An integrated optical signal receiver and method

Abstract

The present invention discloses an integrated optical signal receiver and method. Specifically, the present invention relates to the field of optical communication, and is used to solve the problem of the degradation of the video stream transmission quality caused by the distortion of optical signals in an environment with a low optical signal-to-noise ratio. By constructing a dynamic polarization state transfer matrix to eliminate the cross-coupling caused by the birefringence effect, a clear independent polarization state electrical signal sequence is generated; performing time-frequency domain processing on these electrical signal sequences to compensate for the linear phase noise caused by the laser linewidth and the phase shift caused by the fiber nonlinear effect, and outputting signals with higher phase accuracy; subsequently mapping these signals to a four-dimensional constellation space, using a clustering method to generate an adaptive decision boundary, optimizing symbol decision and enhancing the noise resistance ability; through the collaborative work of the inner and outer loops, dynamically adjusting system parameters according to the signal characteristics, enhancing the processing stability and adaptability; finally performing soft decision decoding on the signals, and combining phase residual correction to improve the decoding accuracy, and outputting the reconstructed video data stream.

Description

Technical Field

[0001] The present invention relates to the field of optical communication, and more particularly, to an integrated optical signal receiver and method. Background Art

[0002] In an integrated optical signal receiver for high-speed wireless video transmission, the optical communication module needs to process polarization multiplexed high-order modulation signals to achieve ultra-high-definition video stream transmission. However, in a low optical signal-to-noise ratio environment, the spontaneous emission noise of the optical amplifier is coupled with the phase noise caused by fiber nonlinear effects (such as the Kerr effect and the birefringence effect), resulting in asymmetric distortion of the signal in the joint space of polarization and phase. The limited linewidth of the laser and the nonlinear phase rotation caused by the fiber Kerr effect cause the constellation points to randomly drift around the origin, destroying the determinacy of the symbol phase information; the polarization state crosstalk caused by the birefringence effect leads to cross-coupling of the electric field components of two orthogonal polarization states, further blurring the clustering boundary of the constellation points. The traditional carrier recovery algorithm based on the phase-locked loop tracking mechanism under the assumption of signal phase stationarity loses lock under low optical signal-to-noise ratio conditions because the phase jump rate dominated by noise exceeds the loop bandwidth, resulting in irreversible dispersion of the clustering center and overlapping of the decision regions in both the polarization and phase dimensions of the constellation diagram, causing a non-linear steep rise in the bit error rate and severely restricting the effective transmission distance and marginal performance reliability of the high-order modulation system.

[0003] To solve the above problems, a technical solution is provided now. Summary of the Invention

[0004] To overcome the above-mentioned defects of the prior art, embodiments of the present invention provide an integrated optical signal receiver and method. By constructing a dynamic polarization state transfer matrix to eliminate the cross-coupling caused by the birefringence effect and generating a clear independent polarization state electrical signal sequence, it lays a foundation for subsequent processing; then performing time-frequency domain processing on these electrical signal sequences to compensate for the linear phase noise caused by the laser linewidth and the phase shift caused by the fiber nonlinear effect, and outputting a signal with higher phase accuracy; subsequently mapping these signals to a four-dimensional constellation space and using a clustering method to generate an adaptive decision boundary to further optimize symbol decision and improve the noise resistance ability; through the collaborative work of the inner and outer loops, dynamically adjusting system parameters according to the signal characteristics to enhance the processing stability and adaptability; finally performing soft decision decoding on the signal and combining phase residual correction to improve the decoding accuracy and outputting a reconstructed video data stream to solve the problems raised in the above background art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An integrated optical signal receiving method, comprising the steps:

[0007] S1: Separate the two orthogonally polarized signals transmitted through the optical fiber, construct a dynamic polarization state transfer matrix, and output an independent polarization state electrical signal sequence;

[0008] S2: Perform time-frequency domain processing on the independent polarization state electrical signal sequence and output a phase compensation signal;

[0009] S3: Map the phase compensation signal to a four-dimensional constellation space, perform constellation point clustering, generate a decision boundary and update the equalizer tap coefficients, and output an equalized signal;

[0010] S4: Under the collaborative mechanism of the outer and inner loops, the outer loop updates the polarization state transfer matrix according to the dispersion degree of the clustering center, and the inner loop adjusts the phase compensation step size based on the symbol error rate;

[0011] S5: Perform belief propagation-assisted soft decision decoding on the equalized signal, construct a phase residual model to correct the symbol confidence, and output the reconstructed video data stream.

[0012] In a preferred embodiment, step S1 includes the following:

[0013] First, receive the complex electric field components of the X polarization state and the Y polarization state from the optical fiber transmission, and convert the complex electric field components of the X polarization state and the Y polarization state into a discrete signal sequence of the X polarization state and a discrete signal sequence of the Y polarization state through analog-to-digital conversion;

[0014] Then, for the birefringence effect in the optical fiber, introduce a polarization state transfer matrix to describe the dynamic cross-coupling relationship between the X polarization state and the Y polarization state. The polarization state transfer matrix is a two-dimensional complex matrix and is initialized according to prior measurement data;

[0015] Next, apply the Kalman filtering algorithm to dynamically estimate the polarization state transfer matrix. By defining the state vector, state transition model, and observation model, use the prediction step and update step to iteratively optimize the estimated value of the polarization state transfer matrix;

[0016] Finally, use the inverse matrix of the estimated polarization state transfer matrix to decouple the discrete signal sequence of the X polarization state and the discrete signal sequence of the Y polarization state, and output an independent polarization state electrical signal sequence, including an independent electrical signal sequence of the X polarization state and an independent electrical signal sequence of the Y polarization state.

[0017] In a preferred embodiment, step S2 includes the following:

[0018] Estimate the phase drift rate through adaptive filtering technology, and use the estimated phase drift rate to reversely adjust the phase components of the X polarization state independent electrical signal sequence and the Y polarization state independent electrical signal sequence to generate the X polarization state signal after time domain compensation and the Y polarization state signal after time domain compensation.

