Method and apparatus for compensating multipath crosstalk in an optical communication link

Abstract

The present invention provides a method and apparatus for compensating multipath crosstalk in an optical communication link. The method includes: an analog-to-digital converter samples the signal output by an analog front end, converts it into a digital signal, and then outputs it to a feedforward equalizer; the feedforward equalizer calculates a feedforward equalization error value by subtracting the target value used in the feedforward equalization process from the data after feedforward equalization; determines whether the feedforward equalizer converges. If it converges, the feedforward equalization error value is output to a filter, and the filter filters the feedforward equalization error value to obtain a multipath crosstalk noise amount; compensating calculation is performed using the multipath crosstalk noise amount, and the compensated data is output to a decision feedback equalizer; wherein, when the analog-to-digital converter samples the signal output by the analog front end, the signal output by the analog front end is decimated at a preset ratio in each clock cycle. The apparatus can implement the above method. The present invention can adapt to the rapid change of multipath crosstalk noise in an optical communication link and perform effective compensation processing.

Description

Technical Field

[0001] The present invention relates to the technical field of optical communication. Specifically, it is a method for compensating multi-path crosstalk in an optical communication link and a device for implementing this method. Background Art

[0002] With the development of 5G communication technology, artificial intelligence technology, and big data technology, higher requirements are put forward for the computing power of chips, and the demand for network bandwidth has also increased exponentially. Whether in data centers or telecommunications networks, the optical interconnection solutions between switches almost evolve one generation forward every three years. With the rapid improvement and iteration of the capacity of switching chips in data centers and telecommunications, the high-speed connections between all levels in the network architecture need to be upgraded accordingly. Due to the rapid evolution of data rates, more advanced modulation techniques are required, that is, the evolution from NRZ to PAM4 technology has been achieved currently. For 100G PAM4 direct modulation and direct detection optical fiber communication, the impact of multi-path interference (MPI) on signals has received increasing attention, and optical communication systems often need to compensate for the multi-path crosstalk situation.

[0003] See Figure 1 , a typical high-speed optical communication system uses the first optical module 11 and the second optical module 21 as carriers for signal transmission and reception. The host 10 realizes communication with the slave 30 through the first optical module 11, the optical fiber 20, and the second optical module 21. A first DSP chip 12 is provided in the first optical module 11, and a first transmit DSP 13 and a first receive DSP 14 are provided in the first DSP chip 12. A second DSP chip 22 is provided in the second optical module 21, and a second transmit DSP 23 and a second receive DSP 24 are provided in the second DSP chip 22. In the downlink, the signal sent by the host 20 passes through the first transmit DSP 13 and the first optoelectronic device 15 in sequence to reach the optical fiber 20, and then passes through the third optoelectronic device 25 and the second transmit DSP 23 to reach the slave 30. In the uplink, the signal sent by the slave 30 passes through the second receive DSP 24 and the fourth optoelectronic device 26 in sequence to reach the optical fiber 20, and passes through the second optoelectronic device 16 and the first receive DSP 14 to reach the host 10. Since the optical communication channel formed by each optoelectronic device and the optical fiber 20 is not an ideal channel, various impairments and noises will cause signal distortion or deterioration of the signal-to-noise ratio.

[0004] See Figure 2, in a simplified typical multi-joint optical fiber transmission channel for a data center and a telecommunication network with a distance ranging from 500 m to 40 km, there are multiple optical fiber joints between the end face 31 of the first optical module and the end face 35 of the second optical module. For example, it includes a first optical fiber joint 32, a second optical fiber joint 33, and a third optical fiber joint 34. Each optical fiber joint is connected by an optical fiber. Due to the reflection of each optical fiber joint, and with the occurrence of situations such as optical fiber aging and dirt on the optical fiber joints, the optical reflection will be more serious, multiple reflections will be formed in the optical fiber channel, and finally multi-path crosstalk noise will be formed, causing multi-path crosstalk to the received signal. Since there are no very strict regulations and requirements for the number of optical fiber joints, the end face reflectivity, and the length of the optical fiber segments, there will be various combinations to adapt to the actual application scenarios. The number of connectors will lead to the uncertainty of the noise variables, the increase in reflectivity will lead to the increase in noise energy, and the change in the optical fiber length will lead to the change in the crosstalk symbol delay. Therefore, for the DSP, there is no way to compensate and detect the multi-path crosstalk noise by finding the source position. When there are multiple large delays, there is no way to use the traditional feed-forward equalizer (FFE) scheme for tracking compensation in engineering applications.

