Note: Descriptions are shown in the official language in which they were submitted.
CA 022321~9 1998-03-13
W O 97/10658 PCT/GB96/02278
- 1 --
METHOD FOR INDEPENDENTLY CONTROLLING THE WAVELENGTH COMPONENT POWERS IN AN
OPTICAL WAYELENGTH DIVISION MULTIPLEXED TRANSMISSION SYSTEM
This invention relates to a method of and
apparatus for controlling the relative amplitudes of
the individual wavelength components of a wavelength
division multiplexed optical signal.
The use of wavelength division multiplexing on
optical communications systems is rapidly expanding in
order to increase the information carrying capacity of
a system. In addition, such multiplexing also allows
the provision of network switching and protection
functions in an effective and economic manner. A
wavelength division multiplexed optical signal
propagated along optical fibre carries several channels
at different wavelengths. At transmission, the
wavelength component for any one channel normally has
the same amplitude as that of the other channels, but
as these wavelength components are processed through a
network, the relative amplitudes of the channels become
unbalanced.
The optical power per channei at key points in a
network will vary depending upon the path taken to
reach a given key point. Moreover, the optical power
will vary dynamically should network reconfiguration o
re-routing take place. Any initial wavelength
imbalance in a transmitter will exacerbate the
variation in channel optical power following processing
of the optical signal through the network.
A further problem is that a typical optical
amplifier has a non linear transfer function, in that
the gain varies dependent upon the wavelength being
amplified. If a wavelength division multiplexed
optical signal is passed through such a non linear
amplifier, any imbalance in the input signal will be
worsened, so degrading the network performance.
The present invention aims at providing both a
method of and apparatus for addressing the difficulties
CA 022321~9 1998-03-13
W O 97/10658 PCT/GB96/02278
-- 2 --
arising from optical power variation across the
channels of a wavelength division multiplexed optical
signal consequent upon the propagation of the
wavelength components of that signal through a network.
According to one aspect of the present invention,
there is provided a method of controlling the relative
amplitudes of the individual wavelength components of a
wavelength division multiplexed optical signal, in
which method the wa~;elength division multiplexed
optical signal is processed through an adaptive optical
wavelength filter, the relative powers of each
wavelength component of the optical signal are
determined, and a complex control signal is supplied to
the adaptive optical wavelength filter which control
signal includes a control component for each wavelength
component of the multiplexed optical signal, the
magnitude of each control component being adjusted
dependent upon the determined power of the respective
optical signal wavelength component.
The present invention makes use of the transfer
characte}istics of an adaptive optical wavelength
filter, known per se Embodiments of such filters
include acousto-optic and electro-optic tunable
filters. In the case of either of these types of
filter, there is an optical waveguide for an optical
signal and the transfer function for an optical signal
is defined by virtue of a stress-induced birefringence
The interaction between the optical signal and the
stressed waveguide results in a polarisation conversion
of the optical signal. As a result, if polarisation
selective elements are added before and/or after the
interactive section of the filter, the passband on the
(or each) output port of the filter will be governed by
whether polarisation conversion has occurred.
In the case of an acousto-optic filter, the
stress-induced birefringence is defined by injecting
-
CA 022321~9 1998-03-13
W O 97/10658 PCT/GB96/02278 - 3 -
into the filter a lower ~requency waveform, i.e.
electromagnetic energy typically in the range of a few
hundred MHz In the case of an electro-optic filter,
the birefrigence is induced by an electrode structure
arranged along the wave guide.
An adaptive optical wavelength filter as described
above allows for the selective addition or subtraction
of a wavelength division multiplexed channel from an
optical communication network. The simultaneous
addition or subtraction of multiple channels can be
achieved by the use of an appropriately configured
filter of this kind.