[0019] In a preferred embodiment, step S2 further includes the following:

[0020] The time-domain compensated X-polarization state signal and the time-domain compensated Y-polarization state signal are converted into frequency-domain signals through fast Fourier transform. The split-step Fourier method is used to estimate the nonlinear phase shift. The phase of the frequency-domain signal is adjusted inversely using the estimated nonlinear phase shift to generate the frequency-domain compensated X-polarization state signal and the frequency-domain compensated Y-polarization state signal. Then, the frequency-domain compensated X-polarization state signal and the frequency-domain compensated Y-polarization state signal are converted back to the time domain through inverse fast Fourier transform to obtain the phase-compensated X-polarization state signal and the phase-compensated Y-polarization state signal; the phase-compensated X-polarization state signal and the phase-compensated Y-polarization state signal are integrated into a two-dimensional complex signal sequence.

[0021] In a preferred embodiment, step S3 includes the following:

[0022] The X-polarization state phase-compensated complex signal and the Y-polarization state phase-compensated complex signal are decomposed into the real part of the X-polarization state, the imaginary part of the X-polarization state, the real part of the Y-polarization state, and the imaginary part of the Y-polarization state. The real part of the X-polarization state, the imaginary part of the X-polarization state, the real part of the Y-polarization state, and the imaginary part of the Y-polarization state at each sampling point are combined into a four-dimensional vector to form a signal point set in the four-dimensional constellation space; then, the fuzzy C-means algorithm is applied to the signal point set in the four-dimensional constellation space for clustering. The number of clusters is set according to the modulation format order, the cluster centers are initialized, the membership degrees of each signal point to the cluster centers are calculated and the cluster centers are updated, and the iteration continues until the cluster centers converge to generate a set of cluster centers.

[0023] In a preferred embodiment, step S3 further includes the following:

[0024] Then, the perpendicular bisecting hyperplane between each pair of cluster centers is calculated based on the set of cluster centers to form a set of decision boundaries; subsequently, the equalizer tap coefficients are defined, and the equalizer tap coefficients are updated through the least mean square error objective function and the gradient descent method, and the iteration is optimized until the objective function converges to obtain the optimal equalizer tap coefficients; then, the four-dimensional signal points are equalized using the optimal equalizer tap coefficients to generate the equalized four-dimensional signal, and the equalized four-dimensional signal is converted into an equalized signal including the X-polarization state complex signal and the Y-polarization state complex signal.

[0025] In a preferred embodiment, step S4 includes the following:

[0026] Calculate the dispersion of the clustering center set in the outer loop. By comparing the dispersion with a preset threshold, update the polarization state transfer matrix using the gradient descent method. Then, perform a hard decision on the equalized signal in the inner loop, calculate the symbol error rate. By comparing the symbol error rate with a preset threshold, adjust the time-domain phase compensation step size and the frequency-domain phase compensation step size using an adaptive strategy. Subsequently, feedback the updated polarization state transfer matrix in the outer loop to step S1 for the next round of polarization state decoupling processing. Feedback the adjusted time-domain phase compensation step size and the frequency-domain phase compensation step size in the inner loop to step S2 for the next round of phase noise compensation processing.

[0027] In a preferred embodiment, step S5 includes the following:

[0028] Perform belief propagation-assisted soft decision decoding on the equalized signal to generate a set of soft information of symbols. Based on the equalized signal and the set of soft information of symbols, construct a phase residual model, calculate the actual phase, the estimated phase, and the phase residual, and optimize and solve the parameters of the phase residual model. Use the phase residual model to adjust the set of soft information of symbols, calculate the phase residual prediction value, and correct the soft information of symbols. Perform a hard decision on the corrected set of soft information of symbols to generate an estimated symbol sequence, convert it into a binary bit stream, and perform decoding and deinterleaving operations to output the reconstructed video data stream.

[0029] An integrated optical signal receiver includes: a polarization decoupling unit, a phase compensation unit, a constellation equalization unit, a loop optimization unit, and a soft decision decoding unit;

[0030] Polarization decoupling unit: Separate two orthogonally polarized state signals transmitted through an optical fiber, use the Kalman filtering algorithm to construct a dynamic polarization state transfer matrix, eliminate the cross-coupling caused by the birefringence effect, and output an independent polarized state electrical signal sequence;

[0031] Phase compensation unit: Perform time-frequency domain processing on the independent polarized state electrical signal sequence, compensate for the linear phase noise caused by the laser linewidth and the non-linear phase shift caused by the fiber Kerr effect, and output a phase compensation signal;

[0032] Constellation equalization unit: Map the phase compensation signal to a four-dimensional constellation space, use the fuzzy C-means algorithm to perform constellation point clustering, generate an adaptive decision boundary, and update the equalizer tap coefficients, and output an equalized signal;

[0033] Loop optimization unit: Update the polarization state transfer matrix according to the dispersion of the clustering center, and adjust the phase compensation step size based on the symbol error rate;

[0034] Soft decision decoding unit: Perform belief propagation-assisted soft decision decoding on the equalized signal, construct a phase residual model to correct the symbol confidence, and output the reconstructed video data stream.

[0035] Technical effects and advantages of an integrated optical signal receiver and method of the present invention:

[0036] By constructing a dynamic polarization state transfer matrix to eliminate the cross-coupling caused by the birefringence effect, a clear sequence of independent polarization state electrical signals is generated, laying a foundation for subsequent processing; then, time-frequency domain processing is performed on these electrical signal sequences to compensate for the linear phase noise caused by the laser linewidth and the phase shift caused by the fiber nonlinear effect, and signals with higher phase accuracy are output; subsequently, these signals are mapped into a four-dimensional constellation space, and a clustering method is used to generate an adaptive decision boundary to further optimize symbol decision and improve the noise resistance; through the collaborative work of the inner and outer loops, the system parameters are dynamically adjusted according to the signal characteristics to enhance the processing stability and adaptability; finally, soft decision decoding is performed on the signals, and the decoding accuracy is improved by combining phase residual correction, and the reconstructed video data stream is output. Description of the Drawings

[0037] Figure 1 It is a schematic flowchart of an integrated optical signal receiving method of the present invention.