[0005] After the multi-path crosstalk noise passes through the square detection at the receiving end, both high-frequency and low-frequency noises exist. However, due to the interference between the signal and the noise, the high-frequency noise weakens, and the influence of the multi-path crosstalk noise is finally mainly concentrated in the low frequency. Through the analysis of the experimental data of the application scenario, the frequency of the multi-path crosstalk noise is usually less than 100 MHz, and in severe cases, the noise-to-signal ratio will be greater than 5%. It is necessary to eliminate the low-frequency noise of the multi-path crosstalk noise through DSP algorithm processing.

[0006] See Figure 3 , an existing method for compensating multi-path crosstalk noise is that in the DSP at the receiving end, the analog-to-digital converter 42 is used to collect the data of the analog front end 41, and then the collected data is segmented and averaged or processed by sliding window averaging through the low-frequency noise compensator 43, which is approximately a low-pass filtering process, to obtain the low-frequency noise, and then the signal is subtracted from the low-frequency noise to achieve the compensation of the multi-path crosstalk noise. The compensated data is output to the feed-forward equalizer 44 and finally output to the decision feedback equalizer 45 for further equalization processing. However, the scheme of segmenting and averaging the data has problems of large tracking error and large jitter, while the scheme of sliding window averaging has problems of too high complexity and great difficulty in implementation.

[0007] On the other hand, the invention patent application with the publication number CN118554892A discloses a noise compensation method for a transimpedance amplifier. This method uses a filter to filter the signal, but this method is mainly applied to the noise compensation of the transimpedance amplifier. In the application scenario of an optical communication link, due to the fact that the change period of the multipath crosstalk noise is faster than that of the noise of the transimpedance amplifier, this method cannot meet the requirements of multipath crosstalk noise compensation in the optical communication link. Summary of the Invention

[0008] The first object of the present invention is to provide a compensation method for multipath crosstalk in an optical communication link that can meet the change period requirements of multipath crosstalk noise in the optical communication link.

[0009] The second object of the present invention is to provide a compensation device for multipath crosstalk that implements the above-mentioned compensation method for multipath crosstalk in an optical communication link.

[0010] To achieve the first object of the present invention, the compensation method for multipath crosstalk in the optical communication link provided by the present invention includes: an analog-to-digital converter collects the signal output by the analog front end, converts it into a digital signal, and then outputs it to the feed-forward equalizer; the feed-forward equalizer calculates the feed-forward equalization error value by subtracting the target value used in the feed-forward equalization process from the data after feed-forward equalization; determines whether the feed-forward equalizer converges. If it converges, the feed-forward equalization error value is output to the filter, and the filter filters the feed-forward equalization error value to obtain the multipath crosstalk noise amount; uses the multipath crosstalk noise amount for compensation calculation, and outputs the compensated data to the decision feedback equalizer; wherein, when the analog-to-digital converter collects the signal output by the analog front end, the signal output by the analog front end is sampled at a preset ratio in each clock cycle.

[0011] As can be seen from the above solution, in view of the characteristic that the change period of the multipath crosstalk noise in the optical communication link is relatively fast, the analog-to-digital converter of the present invention samples the signal of the analog front end by extracting a certain proportion of the signal in each clock cycle, so that the amount of data to be processed is greatly reduced. Moreover, since a certain amount of signal is extracted in each clock cycle, it is possible to ensure that a certain amount of signal is collected in each clock cycle, thereby ensuring the accuracy of multipath crosstalk noise processing.

