METHOD FOR COMPENSATING CHROMATICDISPERSION AND
ASSOCIATED EQUIPMENT
BACKGROUND OF THE INVENTION
The present invention relates to the field of wavelength division multiplexing optical
networks with packet granularity capability and coherent detection and more particularly to the
chromatic dispersion compensation in such networks.
A wavelength division multiplexing optical network refers to a network comprising a
plurality of nodes connected by optical links wherein data signals are transmitted through a
plurality of channels having different wavelengths and which are multiplexed to be transmitted
through the optical links of the network.
Packet granularity or packet switching granularity refers to the possibility to add or drop
one or several optical packets of a signal in an intermediary node while transmitting the other
packets transparently. In such networks, the packets are usually transmitted within time slots so
that the packets of the different channels are synchronized.
Equipments with packet granularity capability are more and more implemented in the
optical communication network in order to enhance the flexibility of communication networks.
Such equipments are referred to as packet optical add-drop multiplexers (POADMs). However, a
POADM requires compensating for the chromatic dispersion induced by the transmission of
optical packets through links of the network.
Indeed, one aspect of the chromatic dispersion called inter-channel chromatic dispersion
refers to the fact that packets transmitted in channels of different wavelengths experience
different travelling speeds so that time shifts or time offsets are introduced between packets
emitted simultaneously. Need is then to resynchronize the packets to enable their processing at
the receiver.
Besides, another aspect of the chromatic dispersion called intra-channel chromatic
dispersion refers to the distortion undergone by the signal representing the bit coding of a packet
during its transmission through the links of the network, rendering the bit decoding difficult and
possibly erroneous.
One way to compensate for both aspects of the chromatic dispersion is to use in-line
compensators located along the links of the network. However, in-line components introduce
additional losses that need to be compensated by additional amplifiers. Moreover, such amplifiers
introduce additional costs and generate additional noise so that the distance that can be reached
transparently with a given quality of signal may be reduced.
SUMMARY OF THE INVENTION
It is therefore an object of the present invention to overcome the above mentioned
drawbacks of the state of the art and to provide a solution to compensate for the chromatic
dispersion in wavelength division multiplexing without using in-line components.
Thus, the present invention refers to a method for compensating, within a node of an
optical network, chromatic dispersion undergone by optical packets transmitted within time
slots of wavelength division multiplexed channels along at least one link of the optical
network, a time slot duration corresponding to the sum of a packet duration and an interpacket
gap duration, the said method comprising the foliowings steps:
- demultiplexing the wavelength division multiplexed channels into a plurality of
bands, a band comprising a predetermined number of adjacent wavelength channels,
- transmitting the said plurality of bands, via a respective plurality of delay lines
having predetermined delays, toward a respective plurality of packet add/drop
structures comprising a coherent receiver,
wherein the said predetermined number of channels of one band is determined so that
a first time shift, due to the effect of the chromatic dispersion along transmission
through the network, between two optical packets of the same time slot sent
respectively in different channels of the same band, remains shorter than an interpacket
gap duration and so that the coherent receiver is capable of processing the said
predetermined number of channels of one band,
wherein the predetermined delay of a delay line associated with a band of channels
corresponds to a second time shift between a channel of the said associated band and a
reference channel, the said second time shift being due to the effects of chromatic dispersion
along the last crossed link.
The embodiments of the present invention also refer to a packet optical add/drop
multiplexer located in a node of a wavelength division multiplexing optical network and
configured to process optical packets transmitted within time slots having a duration
corresponding to a packet duration and an inter-packet gap duration along links of the optical
network and received from remote nodes of the optical network, the said packet optical
add/drop multiplexer comprising:
- a plurality of packet add/drop structures comprising a coherent receiver,
- a band demultiplexer configured for demultiplexing the received multiplexed
channels into a plurality of bands, a band comprising a predetermined number of
adjacent channels, the said predetermined number of channels being determined so
that a first time shift, due to the effect of the chromatic dispersion along transmission
through the network, between two packets sent respectively in a first and a second
channel of the band, remains shorter than an inter-packet gap and so that the coherent
receiver is capable of processing the said predetermined number of channels,
- a plurality of delay lines having predetermined delays, the plurality of bands being
transmitted respectively to the plurality of packet add/drop structures via the said
plurality of delay lines, the predetermined delay of a delay line associated with a band
being determined according to a second time shift between a channel of the associated
band and a reference channel, the said second time shift being due to the effects of
chromatic dispersion along the last crossed link.
The embodiments of the present invention also refer to an optical node of a
wavelength division multiplexing optical network comprising a plurality of nodes
linked by optical links comprising:
- a data repository configured for storing information about the topography of the
links adjacent to the node,
- a packet optical add/drop multiplexer wherein a dedicated receiver is configured for
updating information about the chromatic dispersion undergone by the optical packets
transmitted on other channels than the control channel based on the information about
the topography of the links adjacent to the node stored in the data repository.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG.l is a diagram of the arrangement of the packets in their time slots in the different channels
of the different bands of a WDM signal;
FIG.2 is a diagram of a packet optical add-drop multiplexer according to an embodiment of the
present invention;
FIG.3 is a diagram of a packet optical add-drop multiplexer according to another embodiment of
the present invention;
FIG.4 is a diagram of a packet optical add-drop structure according to an embodiment of the
present invention;
FIG.5 is a diagram of a coherent receiver according to an embodiment of the present invention;
FIG.6 is a diagram of multiple input multiple output adaptive equalizer with four finite impulse
response (FIR) filters;
FIG.7 is a diagram of a spectrum grid comprising a plurality of wavelength channels;
FIG.8 is a diagram of the arrangements of the packets in the time slots of a wavelength
channels;
FIG.9 is a diagram of a network portion;
FIG.IO is a diagram of the packets in their time slots at the ingress node;
FIG.ll is a diagram of the packets in their time slots at the input of an intermediary node;
FIG.12 is a diagram of the packets in their time slots at the output of an intermediary node;
FIG.13 is a diagram of the packets in their time slots at the input of an egress node;
FIG.14 is a diagram of a node comprising a combination of ROADM and POADM according to
a first embodiment;
FIG.15 is a diagram of a node comprising a combination of ROADM and POADM according to
a second embodiment;
In these drawings, the elements having the same reference correspond to elements having
a similar function. When a reference is composed of a reference number and an index, the
reference number represent a class of elements having a similar function while the index
designate a particular element of the class. For example, the elements 13 and 132 refer both to
delay lines but the element 13 may have a delay that is different than the element 32.
