RU2816843C2 - Signal processing method and device and coherent receiver - Google Patents

Abstract

FIELD: communication equipment.

SUBSTANCE: invention relates to a signal processing method and device and a coherent receiver. Method comprises obtaining P real-valued signals; performing at least number-theoretic transformation (NTT) processing for the P real-valued signals to obtain P first real-valued signals in the transform domain; performing at least reconstruction of clock frequency for P first real-valued signals in conversion domain to obtain P second real-valued signals in conversion domain; performing at least polarization compensation and inverse number-theoretic transformation (INTT) processing for P second real-valued signals in the transformation domain to obtain m complex-valued signals X in the time domain and m complex-valued signals Y in the time domain, wherein performing at least polarization compensation and inverse number-theoretic transformation (INTT) processing for the P second real-valued signals in the transformation domain includes sequential execution of polarization compensation processing and INTT processing for P second real-valued signals in conversion domain or sequentially performing INTT processing and polarization compensation processing for P second real-valued signals in transformation domain; and performing phase recovery and decoding of m complex-valued signals in time domain and m complex-valued signals Y in time domain to obtain bit signals, wherein m and P are natural numbers, wherein polarization compensation includes polarization demultiplexing and distortion equalization.

EFFECT: improved efficiency of signal processing due to use of NTT and INTT processing during signal processing procedure.

26 cl, 20 dwg

Description

Field of technology

[0001] The present invention relates to the field of optical communications and, in particular, to a signal processing method and apparatus and a coherent receiver.

State of the art

[0002] The performance of a high-throughput optical communication system may be degraded due to optical fiber attenuation, optical fiber chromatic dispersion, polarization mode chromatic dispersion, laser phase noise, optical fiber nonlinearity, and the like. The presently proposed coherent polarization multiplexing receiver can effectively compensate for the above degradation factors in the electrical domain during the electrical signal processing procedure.

[0003] During the electrical signal processing procedure of the existing coherent polarization multiplexing receiver, time domain signal convolution and inverse channel response function based on Fourier transform are performed to realize the above effective compensation in the electrical domain. Therefore, it is necessary to introduce a fast Fourier transform (FFT) module and an inverse fast Fourier transform (IFFT) module into the structure of a coherent polarization multiplexing receiver. However, the Fourier transform has high complexity and low accuracy. As a result, the electrical signal processing procedure is very complex, and the accuracy of the reconstructed signal is low. Therefore, the question of how to improve the performance of a coherent polarization multiplexing receiver in electrical signal processing becomes an urgent problem that needs to be solved.

The essence of the invention

[0004] The present invention provides a signal processing method and apparatus and a coherent receiver for improving signal processing performance.

[0005] According to a first aspect, a signal processing method is provided. The signal processing method may be performed by a coherent receiver or may be performed by a chip or circuit located in the coherent receiver. This is not limited to in the present invention.

[0006] The signal processing method includes:

first obtaining P real-valued signals, where the P real-valued signals include P real-valued signals obtained through analog-to-digital conversion, or other P real-valued signals that require digital signal processing, and this is not limited to the present invention;

then performing at least number theoretical transform (NTT) processing on the P real-valued signals to obtain P first real-valued signals in the transform domain;

then performing at least clock recovery on the P first real-valued signals in the transform region to obtain P second real-valued signals in the transform region; And

then performing at least polarization compensation (compensation for polarization effects) and inverse number theoretical transform (INTT) processing on the P second real-valued signals in the transform domain to obtain m complex-valued signals X in the time domain and m complex-valued signals Y in the time domain, where m and P are natural numbers.

[0007] In addition, if INTT and combining processing are performed first, and then polarization compensation processing is performed, in the signal processing method provided in the present invention, phase recovery and decoding can be performed on m complex valued time domain signals X and m complex valued signals Y in the time domain, obtained through polarization compensation, to obtain bit signals.

[0008] It should be understood that if INTT and combining processing are performed first, and then polarization compensation processing is performed, the polarization compensation processing is performed based on the time domain signal obtained by the INTT processing. In particular, the ability to withstand loop delay can be improved by performing polarization compensation processing in the time domain.

[0009] Alternatively, if polarization compensation processing is performed first, and then INTT and combining processing are performed, in the signal processing method provided in the present invention, phase recovery and decoding can be performed on m complex-valued time domain signals X in the first polarization direction and m complex-valued time domain signals Y in the second polarization direction, which are obtained by combining to obtain bit signals to realize signal reconstruction.

[0010] It should be understood that if polarization compensation processing is performed first, and then INTT and combining processing are performed, polarization compensation processing is performed based on the signal in the transform domain. In particular, distortion compensation by polarization equalization is performed in the transform region, and multiplication is used to replace convolution, so that power consumption can be reduced.

[0011] In the signal processing method presented in the present invention, signals in two polarization directions are obtained through analog-to-digital conversion, and signals in two polarization directions are input to a digital signal processing (DSP) device of a receiver (Rx). During the Rx DSP processing procedure, number theoretic transform (NTT) processing and inverse number theoretic transform (INTT) processing are used to replace fast Fourier transform (FFT) processing and inverse fast Fourier transform (IFFT) processing to avoid high complexity and low precision caused by FFT and IFFT processing, thereby improving signal processing performance.

[0012] With respect to the first aspect, in some implementations of the first aspect, performing at least number theoretic transform (NTT) processing on the P real-valued signals to obtain P first real-valued signals in the transform domain includes: performing NTT processing on the P real-valued signals for receiving P first real-valued signals in the conversion domain; or

performing digital back propagation (DBP) processing on the P real-valued signals to obtain P tenths of real-valued signals in the time domain, and separately performing NTT processing on P tenths of real-valued signals in the time domain to obtain P first real-valued signals in the transform domain.

[0013] If P real-valued signals are subject to DBP processing, the DBP module is used to implement compensation for nonlinear effects to increase the signal transmission distance.

[0014] In the signal processing method presented in the present invention, P first real-valued signals in the conversion domain can be obtained by inputting P real-valued signals into an NTT module and performing NTT processing; or P first real-valued signals in the transform domain can be obtained by first inputting P real-valued signals into the DBP module and performing DBP processing, and then performing NTT processing. Various methods are provided for obtaining P first real-valued signals in the conversion domain, so that the structural design of the Rx DSP device is more flexible.

[0015] With respect to the first aspect, in some implementations of the first aspect, when P real-valued signals are not subject to DBP processing, chromatic dispersion compensation must be performed when P second real-valued signals in the transform domain are derived from P first real-valued signals in the transform domain. More specifically, performing at least clock recovery on P first real-valued signals in the transform region to obtain P second real-valued signals in the transform region includes: performing chromatic dispersion compensation on the P first real-valued signals in the transform region to obtain P third real-valued signals in transformation areas; and performing clock recovery on the P third real-valued signals in the transform region to obtain P second real-valued signals in the transform region.

[0016] Alternatively, when P real-valued signals are subjected to DBP processing, since the DBP module has a chromatic dispersion compensation function, no additional chromatic dispersion compensation is required when the P second real-valued signals in the transform domain are obtained based on the P first real-valued signals in the transform domain. More specifically, performing at least clock recovery on the P first real-valued signals in the transform region to obtain P second real-valued signals in the transform region includes: performing chromatic dispersion compensation on the P first real-valued signals in the transform region to obtain the P second real-valued signals in transformation areas.

[0017] With respect to the first aspect, in some implementations of the first aspect, the P real-valued signals include an in-phase real-valued signal I ^m _x and a quadrature real-valued signal Q ^m _x in a first polarization direction and an in-phase real signal I ^m _y and a quadrature real-valued signal Q ^m _y in the second polarization direction. In the above case where additional chromatic dispersion compensation is required, the chromatic dispersion compensation process includes: performing NTT processing on the chromatic dispersion impulse response I _h (t) corresponding to the in-phase real-valued signal, and on the chromatic dispersion impulse response Q _h (t), corresponding to the quadrature real-valued signal, to obtain a chromatic dispersion equalization function I _h (w) in the transformation region corresponding to the in-phase real-valued signal, and a chromatic dispersion equalization function Q _h (w) in the transformation region corresponding to the quadrature real-valued signal.

[0018] After obtaining I _h (w) and Q _h (w), 2×m third real-valued signals in the transformation region that are in P third real-valued signals in the transformation region and which are in the first polarization direction can be determined based on I _h (w), Q _h (w) and an in-phase real-valued signal and a quadrature real-valued signal that are in the P first real-valued signals in the transform domain and that are in the first polarization direction, and 2×m third real-valued signals in the transform domain that are in the P third real-valued signals in the transform domain and which are in the second polarization direction can be determined based on I _h (w), Q _h (w) and the in-phase real-valued signal and the quadrature real-valued signal that are in the P first real-valued signals in conversion areas and which are in the second polarization direction to complete the chromatic dispersion compensation.

[0019] With respect to the first aspect, in some implementations of the first aspect, the third real-valued transform domain signals obtained by chromatic dispersion compensation and the first real-valued transform domain signals satisfy the following requirements:

I ^' _x (w) = I _x (w)⋅I _h (w)-Q _x (w)⋅Q _h (w);

Q ^' _x (w) = Q _x (w)⋅I _h (w)+I _x (w)⋅Q _h (w);

I ^' _y (w) = I _y (w)⋅I _h (w)-Q _y (w)⋅Q _h (w); And

Q ^' _y (w) = Q _y (w)⋅I _h (w)+I _y (w)⋅Q _h (w), where

I _x (w) represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the first polarization direction, Q _x (w) represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and that is in first polarization direction, I _y (w) represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the second polarization direction, Q _y (w) represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and which is in the second polarization direction, I ^' _x (w) represents an in-phase real-valued signal that is in the third real-valued signals in the transform region and which is in the first polarization direction, Q ^' _x (w) represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and which is in the first polarization direction, I ^' _y (w) represents the in-phase real-valued signal that is in the third real-valued signals in the transform domain and which is in the second polarization direction, and Q ^' _y (w) represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and that is in the second polarization direction.

[0020] Specifically, when m is 1, determining 2×m real-valued third signals in the transform domain that are in P real-valued third signals in the transform domain and that are in the first polarization direction based on I _h (w), Q _h (w) and an in-phase real-valued signal and a quadrature real-valued signal that are in the P first real-valued signals in the conversion region and that are in the first polarization direction includes:

Ix_3 = Ix_1⋅I _h (w)-Qx_1⋅Q _h (w); Qx_3 = Qx_1⋅I _h (w)+Ix_1⋅Q _h (w), where

Ix_3 represents the in-phase third real-valued signal in the transform domain in the first polarization direction, Ix_1 represents the in-phase real-valued signal that is in the P first real-valued signals in the transform domain and which is in the first polarization direction, Qx_3 represents the quadrature third real-valued signal in the transform domain in the first direction polarization, and Qx_1 represents a quadrature real-valued signal that is in the P first real-valued signals in the transformation region and that is in the first polarization direction; And

determining 2×m real-valued third signals in the transform domain that are in P real-valued third signals in the transform domain and that are in the second polarization direction, based on I _h (w), Q _h (w) and the in-phase real signal and the quadrature real signal , which are in the P first real-valued signals in the conversion domain and which are in the second polarization direction, includes:

Iy_3 = Iy_1⋅I _h (w)-Qy_1⋅Q _h (w); Qy_3 = Qy_1⋅I _h (w)+Iy_1⋅Q _h (w), where

Iy_3 represents the in-phase third real-valued signal in the transform domain in the second polarization direction, Iy_1 represents the in-phase real-valued signal that is in the P first real-valued signals in the transform domain and which is in the second polarization direction, Qy_3 represents the in-phase third real-valued signal in the transform domain in the second direction polarization, and Qy_1 represents an in-phase real-valued signal that is in the P first real-valued signals in the conversion region and that is in the second polarization direction.

[0021] With respect to the first aspect, in some implementations of the first aspect, the DBP processing specifically includes: performing NTT processing on P real-valued signals to obtain P fourth real-valued signals in a transform domain; performing in the transform region chromatic dispersion compensation for P fourth real-valued signals in the transform region to obtain P fifth real-valued signals in the transform region; performing INTT processing on P fifth real-valued signals in the transform domain to obtain P ninth real-valued signals in the time domain; and performing nonlinear compensation (compensation for nonlinear effects) on the P ninth real-valued time domain signals to obtain P tenth real-valued time domain signals.

[0022] With respect to the first aspect, some implementations of the first aspect may first perform polarization compensation and then perform INTT processing to obtain P seventh real-valued time domain signals.

[0023] Specifically, polarization compensation is performed on P second real-valued signals in the conversion domain to obtain P sixth real-valued signals in the conversion domain; and INTT processing is performed on the P sixth real-valued signals in the transform domain to obtain P seventh real-valued signals in the time domain. Equalization and depolarization are performed on the 2×m second real-valued signals in the transform domain that are in the P second real-valued signals in the transform domain and which are in the first polarization direction to obtain 2×m sixth real-valued signals in the transform domain that are in the P sixths real-valued signals in the conversion region and which are in the first polarization direction. Equalization and depolarization are performed on the 2×m second real-valued signals in the transform domain that are in the P second real-valued signals in the transform domain and which are in the second polarization direction to obtain 2×m sixth real-valued signals in the transform domain that are in the P sixths real-valued signals in the conversion region and which are in the second polarization direction.