[0012] A preferred solution is that sampling the signal output by the analog front end at a preset ratio in each clock cycle includes: sampling a preset number of signals output by the analog front end in each clock cycle.

[0013] Thus, since the number of signals sampled in each clock cycle is preset and can be adjusted according to the actual situation, it can meet the usage requirements in different usage scenarios.

[0014] A further solution is that when extracting signals output by the analog front end in a preset quantity in each clock cycle, the positions of the signals extracted in each clock cycle are determined in a polling manner.

[0015] It can be seen that determining the positions of the signals extracted in each clock cycle in a polling manner can avoid the problem of inaccurate calculation of the multipath crosstalk noise caused by extracting signals at fixed positions.

[0016] A further solution is that before calculating the feedforward equalization error value, a clock recovery operation is also performed. By performing the clock recovery operation, it can be ensured that the clock signal remains synchronized during the calculation of the feedforward equalization error value.

[0017] An optional solution is that the filter is a first-order digital low-pass filter; when using the multipath crosstalk noise quantity for compensation calculation, the multipath crosstalk noise quantity is used to compensate the data after feedforward equalization by the feedforward equalizer: the data after feedforward equalization by the feedforward equalizer in the current clock cycle minus the multipath crosstalk noise quantity is used to obtain the compensation data for the current clock cycle.

[0018] It can be seen that the backward compensation method can compensate for the multipath crosstalk noise in the current clock cycle, making the multipath crosstalk noise compensation have high timeliness.

[0019] An optional solution is that the filter is a second-order loop filter; when using the multipath crosstalk noise quantity for compensation calculation, the data after feedforward equalization by the feedforward equalizer in the next clock cycle minus the multipath crosstalk noise quantity calculated in the current cycle is used to obtain the compensation data for the next clock cycle.

[0020] It can be seen that the forward compensation method can compensate the data in the next clock cycle by applying the multipath crosstalk noise quantity in the current clock cycle.

[0021] A further solution is that after calculating the compensation data for the next clock cycle, the following is also performed: in the next clock cycle, calculate the feedforward equalization error value for this clock cycle.

[0022] It can be seen that the feedforward equalization error value for each clock cycle is calculated in each clock cycle, and the multipath crosstalk noise in this clock cycle is compensated.

[0023] To achieve the above second objective, the present invention provides a compensation device for multipath crosstalk in an optical communication link, including: an analog-to-digital converter for collecting the signal output by the analog front end and converting it into a digital signal; a feedforward equalizer for receiving the signal from the analog-to-digital converter and performing feedforward equalization processing, and calculating the feedforward equalization error value by subtracting the target value used in the feedforward equalization process from the data after feedforward equalization; a filter for receiving the feedforward equalization error value when the feedforward equalizer converges and filtering the feedforward equalization error value to obtain the multipath crosstalk noise amount; an adder for performing compensation calculation using the multipath crosstalk noise amount; a decision feedback equalizer for receiving the compensated data and performing equalization processing on the compensated data again; wherein, when the analog-to-digital converter collects the signal output by the analog front end, it extracts the signal output by the analog front end at a preset ratio in each clock cycle.

[0024] As can be seen from the above solution, since the optical communication link has the characteristic of a relatively fast change period of multipath crosstalk noise, the present invention extracts a certain proportion of signals in each clock cycle when the analog-to-digital converter collects the analog front end signal, so that the amount of data to be processed is greatly reduced. And, since a certain amount of signal is extracted in each clock cycle, it is possible to ensure that a certain amount of signal is collected in each clock cycle, thereby ensuring the accuracy of multipath crosstalk noise processing.

[0025] A preferred solution is that the sampling rate of the analog-to-digital converter is greater than or equal to the baud rate of the signal output by the analog front end.

[0026] Thus, the sampling rate of the analog-to-digital converter is higher than the baud rate of the signal output by the analog front end, ensuring that the analog-to-digital converter can accurately extract the signals in each clock cycle.