DETAILED DESCRIPTION OF THE INVENTION
As used herein, the term "WSS" refers to the acronym Wavelength Selective Switch;
As used herein, the term "ROADM" refers to the acronym Reconfigurable Optical Add-
Drop Multiplexer;
As used herein, the term "POADM" refers to the acronym Packet Optical Add-Drop
Multiplexer;
As used herein, the term "SOA" refers to the acronym Semiconductor Optical Amplifier;
As used herein the term "inter-packet gap" in a packet stream refers to the guard interval
which separates two consecutive packets transmitted on a common wavelength channel;
As used herein the term "band" refers to a wavelength interval, usually gathering a
plurality of wavelength channels;
As used herein, the term "time slot" in a packet stream transmitted in a channel refers to
a time interval wherein a packet is inserted. The time slot duration corresponds to the sum of a
packet duration and an inter-packet gap duration.
As used herein the term "FIR filter" refers to the acronym Finite Impulse Response filter
which is a filter whose impulse response (or response to any finite length input) is of finite
duration.
As used herein, the terms "ingress node and egress node" of a signal refer respectively to
the source node from which the signal is emitted (after being converted from the electrical to the
optical domain) and the destination node in which the signal is received (and is converted from
the optical to the electrical domain).
As used herein, the term "transparent" to qualify a path or a transmission refers to the
transmission (or path) of an optical signal without performing any optical-electrical-optical
(OEO) conversion.
As used herein, the term "optical packet" refers to an optical signal corresponding to a
block of a predetermined amount of data (or predetermined number of bits) encoded according
to a coding scheme and modulated according to a predetermined modulation format.
As used herein, the term "dropped packet" refers to a packet for which the current node
is the egress node so that the said packet is detected by a receiver and the data of the packet are
decoded by the receiver. Inversely, the non-dropped packets are transmitted transparently toward
another node.
As used herein the expression "Gb/ s" refers to the unit giga-bit per second.
The embodiments of the present invention refer to a method for compensating
chromatic dispersion undergone by packets transmitted in time slots of wavelength division
multiplexed (WDM) channels through the links of a network wherein inter-channel and intrachannel
chromatic dispersion are compensated in the nodes of the network and separately, the
inter-channel chromatic dispersion being compensated thanks to delay lines while the intrachannel
chromatic dispersion being compensated by digital signal processing techniques.
Furthermore, an aspect of the present invention is to process the received WDM channels per
band in order to reduce the capital expenditure required to implement the inter-channel
chromatic dispersion compensation.
The method is based on coherent detection wherein the packets transmitted at different
times by several channels can be detected by a single coherent receiver without requiring
additional filtering elements.
With time slotted transmissions, optical packets belonging to a common time slot (and
therefore to different channels) are emitted simultaneously (with different transmitters) whereas
optical packets belonging to different time slots (on the same or on different channels) are
emitted at different times as described in fig.l where a set of j wavelength channels l distributed
in k bands B of i channels is represented. Each channel is timely divided into successive time
slots TS of duration DT wherein the packets P are introduced. An additional channel t refers to
the control channel that transmits the header of the packets P transmitted in the other channels
li... than the control channel Xct. The control channel Xct is a circuit switching channel and
therefore does not comprise packets but a continuous flow of data organized in frames T of
length TS.
Nevertheless, the time shifts or time offsets introduced between the channels by the
effects of the inter-channel chromatic dispersion during transmission along the links of the
network may lead to a temporal overlapping of the packets transmitted in different time slots
through different channels.
However, as the introduced time shift is proportional to the wavelength difference of the
channels, for adjacent channels having close wavelengths, herein the channels of a common band
B, the introduced time offset remains very small.
Furthermore, a packet stream transmitted within a channel comprises a succession of
time slots, each time slot comprising a data packet and an inter-packet gap or guard interval
which is used to separate two successive packets. Thus, if a packet is time shifted of a duration
that is shorter than the inter-packet gap, this time shifted packet remains in the time slot and can
therefore be processed normally by the receiver. The inter-packet gap can be seen as a tolerance
in the synchronization of the packets. As a consequence, if the time shift introduced by interchannel
chromatic dispersion between channels having close wavelengths remains smaller than
the inter-packet gap, the receiver is still capable of processing correctly the time shifted packets.
Besides, the number of channels that a coherent receiver is able to process without
introducing too much loss is also limited, for example ten channels with coherent receivers of the
state of the art.
As a consequence, if the time shift introduced within ten adjacent channels along the
transmission is less than an inter-packet gap, the packets transmitted in these channels can be
detected by a receiver without requiring compensating individually for the inter-channel
chromatic dispersion of each channel.
Fig.2 represent a packet add/drop multiplexer (POADM) 1 according to an embodiment
of the present invention. The POADM 1 is aimed at being implemented in the nodes of the
network and comprises an input 3 which is linked to an optical link 4, generally implemented as
an optical fibre, along which the signals comprising the Wavelength Division Multiplexed (WDM)
channels are transmitted. The input 3 of the POADM 1 is linked to an optical coupler 5 that
transmits the WDM channels on one side toward a dedicated receiver 7 and on the other side to
an input of a band demultiplexer 9. This coupler may be replaced by a channel demultiplexer
with the advantage of presenting a lower loss.
The dedicated receiver 7 is aimed at detecting the control channel, which is preferably a
channel located at a border of the spectrum comprising all the transmitted WDM channels. The
control channel transmits information concerning the packets transmitted on the other
wavelengths than the control channel, i.e. their header, and notably the value of the chromatic
dispersion undergone by the packets of the other channels than the control channel. The control
channel may be a channel with a reduced throughput with respect to the other channels, for
example 2.5 Gb/s for the control channel while the other channels may have a 10 Gb/s
throughput.
