Optical: systems and elements – Optical amplifier – Correction of deleterious effects
Reexamination Certificate
1999-07-27
2001-08-14
Tarcza, Thomas H. (Department: 3662)
Optical: systems and elements
Optical amplifier
Correction of deleterious effects
C359S341430, C372S020000
Reexamination Certificate
active
06275328
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method and apparatus for determining control signals of a filter for optimisation of the gain profile of an amplifier, particularly though not exclusively for use in optical communications systems, and a system incorporating the same.
BACKGROUND TO THE INVENTION
The control of optical power levels in optical communications systems is critical in obtaining optimum performance. The power level needs to be sufficient to establish a signal to noise ratio which will provide an acceptable bit error rate but without the power level exceeding a level at which limiting factors (e.g. the onset of non-linear effects) result in degradation of the signal. In wavelength division multiplexed (WDM) transmission, it is desirable to maintain each of the power levels of the individual wavelength components at substantially the same level.
FIG. 1
illustrates a typical WDM transmission system, in which optical signals are transmitted from the multiplexer
10
to the demultiplexer
12
via optical fibre
14
. The individual wavelength components for each channel are generated by the transmitters
16
(Tx) and sent to the receivers
18
(Rx). In order to ensure that optical power is maintained within each of the transmitted channels, one or more line amplifiers
20
are located along the optical fibre transmission path to compensate for power losses in the transmission system.
A typical line amplifier
20
comprises two EDFA (Erbium Doped Filter Amplifier) amplifying elements
22
,
22
′ separated by one or more filters
24
.
The gain of the EDFA (and hence the output
28
from line amplifier) depends upon both the optical power in the transmitted input signal
26
and the optical power from the pump laser (not shown). As
FIG. 2
illustrates, the shape of the gain (the gain profile) of an EDFA changes with the gain of the EDFA. The gain profile may also be affected by temperature, age and other effects. In order to maintain each of the power levels of the individual wavelength components at substantially the same level, it is desirable to have a flat gain profile over the wavelength range of the transmitted channels. A fixed filter cannot flatten the profile of the amplifier for all gain conditions. It is therefore desirable to provide an adaptive filter for the line amplifier to provide compensation for (i.e. to flatten) the varying gain profiles.
The article “Tunable Gain Equalisation Using a Mach-Zehnder Optical Filter in Multistage Fibre amplifiers” (Reference IEEE Photonics Technology Letters, Vol. 3. No. 8, August 1991, Pg718; Kyo Inoue, Toshimi Kominato, and Hiromu Toba) indicates how a tunable signal gain equalisation may be demonstrated using a waveguide type Mach-Zehnder (MZ) optical filter, such that by adjusting the MZ transmittance with the external control current, tuneable gain equalisation may be achieved at the output of each of the amplifier stages. Further, EP794,599 discloses a gain equaliser which includes a plurality of periodic optical filters for equalising the gain of an optical amplifier. This application suggests that the wavelength, phase and amplitude (attenuation) of the transparency characteristics of the filters may be adjusted to allow the optical SNR (Signal to Noise Ratio) in the transmission system to be equalised.
Neither of the above documents discloses the control strategies appropriate for such tunable filters.
As current system designs are approaching the limit of what it is possible to achieve with fixed filters, there is an increasing requirement for a device that will equalise the optical powers in the transmission system channels, and compensate for any non-flat losses in the system.
An ideal control algorithm must be both quick and stable. It must be robust to variations in the number of channels, and must be flexible enough to allow good control over the complete transmission system.
Various control techniques for obtaining the best fit of one or more periodic filters to such an amplification system would be apparent to a skilled person.
For instance, a figure of merit (a single value relating to how well the filters are achieving the required target spectral shape) might be defined by summing the square of the difference between the target profile and the actual profile at each wavelength. This sum of squares of the errors provides a single figure that illustrates how well the filters are fitting to the target shape, and penalises points that deviate from the target.
A brute force technique of stepping through all the possible values of the control variables, and optimising each one, could then be utilised to achieve the target amplifier profile. Equally, a simple optimisation of measuring the local gradient, and finding the minimum figure of merit would allow for a slow convergence on the optimum profile, although this might suffer by finding a false minimum of the figure of merit that provides a stable solution which is not optimum.
More sophisticated optimisations are also known, such as Golden Section searches and parabolic fits. Such optimisations can improve the speed of convergence by reducing the number of iterations required for each fit. However, such approaches will also suffer from errors due to oscillations around the desired target results.
Such oscillations may be caused by the inter-dependence of the various control axes (each axis representing the possible range of values of each control variable) on the figure of merit. Optimising one axis may well disturb several of the other axes from their optimum points.
A solution to the problem is to use a more complex figure of merit, for example to take a Fourier Transform of the error spectrum in order to measure the contributions of each of the filtering elements. Such measurements can then be utilised to modify the control parameters. The drawback of this particular method is that the Fourier Transform is not robust to gaps in error spectrum (the error spectrum being derived from the results from the individual quantised channels, with some systems only utilising some of the total possible number of channels). Also, it is not particularly tolerant of the response characteristics of real filters. However, it does allow for a number of axes to be optimised simultaneously, so that although the number of iterations does not reduce, the time for each iteration is normally reduced.
The object of the present invention is to provide a method for calculating a control signal for a filter arranged to alter the output profile of an amplifier that substantially addresses at least one of the problems of the prior art.
SUMMARY OF THE INVENTION
In one aspect, the present invention provides a method of generating at least one control signal for at least one periodic filter arranged to alter the output gain profile of an amplifier, at least one of the phase and amplitude of the filter response being tuneable, the method comprising the steps of: determining an error profile from the amplifier output gain profile and a predetermined target output gain profile; determining the convolution of the error profile and a function of similar period and form to the response of the filter; and determining at least one control signal for tuning of said filter from the calculated convolution.
Such a convolutional technique allows the amplifier control signals to be directly calculated from the convolution. Once the magnitude or value of the control signal is determined, the control signal may subsequently be generated. This technique reduces the number of iterations required for the amplifier to converge on the target gain profile. This is particularly important in systems where a rapid response is desired, but the error signal is updated relatively slowly compared with the control signal computational time. Additionally, the technique has been shown to be stable/robust to varying number of optical channels, which can result in gaps within the error spectrum and varying amplifier input power levels.
In another aspect, the present invention provides a meth
King Jonathan
Parry Simon
Robinson Alan
Hughes Deandra M.
Lee Mann Smith McWilliams Sweeney & Ohlson
Nortel Networks Limited
Tarcza Thomas H.
LandOfFree
Amplifier control does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Amplifier control, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Amplifier control will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2470851