[0027] A further solution is that when the analog-to-digital converter extracts the signal output by the analog front end at a preset ratio in each clock cycle, it extracts a preset number of signals output by the analog front end in each clock cycle. Description of the Drawings

[0028] Figure 1 is a block diagram of the structure of an existing high-speed optical communication system.

[0029] Figure 2 is a block diagram of the structure of a multi-joint optical fiber transmission channel in an existing data center and telecommunication network.

[0030] Figure 3 is a schematic diagram of an existing method for compensating multipath crosstalk noise.

[0031] Figure 4 is a block diagram of the structure of the first embodiment of the compensation device for multipath crosstalk in the optical communication link of the present invention.

[0032] Figure 5 It is a flowchart of the first embodiment of the method for compensating multipath crosstalk in the optical communication link of the present invention.

[0033] Figure 6 It is a structural block diagram of the second embodiment of the apparatus for compensating multipath crosstalk in the optical communication link of the present invention.

[0034] Figure 7 It is a flowchart of the second embodiment of the method for compensating multipath crosstalk in the optical communication link of the present invention.

[0035] The present invention will be further described below in conjunction with the accompanying drawings and embodiments. Specific embodiments

[0036] The present invention is directed to the situation of multipath crosstalk noise existing in the optical communication link, calculating and compensating the multipath crosstalk noise, thereby reducing the signal distortion in the optical communication link and ensuring the signal transmission quality of the optical communication link.

[0037] The first embodiment:

[0038] Referring to Figure 4 , the apparatus for compensating multipath crosstalk in the optical communication link of this embodiment has a first analog-to-digital converter 51, a first feed-forward equalizer 52, a first-order digital low-pass filter 53, a first decision feedback equalizer 54, and a first adder 55. Among them, the first analog-to-digital converter 51 is used to collect the analog signal output by the analog front end and convert the collected analog signal into a digital signal. Since in the optical communication link, the signal change rate of the analog front end is very fast, in order to ensure accurate sampling of the data, the sampling rate of the first analog-to-digital converter 51 needs to be greater than or equal to the baud rate of the signal output by the analog front end. In addition, while converting the collected analog signal into a digital signal, the first analog-to-digital converter 51 will also convert the serial signal into parallel data for the DSP chip to process the parallel data.

[0039] The first feed-forward equalizer 52 is used to receive the digital signal output by the first analog-to-digital converter 51 and is used to compensate for the inter-symbol interference (ISI) damage of the received digital signal. Specifically, the first feed-forward equalizer 52 obtains a stable feed-forward equalization error through adaptive convergence calculation, denoted as FFE error. Among them, the process of feed-forward equalization is to perform feed-forward equalization calculation on the received data, such as performing convolution calculation on the input data and the feed-forward equalization coefficient. The feed-forward equalization coefficient used by the first feed-forward equalizer 52 in the feed-forward equalization calculation process is adaptively obtained by using the least mean square algorithm. After the feed-forward equalization coefficient converges and stabilizes, the mean square error value of the least mean square algorithm will tend to be stable.

[0040] The first-order digital low-pass filter 53 is a first-order filter with fixed coefficients, which is used to perform smoothing averaging on the input signal and the historical signal, so as to achieve the effect of low-pass filtering on the input signal. In this embodiment, by reasonably configuring the filter coefficients of the first-order digital low-pass filter 53, the bandwidth of the first-order digital low-pass filter 53 can be configured and adjusted to meet the usage requirements of different application scenarios. Since this embodiment is applied in an optical communication link and it is necessary to track and compensate for the multipath crosstalk noise, the bandwidth of the first-order digital low-pass filter 53 needs to be greater than 10 MHz.

[0041] The first adder 55 is used to receive the signals output by the first feed-forward equalizer 52 and the first-order digital low-pass filter 53, and perform addition and subtraction processing on the two signals. Specifically, the signal of the first feed-forward equalizer 52 is subtracted from the signal output by the first-order digital low-pass filter 53, and the calculation result is output to the first decision feedback equalizer 54.