The two types of adaptive optical wavelength
filter usually employed in optical networks are known
as acousto-optic and electro-optic tunable filters. In
the former, a relatively low frequency control signal
is applied to the filter, and in the latter, a d.c.
control signal is applied to the filter electrode
structure, there being a separate electrode structure
for each channel. Either kind of filter may be used in
the present invention. Both kinds of filter may be
configured to have more than one output port and it is
preferred to use such a filter, with the channel power
determination being performed on the output from one
port, and the principal optical signal being propagated
from the other output port. Conveniently, for the
generation of an appropriate control signal, the
output at said one port is the inverse of the output at
said other port. Alternatively, the filter may have a
single output port and the channel power determination
is performed on the optical signal obtained from a
~ passive tapping on the output from that port.
In performing the method of this invention,
consideration must be given to the locking, in terms of
absolute wavelengths, of the filter performing the
adaptive filtering and the elements performing the
CA 022321~9 l99X-03-13
W O 97/10658 PCT/GB96/02278 - 4 -
determination of power levels within the different
channels to define the adaption required in the filter.
A 'window' could be set, which would be a fraction
(perhaps 40~) of the channel separation between the
channels. The optical signals transmitted may
therefore be anywhere within this 'window' for each
channel, and so both the optical filter and the network
will need fairly broad responses to these 'windows !,
As a result, the filter and char.nel power determining
elements (analyser) will need to be locked tightly only
if the responses are well matched. A more relaxed
locking requirement would be possible if ~he analyser
has slightly wider 'windows' than the filter. In a
case where the analyser takes the form of a passive
demultiplexer, this may be achieved by changing the
specification of the filters and other components
within the demultiplexer. Alternatively, where the
analyser comprises a second active filter, a small
design change to widen the filter respon~e may be all
that is required to achieve a relaxed locking
regime, as both filters would, in such an arrangement,
be locked to the same driver circuit.
The transmission standards for an optical
communication network define specific centre
wavelengths for the network. Consequently, no locking
of the adaptive filters to each other, across the
network, should be required.
According to a second aspect of the present
invention, there is provided apparatus for controlling
the relative amplitudes of the individual wavelength
components of a wavelength division multiplexed optical
signal, which apparatus comprises an adaptive optical
wavelength filter through which the optical signal is
passed for processing therein, means to determine the
relative powers of the wavelength components of the
processed optical signal, control means responsive to
CA 022321~9 1998-03-13
W O 97/106~8 PCT/~,~ 22
- 5 -
the determined powers of each wavelength component and
providing a complex control signal to the adaptive
optical wavelength filter which complex control signal
includes a control component for each wavelength
component of the multiplexed optical signal, the
control means controlling the magnitude of each
control component dependent upon the determined power
of the respective optical signal wavelength component.
In the present invention, the complex control
signal has a control component for each wavelength
channel of the optical signal, the energy or magnitude
of each such component being controlled dependent upon
the detected power of each wavelength component of the
optical signal. In order to achieve a closed loop
feedback system, the relative powers of the individual
wavelength components of the optical signal should be
determined following the processing of that signal in
the adaptive optical wavelength filter.
When an acousto-optic tunable filter is employed,
the control signal will have a frequency component for
each channel of the optical signal, the energy of each
such frequency component being adjusted in order to
control the power of the respective wavelength
component of the optical signal. If an electro-optic
tunable filter is employed, the control signal will
have a d.c. component for each channel of the optical
signal, each d.c. component being applied to the
respective electrode structure of the filter to control
the power of the respective channel of the optical
signal, dependent upon the voltage of the applied d.c.
control component.
~ The determination of the power of each channel of
a wavelength division multiplexed optical signal may be
performed by any suitable manner known in the art. For
example, in-line optical filters may be employed to
separate a portion of the optical signal into its
CA 02232lS9 l998-03-l3
W O 97/10658 PCT/GB96/02278 - 6 -
individual wavelength components, the amplitude of each
of which then being determined for instance by an
individual photo-detector for each channel Instead of
the use of optical filters, a wavelength division
demultiplexer may be employed to separate a portion of
the optical signal into individual wavelength
components.
