JP2024122745A - Receiver and adaptive equalization processing method - Google Patents

Abstract

[Problem] To prevent an increase in circuit size and power consumption, and filter delay, and to easily compensate for polarization mode dispersion in response to an increase in optical transmission speed. [Solution] An adaptive equalization processing unit (100) provided in a receiver adaptively compensates for waveform distortion of data. The adaptive equalization processing unit (100) includes a frequency domain compensation unit (101) that compensates for waveform distortion of received data using a frequency domain filter (122), and a filter coefficient update unit (102) that includes a time domain processing filter (125) and calculates filter coefficients based on the received data input to the frequency domain compensation unit (101). In addition, a frequency domain coefficient conversion unit (103) converts the time domain filter coefficients calculated by the filter coefficient update unit (102) into the frequency domain, and outputs them to the frequency domain filter. The filter coefficient update unit (102) then calculates filter coefficients using only a portion of the received data input to the frequency domain filter (122). [Selected Figure] FIG.

Description

The present invention relates to a receiver and an adaptive equalization processing method.

Optical transmission transceivers are becoming faster and have larger capacities, and symbol rates are expected to rise further in the future. Symbol rates have risen from the previous level of around 30 GBaud to over 100 GBaud, and are expected to continue to rise. The receiver compensates for waveform distortion in the received signal by processing the received signal using digital coherent receiving technology. For example, it compensates for waveform distortion (polarization fluctuations, polarization mode dispersion (PMD), polarization dependent loss) that occurs in the received signal due to changes in the polarization state. PMD stands for Polarization Mode Dispersion.

The following patent documents disclose prior art related to polarization mode dispersion compensation. For example, there is a technique in which I/Q signals with the same polarization are chromatic dispersion compensated, signals with different polarizations are processed independently to recover the signals, PMD adaptation coefficients are periodically estimated, and the signals are converted to the frequency domain by FFT and converted back to the time domain by parallel-serial conversion. There is also a technique in which a coherent optical receiver is equipped with modules for chromatic dispersion (CD) compensation and PMD compensation, an adaptive block least mean square (LMS) equalizer generates and estimates filter tap updates for each block of samples, and includes FFT and IFFT modules. CD stands for Chromatic Dispersion, FFT stands for Fast Fourier Transform, and IFFT stands for Inverse FFT. There is also a technique for generating a discrete spectrum by converting a block of samples into the frequency domain, filtering the discrete spectrum with a static filter, converting the time coefficients into the frequency domain to generate the frequency coefficients of the filter, and converting the spectrum into the time domain. There is also a technique for a time domain decision feedback equalizer in which a feedback type time domain filter is used to perform sample-by-sample time domain updates in a frequency domain/time domain hybrid equalizer, and a frequency domain equalizer is included in the forward path (see, for example, Patent Documents 1 to 4 below).

US Patent Application Publication No. 2010/0142952 US Patent Application Publication No. 2020/0204267 Special table 2018-530974 publication Special Publication No. 2004-530365

In conventional time-domain processing, power consumption increases depending on the ability to compensate for polarization mode dispersion. For example, as the symbol rate increases, the number of taps (circuit size) of the FIR filter that compensates for waveform distortion caused by polarization mode dispersion increases, so in the future, it may become impossible to compensate for polarization mode dispersion with realistic power.

On the other hand, when time domain processing is converted to frequency domain processing using FFT (IFFT), the increase in symbol rate increases filter delay, which leads to a new issue of reduced polarization fluctuation resistance. For example, in frequency domain processing, the efficiency of filter processing is improved by converting input data blocks into the frequency domain all at once using FFT (IFFT), which reduces polarization fluctuation tracking ability.

In one aspect, the present invention aims to prevent increases in circuit size and power consumption, as well as filter delays, and to easily compensate for polarization mode dispersion in response to increasing optical transmission speeds.

According to one aspect of the present invention, in a receiver including an adaptive equalization processing unit that adaptively compensates for waveform distortion of received data, the adaptive equalization processing unit includes a frequency domain compensation unit that compensates for the waveform distortion of the received data using a frequency domain filter, a filter coefficient update unit that includes a time domain processing filter and calculates filter coefficients based on the received data input to the frequency domain compensation unit, and a frequency domain coefficient conversion unit that converts the filter coefficients in the time domain calculated by the filter coefficient update unit into the frequency domain and outputs them to the frequency domain filter, and the filter coefficient update unit calculates the filter coefficients using only a portion of the received data input to the frequency domain filter.

According to one aspect of the present invention, it is possible to prevent increases in circuit size and power consumption, as well as filter delays, and to easily compensate for polarization mode dispersion in response to increasing optical transmission speeds.

