architecture
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 27
OXCs• OXCs features
– Functions are similar to OADMs but on a much larger scale in terms of number of ports and wavelengths involved,
– are deployed in mesh topologies or in order to interconnect multiple rings– allow a fast reconfiguration of the network lightpaths
• OXC architectures– Opaque (O/E and E/O conversions are used inside the OXC)
• Signal regeneration is possible • Wavelength conversion is possible• Limited capacity and just one bit rate is used
– Transparent (all-optical)• Several architectures have been proposed with and without wavelength conversion
Remark
• OADM and OXC performances are mainly degraded by intrachannel XTalk• Signals of same optical frequency are combined in OADMs and OXCs
due to the imperfect isolation between ports of the optical switches
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 28
All-optical OXC Architectures
Remark: OLTs of OXC do not use transponders
Optical Switch
λ1λ2λ3
λ1λ2λ3
λ1λ2λ3
λ1λ2λ3
λ1λ2λ3
λ1λ2λ3
OLT
OLT
OLTOLT
OLT
OLTλ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
X
Without wavelength conversion
Optical Switch
λ1λ2λ3
λ1λ2λ3
λ1λ2λ3
OLT
OLT
OLTOLT
OLT
OLTλ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
λ1, λ2, λ3
λ1λ2λ3
λ1λ2λ3
λ1λ2λ3
Wavelength Converters
With wavelength conversion
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 29
Intrachannel XTalk Modeling
( ) ( ) ( )[ ] ( ) ( )[ ]
( )( )
0 S S S 0
S
Electric field at the PIN input (just one channel interferer):
= 2 cos 2 2 cos 2
={0, 1}, depending on whether a 0 or 1 is being sent in the desired channel; = {0, 1},
i i i i
i
E t p d t t t p d t t t
d td t
πν φ ε πν φ⋅ + + ⋅ +
( ) ( )S
S
depending on whether a 0 or 1 is being sent in the XTalk channel; and is the optical frequency of the signal and XTalk carriers; and are the random phases of the signal and XTalk channels
(
i
it tν ν
φ φit is assumed that all channels have an infinite extinction ratio)
( ) ( ) [ ]( )
( ) ( )
S
,1 0
S ,0 0
Worst case XTalk (lower power level for 1 and higher power level for 0)
For bit 1, 1, and 1, cos =-1 (worst case), :
1 2
For bit 0, 0, and 1 (worst case):
i i i
i i
i i i
d t d t
p p
d t d t p p
ε ε
ε
ε
= = ⋅
= −
= = =
�
( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )0 S 0 0 S S S
Power incident on PIN within the receiver bandwidth (proportional to the squared electric field)
= 2 cos 2i i i i i i ip t p d t p d t p d t d t t t tε ε π ν ν φ φ+ + ⋅ − + −
( ) ( ) ( )( ) ( ) ( ) ( )[ ]
S
0 S 0
0 S S
For intrachannel XTalk, :
=
2 cos
i
i i i
i i i
p t p d t p d t
p d t d t t t
ν ν
ε
ε φ φ
=
+ ⋅ +
⋅ −
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 30
Power Penalty due to Intrachannel XTalk
Ideal situation (no intrachannel XTalk)
,1 0 ,0
0
2
no XT
, 0
2
12
i i
i
i
p p p
Q k p k p
Qpk
= = ⇒
= = ⇒
=
Real situation (with intrachannel XTalk): ( )
( )( ) ( )
( )
,1 0 ,0 0
0 0
2
2w/ intra
1 2 ,
1 2 2 1 2
1 12 1 2
i i i i
i i i i i
i
i i
p p p p
Q k p p k p
Qpk
ε ε
ε ε ε ε
ε ε
= − = ⇒
= − − = − −
⇒ =
− −
Intrachannel XTalk power penalty, in dB
( ) ( )1
w/ intraintra 10 10 10
no intra
10log 20log 1 2 20log 1 2ii
i i ii
pP
p
εε ε ε
∆ = = − − − ≈ − −
�
Q factor dependence on power levels of bits 1 and 0 (for dominance of signal-ASE beat noise)
( ) ( ),1 ,0,1 ,0
,1 ,0
n i ii i
r i r i
k p pQ k p p
k p k p−
= = −+
Definition of power penalty, in dB
real w/ intraintraXT 10 10
ideal no intra
10log 10logi i
i i
p pP
p p
∆ = =
Exercise: derive expression for the intrachannel Xtalk power penalty in case of an unamplified system
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 31
Example of Application
( ) ( )1 2
220 20intra intra10
When the two interferers result from the same isolation,
1 10 16 Switch isolation (dB)= 20log 1 10 12P P
ε ε ε
ε −∆ −∆
= =
= − ⇒ − − +
intra 1 dB 0.0118 4 Switch isolation 25.27 dBiP ε⇓
∆ ≤ ⇒ ≤ ⇒ ≥
Calculation of the switch required isolation (dB) for an intrachannel XTalk penalty not exceeding 1 dB.