[0024] The P seventh real-valued time domain signals obtained by INTT processing are then combined. The 2×m seventh real-valued time domain signals that are in the P seventh real-valued time domain signals and that are in the first polarization direction are combined into m complex-valued time domain signals X in the first polarization direction. The 2×m seventh real-valued time domain signals that are in the P seventh real-valued time domain signals and which are in the second polarization direction are combined into m complex-valued Y time domain signals in the second polarization direction.

[0025] With respect to the first aspect, some implementations of the first aspect may first perform INTT processing and then perform polarization compensation to obtain a complex valued time domain signal X and a complex valued time domain signal Y that have undergone polarization compensation.

[0026] Specifically, INTT processing is performed on P second real-valued signals in the transform domain to obtain P eighth real-valued signals in the time domain. The 2×m eighth real time domain signals that are in the P eighth real time domain signals and which are in the first polarization direction are combined into m complex valued time domain signals X in the first polarization direction. The 2×m eighth real time domain signals that are in the P eighth real time domain signals and which are in the second polarization direction are combined into m complex valued Y time domain signals in the second polarization direction. Time-domain polarization compensation is performed on m complex-valued time-domain signals X in the first polarization direction and m complex-valued time-domain Y signals in the second polarization direction to obtain m complex-valued time-domain signals X and m complex-valued time-domain signals Y that have been subjected to polarization compensation.

[0027] With respect to the first aspect, in some implementations of the first aspect, the P real-valued signals include 4×m real-valued signals in m transmission modes. m can be a value equal to 1 or greater than 1. P and m satisfy P=4×m. When m is equal to 1, it can be understood as single-mode transmission; or when m is greater than 1, it can be understood as multimode transmission.

[0028] According to a second aspect, a signal processing apparatus is provided including a processor. The processor is associated with memory and may be configured to execute an instruction in memory to implement the method of any of the first aspect or possible implementations of the first aspect.

[0029] Optionally, the device further includes a communication interface, and the processor is connected to the communication interface.

[0030] In an embodiment, the device is a digital signal processor. When the device is a digital signal processor, the communication interface may be a transceiver or an input/output interface.

[0031] In another embodiment, the device is a chip configured as a digital signal processor. When the device is a chip configured as a digital signal processor, the communication interface may be an input/output interface.

[0032] In another embodiment, the device is a chip or system of chips.

[0033] Optionally, the transceiver may be transceiver circuitry.

[0034] Optionally, the I/O interface may be an I/O circuit.

[0035] Specifically, the signal processing apparatus includes:

a first number theoretic transform (NTT) module configured to perform NTT processing on the P input signals to obtain P first real-valued signals in the transform domain;

a clock recovery unit configured to perform clock recovery for P first real-valued signals in the transformation region or P first real-valued signals in the transformation region obtained by chromatic dispersion compensation to obtain P second real-valued signals in the transformation region;

a polarization compensation module and a first inverse number theoretic transform (INTT) module configured to process P second real-valued signals in the transform domain to obtain m complex-valued time-domain signals X and m complex-valued time-domain Y signals; And

a phase recovery module and a decoding module configured to perform phase recovery and decoding of the complex valued time domain signal X and the complex valued time domain signal Y to obtain bit signals, where

m and P are natural numbers.

[0036] With respect to the second aspect, in some implementations of the second aspect, the P input signals include P real-valued signals, and that the first number theoretic transform (NTT) module is configured to perform NTT processing on the P input signals to obtain P first real-valued signals signals in a transform domain includes the following: a first number theoretical transform (NTT) module is configured to perform a number theoretical transform on the P real-valued signals to obtain P first real-valued signals in the transform domain.

[0037] For example, the P real-valued signals include 4×m real-valued signals in m transmission modes. m can be a value equal to 1 or greater than 1. P and m satisfy P=4×m. When m is equal to 1, it can be understood as single-mode transmission; or when m is greater than 1, it can be understood as multimode transmission.

[0038] With respect to the second aspect, in some implementations of the second aspect, the device further includes a first chromatic dispersion compensation module configured to perform chromatic dispersion compensation for P first real-valued signals in the transform domain to obtain P third real-valued signals in the transform domain; and the clock recovery unit is configured to perform clock recovery for the P third real-valued signals in the conversion domain to obtain the P second real-valued signals in the conversion domain.

[0039] With respect to the second aspect, in some implementations of the second aspect, the first chromatic dispersion compensation module is configured to perform chromatic dispersion compensation for P first real-valued signals in the transform domain to obtain P third real-valued signals in the transform domain, includes the following :

the first chromatic dispersion compensation module is configured to determine m real-valued third signals in the transform domain that are in the P third real-valued signals in the transform domain and that are in the first polarization direction based on the transform-domain chromatic dispersion equalization function I _h (w) corresponding to an in-phase real-valued signal, a chromatic dispersion equalization function Q _h (w) in the transformation region corresponding to the quadrature real-valued signal, and an in-phase real-valued signal and a quadrature real-valued signal that are in the P first real-valued signals in the transformation region and that are in the first polarization direction; And

the first chromatic dispersion compensation module is configured to determine 2×m real-valued third signals in the transform domain that are in the P third real-valued signals in the transform domain and that are in the second polarization direction based on I _h (w), Q _h (w), and an in-phase real-valued signal and a quadrature real-valued signal that are in the P first real-valued signals in the transform region and that are in the second polarization direction, where I _h (w) and Q _h (w) are obtained by performing NTT processing on the impulse response I _h (t ) chromatic dispersion corresponding to the in-phase real-valued signal, and for the impulse response Q _h (t) chromatic dispersion corresponding to the quadrature real-valued signal, respectively.

[0040] With respect to the second aspect, in some implementations of the second aspect, the third real-valued transform domain signals obtained by chromatic dispersion compensation and the first real-valued transform domain signals satisfy the following requirements:

I ^' _x (w) = I _x (w)⋅I _h (w)-Q _x (w)⋅Q _h (w);

Q ^' _x (w) = Q _x (w)⋅I _h (w)+I _x (w)⋅Q _h (w);

I ^' _y (w) = I _y (w)⋅I _h (w)-Q _y (w)⋅Q _h (w); And

Q ^' _y (w) = Q _y (w)⋅I _h (w)+I _y (w)⋅Q _h (w), where

I _x (w) represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the first polarization direction, Q _x (w) represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and that is in first polarization direction, I _y (w) represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the second polarization direction, Q _y (w) represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and which is in the second polarization direction, I ^' _x (w) represents an in-phase real-valued signal that is in the third real-valued signals in the transform region and which is in the first polarization direction, Q ^' _x (w) represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and which is in the first polarization direction, I ^' _y (w) represents the in-phase real-valued signal that is in the third real-valued signals in the transform domain and which is in the second polarization direction, and Q ^' _y (w) represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and that is in the second polarization direction.

[0041] With respect to the second aspect, in some implementations of the second aspect, the first chromatic dispersion compensation module includes a third NTT module, a combiner, and a multiplier. The third NTT module is configured to perform NTT processing on a chromatic dispersion impulse response in a time domain to obtain a chromatic dispersion equalization function and perform NTT processing on the P first real-valued signals in the transform domain. The multiplying unit is configured to multiply the chromatic dispersion equalization function by P first real-valued signals in the transform domain obtained by NTT processing. The combining module is configured to combine signals obtained through processing by the multiplying module.

[0042] With respect to the second aspect, in some implementations of the second aspect, the P input signals include P tenths real-valued time domain signals; and the apparatus further includes a digital backpropagation (DBP) calculation module configured to perform DBP processing on the P digital signals to obtain P tenths real-valued time domain signals.

[0043] When a device includes a DBP module, the device uses the DBP module to implement compensation for nonlinear effects based on linear distortion to increase the signal transmission distance.

[0044] With respect to the second aspect, in some implementations of the second aspect, the DBP module sequentially includes:

a second NTT module, a second chromatic dispersion compensation module, a second INTT module and a non-linear compensation module, wherein

that the DBP module is configured to perform DBP processing on P digital signals to obtain P tenths real-valued signals in the time domain includes the following:

a second NTT module is configured to perform NTT processing on P digital signals to obtain P fourth real-valued signals in the conversion domain;

a second chromatic dispersion compensation module is configured to perform, in the transform domain, chromatic dispersion compensation for P fourth real-valued signals in the transform domain to obtain P fifth real-valued signals in the transform domain;

a second INTT module is configured to perform INTT processing on P fifth real-valued signals in the transform domain to obtain P ninth real-valued signals in the time domain; And

the nonlinear compensation module is configured to perform nonlinear compensation for P ninth real-valued time domain signals to obtain P tenth real-valued time domain signals.

[0045] With respect to the second aspect, in some implementations of the second aspect, the clock recovery module is configured to perform clock recovery on P first real-valued signals in the transform domain to obtain P second real-valued signals in the transform domain.

[0046] With respect to the second aspect, in some implementations of the second aspect, the device further includes:

merging module, while

a polarization compensation module is configured to perform polarization compensation for P second real-valued signals in the conversion domain to obtain P sixth real-valued signals in the conversion domain;

a first inverse number theoretic transform (INTT) module is configured to perform INTT processing on P sixth real-valued signals in a transform domain to obtain P seventh real-valued signals in a time domain; And

a combining module is configured to combine each two of the 2×m seventh real-valued time domain signals that are in the P seventh real-valued time domain signals and that are in the first polarization direction to obtain m complex-valued time domain signals X in the first polarization direction; And

combining each two of the 2×m seventh real-valued time-domain signals that are in the P seventh real-valued time-domain signals and that are in the second polarization direction to obtain m complex-valued Y time-domain signals in the second polarization direction.

[0047] For example, the polarization compensation module includes a first butterfly filter and a second butterfly filter.

[0048] The first butterfly filter is configured to perform equalization and depolarization on 2×m second real-valued signals in the transform domain that are in P second real-valued signals in the transform domain and that are in the first polarization direction to obtain 2×m sixth real-valued signals in the transformation region, which are in P sixth real-valued signals in the transformation region and which are in the first polarization direction.

[0049] The second butterfly filter is configured to perform equalization and depolarization on 2×m second real-valued signals in the transform domain that are in P second real-valued signals in the transform domain and that are in the second polarization direction to obtain 2×m sixth real-valued signals in the transformation region, which are in P sixth real-valued signals in the transformation region and which are in the second polarization direction.

[0050] Specifically, during the signal processing procedure, when the device first performs polarization compensation processing based on the polarization compensation unit, and then performs INTT processing based on the first INTT unit, polarization compensation processing is performed based on the signal in the transform domain. Distortion compensation by polarization equalization is performed in the transform domain, and multiplication is used to replace convolution so that power consumption can be reduced.

[0051] Alternatively, the apparatus further includes a combiner, wherein the first inverse number theoretic transform (INTT) module is configured to perform INTT processing on P second real-valued signals in a transform domain to obtain P eighth real-valued signals in a time domain;

a combining unit is configured to combine each two of the 2×m eighth real-valued time domain signals that are in the P eighth real-valued time domain signals and that are in a first polarization direction to obtain m complex-valued time domain signals X in the first polarization direction; And

combining each two of the 2×m eighth real-valued time domain signals that are in the P eighth real-valued time domain signals and that are in the second polarization direction to obtain m complex-valued Y time domain signals in the second polarization direction; And

the polarization compensation module is configured to perform polarization compensation in the time domain for the m complex-valued signals X and the m complex-valued signals Y.

[0052] For example, the polarization compensation module includes a third butterfly filter. The third butterfly filter is configured to perform time-domain polarization compensation for m complex-valued time-domain signals X in a first polarization direction and m complex-valued time-domain Y signals in a second polarization direction to obtain m complex-valued time-domain signals X and m complex-valued signals Y in the time domain that have undergone polarization compensation.

[0053] Specifically, during the signal processing procedure, when the device first performs INTT processing based on the first INTT module and then performs polarization compensation processing based on the polarization compensation module, polarization compensation processing is performed based on the time domain signal obtained by processing INTT. The ability to withstand loop delay can be improved by performing polarization compensation processing in the time domain.

[0054] According to a third aspect, a coherent receiver is provided. The coherent receiver includes signal processing apparatus according to any of the second aspect or possible implementations of the second aspect.

[0055] In addition, the coherent receiver further includes a polarization beam splitter, a frequency mixer, a photoelectric detector, and an analog-to-digital converter. The polarization beam splitter is configured to receive signals in two polarization directions. The frequency mixer is configured to perform frequency mixing processing for signals with the same polarization direction. The photoelectric detector is configured to convert the strength of an optical signal into an electrical signal strength. The analog-to-digital converter is configured to convert an analog signal into a digital signal.

[0056] According to a fourth aspect, a chip is provided. The chip includes a communication interface, memory and processor. The memory is configured to store a computer program. The processor is configured to read and execute a computer program stored in memory such that the chip implements a signal processing method in accordance with any of the first aspect or possible implementations of the first aspect.

[0057] According to a fifth aspect, a signal processing apparatus is provided including a processor and a communication interface. The processor is connected to memory. The processor may be configured to execute program code in memory to implement a signal processing method according to any of the first aspect or possible implementations of the first aspect.