FIG. 1 is a diagram illustrating an adaptive equalization processing unit of a receiver according to an embodiment. FIG. 2 is a diagram showing an example of a configuration for polarization mode dispersion compensation using existing time domain processing. FIG. 3 is a table showing the characteristics of a polarization mode dispersion compensation circuit using an existing FIR filter. FIG. 4 is a diagram showing examples of existing polarization mode dispersion compensation processes. FIG. 5 is a diagram showing the polarization mode dispersion compensation performance in the existing time domain processing and frequency domain processing. FIG. 6 is a diagram illustrating problems with the existing time domain processing and frequency domain processing. FIG. 7 is a diagram illustrating an example of a configuration of a filter coefficient update unit of the adaptive equalization processor according to the embodiment. FIG. 8 is a diagram illustrating an example of a hardware configuration of a control unit of the adaptive equalization processing unit. FIG. 9 is a flowchart illustrating an example of processing by the adaptive equalization processor according to the embodiment. FIG. 10 is a diagram illustrating an example of a configuration of an optical transmission and reception system including an adaptive equalization processing unit according to an embodiment. FIG. 11 is a diagram illustrating an adaptive equalization processing unit of a receiver according to another embodiment. FIG. 12 is a diagram illustrating an adaptive equalization processing unit of a receiver according to another embodiment.

The following describes in detail an embodiment of the disclosed receiver and adaptive equalization method with reference to the drawings. The adaptive equalization method of the embodiment compensates for waveform distortion, such as polarization mode dispersion (PMD), in a received signal (input data). The adaptive equalization method of the embodiment uses hybrid processing that combines time domain processing and frequency domain processing, and by optimizing each of the time domain processing and frequency domain processing, it compensates for polarization mode dispersion when the transmission speed is increased with a simple configuration.

FIG. 1 is a diagram showing an adaptive equalization processing unit of a receiver according to an embodiment. The adaptive equalization processing unit 100 includes a frequency domain compensation unit 101, a filter coefficient update unit 102, a frequency domain coefficient conversion unit 103, a control unit 110, and a data selection unit 111.

The adaptive equalization processing unit 100 compensates for waveform distortion caused by changes in the polarization state that adaptively change due to fluctuations in the optical fiber transmission line, temperature, etc. The adaptive equalization processing unit 100 of the embodiment intermittently branches and outputs some symbols (blocks) of input data to the filter coefficient update unit 102 for filter coefficient calculation. The filter coefficient update unit 102 intermittently performs update calculation of the filter coefficient (corresponding to the adaptive equalization processing coefficient) by a time domain filter (time domain processing) in time. The filter coefficient update unit 102 includes a filter coefficient control unit 124 and an FIR filter circuit 125. The FIR filter circuit 125 has a 2-input x 2-output 2 x 2 FIR filter corresponding to the input data of orthogonal polarization (H, V). The connection configuration of the 2 x 2 FIR filter can be configured using an existing FIR filter described later in FIG. 2, etc.

The frequency domain coefficient conversion unit 103 converts the time domain filter coefficients calculated by the filter coefficient update unit 102 into frequency domain filter coefficients. The frequency domain coefficient conversion unit 103 can perform coefficient conversion using a general-purpose FFT, DFT (discrete Fourier transform), etc.

The frequency domain compensation unit 101 includes an FFT 121, a frequency domain filter processing unit 122, and an IFFT 123. The FFT 121 converts a predetermined block of input data into the frequency domain by FFT processing all at once, and outputs the converted data to the frequency domain filter processing unit 122. The frequency domain filter processing unit 122 performs filtering in the frequency domain (waveform distortion compensation for polarization mode dispersion) using the frequency domain filter coefficients output by the frequency domain coefficient conversion unit 103. The IFFT 123 performs IFFT on the signal after filtering, and outputs the data of the orthogonal polarization (H', V') after conversion to time domain data.

The adaptive equalization processing unit 100 of the embodiment controls the timing of selecting data to be used for the filter coefficients from the input data using the control unit 110. The control unit 110 notifies the data selection unit 111 and the filter coefficient update unit 102 of the selection timing.

The data selection unit 111 selects data to be used in the filter coefficient update unit 102 based on the selection timing notified by the control unit 110, and branches and outputs the selected data to the filter coefficient update unit 102. For symbols n and n+1 at a predetermined interval in the input data, the data selection unit 111 intermittently selects multiple symbols (5 symbols) around symbols n and n+1 in the center as a block of input data, and branches and outputs the blocks to the filter coefficient update unit 102. Note that under the control of the control unit 110, the data selection unit 111 outputs all of the continuously input input data to the frequency domain compensation unit 101 as is.

The filter coefficient update unit 102 performs calculations to update the filter coefficients based on the selection timing notified by the control unit 110 and the input data.

For updating the filter coefficients, for example, a known signal (known pattern) inserted between data signals may be used. In this case, as shown in FIG. 1, the arrival timing of the known signal (known pattern) is detected by a known pattern detection unit 130 provided in the receiver. Based on the arrival timing of the known signal detected by the known pattern detection unit 130, the control unit 110 determines the selection timing of data to be branched and output to the filter coefficient update unit 102 from the input data, and outputs the data to the data selection unit 111.

For example, one method of detecting a known signal (pattern) is to insert a special known symbol sequence into a data signal in order to detect the arrival timing of the known pattern. Then, by performing a correlation calculation between the known symbol sequence and the received data, the timing with the highest correlation value is searched for, and the position of the known pattern in the transmitted data is searched for (see, for example, Patent No. 6123584). Known signals (patterns) can be detected by a variety of existing methods, and are not limited to this one method.

