CONTROL OF WAVELENGTH SELECTABLE FILTERS IN OPTICAL NETWORKS
This invention relates to the control of tuneable filters in optical networks. It is particularly concerned with such filters used in Wave Division Multiplex (WDM) networks, especially, but not exclusively Dense WDM (DWDM) networks.
In DWDM networks it is well known to select an individual channel from the DWDM multiplex by using optical bandpass filters to demultiplex the individual channels. Typically, the composite signal is dropped from the network and split in a 1:N splitter where N is the number of channels in the multiplex. Each of the N outputs of the splitter has an optical bandpass filter which is tuned to the frequency of a given channel so that only that channel is passed. Usually, these filters are non-adjustable but it has been proposed to use tuneable wavelength selectable filters. If wavelength selectable filters are used, it is essential that they are precisely tuned to the centre frequency of the channel to be received. Failure to tune the filter sufficiently precisely will lead to an unacceptably high error rate in the received traffic. It is possible to use open loop control for the filter. However, this is disadvantageous as any variation in the filter centre frequency will lead to a degraded performance.
We have appreciated that it would be preferable to use closed loop control of the bandpass filter. However, in order to search for the exact centre frequency for the filter it is necessary to search on both sides of the actual operating point, at least by a small amount. This has the disadvantage that system performance will be degraded while the correct operating point is being determined. As a consequence, the maximum capacity of the
system cannot be exploited as the system must be designed to cope with errors caused by the filters being of the optimum frequency during tuning.
There is a need, therefore, for a system in which filters are tuned in a manner which does not degrade system performance and which is transparent to the user.
The invention overcomes the problems mentioned above by measuring the bit error rate of the data at a preset filter position and then dithering the filter frequency and measuring the error rate at each dithered frequency. The filter can be set to the frequency at which the error rate is a minimum. As the signal is sent with forward error control data, the process is transparent to the user.
More specifically, there is provided a method of tuning an adjustable optical filter which passes optical information from a DWDM optical network, the optical information being encoded using a forward error correction (FEC) code, the method comprising setting the adjustable filter to a predetermined frequency and measuring the bit error rate in information passed by the filter; adjusting the filter tuning incrementally to further frequencies on each side of the predetermined frequency and measuring the bit error rate at each further frequency; and setting the filter to the frequency at which the bit error rate is a minimum.
The invention further provides a tuneable optical filter system for use in an optical network in which optical information on the network is encoded with a forward error correction (FEC) code, comprising: a tuneable optical filter for receiving a DWDM signal from the
network and for passing a selected band of the signal; a forward error correction decoder for decoding information passed by the filter and generating an error signal indicative of the error rate in the decoded information; and a controller for varying the tuning of the optical filter incrementally about a predetermined frequency, for comparing the error signal at each frequency to which the filter is tuned and for setting the filter at the frequency at which the error signal is indicative of the lowest error rate.
Embodiments of the invention have the advantage that the exact optimum setting of the filter can be found in a manner that is transparent to the user. As embodiments of the invention use FEC bit error rate variations, the errors which are induced by dithering the tuning frequency are corrected by the FEC decoder and corrector before they reach the user.
As the optimum frequency of the tuneable filter can be found and maintained, the system can be operated at its maximum capacity so enhancing its usefulness.
A preferred embodiment of the invention will now be described, by way of the accompanying drawings, in which:
Figure 1 is a schematic illustration of an optical ring network;
Figure 2 is a schematic block diagram of an embodiment of the present invention; and
Figure 3 is a flow chart illustrating operation of a method embodying the invention.
Figure 1 shows, schematically, a two fibre 10GHz optical network which carries a 32 channel DWDM signal. A number of nodes 10 are arranged on the network. Traffic is sent on the network from a source node to a target node where it is dropped from the network.
Figure 2 shows a portion of one of the nodes 10 on the ring network. The optical DWDM signal is split from the network by a signal splitter 12 to produce a dropped signal output 14. The other output of the splitter is through traffic labelled T which remains on the network. The dropped optical signal is passed to a 32:1 splitter 16 which outputs 32 signals, each of which comprises the 32 channel multiplex. Figure 2 only shows how one of these channels is processed but it will be understood that similar processing is performed on each of the channels.
The output 18 from the signal splitter 16 is input to a tuneable filter 2Q. The filter is an essentially mechanical component whose centre frequency is controlled by a servo motor. The centre frequency can vary with time or with operating conditions such as temperature.