[0042] The first decision feedback equalizer 54 is used to perform equalization processing on the signal after feed-forward processing again. According to the decision result of the signal output by the first feed-forward equalizer 52, the decision feedback equalization coefficient is adjusted, and decision feedback equalization processing is performed accordingly, so as to reduce the signal distortion. Specifically, after determining the current decision symbol, the first decision feedback equalizer 54 needs to adjust the coefficient of the decision feedback equalization according to the decision result of the previous decision symbol, and apply it to the current decision symbol, and then send the compensated current decision symbol to the decision device for decision, and output the decision result of the current decision symbol.

[0043] The following combines Figure 5 to introduce the specific process of the multipath crosstalk compensation method for the optical communication link in this embodiment. First, step S1 is executed. The first analog-to-digital converter 51 samples the analog signal output by the analog front end, performs analog-to-digital conversion on the analog signal to obtain a digital signal, and converts the serial signal into a parallel signal. When the first analog-to-digital converter 51 samples the signal, the signal is decimated at a certain ratio in each clock cycle. For example, in one clock cycle, the analog front end will output 64 or 32 signals, and the first analog-to-digital converter 51 will decimate the signals output by the analog front end at a certain ratio or quantity. For example, 1 to 4 signals are decimated in each clock cycle. The quantity or ratio of the signals decimated by the first analog-to-digital converter 51 can be adjusted according to actual needs. The fewer the signals decimated, the less the subsequent calculation amount, which can improve the calculation speed and reduce the compensation delay. However, the fewer the decimated signals, the more it will affect the accuracy of the compensation calculation. Therefore, it is necessary to balance the speed and accuracy of the compensation calculation and reasonably configure the decimated quantity.

[0044] In addition, for the positions of the signals extracted in each clock cycle, they can be determined by sampling and polling. For example, the number of signals extracted in each clock cycle is 4. In the first clock cycle, the signals at positions 1, 17, 33, and 49 are extracted. In the second clock cycle, the signals extracted are located at positions 2, 18, 34, and 50 respectively. In the third clock cycle, the signals extracted are located at positions 3, 19, 35, and 51 respectively, and so on. In this way, within multiple adjacent clock cycles, the positions of the signals extracted in each clock cycle are different, which can avoid the problem that the calculation of the multipath crosstalk noise is not accurate enough due to repeatedly extracting signals at the same position, improve the calculation accuracy of the multipath crosstalk noise, and thus the calculated compensation result is more accurate.

[0045] Then, step S2 is executed. The first feed-forward equalizer 52 is started, and an operation of clock and data recovery is performed. The first feed-forward equalizer 52 will calculate the feed-forward equalization error value (FFE error). Specifically, the first feed-forward equalizer 52 uses the data obtained after feed-forward equalization minus the target value used in the feed-forward equalization process to calculate the feed-forward equalization error value. Among them, the target value used in the feed-forward equalization process is the data decision value after feed-forward equalization.

[0046] Next, step S3 is executed to determine whether the first feed-forward equalizer 52 has converged. If it has not converged, it means that the data error is large and compensation cannot be performed, so step S7 is executed to directly output the data to the first decision feedback equalizer 54. If the first feed-forward equalizer 52 has converged, then step S4 is executed to turn on the first-order digital low-pass filter 53, and the feed-forward equalization error value calculated in step S2 is input to the first-order digital low-pass filter 53. In order to better filter the feed-forward equalization error value, reasonable parameters of the first-order digital low-pass filter 53 need to be configured according to the application scenario, so that the first-order digital low-pass filter 53 has a reasonable low-pass filtering bandwidth. After receiving the feed-forward equalization error value, the first-order digital low-pass filter 53 filters the feed-forward equalization error value and obtains a filtering result, and estimates the amount of multipath crosstalk noise based on the filtering result.

