A distributed polarization mode dispersion detection network generation method and device

Abstract

The application relates to the technical field of optical communication, and discloses a distributed polarization mode dispersion detection network generation method and device. The application can receive end optical signals of a sample optical fiber link and convert the end optical signals into electrical signals, pre-process the electrical signals to obtain pre-processed electrical signals, and determine original electrical signals; the sample optical fiber link is divided into N first segmented links, and the polarization mode dispersion values of each first segmented link are set; a distributed polarization mode dispersion compensation algorithm is constructed and mapped into a corresponding to-be-trained neural network; the to-be-trained neural network is trained according to the pre-processed electrical signals and the original electrical signals, so as to obtain a trained target neural network; and the target neural network is used for detecting the distributed polarization mode dispersion value of a to-be-measured optical fiber link. The application can generate the target neural network for distributed polarization mode dispersion detection of the to-be-measured optical fiber link, can realize detection of local polarization mode dispersion of the optical fiber link, and increases the diversity of the local polarization mode dispersion detection mode of the optical fiber link.

Description

Technical Field

[0001] This invention relates to the field of optical communication technology, and in particular to a method and apparatus for generating a distributed polarization mode dispersion detection network. Background Technology

[0002] Polarization mode dispersion in fiber optic links affects the quality of optical signal transmission. At the same time, polarization mode dispersion can also be used to detect whether there are abnormalities in the optical signal transmission of fiber optic links.

[0003] Related technologies can be used in various ways to determine the polarization mode dispersion of fiber optic links, such as prediction methods based on the degree of polarization, and prediction methods based on the power spectrum of radio frequency signals.

[0004] However, the relevant techniques are generally used to determine the cumulative polarization mode dispersion of optical fiber links, and there are fewer ways to determine the local polarization mode dispersion value. Summary of the Invention

[0005] This invention provides a method and apparatus for generating a distributed polarization mode dispersion detection network, which addresses the deficiency of limited methods for determining local polarization mode dispersion in related technologies. It generates a target neural network that can be used to detect distributed polarization mode dispersion values ​​in optical fiber links, thereby diversifying the methods for determining local polarization mode dispersion.

[0006] In a first aspect, the present invention provides a method for generating a distributed polarization mode dispersion detection network, the method comprising:

[0007] Receive the optical signal at the end of the sample fiber optic link and convert it into a corresponding electrical signal;

[0008] The electrical signal is preprocessed to obtain a preprocessed electrical signal, and the original electrical signal, the cumulative polarization mode dispersion value of the sample fiber link, and the polarization principal axis direction are determined based on the electrical signal.

[0009] The sample fiber optic link is divided into N first segment links, and the polarization mode dispersion value of each first segment link is set based on the cumulative value of polarization mode dispersion.

[0010] A distributed polarization mode dispersion compensation algorithm is constructed by utilizing the polarization mode dispersion value of each first segment link, the polarization principal axis direction, and the basic signal compensation algorithm.

[0011] The distributed polarization mode dispersion compensation algorithm is mapped to a corresponding neural network to be trained.

[0012] The neural network to be trained is trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed polarization mode dispersion value of the optical fiber link under test.

[0013] Optionally, training the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network includes:

[0014] The preprocessed electrical signal and the original electrical signal are used as training samples and sample labels corresponding to the training samples, respectively.

[0015] The training samples and the sample labels are used to train the neural network to obtain a trained target neural network. The target neural network is used to detect the polarization mode dispersion values ​​of N second segment links in the fiber optic link under test. The length of the second segment link corresponds to the length of the first segment link.

[0016] The polarization mode dispersion value of the second segment link is used to compare with the polarization mode dispersion reference value of the second segment link to determine whether the working state of the second segment link is abnormal. The polarization mode dispersion reference value is obtained based on the target neural network and the fiber optic link under test in normal working state.

[0017] Optionally, when the length of each of the first segment links is the same, setting the polarization mode dispersion value of each of the first segment links based on the cumulative polarization mode dispersion value includes:

[0018] Determine the position number of each first segment link, the position number being used to identify the position of the first segment link in each of the first segment links of the sample optical fiber link;

[0019] Calculate the ratio between the cumulative polarization mode dispersion value and N;

[0020] For any of the first segment links, the ratio is multiplied by the position number of the first segment link to obtain the corresponding product, which is used as the polarization mode dispersion value of the first segment link.

[0021] Optionally, the step of constructing a distributed polarization mode dispersion compensation algorithm using the polarization mode dispersion value of each of the first segment links, the polarization principal axis direction, and the basic signal compensation algorithm includes:

[0022] Perform a Fourier transform on the electrical signal variable to obtain the frequency domain signal variable;

[0023] The frequency domain signal variable is transformed from the original coordinate axis direction to the polarization principal axis direction to obtain the principal axis signal variable;

[0024] Based on the main axis signal variable and the polarization mode dispersion value of each first segment link, a polarization mode dispersion compensation factor is generated for each first segment link.

[0025] The distributed polarization mode dispersion compensation algorithm is constructed based on the polarization mode dispersion compensation factor and the basic signal compensation algorithm corresponding to each of the first segment links.

[0026] Optionally, when the basic signal compensation algorithm includes a frequency domain conversion module, a principal axis conversion module, a digital back-transmission module, a coordinate axis conversion module, and a time domain conversion module connected in sequence, the step of constructing the distributed polarization mode dispersion compensation algorithm based on the polarization mode dispersion compensation factor corresponding to each first segment link and the basic signal compensation algorithm includes:

[0027] A distributed polarization mode dispersion compensation module is constructed using the polarization mode dispersion compensation factor.

[0028] The distributed polarization mode dispersion compensation module is placed between the main axis conversion module and the digital back transmission module of the basic signal compensation algorithm to obtain the distributed polarization mode dispersion compensation algorithm. The distributed polarization mode dispersion compensation algorithm includes the frequency domain conversion module, the main axis conversion module, the distributed polarization mode dispersion compensation module, the digital back transmission module, the coordinate axis conversion module, and the time domain conversion module connected in sequence.

[0029] Optionally, the distributed polarization mode dispersion compensation module is used to perform polarization mode dispersion compensation on the signal from the main axis conversion module according to the first polarization mode dispersion compensation factor to obtain the first polarization mode compensated signal; wherein, the first polarization mode dispersion compensation factor is the polarization mode dispersion compensation factor corresponding to the first segment link with position number 1.

[0030] The digital back-transmission module is used to perform dispersion compensation and nonlinear compensation on the first polarization mode compensated signal to obtain a first-stage compensated signal and return it to the distributed polarization mode dispersion compensation module. This allows the distributed polarization mode dispersion compensation module to perform polarization mode dispersion compensation on the first-stage compensated signal according to a second polarization mode dispersion compensation factor until polarization mode dispersion compensation is performed using the polarization mode dispersion compensation factors corresponding to all the segmented links, and the final stage compensated signal is obtained. The final stage compensated signal is then sent to the coordinate axis transformation module. The second polarization mode dispersion compensation factor is the polarization mode dispersion compensation factor corresponding to the first segmented link with position number 2.