The determination of the power of each channel of
the optical signal may instead be determined by a
further adaptive optical wavelength filter similar to
that employed to control the magnitude of each
wavelength component of the optical signal. Such a
further filter may sample each channel of the optical
signal, in sequence, the output of the further
filter sequentially corresponding to the power of each
channel of the optical signal. By linking the two
filters to the same driver circu-t, the operation of
the two filters will be closely locked to each other
and the signal wavelengths being analysed. With this
arrangement, the two filters could be arranged on one
integrated circuit.
The magnitude of each control component of the
control signal may be adjusted such that the relative
powers of each wavelength component of the optical
signal are, after processing in the filter,
substantially the same. Alternatively, when the
processed optical signal is subsequently to be
processed through a non-linear component such as an
optical amplifier, each control component of the
control signal may be adjusted having regard to the
transfer function of the subsequent non-linear
component. In this way, the power of all wavelength
channels of the optical signal may be controlled so as
to be essentially the same, following processing
3~ through the non-linear component.
By way of example only, the invention will now be
CA 022321~9 1998-03-13
W O 97/10658 PCT/G B96/02278
-- 7 -
described in greater detail and certain specific
examples thereof given, reference being made to the
accompanying drawings, in which:-
Figure 1 schematically shows a network for the
processing of wavelength division multiplexed opticalsignals;
Figure 2 diagrammatically illustrates a first
example of a method of this invention;
Figure 3 diagrammatically illustrates a second
example cf a method of this invention, similar to that
of Figure 2;
Figure 4 diagrammatically illustrates a third
example, using a filter to analyse the channel
wavelengths;
Figure 5 illustrates yet another example, similar
to that of Figure 4;
Figure 6 shows the transfer function of an
adaptive filter, showing the slewing of power between
the secondary and main outputs of the filter, as used
in the embodiment of Figure 2; and
Figures 7A and 7B compare the output signals from
a non-linear optical amplifier, respectively without
and with adaptive balancing according to this
invention.
Figure 1 diagrammatically illustrates a network
including a plurality of switching nodes 10 for optical
signals propagated along optical fibres 11. The
signals may be multi-channel wavelength division
multiplexed signals and so there may be a plurality of
wavelengths appearing at any one or more of the nodes
10. For example, fibre 12 may be carrying a channel of
wavelength ~ 1 and of amplitude ~1, and fibre 13 a
channel of wavelength 2 2 and amplitude ~2. If these
channels are switched both to appear on fibre 14 as a
wavelength division multiplexed signal, though both
signals had the same amplitude on entering the network,
CA 022321~9 l99X-03-13
W O 97/10658 PCT/GB96/02278 - 8 -
following the processing through the nodes, the
relative amplitudes of the two channels will be as
shown at ~3. Upon subsequent processing of that
multiplexed signal, the dif~ference in the amplitudes of
5 the two channels will be exacerbated, leading to
possible difficulties in recovering the smaller
amplitude channel.
Figure 2 shows the processing of a wavelength
division multiplexed signal with channel amplitudes out
10 of balance, such as the signal on fibre 14 of Figure 1.
An electro--optic or acousto--optic adaptive filter 20
has an input port 21 and main and secondary output
ports 22 and 23 respectively. The filter further has a
control port 24. Such a filter is known per se in the
15 art and will not be described in further detail here.
Optical fibre 14 carrying a wavelength division
multiplexed signal is connected to the input port 21
and a further fibre to the main output port 22. All of
the input channels appear at both the main and
20 secondary ports, but the signal f~rom the secondary port
23 is supplied to a wavelength demultiplexer (not
shown, but known per se in the art) in order to provide
individual channel components to a group of photo--
detectors 25, with one channel component supplied to
25 each photo--detector respectively. The photo--detectors
each determine the power of the channel component
supplied thereto and in turn provide an output to a
control circuit 26. That circuit 26 controls the
operation of a plurality of oscillators 27, one
30 oscillator for each wavelength channel of the input
signal; the outputs of the oscillators 27 are combined
at 28 and supplied to the control port 24 of the
filter.