[0038] Figure 2 It is a schematic structural diagram of an integrated optical signal receiver of the present invention. Detailed Embodiments

[0039] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0040] Embodiment 1: Figure 1 An integrated optical signal receiving method of the present invention is given, including:

[0041] S1: Separate two orthogonally polarized state signals transmitted through the optical fiber, and use the Kalman filtering algorithm to construct a dynamic polarization state transfer matrix to output a sequence of independent polarization state electrical signals.

[0042] S2: Perform time-frequency domain processing on the sequence of independent polarization state electrical signals to output phase compensation signals.

[0043] S3: Map the phase compensation signals into a four-dimensional constellation space, perform constellation point clustering, generate a decision boundary and update the equalizer tap coefficients, and output equalized signals.

[0044] S4: Under the collaborative mechanism of the outer and inner loops, the outer loop updates the polarization state transfer matrix according to the dispersion degree of the clustering center, and the inner loop adjusts the phase compensation step size based on the symbol error rate.

[0045] S5: Perform belief propagation-assisted soft decision decoding on the equalized signal, construct a phase residual model to correct the symbol confidence, and output the reconstructed video data stream.

[0046] In a high-speed wireless video transmission chip, the optical communication module needs to process polarization multiplexed high-order modulation signals to achieve the transmission of ultra-high-definition video streams. According to the invention background, in a low optical signal-to-noise ratio environment, the spontaneous emission noise of the optical amplifier and the phase noise caused by the fiber nonlinear effect are coupled with each other, resulting in asymmetric distortion of the signal in the joint polarization and phase space. Specifically, the birefringence effect in the optical fiber transmission link causes cross-coupling of the electric field components of two orthogonal polarization states, making the clustering boundary of the received signal constellation points blurred and seriously affecting the accuracy of symbol decision. To solve this problem, the processing solution proposes step S1, which separates the two orthogonal polarization state signals transmitted through the optical fiber and constructs a dynamic polarization state transfer matrix using the Kalman filter algorithm to eliminate the cross-coupling caused by the birefringence effect and output an independent polarization state electrical signal sequence.

[0047] Step S1 includes the following:

[0048] S1. Signal reception and preprocessing:

[0049] First, receive two orthogonal polarization state signals from the optical fiber transmission link. These two signals are respectively represented as the complex electric field components of the X polarization state and the Y polarization state. Perform analog-to-digital conversion processing on these two orthogonal polarization state signals to convert them from continuous-time signals to discrete-time sequences. Specifically, through the sampling operation, the complex electric field component of the X polarization state is converted into a discrete signal sequence of the X polarization state, and the complex electric field component of the Y polarization state is converted into a discrete signal sequence of the Y polarization state. Each sampling point is identified by a time index. After completing the analog-to-digital conversion, the discrete signal sequences of the X polarization state and the Y polarization state become the basic inputs for subsequent digital signal processing. The purpose of converting the continuous-time signal into a discrete-time sequence is to achieve the digital representation of the signal, which is convenient for applying precise signal processing techniques in the digital domain, such as filtering and matrix operations.

[0050] S2. Polarization state transfer matrix modeling:

[0051] In the polarization state transfer matrix modeling stage, aiming at the cross-coupling phenomenon between the X polarization state and the Y polarization state caused by the birefringence effect in optical fiber transmission, this dynamic relationship is described by the polarization state transfer matrix. The polarization state transfer matrix is a two-dimensional complex matrix with both the number of rows and columns being 2, which is used to represent the coupling characteristics between the X polarization state and the Y polarization state. Specifically, each element of the polarization state transfer matrix is a complex number, which respectively characterizes the coupling amplitude and phase change from the X polarization state to the X polarization state, from the X polarization state to the Y polarization state, from the Y polarization state to the X polarization state, and from the Y polarization state to the Y polarization state. In the initial stage, the polarization state transfer matrix can be set as an identity matrix, that is, assuming no coupling between the X polarization state and the Y polarization state, or the matrix elements can be initialized according to prior measurement data to be closer to the actual coupling state.

[0052] The purpose of introducing the polarization state transfer matrix is to quantitatively describe the polarization state cross-coupling caused by the birefringence effect of optical fiber through a mathematical model, providing a theoretical basis for signal decoupling. This modeling method can transform complex physical phenomena into a computable matrix form, facilitating the application of digital signal processing technology. At the same time, the flexibility of initialization enables the model to adapt to different optical fiber transmission conditions, improving the adaptability and accuracy of processing.

[0053] S3. Application of Kalman filtering algorithm:

[0054] In the application stage of the Kalman filtering algorithm, in order to dynamically estimate the polarization state transfer matrix, first, the four complex elements of the polarization state transfer matrix are flattened into a column vector, which is called the state vector. Then, the state transition model is defined, assuming that the state vector is only affected by process noise between adjacent time indices, and the process noise has a zero mean and follows a Gaussian distribution characteristic. Next, the observation model is defined, indicating that the received discrete signal sequences of the X polarization state and the Y polarization state are the results of the polarization state transfer matrix acting on the independent polarization state signals at the transmitter end, and then adding observation noise, where the observation noise also has a zero mean and follows a Gaussian distribution characteristic. Since the independent polarization state signals at the transmitter end are unknown, the initial estimate can be obtained by iterative calculation through the decoupling result or reference signal of the previous time index. Based on the above models, the Kalman filtering algorithm dynamically updates the estimated value of the state vector by combining the state transition model and the observation model, thereby realizing the real-time optimization of the polarization state transfer matrix.

[0055] The purpose of adopting the Kalman filtering algorithm is to utilize its state estimation ability in time-varying systems, and dynamically track the change of the polarization state transfer matrix by fusing the estimation results of the previous time index and the observation data of the current time index.