[0031] Optionally, the neural network to be trained includes a frequency domain transformation layer, a principal axis transformation layer, a coordinate axis transformation layer, a time domain transformation layer, and multiple signal compensation layers;

[0032] The frequency domain transformation layer, the principal axis transformation layer, the coordinate axis transformation layer, and the time domain transformation layer correspond to the frequency domain transformation module, the principal axis transformation module, the coordinate axis transformation module, and the time domain transformation module, respectively.

[0033] Each of the signal compensation layers is used to perform polarization mode dispersion compensation, dispersion compensation and nonlinear compensation on the received signal, so as to realize a stage compensation based on the distributed polarization mode dispersion compensation module and the digital back transmission module.

[0034] The frequency domain conversion layer, the principal axis conversion layer, and the signal compensation layer used to implement the first stage compensation are connected in sequence; all the signal compensation layers are connected in sequence according to the stage compensation order, and the output of one signal compensation layer is the input of the next signal compensation layer; the signal compensation layer used to implement the last stage compensation, the coordinate axis conversion layer, and the time domain conversion layer are connected in sequence.

[0035] Optionally, each of the signal compensation layers includes polarization mode dispersion estimation parameters to be trained, which are used to estimate polarization mode dispersion compensation values; the principal axis transformation layer and the coordinate axis transformation layer include principal axis azimuth and principal axis elevation angles to be trained;

[0036] The step of training the neural network to be trained based on the preprocessed electrical signal and the original electrical signal includes:

[0037] The preprocessed electrical signal is used as a training sample and input into the neural network to be trained for polarization mode dispersion compensation, dispersion compensation and nonlinear compensation, so as to obtain the compensated signal output by the neural network to be trained.

[0038] The loss function value is calculated based on the difference between the compensated signal and the original electrical signal;

[0039] The polarization mode dispersion estimation parameters, principal axis azimuth angle, and / or principal axis elevation angle in the neural network to be trained are updated based on the loss function value.

[0040] Optionally, determining the original electrical signal, the cumulative polarization mode dispersion value of the sample fiber link, and the polarization principal axis direction based on the electrical signal includes:

[0041] The electrical signal is sequentially subjected to bias correction processing, orthogonalization processing, and dispersion compensation processing to obtain the corresponding first processed signal;

[0042] The first processed signal is polarization demultiplexed and polarization mode dispersion compensated using an adaptive equalization filter to obtain the second processed signal and to acquire multiple tap coefficients in the adaptive equalization filter.

[0043] Carrier phase recovery is performed on the second processed signal to obtain the phase-recovered signal and phase noise estimate;

[0044] The phase-recovered signal is processed to obtain the original symbol signal;

[0045] The original symbol signal is pulse-shaped using a Nyquist filter to obtain a pulse-shaped signal.

[0046] The pulse-shaped signal is subjected to phase noise loading based on the phase noise estimate to obtain the original electrical signal;

[0047] The cumulative value of polarization mode dispersion and the direction of the principal polarization axis are determined using the plurality of tap coefficients.

[0048] In a second aspect, the present invention provides a distributed polarization mode dispersion detection network generation device, the device comprising:

[0049] The receiving unit is used to receive the optical signal at the end of the sample fiber optic link;

[0050] A conversion unit is used to convert the terminal optical signal into a corresponding electrical signal;

[0051] A preprocessing unit is used to preprocess the electrical signal to obtain a preprocessed electrical signal;

[0052] The determining unit is used to determine the original electrical signal, the cumulative polarization mode dispersion value of the sample optical fiber link, and the polarization principal axis direction based on the electrical signal.

[0053] A partitioning unit is used to divide the sample optical fiber link into N first segment links;

[0054] The setting unit is used to set the polarization mode dispersion value of each first segment link based on the accumulated polarization mode dispersion value;

[0055] The construction unit is used to construct a distributed polarization mode dispersion compensation algorithm by utilizing the polarization mode dispersion value of each of the first segment links, the polarization principal axis direction, and the basic signal compensation algorithm;

[0056] A mapping unit is used to map the distributed polarization mode dispersion compensation algorithm to a corresponding neural network to be trained.

[0057] The training unit is used to train the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed polarization mode dispersion value of the optical fiber link under test.

[0058] The distributed polarization mode dispersion (PDDM) detection network generation method and apparatus provided by this invention can receive the end optical signal of a sample optical fiber link and convert it into a corresponding electrical signal; preprocess the electrical signal to obtain a preprocessed electrical signal, and determine the original electrical signal, the cumulative value of polarization mode dispersion (PMD) of the sample optical fiber link, and the polarization principal axis direction based on the electrical signal; divide the sample optical fiber link into N first segment links, and set the PMD value of each first segment link based on the cumulative PMD value; construct a distributed PMD compensation algorithm using the PMD value, polarization principal axis direction, and basic signal compensation algorithm of each first segment link; map the distributed PMD compensation algorithm to a corresponding neural network to be trained; train the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed PMD value of the optical fiber link under test. This invention can generate a target neural network for distributed PMD detection of the optical fiber link under test, enabling the detection of local PMD of the optical fiber link and increasing the diversity of methods for detecting local PMD of the optical fiber link. Attached Figure Description

[0059] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0060] Figure 1 A flowchart of a method for generating a distributed polarization mode dispersion detection network provided in an embodiment of the present invention;

[0061] Figure 2 A flowchart illustrating a method for monitoring the operational status of a fiber optic link under test, as provided in an embodiment of the present invention;

[0062] Figure 3 A flowchart of electrical signal processing provided in an embodiment of the present invention;

[0063] Figure 4 This is a schematic diagram illustrating the setting of polarization mode dispersion values ​​for a segmented link, provided in an embodiment of the present invention.

[0064] Figure 5 This is a schematic diagram of a distributed polarization mode dispersion compensation algorithm provided in an embodiment of the present invention;

[0065] Figure 6 This is a schematic diagram illustrating the mapping relationship between a distributed polarization mode dispersion compensation algorithm and a neural network to be trained, provided in an embodiment of the present invention.

[0066] Figure 7 Another electrical signal processing flowchart provided in this embodiment of the invention;

[0067] Figure 8 This is a schematic diagram of a distributed polarization mode dispersion detection network generation device provided in an embodiment of the present invention;

[0068] Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0069] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0070] The following is combined Figures 1-7 The present invention describes a method for generating a distributed polarization mode dispersion detection network.

[0071] like Figure 1 As shown, this embodiment proposes a first method for generating a distributed polarization mode dispersion detection network, which may include the following steps:

[0072] S101, Receive the end optical signal of the sample fiber optic link.

[0073] The sample fiber optic link can be a fiber optic link of a certain length that is in normal working condition.

[0074] Among them, the terminal optical signal is the optical signal at the end of the sample optical fiber link.

[0075] Specifically, in this embodiment, a polarization-multiplexed coherent optical communication receiver can be used to receive the end optical signal of the sample fiber optic link.

[0076] The polarization-multiplexed coherent optical communication receiver may include a pre-amplified erbium-doped fiber amplifier, an optical bandpass filter, a 90° mixer, and a balanced photodetector.

[0077] S102. Convert the optical signal at the end of the sample fiber optic link into the corresponding electrical signal.

[0078] Specifically, in this embodiment, a balanced photodetector in a polarization multiplexed coherent optical communication receiver can be used to convert the end optical signal into a corresponding electrical signal.