The dedicated receiver 7 is preferably a non-coherent receiver in order to reduce the cost
of this dedicated receiver 7. In such case, a filter is implemented at the input of the dedicated
receiver 7 in order to filter out the channels that do not correspond to the control channel. If the
coupler 5 is replaced by a channel demultiplexer, no filter is necessary. Furthermore, the control
channel is demodulated and decoded at each node. As a consequence, the cumulated chromatic
dispersion of the control channel is limited and corresponds to the chromatic dispersion
undergone across the last link. Thus, if the throughput is reduced, a modulation format which is
robust to impairments such as an on-off keying (OOK) format can be used. In such case,
detection can be achieved without chromatic dispersion compensation. However, if a higher
throughput is requested, or if the length of the links is longer than what is typically used in core
networks, compensation techniques of the state of the art can be implemented such as a fibre
Bragg grating or a maximum-likelihood sequence estimation.
Besides, as the time for the dedicated receiver to detect and process the data transmitted
in the control channel c t may not be negligible with respect to a packet duration, these detection
and processing need to be performed in advance with respect to the processing of the packets
transmitted in the other channels for which information about the chromatic dispersion is
transmitted in the control channel. Two solutions may be applied to solve this issue, either
information about the chromatic dispersion undergone by dropped packets transmitted in a given
time slot is transmitted in a previous time slot of the control channel or a delay line is added, for
example between the optical coupler 5 and the band demultiplexer 9 in the POADM 1 presented
in Fig.2, the delay of this delay line corresponding to the time necessary for the dedicated receiver
to process a packet of the control channel.
The band demultiplexer 9 comprises a plurality of outputs linked respectively to a
plurality of packet add/drop structures 11 via a respective plurality of delay lines 13 (x= l,..,N)
having predetermined delays. The fundamental idea is to gather in one band the channels having
undergone a relative time offset between each other due to the inter-channel chromatic
dispersion that is smaller than the inter-packet gap duration D ΐ . The WDM channels are
demultiplexed by bands, each band comprising a predetermined number of channels, and one
band is transmitted toward one packet add/drop structure 11 via one delay line 13 (x= l,..,N)
implemented for instance as a piece of optical fibre of a predetermined length. The
predetermined delay, i.e. the length, of a delay line 13 (x= l,..,N) associated with one band is
determined based on the inter-channel chromatic dispersion undergone by the channels of that
band on the last crossed link, i.e. from the previous node. The delay of a delay line 13 (x= l,..,N)
corresponds to the time offset introduced by the inter-channel chromatic dispersion between one
channel of the band and a reference channel, for instance the control channel.
The output of the packet add/ drop structures are linked respectively to the inputs of a
band multiplexer 5 that re-multiplex the plurality of bands in a single WDM signal. The output
of the band multiplexer 5 is linked to a first input of an optical coupler 17. A second input of
the optical coupler 17 is linked to a dedicated transmitter 19 configured for encoding and
modulating the packets of the control channel. The output of the optical coupler 7 is linked to
the output 2 1 of the POADM 1 towards an optical link 22.
According to another embodiment represented in Fig.3, the input 3 of the POADM 1 is
linked directly to a band demultiplexer 9 and the dedicated receiver 7 is located behind the band
demultiplexer 9 so that one output of the band demultiplexer 9 is linked to the dedicated receiver
7. The band demultiplexer 9 is then configured to send the control channel to the output linked
to the dedicated receiver 7 and the plurality of bands towards the respective plurality of packet
add/ drop structures via the plurality of delay line. In such configuration, no optical coupler 5
and no filter is required at the input of the dedicated receiver 7 as only the control channel is
transmitted by the band demultiplexer 9. In the same way, the dedicated transmitter 9 is linked
to an input of the band multiplexer 5 to be re-multiplexed with the plurality of bands and the
output of the band multiplexer 5 is linked directly to the output 2 1 of the POADM 1 so that
the optical coupler 17 is not necessary anymore.
4 represents a diagram of a packet add/drop structure 11 according to an
embodiment of the present invention. It comprises an optical coupler 23 having an input that
corresponds to the input of the packet add/drop structure 11 and is therefore linked to a delay
line 13 (x=l,..,N). The optical coupler 23 has two outputs, one linked to a coherent receiver 25
and the other linked to a demultiplexer 27. The use of a coherent receiver is important in order
to compensate for the intra-channel chromatic dispersion which will be described in details in the
following of the description. Thus, the band received from the delay line 13 (x=l,..,N) is sent
toward both the coherent receiver 25 and the demultiplexer 27. The demultiplexer 27 is
configured for demultiplexing the band received at its input into a plurality of individual
channels. The outputs of the demultiplexer 27 are linked respectively to the inputs of a
multiplexer 3 1 via a plurality of optical gates 29. The optical gates 29 could be configured to
block the data of the time slots corresponding to the packets being dropped in the node in order
to "free" these time slots and to enable adding new packets in this time slots and to let the nondropped
packet through. The dropped packets are detected by the coherent receiver 25. The
optical gates 29 are implemented preferably as semiconductor optical amplifier (SOA) gates.
Indeed, other technologies such as Mach-Zehnder modulators (MZM), ring resonators, acoustooptic
switches, liquid crystal on silicon (LCoS) or micro-electromechanical systems (MEMS)
could also be used but in the state of the art, these components have drawbacks for the present
application such as a slow functioning with respect to a packet duration or a low blocking
efficiency that currently prevent their utilization.
The multiplexer 3 1 is configured to re-multiplex the individual channels in a band. The
output of the multiplexer 3 1 is linked to an input of an optical coupler 33 which has a second
input linked to a transmitter 35. The transmitter 35 is configured to transmit packets aimed at
being introduced in the free time slots of the band. The continuous wave (CW) laser used in the
transmitter 35 may be implemented as a fast tunable CW laser. Alternatively, an array of lasers
emitting at wavelength corresponding to the channels of the band and coupled to a fast selector
that selects, for each time slot, the laser corresponding to the wavelength that needs to be
transmitted can be implemented. The optical coupler 33 mixes the optical signals received from
the multiplexer 3 1 and from the transmitter 35 so that the packets coming from the transmitter
35 are introduced within the free time slots of the band received from the multiplexer 31. The
output of the optical coupler 33 corresponds to the output of the packet add/ drop structure
and is linked to the band multiplexer 15.