[0056] S4. Kalman filtering iteration:

[0057] In the Kalman filter iteration stage, the estimation process of the polarization state transfer matrix is divided into two parts: the prediction step and the update step. In the prediction step, the state vector value at the current time index is predicted using the estimated value of the state vector at the previous time index, and the covariance matrix of the prediction error is calculated. The covariance matrix characterizes the uncertainty degree of the prediction result. In the update step, according to the discrete signal sequences of the X polarization state and the Y polarization state received at the current time index, the Kalman gain is calculated. The Kalman gain is a weighting factor used to balance the influence of the prediction result and the observed data. When calculating the Kalman gain, it depends on the observation matrix, which is determined by the linearization approximation of the independent polarization state signals at the transmitting end. Subsequently, the estimated value of the state vector is updated using the Kalman gain, and the error covariance matrix is updated. By repeatedly executing the prediction step and the update step, the estimated value of the polarization state transfer matrix gradually approaches the true value.

[0058] The purpose of the Kalman filter iteration process is to gradually reduce the error of the estimated value of the polarization state transfer matrix through cyclic optimization, ensuring that the estimated result can accurately reflect the dynamic coupling characteristics in optical fiber transmission. This iterative method can effectively cope with time-varying noise and nonlinear effects, improve the stability and accuracy of the decoupling process, and provide high-quality matrix parameters for subsequent steps.

[0059] S5. Polarization state decoupling:

[0060] In the polarization state decoupling stage, the discrete signal sequences of the X polarization state and the Y polarization state received are decoupled using the polarization state transfer matrix estimated by Kalman filter iteration. Specifically, by calculating the inverse matrix of the polarization state transfer matrix, the coupling effect of the received signal is eliminated reversely, and independent electrical signal sequences of the X polarization state and the Y polarization state are generated through conversion. The decoupling process is based on the principle of linear algebra, that is, through inverse matrix operation, the received signal is restored to the independent polarization state signals at the transmitting end without cross-coupling influence. After decoupling, independent polarization state electrical signal sequences are output, including the independent electrical signal sequence of the X polarization state and the independent electrical signal sequence of the Y polarization state.

[0061] The purpose of performing polarization state decoupling is to eliminate the cross-coupling between the X polarization state and the Y polarization state caused by the fiber birefringence effect, enabling subsequent signal processing to operate on each independent polarization state separately. This decoupling method not only improves the efficiency of signal processing but also significantly enhances the accuracy of symbol decision, laying a foundation for the reliability of ultra-high-definition video stream transmission.

[0062] In a low optical signal-to-noise ratio environment, the spontaneous emission noise of the optical amplifier and the phase noise caused by the fiber nonlinear effect are coupled with each other, resulting in asymmetric distortion of the signal in the joint space of polarization and phase. The limited linewidth of the laser causes linear phase drift, and the fiber Kerr effect causes nonlinear phase rotation, making the constellation points randomly drift around the origin, destroying the determinacy of the symbol phase information. Step S1 has eliminated the cross-coupling caused by the birefringence effect through the Kalman filtering algorithm and output an independent polarization state electrical signal sequence. Step S2 is based on the output of Step S1 and performs time-frequency domain processing on the linear and nonlinear phase noises to output a phase compensation signal.

[0063] Step S2 includes the following:

[0064] S2.1, Time-domain linear phase noise compensation:

[0065] In the time-domain linear phase noise compensation stage, the linear phase noise caused by the laser linewidth is processed. The linear phase noise is manifested as a linear drift of the signal phase over time, directly affecting the accuracy of the symbol phase information. First, the phase extraction is performed on the independent electrical signal sequences of the X polarization state and the Y polarization state respectively to obtain the phase components of each signal sequence. The phase component consists of three parts: the initial phase, the phase drift rate, and the residual noise, where the phase drift rate is caused by the laser linewidth and is the main source of the linear phase drift. Then, the phase drift rate is estimated through adaptive filtering technology. The specific method is to analyze the relationship between the phase component and the time index and calculate the average rate of change of the phase over time. Then, the phase components of the independent electrical signal sequences of the X polarization state and the Y polarization state are adjusted backward using the estimated phase drift rate to generate the X polarization state signal after time-domain compensation and the Y polarization state signal after time-domain compensation.

[0066] For example, the processing process is as follows:

[0067] The independent electrical signal sequence of the X polarization state and the independent electrical signal sequence of the Y polarization state are respectively processed in the time domain.

[0068] For each polarization state signal , calculate its phase .

[0069] Assume the phase noise model is , where:

[0070] represents the initial phase;

[0071] represents the phase drift rate, caused by the laser linewidth;

[0072] Indicates the residual noise.

[0073] Use a Wiener filter to estimate the phase drift rate , and compensate for the phase drift through the following formula:

[0074] ;

[0075] where is the estimated phase drift rate.

[0076] Output: The signal after time-domain compensation and .

[0077] The purpose of time-domain linear phase noise compensation is to restore the phase stability of the independent electrical signal sequences of the X polarization state and the Y polarization state by estimating and canceling the phase drift rate.

[0078] S2.2, Frequency-domain non-linear phase offset cancellation:

[0079] In the frequency-domain non-linear phase offset cancellation stage, compensation is performed for the non-linear phase offset caused by the fiber Kerr effect. The non-linear phase offset is closely related to the instantaneous power of the signal, resulting in non-linear rotation of the signal phase. First, the time-domain compensated X-polarization state signal and the time-domain compensated Y-polarization state signal are converted to the frequency domain through fast Fourier transform, respectively obtaining the frequency-domain X-polarization state signal and the frequency-domain Y-polarization state signal. Then, the split-step Fourier method is used to simulate the reverse process of fiber transmission to estimate the non-linear phase offset. Specifically, the magnitude of the non-linear phase offset is jointly determined by the fiber non-linear coefficient, the effective fiber length, and the square of the signal power spectrum, and the phase rotation amount of each frequency component is obtained through numerical calculation. Then, the estimated non-linear phase offset is used to inversely adjust the phases of the frequency-domain X-polarization state signal and the frequency-domain Y-polarization state signal to generate the frequency-domain compensated X-polarization state signal and the frequency-domain compensated Y-polarization state signal. Finally, the frequency-domain compensated X-polarization state signal and the frequency-domain compensated Y-polarization state signal are converted back to the time domain through inverse fast Fourier transform, respectively obtaining the phase-compensated X-polarization state signal and the phase-compensated Y-polarization state signal.