[0047] Next, step S5 is executed to compensate the data after feed-forward equalization using the amount of multipath crosstalk noise. Specifically, the data calculated by the first feed-forward equalizer 52 through feed-forward equalization is output to the first adder 55, and the amount of multipath crosstalk noise calculated by the first-order digital low-pass filter 53 is output to the first adder 55. The first adder 55 uses the data after feed-forward equalization minus the amount of multipath crosstalk noise to obtain the compensated data.

[0048] Finally, step S6 is executed. The first adder 55 outputs the compensated data to the first decision feedback equalizer 54, and the first decision feedback equalizer 54 equalizes the data again.

[0049] It can be seen that the multi-path crosstalk compensation device for the optical communication link in this embodiment is a backward compensation scheme, that is, the compensation for multi-path crosstalk noise is performed after the first feed-forward equalizer 52. This scheme can compensate for the multi-path crosstalk noise in the current clock cycle within the same clock cycle, ensuring the timeliness of multi-path crosstalk noise compensation.

[0050] Second Embodiment:

[0051] See Figure 6 , the multi-path crosstalk compensation device for the optical communication link in this embodiment includes a second analog-to-digital converter 61, a second feed-forward equalizer 62, a second-order loop filter 63, a second decision feedback equalizer 64, and a second adder 65. Among them, the second analog-to-digital converter 61 is used to collect the analog signal output by the analog front end and convert the collected analog signal into a digital signal. Since in the optical communication link, the signal change rate of the analog front end is very fast, in order to ensure accurate sampling of the data, the sampling rate of the second analog-to-digital converter 61 needs to be greater than or equal to the baud rate of the analog front end output signal. In addition, while converting the collected analog signal into a digital signal, the second analog-to-digital converter 61 also converts the serial signal into parallel data for the DSP chip to process the parallel data.

[0052] The second feed-forward equalizer 62 is used to receive the digital signal output by the second analog-to-digital converter 61 and is used to compensate for the inter-symbol interference damage of the received digital signal. For example, the second feed-forward equalizer 62 obtains a stable feed-forward equalization error through adaptive convergence calculation. The process of feed-forward equalization is to perform feed-forward equalization calculation on the received data, such as performing convolution calculation on the input data and the feed-forward equalization coefficient.

[0053] The second-order loop filter 63 uses two coefficients to form a fixed second-order filter. The two coefficients are the integral coefficient ki and the proportional coefficient kp. Moreover, the value of the proportional coefficient kp needs to be greater than or equal to the value of the integral coefficient ki to ensure the stability of the loop, realize the smoothing and averaging process of the input signal and the historical signal, and play the role of low-pass filtering. In this embodiment, the bandwidth of the second-order loop filter 63 can be adjusted by configuring different filter coefficients to meet the usage requirements in various scenarios. In order to meet the need to track multi-path crosstalk noise, the bandwidth of the second-order loop filter 63 in this embodiment is greater than 10 MHz.

[0054] The second-order loop filter 63 can receive the data output by the second feed-forward equalizer 62, that is, receive the data obtained after feed-forward equalization, and perform filtering processing on the data after feed-forward equalization. The result of the filtering processing is output to the second adder 65. The second adder 65 is used to receive the signals output by the second analog-to-digital converter 61 and the second-order loop filter 63, and perform addition and subtraction processing on the two signals. Specifically, the signal output by the second analog-to-digital converter 61 is subtracted from the signal output by the second-order loop filter 63, and the calculation result is fed back to the second feed-forward equalizer 62.

[0055] The second decision feedback equalizer 64 is used to perform equalization processing on the signal after feed-forward processing again. According to the decision result of the signal output by the second feed-forward equalizer 62, the decision feedback equalization coefficient is adjusted, and decision feedback equalization processing is performed accordingly, so as to reduce the signal distortion. Specifically, after determining the current decision symbol, the second decision feedback equalizer 64 needs to adjust the coefficient of the decision feedback equalization according to the decision result of the previous decision symbol, and apply it to the current decision symbol, and then send the compensated current decision symbol to the decision device for decision, and output the decision result of the current decision symbol.