1. Assuming just one interferer
( ) ( )220 20intra intra101 10 4 Switch isolation (dB)= 20log 1 10 6P P
iε −∆ −∆= − ⇒ − − +
( )1
1intra 10
In case of interferers, should be replaced by :
20log 1 2
Nii i
Ni i
N
P
ε ε
ε=
=
∑
∆ = − − ∑
intra 1 dB 0.0118 16 Switch isolation 31.27 dBiP ε⇓
∆ ≤ ⇒ ≤ ⇒ ≥
2. Assuming the intrachannel XTalk comes from two interferersresulting from the same switch isolation level
Required isolation
increases with the square of the
number of interferers
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 32
OTN – Optical Transport Network• OTN is a recent ITU-T standardisation (G.709)
– designed to transport data packet traffic such as IP and Ethernet over fibre optics, as well as legacy traffic (SDH).
– the target is long distance transmission with data bit rates from 2.5 Gbit/s up to 40 Gbit/s
– defines an Optical Transport Hierarchy (OTH) similar to SDH with two stages: first is electrical (mapping of tributary signals and overhead insertion) and second is optics (creation of optical channels and WDM structure)
– Capabilities: FEC - RS(255,239) -, management, protocol transparency and asynchronous timing
39.813 Gbit/sSTM-25643.018 Gbit/sOTU39.953 Gbit/sSTM-6410.709 Gbit/sOTU22.488 Gbit/sSTM-162.666 Gbit/sOTU1Line ratesSDHLine ratesOTN (G.709)
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 33
Forward Error Correction (FEC) for OFTS
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 34
Why Forward Error Correction?
• It is entirely possible that a specified BER cannot be achieved.
• Only viable alternative � use an error-correction scheme.
• In one approach, errors are detected but not corrected.
– Suitable when packet switching is used (Internet protocol).
• In FEC, errors are detected and corrected at the receiver without any retransmission of bits.
• This scheme is best suited for lightwave systems operating with SONET or SDH protocol (synchronous transmission).
• Historically, lightwave systems did not employ FEC until the useof in-line optical amplifiers became common.
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 35
Basics of Error-Correcting Codes
• Basic idea: add extra bits at transmitter using a suitable code.
• At the receiver end, a decoder uses these control bits to detect and correct errors.
• How many errors can be corrected depends on the coding scheme employed.
• In general, more errors can be corrected by adding more control bits to the signal.
• There is a limit to this process since line bit rate of the system increases after the FEC coder.
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 36
FEC Code Overhead and Redundancy
• Code redundancy: ρFEC= Rb,l / Rb,i −1
• Code ratio: rFEC = Rb,i / Rb,l
FEC coder FEC decoderTransmissionpathRb,lRb,i Rb,l Rb,i
Tx Rx
Effective (line) bit rateInformation
(data) bit rateOptical
transmitterOpticalreceiver
Optical channel
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 37
Types of Error-Correcting Codes
• Classified under names such as:
– linear,
– cyclic,
– Hamming,
– Reed–Solomon,
– convolutional,
– product, and
– turbo codes.
• Among these, Reed–Solomon (RS) codes have attracted most attention for lightwave systems.
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 38
Characterisation of RS Codes
• Reed-Solomon (RS) codes do not operate on bits but on groups of bits �symbols
• RS code considers blocks of k data symbols and calculates r additional symbols with redundant information (FEC overhead), based on the code.
• The transmitter sends the blocks of n = k + r symbols to the receiver � RS(n,k).
• k + r coded symbols have to be transmitted in the same duration as k information symbols, each coded symbol has k/(k+r) the duration of uncoded symbol � line bit rate increases by n/k � rFEC = k / n
• RS(n,k) codes have the restriction that if a symbol consists of b bits � length of the code: n = 2b-1.– code length of n = 255 if (8-bit) bytes are used as symbols.
• The number of redundant bits r can take any even value.
RS(n,k)coder
RS(n,k)decoder
Transmissionpath
Tx RxBlocks ofk symbols
Blocks of n=k+r
symbols
Blocks of n=k+r
symbolsBlocks ofk symbols
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 39
Performance of RS Codes
• The receiver considers a block of n symbols, and knowing the code used by the transmitter, it can correctly decode the k data symbols even if up to r/2 of the n symbols are in error.