[0080] For example, the processing procedure is as follows:

[0081] Convert the time-domain compensated signal to the frequency domain:

[0082] ;

[0083] where represents the frequency index.

[0084] Estimate the non - linear phase shift , adopt the split - step Fourier method (SSFM) to simulate the reverse process of optical fiber transmission, and the calculation formula is:

[0085] ;

[0086] Where:

[0087] represents the optical fiber non - linear coefficient;

[0088] represents the effective optical fiber length;

[0089] represents the signal power spectrum.

[0090] Cancel the non - linear phase shift:

[0091] ;

[0092] Convert the signal after frequency - domain compensation back to the time domain:

[0093] ;

[0094] Output: phase - compensated signal and .

[0095] The purpose of non - linear phase - shift cancellation in the frequency domain is to accurately eliminate the non - linear phase - rotation effect by simulating the reverse process of optical fiber transmission. As an efficient numerical method, the split - step Fourier method can accurately estimate the non - linear phase shift and improve the compensation accuracy. Combining with the processing method of time - domain compensation can comprehensively solve the phase - noise problem.

[0096] S2.3, Phase - compensated signal integration:

[0097] In the phase - compensated signal integration stage, integrate the phase - compensated X - polarization state signal and the phase - compensated Y - polarization state signal into a composite signal sequence. Specifically, align the phase - compensated X - polarization state signal and the phase - compensated Y - polarization state signal according to the time index, and combine them into a two - dimensional complex signal sequence, where each time index corresponds to a vector containing X - polarization state and Y - polarization state signal components. The integrated two - dimensional complex signal sequence is convenient for unified processing and management in subsequent steps.

[0098] The purpose of integrating the phase - compensated X - polarization state signal and the phase - compensated Y - polarization state signal is to provide a standardized input format for subsequent constellation - space mapping and clustering algorithms, ensure the unity and efficiency of signal processing, reduce data redundancy, and improve processing efficiency.

[0099] Step S2 receives the independent polarization state electrical signal sequence output by step S1, compensates the linear phase noise caused by the laser linewidth through a time-domain Wiener filter, and then cancels the nonlinear phase shift caused by the fiber Kerr effect through a frequency-domain split-step Fourier method, and finally outputs a phase compensation signal. This process effectively alleviates the problem of constellation point drift caused by phase noise in a low optical signal-to-noise ratio environment, provides a stable signal input for the four-dimensional constellation space mapping and fuzzy C-means clustering in step S3, and ensures the accuracy of the phase information for ultra-high-definition video stream transmission.

[0100] Step S3 receives the phase compensation signal output by step S2, and outputs an equalization signal through mapping to a four-dimensional constellation space, constellation point clustering, decision boundary generation, and equalizer coefficient update.

[0101] Step S3 includes the following:

[0102] S3.1, Signal mapping to a four-dimensional constellation space:

[0103] In the stage of signal mapping to a four-dimensional constellation space, the phase compensation signal from step S2 is received. This phase compensation signal consists of a phase compensation complex signal in the X polarization state and a phase compensation complex signal in the Y polarization state, and each signal is identified by a sampling point index. First, the phase compensation complex signal in the X polarization state is decomposed into the real part of the X polarization state and the imaginary part of the X polarization state, and at the same time, the phase compensation complex signal in the Y polarization state is decomposed into the real part of the Y polarization state and the imaginary part of the Y polarization state. Then, for each sampling point, the real part of the X polarization state, the imaginary part of the X polarization state, the real part of the Y polarization state, and the imaginary part of the Y polarization state are combined into a four-dimensional vector, and this four-dimensional vector represents a signal point in the four-dimensional constellation space. The set of four-dimensional vectors corresponding to all sampling points constitutes the signal point set in the four-dimensional constellation space.

[0104] The four-dimensional constellation space refers to a mathematical representation method used to describe the signal distribution state of polarization multiplexed high-order modulation signals after being processed at the receiving end. A four-dimensional vector corresponds to a point in a four-dimensional Euclidean space. The set of four-dimensional vectors corresponding to all sampling points forms the constellation diagram of the signal, and the four-dimensional space where this constellation diagram is located is the four-dimensional constellation space. The four-dimensional constellation space can completely represent the amplitude and phase information of the polarization multiplexed signal, facilitating subsequent clustering analysis, decision boundary generation, and equalization processing, thereby supporting the accurate demodulation and data recovery of high-order modulation formats.

[0105] S3.2, Constellation point clustering using the fuzzy C-means algorithm:

[0106] In the stage of performing constellation point clustering using the fuzzy C - means algorithm, clustering processing is carried out on the signal point set in the four - dimensional constellation space to determine the clustering centers of the constellation points. First, set the number of clusters to the order of the modulation format. For example, for M - order modulation, the number of clusters is set to M. Then, initialize the clustering centers, which can be set by randomly selecting signal points or based on the prior knowledge of the modulation format. Next, calculate the membership degree of each signal point belonging to each cluster. The membership degree represents the similarity between the signal point and the clustering center. Then, update the clustering centers. The specific method is to perform a weighted sum on all signal points. The weight during weighting is the square of the membership degree, and finally, normalize the weighted sum to obtain the new clustering centers. Repeat the process of calculating the membership degree and updating the clustering centers until the change in the clustering centers is less than a preset threshold or reaches a predetermined maximum number of iterations, and finally obtain the set of clustering centers.

[0107] The purpose of using the fuzzy C - means algorithm for constellation point clustering is to effectively handle the situation of blurred constellation point boundaries in a low optical signal - to - noise ratio environment. The fuzzy C - means algorithm assigns membership degrees of multiple clusters to each signal point, enhancing the robustness and accuracy of clustering, and providing reliable clustering center support for subsequent generation of decision boundaries.