[0056] The following combines Figure 7 to introduce the specific process of the compensation method for multipath crosstalk of the optical communication link in this embodiment. First, step S11 is executed. The second analog-to-digital converter 61 samples the analog signal output by the analog front end, performs analog-to-digital conversion on the analog signal to obtain a digital signal, and converts the serial signal into a parallel signal. When the second analog-to-digital converter 61 samples the signal, the signal is decimated at a certain ratio in each clock cycle. For example, 1 to 4 signals are decimated in one clock cycle. And the number or ratio of the signals decimated by the second analog-to-digital converter 61 can be adjusted according to actual needs. In addition, for the positions of the signals decimated in each clock cycle, a polling method can be used for determination, that is, in adjacent multiple clock cycles, the positions of the signals decimated in each clock cycle change in turn, such as moving one bit backward. This method can avoid the problem that the calculation of the multipath crosstalk noise is inaccurate due to repeatedly sampling signals at the same position.

[0057] Then, step S12 is executed. The second feed-forward equalizer 62 is started, and an operation of clock and data recovery is performed. And the second feed-forward equalizer 62 calculates the feed-forward equalization error value (FFE error). Specifically, the second feed-forward equalizer 62 uses the data obtained after feed-forward equalization in the current clock cycle minus the target value used in the feed-forward equalization process in the current clock cycle to calculate the feed-forward equalization error value.

[0058] Next, step S13 is executed to determine whether the second feedforward equalizer 62 has converged. If it has not converged, indicating that the data error is large and compensation cannot be performed, then step S19 is executed to directly output the data to the second decision feedback equalizer 64. If the second feedforward equalizer 62 has converged, then step S14 is executed to turn on the second-order loop filter 63, and the feedforward equalization error value calculated in step S12 is input to the second-order loop filter 63. To better filter the feedforward equalization error value, reasonable parameters of the second-order loop filter 63 need to be configured according to the application scenario, so that the second-order loop filter 63 has a reasonable low-pass filtering bandwidth. After receiving the feedforward equalization error value, the second-order loop filter 63 filters the feedforward equalization error value and obtains a filtering result, and estimates the multipath crosstalk noise amount of the current clock cycle based on the filtering result.

[0059] On the other hand, in the next clock cycle, the second analog-to-digital converter 61 will collect the analog signal output by the analog front end in the next clock cycle, that is, step S15 is executed, and then step S16 is executed to compensate the data in the next clock cycle using the multipath crosstalk noise amount calculated in the current clock cycle. Specifically, the second adder 65 uses the data obtained by the second analog-to-digital converter 61 in the next clock cycle minus the multipath crosstalk noise amount calculated in the current clock cycle to calculate the compensation data in the next clock cycle, that is, uses the data obtained in step S15 minus the data output by the second-order loop filter 63 in step S14 to obtain the compensation data in the next clock cycle.

[0060] Then, step S17 is executed, and the second feedforward equalizer 62 calculates a new feedforward equalization error value in the next clock cycle, and at the same time performs the operation of clock recovery. In the next clock cycle, the data received by the second feedforward equalizer 62 is the data output by the second adder 65. Therefore, this data is the calculation result of using the data obtained by the second analog-to-digital converter 61 in the next clock cycle minus the multipath crosstalk noise amount calculated in the current clock cycle. Finally, step S18 is executed to output the compensated data to the second decision feedback equalizer 64, and the second decision feedback equalizer 64 performs equalization processing on the data again.

[0061] It can be seen that the compensation device for multipath crosstalk in the optical communication link of this embodiment is a forward compensation scheme, that is, the multipath crosstalk noise calculated in the current clock cycle is used to compensate before the next clock cycle is fed forward equalized by the second feedforward equalizer 62.

