WO2015055225A1

WO2015055225A1 - An optical receiver

Info

Publication number: WO2015055225A1
Application number: PCT/EP2013/071420
Authority: WO
Inventors: Valery NOBI ROSENTAL
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2013-10-14
Filing date: 2013-10-14
Publication date: 2015-04-23

Abstract

There is provided a device for processing a stream of traffic samples generated by digitally sampling an optical signal carrying traffic transmitted over an optical channel, the traffic having a symbol rate. The device comprises: a plurality of equaliser branches each comprising one or more equalisers configured to process the stream of traffic samples at a respective traffic sample rate. The device further comprises a selecting unit configured to select one of the plurality of equaliser branches to process the stream of traffic samples based on the symbol rate of the traffic and to selectively cause the stream of traffic samples to be processed by the selected one of the plurality of equaliser branches. There is also provided an optical receiver comprising the device, a method for processing, at an optical receiver, a stream of traffic samples generated by digitally sampling an optical signal carrying traffic transmitted over an optical channel, a computer program product and a computer comprising the computer program product.

Description

An Optical Receiver

Technical Field The present invention relates to a device for processing a stream of traffic samples generated by digitally sampling an optical signal carrying traffic transmitted over an optical channel. The present invention also relates to an optical receiver for receiving an optical signal carrying traffic transmitted over an optical link comprising a device for processing a stream of traffic samples generated by digitally sampling the received optical signal, a method for processing, at an optical receiver, a stream of digital traffic samples generated from an optical signal carrying traffic transmitted over an optical link, a computer program product which, when run on a computer, carries out the method and a computer comprising the computer program product. Background

A proposal to transmit traffic over an optical link at different symbol rates dependent on traffic demand is disclosed in an article co-authored by the inventors, titled "Cognitive Power Management in 100G Optical Transponders" by V.N Rozental et al published by the IEEE. In this proposal, although traffic may be transmitted over an optical channel or link at different symbol rates, system parameters such as the analogue-to-digital signal conversion rate at the optical receiver remain unchanged.

Figure 1 illustrates an example of this method. An optical transmitter 10 receives, in this example, a plurality of 1 Gb/s or 10 Gb/s Ethernet streams from a plurality of IP (Internet Protocol) Grouters 12. The optical transmitter 10 is capable of multiplexing the incoming information into a payload for transmission at 100 Gb/s, 50 Gb/s, 25 Gb/s or 12.5 Gb/s. The payload is mapped onto an optical channel for transmission. At the optical receiver (not shown), digital traffic samples are generated from the received optical signal by an analogue-to-digital signal convertor (also not shown) at a rate of two samples per minimum symbol period T_s. When the payload is transmitted at 100 Gb/s, one symbol is transmitted over minimum symbol period T_s. Thus, two samples are acquired per symbol. When however the traffic is transmitted at 50 Gb/s for example, as indicated in Figure 1 by crosses through the arrows, every second symbol is identical to the previous symbol (i.e. each symbol is repeated twice). This means that one symbol is transmitted over double the minimum symbol period T_s, such that symbols are transmitted at half the maximum symbol rate. Thus, in this example, four samples per symbol are acquired at the optical receiver. The samples are then however downsampled to leave two samples per symbol. Thus, in this example, every second sample is discarded and the traffic sample rate is halved.

In the article identified above it is suggested that, when the symbol rate of traffic received at the optical receiver changes, the filter length of a chromatic dispersion equaliser which processes the resultant traffic samples may be adjusted. Thus, if the symbol rate (and therefore the sample rate) of the traffic decreases, the power required by the chromatic dispersion equaliser to process the traffic samples may be reduced.

The inventors have appreciated however that it is complex to adjust the filter size of a chromatic dispersion equaliser and, whilst the filter size of the chromatic dispersion equaliser is being adjusted, data carried by the traffic may be lost.

Summary

According to the present invention, there is provided a device for processing a stream of traffic samples generated by digitally sampling an optical signal carrying traffic transmitted over an optical channel, the traffic having a symbol rate. The device comprises a plurality of equaliser branches each comprising one or more equalisers configured to process the stream of traffic samples at a respective traffic sample rate. The device further comprises a selecting unit configured to select one of the plurality of equaliser branches to process the stream of traffic samples based on the symbol rate of the traffic and to selectively cause the selected one of the plurality of equaliser branches to process the stream of traffic samples.