[0058] According to a sixth aspect, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program. When a computer program is executed by a device, the device may implement a signal processing method according to any of the first aspect or possible implementations of the first aspect.

[0059] According to a seventh aspect, a computer program product is provided including instructions. When the instruction is executed by the computer, the device is provided with the ability to implement a signal processing method according to any of the first aspect or possible implementations of the first aspect.

Brief description of drawings

[0060] FIG. 1(a) and Fig. 1(b) is a schematic illustration of a coherent polarization multiplexing receiver;

[0061] FIG. 2 is a schematic diagram of an Rx DSP architecture according to an embodiment of the present invention;

[0062] FIG. 3 is a schematic diagram of another Rx DSP architecture according to an embodiment of the present invention;

[0063] FIG. 4(a) is a schematic illustration of an implementation of chromatic dispersion compensation based on FFT transform; Fig. 4(b) is a schematic block diagram of a chromatic dispersion compensation module according to an embodiment of the present invention;

[0064] FIG. 5 is a schematic diagram of yet another Rx DSP architecture according to an embodiment of the present invention;

[0065] FIG. 6 is a schematic diagram of yet another Rx DSP architecture according to an embodiment of the present invention;

[0066] FIG. 7 is a schematic block diagram of a DBP module according to an embodiment of the present invention;

[0067] FIG. 8(a) is a schematic block diagram of a first polarization compensation module according to an embodiment of the present invention; Fig. 8(b) is a schematic block diagram of a second polarization compensation module according to an embodiment of the present invention;

[0068] FIG. 9(a) is a schematic diagram of an Rx DSP architecture in a multimode transmission scenario according to an embodiment of the present invention; Fig. 9(b) is a schematic diagram of an Rx DSP architecture in another multimode transmission scenario according to an embodiment of the present invention; Fig. 9(c) is a schematic diagram of an Rx DSP architecture in yet another multimode transmission scenario according to an embodiment of the present invention; Fig. 9(d) is a schematic diagram of an Rx DSP architecture in another multimode transmission scenario according to an embodiment of the present invention; Fig. 9(e) is a schematic block diagram of another DBP module according to an embodiment of the present invention;

[0069] FIG. 10(a) is a schematic block diagram of a signal processing method according to an embodiment of the present invention; Fig. 10(b) is a schematic block diagram of DBP processing according to an embodiment of the present invention; Fig. 10(c) is a schematic block diagram of obtaining a complex-valued time domain signal according to an embodiment of the present invention; Fig. 10(d) is another schematic block diagram of obtaining a complex valued time domain signal according to an embodiment of the present invention; And

[0070] FIG. 11 is a schematic diagram of a chip according to an embodiment of the present invention.

Description of Embodiments

[0071] The technical solutions of the present invention will now be described with reference to the accompanying drawings.

[0072] Technical solutions in embodiments of the present invention relate to improvements in a coherent polarization multiplexing receiver. The following first describes a conventional polarization multiplexing coherent receiver with reference to FIG. 1(a) and Fig. 1(b). Fig. 1(a) and Fig. 1(b) is a schematic illustration of a single-mode coherent polarization multiplexing receiver.

[0073] As can be understood from FIG. 1(a) and Fig. 1(b) it can be seen that the signal processing procedure of a single-mode coherent polarization multiplexing receiver includes the following steps.

[0074] The optical signal is separated into signals x1 and y1 by processing by polarization beam splitter #1 in the Polarization beam splitter 101. x1 is sent to frequency mixer #1 in frequency mixer (hybrid) 102, and y1 is sent to frequency mixer #2 There are 102 frequencies in the mixer.

[0075] Light from the local oscillator (eg, generated by a local laser or transmitter) is separated into signals x2 and y2 by processing by polarization beam splitter #2 in polarization beam splitter 101. x2 is sent to frequency mixer #1 in frequency mixer 102, and y2 sent to frequency mixer #2 in frequency mixer 102.

[0076] It should be noted that the optical signal or light from a local oscillator has two polarization modes: transverse electric (TE) and transverse magnetic (TM), which may also be called the X polarization state and the Y polarization state, or also may be called the X polarization direction and the Y polarization direction.

[0077] The X polarization state and the Y polarization state are orthogonal to each other. In particular, after polarization rotation, a light ray in one polarization state (Y polarization state) becomes a light ray in an X polarization state. It should be noted that X and Y in this case are not the x-axis and y-axis in the narrow sense, but they are two orthogonal directions in a broad sense, for example, horizontal direction 45° and vertical direction 45°.

[0078] Frequency mixer #1 and frequency mixer #2 output four signals. These four signals are used as input signals of the photoelectric detector (photodiode) 103 and are input separately into the photoelectric detector #1 to the photoelectric detector #4. The photoelectric detector is configured to convert the strength of an optical signal into an electrical signal strength.

[0079] The output signal of photoelectric detector #1 is an Ix signal, the output signal of photoelectric detector #2 is a Qx signal, the output signal of photoelectric detector #3 is an Iy signal, and the output signal of photoelectric detector #4 is a Qy signal.

[0080] Ix, Qx, Iy and Qy are converted by A/D converter #1 to A/D converter #4 in analog-to-digital converter (ADC) 104, respectively, and digital signals Ix, Qx are output , Iy and Qy obtained through analog-to-digital conversion. Ix and Qx are combined by a combiner #1 in the combiner 105 to obtain a complex-valued signal X. Iy and Qy are combined by a combiner #2 in the combiner 105 to obtain a complex-valued signal Y. For example, the combiner in the present invention may also be called an adder.

[0081] The complex valued signal X is input to FFT module #1 in FFT module 106 and processed by FFT module #1 to obtain a frequency domain signal X that has undergone a fast Fourier transform. The complex-valued signal Y is input to FFT unit #2 in FFT unit 106 and processed by FFT unit #2 to obtain a frequency domain signal Y that has undergone a fast Fourier transform.

[0082] The frequency domain signal X is input to the chromatic dispersion compensation unit #1 in the chromatic dispersion compensation (CDC) unit 112, and chromatic dispersion compensation is performed in the frequency domain to obtain a frequency domain signal X that has undergone dispersion compensation in the frequency domain. The frequency domain signal Y is input to the chromatic dispersion compensation unit #2 in the chromatic dispersion compensation unit 112, and chromatic dispersion compensation is performed in the frequency domain to obtain a frequency domain signal Y that has undergone frequency domain chromatic dispersion compensation.

[0083] The frequency domain signal X obtained by frequency domain chromatic dispersion compensation is input to the clock recovery unit #1 in the clock recovery unit 107, and clock recovery is performed to obtain the frequency domain signal X that has undergone clock recovery frequencies. The frequency domain signal Y obtained by frequency domain chromatic dispersion compensation is input to the clock recovery unit #2 in the clock recovery unit 107, and clock recovery is performed to obtain the frequency domain signal Y that has undergone clock recovery.

[0084] The frequency domain signal X obtained by clock recovery is input to IFFT unit #1 in IFFT unit 113 and processed by IFFT unit #1 to obtain a time domain signal X that has undergone an inverse fast Fourier transform. The frequency domain signal Y obtained by clock recovery is input to IFFT unit #2 in IFFT unit 113 and processed by IFFT unit #2 to obtain a time domain signal Y that has undergone an inverse fast Fourier transform.

[0085] Both the time domain signal X and the time domain signal Y are input to the polarization compensation unit 108 . Optionally, the polarization compensation module includes 2x2 butterfly filters and performs polarization demultiplexing and distortion equalization to obtain a time domain X signal and a time domain Y signal that have undergone polarization compensation. In particular, as shown in FIG. 1(a) and Fig. 1(b), the polarization compensation unit 108 further includes a coefficient updating unit 109. The coefficient updating unit 109 is configured to update the coefficient of the filter included in the polarization compensation unit 108 .

[0086] The complex value signal X obtained by polarization compensation is sequentially sent to the phase recovery unit #1 in the phase recovery unit 110 and to the decoding unit #1 in the decoder unit 111 to obtain a bit signal. The complex value signal Y obtained by polarization compensation is sequentially sent to the phase recovery unit #2 in the phase recovery unit 110 and to the decoding unit #2 in the decoding unit 111 to obtain a bit signal.

[0087] It should be noted that since the FFT module 106 and the IFFT module 113 are introduced into the current coherent receiver signal processing routine, the following cases may occur:

[0088] (1) The complexity of the FFT transformation increases with the number of transformation points at a rate of Nlog2N.

[0089] (2) FFT transformation involves a large number of multiplication operations, and the complexity of multiplication is quite high.

[0090] (3) The FFT transformation matrix is a complex number matrix based on the exponent e, and cannot accurately indicate the truncation error and fixed-point calculation error on a computer with a limited number of bits.

[0091] This embodiment of the present invention mainly relates to improvements in the post-ADC processing procedure. Therefore, for ease of description, the post-ADC processing procedure is collectively referred to as Receiver Digital Signal Processing (Rx DSP), for example, the Rx DSP signal processing procedure included in the large area outlined by the dotted line in FIG. 1(a) and Fig. 1(b).

[0092] For ease of understanding, the present invention briefly describes several basic concepts.

[0093] 1. Polarization multiplexing

[0094] The transmission mode, that is, the HE ₁₁ mode, for an optical signal in a single-mode optical fiber includes two submodes: HE ^x ₁₁ and HE ^y ₁₁ . These two submodes are independent of each other, and their polarization directions are orthogonal to each other. These two submodes present different shapes during transmission: linear polarization, elliptical polarization and circular polarization, but always remain orthogonal. X-polarization and Y-polarization in the present invention are two orthogonal polarization states in a multiplexing single-mode optical fiber.

[0095] Compared to single-mode optical fiber, multimode optical fiber can transmit signals in a variety of modes. Currently, the LP _pq mode is commonly used. The values of p and q represent different mode field characteristics of the LP mode.

[0096] 2. NTT processing and INTT processing

[0097] Both NTT processing and FFT processing in the present invention relate to transformation for realizing fast convolution. FFT processing is a complex number transformation whose kernel is K=exp(jα). exp(jα) is a complex number and has a storage truncation error. In addition, the difficulty of multiplying a complex number is quite high. An NTT transformation is a transformation that is defined on a finite field and whose kernel is K (K is an integer). Typically K is 2. Therefore, the transformation matrix does not have a truncation error. On a binary computer, both multiplication and division by 2 can be implemented through shifting, so that multiplication can be avoided and power consumption reduced.

[0098] In addition, INTT processing is the inverse process of NTT processing, and IFFT is the inverse process of FFT processing. Details are not described in the present invention.

[0099] 3. Transformation area and time area

[0100] The time domain signal in embodiments of the present invention is a signal that is input to the Rx DSP process through analog-to-digital conversion. The signal obtained by performing FFT processing on a time domain signal is called a frequency domain signal, and the signal obtained by NTT processing is called a transform domain signal.

[0101] In addition, a time domain signal is obtained by performing IFFT processing on a frequency domain signal, and a time domain signal is obtained by performing INTT processing on a transform domain signal.

[0102] More specifically, a time domain signal in the present invention can be understood as a signal with time detail and can change as time changes; a frequency domain signal can be understood as a frequency domain representation into which a time domain signal is converted by time domain to frequency domain conversion; and similarly, a transform domain signal can be understood as a transform domain representation into which a time domain signal is converted by a time domain to transform domain transform. During the time domain to conversion process, the time domain signal is usually segmented, then a portion is superimposed at the beginning and end, and then the signal is sent to the NTT/INTT conversion module. For example, a time domain data segment includes 1024 points. The time domain signal is first divided into 512 points, then 256 points are superimposed at the beginning and end, and then 1024 points are sent to the NTT/INTT conversion module. 512, 256 and 1024 are just examples here. A certain overlay proportion and a certain number of points in a segment are not limited in the present invention. The number of data conversion points is the length of the array when converting data. Generally, a higher number of points results in better conversion performance.

[0103] 4. In-phase real signal and quadrature real signal

[0104] Quadrature amplitude modulation is widely used in modern coherent communications. Quadrature amplitude modulation is a modulation scheme in which amplitude modulation is performed on two orthogonal carriers. The two carriers are usually sine waves with a phase difference of 90° and are therefore called orthogonal carriers. The real part of a complex-valued signal is called the codirectional component, and the imaginary part of the signal is called the quadrature component.

[0105] It should be noted that both the real part and the imaginary part of a complex-valued signal (eg, X=a+bj) are real numbers (eg, both a and b are real numbers). Therefore, in the present invention, the co-directional component is referred to as an in-phase real signal, and the quadrature component is referred to as a quadrature real signal.

[0106] From the above descriptions, it can be understood that introducing an FFT module and an IFFT module into an electrical signal processing procedure of a coherent receiver may increase the complexity of the signal processing procedure. To solve a problem in the existing electrical signal processing procedure of a coherent receiver and reduce the power consumption of the algorithm, embodiments of the present invention provide a signal processing method and apparatus in which an electrical signal is processed without introducing an FFT module or an IFFT module, thereby improving the performance of a polarization multiplexing coherent receiver at electrical signal processing and reducing complexity.