[0062] Since the analog-to-digital converter of the present invention samples the data output by the analog front end by sampling, it can reduce the amount of data for subsequent processing, and by polling to determine the position of the data extracted in each clock cycle, it is beneficial to improve the accuracy of multipath crosstalk noise compensation.

[0063] Finally, it should be emphasized that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. Method for compensating multipath crosstalk in an optical communication link, comprising: An analog-to-digital converter samples the signal output by the analog front end and converts it into a digital signal, and then outputs it to a feedforward equalizer; Characterized in that: The feedforward equalizer calculates a feedforward equalization error value by subtracting the target value used in the feedforward equalization process from the data after feedforward equalization; Determine whether the feedforward equalizer converges. If it converges, output the feedforward equalization error value to a filter, and the filter filters the feedforward equalization error value to obtain the multipath crosstalk noise amount; Perform compensation calculation using the multipath crosstalk noise amount, and output the compensated data to a decision feedback equalizer; Wherein, when the analog-to-digital converter samples the signal output by the analog front end, the signal output by the analog front end is sampled at a preset ratio in each clock cycle, and the position of the signal sampled in each clock cycle is determined in a polling manner.

2. The method for compensating multipath crosstalk in an optical communication link according to claim 1, characterized in that: Sampling the signal output by the analog front end at a preset ratio in each clock cycle includes: Sampling a preset number of the signals output by the analog front end in each clock cycle.

3. The method for compensating multipath crosstalk in an optical communication link according to claim 1 or 2, characterized in that: Before calculating the feedforward equalization error value, a clock recovery operation is also performed.

4. The method for compensating multipath crosstalk in an optical communication link according to claim 1 or 2, characterized in that: The filter is a first-order digital low-pass filter; When performing compensation calculation using the multipath crosstalk noise amount, use the multipath crosstalk noise amount to compensate the data after feedforward equalization of the feedforward equalizer: subtract the multipath crosstalk noise amount from the data after feedforward equalization of the feedforward equalizer in the current clock cycle to obtain the compensated data in the current clock cycle.

5. The method for compensating multipath crosstalk in an optical communication link according to claim 1 or 2, characterized in that: The filter is a second-order loop filter; When performing compensation calculation using the multipath crosstalk noise amount, subtract the multipath crosstalk noise amount calculated in the current cycle from the data sampled by the analog-to-digital converter in the next clock cycle to obtain the compensated data in the next clock cycle.

6. The method for compensating multipath crosstalk in an optical communication link according to claim 5, characterized in that: After calculating the compensated data in the next clock cycle, further perform: In the next clock cycle, calculate the feedforward equalization error value of this clock cycle.

7. Device for compensating multipath crosstalk in an optical communication link, comprising: An analog-to-digital converter for sampling the signal output by the analog front end and converting it into a digital signal; Characterized in that: A feedforward equalizer for receiving the signal of the analog-to-digital converter and performing feedforward equalization processing, and calculating a feedforward equalization error value by subtracting the target value used in the feedforward equalization process from the data after feedforward equalization; A filter, when the feedforward equalizer converges, receiving the feedforward equalization error value input, and filtering the feedforward equalization error value to obtain the multipath crosstalk noise amount; An adder for performing compensation calculation using the multipath crosstalk noise amount; A decision feedback equalizer that receives the compensated data and performs equalization processing on the compensated data again; Wherein, when the analog-to-digital converter samples the signal output by the analog front end, the signal output by the analog front end is decimated at a preset ratio in each clock cycle, and the position of the signal decimated in each clock cycle is determined in a polling manner.

8. The compensation device for multipath crosstalk of an optical communication link according to claim 7, wherein: The sampling rate of the analog-to-digital converter is greater than or equal to the baud rate of the signal output by the analog front end.

9. The compensation device for multipath crosstalk of an optical communication link according to claim 7 or 8, wherein: When the analog-to-digital converter decimates the signal output by the analog front end at a preset ratio in each clock cycle, a preset number of the signals output by the analog front end are decimated in each clock cycle.

Images