[0079] S103. Preprocess the electrical signal to obtain the preprocessed electrical signal.

[0080] The preprocessed electrical signal is the signal obtained after preprocessing the electrical signal.

[0081] It should be noted that preprocessing can improve the signal quality of electrical signals, reduce noise interference, and enhance signal reliability and stability. Furthermore, preprocessing can adjust and standardize signals to facilitate subsequent use in training neural networks.

[0082] S104. Determine the original electrical signal, the cumulative value of polarization mode dispersion, and the direction of the polarization principal axis of the sample fiber link based on the electrical signal.

[0083] Specifically, this embodiment can demodulate the electrical signal to restore the original electrical signal.

[0084] Specifically, in this embodiment, the cumulative value of polarization mode dispersion and the direction of polarization principal axis of the sample optical fiber link can be calculated based on the electrical signal.

[0085] S105. Divide the sample fiber optic link into N first segment links.

[0086] The first segment link refers to a segment of the sample fiber optic link. Specifically, the lengths of different first segment links can be the same or different.

[0087] It should be noted that there is no overlap between each first segment link.

[0088] S106. Set the polarization mode dispersion value for each first segment link based on the cumulative value of polarization mode dispersion.

[0089] Specifically, in this embodiment, the polarization mode dispersion value of each first segment link can be set according to the cumulative value of polarization mode dispersion.

[0090] S107. Construct a distributed polarization mode dispersion compensation algorithm by utilizing the polarization mode dispersion value, polarization principal axis direction, and basic signal compensation algorithm of each first segment link.

[0091] The basic signal compensation algorithm can include Fourier transform algorithm and digital backpropagation algorithm, among other basic algorithms used for signal compensation.

[0092] Specifically, in this embodiment, a distributed polarization mode dispersion compensation algorithm can be constructed based on the basic signal compensation algorithm according to the cumulative value of polarization mode dispersion, the direction of the polarization principal axis, and the length of each segment link.

[0093] S108. Map the distributed polarization mode dispersion compensation algorithm to the corresponding neural network to be trained.

[0094] Specifically, in this embodiment, the distributed polarization mode dispersion compensation algorithm can be mapped to a corresponding neural network and used as the neural network to be trained.

[0095] S109. Train the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain the trained target neural network. The target neural network is used to detect the distributed polarization mode dispersion value of the optical fiber link under test.

[0096] It is understandable that the target neural network is a distributed polarization mode dispersion detection network.

[0097] Specifically, in this embodiment, the preprocessed electrical signal can be used as a training sample, and the original electrical signal can be used as the corresponding sample label. The training sample and sample label are used to train the neural network to be trained, and the parameters in the neural network to be trained are updated to train the distributed polarization mode dispersion detection capability of the neural network to be trained.

[0098] Specifically, in this embodiment, multiple training samples and corresponding sample labels are collected to train the neural network to be trained, in accordance with steps S101 to S108. This ensures the accuracy and stability of the target neural network and other network performance, thereby effectively ensuring the accuracy and reliability of distributed polarization mode dispersion detection of the fiber optic link under test.

[0099] Optionally, this embodiment can collect the end optical signals of the sample fiber optic link at different times to generate multiple training samples and sample labels, or collect the end optical signals of different sample fiber optic lines to generate multiple training samples and sample labels.

[0100] The fiber optic link under test is the fiber optic link for which distributed polarization mode dispersion detection is required. It is understood that the fiber optic link under test can be in a normal operating state or an abnormal operating state; this embodiment does not limit this.

[0101] The distributed polarization mode dispersion value can include the polarization mode dispersion values ​​of different parts of the fiber optic link under test.

[0102] The distributed polarization mode dispersion (PDDM) detection network generation method proposed in this embodiment can receive the end optical signal of a sample fiber optic link and convert it into a corresponding electrical signal; preprocess the electrical signal to obtain a preprocessed electrical signal, and determine the original electrical signal, the cumulative value of polarization mode dispersion (PMD) of the sample fiber optic link, and the direction of the polarization principal axis based on the electrical signal; divide the sample fiber optic link into N first segment links, and set the PMD value of each first segment link based on the cumulative PMD value; construct a distributed PMD compensation algorithm using the PMD value, the direction of the polarization principal axis, and the basic signal compensation algorithm of each first segment link; map the distributed PMD compensation algorithm to a corresponding neural network to be trained; train the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed PMD value of the fiber optic link under test. This embodiment can generate a target neural network for distributed PMD detection of the fiber optic link under test, enabling the detection of local PMD of the fiber optic link and increasing the diversity of methods for detecting local PMD of the fiber optic link.

[0103] based on Figure 1 This embodiment proposes a second method for generating a distributed polarization mode dispersion detection network. In this method, step S109 may include:

[0104] The preprocessed electrical signal and the original electrical signal are used as training samples and the sample labels corresponding to the training samples, respectively.

[0105] The training samples and sample labels are used to train the neural network to obtain the trained target neural network. The target neural network is used to detect the polarization mode dispersion values ​​of N second segment links in the fiber optic link under test. The length of the second segment link corresponds to the length of the first segment link.

[0106] The polarization mode dispersion value of the second segment link is used to compare with the polarization mode dispersion reference value of the second segment link to determine whether the working state of the second segment link is abnormal. The polarization mode dispersion reference value is obtained based on the target neural network and the fiber optic link under test in normal working state.

[0107] Specifically, the lengths of the aforementioned N first segment links and the aforementioned N second segment links can be in a one-to-one correspondence, and the ratio between the first and second segment links in the same position can be a fixed value. For example, when the sample fiber link and the fiber link under test are of equal length, and the first segment link is obtained by dividing the sample fiber link into N equal parts, the second segment link can also be obtained by dividing the fiber link under test into N equal parts. In this case, the lengths of the first and second segment links in the same position are equal, i.e., the ratio is equal to 1. Of course, the ratio between the first and second segment links in the same position can also be other values, such as 0.5. It can be understood that in this case, the length of the fiber link under test can be twice that of the sample fiber link.

[0108] like Figure 2 As shown, in a long-distance optical fiber communication system, this embodiment can detect the polarization mode dispersion (PMD) values ​​of each second segment link in the optical fiber link under test when the link under test is in normal working condition, such as when the link under test and the surrounding environment are in normal condition, using a target neural network and storing them as reference values ​​for each second segment link. To promptly detect anomalies, this embodiment can perform periodic inspections, collecting the end optical signals of the optical fiber link under test and using the target neural network to detect the PMD values ​​of each second segment link. To eliminate the influence of noise, interference, and other factors, this embodiment should collect multiple sets of end optical signals each time it is tested, and take the average of the PMD values ​​as the final PMD value of the test result for that round. For example, for a certain second segment link, this embodiment can use the average of the detected PMD values ​​of each second segment link as the final PMD value of that second segment link. Then, when comparing differences, the final PMD value of the second segment link can be compared with the reference value of the PMD value of that second segment link. Among them, distributed polarization mode dispersion value 1, distributed polarization mode dispersion value 2, and distributed polarization mode dispersion value n are the distributed polarization mode dispersion values ​​corresponding to the terminal optical signals collected at different times. This embodiment can compare the results of this detection with reference values ​​to determine whether the difference exceeds a certain threshold. If so, it is determined that there is an anomaly in the corresponding second segment link, the second segment link is located, and staff are notified to perform line maintenance. If the difference does not exceed the threshold, it can be determined that the fiber optic link under test is in normal operating condition.