[0108] S3.3, Generate an adaptive decision boundary:

[0109] In the stage of generating an adaptive decision boundary, use the set of clustering centers to generate decision boundaries for subsequent symbol decisions. First, for any two different clustering centers in the four - dimensional constellation space, calculate the mid - perpendicular hyperplane between them as the decision boundary. The specific calculation method is as follows: Use the difference vector between the two clustering centers as the normal vector of the mid - perpendicular hyperplane, and the mid - perpendicular hyperplane passes through the mid - point of the two clustering centers. The set of mid - perpendicular hyperplanes corresponding to all pairs of clustering centers constitutes the decision boundary set. This decision boundary set divides the four - dimensional constellation space into multiple regions, each region corresponding to a clustering center, and is used to determine the symbol category to which the signal point belongs.

[0110] The purpose of generating an adaptive decision boundary is to adaptively adjust the decision region according to the actual clustering distribution of the received signal to adapt to the influence of signal distortion and noise. Compared with fixed decision boundaries, adaptive decision boundaries can significantly improve the accuracy of symbol decisions and reduce the bit error rate, thus providing technical guarantee for the reliable operation of high - order modulation systems in complex transmission environments.

[0111] S3.4, Update the equalizer tap coefficients:

[0112] During the stage of updating the equalizer tap coefficients, the equalizer tap coefficients are optimized based on the cluster center set and the decision boundary set to improve the signal quality. First, the equalizer tap coefficients are defined as a vector, and the equalized signal is the inner product of this tap coefficient and the four-dimensional signal points. Then, the objective function is set as the minimum mean square error, specifically the expected value of the square of the Euclidean distance between the equalized signal and its affiliated cluster center, where the affiliated cluster center is determined by the nearest neighbor principle, that is, the cluster center closest to the equalized signal is selected. Next, the gradient descent method is used to update the tap coefficients. The specific method is as follows: calculate the gradient of the objective function with respect to the tap coefficients, which is determined by the expected value of the product of the difference between the equalized signal and its affiliated cluster center and the four-dimensional signal points, and then update the tap coefficients along the opposite direction of the gradient with a preset step size. Repeat the update process until the objective function converges, and finally obtain the optimal equalizer tap coefficients.

[0113] The purpose of updating the equalizer tap coefficients is to optimize the equalizer parameters, reduce the influence of signal distortion and noise on the constellation point distribution, thereby improving the clustering tightness and decision accuracy of the signal. As an efficient optimization method, the gradient descent method can quickly converge to the optimal solution, ensuring the real-time performance and effectiveness of the equalization process, and providing a high-quality basis for subsequent signal output.

[0114] The four-dimensional signal points are equalized using the optimal equalizer tap coefficients to generate the equalized four-dimensional signal. The specific method is to calculate the inner product of the tap coefficients and the four-dimensional signal points to obtain the equalized four-dimensional signal. Then, the equalized four-dimensional signal is converted into the form of a two-dimensional complex signal. The specific conversion method is as follows: the first two components of the equalized four-dimensional signal are respectively used as the real part and the imaginary part of the X polarization state, and the last two components are respectively used as the real part and the imaginary part of the Y polarization state, and are respectively combined into the complex signal of the X polarization state and the complex signal of the Y polarization state, and finally the equalized signal is formed.

[0115] Step S3 receives the phase compensation signal output by step S2, maps it to the four-dimensional constellation space, uses the fuzzy C-means algorithm for constellation point clustering, generates an adaptive decision boundary, and updates the equalizer tap coefficients, and finally outputs the equalized signal. This process effectively solves the problem of blurred constellation point boundaries in a low optical signal-to-noise ratio environment, improves the decision accuracy of the signal, provides a high-quality equalized signal for the outer and inner loop cooperation mechanisms in step S4, and ensures the reliability of ultra-high-definition video stream transmission.

[0116] Step S4 includes the following content:

[0117] Step S4 receives the equalized signal and the cluster center set output by step S3, and dynamically updates the polarization state transfer matrix and the phase compensation step size through the cooperation mechanism of the outer and inner loops. The following is the specific processing technical logic:

[0118] Outer loop: Update the polarization state transfer matrix according to the dispersion degree of the clustering centers.

[0119] In the outer loop stage, adjust the polarization state transfer matrix according to the dispersion degree of the clustering center set to reduce the spread of the constellation point distribution. First, calculate the dispersion degree of the clustering centers. The specific method is: calculate the average Euclidean distance from all clustering centers to their geometric center, where the geometric center is the average value of all clustering center vectors. Then, set a dispersion degree threshold. When the calculated dispersion degree of the clustering centers is greater than the dispersion degree threshold, trigger the update of the polarization state transfer matrix. The gradient descent method is used to update the polarization state transfer matrix. Specifically: by calculating the gradient of the dispersion degree of the clustering centers with respect to the polarization state transfer matrix, adjust the polarization state transfer matrix along the opposite direction of the gradient with a preset learning rate. The learning rate is a small positive number used to control the adjustment step size.

[0120] The purpose of updating the polarization state transfer matrix according to the dispersion degree of the clustering centers is to improve the accuracy of symbol decision by reducing the spread of the constellation point distribution. The gradient descent method adaptively adjusts the polarization state transfer matrix according to the change of the dispersion degree, and can maintain the best polarization state decoupling effect in a dynamic environment, thus enhancing the robustness of the system and its adaptability to complex channel environments.

[0121] Inner loop: Adjust the phase compensation step size based on the symbol error rate.

[0122] In the inner loop stage, adjust the phase compensation step size based on the symbol error rate to optimize the phase noise compensation effect. First, perform hard decision on the equalized signal to generate an estimated symbol sequence. Then, calculate the current symbol error rate. The specific method is: calculate the ratio of the number of error symbols to the total number of symbols. Next, set a symbol error rate threshold. When the calculated symbol error rate is greater than the symbol error rate threshold, adjust the phase compensation step size. The time-domain phase compensation step size and the frequency-domain phase compensation step size are adjusted using an adaptive strategy. Specifically: multiply the current time-domain phase compensation step size and the frequency-domain phase compensation step size by a factor greater than 1 respectively. The adjustment factor is proportional to the symbol error rate, and the adjustment coefficient is used to control the sensitivity of the adjustment. The purpose of adjusting the phase compensation step size based on the symbol error rate is to optimize the phase noise compensation effect and reduce the symbol error rate by adaptively adjusting the time-domain phase compensation step size and the frequency-domain phase compensation step size. The adaptive strategy dynamically adjusts the phase compensation step size according to the actual signal quality to ensure the best compensation effect can be obtained in different noise environments, thus enhancing the stability of the system and its signal processing performance.