The band demultiplexer 9 and the band multiplexer 5, will be preferably implemented as
low cost fixed band demultiplexers based on thin film filter or silica technology. The
demultiplexer 27 and the multiplexer 3 1 will be preferably implemented as array waveguide
gratings (AWG). This AWG could be realized with different technology such as III-V
semiconductor or silicon photonics. These two technologies could enable the complete
integration of the multiplexer, the demultiplexer and the optical gates. Alternatively, these
equipments may also be implemented as Wavelength Selective Switches (WSS) based on
electromechanical systems (MEMS) or liquid crystals on silicon (LcoS).
Thus, a band transmitted along a delay line 3 is received by the coherent receiver 25 and
the packets aimed at being dropped are detected by this coherent receiver 25. It has to be noted
that, if a packet add/drop structure comprises only one coherent receiver 25, within one time
slot, only one packet of one channel can be detected so that if two packets have a common
egress node, these two packets have to be transmitted either within two different bands or in two
different time slots. Such issue may obviously be overcome by implementing a plurality of
receivers per packet add/drop structure 11.
Fig.5 represents the functional elements of an embodiment of a coherent receiver 25
located in a packet add/ drop structure 11 described in Fig.4.
The coherent receiver 25 comprises an input 37 which is connected to the optical coupler
23 and that receives a band comprising a predetermined number of multiplexed channels. The
input 37 is linked to a first input of a coherent mixer 39. The second input of the coherent mixer
39 is linked to a local oscillator 4 1 implemented as a fast tunable laser which is tuned, for each
time slot, to the wavelength corresponding to the channel of the band that comprises a packet
that needs to be dropped. Similarly to a transmitter 35, the local oscillator 4 1 may be
implemented by an array of lasers emitting a set of wavelengths corresponding to the channels
of the band and coupled to a fast selector (the number of lasers in the array being equal to the
number of channels in the band).
The coherent mixer 39 comprises for instance a polarization beam splitter (PBS), a 50/50
optical splitter, and two 90° optical hybrids. The polarization beam splitter is configured for
splitting the signal received at the input into two signals having orthogonal polarizations. The
50/50 optical splitter is configured to split the signal received from the local oscillator 4 1 in two
signals having half power each. One output of the PBS and one output of the 50/ 50 splitter are
sent to a 90° hybrid coupler. The other PBS output and the other output of the 50/ 50 splitter are
sent to the second 90° optical hybrid. Therefore, the inphase and quadrature components of
both polarizations are retrieved at the outputs of the coherent mixer 39. These four components
are then detected by four photo-detectors 43, generally implemented as balanced photodiodes,
which are linked respectively to four analogical to digital (A/D) converters 45. The obtained four
digital signals are then used to feed digital signal processing means 47.
The digital processing means 47 comprise an electronic dispersion compensation module
and an adaptive equalizer.
The electronic dispersion compensation module comprises a digital filter which is configured to
compensate for the degradations (i.e. distortions) of the received signal due to the intra-channel
chromatic dispersion. These distortions depend on the total intra-channel chromatic dispersion
accumulated by a packet during its transparent propagation along the links of the network. These
distortions can therefore be different for each packet depending on the path that has been
followed.
The intra-channel chromatic dispersion can be described in the frequency domain as an all-pass
transfer function herein noted HDISP and defined by:
= e 4
with c the speed of light in vacuum, l the wavelength of the signal, w the angular frequency
and D the chromatic dispersion value defined by D=L with L the length of the optical fibre,
b a constant that depends on the type of the optical fibre and j the complex number with unit
modulus and angle of p/2.
Thus, in order to compensate for the effects of the intra-channel chromatic dispersion, the
digital filter of the electronic dispersion compensation module is configured to have a transfer
function that is the inverse of sp(i.e. HDISP ) - Such filter may be implemented in the time
or the frequency domain, using recursive or non-recursive filters. Furthermore, to configure
the digital filter, the chromatic dispersion value D needs to be known. However, such value
cannot be measured in a packet granularity application due to the too long duration of the
measurement with respect to a packet duration. In order to overcome this problem, the value
of the chromatic dispersion is transmitted within the control channel. Indeed, the dedicated
receiver 7 is configured to retrieve the information transmitted in the control channel and in
particular the value of the chromatic dispersion undergone by the dropped packets and also to
transmit this retrieved value to the coherent receiver 25 which detects these dropped packets.
As a consequence, the chromatic dispersion value provided by the dedicated receiver is used
by the electronic dispersion compensation module of the coherent receiver 25 to configure its
digital filter and to compute the value of D in the transfer function.
In order to obtain, within the control channel, an estimation of the chromatic dispersion
undergone by a packet at its egress node, the information concerning this chromatic dispersion is
initially set to 0 and is updated in each node along the path of the signal.
Indeed, it is assumed that the topography of the network (length and type of the optical
fibres along the links) is determined and stored in a data repository at network building time.
This data repository may be part of a centralized entity of the network such as a network
management system that distributes the local topologies to the nodes of the network via control
plane mechanisms. Such organization enables the storage, within each node, of the topography
of the adjacent links. As a consequence, the estimation of the chromatic dispersion undergone by
the signals along the last (or the next) crossed link can be determined within each node based on
the topography information stored in a data repository of the node.
Thus, after (or before) each link of the path, the values of the chromatic dispersion
undergone by the packets of the other channels than the control channel which are encoded in
the control channel are updated by adding the value associated with the last (or next) crossed link.
The cumulated value of the chromatic dispersion along the path is therefore obtained at the
egress node. Indeed, as the control channel is detected in each node, the values transmitted in the
control channel can be updated (by adding the value corresponding to the last link) and such
updates do not introduce any additional conversion or loss for the data packets transmitted on
the other channels (which can still be transmitted transparently across the network).
In order for the digital filter to produce a transfer function that is the inverse of HDISP , its
taps weights have to be determined.
In the case of a non-recursive filter implemented in the time domain with an odd
number N of taps, the tap weights are given by:
k — —
For k=l . ..N, where T is a symbol duration, and 2 2 where 2 the integer
part of N/2 rounded towards minus infinity. Thus, the tap weights can be computed based on
the chromatic dispersion value transmitted provided in the control channel. If this
computation is too long, a set of possible chromatic dispersion values and the associated tap
weights may be stored in a data repository such as a look-up table of the node. As a
consequence, as no measurement of the chromatic dispersion is needed and as only limited or
no computation is required to determine the tap weights of the FIR filter, the electronic
dispersion compensation module described herein enables a fast compensation of the intrachannel
chromatic dispersion.