Embodiments of the present invention have the advantage that traffic at a lower symbol rate than a maximum symbol rate may be processed by one or more equalisers configured to process a stream of traffic samples at a traffic sample rate lower than the maximum traffic sample rate. This may enable power to be saved, in comparison to processing the traffic at a higher traffic sample rate. However, in contrast to the above proposal, the optical receiver may be less complex, and data may not be lost when the symbol rate of traffic received by the optical receiver changes. As will be understood by those skilled in the art, an "equaliser" is device suitable for processing a stream of traffic samples to compensate at least partially for one or more transmission effects, such as but not limited to chromatic dispersion, caused by transmission of the traffic over an optical channel or link.

In a preferred embodiment the selecting unit is configured to selectively cause the selected one of the plurality of equaliser branches to process the stream of traffic samples by causing the selected one of the plurality of equaliser branches to transition to a power up (i.e. active) state. If any of the other one or more equaliser branches is at a power up state, those equaliser branches may be transitioned to a power down (inactive) state. In this way, the non-selected equaliser branches may act as open circuits and therefore block the stream of traffic samples, such that the stream of traffic samples may only processed by the selected one of the plurality of equaliser branches. This implementation has the advantage that it may be simple. However, it should be appreciated that the selecting unit may be configured to selectively cause the selected one of the plurality of equaliser branches to process the stream of traffic samples in other ways, as will be discussed below.

Preferably, the respective traffic sample rate of the selected one of the plurality of equaliser branches is equal to the symbol rate of the traffic multiplied by a predefined sample per symbol factor. This may enable the traffic to be processed at the lowest possible traffic sample rate, whilst maintaining a predefined number of samples per symbol, and therefore the minimum amount of power to be used to process the stream of traffic samples.

In a preferred embodiment of the present invention wherein the stream of traffic samples is generated at a nominal traffic sample rate, at least one of the plurality of equaliser branches further comprises a downsampler configured to downsample the stream of traffic samples to its respective traffic sample rate. This has the advantage that, in embodiments where the stream of traffic samples is generated at a nominal traffic sample rate, a downsampler also does not need to be adjusted to downsample a stream of traffic samples by a different downsampling factor, or not at all, when the symbol rate of the received traffic changes.

Preferably, the at least one of the plurality of equaliser branches further comprises a digital low pass filter arranged to low pass filter the stream of traffic samples before the stream of traffic samples is downsampled by the downsampler. This has the advantage that the quality of the downsampled stream of traffic samples can be improved beyond that of the higher rate stream of traffic samples. This may be desirable in some applications. However, it should be appreciated that, if it is not desired to increase the quality of the signal beyond that of the higher rate signal, a low pass filter is not necessary.

In a preferred example, the downsampler is configured to downsample the stream of traffic samples by a downsampling factor, and the digital low pass filter has a cutoff frequency (normalised by the nominal sample rate divided by a predefined sample per symbol factor) equal to one divided by the downsampling factor. This may enable the low pass filter to optimise the quality of downsampled stream of traffic samples. The selecting unit may be configured to: receive a signal having an indication of the symbol rate of the traffic; and select one of the plurality of equaliser branches to process the stream of traffic samples based on the indication. This has the advantage that, in particular in embodiments where the stream of traffic samples is generated at a nominal traffic sample rate (and subsequently downsampled where necessary), the symbol rate of the traffic may be easily determined by the selecting unit.

Each of the plurality of equaliser branches may comprise a chromatic dispersion equaliser. A "chromatic dispersion" equaliser is an equaliser suitable for processing a stream of traffic samples to compensate at least partially for chromatic dispersion caused by transmission of the traffic over an optical link or channel. A chromatic dispersion equaliser may be a "bulk" equaliser, as will be described in more detail below, or any other type of equaliser suitable for at least partially compensating for chromatic dispersion. For example, at low traffic sample rates an "adaptive" equaliser which is primarily intended to compensate for other transmission effects such as polarisation mode dispersion may be used to compensate for chromatic dispersion to a satisfactory degree.

In a preferred embodiment of the present invention, each of at least a first equaliser branch and a second equaliser branch of the plurality of equaliser branches comprises an adaptive equaliser having one or more parameters having an adaptable value. An "adaptive equaliser" is any equaliser which has one or more parameters having adaptable values, whereby those parameter values may change whilst the adaptive equaliser processes a stream of traffic samples.