[0109] The distributed polarization mode dispersion detection network generation method proposed in this embodiment uses a target neural network to monitor the distributed polarization mode dispersion in the fiber optic link under test, determine whether the working status of the second segment link is normal, locate the second segment link with abnormality, improve the efficiency of anomaly detection and location, and thus improve the efficiency of anomaly handling.

[0110] based on Figure 1 This embodiment proposes a third method for generating a distributed polarization mode dispersion detection network. In this method, step S104 may include:

[0111] The electrical signal is sequentially subjected to bias correction, orthogonalization, and dispersion compensation to obtain the corresponding first-processed signal.

[0112] The first processed signal is polarization demultiplexed and polarization mode dispersion is compensated by using an adaptive equalization filter to obtain the second processed signal and obtain multiple tap coefficients in the adaptive equalization filter.

[0113] Carrier phase recovery is performed on the second processed signal to obtain the phase-recovered signal and phase noise estimates.

[0114] The phase-recovered signal is processed to obtain the original symbol signal;

[0115] The original symbol signal is pulse-shaped using a Nyquist filter to obtain the pulse-shaped signal.

[0116] The original electrical signal is obtained by applying phase noise to the pulse-shaped signal based on the phase noise estimate.

[0117] Multiple tap coefficients are used to determine the cumulative value of polarization mode dispersion and the direction of the principal polarization axis.

[0118] Specifically, such as Figure 3 As shown, this embodiment can input the end optical signal of the sample fiber optic link to the optical receiver front end, and input the signal output from the local oscillator laser to the optical receiver front end (the signal output from the local oscillator laser is used to generate an electrical signal), and process it using a transconductance amplifier (TIA) and an analog-to-digital converter (ADC). The ADC downsamples the electrical signal, transforming it into a digital signal that is upsampled by a factor of two. After polarization correction and orthogonalization processing, the amplitude and phase imbalance of the I and Q paths (I and Q are two signal components in digital communication, used to represent the real and imaginary parts of the modulated signal) are corrected. Next, dispersion estimation and compensation are performed, using an adaptive equalization filter for polarization demultiplexing to compensate for the damage caused by polarization mode dispersion. Then, frequency offset estimation and compensation, and laser phase noise compensation (i.e., carrier phase recovery) are performed to obtain the phase-recovered signal and phase noise estimate. The phase-recovered signal is then decided and decoded to obtain the original symbol signal. Next, the original symbol signal is pulse-shaped using a Nyquist filter to obtain the pulse-shaped signal. Phase noise is then applied to the pulse-shaped signal based on the estimated phase noise value to obtain the original electrical signal.

[0119] Among them, after the adaptive equalization filter completes the polarization demultiplexing and polarization mode dispersion compensation tasks, it obtains four trained tap coefficients h. xx h xy h yx and h yy Based on the four trained tap coefficients h xx h xy h yx and h yy Determine the cumulative value of polarization mode dispersion and the direction of the principal polarization axis.

[0120] Specifically, in this embodiment, the following matrix can be constructed first:

[0121]

[0122]

[0123] Where ω is the angular frequency, M(ω) is a user-defined matrix, DFT represents the Discrete Fourier Transform, which can be implemented using the Fast Fourier Transform (FFT). H(ω) is the channel frequency response matrix corresponding to the polarization mode dispersion. -1 (ω) is the inverse of this matrix. h p (m) represents the tap coefficient of the adaptive equalization filter corresponding to the m-th signal sampling point, where p takes values ​​from x, xy, yx, and yy, and the number of tap coefficients is N. taps After the DFT operation, each frequency component also has four weighting coefficients.

[0124] Let τ be the cumulative amount of polarization mode dispersion, then the following relationship holds:

[0125]

[0126]

[0127] Here, R1(ω) is a rotation matrix that rotates the signal polarization direction to the direction of the principal polarization axis. ζ and 2φ are the azimuth and elevation angles of the principal polarization axis at the end of the sample fiber link on the Poincaré sphere. D(ω) is the channel frequency response matrix corresponding to the dispersion effect. Δω represents the frequency offset. The cumulative polarization mode dispersion τ can be represented by the matrix M(ω+Δω)M(ω). -1 The eigenvalues ​​ρ1 and ρ2 are used for calculation. That is, by solving the eigenvalue equation M(ω+Δω)M(ω). -1 From t = ρt, we can obtain two eigenvalues ​​ρ1 and ρ2, and two corresponding eigenvectors, where t represents the eigenvector of the eigenvalue equation. Using the eigenvalues, we can obtain:

[0128]

[0129] The two eigenvectors correspond to orthogonal polarization principal states t slow and t fast .

[0130] Subsequently, this embodiment can be based on the polarization dominant state t slow and t fast Determine the direction of the principal polarization axis.

[0131] The distributed polarization mode dispersion detection network generation method proposed in this embodiment can effectively determine the cumulative value of polarization mode dispersion and the polarization principal axis direction of the original electrical signal and the sample fiber link based on the electrical signal, thereby effectively ensuring the construction and generation of subsequent algorithms and neural networks.

[0132] based on Figure 1 This embodiment proposes a fourth method for generating a distributed polarization mode dispersion detection network. In this method, when the length of each first segment link is the same, step S106 above may include:

[0133] Determine the position number of each first segment link. The position number is used to identify the position of the first segment link in each first segment link of the sample fiber optic link.

[0134] Calculate the ratio between the cumulative value of polarization mode dispersion and N;

[0135] For any first segment link, the ratio is multiplied by the position number of the first segment link to obtain the corresponding product, which is used as the polarization mode dispersion value of the first segment link.

[0136] Specifically, the position number is the position number of a certain first segment link in the sample fiber optic link from the beginning to the end of the sample fiber optic link.

[0137] It should be noted that, in essence, polarization mode dispersion (PMD) values ​​vary randomly. However, it can be approximated that the average PMD value of a certain segment of the fiber optic link remains constant over a period of time. The cumulative PMD value and principal axis direction measured at the end of the fiber optic link can be used to describe the impact of PMD on the entire fiber optic link, but they are insufficient to describe the details of the evolution of local PMD values.

[0138] In this embodiment, the evolution of polarization mode dispersion along the fiber optic link can be approximated as a step function. The fiber optic link is divided into N first segment links, and the polarization mode dispersion of each first segment link can be calculated according to the following formula:

[0139] τ i =iτ step ;

[0140] τ step =τ / N;

[0141] Where τ is the cumulative polarization mode dispersion, i is the position number of the first segment link, and τ i It is the polarization mode dispersion value of the first segment link with position number i.

[0142] like Figure 4 As shown in the figure, this embodiment illustrates the setting diagram of the polarization mode dispersion value for each first segment link.

[0143] Figure 4 The data includes the actual polarization mode dispersion variation curves in the sample fiber optic links, as well as the polarization mode dispersion values ​​for each set first segment link. The polarization mode dispersion value of each first segment link is constant within the corresponding first segment link, exhibiting a step function variation.