[0107] The signal processing method and apparatus provided in the embodiments of the present invention are described in detail below with reference to the accompanying drawings. The signal processing method provided in embodiments of the present invention can be used by a coherent polarization multiplexing receiver to process an electrical signal in a single-mode or multi-mode transmission scenario.

[0108] FIG. 2 is a schematic diagram of an Rx DSP architecture according to an embodiment of the present invention. The 200 Rx DSP architecture includes:

a first NTT unit 211, a first chromatic dispersion compensation unit 212, a clock recovery unit 213, a first polarization compensation unit 214, a first INTT unit 215, a combining unit 216, a phase recovery unit 217, and a decoding unit 218.

[0109] The first NTT module includes NTT module #1, NTT module #2, NTT module #3, and NTT module #4. The signals I ¹ _x , Q ¹ _x , I ¹ _y and Q ¹ _y are input to NTT module #1, NTT module #2, NTT module #3 and NTT module #4, respectively.

[0110] Optionally, the signals I ¹ _x , Q ¹ _x , I ¹ _y and Q ¹ _y are real-valued signals obtained by ADC conversion.

[0111] NTT module #1, NTT module #2, NTT module #3 and NTT module #4 perform NTT processing for I ¹ _x , Q ¹ _x , I ¹ _y and Q ¹ _y , respectively, and output the first real-valued signals Ix_1 , Qx_1, Iy_1 and Qy_1 in the conversion region to the first chromatic dispersion compensation unit 212.

[0112] The first chromatic dispersion compensation unit 212 includes a chromatic dispersion compensation unit #1 and a chromatic dispersion compensation unit #2. Ix_1 and Qx_1 are input to the chromatic dispersion compensation unit #1, and chromatic dispersion compensation is performed to obtain the third real-valued signals Ix_3 and Qx_3 in the conversion region. Iy_1 and Qy_1 are input to the chromatic dispersion compensation unit #2, and chromatic dispersion compensation is performed to obtain the third real-valued signals Iy_3 and Qy_3 in the conversion region.

[0113] The clock recovery unit 213 includes a clock recovery unit #1 and a clock recovery unit #2. Ix_3 and Qx_3 are input to clock recovery unit #1, and clock recovery is performed to obtain second real-valued signals Ix_2 and Qx_2 in the conversion region. Iy_3 and Qy_3 are input to clock recovery unit #2, and clock recovery is performed to obtain second real-valued signals Iy_2 and Qy_2 in the conversion region.

[0114] The second real-valued signals Ix_2, Qx_2, Iy_2 and Qy_2 are input to the first polarization compensation unit 214, and polarization demultiplexing and distortion equalization are performed to obtain the sixth real-valued signals Ix_5, Qx_5, Iy_5 and Qy_5 in the transform domain. Optionally, the first polarization compensation unit 214 further includes a coefficient updating unit 2141. The coefficient update unit 2141 is configured to update the coefficient of the filter included in the first polarization compensation unit 214.

[0115] The first INTT module 215 includes an INTT module #1, an INTT module #2, an INTT module #3, and an INTT module #4. The signals Ix_5, Qx_5, Iy_5 and Qy_5 are input to INTT module #1, INTT module #2, INTT module #3 and INTT module #4, respectively.

[0116] INTT module #1, INTT module #2, INTT module #3 and INTT module #4 perform INTT processing for Ix_5, Qx_5, Iy_5 and Qy_5, respectively, and output the seventh real-valued signals Ix_6, Qx_6, Iy_6 and Qy_6 in time areas into the merging module 216.

[0117] The combining module 216 includes a combining module #1 and a combining module #2. Ix_6 and Qx_6 are input to combiner #1 and combined to obtain a complex-valued time domain signal X. Iy_6 and Qy_6 are input to combiner #2 and combined to obtain a complex-valued time domain signal Y.

[0118] The phase restoration unit 217 includes a phase restoration unit #1 and a phase restoration unit #2. The decoding unit 111 includes a decoding unit #1 and a decoding unit #2. The complex valued time domain signal X is sequentially input to the phase recovery unit #1 and the decoding unit #1 to obtain a reconstructed bit signal. The complex valued time domain signal Y is sequentially input to the phase recovery unit #2 and the decoding unit #2 to obtain a reconstructed bit signal.

[0119] The 200 Rx DSP architecture shown in FIG. 2 can be called an Rx DSP architecture based entirely on the conversion domain. The Rx DSP architecture, which is entirely based on the conversion domain, not only performs NTT/INTT-based chromatic dispersion compensation, but also performs NTT/INTT-based polarization compensation, that is, performs chromatic dispersion compensation and polarization compensation in the transform domain. In particular, distortion compensation by polarization equalization is performed in the transform region, and multiplication is used to replace convolution, so that power consumption can be further reduced. In addition, the number (size) of transformation points can be increased so that distortion is effectively leveled out and performance is improved.

[0120] FIG. 3 is a schematic diagram of another Rx DSP architecture according to an embodiment of the present invention. The 300 Rx DSP architecture includes:

a first NTT unit 211, a first chromatic dispersion compensation unit 212, a clock recovery unit 213, a second polarization compensation unit 310, a coefficient updating unit 311, a first INTT unit 215, a combining unit 216, a phase recovery unit 320, and a decoding unit 330.

[0121] The pre-clock recovery modules in architecture 300 are the same as those in architecture 200. The difference is that the output of clock recovery module 213 is first processed by INTT module 215 and combined by combiner 216, and then inputted into the second module. 310 polarization compensation. The process until the clock frequency is restored is again not described.

[0122] The first INTT module 215 includes an INTT module #1, an INTT module #2, an INTT module #3, and an INTT module #4. The second real-valued signals Ix_2, Qx_2, Iy_2 and Qy_2 are input to INTT module #1, INTT module #2, INTT module #3 and INTT module #4, respectively.

[0123] INTT module #1, INTT module #2, INTT module #3 and INTT module #4 perform INTT processing for Ix_2, Qx_2, Iy_2 and Qy_2, respectively, and output the eighth real-valued signals Ix_7, Qx_7, Iy_7 and Qy_7 in time areas into the merging module 216.

[0124] The combining unit 216 includes a combining unit #1 and a combining unit #2. Ix_7 and Qx_7 are input to combiner #1 and combined to obtain a complex-valued time domain signal X. Iy_7 and Qy_7 are input to combiner #2 and combined to obtain a complex-valued time domain signal Y.

[0125] The complex time domain signal X and the complex time domain signal Y are input to the second polarization compensation unit 310, and polarization demultiplexing and distortion equalization are performed to obtain the complex time domain signal X and the complex time domain signal Y that have undergone compensation. polarization. The second polarization compensation unit 310 further includes a coefficient updating unit 311. The coefficient updating unit 311 is configured to update the coefficient of the filter included in the second polarization compensation unit 310.

[0126] The phase restoration unit 320 includes a phase restoration unit #1 and a phase restoration unit #2. The decoding unit 111 includes a decoding unit #1 and a decoding unit #2. The complex valued time domain signal X obtained by polarization compensation is sequentially input to the phase recovery unit #1 and the decoding unit #1 to obtain the reconstructed bit signal. The complex valued time domain signal Y obtained by polarization compensation is sequentially input to the phase recovery unit #2 and the decoding unit #2 to obtain the reconstructed bit signal.

[0127] The 300 Rx DSP architecture shown in FIG. 3 may be called an Rx DSP architecture combining time domain and conversion domain. In the Rx DSP architecture combining time domain and conversion domain, chromatic dispersion compensation is performed on an NTT/INTT basis, and polarization compensation is still performed in the time domain, that is, chromatic dispersion compensation is performed in the conversion domain, but polarization compensation is performed in the time domain . In particular, the chromatic dispersion compensation module with quite high power consumption realizes compensation in the conversion domain, so that the overall power consumption can be effectively reduced, and the polarization equalization module remains in the time domain to perform compensation, so that the ability of the equalization module to withstand loop delay is increased.

[0128] FIG. 4(a) is a schematic block diagram of a chromatic dispersion compensation module 400. The chromatic dispersion compensation module 400 may be applied to the architecture shown in FIG. 1(a) and Fig. 1(b). The chromatic dispersion compensation unit includes an FFT unit 402, an FFT unit 404, a combining unit 401, a combining unit 403, and a multiplying unit 405.

[0129] Signals are transmitted via optical fiber. The effect of the chromatic dispersion of an optical fiber on signals can be described as the impulse response. When equalizing chromatic dispersion in the time domain, finite impulse response (FIR) is used for compensation. The FIR ratio is as follows: . indicates the chromatic dispersion coefficient of the optical fiber, indicates the signal transmission distance in the optical fiber, indicates an imaginary number and indicates the time. Since the signal in the x polarization direction and the signal in the y polarization direction experience the same distortion due to chromatic dispersion, the same function h(t) is used during equalization.

[0130] In the FFT-based chromatic dispersion compensation module in the signal processing procedure shown in FIG. 1(a) and Fig. 1(b), the input of the chromatic dispersion compensation module is a complex-valued signal X(ω) in the frequency domain.

[0131] For example, I _h (t) and Q _h (t) are time domain chromatic dispersion impulse responses, I _h (t) and Q _h (t) are combined by the combiner 403 into a complex valued impulse response H (t), H(t) is input to the FFT module 404 to obtain the frequency domain equalization function H(ω), and H(ω) and X(ω) are input to the multiplying module 405 and multiplied to obtain the output of the chromatic dispersion compensation module: Xout( ω) = H(ω)⋅X(ω).

[0132] FIG. 4(b) is a schematic block diagram of a chromatic dispersion compensation module 410 according to an embodiment of the present invention. The chromatic dispersion compensation module 410 may be applied to the 200 and 300 RX DSP architectures. The chromatic dispersion compensation module includes an NTT module 411, an NTT module 412, a NTT module 419, an NTT module 420, a combiner 414, a combiner 418, a multiplier 413, a multiply 415, a multiplier 416, and a multiplier 417.

[0133] In the signal processing procedures shown in FIG. 2 and Fig. 3, the NTT-based chromatic dispersion compensation module separately performs chromatic dispersion compensation for the signal in the x polarization direction and the signal in the y polarization direction. The chromatic dispersion compensation process (as shown in FIG. 4(b)) is described using chromatic dispersion compensation in the polarization direction x as an example.

[0134] For example, impulse response chromatic dispersion in the polarization direction x is divided into I _h (t) and Q _h (t), where And .

[0135] I _h (t) and Q _h (t) are input to NTT module 419 and NTT module 420, respectively, and converted to obtain chromatic dispersion equalization functions I _h (w) and Q _h (w) in the NTT transform region. The co-directional signal I _x (t) and the quadrature signal Q _x (t) in the polarization x direction are input to NTT module 411 and NTT module 412, respectively, and are converted to obtain I _x (w) and Q _x (w) signals in the conversion domain .

[0136] As can be understood from FIG. 4(b) it can be seen that the result obtained by inputting the co-directional signal I _x (w) in the transform region and the co-directional component I _h (w) of the equalization function into the multiplication unit 413 and performing the multiplication, and the result obtained by inputting the quadrature signal Q _x (w) in the transform domain and the quadrature component Q _h (w) the alignment functions into the multiplication unit 415 and performing the multiplication are combined by the combine unit 414 as follows: Ix_out(w) = I _x (w)⋅I _h (w)-Q _x (w)⋅Q _h (w).

[0137] The result obtained by inputting Q _x (w) and I _h (w) into the multiplication and execution unit 416, and the result obtained by inputting I _x (w) and Q _h (w) into the multiplication and execution unit 417 multiplications are combined by the combiner 418 as follows: Qx_out(w) = Q _x (w)⋅I _h (w)-I _x (w)⋅Q _h (w).

[0138] Similarly, when chromatic dispersion compensation is performed in the y polarization direction, the co-directional signal I _y (t) and the quadrature signal Q _y (t) in the y polarization direction are input to the NTT modules and converted to obtain the I _y (w) and Q _y (w) in the transformation domain, respectively. The process of compensating chromatic dispersion in the polarization direction y can be described using the following formulas:

Iy_out(w) = I _y (w)⋅I _h (w)-Q _y (w)⋅Q _h (w); And

Qy_out(w) = Q _y (w)⋅I _h (w)+I _y (w)⋅Q _h (w).

[0139] In actual operation, the alignment function H(ω) in the Fourier transform domain in FIG. 4(a), and the chromatic dispersion equalization functions I _h (w) and Q _h (w) in the number theoretic transformation region in FIG. 4(b) only need to be calculated once and stored in memory, and can be called up repeatedly for different input signals.

[0140] For example, one NTT transform and one FFT transform are used as examples to explain the difference between the two types of transforms:

[0141] Both FFT and NTT transformation matrices can be represented as follows:

.

[0142] The difference lies in the FFT transformation kernel: . NTT transformation kernel: α - power of two. Typically α is 2 (or may be or other meaning which is not limited in the present invention). For example, . Each element in the transformation matrix still has a power of 2, and multiplication by 2 or power of 2 in a binary operation can be done by shifting the bits.

[0143] Accordingly, for example, the core . The H transformation matrix for NTT and the H ^-1 transformation matrix for INTT are respectively as follows:

;

.