In addition, the number of taps of each FIR filter of the FIR filter circuit 125 of the embodiment does not have to be the same as that of the frequency domain filter processing unit 122. The circuit size of the FIR filter changes linearly depending on the number of taps, but the number of taps can be the minimum number that matches the compensation requirement value.

According to the adaptive equalization processing unit 100 configured as above, polarization mode dispersion compensation is performed by hybrid processing of time domain processing and frequency domain processing. The filter coefficient update unit 102 corresponds to the existing time domain processing, and calculates the filter coefficients using data that is selected and output only from the symbol data required for updating the filter coefficients, so that the number of filter operations, i.e., FIR filter circuits 125, can be reduced.

For example, in the example of FIG. 1, only two FIR filter circuits 125 are required: an FIR filter circuit 125 for symbol n and a filter circuit 125 for symbol n+1, which reduces the circuit size. This reduction in the number of FIR filters will be described later. This allows the adaptive equalization processing unit 100 to suppress the increase in circuit size and power consumption that occurs when adaptive equalization processing is performed using only time domain processing.

Furthermore, the frequency domain compensation unit 101 shown in FIG. 1 is similar to existing frequency domain processing, but performs distortion compensation on input data based on the filter coefficients calculated by the filter coefficient update unit 102. This makes it possible to prevent filter delays that occur when adaptive equalization processing is performed using only frequency domain processing. As a result, the adaptive equalization processing unit 100 of the embodiment can compensate for polarization mode dispersion when the transmission speed is increased with a simple configuration.

(Regarding existing adaptive equalization processing)
Here, existing adaptive equalization processing will be explained. Here, existing time domain processing and frequency domain processing will be explained separately.

FIG. 2 is a diagram showing a configuration example of polarization mode dispersion compensation by existing time domain processing. FIG. 2 shows an example of a polarization mode dispersion compensation circuit in adaptive equalization processing of MIMO (Multi Input Multi Output). The polarization mode dispersion compensation circuit 200 shown in FIG. 2(a) has an FIR filter circuit 201 and a filter coefficient calculation unit 202. In the FIR filter circuit 201, orthogonal X-polarized data r _x,n and Y-polarized data r _y,n are input to four FIR filters 201. The X-polarized data r _x,n is input to the FIR filter 201a and the cross-connected FIR filter 201c. The Y-polarized data r _y,n is input to the cross-connected FIR filter 201b and the FIR filter 201d.

The filter coefficient calculation unit 202 calculates the filter coefficient Wxx for the FIR filter circuit 201 based on the input data using the following formula (1).

The outputs S _x,n , S _y,n of the FIR filter circuit 201 shown in equation (1) are represented by the determinant of the four transfer functions W _xx (W _hh , W _hv , W _vh , W _vv ) of the FIR filters 201a to 201d and the inputs r _x,n , r _y,n , and the filter coefficients indicate the inverse characteristics of the transmission path distortion.

The transfer functions output by the FIR filters 201a and 201b are output as a compensation output _Sx,n for the crosstalk and delay difference due to the addition of the x and y components. The transfer functions output by the FIR filters 201c and 201d are output as a compensation output Sy _,n for the crosstalk and delay difference due to the addition of the x and y components. The compensation outputs Sx _,n and _Sy,n are output to a filter coefficient calculation unit 202, which follows the time fluctuation of the transmission path distortion.

FIG. 2B shows an example of the internal configuration of one FIR filter 201, in which a plurality of delay elements T are connected in series to the input with a predetermined number of taps, and the multiplication results of the filter coefficient _Wxx between the input and output of each delay element are added together for the plurality of taps and output.

Figure 3 is a diagram showing the characteristics of a polarization mode dispersion compensation circuit using an existing FIR filter. Figure 3(a) is a characteristic diagram of polarization mode dispersion compensation performance (ps), with the horizontal axis showing the symbol rate and the vertical axis showing the polarization mode dispersion compensation performance. When the number of taps is 17, the polarization mode dispersion compensation performance decreases with an increase in the symbol rate. For example, the polarization mode dispersion compensation performance is 180 ps when the symbol rate is 32 GBaud, but when the symbol rate becomes 64 GBaud, the polarization mode dispersion compensation performance is halved to 90 ps. Furthermore, when the symbol rate becomes 128 GBaud, the polarization mode dispersion compensation performance is halved to 45 ps.

Figure 3(b) is a power characteristic diagram of the polarization mode dispersion compensation circuit 200 when the number of taps is increased, with the horizontal axis showing the number of taps and the vertical axis showing the power. The more the number of taps increases, the greater the power. If the baud rate increases beyond the current symbol rate of 120 GBaud, power consumption will increase in accordance with the increase in the number of taps, and it may become impossible to perform polarization mode dispersion compensation with a realistic power.