( ), , , ,
, ,
Coded (line) symbol error probability
1 1
: line bit error probability
be s l e b l
e b l
P P
P
= − −
( ), , , , , ,2 1
, ,
Information symbol error probability
1
: line symbol error probability
: number of -combinations from elements: greater integer contained in
n n in ie s i i e s l e s l
i r
e s lni
iP C P Pn
P
C i nx x
−
= +
= ⋅ ⋅ −∑
( ) ( )1, , , , , , , ,
, ,
Information bit error probability
1 1 1 1
: information symbol error probability
b be b i e s i e s i e b i
e s i
P P P P
P
= − − ⇔ = − −
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 40
Useful Approximation to Calculate the Data Bit Error Probability
( ) 2 12 1, , , , , ,2 1 , ,
Approximation for information symbol error probability2 1
1 for <<1n rrn
e s i e s l e s lr e s lr
P C P P nPn
− − + +
+ = ⋅ ⋅ −
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 41
RS Codes for OFTS
G.709 ITU (OTN) recommendation
• RS(255, 239) with b = 8: FEC overhead of 6.7%.
• RS(255, 223) with b = 8: FEC overhead of 14.4%
Improvement in BER
due to FEC is quantified
through the coding gain
Single RS
codes
RSproduct
codes
Conca-tenated
RScodes
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 42
Coding Gain (1)
• Coding gain: a measure of improvement in bit error probability through FEC.
– It is expressed in terms of the equivalent value of Q (for a given bit error probability) as
• Factor of 20 is used in place of 10 because performance is often quantified through Q2.
• If FEC decoder improves BER from 10−3 to 10−9, Q increases from 3 to 6, resulting in a coding gain of 6 dB.
• Magnitude of coding gain increases with the FEC overhead.
min10 min 10 FEC,min 10
FEC,min
FEC,min
min
Coding gain
20 log 20 log 20 log
: minimum factor required at receiver input
with FEC to achieve the required data bit error probability: minimum fa
cQG Q Q
Q
Q Q
Q Q
= − =
ctor required at receiver input without FEC to achieve the required data bit error probability
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 43
Coding Gain (2)
10-9
Coding gain of RS(255,239)
Coding gain of RS(255,223)
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 44
Coding Gain of RS codes
• For single RS codes, coding gain is 5.5 dB for 10% overhead and increases sublinearly, reaching 8 dB for 50% overhead.
• It can be improved by concatenating two or more RS codes or by employing the RS product codes.
@ bit error probability
of 10-9
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 45
Compromise on FEC Implementation
• While implementing FEC, one faces a dilemma.
– As the overhead is increased to realize more coding gain, line bit rate increases.
– Since Q factor realized at the receiver depends on the bit rate, its value is reduced, and BER actually worsens.
– Decoder improves it but it first has to overcome the degradation caused by the increased bit rate.
• If an aggressive FEC scheme is employed, BER may degrade so muchthat the system is not operable even with the FEC coder.
• An optimum range of coding overhead exists for every system designed to operate at a specific bit rate over a certain distance.
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 46
Gross vs. Net Coding Gain
( )2
,2
, 10 min 10 FEC,min
Gross coding
Using the approximation for the required OSNR
1
1
and, for the same extinction ratio and optical filter with and without FEC, we can write
20log 20log
e nR
o
c N
Q B rosnrB r
G Q Q
+= ⋅−
= −
,
, ,FEC10
, gain,
10log
c G
e n
e nG
BB
−
���������������
, min FEC,min
FEC,min
min
Net coding gain
: minimum OSNR required at receiver input
with FEC to achieve the required data bit error probability: minimum OSNR required at receiver input
c NG OSNR OSNROSNR
OSNR
= −
without FEC to achieve the required data bit error probability
What really matters from the viewpoint of system design
(achievable distance)!
, ,FEC, , 10
,10log e n
c N c Ge n
BG G
B
= −
, , , ,FEC ,
,, , 10
,
For and
(in the same proportionality)
10 log
e n b i e n b l
b lc N c G
b i
B R B R
RG G
R
∝ ∝
= −
© Adolfo Cartaxo Chapter 4, Optical Fibre Telecommunication Systems, 2012 47
Symbol Interleaving
• Without interleaving, the symbols would be transmitted in row order � symbols in row 1 are transmitted, followed by the symbols in row 2, and so on.
• The idea of interleaving is to transmit the first d symbols in column 1, followed by the first d symbols in column 2, and so on. Thus, symbol 1 would be followed by symbol k + 1.
• When d symbols have been transmitted from all n columns, we transmit the next d symbols in column 1 - from rows (d + 1) to 2d -, followed by the next d symbols in column 2, and so on. The parameter d is called the interleaving depth.
• Suppose there is a burst of b symbol errors. Only ceil(b/d) of these symbols will occur in the same row due to interleaving � a (255,223) Reed-Solomon code will be able to correct any burst of b errors when interleaving to depth d is used, provided ceil(b/d) < 16.
n-k redundant symbolsdk...(d-1)k+3(d-1)k+2(d-1)k+1d
n-k redundant symbols...............
n-k redundant symbols3k...2k+32k+22k+13
n-k redundant symbols2k...k+2k+2k+12
n-k redundant symbolsk...3211
n...k...321Indexes
Information symbols Redundancy symbols
RS(n,k)