Figure 3 shows an arrangement similar to that of
35 Figure 2, but the adaptive filter 20 has only a main
output port 22. The input to the channel power
CA 022321~9 1998-03-13
W O 97/10658 PCT/GB96/02278
analyser section of the arrangement is taken from a
passive tapping 29 on the main output from the filter.
In all other respects this arrangement corresponds to
that of Figure 2 and will not be described in further
detail here.
The third embodiment shown in Figure 4 employs a
channel power analyser section utilising a second
adaptive wavelength filter 30 the input port 31 of
which is connected to the secondary output port 23 of
the principal adaptive filter 20. A single control
circuit 32 controls the operation of two separate
sets 33 and 34 of oscillators, the two sets of
oscillators being associated with the two adaptive
filters 20 and 30 respectively. The outputs of the
oscillators of each set are combined at 35 and 36
respectively and the resultant control signals ar~
supplied to the control ports 24 and 37 respectively,
of the two adaptive filters 20 and 30.
The output of the adaptive filter 30 is supplied
to a single photo-detector 38 and the signal indicative
of the power of the channel component instantaneously
supplied to the photo-detector 38 is fed to the control
circuit 32, in order to control the appropriate
oscil~ator of the set 33 associated with the filter 20.
In this way, the powers of the various wavelengths in
the signal leaving the main output port of the filter
20 may be balanced as required, with the analysis of
the powers of the channels being performed using the
further adaptive filter 30 for sampling the channels
one at a time, in sequence, under the control of
circuit 32. By using a single control circuit 32, the
operation of the embodiment may properly be
synchronised to ensure that the transfer function of
the filter 20 for each channel is properly matched to
the detected power of that channel.
Figure 5 illustrates a further embodiment similar
CA 022321~9 1998-03-13
W O 97/10658 PCT/~,CJ~227
- 1() -
to that of Figure 4 in that a second adaptive filter 30
is employed to perform the analysis of the multi-
channel signal passing through the filter 20. In this
case, rather than providing two separate sets of
oscillators a single set 40 is arranged to control both
filters 20 and 30. For controlling the filter 20, the
outputs from the oscillators are passed through
elements 41 the effective resistance of which can be
varied by the control circuit 32, before the outputs
are combined and supplied to the control port 24 of
filter 20. An output is also taken from each
oscillator to a switch circuit 43, the operation of the
switch being performed by the control circuit 32 so
that the appropriate oscillator output is supplied to
the control port of filter 30 in a timed relationship
to the operation of the oscillators.
In other respects, the operation of the embodiment
of Figure 5 corresponds to that of Figure 4 and will
not be described further here.
Figure 6 shows, for the embodiments any of Figures
2,4 or 5, the relationship between the optical powers
appearing at the main and secondary ports 22 and 23,
for any one wavelength channel of a signal supplied to
the input port Z1, when a suitable high frequency
control signal is supplied to the control port 24. By
adjusting the energy of the control signal, the
transfer function of the filter for the channel
associated with the frequency of the control signal can
be varied. The power of the output signal at the
secondary port 23 is the inverse of the power of the
signal at the main output port 22.
By suitably varying the energy of the control
signal, the power of the optical signal at the main
output port 22 may be controlled to have a desired
value. Such control is performed dependent upon the
determined power of the signal from the secondary
CA 022321~9 1998-03-13
W O 97/10658 PCT/G~022/8
1 1 _
,
output port 23. The control circuit may thus be
configured to define a closed loop system to ensure the
power of the optical system at the main output port is
maintained constant at a desired value, irrespective of
variations in the power of the optical signal at the
input port 21.
Figure 7A shows the effect of a typical non-linear
optical amplifier 45 on a wavelength division
multiplexed optica' signal the relative powers of the
individual channels of which are as shown at 46. As
can be seen at 47, following processing by the
amplifier 45, the relative imbalance in the channel
powers is increased. However, by subjecting the input
signal to the amplifier 45 to adaptive balancing by the
method and apparatus of this invention as described
above, the relative powers of the channels of the
output signal from the amplifier 45 may all be
substantially the same, as shown in Figure 7B at 48.