[0144] The internal working processes of clock recovery module #1 and clock recovery module #2 are similar to those in FIG. 4(b), and only the chromatic dispersion equalization functions I _h (w) and Q _h (w) need to be replaced by functions for equalizing clock delays. The delay caused by the clock frequency is expressed as using finite time domain impulse response. In the clock recovery module, I _h (w) and Q _h (w) can be replaced by NTT conversion . The details are not described here.

[0145] FIG. 5 is a schematic diagram of yet another Rx DSP architecture according to an embodiment of the present invention. The 500 Rx DSP architecture includes:

a digital back propagation (DBP) module 510, a first NTT module 211, a clock recovery module 213, a first polarization compensation module 214, a first INTT module 215, a combining module 216, a phase recovery module 217, and a decoding module 218.

[0146] The architecture after the first polarization compensation module 214 in the 500 Rx DSP architecture is the same as in the 200 Rx DSP architecture. The difference is that the signal undergoes DBP processing before polarization compensation. Since the DBP module 510 has a chromatic dispersion compensation function, the process after the DBP module may not require the chromatic dispersion compensation module 212.

[0147] The I ¹ _x , Q ¹ _x , I ¹ _y and Q ¹ _y signals are input to the DBP module 510, DBP processing is performed, and the time domain signals Ix_4, Qx_4, Iy_4 and Qy_4 are output to the first NTT module 211.

[0148] The first NTT module 211 includes NTT module #1, NTT module #2, NTT module #3, and NTT module #4. Ix_4, Qx_4, Iy_4 and Qy_4 are input to NTT module #1, NTT module #2, NTT module #3 and NTT module #4, respectively.

[0149] NTT module #1, NTT module #2, NTT module #3 and NTT module #4 perform NTT processing for Ix_4, Qx_4, Iy_4 and Qy_4, respectively, and output the first real-valued signals Ix_1, Qx_1, Iy_1 and Qy_1 in the region conversion to the clock recovery module 213.

[0150] The clock recovery unit 213 includes a clock recovery unit #1 and a clock recovery unit #2. Ix_1 and Qx_1 are input to clock recovery unit #1, and clock recovery is performed to obtain second real-valued signals Ix_2 and Qx_2 in the conversion region. Iy_1 and Qy_1 are input to clock recovery unit #2, and clock recovery is performed to obtain second real-valued signals Iy_2 and Qy_2 in the conversion region.

[0151] For the process after the clock recovery unit 213, see the embodiment shown in FIG. 2.

[0152] The 500 Rx DSP architecture shown in FIG. 5 can be called an Rx DSP architecture based entirely on the conversion domain in which DBP is added. In contrast to the entirely transform domain based Rx DSP architecture shown in FIG. 2 is that the DBP module has been added. It should be understood that in an Rx DSP architecture based entirely on the conversion domain shown in FIG. 2, only the linear distortion of the optical fiber is compensated, for example, by chromatic dispersion compensation. In the Rx DSP architecture, which is completely based on the conversion domain, in which the DBP module is added, the DBP module is used to realize linear distortion based non-linear effects compensation to increase the signal transmission distance.

[0153] FIG. 6 is a schematic diagram of yet another Rx DSP architecture according to an embodiment of the present invention. The 600 Rx DSP architecture includes:

DBP module 510, first NTT module 211, clock recovery module 213, second polarization compensation module 310, coefficient updating module 311, first INTT module 215, combining module 216, phase recovery module 320 and decoding module 330.

[0154] The architecture after the second polarization compensation module 310 in the 600 Rx DSP architecture is the same as in the 300 Rx DSP architecture. The architecture in front of the second polarization compensation module 310 is the same as the 500 Rx DSP architecture. More specifically, the received signal obtained by ADC conversion is first subjected to DBP processing. For the process before DBP processing and clock recovery, see the embodiment shown in FIG. 5. For the process after the clock recovery unit 213, see the embodiment shown in FIG. 3.

[0155] The 600 Rx DSP architecture shown in FIG. 6 may be referred to as an Rx DSP architecture combining a time domain and a conversion domain to which DBP is added. Unlike the Rx DSP architecture combining time domain and transform domain shown in FIG. 3 is that the DBP module has been added. It should be understood that in the Rx DSP architecture combining time domain and transform domain shown in FIG. 3, only compensates for the linear distortion of the optical fiber, such as by chromatic dispersion compensation. In the Rx DSP architecture, which combines time domain and transform domain, to which a DBP module is added, the DBP module is used to realize linear distortion based non-linear effects compensation to increase the signal transmission distance.

[0156] FIG. 7 is a schematic block diagram of a DBP module according to an embodiment of the present invention. The 700 DBP module includes:

a second NTT module 711, a second chromatic dispersion compensation module 712, a second INTT module 713, and a non-linear compensation module 714, wherein the structure of the second chromatic dispersion compensation module 712 is shown in FIG. 4(b) and is not described in detail here.

[0157] As can be seen from the block diagram of the DBP module shown in FIG. 7, performing DBP processing on I ¹ _x , Q ¹ _x , I ¹ _y and Q ¹ _y to obtain the real-valued signals Ix_4, Qx_4, Iy_4 and Qy_4 shown in FIG. 5 and Fig. 6, in particular, includes the following steps.

[0158] The second NTT module 711 includes NTT module #1, NTT module #2, NTT module #3, and NTT module #4. I ¹ _x , Q ¹ _x , I ¹ _y and Q ¹ _y are input to NTT module #1, NTT module #2, NTT module #3 and NTT module #4, respectively.

[0159] NTT module #1, NTT module #2, NTT module #3 and NTT module #4 perform NTT processing for I ¹ _x , Q ¹ _x , I ¹ _y and Q ¹ _y , respectively, and output the fourth real-valued signals Ix_4' , Qx_4', Iy_4' and Qy_4' to the second chromatic dispersion compensation module 712.

[0160] The second chromatic dispersion compensation unit 712 includes a chromatic dispersion compensation unit #1 and a chromatic dispersion compensation unit #2. Ix_4' and Qx_4' are input to chromatic dispersion compensation unit #1, and chromatic dispersion compensation is performed in the transform domain to obtain fifth real-valued signals Ix_4'' and Qx_4'' in the transform domain. Iy_4' and Qy_4' are input to chromatic dispersion compensation unit #2, and chromatic dispersion compensation is performed in the transform domain to obtain fifth real-valued signals Iy_4'' and Qy_4'' in the transform domain.

[0161] The second INTT module 713 includes an INTT module #1, an INTT module #2, an INTT module #3, and an INTT module #4. Ix_4'', Qx_4'', Iy_4'' and Qy_4'' are input to INTT module #1, INTT module #2, INTT module #3 and INTT module #4, respectively.

[0162] INTT module #1, INTT module #2, INTT module #3 and INTT module #4 perform INTT processing for Ix_4'', Qx_4'', Iy_4'' and Qy_4'', respectively, and output the ninth real-valued signals Ix_4 ''', Qx_4''', Iy_4''' and Qy_4''' time domain to the nonlinear compensation module 714.

[0163] The nonlinear compensation unit 714 includes a nonlinear compensation unit #1 and a nonlinear compensation unit #2. Ix_4' and Qx_4' are input to nonlinear compensation module #1, and nonlinear compensation is performed to obtain Ix_4 and Qx_4. Iy_4' and Qy_4' are input to nonlinear compensation module #2, and nonlinear compensation is performed to obtain Iy_4 and Qy_4.

[0164] It should be noted that the above processing procedure in the DBP module needs to be repeated N times. To ensure performance as the communication span increases, the number N of DBP iterations must also be increased. The various DBP solutions are not limited here. Depending on the different power consumption required by scenarios, N can be set to the total link length, or single-hop DBP can be set. In the DBP module in this embodiment of the present invention, neither the FFT module nor the IFFT module is used; instead, the NTT module and the INTT module are used. Thus, during the signal processing procedure of the DBP module, repeated N transformations do not cause error accumulation.

[0165] FIG. 8(a) is a schematic block diagram of a first polarization compensation unit 214 according to an embodiment of the present invention. Fig. 8(b) is a schematic block diagram of a second polarization compensation unit 310 according to an embodiment of the present invention.

[0166] As can be understood from FIG. 8(a) it can be seen that the first polarization compensation module 214 in this embodiment of the present invention includes a butterfly filter 810 and a butterfly filter 820. The butterfly filter 810 includes a filter 801, a filter 802, a filter 803, a filter 804, an adder 805, and an adder 806. A butterfly filter 820 includes a filter 807, a filter 808, a filter 809, a filter 8010, an adder 8011, and an adder 8012.

[0167] The output of filter 801, the output of filter 807, the output of filter 803, and the output of filter 809 are negative. The above four signals are input to the adder 805 to obtain the signal Ix_5, which has undergone chromatic dispersion compensation. The output signal of the filter 801, the output signal of the filter 807, the output signal of the filter 803 and the output signal of the filter 809 are input to the adder 805 to obtain the signal Qx_5, which has undergone chromatic dispersion compensation. The output of filter 802, the output of filter 808, the output of filter 804, and the output of filter 8010 are negative. The above four signals are input to the adder 806 to obtain the signal Iy_5, which has undergone chromatic dispersion compensation. The output signal of the filter 802, the output signal of the filter 808, the output signal of the filter 804 and the output signal of the filter 8010 are output to the adder 805 to obtain the signal Qy_5, which has undergone chromatic dispersion compensation.

[0168] The clock recovery module outputs four signals. Four signals Ix_2, Qx_2, Iy_2 and Qy_2 are input into the polarization compensation module. The polarization compensation module includes two butterfly filters. One butterfly filter is a 2x2 system with two inputs and two outputs. Two complex-valued signals are input, and two complex-valued signals are output. Since the NTT transform processes real-valued signals, the polarization equalization part is a 4x4 system with four inputs and four outputs.

[0169] The first butterfly filter 810 is for equalizing and depolarizing the I-path signals (including the x-polarized I-path signal Ix_2 and the y-polarized I-path signal Iy_2), and the transform matrix is . The second butterfly filter 820 is for Q-path signals (including x-polarized Q-path signal Qx_2 and y-polarized Q-path signal Qy_2), and the transform matrix is .

[0170] The polarization compensation module provides four real-valued signals: Ix_5, Qx_5, Iy_5 and Qy_5. The process of performing polarization compensation by the first polarization compensation unit 214 is described with reference to FIG. 8(a).

[0171] The input signals of the first polarization compensation unit 214 are four real-valued signals: Ix_2, Qx_2, Iy_2, and Qy_2, which are a codirectional signal Ix_2 in the x polarization direction, a quadrature signal Qx_2 in the x polarization direction, a codirectional signal Iy_2 in the y polarization direction, and quadrature signal Qy_2 in the polarization direction y, respectively.

[0172] Specifically, the result of multiplying Ix_2 and is entered into the adder 805, the negative value of the result of the multiplication Qx_2 and is entered into the adder 805, the result of multiplying Iy_2 and is input into the adder 805, the negative value of the multiplication result Qy_2 and is input to the adder 805, and the adder 805 outputs a co-directional signal Ix_5, which is in the x polarization direction and which is obtained by polarization compensation.

[0173] Result of multiplying Ix_2 and is entered into the adder 8011, the negative value of the result of the multiplication Qx_2 and is entered into the adder 8011, the result of multiplying Iy_2 and is entered into the adder 8011, the result of multiplying Qy_2 and is input to the adder 8011, and the adder 8011 outputs a quadrature signal Qx_5, which is in the x polarization direction and which is obtained by polarization compensation.

[0174] Result of multiplying Ix_2 and is input into the adder 806, the negative value of the result of the multiplication Qx_2 and is entered into the adder 806, the result of multiplying Iy_2 and is input into the adder 806, the negative value of the result of the multiplication Qy_2 and is input to the adder 806, and the adder 806 outputs a co-directional signal Iy_5, which is in the y polarization direction and which is obtained by polarization compensation.

[0175] Result of multiplying Ix_2 and is entered into the adder 8012, the result of multiplying Qx_2 and is entered into the adder 8012, the result of multiplying Iy_2 and is entered into the adder 8012, the result of multiplying Qy_2 and is entered into the adder. 8012, and the adder 8012 outputs a quadrature signal Qy_5, which is in the y polarization direction and which is obtained by polarization compensation.

[0176] The process in which the output signals Ix_5, Qx_5, Iy_5 and Qy_5 of the first polarization compensation unit 214 interact with the equalization matrix and the input signals Ix_2, Qx_2, Iy_2 and Qy_2 is described by the following formula:

[0177] For example, the first polarization compensation unit 214 further includes a coefficient updating unit 2141 (not shown in the figure). The coefficient update unit 2141 is configured to update the coefficient of the filter included in the first polarization compensation unit 214. In this embodiment of the present invention, methods for updating alignment coefficient matrices And Polarization compensation modules include blind updating, learning updating, updating without decision feedback, updating with decision feedback, and the like. The method for updating the polarization compensation module coefficient is not limited in this embodiment of the present invention. The details are not described here.

[0178] As can be understood from FIG. 8(b), the second polarization compensation module 310 in this embodiment of the present invention includes 2×2 butterfly filters, and the butterfly filters include a filter 801, a filter 802, a filter 803, a fourth filter 804, a summerer 805, and an adder 806.