FIG. 4 is a diagram showing examples of each process of the existing polarization mode dispersion compensation. FIG. 4(a) shows a polarization mode dispersion compensation circuit 200 by time domain processing using the above-mentioned FIR filter. The filter coefficient calculation unit 202 of the polarization mode dispersion compensation circuit 200 shown in FIG. 4(a) calculates filter coefficients using input data r _x,n , _ry,n, and follows the time fluctuation of the transmission line distortion by the compensation output S _x,n , S _y,n . However, an increase in the baud rate (number of taps) increases the delay amount of the FIR filter circuit 201. In this case, it takes a long time to calculate new filter coefficients in the filter coefficient calculation unit 202, and it is not possible to respond to the fluctuation of the transmission line distortion, resulting in a decrease in polarization fluctuation resistance performance.

Figure 4(b) shows a polarization mode dispersion compensation circuit 400 using frequency domain processing. In order to solve the problems that arise in the polarization mode dispersion compensation circuit 200 using time domain processing shown in Figure 4(a), it is possible to use the polarization mode dispersion compensation circuit 400 using frequency domain processing shown in Figure 4(b).

4B includes an FFT 401, a frequency domain filter processor 402, and an IFFT 403, and converts a plurality of symbols of time domain input data r _x,n , _ry,n into the frequency domain all at once by the FFT 301. Then, the filter coefficients calculated by the filter coefficient calculation unit 404 are converted into frequency domain filter coefficients by the frequency domain coefficient converter 405, and output to the frequency domain filter processor 402. The frequency domain filter processor 402 performs filtering based on the frequency domain filter coefficients converted by the frequency domain coefficient converter 405, and obtains an output via the IFFT 403.

Figure 5 is a graph showing the polarization mode dispersion compensation performance in existing time domain processing and frequency domain processing. The horizontal axis is the polarization mode dispersion compensation performance, which corresponds to the filter memory length (number of taps, FFT/IFFT size). The vertical axis is the power and circuit size.

The dotted line in FIG. 5 is the characteristic line of compensation by time domain processing shown in FIG. 4(a), and is located in region A when the symbol rate is small. However, in the characteristic line T by time domain processing, the power (circuit size) increases as the filter memory length increases. In contrast, the solid line in FIG. 5 shows the characteristic line F of compensation by frequency domain processing shown in FIG. 4(b). In frequency domain processing, the increase in power (circuit size) with an increase in filter memory length can be suppressed compared to time domain processing, and even if the baud rate increases in the future, the increase in power (circuit size) is suppressed even if the filter memory length is increased. The characteristic line F of frequency domain processing shown in FIG. 5 shows that the power (circuit size) is reversed in the filter memory length part exceeding region A compared to the characteristic line T of time domain processing, and it is possible that polarization mode dispersion compensation by frequency domain processing may be advantageous.

Figure 6 is an explanatory diagram of the problems with existing time domain processing and frequency domain processing. Both the time domain processing and frequency domain processing described above have the following problems.

FIG. 6A is a diagram showing an example of filter processing in the polarization mode dispersion compensation circuit 200 by time domain processing. For one symbol n of input data r _x,n , one FIR filter circuit 201 described above is used. In optical communications with a data rate of several tens to several hundreds of GHz, the data rate is faster than the processing rate of the circuit, so parallel signal processing by the FIR filter circuit 201 for each input symbol is required. In the case of FIG. 6A, if the data rate that can be processed by one FIR filter circuit 201 is 1 GBaud, parallel signal processing by 128 FIR filter circuits 201 is required for 128 GBaud. In the polarization mode dispersion compensation circuit 200 by time domain processing, the circuit size and power consumption of the FIR filter circuit 201 increase as the baud rate increases due to an increase in not only the number of taps but also the number of parallel signal processes.

Figure 6(b) is a diagram showing an example of filter processing by the polarization mode dispersion compensation circuit 400 using frequency domain processing. As described above, the polarization mode dispersion compensation circuit 400 performs frequency domain processing using input data of multiple symbols, for example, the entire 128 symbols as input. The FFT 401 converts a block of multiple symbols of input data into the frequency domain all at once, and only one filter process is required for every 128 symbols. This filter process is advantageous when the filter memory length (corresponding to the tap length of the FIR filter) is large because it processes all at once. On the other hand, the polarization mode dispersion compensation circuit 400 makes filter processing more efficient by processing multiple symbols together, which results in large filter delay.

Here, an example of filter coefficient control will be explained. One example of filter coefficient control is CMA (Constant Modulus Algorithm). In this CMA, the tap coefficients are controlled by the stochastic gradient descent method so that the evaluation function approaches 0. The stochastic gradient descent algorithm is expressed by the following equation (2).

where w _n : adaptive equalizer tap coefficient vector (w _hh , w _vh , w _hv , w _vv ) at time n, r _n : input signal vector of the adaptive equalizer, s _n : output signal electric field of the adaptive equalizer, μ : step size (controls the strength of feedback), g : target signal amplitude after equalization (constant).

As stated above, as the circuit delay of the filter increases, the update period slows down. This reduces the polarization fluctuation tolerance performance.

As described above, both time domain processing and frequency domain processing have their own issues. Attempting to perform polarization mode dispersion compensation using only time domain processing increases the circuit size and power consumption. Furthermore, attempting to perform compensation using only frequency domain processing results in filter delay.