In this preferred embodiment the selecting unit may further be configured to cause the second one of the plurality of equaliser branches, as well as the selected one of the plurality of equaliser branches, to process a last portion of the stream of traffic samples. The device further comprises a processing unit configured to determine a difference between an output of the second one of the plurality of equaliser branches and an output of the selected one of the plurality of equaliser branches, whereby, based on the determining, the adaptive equaliser on the second one of the plurality of equaliser branches can update its one or more parameters.

This preferred embodiment has the advantage that up-to-date parameter values may not be lost when a switch is made to process a subsequent stream of traffic samples on a different one of the equaliser branches from a previous stream of traffic samples and, when the switch is made, the subsequent stream of traffic samples may be immediately processed using up-to-date parameter values.

In a preferred embodiment, the selecting unit is configured to cause the second one of the plurality of equaliser branches to process the last portion of the stream of traffic samples, only when the second one of the plurality of equaliser branches is selected by the selecting unit to process a subsequent stream of traffic samples.

This has the advantage that the second one of the plurality of equaliser branches is only activated, in addition to the selected one of the plurality of equaliser branches, when it is determined that the second one of the plurality of equaliser branches is to process a subsequent or following stream of traffic samples. This means that, advantageously, the period in which the two equaliser branches are active simultaneously, and therefore both consuming power, can be reduced. There is also provided an optical receiver for receiving an optical signal carrying traffic transmitted over an optical channel, the optical receiver comprising a device as described above.

There is further provided a method for processing, at an optical receiver, a stream of traffic samples generated from an optical signal carrying traffic transmitted over an optical channel, the traffic having a symbol rate, the optical receiver comprising a plurality of equaliser branches each comprising one or more equalisers configured to process a stream of traffic samples at a respective traffic sample rate. The method comprises: selecting one of the plurality of equaliser branches to process the stream of traffic samples based on the symbol rate of the traffic represented by the traffic samples; and selectively causing the stream of traffic samples to be processed by the selected one of the plurality of equaliser branches. There is also provided a computer program product configured to, when run on a computer, perform the above method. The computer program product could be in the form of a signal, for example a downloadable signal, or in any other form. The computer program product may be stored on a computer-readable medium.

There is also provided a computer comprising the computer program product. Description of the Drawings

A preferred embodiment of the present invention will now be described, by way of example only, and with reference to the accompanying drawings in which:

Figure 1 illustrates an optical transmitter according to the prior art;

Figure 2 illustrates a schematic diagram of an optical receiver according to an embodiment of the present invention;

Figure 3 illustrates a portion of an optical receiver according to a preferred embodiment of the present invention

Figure 4 is a flow chart according to an embodiment of the present invention; and

Figure 5 illustrates a portion of an optical receiver according to a preferred embodiment of the present invention.

Description

Figure 2 illustrates a schematic diagram showing modules of an optical receiver 20 according to an embodiment of the present invention. The optical receiver 20 may be a coherent optical receiver. The optical receiver 20 is configured to process a received optical signal in the digital domain. The optical receiver 20 comprises an optical "front end" 22 configured to receive an optical signal carrying traffic transmitted over an optical channel or link, such as an optical fibre, by an optical transmitter (not shown). As will be understood by those skilled in the art, the traffic is made up from a number of symbols which may each, dependent on the modulation format used (for example on-off keying (OOK) or binary phase- shift keying (BPSK)), represent one or more bits. The optical "front end" comprises an analogue-to-digital signal convertor ADC (not shown) configured to generate (digital) traffic samples from the received optical signal by digitally sampling the received optical signal. In this way, the continuous optical signal is converted into a series of discrete digital samples. In this example, the ADC generates traffic samples at a constant traffic sample rate (in this example two samples per minimum symbol period T_s), regardless of the symbol rate of the traffic carried by the optical signal. However, it should be appreciated that, in alternative embodiments of the present invention, it is possible that the ADC may generate traffic samples at a variable or adaptive traffic sample rate, dependent on the symbol rate of the traffic carried by the optical signal.

The stream of traffic samples generated by the ADC is passed to an equaliser module 24.

An example of an equaliser module 24 according to a preferred embodiment of the present invention is illustrated in Figure 3. In this example, it is seen that two streams of traffic samples are received by the equaliser module 24, labelled Xi and x₂. These streams are generated from respective polarisations of the optical signal received by the optical "front end" 22. However, it should be appreciated that it is possible that, alternatively, only one stream of traffic samples is received by the equaliser module 24, and for ease of explanation in the following the processing of only one stream of traffic samples is described.