Besides, other physical impairments such as polarization mode dispersion or
transceiver induced inter- symbol interference introduce signal degradation and need to be
compensated for.
This is achieved by an adaptive equalizer which is located at the output of the electronic
compensation module. The adaptive equalizer can be implemented with a multiple-input multiple
output (MIMO) time domain array of complex adaptive finite impulse response (FIR) digital
filters arranged in a butterfly structure such as described in Fig.6 with an array of four FIR filters
noted Hxx, Hxy, Hyy and Hyy. In the present example, the adaptive equalizer has two inputs
noted Xin and Yin that corresponds to the two polarizations and that contains the two
quadrature components (real part and imaginary part). The outputs Xout and Yout of the
adaptive equalizer are given by:
N (H [l]Xin[k+ l]+ H y [l]Yin [k+ l ] \Hyx [l]Xin[k+ l]+ l ]Yin [k+l]J
where N is the number of taps in the FIR filters, Hxx, Hxy, Hyx and Hyy are vectors of length N
comprising the tap weights, Xin and Yin are sliding blocks of N samples to which the filter is
applied, k is the sampling time index and 1the filter tap index.
The taps of the FIR filters of the adaptive equalizer are updated by an equalization algorithm
such as a constant modulus algorithm (CMA). CMA is a blind adaptation algorithm (the bits to
decode are not known) that adjusts the filter coefficients of the equalizer to reduce the intersymbol
interference of the received signal. The algorithm assumes that the transmitted signal is a
constant modulus signal, i.e. its amplitude is constant (this is the case for instance with quadrature
phase shift keying (QPSK) modulation format). The tap weights are then updated by:
k+l,l]= Hxx [k+ l,l]+ m 5 i Xout [k]Xm[k+ I ]
xy [k+ l,l]= Hxy [k+ 1, ί ]+ m d e Co ί [ +I ]
Hy [k+ l,l]= Hyx [k+ 1,1]+ m I ]
H yy + l,l]= Hyy [k+ 1,1]+ m 5 Yout [k]7m[k+ ]
with n the complex conjugate of Xin, m the convergence parameter, de1 and de2 are estimates
of the derivative of the modulus errors in the produced complex digital signal values and are
given by
d e 2- Xout 2, d e 2= a 2- Yout 2
where a is the targeted signal amplitude.
Alternate equalization algorithms may also be applied instead of the CMA such as a least-mean
square (LMS) algorithm, a decision directed (DD) algorithm or a zero-forcing (ZF) algorithm.
Besides, it has to be noted that the adaptive equalizer is also capable of compensating for
potential residual degradations due to chromatic dispersion, the amount of degradations due to
chromatic dispersion the adaptive equalizer is capable of processing depending on the number of
taps (the higher the number of taps and the higher the amount of degradations due chromatic
dispersion that can be compensated for). Indeed, as the amount of chromatic dispersion
transmitted in the control channel is only an estimation of the real amount of chromatic
dispersion undergone by a packet, a small amount of degradations due to chromatic dispersion
may still remain at the output of the electronic dispersion compensation module and the
adaptive equalizer may be configured to compensate for these remaining degradations due to
chromatic dispersion. Besides, as these remaining degradations due to intra-channel chromatic
dispersion are low, the convergence time of the adaptive equalizer is greatly reduced with respect
to the convergence time in the case of large degradations due to a high amount of intra-channel
chromatic dispersion (as it is the case at the input of the electronic dispersion compensation
module) so that the adaptive equalizer applies a fine compensation of the remaining degradations
due to intra-channel chromatic dispersion in a small amount of time.
The digital processing means 47 described in fig.5 previously are provided through the
use of a dedicated hardware as well as hardware capable of executing software in association
with appropriate software dedicated to the signal processing. When provided by a processor,
the digital processing means 47 may be provided by a single dedicated processor, by a single
shared processor, or by a plurality of individual processors, some of which may be shared.
Moreover, explicit use of the term "processor" should not be construed to refer exclusively to
hardware capable of executing software, and may implicitly include, without limitation,
digital signal processor (DSP) hardware, network processor, application specific integrated
circuit (ASIC), field programmable gate array (FPGA), read-only memory (ROM) for storing
software, random access memory (RAM), and non-volatile storage. Other hardwares,
conventional and/or custom, may also be included.
In order to better understand the functioning of the POADM 1 described previously, an
example will now be described based on a WDM signal with sixty-five channels noted from
lΐ to l65, and distributed in a spectral grid with a channel spacing Dl along a wavelength axis
l as represented schematically in fig. . The channel l65 is the control channel l ί . Fig.8
represents the arrangement of packets, noted P1...P6 within the corresponding time slots,
noted TS1..TS6, of a channel, lΐ in the present case along the axis of an optical fiber x. A
time slot duration AT corresponds to a packet duration tp and an inter-packet gap duration or
guard-band duration At.
The first step which is performed at the configuration of the network is the
determination of the maximum number of channels that can be gathered in a band. As
described previously, two parameters need to be taken into account for this determination.
First, the maximum number of channels that can be processed with the implemented
coherent receivers without introducing too much penalties is needed. This number depends on
the technology of the coherent receivers and is typically in the coherent receivers of the state
of the art equal to ten, which means that no more than ten channels can be gathered in a band.
Then, the time offset introduced by the chromatic dispersion between two channels of
a band along any transparent path of the network has to remain shorter than an inter-packet
gap At.
Thus, the length of the longest transparent path that a packet may possibly traveled within the
network is determined. Knowing this maximum length and the features of the links (induced
chromatic dispersion per length unit), the time offset introduced by the chromatic dispersion
along this maximum length between two channels spaced apart from a given wavelength
interval can be determined and compared to the inter-packet gap At. In the present example,
the maximum wavelength interval that produces an offset shorter than At corresponds for
instance to seven channel spacings Dl. In such case, the maximum number of channels in a
band has to be limited to eight. Thus, the POADM 1 of the network is configured to process
bands having a maximum of eight channels. In the present case, with sixty-five channels, the
channels can be gathered in eight bands having each eight channels plus the control channel
c As a consequence, eight packet add/drop structures 11 are required in the POADM 1 of
each node of the network to process the sixty-four data channels.