In contrast, the adaptive equalization processing unit 100 of the embodiment has a hybrid configuration of time domain processing and frequency domain processing, and passes only a portion of the input data to the filter coefficient update unit 102. The filter coefficient update unit 102 intermittently performs update calculations of the filter coefficients (corresponding to adaptive equalization processing coefficients) using a time domain filter (time domain processing). Although the FIR filter circuit 125 is a time domain filter, delays can be reduced by reducing the number of taps, and the filter coefficient update unit 102 can quickly calculate the filter coefficients. Furthermore, the filter coefficient update unit 102 only needs to perform coefficient updates using only a portion of the input data, and performs coefficient updates once every 128 times, for example.

And according to the embodiment, the parallel connection of 128 FIR filter circuits 201 required in the polarization mode dispersion compensation circuit 200 using time domain processing shown in FIG. 6(a) is no longer necessary, and the number of filter circuits can be reduced.

Figure 7 is a diagram showing an example of the configuration of a filter coefficient update unit of an adaptive equalization processing unit according to an embodiment. The FIR filter circuit 125 of the filter coefficient update unit 102 according to an embodiment can be configured using four connected 2x2 filters 125a to 125d, similar to the FIR filter circuit 201 described in the existing time-domain processing polarization mode dispersion compensation circuit 200 (see Figure 6(a)).

Here, the FIR filter circuit 125 of the filter coefficient update unit 102 in the embodiment only needs to process the symbol data required for the coefficient update calculation, so the filter calculation can be omitted. In the example of FIG. 7, it is sufficient to use two FIR filter circuits 125 for two symbol data (n, n+1). As a result, for example, compared to the 128 FIR filter circuits 201 used in the existing time domain processing shown in FIG. 6(a), in the embodiment, only two filter circuits 125 are required.

Furthermore, a known pattern may be inserted between the data symbols described in FIG. 1, and the coefficients may be updated using the known pattern. For example, a known pattern (corresponding to one symbol of n, n+1) is inserted into consecutive data symbols at a predetermined time interval T0, such as every 32 symbols, every 62 symbols, or every 128 symbols. In this case, the FIR filter circuit 125 of the filter coefficient update unit 102 only needs to filter the known pattern. Furthermore, if a slow interval for updating the coefficients is acceptable, that is, if the symbol interval (time interval) of symbols n and n+1 is wide, time-division processing can be performed using only one FIR filter circuit 125, and the FIR filter circuit 125 can be further simplified.

In this way, the FIR filter circuit 125 of the filter coefficient update unit 102 according to the embodiment can reduce the operation rate even if the number of filter taps is large. In the embodiment, by using the FIR filter circuit 125 that performs time domain processing as one function of the filter coefficient calculation in the filter coefficient update unit 102, it is possible to improve power efficiency while balancing the tracking ability and compensation performance for polarization mode dispersion distortion.

Incidentally, the adaptive equalization processing unit 100 shown in FIG. 1, including the filter coefficient update unit 102 shown in FIG. 7, requires high-speed signal processing, so currently it is configured using dedicated ASIC hardware. However, in the future, the adaptive equalization processing unit 100 can also be configured using an FPGA, DSP, or CPU that supports high-speed processing. ASIC stands for Application Specific Integrated Circuit, FPGA stands for Field Programmable Gate Array, and DSP stands for Digital Signal Processor.

(Example of Hardware Configuration of Control Unit of Adaptive Equalization Processing Unit)
Fig. 8 is a diagram showing an example of the hardware configuration of a control unit of an adaptive equalization processing unit. For example, the control unit 110 shown in Fig. 1 is a configuration related to input data selection timing control, and can be configured by, for example, the hardware shown in Fig. 8.

For example, the control unit 110 has a processor 801 such as a CPU (Central Processing Unit), a memory 802, a network IF 803, a recording medium IF 804, and a recording medium 805. In addition, each component is connected to each other via a bus 800.

Here, the processor 801 is a control unit that controls the entire control unit 110. The processor 801 may have multiple cores. The memory 802 includes, for example, a ROM (Read Only Memory), a RAM (Random Access Memory), and a flash ROM. Specifically, for example, the flash ROM stores a control program, the ROM stores an application program, and the RAM is used as a work area for the processor 801. The programs stored in the memory 802 are loaded into the processor 801, causing the processor 801 to execute the coded processing.

The network IF 803 serves as the interface between the network NW and the inside of the device, and controls the input and output of information.

The recording medium IF 804 controls the reading/writing of data from/to the recording medium 805 under the control of the processor 801. The recording medium 805 stores the data written under the control of the recording medium IF 804.

In addition to the components described above, the control unit 110 may also be capable of connecting, for example, an input device, a display, etc. via an IF.

The processor 801 shown in FIG. 8 executes a program to, for example, determine the timing of input data selection by the control unit 110 shown in FIG. 1 and control the data selection unit 111.