The equaliser module 24 comprises a plurality of equaliser branches, each comprising one or more (digital signal) equalisers 35, 36 configured to process a stream of traffic samples at a respective traffic sample rate. In this example there are four equaliser branches. These equaliser branches are arranged in parallel.

As indicated in Figure 3, a first of the equaliser branches is configured to process a stream of traffic samples at a maximum sample rate, 56G samples per second (corresponding to 28Gbaud, where there are two samples per symbol). A second of the equaliser branches is configured to process a stream of traffic samples at half the maximum sample rate, 28G samples per second (corresponding to 14Gbaud where there are two samples per symbol), a third of the equaliser branches is configured to process a stream of traffic samples at a quarter of the nominal sample rate, 14G samples per second (corresponding to 7Gbaud where there are two samples per symbol), and the fourth of the equaliser branches is configured to process a stream of traffic samples at an eighth of the nominal sample rate, 7G samples per second (corresponding to 3.5Gbaud where there are two samples per symbol). Each of the equaliser(s) on the respective equaliser branches are driven by a clock signal, indicated by a dashed line, at a respective clock rate. In this example, a clock source (not shown) generates a clock signal at a master clock rate. This clock signal is provided, at the master clock rate, to the equaliser on the first equaliser branch. The clock signal is however downsampled by one or more downsamplers 26 before being provided to the equalisers on each of the second to fourth equaliser branches, by a downsampling factor of 2, 4 or 8 respectively.

The equaliser module 24 further comprises a selecting unit 28 which is configured to select one of the plurality of equaliser branches to process a stream of traffic samples generated from the received optical signal, based on the symbol rate of the traffic represented by those traffic samples. Note that different equaliser branches may be selected to process successive portions of the stream of traffic samples generated by the ADC (from successive portions of the continuous optical signal received by the optical "front end"), where these successive portions carry traffic at different symbol rates. The selecting unit 28 may comprise any combination of hardware and or software. For example, the selecting unit 28 may comprise a processor.

In an example, the selecting unit 28 is configured to receive an indication of the symbol rate of the traffic represented by the traffic samples, for example from the optical transmitter which transmitted the optical signal or from a network control entity such as a network management system (NMS) or an SDN controller, and to select one of the plurality of equaliser branches to process the stream of traffic samples based on the indication.

However, other embodiments are possible.

The selecting unit 28 is further configured to selectively cause the selected one of the plurality of equaliser branches to process the stream of traffic samples. This may be achieved in several ways. For example, in a preferred embodiment the selecting unit 28 may selectively cause the selected one of the plurality of equaliser branches to process a stream of traffic samples by causing the selected one of the plurality of equaliser branches to transition to a power up state (also referred to as an active state). Where a different one of the plurality of equaliser branches has processed a previous stream of traffic samples, and is therefore at a power up state, the selecting unit 28 may further cause that equaliser branch to transition to a power down state (also referred to as an inactive state). In this way, each of the non-selected equaliser branches (which are at a power down state) may act as open circuits and therefore block the stream of traffic samples from passing through those equaliser branches.

As discussed above, this embodiment has the advantage that it may be easy to implement. However, other embodiments are possible, and for example the equaliser module 24 may comprise switching circuitry (not shown) capable of selectively causing a selected one of the plurality of equaliser branches to process a stream of traffic samples. The selecting unit 28 may be configured to cause the switching circuitry to cause the selected one of the plurality of equaliser branches to process a stream of traffic samples. For example, the switching apparatus may comprise an input switch which is arranged to receive the stream of traffic samples passed from the ADC to the equaliser module 24 and is operable to selectively pass the stream of traffic samples to the selected one of the plurality of equaliser branches.

Preferably, the selecting unit 28 selects the equaliser branch to process a stream of traffic samples which is configured to process the stream of traffic samples at a traffic sample rate equal to the symbol rate of the traffic multiplied by the predefined sample per symbol factor (in this example two samples per symbol).

Thus, in this example, if the symbol rate of the traffic is 28Gbaud, the stream of traffic samples is processed by the first equaliser branch. Similarly, if the symbol rate of the traffic is at 14Gaud, 7Gbaud or 3.5Gbaud, the stream of traffic samples is processed by the second to fourth equaliser branches respectively. However, it should be appreciated that traffic having a symbol rate of 7Gbaud for example could alternatively be processed at a higher traffic sample rate for example at 14Gbaud, without loss of data, and yet still at a lower traffic sample rate than the maximum traffic sample rate, and therefore using less power than if the traffic were processed at the maximum traffic sample rate.