Fig.9 represents an example of a network path with four nodes noted Nl (1=1, 2...4)
linked by three links L -2, L2-3 and L3-4 respectively between nodes N1-N2, N2-N3 and N3-
N4. The optical nodes Nl (1=1. ..4) comprises a POADM 1 implemented as described in figures 1
to 3. The sixty-five multiplexed channels are transmitted from node Nl toward node N4 through
nodes N2 and N3. At a given time, two packets need to be transmitted at the same time from
node Nl to node N3.
These two packets are placed in channels lΐ and l9 which correspond to two different bands.
Indeed, as described previously, only packets of different bands can be dropped simultaneously.
Fig. 10 represents the arrangement of the packets in their time slots at the ingress node
Nl for the set of sixty-five channels. For sake of clarity, only 6 channels noted l , l2, l8, l9,
lΐ q and l65 are represented. Channels lΐ , l2 and l8 belong to band Bl, channels l9 and lΐ q
belong to band B2 and channel l65 is the control channel l ί . The represented time interval
corresponds to four time slots noted TSl, TS2, TS3 and TS4. The two packets of interest that
need to be transmitted to node N3 are coloured in black and are noted RG and P9' . These two
packets are sent within time slot TS2 in channels lΐ and l9 respectively. The other packets are
aimed to node N4.
These channels are multiplexed and sent by node N l to node N2 through the link Ll-2.
Fig. 11 represents the arrangements of the packets when they are received at the input of the
node N2. The control channel l65 is considered to be the reference so that the time slots in
fig. 11 are set according to the frames T of the control channel l65. Due to the inter-channel
chromatic dispersion, the other channels are time shifted with respect to the control channel. The
packets of channel lΐ that has the largest wavelength difference with respect to the control
channel l65 have the largest time offset with respect to the start of theframe T of the control
channel l65. However, the time shift or time offset between two channels within a band, for
example the time shift ts between channel l and channel l8 remains shorter than the interpacket
gap At (in practice, packet R is still ahead of packet P8" which belongs to the next time
slot TS3).
The data transmitted by the control channel l65 are then detected by the dedicated receiver.
Indeed, the control channel l65 is demodulated and the data transmitted in the control channel
l65 are detected in each node. The information about the chromatic dispersion is updated with
the chromatic dispersion undergone along the link N1-N2 for each of the packets transmitted
along the other channels (channels lΐ to l64). As no packet is dropped in node N2, these
updated information are encoded, modulated and emitted by the dedicated transmitter to be remultiplexed
with the other channels (channels lΐ to l64) which have been transmitted
transparently by the POADM 1 of node N2. However, due to the delay lines 13, the bands Bl,
B2...B8 are re-synchronized with respect to the control channel, for instance, the last channel of
each band is re-synchronized with the control channel as represented in fig.12 so that only the
small offsets between channels of a common band still remain.
The multiplexed channels are then transmitted from node N2 to node N3.
Fig. 13 represents the arrangements of the packets when they are received at the input
of node N3. The time shift ts between channels lΐ and channel l8 is longer than at the
reception in node N2 due to the inter-channel chromatic dispersion undergone between node N2
and node N3 but remains shorter than the inter-packet gap At. The data of the control channel
l65 transmitted in the frame T of the time slot TS1 and that comprises the information about
the chromatic dispersion of packets PI' and P9' (the transmission of the control data in the
previous time slot allows the dedicated receiver 7 to have time to detect and process the data of
the control channel corresponding to packets R and P9' before the detection of the data
packets PI' and P9'). The information about the chromatic dispersion undergone by packets PI'
and P9' transmitted in the control channel is retrieved by the dedicated receiver 7. The retrieved
value is updated with the value of the chromatic dispersion undergone between node N2 and
node N3. The updated values corresponding to packet RG and P9' are then transmitted
respectively to the coherent receivers processing the band Bl and the band B2. When the time
slot TS2 is received by the coherent receiver 25 processing the band Bl, the local oscillator 4 1
is tuned on the wavelength corresponding to lΐ and the value transmitted by the dedicated
receiver 7 concerning the chromatic dispersion undergone by packet PI' along its travelling
through the network (between node N and node N3 in the present case) is used to configure the
electronic dispersion compensation module of the coherent receiver 25 in order to compensate
for the intra-channel chromatic dispersion and to retrieve the data encoded in packet PI'. In the
same way, the data encoded in packet P9' are also retrieved by the coherent receiver 25 processing
the band B2. Furthermore, the data transmitted in the channels l and l9 during time slot TS2
are blocked by the optical gates 29 of the packet add/drop structures processing bands Bl and
B2 so that new data packets that need to be transmitted toward N4 can be added within time
slot TS2 in the channels l and l9 by the transmitter 35. Besides, the packets that are not
dropped in node N3, for instance packet P2', are transmitted transparently through node N3
toward node N4. The information concerning the chromatic dispersion undergone by these
non-dropped packets is updated with the value corresponding to the link N2-N3 and the
updated value is re-encoded within the control channel to be sent toward N4. Furthermore, the
information about the chromatic dispersion undergone by the packets transmitted in channels l
and l9 and time slot TS2 is reset to zero as new packets are emitted from node N3 in these
time slots. Thus, in each node the inter-channels chromatic dispersion is compensated per
band and the information about the undergone chromatic dispersion is updated in the control
channel in order to process the intra-channel chromatic dispersion at destination, allowing
thus to deal with both effects of the chromatic dispersion.
Besides, it has to be noticed that the configuration described above enables a
compensation of the chromatic dispersion without requiring in-line components such as in
line compensation fibres. As a consequence, such configuration is particularly adapted in the
case of a network comprising a combination of equipments providing wavelength granularity
capability with equipments providing packets granularity capability. Indeed, structures with
packet granularity such as POADMs 1 are more and more implemented due to their higher
flexibility with respect to the wavelength packet granularity structures such as the
Reconfigurable Optical Add/Drop Multiplexers (ROADMs). However, as ROADM are
already implemented and as the packet granularity is interested in the case of low or bursty
traffic to optimize the network capacity, a combination of both ROADM and POADM
appears to be a good trade-off to limit the capital expenditure while increasing the flexibility
of the network.