[0144] The horizontal axis represents the distance between the first segment link and the beginning of the sample fiber link, in kilometers; the vertical axis represents the polarization mode dispersion value, in picoseconds. `step` represents the step size, indicating the length of each first segment link.

[0145] Optionally, step S107 above may include:

[0146] Perform a Fourier transform on the electrical signal variable to obtain the frequency domain signal variable;

[0147] The frequency domain signal variable is transformed from the original coordinate axis direction to the polarization principal axis direction to obtain the principal axis signal variable;

[0148] Based on the main axis signal variable and the polarization mode dispersion value of each first segment link, a polarization mode dispersion compensation factor corresponding to each first segment link is generated.

[0149] A distributed polarization mode dispersion compensation algorithm is constructed based on the polarization mode dispersion compensation factor and the basic signal compensation algorithm corresponding to each first segment link.

[0150] Optionally, when the basic signal compensation algorithm includes a frequency domain conversion module, a principal axis conversion module, a digital back-transmission module, a coordinate axis conversion module, and a time domain conversion module connected in sequence, the above-mentioned distributed polarization mode dispersion compensation algorithm constructed based on the polarization mode dispersion compensation factor corresponding to each first segment link and the basic signal compensation algorithm may include:

[0151] A distributed polarization mode dispersion compensation module is constructed using a polarization mode dispersion compensation factor;

[0152] By placing the distributed polarization mode dispersion compensation module between the main axis conversion module and the digital back transmission module of the basic signal compensation algorithm, a distributed polarization mode dispersion compensation algorithm is obtained. The distributed polarization mode dispersion compensation algorithm includes a frequency domain conversion module, a main axis conversion module, a distributed polarization mode dispersion compensation module, a digital back transmission module, a coordinate axis conversion module, and a time domain conversion module connected in sequence.

[0153] Optionally, a distributed polarization mode dispersion compensation module is used to perform polarization mode dispersion compensation on the signal from the main axis conversion module according to the first polarization mode dispersion compensation factor to obtain the first polarization mode compensated signal; wherein, the first polarization mode dispersion compensation factor is the polarization mode dispersion compensation factor corresponding to the first segment link with position number 1.

[0154] The digital back-transmission module is used to perform dispersion compensation and nonlinear compensation on the first polarization mode compensated signal to obtain the first stage compensated signal and return it to the distributed polarization mode dispersion compensation module. The distributed polarization mode dispersion compensation module then performs polarization mode dispersion compensation on the first stage compensated signal according to the second polarization mode dispersion compensation factor until polarization mode dispersion compensation is performed using the polarization mode dispersion compensation factors corresponding to all segmented links and the final stage compensated signal is obtained. The final stage compensated signal is then sent to the coordinate axis transformation module.

[0155] Specifically, the digital back-transmission module is the module used to perform the digital back-transmission algorithm-related processing.

[0156] It should be noted that the distributed polarization mode dispersion compensation algorithm is an algorithm that takes polarization mode dispersion into account and is built on the basis of the digital back-transmission algorithm.

[0157] See Figure 5 This embodiment can perform an FFT operation on the electrical signal variable to transform it into the frequency domain, obtaining the corresponding frequency domain signal variable. Then, this embodiment can convert the received signal A on the X and Y axes of the original coordinate system in the laboratory into the frequency domain signal variable. Rx,x and A Rx,y To convert to the polarization principal axis and obtain the principal axis signal A PSP,slow and A PSP,fast Specifically:

[0158]

[0159] Apply the polarization mode dispersion value of each step size, i.e., each first segment link, to A. PSP,slow and A PSP,fast We can obtain:

[0160]

[0161] Here, half of the delay applies to the slow component (slow principal axis component), and half applies to the fast component (fast principal axis component). The entire delay can also be applied to a single component, i.e.:

[0162]

[0163] in, This is the polarization mode dispersion compensation factor corresponding to the first segment link with position number i.

[0164] This embodiment can perform traditional digital backpropagation processing after inverting the polarization mode dispersion value, first compensating for dispersion, then transforming to the time domain, compensating for nonlinear damage, and finally transforming to the frequency domain.

[0165]

[0166]

[0167]

[0168] in, Represents the dispersion compensation operator. This represents the nonlinear compensation operator. α is the power loss coefficient of the optical fiber, j is the imaginary unit, β² is the group velocity dispersion parameter of the optical fiber, γ is the nonlinear coefficient of the optical fiber, and l... step It is the length of each step, l step,eff Is with l step The corresponding effective power length. After all steps are completed, adjust the signal principal axis coordinate system to the X and Y coordinate systems, and use IFFT operation to transform it into a time-domain signal.

[0169] It should be noted that, in Figure 5 In this context, PSP adjustment refers to the transformation of the signal along the coordinate axis.

[0170] Optionally, the neural network to be trained includes a frequency domain transformation layer, a principal axis transformation layer, a coordinate axis transformation layer, a time domain transformation layer, and multiple signal compensation layers;

[0171] The frequency domain transformation layer, principal axis transformation layer, coordinate axis transformation layer, and time domain transformation layer correspond to the frequency domain transformation module, principal axis transformation module, coordinate axis transformation module, and time domain transformation module, respectively.

[0172] Each signal compensation layer is used to perform polarization mode dispersion compensation, dispersion compensation and nonlinear compensation on the received signal, so as to realize a stage compensation based on the distributed polarization mode dispersion compensation module and the digital back transmission module.

[0173] The frequency domain conversion layer, the principal axis conversion layer, and the signal compensation layer used to implement the first stage compensation are connected in sequence; all signal compensation layers are connected in sequence according to the stage compensation order, and the output of one signal compensation layer is the input of the next signal compensation layer; the signal compensation layer, the coordinate axis conversion layer, and the time domain conversion layer used to implement the last stage compensation are connected in sequence.

[0174] Optionally, each signal compensation layer includes polarization mode dispersion estimation parameters to be trained, which are used to estimate the polarization mode dispersion compensation value; the principal axis transformation layer and coordinate axis transformation layer include the principal axis azimuth and principal axis elevation angles to be trained. The above training of the neural network to be trained based on the preprocessed electrical signal and the original electrical signal includes:

[0175] The preprocessed electrical signal is used as a training sample and input into the neural network to be trained for polarization mode dispersion compensation, dispersion compensation and nonlinear compensation, so as to obtain the compensated signal output by the neural network to be trained.

[0176] The loss function value is calculated based on the difference between the compensated signal and the original electrical signal;

[0177] The polarization mode dispersion estimation parameters, principal axis azimuth angle, and / or principal axis elevation angle in the neural network under training are updated based on the loss function value.

[0178] See Figure 6 The diagram shows the mapping relationship between the distributed polarization mode dispersion compensation algorithm and the neural network to be trained.