In this embodiment, as mentioned above, the traffic samples received by the equaliser module 24 (generated by the ADM) are at a nominal traffic sample rate (corresponding to the maximum traffic sample rate). Each of the second to fourth equaliser branches comprises a downsampler 32 configured to downsample a stream of traffic samples passed to that equaliser branch to its respective traffic sample rate (before the stream is processed by its one or more equalisers). Thus, in this example, the downsamplers 32 on the second to fourth branches have downsampling factors of 2, 4 and 8 respectively. In this example, each of the second to fourth equaliser branches further comprise a low pass filter 34 arranged to low pass filter the stream of traffic samples before the downsampler downsamples the stream of traffic samples. Each low pass filter 34 is arranged to receive a stream of traffic samples which has been passed to the equaliser branch, low pass filter the stream of traffic samples to produce a low pass filtered stream of traffic samples, and pass the low pass filtered stream of traffic samples to the respective downsampler 32. The downsampler 32, in turn, is arranged to downsample the stream of traffic samples to produce a stream of traffic samples at a lower traffic sample rate (the respective traffic sample rate) and pass the stream of traffic samples to the first equaliser on its equaliser branch.

In this preferred example, as indicated in Figure 3, the digital low pass filter 34 has a cutoff frequency normalized by the maximum symbol rate (i.e. normalised by the nominal sample rate divided by the predefined sample per symbol factor) equal to one divided by the downsampling factor used by the respective downsampler 32. This enables the low pass filtering to be optimised to the respective traffic sample rate. However, other configurations are possible as will be appreciated by those skilled in the art, and in particular different types of low pass filters have different characteristics and so may be configured to optimise the signal-to-noise ratio in a different way, based on the bandwidth of the respective equaliser branch.

Once a stream of traffic samples has been processed by a selected one of the plurality of equaliser branches, the processed stream of traffic samples (which may now be at a different traffic sample rate from the input stream of traffic samples), labelled y ( yi and y₂), is output at a common output.

The processed stream of traffic samples is then passed to a symbol decision block 37 which is configured to identify symbols in the processed stream of traffic samples, as will be understood by those skilled in the art. Once the symbols are identified, a decoder may then recover the binary sequence represented by the symbols.

A flow chart showing steps for processing a stream of traffic samples generated by digitally sampling an optical signal carrying traffic transmitted over an optical channel according to an embodiment of the present invention is shown in Figure 4. The method comprises, at step 400, selecting one of the plurality of equaliser branches to process the stream of traffic samples based on the symbol rate of the traffic. At step 410, selectively causing the selected one of the plurality of equaliser branches to process the stream of traffic samples, and at step 420 processing the stream of traffic samples using the selected equaliser branch.

In this example, each of the plurality of equaliser branches comprises a chromatic dispersion equaliser. In particular, the first to third equaliser branches comprise a "bulk" equaliser 35, and in addition an "adaptive" equaliser 36 which is primarily intended to compensate for other transmission effects such as polarisation mode dispersion. The fourth equaliser branch comprises only an adaptive equaliser 36. As will be appreciated by those skilled in the art, at low samples rates, as mentioned above, such an adaptive equaliser 36 may be capable of compensating for chromatic dispersion to a satisfactory degree.

Bulk equalisers are known in the art and are equalisers intended to compensate for chromatic dispersion caused by transmission of the traffic over an optical link or channel. Bulk equalisers comprise one or more digital filters having a number of taps (or parameters or coefficients).

As seen in the following equation disclosed in a paper by S. Savory: Digital Filters for Coherent Optical Receivers, Optics Express vol. 16, pp.804-817, 2008, the number of taps required by a bulk equaliser to compensate for chromatic dispersion is proportional to the square of the traffic sample rate of the stream of traffic samples to be processed:

Όλ L

N, CD 2 x + 1 .

2cT smp

where N_CD is the number of equaliser taps required for chromatic dispersion compensation, D is the fibre dispersion coefficient, λ is the carrier wavelength, L is the length of the optical link T_sm the sample period and c is the speed of light.