[0179] Functions of the second polarization compensation module 310 include optical fiber channel effects such as equalization compensation, differential group delay (DGD), and residual chromatic dispersion. The effect of an optical fiber can be described using a 2x2 distortion matrix. The inverse of the distortion matrix can be obtained using various algorithms. The time domain output coefficients of the butterfly filter 108 are hxx, hxy, hyx and hyy. The X and Y signals are convolved using the tap coefficient matrices hxx, hxy, hyx and hyy to obtain the Xout and Yout signals, which have undergone polarization distortion compensation. Common algorithms for calculating the tap coefficient matrix include the constant modulus algorithm (CMA), Least Mean Square (LMS), and various modified versions of LMS or other forms such as data-driven LMS.

[0180] The combining unit 216 outputs a complex-valued signal X in the X polarization direction, and the complex-valued signal X is input to the filter 801 and the filter 803. The combining unit 216 outputs the complex-valued signal Y in the Y polarization direction, and the complex-valued signal Y is input to the filter 802 and the filter 804 Filter 801 and filter 803 provide output to adder 805. Filter 802 and filter 804 provide output to adder 806.

[0181] hxx, hxy, hyx and hyy in FIG. 8(b) represent the coefficients of filter 801, filter 802, filter 803, and filter 804, respectively.

[0182] It should be noted that although FIG. 8(a) and Fig. 8(b) are used as examples to describe the butterfly filters in the present invention, the butterfly filters shown in FIG. 8(a) and Fig. 8(b) are merely examples and do not limit the scope of protection of the present invention. Butterfly filters in embodiments of the present invention are to be understood in a broad sense, and a filter or combination of filters that can implement a deconvolution function may be referred to as a butterfly filter.

[0183] The Rx DSP provided in this embodiment of the present invention can be further applied to a multimode transmission scenario. For example, Rx DSP applies to m transmission modes. In this case, the Rx DSP accepts P real-valued signals.

[0184] FIG. 9(a) is a schematic diagram of an Rx DSP architecture in a multimode transmission scenario according to an embodiment of the present invention. The 900 Rx DSP architecture includes:

a first NTT unit 211, a first chromatic dispersion compensation unit 212, a clock recovery unit 213, a first polarization compensation unit 214, a first INTT unit 215, a combining unit 216, a phase recovery unit 217, and a decoding unit 218.

[0185] P real-valued signals are input to the first NTT module 211. The first NTT module separately performs NTT processing on the P real-valued signals and outputs the P first real-valued signals in the conversion region to the first chromatic dispersion compensation module 212 .

[0186] Optionally, the P real-valued signals are real-valued signals obtained by ADC conversion. The P real-valued signals include 4×m real-valued signals in m transmission modes.

[0187] The first chromatic dispersion compensation unit 212 separately performs chromatic dispersion compensation for the P first real-valued signals in the conversion area and outputs the P third real-valued signals in the conversion area to the clock recovery unit 213 .

[0188] The clock recovery unit 213 separately performs clock recovery for the P third real-valued signals in the conversion domain and outputs the P second real-valued signals in the conversion domain to the first polarization compensation unit 214 .

[0189] The first polarization compensation unit 214 performs polarization compensation processing for the P second real-valued signals in the transform domain and outputs the P sixth real-valued signals in the transform domain to the first INTT unit 215 .

[0190] The first INTT unit 215 separately performs INTT processing on the P sixth real-valued signals in the transform domain and outputs the P seventh real-valued signals in the time domain to the combining unit 216 .

[0191] The combining unit 216 combines P seventh real-valued signals in the time domain and outputs m complex-valued signals X and m complex-valued signals Y to the phase recovery unit 217 and the decoding unit 218.

[0192] The phase restoration unit 217 and the decoding unit 218 perform phase restoration and decoding processing for m complex-valued signals X and m complex-valued signals Y to obtain reconstructed bit signals.

[0193] FIG. 9(b) is a schematic diagram of an Rx DSP architecture in another multimode transmission scenario according to an embodiment of the present invention. The 910 Rx DSP architecture includes:

a first NTT module 211, a first chromatic dispersion compensation module 212, a clock recovery module 213, a second polarization compensation module 108, a first INTT module 215, a combining module 216, a phase recovery module 320, and a decoding module 330.

[0194] The pre-clock recovery modules in architecture 910 are the same as those in architecture 900. The difference is that the output of the clock recovery module is first processed by the INTT module and combined by the combiner, and then input to the second polarization compensation module 108 . The process until the clock frequency is restored is again not described.

[0195] The first INTT module 215 separately performs INTT processing on the P second real-valued signals in the transform domain and outputs the P eighth real-valued signals in the time domain to the combining module 216 . The combining unit 216 combines the P eighth real-valued signals in the time domain and outputs the m complex-valued signals X and the m complex-valued signals Y to the second polarization compensation unit 310 .

[0196] The second polarization compensation unit 310 performs polarization compensation processing for m complex-valued signals X and m complex-valued signals Y, and outputs m complex-valued time-domain signals X and m complex-valued time-domain signals Y obtained by polarization compensation to the phase recovery unit 320 and a decoding module 330. The second polarization compensation unit 310 further includes a coefficient updating unit 311 (not shown in the drawing). The coefficient updating unit 311 is configured to update the coefficient of the filter included in the second polarization compensation unit 310.

[0197] The phase restoration unit 320 and the decoding unit 330 perform phase restoration and decoding processing for m complex-valued signals X and m complex-valued signals Y to obtain reconstructed bit signals.

[0198] FIG. 9(c) is a schematic diagram of an Rx DSP architecture in yet another multimode transmission scenario according to an embodiment of the present invention. The 920 Rx DSP architecture includes:

a DBP module 510, a first NTT module 211, a clock recovery module 213, a first polarization compensation module 214, a first INTT module 215, a combining module 216, a phase recovery module 217 and a decoding module 218.

[0199] The architecture after the first polarization compensation module 214 in the 920 Rx DSP architecture is the same as in the 900 Rx DSP architecture. The difference is that the signal undergoes DBP processing before polarization compensation. Since the DBP module 510 has a chromatic dispersion compensation function, the process after the DBP module may not require the chromatic dispersion compensation module 212.

[0200] P real-valued signals are input to the DBP module 510, DBP processing is performed, and P tenth real-valued time domain signals are output to the first NTT module 211.

[0201] The first NTT unit 211 separately performs NTT processing on the tenth real-valued signals in the time domain and outputs P first real-valued signals in the conversion domain to the clock recovery unit 213 .

[0202] The clock recovery unit 213 separately performs clock recovery processing for the P first real-valued signals in the conversion domain and outputs the P second real-valued signals in the conversion domain to the first polarization compensation unit 214 .

[0203] For the process after the clock recovery unit 213, see the embodiment shown in FIG. 9(a).

[0204] FIG. 9(d) is a schematic diagram of an Rx DSP architecture in another multimode transmission scenario according to an embodiment of the present invention. The 930 Rx DSP architecture includes:

a DBP module 510, a first NTT module 211, a clock recovery module 213, a second polarization compensation module 310, a first INTT module 215, a combining module 216, a phase recovery module 320, and a decoding module 330.

[0205] The architecture after the second polarization compensation module 310 in the Rx DSP architecture 930 is the same as in the Rx DSP architecture 910. The architecture in front of the second polarization compensation module 310 is the same as the 920 Rx DSP architecture. More specifically, the received signal obtained by ADC conversion is first subjected to DBP processing. For the process before DBP processing and clock recovery, see the embodiment shown in FIG. 9(c). For the process after the clock recovery unit 213, see the embodiment shown in FIG. 9(b).

[0206] For example, the DBP module 510 shown in FIG. 9(c) and Fig. 9(d) is shown in FIG. 9(e). Fig. 9(e) is a schematic block diagram of another DBP module according to an embodiment of the present invention. The 940 DBP module includes:

a second NTT module 711, a second chromatic dispersion compensation module 712, a second INTT module 713, and a nonlinear compensation module 714.

[0207] That P real-valued signals are input to the DBP module 510, and DBP processing is performed to obtain P tenth real-valued signals in the time domain, specifically includes the following steps.

[0208] P real-valued signals are respectively input to P second NTT modules 711, NTT processing is performed, and P fourth real-valued signals in the conversion domain are output to second chromatic dispersion compensation module 712.

[0209] The second chromatic dispersion compensation unit 712 performs chromatic dispersion compensation processing for the P fourth real-valued signals in the transform domain and outputs the P fifth real-valued signals in the transform domain to the second INTT unit 713 .

[0210] The second INTT module 713 performs INTT processing on the P fifth real-valued signals in the transform domain and outputs the P ninth real-valued signals in the time domain to the nonlinear compensation module 714 .

[0211] Nonlinear compensation unit 714 performs nonlinear compensation for P ninth real-valued time domain signals to obtain P tenth real-valued time domain signals.

[0212] In particular, compared with performing signal processing using an FFT module and an IFFT module, performing signal processing using an NTT module and an INTT module has the following advantages.

[0213] (1) When signal processing is performed based on the NTT module and the INTT module, the complexity of the Rx DSP architecture is reduced.

[0214] When signal processing is performed based on the FFT module and the IFFT module, during the signal transmission process, the power consumption of chromatic dispersion compensation alone is half of the total power consumption of the Rx DSP, and the complexity of the N-point FFT is proportional to Nlog2N multiplications. Depending on the different bit depth of these operations, the complexity of multiplication is from several to tens of times greater than the complexity of addition.

[0215] However, the complexity of signal processing performed based on the NTT module and the INTT module is proportional to 2Nlog2N additions. Depending on the different data widths and the different number N of conversion points, the complexity of signal processing performed based on the NTT module and the INTT module is reduced to different degrees. For example, if the number of conversion points is 256, NTT's power consumption is reduced to about 1/4 that of FFT's.

[0216] (2) When signal processing is performed based on the NTT module and the INTT module, the transformation matrix does not have a truncation error or a fixed-point calculation error.

[0217] When signal processing is performed based on the FFT module and the IFFT module, since the FFT transformation kernel is an exponential function, the truncation error cannot be accurately expressed on the binary computer. In addition, the width of the transformation matrix is not fixed, which leads to a fixed-point calculation error.

[0218] However, in the signal processing performed based on the NTT module and the INTT module, since the NTT conversion kernel is an integer, and the conversion kernel is usually 2 or a power of 2, the conversion kernel can be accurately expressed in memory. In addition, the bit width of the transformation matrix is fixed, without the error of fixed-point calculations.

[0219] (3) When signal processing is performed based on the NTT module and the INTT module, errors are not accumulated.

[0220] A truncation error and a fixed-point calculation error in a single FFT or IFFT transform are accumulated multiple times during the DBP propagation calculation. As a result, the error in the final result is larger.

[0221] However, there is no truncation error or fixed-point error in a single NTT or INTT transformation. Therefore, repeated N transformations do not cause error accumulation.

[0222] (4) When signal processing is performed based on the NTT module and the INTT module, the memory size can be reduced.

[0223] The NTT transformation kernel is 2, and the transformation matrix includes 2 or a power of two. In the binary system, a number is multiplied by a power of 2 simply by shifting it. Therefore, the DSP process does not need to store a huge transformation matrix, only the transformation kernel needs to be stored.

[0224] FIG. 10(a) is a schematic block diagram of a signal processing method according to an embodiment of the present invention. The following steps are included: S1010 to S1050.

[0225] S1010. P real-valued signals are obtained.

[0226] For example, the P real-valued signals include real-valued signals that have two polarization directions and correspond to each of the m modes. When m is 1, it indicates single-mode transmission; or when m is greater than 1, it indicates multimode transmission. m and P are natural numbers.

[0227] S1020. The P first real-valued signals in the transformation domain are obtained.

[0228] At least NTT processing is performed on P real-valued signals to obtain P first real-valued signals in the transform domain. The following two possible implementations are included.

[0229] Implementation 1.1: NTT processing is performed on P real-valued signals to obtain P first real-valued signals in the transform domain.

[0230] Implementation 1.2: DBP and NTT processing are sequentially performed on P real-valued signals to obtain P first real-valued signals in the transform domain.

[0231] In embodiment 1.2, DBP processing is first performed on P real-valued signals to obtain P tenths of real-valued signals in the time domain, and then NTT processing is performed on P tenths of real-valued time domain signals to obtain P first real-valued signals in the transform domain.

[0232] Performing NTT processing on P real-valued signals in Embodiment 1.1 and performing NTT processing on P tenths real-valued signals in the time domain that are obtained by performing DBP processing on P real-valued signals in Embodiment 1.2 may collectively be referred to as performing NTT processing on P input signals. In other words, in this embodiment of the present invention, the P input signals include P real-valued signals or P tenths of real-valued signals in the time domain, which are obtained by performing DBP processing on the P real-valued signals.

[0233] For ease of understanding, the process of performing DBP processing on P real-valued signals to obtain P tenths real-valued signals in the time domain is briefly described with reference to FIG. 10(b). Fig. 10(b) is a schematic block diagram of DBP processing according to an embodiment of the present invention. The following steps are included: S1021 to S1024.

[0234] S1021. P fourth real-valued signals in the transform domain are obtained.