[0123] Loop cooperation mechanism:

[0124] In the loop cooperation mechanism stage, the outer loop and the inner loop work together to dynamically optimize the polarization state transfer matrix and the phase compensation step size. Specifically, the outer loop updates the polarization state transfer matrix once every fixed time interval, and the inner loop adjusts the time-domain phase compensation step size and the frequency-domain phase compensation step size within each symbol period. Then, the updated polarization state transfer matrix of the outer loop is fed back to step S1 for the next round of polarization state decoupling processing; the adjusted time-domain phase compensation step size and frequency-domain phase compensation step size of the inner loop are fed back to step S2 for the next round of phase noise compensation processing.

[0125] Step S5 includes the following:

[0126] Step S5 receives the equalized signal passed from step S4, and needs to construct a phase residual model to correct the symbol confidence through belief propagation-assisted soft decision decoding, and output the reconstructed video data stream. The following elaborates in detail on the specific processing technical logic of step S5.

[0127] S5.1, Belief propagation-assisted soft decision decoding:

[0128] In the belief propagation-assisted soft decision decoding stage, soft decision decoding is performed on the equalized signal to generate soft information for each symbol. First, the equalized signal is input into the belief propagation decoder. The belief propagation decoder iteratively calculates the marginal posterior probability of each symbol through a message passing mechanism according to the prior distribution of the signal and the channel model. The specific calculation process is as follows: In the factor graph, the variable nodes and the check nodes exchange messages, and gradually update the symbol confidence until the messages converge or reach the preset maximum number of iterations. Then, the log-likelihood ratio of each symbol is calculated as the soft information, and the log-likelihood ratio represents the logarithm of the ratio of the probability that the symbol is 1 to the probability that the symbol is 0.

[0129] The purpose of using belief propagation-assisted soft decision decoding is to improve the accuracy of decoding through iterative optimization. Especially in the low optical signal-to-noise ratio environment, it can effectively cope with noise interference, improve the reliability of symbol confidence, and provide high-quality soft information support for the subsequent construction of the phase residual model.

[0130] S5.2, Constructing the phase residual model:

[0131] In the stage of constructing the phase residual model, the phase error is quantified based on the equalized signal and the soft information set. First, the actual phase of each symbol is calculated according to the equalized signal. Specifically, the arctangent of the ratio of the imaginary part to the real part of the complex value of the equalized signal is taken to obtain the actual phase angle. Then, the estimated phase is inferred according to the soft information set, and the ideal constellation point phase of the symbol is determined by the maximum a posteriori probability. Next, the phase residual is calculated, which is the difference between the actual phase and the estimated phase. After that, a phase residual model is constructed, expressing the phase residual as a linear combination of feature vectors plus a model error. The feature vectors include the amplitude of the symbol and the phase influence of neighboring symbols. Finally, the model parameters are optimized by solving. Specifically, the sum of the squares of the difference between the phase residual and the model prediction value is minimized to obtain the optimal model parameters.

[0132] The purpose of constructing the phase residual model is to quantify the influence of phase noise on symbol confidence. By fitting the phase residual with a linear model, the phase error can be accurately estimated, providing a basis for correcting the symbol confidence.

[0133] S5.3, Correct the symbol confidence:

[0134] In the stage of correcting the symbol confidence, the phase residual model is used to adjust the soft information set to improve the accuracy of the symbol confidence. First, the phase residual prediction value is calculated for each symbol. Specifically, the inner product of the feature vector and the model parameters is used as the prediction value. Then, the soft information set is adjusted. Specifically, the original soft information is added with the negative value of the phase residual prediction value multiplied by an adjustment factor. The adjustment factor is used to control the amplitude of the correction, and its value is preset according to system requirements.

[0135] The purpose of correcting the symbol confidence is to improve the accuracy of the soft information set by offsetting the influence of the phase error on the confidence. The adjusted soft information can more truly reflect the probability distribution of the symbol, thereby improving the correct rate of subsequent hard decisions, ensuring the reconstruction quality of the video data stream, and providing support for high-quality video transmission.

[0136] S5.4, Output the reconstructed video data stream:

[0137] In the stage of outputting the reconstructed video data stream, hard decisions are made on the corrected soft information set to generate an estimated symbol sequence. Specifically, when the soft information is greater than 0, the symbol is judged as 1, otherwise it is judged as 0. Then, the hard decision result is converted into a binary bit stream. Next, decoding and deinterleaving operations are performed on the binary bit stream to restore the original video data stream, and the decoding process is carried out according to the preset coding rules.

[0138] Embodiment 2: Figure 2 An integrated optical signal receiver of the present invention is given, including: a polarization decoupling unit, a phase compensation unit, a constellation equalization unit, a loop optimization unit, and a soft decision decoding unit;

[0139] Polarization decoupling unit: It separates the two orthogonally polarized state signals transmitted through the optical fiber, constructs a dynamic polarization state transfer matrix by using the Kalman filtering algorithm, eliminates the cross-coupling caused by the birefringence effect, and outputs a sequence of independent polarization state electrical signals;

[0140] Phase compensation unit: It performs time-frequency domain processing on the sequence of independent polarization state electrical signals, compensates the linear phase noise caused by the laser linewidth and the non-linear phase shift caused by the fiber Kerr effect, and outputs a phase compensation signal;

[0141] Constellation equalization unit: It maps the phase compensation signal to a four-dimensional constellation space, uses the fuzzy C-means algorithm for constellation point clustering, generates an adaptive decision boundary and updates the equalizer tap coefficients, and outputs an equalized signal;

[0142] Loop optimization unit: It updates the polarization state transfer matrix according to the dispersion degree of the clustering center, and adjusts the phase compensation step size based on the symbol error rate;

[0143] Soft decision decoding unit: It performs soft decision decoding assisted by belief propagation on the equalized signal, constructs a phase residual model to correct the symbol confidence, and outputs a reconstructed video data stream.