(Processing example of adaptive equalization processing unit)
9 is a flowchart showing an example of processing by the adaptive equalization processing unit according to the embodiment. The processing examples of the functions of the adaptive equalization processing unit 100 shown in FIG. 1 are shown. First, the control unit 110 determines the timing of data selection (step S901). For example, the control unit 110 waits for the arrival timing of a known pattern (n, n+1) detected by the known pattern detection unit 130 (step S901: No loop). Then, the control unit 110 determines the arrival timing of the known pattern (n, n+1) as the selection timing (step S901: Yes) and controls the processing from step S902 onward for each selection timing.

The input data (block) of multiple symbols including a known pattern that is determined to be the selection timing is branched and output by the data selection unit 111 to the filter coefficient update unit 102. The data selection unit 111 extracts data required for adaptive equalization coefficient calculation from the intermittent input data (block) selected at the selection timing among the continuous input data (step S902).

Next, using the data extracted by the data selection unit 111, the filter coefficient update unit 102, which is configured as a time domain filter, calculates adaptive equalization coefficients (filter coefficients) independently of the frequency domain compensation unit 101 (step S903).

Next, the adaptive equalization coefficients calculated by the filter coefficient update unit 102 are converted into frequency domain filter coefficients by the frequency domain coefficient conversion unit 103 (step S904). The frequency domain coefficient conversion unit 103 outputs the frequency domain filter coefficients to the frequency domain compensation unit 101.

Then, the frequency domain compensation unit 101 compensates for the waveform distortion (polarization mode dispersion) of the input data based on the frequency domain filter coefficients (step S905). At this time, the frequency domain compensation unit 101 updates the frequency domain filter coefficients after compensation.

As a result, adaptive equalization processing for one symbol n is performed, and the adaptive equalization processing unit 100 returns to the processing of step S901. As a result, the adaptive equalization processing unit 100 performs polarization mode dispersion compensation processing for the continuously input input data based on the filter coefficients updated at each selection timing.

(Example of optical transmission/reception system and receiver configuration)
10 is a diagram showing a configuration example of an optical transmission and reception system including an adaptive equalization processing unit according to an embodiment. Fig. 10(a) shows a configuration example of an optical transmission and reception system 1000, in which a transmitter (Tx) 1001 transmits an optical signal by an optical phase modulation method such as QPSK. QPSK is an abbreviation for Quadruple Phase Shift Keying. The optical signal transmitted from the transmitter 1001 is transmitted via a set of optical amplifiers 1002 and an optical fiber transmission line 1003 arranged at a predetermined distance or span, and is received by a receiver (Rx) 1004.

Figure 10 (b) shows an example of the internal configuration of the receiver 1004. The receiver 1004 has various units including the adaptive equalization processing unit 100 of the embodiment. The received signal and the local light of the local light source 1005 are input to the coherent reception front end 1011 provided in front of the receiver 1004, and the optical signal received by the optical coherent method is converted into an electrical signal.

The ADC 1012 converts the received signal into a digital signal and outputs it to the received signal compensation unit 1013. The received signal compensation unit 1013 includes a dispersion compensation unit 1014, the adaptive equalization processing unit 100 of the above-mentioned embodiment, a carrier frequency/phase compensation unit 1015, and an error correction unit 1016 as functions for compensating for the received signal.

The dispersion compensation unit 1014 compensates for chromatic dispersion of the received signal. The adaptive equalization processing unit 100 compensates for polarization mode dispersion of the received signal (input data) by the adaptive equalization processing described above. The carrier frequency/phase compensation unit 1015 compensates for frequency and phase deviations that prevent constellation rotation due to optical phase deviation and optical carrier frequency deviation. The error correction unit 1016 performs error correction (FEC: Forward Error Correction) on the received signal (input data) and outputs a demodulated signal.

Other Embodiments
Fig. 11 is a diagram showing an adaptive equalization processing unit of a receiver according to another embodiment. The adaptive equalization processing unit 1100 shown in Fig. 11 is a configuration example corresponding to a fractional rate input signal. In Fig. 11, the same components as those in the adaptive equalization processing unit 100 (see Fig. 1) described above are given the same reference numerals. Also, the control unit 110 and the data selection unit 111 are omitted from the illustration.

When the input signal is a fractional rate, i.e., when sampling is a non-integer multiple of the symbol rate, the adaptive equalization processing unit 1100 simultaneously performs resampling to integer multiple sampling. This adaptive equalization process requires not only waveform distortion of the transmission path, but also resampling, i.e., processing to convert a signal input at a frequency that is a non-integer multiple higher than the symbol rate into samples that are an integer multiple of the symbol rate.

In the configuration example of FIG. 11, the FFT 121 at the input of the frequency domain compensation unit 101 and the IFFT 123 at the output are set to the same block length, and a phase shift resampling unit 1101 is provided after the IFFT 123 to perform resampling. In addition, in the case of fractional input, the phase shift resampling unit 1101 performs resampling with a phase shift rather than simple decimation.

This allows the output of the frequency domain compensation unit 101 to be normally output at an integer multiple of the symbol rate (e.g., 1x, 2x, etc.) even if the input data is sampled at a non-integer multiple of the symbol rate, e.g., 1.5x.