The output of the optical bandpass filter is a single wavelength signal which is passed to a receive side transponder 22 where the optical single wavelength signal is converted into an electrical signal. This electrical signal forms the input to a forward error control chip 24.
Forward error control is a well known method used in optical communications to enable a receiver to correct errors in the received data. A degree of redundant coding is introduced into the data stream according to one of a number of coding algorithms. One such algorithm is the Reed-Solomon algorithm which is a block based error correction code
which encodes an array of k data symbols as in input and returns an array of n symbols. Operating in a 10GHz environment, a Reed Solomon encoder will convert a 9.9Gbps symbol stream into a 10.7Gbps stream.
The forward error correction encoding is introduced when traffic at 9.9Gbps is converted to the 10.7Gbps symbol stream when it is input into the optical network. At the receiving node 10, this 10.7Gbps stream passes through the tuneable filter 20 to the FEC chip 24.
The FEC chip 24 comprises a decoder 26 which decodes the electrical data symbols and a corrector 28 which operates under the control of correction data received from the decoder i to correct the decoded data. The corrected data is then output to its intended destination as the 9.9Gbps stream.
The decoder 26 also calculates a bit error rate (BER) which is used as a control signal to control the tuning of the bandpass filter. In practice, the signal controls the filter servo. The bit error rate is a measure of the number of errors in the data stream and varies with the tuning of the filter. When the filter is in its optimum position, the bit error rate will be at a minimum.
The BER signal is fed to a microcontroller 30 which may typically be a microprocesser controlled device which generates a wavelength control signal to alter the tuning of the bandpass filter. This signal is first converted to an analog signal by a digital to analog converter 32 and then applied to a control input of the filter.
The manner in which the process operates will now be explained with reference to the flowchart of Figure 3.
At step 40, the channel to which the bandpass filter is to pass is set, for example from a remote network manager. For example, channel 20 may be selected. A value for the selected channel is retrieved from a look up table by the microcontroller 30 at step 42. For example for channel 20 this may be an angle of 23°. A control signal is output, converted to an analog signal and applied to the control input of the filter. At step 44, the bit error rate (BER) for this setting is retrieved from the FEC decoder. The microcontroller then determines whether there are any errors at step 46. If there are not, the filter is judged to be correctly positioned and the process loops back to step 40 with the BER being checked periodically. If the BER shows that there are errors the BER is stored and at step 48 the signal to the DAC 32 is incremented or decremented. For example, the signal change could represent +/-1E of filter position. In the example of Figure 3 the signal is incremented by IE to an angle of 24°. The error rate is then retrieved again at step 50 and the error rate at 24° compared with the error rate at 23° at step 52. If the error rate BER(T+i) is not greater than the error rate BER(τ), then the sense of the adjustment of the filter is correct and is moving towards the optimum point. In that case, the process loops back and increments the filter angle by a further 1°. If the error rate is increased such that BER(τ+1> > BER(χ), the adjustment is being made in the wrong direction and at step 54 the filter position is adjusted to a position 1° to the other side of the start position, that is 22°. The BER is retrieved and stored (as BET(T+D) at step 56. At step 58 the stored value is compared with the initial bit error rate. If the rate is falling a further decrement is made at step 54. If it is
increasing the processing reverts back to step 44. Eventually a position will be established where BER is at a minimum. This position corresponds to the optimum filter position.
Because the FEC chip corrects errors in the data passed to the user, the filter can be tuned by deliberately introducing errors. The FEC chip ensures that these errors are transparent to the user and that the user continues to receive error free data. The use of dithering either side of the predetermined optimum point of the filter does not increase the error rate to such an extent that the FEC chip can no longer correct the errors.
The optimisation of the filter is performed gradually and continues while the system is running. If the centre frequency of the filter drifts, for example with time or temperature, the FEC error correction rate will measure a skew, showing that the degradation is greater in one direction than the other when the output is dithered. The filter can then be pulled back to its optimum operating point.
By maintaining the filters precisely at their operating points, the system can be operated at full capacity.
Although the embodiment has been described with reference to a ring network, it should be appreciated that the invention is not limited to the type of optical network with which tuneable filters are used.
Various modifications are possible and will occur to those skilled in the art without departing from the scope of the invention which is defined by the following claims.