Fig. 14 represents an example of an optical node 49 that combines a ROADM 50 with two
POADMs 1. In practise, only one or more than two POADMs may also be gathered with a
ROADM 50.
The represented node 49 comprises two inputs 51a and 51b that receive signals respectively from
optical links 4a and 4b and two outputs 53a and 53b that transmits signals to two optical links 22a
and 22b. The inputs 51a and 51b are linked respectively to amplifiers 55a and 55b such as an
Erbium Doped Fibre Amplifier (EDFA) in order to amplify the received signal. Indeed, as the
transmission through the links 4a and 4b induces losses, the received signal comprising a set of
channels may need to be amplified to enable a good detection at the receivers. The output of the
amplifiers 55a and 55b are respectively linked to demultiplexers 57a and 57b which are configured
to split the received signal comprising a plurality of multiplexed channels into two signals
comprising each a subset of channels, the first subset corresponding to the channels aimed at
being processed by the ROADM 50 and the second subset corresponding to the channels aimed
at being processed by a POADM 1. Thus, the demultiplexer 57a and 57b comprise one input and
two outputs and may be implemented as 1-to-two WSS. For each demultiplexer 57a and 57b, the
first subset is then transmitted to an optical coupler 60a or 60b respectively to transmit the
channels either directly to a multiplexer 58a or 58b if they are not dropped or to a drop structure
59 if they are dropped. For example, the WDM signal received at the input 51a may comprise 73
channels, a first subset of 8 channels (from l66 to l73) is destined to a drop structure 59 and a
second subset of 65 channels (from l to l65) is destined to the POADM 1. However, among
the 8 channels processed by ROADM 50 (from l66 to l73), if channels are not dropped in node
49, they are transmitted directly to the multiplexer 58a or 58b to be transmitted transparently
toward another node through optical links 22a or 22b. The channels of the second subsets are
then transmitted to a first POADM 1 and processed as described previously. The channels of the
first subsets that need to be dropped are transmitted to the drop structures 59 where they are
demultiplexed by a demultiplexer 61, for example a WSS, to be transmitted individually to a
receiver 63 to be detected. In the same way, the channels of the first subset received at the input
51b are transmitted either directly to the multiplexers 58a or 58b or to a drop structure 59 while
the channels of the second subsets received at the input 51b are transmitted to a second
POADM 1. The ROADM 50 also comprises add structures 67 comprising transmitters 69
configured for emitting signals corresponding respectively to the wavelength of the channels l66
to l73 notably. The transmitters 69 are linked to a multiplexer 1 and then to the multiplexers
58a and 58b to be remultiplexed with the channels transmitted transparently and the channels
processed by the POADMs 1. Amplifiers 73a and 73b, such as an EDFA, may also be
implemented at the output of the multiplexers 58a and 58b before the transmission toward
links 22a and 22b in order to compensate for the losses that the WDM signal will undergone
along the links 22a and 22b.
According to another embodiment represented in fig. 15 in the case of only one input
51, one output 53 and one POADM 1, the demultiplexer 57 (57a and 57b in fig. 14) and the
optical coupler 60 (60a and 60b in fig. 14) may be replaced by a single demultiplexer 75
comprising a plurality of outputs linked respectively to the POADM 1, the drop structures 59
and the multiplexer 58 The demultiplexer 75 in then configured to split the received signal in
a plurality of signal comprising the subsets of channels aimed respectively to the POADM 1,
the drop structures 59 and the multiplexer 58. The demultiplexer 75 may be implemented as a
WSS.
Thus, the gathering of adjacent channels in bands, the intra-channel chromatic dispersion
compensation using a delay line per band associated with the intra-channel chromatic dispersion
compensation using digital signal processing means combined with the transmission of an
estimated value of the chromatic dispersion undergone by a packet along its transmission in a
control channel enables compensating for both aspects of the chromatic dispersion within a
node and in an amount of time compatible with the packet granularity constraints. Furthermore,
such compensation does not require any in-line components so that its implementation requires
only limited capital expenditures and is particularly adapted to enhance existing wavelength
switching equipment with packet granularity capability allowing higher flexibility at a reduced
cost.
CLAIMS
1. Method for compensating, within a node of an optical network, chromatic
dispersion undergone by optical packets transmitted within time slots of wavelength
division multiplexed channels along at least one link (4, 22) of the optical network, a
time slot duration (DT) corresponding to the sum of a packet duration (tp) and an
inter-packet gap duration (At), the said method comprising the foliowings steps:
- demultiplexing the wavelength division multiplexed channels into a plurality of
bands (B), a band (B) comprising a predetermined number of adjacent wavelength
channels (l),
- transmitting the said plurality of bands (B), via a respective plurality of delay lines
(13) having predetermined delays, toward a respective plurality of packet add/drop
structures (11) comprising a coherent receiver (25),
wherein the said predetermined number of channels of one band is determined so that
a first time shift, due to the effect of the chromatic dispersion along transmission
through the network, between two optical packets of the same time slot sent
respectively in different channels of the same band, remains shorter than an interpacket
gap duration (At) and so that the coherent receiver (25) is capable of processing
the said predetermined number of channels of one band,
wherein the predetermined delay of a delay line (13) associated with a band of
channels corresponds to a second time shift between a channel of the said associated
band and a reference channel (l65), the said second time shift being due to the effects
of chromatic dispersion along the last crossed link (4).
2. Method in accordance with claim 1 wherein one channel of the wavelength
division multiplexed channels corresponds to a control channel (l ϊ ) and transmits
control optical data comprising information about the chromatic dispersion undergone
by optical packets transmitted on other wavelength division multiplexed channels than
the control channel (l ϊ ), the said control channel (l ϊ ) being demodulated and
processed separately by a dedicated receiver (7) and transmitted separately by a
dedicated transmitter (19).
3. Method in accordance with claim 2 wherein the information transmitted by the
control channel is decoded by a dedicated receiver (7) when received in a node so that
the said information is updated with the value of the chromatic dispersion undergone
by optical packets transmitted on other wavelength division multiplexed channels than
the control channel (l ϊ ) along the last crossed link (4).