Figure 12 is a diagram showing an adaptive equalization processing unit of a receiver according to another embodiment. The adaptive equalization processing unit 1200 shown in Figure 12 is another example of a configuration compatible with fractional rate input, and performs resampling processing in the frequency domain.

In this adaptive equalization processing unit 1200, a resampling unit 1201, DFT 1202, and phase shift unit 1203 are placed after the frequency domain compensation unit 101. Here, since the output of the frequency domain compensation unit 101 differs between the input symbol sequence and the output symbol sequence, the resampling unit 1201 performs FFT processing, for example, power-of-two sample block processing, on the input data, and performs DFT processing on the output data with the DFT 1202.

In this adaptive equalization processing unit 1200, power efficiency decreases due to DFT processing, but resampling can be performed in the frequency domain, making it possible to achieve high-precision resampling.

The receiver of the embodiment described above includes an adaptive equalization processing unit that adaptively compensates for waveform distortion of received data. The adaptive equalization processing unit includes a frequency domain compensation unit that compensates for waveform distortion of received data using a frequency domain filter, a time domain processing filter, and a filter coefficient update unit that calculates filter coefficients based on received data input to the frequency domain compensation unit. It also includes a frequency domain coefficient conversion unit that converts the time domain filter coefficients calculated by the filter coefficient update unit into the frequency domain and outputs them to the frequency domain filter. The filter coefficient update unit calculates filter coefficients using only a portion of the received data input to the frequency domain filter. As a result, the filter coefficient update unit can reduce the operation rate even if the number of taps is large by intermittently calculating filter coefficients using only a portion of the received data. Therefore, even if a time domain filter such as an FIR filter with a large number of taps is used only in the filter coefficient update unit, it is possible to improve tracking performance, waveform distortion compensation performance, and power efficiency.

The receiver of the embodiment also includes a data selection unit that outputs all of the received data to the frequency domain compensation unit, and selects and outputs a portion of the received data to the filter coefficient update unit. It also includes a control unit that instructs the data selection unit to selectively output each time a predetermined portion of the received data is received. This allows both of the hybrid function configurations, the filter coefficient update unit that calculates filter coefficients from the received data by time domain processing, and the frequency domain compensation unit that compensates for waveform distortion of the received data by frequency domain processing, to operate optimally.

In addition, part of the received data in the embodiment can be a known signal inserted into the data signal at a predetermined time interval. In this way, by inserting a general-purpose known signal such as a pilot signal into the data signal at a predetermined time interval, the filter coefficient update unit can intermittently calculate the filter coefficients at this predetermined time interval.

The filter coefficient update unit of the receiver of the embodiment includes multiple FIR filters, and can calculate filter coefficients based on tap coefficients calculated based on a portion of the received data. For example, the time domain processing filter can be configured to include four FIR filters that are directly and cross-connected with two inputs and two outputs. In this way, the filter coefficient update unit can be easily configured using general-purpose FIR filters used in existing time domain processing.

The frequency domain compensation unit of the receiver of the embodiment can be configured to include an FFT, a frequency domain filter, and an IFFT. In this way, the frequency domain compensation unit can be easily configured using general-purpose FFT, IFFT, etc. used in existing frequency domain processing.

The receiver of the embodiment may also be configured to handle the case where the received data is a fractional rate sampled at a non-integer multiple of the symbol rate. In this case, the frequency domain compensation unit may further include a phase shift/resampling unit that performs phase shifting and resampling on the output data of the received data IFFT. The frequency domain compensation unit may also include a resampling unit that performs resampling processing on the output data of the received data IFFT in the frequency domain, a DFT for discrete Fourier transform processing, and a phase shift unit. This makes it possible to perform sampling output at an integer multiple of the symbol rate of the received data even in the case of sampling at a non-integer multiple of the symbol rate.

The following additional notes are provided with respect to the above-described embodiment.

(Supplementary Note 1) In a receiver including an adaptive equalization processing unit that adaptively compensates for waveform distortion of received data,
The adaptive equalization processing unit includes:
a frequency domain compensation unit that compensates for waveform distortion of the received data by a frequency domain filter;
a filter coefficient update unit including a time domain processing filter, the filter coefficient update unit calculating a filter coefficient based on the received data input to the frequency domain compensation unit;
a frequency domain coefficient conversion unit that converts the filter coefficients in the time domain calculated by the filter coefficient update unit into a frequency domain and outputs the frequency domain filter to the frequency domain filter;
the filter coefficient update unit calculates the filter coefficients using only a portion of the received data input to the frequency domain filter.
A receiver comprising:

(Supplementary Note 2) A data selection unit that outputs all of the received data to the frequency domain compensation unit and selects and outputs a portion of the received data to the filter coefficient update unit;
a control unit that instructs the data selection unit to select and output the received data at each timing when a predetermined portion of the received data is received;
2. The receiver of claim 1, comprising:

(Appendix 3) The receiver described in Appendix 2, characterized in that part of the received data is a known signal inserted into the data signal at a predetermined time interval.