[0179] exist Figure 6 In the upper branch, a distributed polarization mode dispersion compensation algorithm is used, performing polarization mode dispersion compensation, dispersion compensation, and nonlinearity compensation at each step. The lower branch is the implementation of the neural network to be trained in the upper branch; the network layer is the signal compensation layer, which calculates τ... i Adding these parameters as training parameters allows for the acquisition of polarization parameter information at local locations along the fiber optic link, while simultaneously achieving polarization mode dispersion compensation. The trainable parameters for each network layer include τ. i The weighting factor w and the nonlinear parameter α. The correspondence between the output and input signals of each layer of the network is as follows:

[0180] x i =g(w(DGD(y) k )));

[0181] Where, the weighting factor w is the tap coefficient of the time-domain finite impulse response filter corresponding to dispersion, and g is the nonlinear activation function of each layer, defined as g (i) (x)=xexp(-jα i |x| 2DGD(·) indicates polarization mode dispersion compensation, w(·) indicates multiplication by a weighting factor, which is to compensate for dispersion; y k This represents the input variables of the k-th layer of the neural network. The frequency domain transformation layer, principal axis transformation layer, coordinate axis transformation layer, and time domain transformation layer in the neural network to be trained can further improve the accuracy and flexibility of polarization mode dispersion estimation. Its trainable parameters include the azimuth angle ζ and the elevation angle 2φ. After the preprocessed signal and the original electrical signal are input into the neural network to be trained, the steepest gradient descent method can be used to optimize each trainable parameter.

[0182] Specifically, the loss function can be chosen as the mean squared error. Where x represents the original electrical signal, This refers to the actual output signal of the neural network. Minimizing the MSE is equivalent to maximizing the signal-to-noise ratio (SNR). Here, E represents the expected value, which is the average value after multiple calculations.

[0183] See Figure 7 This embodiment can receive electrical signals and process them in two paths. One path performs preprocessing operations on the electrical signal, including downsampling, polarization correction, and orthogonalization, to obtain the test signal, i.e., the preprocessed electrical signal. The other path performs preprocessing, channel equalization, carrier phase recovery, symbol decision, and signal reconstruction operations (signal reconstruction includes pulse shaping and adding phase noise) to obtain the reference signal, i.e., the original electrical signal. Then, the test signal and the original electrical signal are used as training data and input to the network to be trained. The network to be trained can output a distributed polarization mode dispersion estimate based on the test signal.

[0184] The distributed polarization mode dispersion detection network generation method proposed in this embodiment can effectively ensure the generation of the target neural network, thereby effectively realizing distributed polarization mode dispersion detection and monitoring of optical fiber links, which is conducive to realizing integrated sensing and communication functions. This method can use forward communication signals to measure polarization mode dispersion, thereby improving the sensing distance. It does not require special signal design, does not rely on redundant information, and has high spectral efficiency.

[0185] like Figure 8 As shown, this embodiment proposes a distributed polarization mode dispersion detection network generation device, which may include:

[0186] The receiving unit 801 is used to receive the end optical signal of the sample optical fiber link;

[0187] The conversion unit 802 is used to convert the terminal optical signal into a corresponding electrical signal;

[0188] Preprocessing unit 803 is used to preprocess the electrical signal to obtain a preprocessed electrical signal;

[0189] The determining unit 804 is used to determine the original electrical signal, the cumulative polarization mode dispersion value of the sample optical fiber link, and the polarization principal axis direction based on the electrical signal.

[0190] The partitioning unit 805 is used to divide the sample optical fiber link into N first segment links;

[0191] Setting unit 806 is used to set the polarization mode dispersion value of each first segment link based on the accumulated polarization mode dispersion value;

[0192] Construction unit 807 is used to construct a distributed polarization mode dispersion compensation algorithm by utilizing the polarization mode dispersion value of each of the first segment links, the polarization principal axis direction, and the basic signal compensation algorithm;

[0193] The mapping unit 808 is used to map the distributed polarization mode dispersion compensation algorithm to a corresponding neural network to be trained.

[0194] The training unit 809 is used to train the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed polarization mode dispersion value of the optical fiber link under test.

[0195] It should be noted that the processing procedures and beneficial effects of the receiving unit 801, conversion unit 802, preprocessing unit 803, determining unit 804, partitioning unit 805, setting unit 806, construction unit 807, mapping unit 808, and training unit 809 can be referred to respectively. Figure 1 Steps S101 to S109 are not described in detail here.

[0196] Optionally, the training unit 809 is further configured to: use the preprocessed electrical signal and the original electrical signal as training samples and sample labels corresponding to the training samples, respectively; train the neural network to be trained using the training samples and the sample labels to obtain a trained target neural network, wherein the target neural network is used to detect the polarization mode dispersion values ​​of N second segment links in the fiber optic link under test, the length of the second segment link corresponding to the length of the first segment link; the polarization mode dispersion value of the second segment link is used to compare with the polarization mode dispersion reference value of the second segment link to determine whether the working state of the second segment link is abnormal, wherein the polarization mode dispersion reference value is obtained based on the target neural network and the fiber optic link under test in normal working state.

[0197] Optionally, when the length of each first segment link is the same, the setting unit 806 is further configured to: determine the position number of each first segment link, the position number being used to identify the position of the first segment link in each of the first segment links of the sample optical fiber link; calculate the ratio between the cumulative polarization mode dispersion value and N; for any first segment link, multiply the ratio by the position number of the first segment link to obtain the corresponding product and use it as the polarization mode dispersion value of the first segment link.

[0198] Optionally, the construction unit 807 is further configured to: perform a Fourier transform on the electrical signal variable to obtain a frequency domain signal variable; transform the frequency domain signal variable from the original coordinate axis direction to the polarization principal axis direction to obtain a principal axis signal variable; generate a polarization mode dispersion compensation factor corresponding to each first segment link based on the principal axis signal variable and the polarization mode dispersion value of each first segment link; and construct the distributed polarization mode dispersion compensation algorithm according to the polarization mode dispersion compensation factor corresponding to each first segment link and the basic signal compensation algorithm.

[0199] Optionally, when the basic signal compensation algorithm includes a frequency domain conversion module, a principal axis conversion module, a digital back-transmission module, a coordinate axis conversion module, and a time domain conversion module connected in sequence, the construction unit 807 is further configured to: construct a distributed polarization mode dispersion compensation module using the polarization mode dispersion compensation factor; and set the distributed polarization mode dispersion compensation module between the principal axis conversion module and the digital back-transmission module of the basic signal compensation algorithm to obtain the distributed polarization mode dispersion compensation algorithm, wherein the distributed polarization mode dispersion compensation algorithm includes the frequency domain conversion module, the principal axis conversion module, the distributed polarization mode dispersion compensation module, the digital back-transmission module, the coordinate axis conversion module, and the time domain conversion module connected in sequence.

[0200] Optionally, the distributed polarization mode dispersion compensation module is used to perform polarization mode dispersion compensation on the signal from the main axis conversion module according to the first polarization mode dispersion compensation factor to obtain the first polarization mode compensated signal; wherein, the first polarization mode dispersion compensation factor is the polarization mode dispersion compensation factor corresponding to the first segment link with position number 1.

[0201] The digital back-transmission module is used to perform dispersion compensation and nonlinear compensation on the first polarization mode compensated signal to obtain a first-stage compensated signal and return it to the distributed polarization mode dispersion compensation module. This allows the distributed polarization mode dispersion compensation module to perform polarization mode dispersion compensation on the first-stage compensated signal according to a second polarization mode dispersion compensation factor until polarization mode dispersion compensation is performed using the polarization mode dispersion compensation factors corresponding to all the segmented links, and the final stage compensated signal is obtained. The final stage compensated signal is then sent to the coordinate axis transformation module. The second polarization mode dispersion compensation factor is the polarization mode dispersion compensation factor corresponding to the first segmented link with position number 2.