4. Method in accordance with claim 2 or 3 wherein the coherent receivers (25)
comprise an electronic dispersion compensation module and wherein the information
about the chromatic dispersion undergone by a dropped optical packet transmitted by
the control channel (l ϊ ) is retrieved by the dedicated receiver (7) and is transmitted to
the coherent receiver (25) receiving the dropped optical packet, the electronic
dispersion compensation module of the said coherent receiver (25) being configured
according to the said information to compensate at least partially for the intra-channel
chromatic dispersion.
5. Method in accordance with claim 4 wherein the coherent receivers (25) comprise
an adaptive equalizer associated with a constant modulus algorithm to compensate for
remaining signal degradations at the output of the electronic dispersion compensation
module.
6. Method in accordance with one of the previous claims wherein a packet add/drop
structure (11) also comprises an optical coupler (23) to transmit the band on one side
toward the coherent receiver (25) and on the other side toward an input of a
demultiplexer (27) configured for demultiplexing channels of a band, the outputs of
the said demultiplexer (27) being linked to the inputs of a respective plurality of
optical gates (29) configured to free the time slots corresponding to dropped optical
packets, the output of the optical gates (29) being connected to the input of a
multiplexer (31) configured to remultiplex the channels of the band, the output of the
multiplexer (31) being linked to an optical coupler (33), a transmitter (35) being also
linked to the said optical coupler (33) which is configured for adding packets received
from the transmitter (35) within the available time slots of the channels of the band,
the output of optical coupler (33) being linked to an output of the packet add/drop
structure (11), the said output of the packet add/drop structure ( 11) being linked to an
input of a band multiplexer (15) configured for remultiplexing the remultiplexed
bands received from the plurality of add/drop structures ( 11).
7. Packet optical add/drop multiplexer (1) located in a node of a wavelength division
multiplexing optical network and configured to process optical packets transmitted
within time slots (TS) having a duration (DT) corresponding to a packet duration (tp)
and an inter-packet gap (At) along links (4, 22) of the optical network and received
from remote nodes of the optical network, the said packet optical add/drop
multiplexer (1) comprising:
- a plurality of packet add/drop structures ( 11) comprising a coherent receiver (25),
- a band demultiplexer (9) configured for demultiplexing the received multiplexed
channels into a plurality of bands (B), a band comprising a predetermined number of
adjacent channels (l), the said predetermined number of channels being determined so
that a first time shift, due to the effect of the chromatic dispersion along transmission
through the network, between two packets sent respectively in a first and a second
channel of the band, remains shorter than an inter-packet gap (At) and so that the
coherent receiver (25) is capable of processing the said predetermined number of
channels,
- a plurality of delay lines (13) having predetermined delays, the plurality of bands (B)
being transmitted respectively to the plurality of packet add/drop structures ( 11) via
the said plurality of delay lines (13), the predetermined delay of a delay line (13)
associated with a band being determined according to a second time shift between a
channel of the associated band and a reference channel (l65), the said second time
shift being due to the effects of chromatic dispersion along the last crossed link (4).
8. Packet optical add/drop multiplexer (1) in accordance with claim 7 wherein it also
comprises:
- a dedicated transmitter (19) configured for transmitting control optical data in a
control channel (l ϊ ),
- a dedicated receiver (7) configured for processing control optical data transmitted in
a control channel (l ϊ ) .
9. Packet optical add/drop multiplexer (1) in accordance with claim 8 wherein the
dedicated receiver (7) is configured for retrieving information about the chromatic
dispersion undergone by the optical packets transmitted on other channels than
the control channel (l ί ), updating the said information and, for the dropped packets,
transmitting the said information to the coherent receivers (25) of the packet add/drop
structures (11) processing the said dropped optical packets, the said coherent receivers
(25) comprising an electronic dispersion compensation module which is configured to
receive, from the dedicated receiver (7), information about the chromatic dispersion
undergone by a dropped optical packet and to process the said dropped optical packet
according to the said information.
10. Packet optical add/drop multiplexer (1) in accordance with claim 9 wherein the
packet add/drop structures ( 11) also comprises:
- a demultiplexer (27),
- a plurality of optical gates (29),
- a multiplexer (31),
- a transmitter (35),
- a first optical coupler (23) to transmit the band on one side toward the coherent
receiver (25) and on the other side toward an input of the demultiplexer (27)
configured for demultiplexing channels of a band, the outputs of the said
demultiplexer (27) being linked respectively to the plurality of optical gates (29)
configured to free the time slots corresponding to dropped optical packets, the output
of the plurality of optical gates (29) being connected to a multiplexer (31) configured
to remultiplex the channels of the band,
- a second optical coupler (33) to receive the band transmitted from the multiplexer
(31) and to insert the optical packets transmitted from the transmitter (35) in the free
time slots of the band,
and wherein the packet optical add/drop multiplexer (1) also comprises a band
multiplexer (15) configured for remultiplexing the bands received from the plurality of
add/drop structures (11).
11. Packet optical add/drop multiplexer (1) in accordance with one of the claims from
7 to 10 wherein the band demultiplexer (9) is implemented as a wavelength selective
switch.
12. Optical node of a wavelength division multiplexing optical network comprising a
plurality of nodes linked by optical links (4, 22) comprising:
- a data repository configured for storing information about the topography of the
links (4, 22) adjacent to the node,
- a packet optical add/drop multiplexer (1) according to one of the claims from 7 to 11
wherein a dedicated receiver (7) is configured for updating information about the
chromatic dispersion undergone by the optical packets transmitted on other channels
(li... ) than the control channel (l ϊ ) based on the information about the topography of
the links adjacent to the node stored in the data repository.
13. Optical node (49) in accordance with claim 11 comprising also:
- a reconfigurable optical add/drop multiplexer (50),
- a demultiplexer (57) configured for separating a first subset of wavelength channels
destined to the packet optical add/drop multiplexer (1) from a second subset of
wavelength channels destined to the reconfigurable optical add/drop multiplexer (50)
and for transmitting the said first and second subsets respectively to the packet optical
add/drop multiplexer (1) and to the reconfigurable optical add/drop multiplexer (50).