(Additional Note 4) The filter coefficient update unit includes a plurality of FIR filters, and calculates the filter coefficients based on tap coefficients calculated based on a portion of the received data.
2. The receiver of claim 1 .

(Appendix 5) The receiver described in Appendix 4, characterized in that the time domain processing filter includes four FIR filters that are directly and cross-connected with two inputs and two outputs.

(Appendix 6) The receiver described in Appendix 1, characterized in that the frequency domain compensation unit includes an FFT, the frequency domain filter, and an IFFT.

(Appendix 7) The receiver described in appendix 6 is characterized in that, in response to the case where the received data is sampled at a non-integer multiple of the symbol rate, the frequency domain compensation unit further has a phase shift and resampling unit that performs phase shifting and resampling on the received data and the output data of the IFFT, and outputs a sampled output at an integer multiple of the symbol rate of the received data.

(Appendix 8) The receiver described in appendix 6 is characterized in that, in response to the case where the received data is sampled at a non-integer multiple of the symbol rate, the frequency domain compensation unit further has a resampling unit that performs resampling processing on the received data and the output data of the IFFT in the frequency domain, a DFT for discrete Fourier transform processing, and a phase shift unit, and performs sampling output at an integer multiple of the symbol rate of the received data.

(Additional Note 9) An adaptive equalization processing method for a receiver that adaptively compensates for waveform distortion of received data, comprising:
Calculating a time domain filter coefficient by time domain processing based on the received data;
Transforming the calculated time domain filter coefficients into a frequency domain;
Compensating for waveform distortion of the received data by frequency domain processing based on the converted filter coefficients in the frequency domain;
The filter coefficients are calculated using only a portion of the received data.
2. An adaptive equalization method comprising:

100, 1100, 1200 Adaptive equalization processing unit 101 Frequency domain compensation unit 102 Filter coefficient update unit 103 Frequency domain coefficient conversion unit 110 Control unit 111 Data selection unit 121 FFT
122 Frequency domain filter processing unit 123 IFFT
124 Filter coefficient control unit 125 FIR filter circuit 130 Known pattern detection unit 800 Bus 801 Processor 802 Memory 805 Recording medium 1000 Optical transmission/reception system 1001 Transmitter 1002 Optical amplifier 1003 Transmission path 1004 Receiver 1005 Local oscillator 1011 Coherent reception front end 1013 Received signal compensation unit 1014 Dispersion compensation unit 1015 Carrier frequency/phase compensation unit 1016 Error correction unit 1101 Phase shift/sampling unit 1201 Resampling unit 1202 DFT
1203 Phase shift unit

Claims (8)

A receiver including an adaptive equalization processor that adaptively compensates for waveform distortion of received data,
The adaptive equalization processing unit includes:
a frequency domain compensation unit that compensates for waveform distortion of the received data by a frequency domain filter;
a filter coefficient update unit including a time domain processing filter, the filter coefficient update unit calculating a filter coefficient based on the received data input to the frequency domain compensation unit;
a frequency domain coefficient conversion unit that converts the filter coefficients in the time domain calculated by the filter coefficient update unit into a frequency domain and outputs the frequency domain filter to the frequency domain filter;
the filter coefficient update unit calculates the filter coefficients using only a portion of the received data input to the frequency domain filter.
A receiver comprising: a data selection unit that outputs all of the received data to the frequency domain compensation unit and selects and outputs a portion of the received data to the filter coefficient update unit;
a control unit that instructs the data selection unit to select and output the received data at each timing when a predetermined portion of the received data is received;
2. The receiver of claim 1, further comprising: The receiver according to claim 2, characterized in that part of the received data is a known signal inserted into the data signal at a predetermined time interval. the filter coefficient update unit includes a plurality of FIR filters, and calculates the filter coefficients based on tap coefficients calculated based on a portion of the received data.
2. The receiver of claim 1 . The receiver according to claim 1, characterized in that the frequency domain compensation unit includes an FFT, the frequency domain filter, and an IFFT. The receiver according to claim 5, characterized in that, in response to the case where the received data is sampled at a non-integer multiple of the symbol rate, the frequency domain compensation unit further has a phase shift/resampling unit that performs phase shifting and resampling on the received data and the output data of the IFFT, and outputs a sampled output at an integer multiple of the symbol rate of the received data. The receiver according to claim 5, characterized in that, in response to the case where the received data is sampled at a non-integer multiple of the symbol rate, the frequency domain compensation unit further includes a resampling unit that performs resampling processing on the received data and the output data of the IFFT in the frequency domain, a DFT for discrete Fourier transform processing, and a phase shift unit, and performs sampling output at an integer multiple of the symbol rate of the received data. An adaptive equalization processing method for a receiver that adaptively compensates for waveform distortion of received data, comprising:
Calculating a time domain filter coefficient by time domain processing based on the received data;
Transforming the calculated time domain filter coefficients into a frequency domain;
Compensating for waveform distortion of the received data by frequency domain processing based on the converted filter coefficients in the frequency domain;
The filter coefficients are calculated using only a portion of the received data.
2. An adaptive equalization method comprising:

Images