Thus, the lower the traffic sample rate at which a bulk equaliser is intended to process a stream of traffic samples, the fewer the taps required, and therefore the smaller the bulk equaliser. For example, if the traffic sample rate is half the maximum traffic sample rate, the number of taps required is reduced by ¼. This may result in less power being required by the bulk equaliser to process the stream of traffic samples. In addition, even though the size of the equaliser module 24 will increase by virtue of including a plurality of equaliser branches, the increase in size may be relatively low. Taking this example, the size of the bulk equaliser on the second branch is 1/4 the size of the bulk equaliser on the first branch, and the size of the bulk equaliser on the third branch is 1/16 the size of the bulk equaliser on the first branch. Thus, the overall increase in the size of the equaliser module 24 is around 1/3. Each of the equaliser branches further comprises an adaptive equaliser. However, as will be appreciated by those skilled in the art, the bulk equaliser is the largest component in an equaliser module, consuming most of the power required by the equaliser module 24, and therefore including a plurality of adaptive equalisers may not significantly increase the size of, or the power consumed by, the equaliser module 24.

As described above, during steady state operation, only one of the plurality of equaliser branches is preferably active at a time. However, as will now be described, where each of at least two of the plurality of equaliser branches comprises an adaptive equaliser 36, as in this example, in a preferred embodiment of the present invention, two equaliser branches may be active at the same time in order to train the adaptive equaliser 36 on the non-selected branch. In comparison to a bulk equaliser, an adaptive equaliser has at least one parameter or coefficient which can adapt or change during processing of a stream of traffic samples.

In this preferred embodiment, the selecting unit 28 may cause a second equaliser branch to process a last portion of a stream of traffic samples, in addition to a first equaliser branch (selected to process the stream of traffic samples). Figure 5 illustrates an arrangement between a first and second equaliser branch according to a preferred embodiment of the present invention.

In this example, in Figure 5a, the upper equaliser branch (referred to as a first equaliser branch) is processing a stream of traffic samples. The adaptive equaliser 36 on that branch (referred to as a first adaptive equaliser 36) is therefore labelled active. Note that there is a switch 38 which is configured such that the output of the first adaptive equaliser 36 (a processed stream of traffic samples y) is passed to symbol decision block 37. As discussed above, symbol decision block 37 identifies the symbols in the traffic. In addition, a second equaliser branch, which comprises a second adaptive equaliser 36, is also processing the stream of traffic samples. This second adaptive equaliser 36 is also therefore labelled as active. Note however that the output of the second adaptive equaliser 36 is not passed to symbol decision block 37.

In this example, the processed stream of traffic samples output by the first equaliser branch (after symbols have been identified in the traffic) is passed to a traffic sample rate converter 40. The traffic sample rate converter 40 may be an upsampler or a downsampler, and is configured to convert the stream of traffic samples to the respective traffic sample rate of the second equaliser branch. In this example, the traffic sample stream is downsampled by a factor of 2 (but alternatively the traffic sample stream could be upsampled by a factor of 2, or upsampled/downsampled by a different factor). After downsampling/upsampling, in this example, the stream of traffic samples is input into the second adaptive equaliser 36, which allows the second adaptive equaliser 36 to update its parameter values. As will be understood by those skilled in the art, in this example, the second adaptive equaliser 36 comprises a processing unit (not shown) configured to determine a difference between an output of the second one of the plurality of equaliser branches and an output of the selected one of the plurality of equaliser branches, whereby, based on the determining, the adaptive equaliser on the second one of the plurality of equaliser branches can update its one or more parameters. However, in alternative embodiments, the processing unit may be located separate from the second adaptive equaliser.

In addition, it is possible that instead of downsampling the output from the first equaliser branch (where the first equaliser branch is configured to process the stream of traffic samples at a higher traffic sample rate than the second equaliser branch), the device may for example comprise circuitry which is configured to pass only a subset of the samples (for example every second sample) to the processing unit. Similarly, where the first equaliser branch is configured to process the stream of traffic samples at a lower traffic sample rate than the second equaliser branch, instead of upsampling the output from the first equaliser branch, the processing unit and or the second adaptive equaliser may be configured such that the updating step is only performed in respect of a subset of the samples, in this example every second sample.

As illustrated in Figure 5b, after the one or more parameters of the second adaptive equaliser have been updated, the second equaliser branch may now begin to produce system output (i.e. to process a subsequent stream of traffic samples, which after processing is passed to symbol decision block 37). The first equaliser branch may be transitioned to a power down (or standby) state. Switch 38 may be configured to allow only an output from the second equaliser branch to reach symbol decision block 37.