[0235] NTT processing is performed on P real-valued signals to obtain P fourth real-valued signals in the transform domain.

[0236] S1022. P fifth real-valued signals in the transform domain are obtained.

[0237] Chromatic dispersion compensation is performed in the transform domain for P fourth real-valued signals in the transform domain to obtain P fifth real-valued signals in the transform domain.

[0238] S1023. P ninth real-valued signals in the time domain are obtained.

[0239] INTT processing is performed on P fifth real-valued signals in the transform domain to obtain P ninth real-valued signals in the time domain.

[0240] S1024. P tenths real-valued signals are obtained in the time domain.

[0241] Nonlinear compensation is performed on P ninth real-valued time domain signals to obtain P tenth real-valued time domain signals.

[0242] S1030. P second real-valued signals in the transform domain are obtained.

[0243] At a minimum, clock recovery processing is performed on the P first real-valued signals in the transform domain to obtain P second real-valued signals in the transform domain. According to the two possibilities, the following two possible implementations are included in step S1020.

[0244] Implementation 2.1: This corresponds to implementation 1.1 in step S1020. The processing procedure includes the following steps.

[0245] First, chromatic dispersion compensation is performed on the P first real-valued signals in the transform domain to obtain P third real-valued signals in the transform domain.

[0246] Clock recovery is then performed separately for the P third real-valued signals in the transform domain to obtain P second real-valued signals in the transform domain.

[0247] Performing chromatic dispersion compensation on the P first real-valued signals in the transform domain to obtain P third real-valued signals in the transform domain specifically includes the following steps:

[0248] First, the chromatic dispersion impulse response I _h (t) corresponding to the in-phase real-valued signal and the chromatic dispersion impulse response Q _h (t) corresponding to the quadrature real-valued signal are determined, and NTT is performed to obtain the chromatic dispersion equalization function I _h (w) in the transformation region corresponding to the in-phase real-valued signal, and the chromatic dispersion equalization function Q _h (w) in the transformation region corresponding to the quadrature real-valued signal.

[0249] Then, 2×m third real-valued signals in the transform domain that are in P third real-valued signals in the transform domain and that are in the first polarization direction are determined based on I _h (w), Q _h (w) and the in-phase real signal and a quadrature real-valued signal that are in the P first real-valued signals in the transform domain and that are in the first polarization direction, and define 2×m third real-valued signals in the transform domain that are in the P third real-valued signals in the transform domain and that are in the second polarization direction, based on I _h (w), Q _h (w) and the in-phase real-valued signal and the quadrature real-valued signal, which are in the P first real-valued signals in the transformation region and which are in the second polarization direction.

[0250] Specifically, the third real-valued transform domain signals obtained by chromatic dispersion compensation and the first real-valued transform domain signals that have not undergone chromatic dispersion compensation satisfy the following requirements:

I ^' _x (w) = I _x (w)⋅I _h (w)-Q _x (w)⋅Q _h (w);

Q ^' _x (w) = Q _x (w)⋅I _h (w)+I _x (w)⋅Q _h (w);

I ^' _y (w) = I _y (w)⋅I _h (w)-Q _y (w)⋅Q _h (w); And

Q ^' _y (w) = Q _y (w)⋅I _h (w)+I _y (w)⋅Q _h (w), where

I _x (w) represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the first polarization direction, Q _x (w) represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and that is in first polarization direction, I _y (w) represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the second polarization direction, Q _y (w) represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and which is in the second polarization direction, I ^' _x (w) represents an in-phase real-valued signal that is in the third real-valued signals in the transform region and which is in the first polarization direction, Q ^' _x (w) represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and which is in the first polarization direction, I ^' _y (w) represents the in-phase real-valued signal that is in the third real-valued signals in the transform domain and which is in the second polarization direction, and Q ^' _y (w) represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and that is in the second polarization direction.

[0251] Implementation Option 2.2:

[0252] This corresponds to implementation 1.2 in step S1020. The processing procedure includes the following stage:

[0253] Separately, clock recovery is performed for the P first real-valued signals in the conversion domain to obtain P second real-valued signals in the conversion domain.

[0254] S1040. Complex-valued signals are obtained in the time domain.

[0255] At least polarization compensation processing and INTT processing are performed on P second real-valued signals in the transform domain to obtain m complex-valued time-domain signals X and m complex-valued time-domain Y signals. The following two possible implementations are included.

[0256] Embodiment 3.1: Polarization compensation processing, INTT processing, and combining processing are sequentially performed on P second real-valued signals in the transform domain to obtain m complex-valued time-domain signals X and m complex-valued time-domain Y signals. Fig. 10(c) is a schematic block diagram of obtaining a complex valued time domain signal according to an embodiment of the present invention. The following steps are included: S1041 to S1043.

[0257] S1041. P sixth real-valued signals in the transform domain are obtained.

[0258] Polarization compensation is performed on P second real-valued signals in the transform domain to obtain P sixth real-valued signals in the transform domain.

[0259] S1042. P seventh real-valued signals in the time domain are obtained.

[0260] INTT processing is performed separately on the P sixth real-valued signals in the transform domain to obtain P seventh real-valued signals in the time domain.

[0261] S1043. Concatenation is performed to obtain m complex-valued signals X and m complex-valued signals Y.

[0262] Each two of the 2×m seventh real-valued time domain signals that are in the P seventh real-valued time domain signals and that are in the first polarization direction are combined to obtain m complex-valued time domain signals X in the first polarization direction.

[0263] Each two of the 2×m seventh real-valued time domain signals that are in the P seventh real-valued time domain signals and that are in the second polarization direction are combined to obtain m complex-valued Y time domain signals in the second polarization direction.

[0264] Embodiment 3.2: INTT processing, combining processing, and polarization compensation processing are sequentially performed for P second real-valued signals in the transform domain. Fig. 10(d) is another schematic block diagram of obtaining a complex valued time domain signal according to an embodiment of the present invention. The following steps are included: S1044 to S1046.

[0265] S1044. P eighth real-valued signals in the time domain are obtained.

[0266] INTT processing is performed on the P second real-valued signals in the transform domain to obtain P eighth real-valued signals in the time domain.

[0267] S1045. m complex-valued signals X and m complex-valued signals Y are obtained.

[0268] Each two of the 2×m eighth real time domain signals that are in the P eighth real time domain signals and that are in the first polarization direction are combined to obtain m complex valued time domain signals X in the first polarization direction.

[0269] Each two of the 2×m eighth real-valued time domain signals that are in the P eighth real-valued time domain signals and that are in the second polarization direction are combined to obtain m complex-valued Y time domain signals in the second polarization direction.

[0270] S1046. m complex-valued signals X and m complex-valued signals Y, which have undergone polarization compensation, are obtained.

[0271] Polarization compensation is performed on m complex time domain signals X in the first polarization direction and m complex time domain Y signals in the second polarization direction to obtain m complex time domain signals X and m complex time domain Y signals that have been subjected to polarization compensation.

[0272] S1050. Receive bit signals.

[0273] To obtain the reconstructed bit signals, it is necessary to perform phase recovery and decoding on the complex-valued time domain signals obtained in step S1040.

[0274] First, phase recovery is performed on m complex time domain signals X and m complex valued time domain signals Y to obtain m complex valued time domain signals X and m complex valued time domain signals Y that have undergone phase recovery.

[0275] Then, the m complex-valued time domain signals X and the m complex-valued time domain signals Y, which are obtained by phase recovery, are decoded to obtain the m complex-valued time domain signals X and the m complex-valued time domain signals Y that have been decoded.

[0276] It should be noted that the method of performing phase recovery and decoding of complex-valued time domain signals is not limited in this embodiment of the present invention. See the current process of performing phase recovery and decoding of an electrical signal by a coherent polarization multiplexing receiver.

[0277] An embodiment of the present invention further provides a coherent receiver including a polarization beam splitter, a frequency mixer, a photoelectric detector, an analog-to-digital converter, and an Rx DSP shown in FIG. 9(a),Fig. 9(b),Fig. 9(c) or FIG. 9(d). The polarization beam splitter, frequency mixer, photoelectric detector, and analog-to-digital converter are similar to those shown in FIG. 1(a) and Fig. 1(b). The details are not described here again.

[0278] An embodiment of the present invention further provides a chip. Fig. 11 is a schematic diagram of a chip 1100 in accordance with the present invention. The chip 1100 includes a processor 1110, a memory 1120, and a communications interface 1130. Processor 1110 is coupled to memory 1120. Memory 1120 is configured to store a computer program or instruction and/or data. The processor 1110 is configured to execute a computer program or instruction and/or data stored in the memory 1120 so that the method according to the above method embodiment is performed.

[0279] In an exemplary implementation, the chip shown in FIG. 11 may be a signal processing device including a processor 1110 and a communications interface 1130. The processor 1110 is coupled to the memory using a communication interface 1130, and the processor 1110 is configured to perform the method according to the above method embodiment.

[0280] It should be understood that the above embodiments are merely examples to describe the signal processing procedure provided in the present invention and do not impose any limitations on the scope of protection of the present invention. All other simple variations of the Rx DSP architecture fall within the scope of protection of the present invention. The difference between the Rx DSP architecture in the present invention and the existing Rx DSP architecture is that the FFT module and the IFFT module are not used for signal processing, but the NTT module and the INTT module are used.

[0281] It should also be understood that in embodiments of the present invention, unless otherwise indicated or in the event of a logical conflict, terms and/or descriptions in different embodiments may be consistent and may mutually refer to each other. Specifications in different embodiments may be combined based on the internal logical relationship of the specifications to form a new embodiment.

[0282] One skilled in the art can clearly understand that for the purpose of conveniently and concisely describing the detailed operating process of the above-mentioned system, apparatus and unit, reference should be made to the corresponding process in the above-mentioned method embodiments, and details will again not be given here.

[0283] When functions are implemented in the form of a software function block and sold or used as an independent product, the functions may be stored on a computer-readable storage medium. Based on this understanding, the essence of the technical solutions in the present invention, or a certain part contributing to the prior art, or some technical solutions can be implemented in the form of a software product. The software product is stored in a storage medium and includes a number of instructions for directing a computing device (which may be a personal computer, a server, or a network device) to perform all or some of the steps of the methods described in the embodiments of the present invention. The above storage media includes: any media that can store program code, such as a USB flash drive, a removable hard drive, a Read-Only Memory (ROM), a Random Access Memory (RAM), magnetic disk or optical disk.

Claims (98)