[0202] Optionally, the neural network to be trained includes a frequency domain transformation layer, a principal axis transformation layer, a coordinate axis transformation layer, a time domain transformation layer, and multiple signal compensation layers;

[0203] The frequency domain transformation layer, the principal axis transformation layer, the coordinate axis transformation layer, and the time domain transformation layer correspond to the frequency domain transformation module, the principal axis transformation module, the coordinate axis transformation module, and the time domain transformation module, respectively.

[0204] Each of the signal compensation layers is used to perform polarization mode dispersion compensation, dispersion compensation and nonlinear compensation on the received signal, so as to realize a stage compensation based on the distributed polarization mode dispersion compensation module and the digital back transmission module.

[0205] The frequency domain conversion layer, the principal axis conversion layer, and the signal compensation layer used to implement the first stage compensation are connected in sequence; all the signal compensation layers are connected in sequence according to the stage compensation order, and the output of one signal compensation layer is the input of the next signal compensation layer; the signal compensation layer used to implement the last stage compensation, the coordinate axis conversion layer, and the time domain conversion layer are connected in sequence.

[0206] Optionally, each of the signal compensation layers includes a polarization mode dispersion estimation parameter to be trained, which is used to estimate the polarization mode dispersion compensation value; the principal axis transformation layer and the coordinate axis transformation layer include a principal axis azimuth angle and a principal axis elevation angle to be trained; the training unit 809 is further configured to: input the preprocessed electrical signal as a training sample into the neural network to be trained for polarization mode dispersion compensation, dispersion compensation and nonlinear compensation, to obtain the compensated signal output by the neural network to be trained; calculate a loss function value based on the difference between the compensated signal and the original electrical signal; and update the polarization mode dispersion estimation parameter, the principal axis azimuth angle and / or the principal axis elevation angle in the neural network to be trained based on the loss function value.

[0207] Optionally, the determining unit 804 is further configured to: sequentially perform polarization correction processing, orthogonalization processing, and dispersion compensation processing on the electrical signal to obtain a corresponding first processed signal; use an adaptive equalization filter to perform polarization demultiplexing and polarization mode dispersion compensation on the first processed signal to obtain a second processed signal and acquire multiple tap coefficients in the adaptive equalization filter; perform carrier phase recovery on the second processed signal to obtain a phase-recovered signal and a phase noise estimate; perform decision processing on the phase-recovered signal to obtain the original symbol signal; use a Nyquist filter to perform pulse shaping processing on the original symbol signal to obtain a pulse-shaped signal; apply phase noise to the pulse-shaped signal according to the phase noise estimate to obtain the original electrical signal; and use the multiple tap coefficients to determine the cumulative value of polarization mode dispersion and the direction of the polarization principal axis.

[0208] The distributed polarization mode dispersion (PDDM) detection network generation device proposed in this embodiment can receive the end optical signal of a sample fiber optic link and convert it into a corresponding electrical signal; preprocess the electrical signal to obtain a preprocessed electrical signal, and determine the original electrical signal, the cumulative value of polarization mode dispersion (PMD) of the sample fiber optic link, and the direction of the polarization principal axis based on the electrical signal; divide the sample fiber optic link into N first segment links, and set the PMD value of each first segment link based on the cumulative PMD value; construct a distributed PMD compensation algorithm using the PMD value, the direction of the polarization principal axis, and the basic signal compensation algorithm of each first segment link; map the distributed PMD compensation algorithm to a corresponding neural network to be trained; train the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed PMD value of the fiber optic link under test. This embodiment can generate a target neural network for distributed PMD detection of the fiber optic link under test, enabling the detection of local PMD of the fiber optic link and increasing the diversity of methods for detecting local PMD of the fiber optic link.

[0209] In this embodiment, the distributed polarization mode dispersion detection network generation device is presented in the form of a functional unit. Here, a unit refers to an application-specific integrated circuit (ASIC), a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0210] This invention also provides a computer device having the above-described features. Figure 8 The distributed polarization mode dispersion detection network generation device shown is shown.

[0211] Please see Figure 9 , Figure 9This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 9 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 9 Take a processor 10 as an example.

[0212] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0213] The memory 20 stores instructions executable by at least one processor 10 to cause at least one processor 10 to perform the method shown in the above embodiments.

[0214] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function. The data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, which can be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0215] Memory 20 may include volatile memory, such as random access memory. Memory may also include non-volatile memory, such as flash memory, hard disk, or solid-state drive. Memory 20 may also include combinations of the above types of memory.

[0216] The computer device also includes a communication interface 30 for communicating with other devices or communication networks.

[0217] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0218] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for generating a distributed polarization mode dispersion detection network, characterized in that, The method includes: Receive the optical signal at the end of the sample fiber optic link and convert it into a corresponding electrical signal; The electrical signal is preprocessed to obtain a preprocessed electrical signal, and the original electrical signal, the cumulative polarization mode dispersion value of the sample fiber link, and the polarization principal axis direction are determined based on the electrical signal. The sample fiber optic link is divided into N first segment links, and the polarization mode dispersion value of each first segment link is set based on the cumulative value of polarization mode dispersion. A distributed polarization mode dispersion compensation algorithm is constructed by utilizing the polarization mode dispersion value of each first segment link, the polarization principal axis direction, and the basic signal compensation algorithm. The distributed polarization mode dispersion compensation algorithm is mapped to a corresponding neural network to be trained. The neural network to be trained is trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed polarization mode dispersion value of the optical fiber link under test. The step of training the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network includes: The preprocessed electrical signal and the original electrical signal are used as training samples and sample labels corresponding to the training samples, respectively. The training samples and the sample labels are used to train the neural network to obtain a trained target neural network. The target neural network is used to detect the polarization mode dispersion values ​​of N second segment links in the fiber optic link under test. The length of the second segment link corresponds to the length of the first segment link. The polarization mode dispersion value of the second segment link is compared with the polarization mode dispersion reference value of the second segment link to determine whether the working state of the second segment link is abnormal. The polarization mode dispersion reference value is obtained based on the target neural network and the fiber optic link under test in normal working state. Wherein, when the length of each of the first segment links is the same, setting the polarization mode dispersion value of each of the first segment links based on the cumulative polarization mode dispersion value includes: Determine the position number of each first segment link, the position number being used to identify the position of the first segment link in each of the first segment links of the sample optical fiber link; Calculate the ratio between the cumulative polarization mode dispersion value and N; For any of the first segment links, the ratio is multiplied by the position number of the first segment link to obtain the corresponding product, which is used as the polarization mode dispersion value of the first segment link.

2. The method according to claim 1, characterized in that, The step of constructing a distributed polarization mode dispersion compensation algorithm using the polarization mode dispersion value of each of the first segment links, the polarization principal axis direction, and the basic signal compensation algorithm includes: Perform a Fourier transform on the electrical signal variable to obtain the frequency domain signal variable; The frequency domain signal variable is transformed from the original coordinate axis direction to the polarization principal axis direction to obtain the principal axis signal variable; Based on the main axis signal variable and the polarization mode dispersion value of each first segment link, a polarization mode dispersion compensation factor is generated for each first segment link. The distributed polarization mode dispersion compensation algorithm is constructed based on the polarization mode dispersion compensation factor and the basic signal compensation algorithm corresponding to each of the first segment links.