In a preferred embodiment, the selecting unit 28 is configured to only cause the second equaliser branch to process a last portion of a stream of traffic samples, in addition to the first equaliser branch, if the second equaliser branch is selected to process a subsequent stream of traffic samples. This enables the amount of time that the two equaliser branches are active simultaneously, and therefore both consuming power, to be minimised.

Furthermore, it should be appreciated that the last portion of the stream of traffic samples need only be long enough for the second adaptive equaliser to update its parameter values (for example a few nanoseconds).

Claims

1 . A device for processing a stream of traffic samples generated by digitally sampling an optical signal carrying traffic transmitted over an optical channel, the traffic having a symbol rate, the device comprising:

a plurality of equaliser branches each comprising one or more equalisers configured to process the stream of traffic samples at a respective traffic sample rate; and

a selecting unit configured to select one of the plurality of equaliser branches to process the stream of traffic samples based on the symbol rate of the traffic and to selectively cause the selected one of the plurality of equaliser branches to

process the stream of traffic samples.

2. A device according to claim 1 , wherein the selecting unit is configured to selectively cause the selected one of the plurality of equaliser branches to process the stream of traffic samples by:

causing the selected one of the plurality of equaliser branches to transition to a power up state.

3. A device according to claim 1 or 2, wherein the respective traffic sample rate of the selected one of the plurality of equaliser branches is equal to the symbol rate of the traffic multiplied by a predefined sample per symbol factor.

4. A device according to any preceding claim, wherein the stream of traffic samples is generated at a nominal traffic sample rate, and

wherein at least one of the plurality of equaliser branches further comprises a downsampler configured to downsample the stream of traffic samples to its respective traffic sample rate.

5. A device according to claim 4, wherein the at least one of the plurality of equaliser branches further comprises a digital low pass filter arranged to low pass filter the stream of traffic samples before the stream of traffic samples is downsampled by the downsampler.

6. A device according to claim 5, wherein the downsampler is configured to

downsample the stream of traffic samples by a downsampling factor, and the digital low pass filter has a cutoff frequency, normalised by the nominal sample rate divided by a predefined sample per symbol factor, equal to one divided by the downsampling factor.

7. A device according to any preceding claim, wherein the selecting unit is configured to:

receive a signal having an indication of the symbol rate of the traffic; and select one of the plurality of equaliser branches to process the stream of traffic samples based on the indication.

8. A device according to any preceding claim, wherein each of the plurality of equaliser branches comprises a chromatic dispersion equaliser.

9. A device according to any preceding claim, wherein each of at least the selected one of the plurality of equaliser branches and a second one of the plurality of equaliser branches different from the selected one of the plurality of equaliser branches comprises an adaptive equaliser having one or more parameters having an adaptable value;

wherein the selecting unit is further configured to cause the second one of the plurality of equaliser branches, as well as the selected one of the plurality of equaliser branches, to process a last portion of the stream of traffic samples; and the device further comprises a processing unit configured to determine a difference between an output of the second one of the plurality of equaliser branches and an output of the selected one of the plurality of equaliser branches, whereby based on the determining, the adaptive equaliser on the second one of the plurality of equaliser branches can update its one or more parameters.

10. A device according to any of claims 9 wherein the selecting unit is configured to

cause the second one of the plurality of equaliser branches to process the last portion of the stream of traffic samples, only when the second one of the plurality of equaliser branches is selected by the selecting unit to process a subsequent stream of traffic samples.

1 1 . An optical receiver for receiving an optical signal carrying traffic transmitted over an optical channel, the optical receiver comprising a device according to any of claims 1 to 10.

12. A method for processing, at an optical receiver, a stream of traffic samples generated by digitally sampling an optical signal carrying traffic transmitted over an optical channel, the traffic having a symbol rate, the optical receiver comprising a plurality of equaliser branches each comprising one or more equalisers configured to process the stream of traffic samples at a respective traffic sample rate, the method comprising:

selecting one of the plurality of equaliser branches to process the stream of traffic samples based on the symbol rate of the traffic; and

selectively causing the stream of traffic samples to be processed by the selected one of the plurality of equaliser branches.

13. A computer program product which, when run on a computer, performs the method according to claim 12.

14. A computer comprising a computer program product according to claim 13.