1. A method for processing a signal, the method comprising the steps of: receive P real-valued signals; performing at least number theoretic transform (NTT) processing on the P real-valued signals to obtain P first real-valued signals in the transform domain; performing at least clock recovery for the P first real-valued signals in the conversion region to obtain P second real-valued signals in the conversion region; performing at least polarization compensation and inverse number theoretic transform (INTT) processing on P second real-valued signals in the transform domain to obtain m complex-valued time-domain signals X and m complex-valued time-domain signals Y, wherein performing at least polarization compensation and Inverse number theoretic transform (INTT) processing for P second real-valued signals in the transform domain includes: sequentially performing polarization compensation processing and INTT processing for P second real-valued signals in the transform domain, or sequentially performing INTT processing and polarization compensation processing for the P second real-valued signals in the transform domain; And performing phase recovery and decoding of m complex valued time domain signals X and m complex valued time domain signals Y to obtain bit signals, wherein m and P are natural numbers, wherein polarization compensation includes polarization demultiplexing and distortion equalization. 2. The method according to claim 1, in which the P real-valued signals contain real-valued signals in m transmission modes, and when m is equal to 1, it indicates single-mode transmission, or when m is greater than 1, it indicates multi-mode transmission. 3. The method of claim 1 or 2, wherein performing at least clock recovery for P first real-valued signals in the transform domain to obtain P second real-valued signals in the transform domain comprises the steps of: performing chromatic dispersion compensation for P first real-valued signals in the transform region to obtain P third real-valued signals in the transform region; And performing clock recovery for the P third real-valued signals in the transform region to obtain P second real-valued signals in the transform region. 4. The method of claim 3, wherein the P real-valued signals comprise an in-phase real-valued signal and quadrature real-valued signal in the first direction of polarization, and the in-phase real-valued signal and quadrature real-valued signal in the second polarization direction; And performing chromatic dispersion compensation on P first real-valued signals in the transform region to obtain P third real-valued signals in the transform region comprises the steps of: perform NTT processing for impulse response chromatic dispersion corresponding to an in-phase real-valued signal, and for impulse response chromatic dispersion corresponding to the quadrature real-valued signal to obtain the function equalization of chromatic dispersion in the transformation region corresponding to the in-phase real-valued signal, and the function equalizing chromatic dispersion in the transformation region corresponding to the quadrature real-valued signal; determine third real-valued signals in the transform domain that are in P third real-valued signals in the transform domain and that are in the first polarization direction, based on , and in-phase real-valued signal and quadrature real-valued signal , which are in the P first real-valued signals in the transformation region and which are in the first polarization direction; And determine third real-valued signals in the transform domain that are in P third real-valued signals in the transform domain and that are in the second polarization direction, based on , and in-phase real-valued signal and quadrature real-valued signal , which are in the P first real-valued signals in the conversion domain and which are in the second polarization direction. 5. The method of claim 4, wherein the third real-valued signals in the transform domain obtained by chromatic dispersion compensation and the first real-valued signals in the transform domain satisfy the following requirements: wherein represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the first polarization direction, represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and that is in the first polarization direction, represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the second polarization direction, represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and that is in the second polarization direction, represents an in-phase real-valued signal that is in the third real-valued signals in the conversion domain and that is in the first polarization direction, represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and that is in the first polarization direction, represents an in-phase real-valued signal that is in the third real-valued signals in the conversion domain and that is in the second polarization direction, and represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and that is in the second polarization direction. 6. The method of claim 1, wherein performing at least number theoretic transform (NTT) processing on the P real-valued signals to obtain P first real-valued signals in the transform domain comprises the steps of: performing digital backpropagation (DBP) processing on the P real-valued signals to obtain P tenths of real-valued signals in the time domain; And separately performing NTT processing on the P tenth real-valued signals in the time domain to obtain P first real-valued signals in the transform domain. 7. The method of claim 6, wherein performing DBP processing on P real-valued signals to obtain P tenths real-valued signals in the time domain comprises the steps of: performing NTT processing on the P real-valued signals to obtain P fourth real-valued signals in the transform region; performing, in the transform region, chromatic dispersion compensation for P fourth real-valued signals in the transform region to obtain P fifth real-valued signals in the transform region; performing INTT processing on P fifth real-valued signals in the transform domain to obtain P ninth real-valued signals in the time domain; And performing nonlinear compensation on P ninth real-valued time domain signals to obtain P tenth real-valued time domain signals. 8. The method according to any one of claims 1 to 7, wherein performing at least polarization compensation and inverse number theoretic transform (INTT) processing on P second real-valued signals in the transform domain to obtain m complex-valued time-domain signals X and m complex-valued signals Y signals in the time domain contains stages in which: performing polarization compensation on P second real-valued signals in the transform region to obtain P sixth real-valued signals in the transform region; performing INTT processing on P sixth real-valued signals in the transform domain to obtain P seventh real-valued signals in the time domain; And combine every two of seventh real-valued time-domain signals that are in P seventh real-valued time-domain signals and that are in the first polarization direction to obtain m complex-valued time-domain signals X in the first polarization direction, and combine every two of seventh real-valued time domain signals that are in P seventh real-valued time domain signals and that are in a second polarization direction to obtain m complex-valued Y time domain signals in the second polarization direction. 9. The method of claim 8, wherein performing polarization compensation on P second real-valued signals in the transform domain to obtain P sixth real-valued signals in the transform domain comprises the steps of: perform equalization and depolarization to second real-valued signals in the transform domain that are in P second real-valued signals in the transform domain and that are in the first polarization direction to obtain sixth real-valued signals in the transformation region that are in P sixth real-valued signals in the transformation region and which are in the first polarization direction; And perform equalization and depolarization to second real-valued signals in the transform domain that are in P second real-valued signals in the transform domain and that are in the second polarization direction to obtain sixth real-valued signals in the transformation region that are in P sixth real-valued signals in the transformation region and which are in the second polarization direction. 10. The method according to any one of claims 1 to 7, wherein performing at least polarization compensation and inverse number theoretic transform (INTT) processing on P second real-valued signals in the transform domain to obtain m complex-valued time-domain signals X and m complex-valued signals Y signals in the time domain contains stages in which: performing INTT processing on P second real-valued signals in the transform domain to obtain P eighth real-valued signals in the time domain; combine every two of eighth real-valued time-domain signals that are in P eighth real-valued time-domain signals and that are in the first polarization direction to obtain m complex-valued time-domain signals X in the first polarization direction, and combine every two of eighth real-valued time domain signals that are in P eighth real-valued time domain signals and that are in a second polarization direction to obtain m complex-valued Y time domain signals in the second polarization direction; And perform time domain polarization compensation for m complex time domain signals X in the first polarization direction and m complex time domain signals Y in the second polarization direction to obtain m complex time domain signals X and m complex time domain signals Y that have been subjected to polarization compensation. 11. Signal processing device containing: a first number theoretic transform (NTT) module configured to perform NTT processing on the P input signals to obtain P first real-valued signals in the transform domain; a clock recovery unit configured to perform clock recovery for P first real-valued signals in the transformation region or P first real-valued signals in the transformation region obtained by chromatic dispersion compensation to obtain P second real-valued signals in the transformation region; a polarization compensation module and a first inverse number theoretic transform (INTT) module configured to process P second real-valued signals in the transform domain to obtain m complex-valued time-domain signals X and m complex-valued time-domain Y signals, the polarization compensation module and the first module inverse number theoretic transform (INTT) devices configured to process P second real-valued signals in the transform domain include: a polarization compensation module and a first inverse number theoretic transform (INTT) module configured to sequentially perform polarization compensation processing and INTT processing for P second real-valued signals in the transform domain, or a polarization compensation module and a first inverse number theoretic transform (INTT) module, configured to sequentially perform INTT processing and polarization compensation processing for P second real-valued signals in the transformation region; And a phase recovery module and a decoding module configured to process m complex valued time domain signals X and m complex valued time domain signals Y to obtain bit signals, wherein m and P are natural numbers, polarization compensation module includes butterfly filters and performs polarization demultiplexing and distortion equalization. 12. The device according to claim 11, in which P input signals comprise real-valued signals in m transmission modes, and when m is equal to 1, it indicates single-mode transmission, or when m is greater than 1, it indicates multi-mode transmission. 13. The device according to claim 11 or 12, wherein the device additionally contains: a first chromatic dispersion compensation module configured to perform chromatic dispersion compensation for P first real-valued signals in the transform domain to obtain P third real-valued signals in the transform domain, wherein the clock recovery module is configured to perform clock recovery for the third real-valued signal to obtain P second real-valued signals in the conversion region. 14. The device of claim 13, wherein the P input signals comprise an in-phase real-valued signal and quadrature real-valued signal in the first direction of polarization, and the in-phase real-valued signal and quadrature real-valued signal in the second polarization direction; and that the first chromatic dispersion compensation module is configured to perform chromatic dispersion compensation for P first real-valued signals in the transform domain to obtain P third real-valued signals in the transform domain, comprising the following: the first chromatic dispersion compensation module is configured to determine third real-valued signals in the transform domain that are in P third real-valued signals in the transform domain and that are in the first polarization direction, based on the function equalization of chromatic dispersion in the transformation region corresponding to the in-phase real-valued signal, function equalizing the chromatic dispersion in the transformation region corresponding to the quadrature real-valued signal and the in-phase real-valued signal and the quadrature real-valued signal that are in the P first real-valued signals in the transformation region and that are in the first polarization direction; And the first chromatic dispersion compensation module is configured to determine third real-valued signals in the transform domain that are in P third real-valued signals in the transform domain and that are in the second polarization direction, based on , and an in-phase real-valued signal and a quadrature real-valued signal that are in the P first real-valued signals in the transformation region and that are in the second polarization direction, wherein And are obtained by performing NTT processing on the impulse response chromatic dispersion corresponding to an in-phase real-valued signal, and for impulse response chromatic dispersion corresponding to the quadrature real-valued signal, respectively. 15. The apparatus of claim 14, wherein the third real-valued signals in the transform domain obtained by chromatic dispersion compensation and the first real-valued signals in the transform domain satisfy the following requirements: wherein represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the first polarization direction, represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and that is in the first polarization direction, represents an in-phase real-valued signal that is in the first real-valued signals in the transform domain and that is in the second polarization direction, represents a quadrature real-valued signal that is in the first real-valued signals in the transform domain and that is in the second polarization direction, represents an in-phase real-valued signal that is in the third real-valued signals in the conversion domain and that is in the first polarization direction, represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and that is in the first polarization direction, represents an in-phase real-valued signal that is in the third real-valued signals in the conversion domain and that is in the second polarization direction, and represents a quadrature real-valued signal that is in the third real-valued signals in the transform domain and that is in the second polarization direction. 16. The device according to any one of claims 13 to 15, wherein the first chromatic dispersion compensation module comprises a third NTT module, a combining module and a multiplication module; wherein the third NTT module is configured to perform NTT processing for a chromatic dispersion impulse response in a time domain to obtain a chromatic dispersion equalization function and perform NTT processing on the P first real-valued signals in the transform domain; wherein the multiplying unit is configured to multiply the chromatic dispersion equalization function by P first real-valued signals in the transform domain obtained by NTT processing; and the combining module is configured to combine the signals obtained by processing by the multiplying module. 17. The device of claim 16, wherein the P input signals comprise P tenths real-valued time domain signals; And the device additionally contains: a digital backpropagation (DBP) calculation module configured to perform DBP processing on P digital signals to obtain P tenths real-valued time domain signals. 18. The device according to claim 17, in which the DBP module sequentially contains: a second NTT module, a second chromatic dispersion compensation module, a second INTT module, and a nonlinear compensation module, wherein that the DBP module is configured to perform DBP processing on P digital signals to obtain P tenths real-valued signals in the time domain comprises the following: a second NTT module is configured to perform NTT processing on P digital signals to obtain P fourth real-valued signals in the conversion domain; a second chromatic dispersion compensation module is configured to perform, in the transform domain, chromatic dispersion compensation for P fourth real-valued signals in the transform domain to obtain P fifth real-valued signals in the transform domain; a second INTT module is configured to perform INTT processing on P fifth real-valued signals in the transform domain to obtain P ninth real-valued signals in the time domain; And the nonlinear compensation module is configured to perform nonlinear compensation for P ninth real-valued time domain signals to obtain P tenth real-valued time domain signals. 19. The device according to any one of claims 14-18, wherein the device additionally contains: merging module, while a polarization compensation module is configured to perform polarization compensation for P second real-valued signals in the conversion domain to obtain P sixth real-valued signals in the conversion domain; the first INTT module is configured to perform INTT processing on P sixth real-valued signals in the transform domain to obtain P seventh real-valued signals in the time domain; And the merge module is configured to combine every two of seventh real-valued time domain signals that are in P seventh real-valued time domain signals and that are in the first polarization direction to obtain m complex-valued time domain signals X in the first polarization direction; And combining every two of seventh real-valued time domain signals that are in P seventh real-valued time domain signals and that are in a second polarization direction to obtain m complex-valued Y time domain signals in the second polarization direction. 20. The apparatus of claim 19, wherein the polarization compensation module comprises: a first butterfly filter and a second butterfly filter; the first butterfly filter is configured to perform equalization and depolarization for second real-valued signals in the transform domain that are in P second real-valued signals in the transform domain and that are in the first polarization direction to obtain sixth real-valued signals in the transformation region that are in P sixth real-valued signals in the transformation region and which are in the first polarization direction; And a second butterfly filter is configured to perform equalization and depolarization for second real-valued signals in the transform domain that are in P second real-valued signals in the transform domain and that are in the second polarization direction to obtain sixth real-valued signals in the transformation region that are in P sixth real-valued signals in the transformation region and which are in the second polarization direction. 21. The device according to any one of claims 14-18, wherein the device additionally contains: merging module, while a first inverse number theoretic transform (INTT) module is configured to perform INTT processing on P second real-valued signals in the transform domain to obtain P eighth real-valued signals in the time domain; the merge module is configured to combine every two of eighth real-valued time domain signals that are in P eighth real-valued time domain signals and that are in the first polarization direction to obtain m complex-valued time domain signals X in the first polarization direction; And combining every two of eighth real-valued time domain signals that are in P eighth real-valued time domain signals and that are in a second polarization direction to obtain m complex-valued Y time domain signals in the second polarization direction; And the polarization compensation module is configured to perform polarization compensation in the time domain for the m complex-valued signals X and the m complex-valued signals Y. 22. The device according to claim 21, wherein the polarization compensation module comprises a third butterfly filter; And a third butterfly filter is configured to perform time-domain polarization compensation for m complex-valued time-domain signals X in a first polarization direction and m complex-valued time-domain Y signals in a second polarization direction to obtain m complex-valued time-domain signals X and m complex-valued signals Y in the time domain that have undergone polarization compensation. 23. Coherent receiver containing: the apparatus of any one of claims 11 to 22, a polarization beam splitter, a frequency mixer, a photoelectric detector, and an analog-to-digital converter, wherein the polarization beam splitter is configured to obtain signals in two polarization directions, the frequency mixer is configured to perform frequency mixing processing for signals with the same polarization direction among the signals in two polarization directions that are output by the polarization beam splitter, the photoelectric detector is configured to convert the strength of the optical signal output by the frequency mixer into that of an electrical signal, and the analog-to-digital converter is configured to perform analog-to-digital conversion for the signal output by the photoelectric detector. 24. A chip comprising a communication interface, a memory, and a processor, wherein the memory is configured to store a computer program, and the processor is configured to read and execute a computer program stored in the memory, such that the chip implements the method according to any one of claims 1 to 10. 25. A signal processing device containing: processor and communication interface, while the processor is coupled to the memory via a communication interface, and the processor is configured to execute program code in the memory to implement the method of any one of claims 1 to 10. 26. Computer-readable storage medium containing: the computer-readable storage medium stores a computer program; And when the computer program is executed on a computer, the computer is configured to perform the method of any one of claims. 1-10.

Images