3. The method according to claim 2, characterized in that, When the basic signal compensation algorithm includes a frequency domain conversion module, a principal axis conversion module, a digital back-transmission module, a coordinate axis conversion module, and a time domain conversion module connected in sequence, the step of constructing the distributed polarization mode dispersion compensation algorithm based on the polarization mode dispersion compensation factor corresponding to each first segment link and the basic signal compensation algorithm includes: A distributed polarization mode dispersion compensation module is constructed using the polarization mode dispersion compensation factor. The distributed polarization mode dispersion compensation module is placed between the main axis conversion module and the digital back transmission module of the basic signal compensation algorithm to obtain the distributed polarization mode dispersion compensation algorithm. The distributed polarization mode dispersion compensation algorithm includes the frequency domain conversion module, the main axis conversion module, the distributed polarization mode dispersion compensation module, the digital back transmission module, the coordinate axis conversion module, and the time domain conversion module connected in sequence.

4. The method according to claim 3, characterized in that, The distributed polarization mode dispersion compensation module is used to perform polarization mode dispersion compensation on the signal from the main axis conversion module according to the first polarization mode dispersion compensation factor to obtain the first polarization mode compensated signal; wherein, the first polarization mode dispersion compensation factor is the polarization mode dispersion compensation factor corresponding to the first segment link with position number 1. The digital back-transmission module is used to perform dispersion compensation and nonlinear compensation on the first polarization mode compensated signal to obtain a first-stage compensated signal and return it to the distributed polarization mode dispersion compensation module. This allows the distributed polarization mode dispersion compensation module to perform polarization mode dispersion compensation on the first-stage compensated signal according to a second polarization mode dispersion compensation factor until polarization mode dispersion compensation is performed using the polarization mode dispersion compensation factors corresponding to all the segmented links, and the final stage compensated signal is obtained. The final stage compensated signal is then sent to the coordinate axis transformation module. The second polarization mode dispersion compensation factor is the polarization mode dispersion compensation factor corresponding to the first segmented link with position number 2.

5. The method according to claim 4, characterized in that, The neural network to be trained includes a frequency domain transformation layer, a principal axis transformation layer, a coordinate axis transformation layer, a time domain transformation layer, and multiple signal compensation layers; The frequency domain transformation layer, the principal axis transformation layer, the coordinate axis transformation layer, and the time domain transformation layer correspond to the frequency domain transformation module, the principal axis transformation module, the coordinate axis transformation module, and the time domain transformation module, respectively. Each of the signal compensation layers is used to perform polarization mode dispersion compensation, dispersion compensation and nonlinear compensation on the received signal, so as to realize a stage compensation based on the distributed polarization mode dispersion compensation module and the digital back transmission module. The frequency domain conversion layer, the principal axis conversion layer, and the signal compensation layer used to implement the first stage compensation are connected in sequence; all the signal compensation layers are connected in sequence according to the stage compensation order, and the output of one signal compensation layer is the input of the next signal compensation layer; the signal compensation layer used to implement the last stage compensation, the coordinate axis conversion layer, and the time domain conversion layer are connected in sequence.

6. The method according to claim 5, characterized in that, Each of the signal compensation layers includes polarization mode dispersion estimation parameters to be trained, which are used to estimate polarization mode dispersion compensation values; the principal axis transformation layer and the coordinate axis transformation layer include principal axis azimuth and principal axis elevation angles to be trained; The step of training the neural network to be trained based on the preprocessed electrical signal and the original electrical signal includes: The preprocessed electrical signal is used as a training sample and input into the neural network to be trained for polarization mode dispersion compensation, dispersion compensation and nonlinear compensation, so as to obtain the compensated signal output by the neural network to be trained. The loss function value is calculated based on the difference between the compensated signal and the original electrical signal; The polarization mode dispersion estimation parameters, principal axis azimuth angle, and / or principal axis elevation angle in the neural network to be trained are updated based on the loss function value.

7. The method according to any one of claims 1 to 6, characterized in that, The step of determining the original electrical signal, the cumulative polarization mode dispersion value of the sample fiber link, and the polarization principal axis direction based on the electrical signal includes: The electrical signal is sequentially subjected to bias correction processing, orthogonalization processing, and dispersion compensation processing to obtain the corresponding first processed signal; The first processed signal is polarization demultiplexed and polarization mode dispersion compensated using an adaptive equalization filter to obtain the second processed signal and to acquire multiple tap coefficients in the adaptive equalization filter. Carrier phase recovery is performed on the second processed signal to obtain the phase-recovered signal and phase noise estimate; The phase-recovered signal is processed to obtain the original symbol signal; The original symbol signal is pulse-shaped using a Nyquist filter to obtain a pulse-shaped signal. The pulse-shaped signal is subjected to phase noise loading based on the phase noise estimate to obtain the original electrical signal; The cumulative value of polarization mode dispersion and the direction of the principal polarization axis are determined using the plurality of tap coefficients.

8. A distributed polarization mode dispersion detection network generation device, characterized in that, The device includes: The receiving unit is used to receive the optical signal at the end of the sample fiber optic link; A conversion unit is used to convert the terminal optical signal into a corresponding electrical signal; A preprocessing unit is used to preprocess the electrical signal to obtain a preprocessed electrical signal; The determining unit is used to determine the original electrical signal, the cumulative polarization mode dispersion value of the sample optical fiber link, and the polarization principal axis direction based on the electrical signal. A partitioning unit is used to divide the sample optical fiber link into N first segment links; The setting unit is used to set the polarization mode dispersion value of each first segment link based on the accumulated polarization mode dispersion value; The construction unit is used to construct a distributed polarization mode dispersion compensation algorithm by utilizing the polarization mode dispersion value of each of the first segment links, the polarization principal axis direction, and the basic signal compensation algorithm; A mapping unit is used to map the distributed polarization mode dispersion compensation algorithm to a corresponding neural network to be trained. The training unit is used to train the neural network to be trained based on the preprocessed electrical signal and the original electrical signal to obtain a trained target neural network, which is used to detect the distributed polarization mode dispersion value of the optical fiber link under test. The training unit is also used for: The preprocessed electrical signal and the original electrical signal are used as training samples and sample labels corresponding to the training samples, respectively. The training samples and the sample labels are used to train the neural network to obtain a trained target neural network. The target neural network is used to detect the polarization mode dispersion values ​​of N second segment links in the fiber optic link under test. The length of the second segment link corresponds to the length of the first segment link. The polarization mode dispersion value of the second segment link is compared with the polarization mode dispersion reference value of the second segment link to determine whether the working state of the second segment link is abnormal. The polarization mode dispersion reference value is obtained based on the target neural network and the fiber optic link under test in normal working state. Wherein, when the length of each of the first segment links is the same, the setting unit is further configured to: Determine the position number of each first segment link, the position number being used to identify the position of the first segment link in each of the first segment links of the sample optical fiber link; Calculate the ratio between the cumulative polarization mode dispersion value and N; For any of the first segment links, the ratio is multiplied by the position number of the first segment link to obtain the corresponding product, which is used as the polarization mode dispersion value of the first segment link.

Images