[0144] All the above formulas are dimensionless and take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the real situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0145] It should be noted that the system of the present invention can be deployed on the device itself to achieve embedded applications, or can also run on a PC or other terminals with a user interface, so as to meet various hardware environments and usage requirements.

[0146] Only some exemplary embodiments of the present invention are described above by way of illustration. Undoubtedly, for those of ordinary skill in the art, without departing from the spirit and scope of the present invention, the described embodiments can be modified in various different ways. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the protection scope of the claims of the present invention.

[0147] It should be noted that in this text, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.

[0148] As described above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claimed rights.

Claims

1. An integrated optical signal receiving method, characterized in that: Includes steps: S1: Separate the two orthogonal polarization state signals transmitted by optical fiber, construct a dynamic polarization state transfer matrix, and output an independent polarization state electrical signal sequence; S2: Perform time-frequency domain processing on the independent polarization state electrical signal sequence and output a phase compensation signal; S3: Map the phase compensation signal to the four-dimensional constellation space, perform constellation point clustering, generate the decision boundary and update the equalizer tap coefficient, and output the equalized signal; S4: Under the coordination mechanism of the outer and inner loops, the outer loop updates the polarization state transfer matrix according to the cluster center dispersion, and the inner loop adjusts the phase compensation step size based on the symbol error rate; S5: Perform confidence propagation-assisted soft decision decoding on the equalized signal, construct a phase residual model to correct the symbol confidence, and output the reconstructed video data stream.

2. The integrated optical signal receiving method according to claim 1, characterized in that: Step S1 includes the following contents: receiving a complex electric field component of an X polarization state and a complex electric field component of a Y polarization state from optical fiber transmission, and converting the complex electric field component of the X polarization state and the complex electric field component of the Y polarization state into a discrete signal sequence of an X polarization state and a discrete signal sequence of a Y polarization state through analog-to-digital conversion; Aiming at the birefringence effect in optical fiber, a polarization state transfer matrix is ​​introduced to describe the dynamic cross-coupling relationship between the X polarization state and the Y polarization state. The polarization state transfer matrix is ​​a two-dimensional complex matrix and is initialized according to the prior measurement data. The Kalman filter algorithm is used to dynamically estimate the polarization state transfer matrix. By defining the state vector, state transfer model and observation model, the estimated value of the polarization state transfer matrix is ​​iteratively optimized using the prediction step and the update step. The inverse matrix of the estimated polarization state transfer matrix is ​​used to decouple the discrete signal sequence of the X polarization state and the discrete signal sequence of the Y polarization state, and an independent polarization state electrical signal sequence is output, including an independent electrical signal sequence of the X polarization state and an independent electrical signal sequence of the Y polarization state.

3. The integrated optical signal receiving method according to claim 2, characterized in that: Step S2 includes the following contents: The phase drift rate is estimated by adaptive filtering technology, and the phase components of the X polarization state independent electrical signal sequence and the Y polarization state independent electrical signal sequence are reversely adjusted using the estimated phase drift rate to generate the X polarization state signal and the Y polarization state signal after time domain compensation.

4. The integrated optical signal receiving method according to claim 3, characterized in that: Step S2 also includes the following contents: The X polarization state signal and the Y polarization state signal after time domain compensation are converted into frequency domain signals through fast Fourier transform, and the nonlinear phase offset is estimated by split-step Fourier method. The phase of the frequency domain signal is reversely adjusted by using the estimated nonlinear phase offset to generate the X polarization state signal and the Y polarization state signal after frequency domain compensation. The X polarization state signal and the Y polarization state signal after frequency domain compensation are then converted back into the time domain through inverse fast Fourier transform to obtain the X polarization state signal and the Y polarization state signal after phase compensation. The X polarization state signal and the Y polarization state signal after phase compensation are integrated into a two-dimensional complex signal sequence.

5. The integrated optical signal receiving method according to claim 4, characterized in that: Step S3 includes the following contents: The X polarization state phase compensation complex signal and the Y polarization state phase compensation complex signal are decomposed into the X polarization state real part, the X polarization state imaginary part, the Y polarization state real part and the Y polarization state imaginary part, and the X polarization state real part, the X polarization state imaginary part, the Y polarization state real part and the Y polarization state imaginary part of each sampling point are combined into a four-dimensional vector to form a signal point set in the four-dimensional constellation space; then, the fuzzy C-means algorithm is applied to cluster the signal point set in the four-dimensional constellation space, the number of clusters is set according to the modulation format order, the cluster center is initialized, the membership degree of each signal point to the cluster center is calculated and the cluster center is updated, and it is iterated until the cluster center converges to generate a cluster center set.

6. The integrated optical signal receiving method according to claim 5, characterized in that: Step S3 also Includes the following: Then, based on the cluster center set, the median hyperplane between each pair of cluster centers is calculated to form a decision boundary set; then, the equalizer tap coefficients are defined, and the equalizer tap coefficients are updated by the minimum mean square error objective function and the gradient descent method. The optimization is iteratively performed until the objective function converges to obtain the optimal equalizer tap coefficients; then, the four-dimensional signal points are equalized using the optimal equalizer tap coefficients to generate an equalized four-dimensional signal, and the equalized four-dimensional signal is converted into an equalized signal containing an X-polarization state complex signal and a Y-polarization state complex signal.

7. The integrated optical signal receiving method according to claim 6, characterized in that: Step S4 includes the following contents: The dispersion of the cluster center set is calculated in the outer loop, and the polarization state transfer matrix is ​​updated by using the gradient descent method by comparing the dispersion with the preset threshold. Then, the equalized signal is hard-determined in the inner loop, the symbol error rate is calculated, and the time domain phase compensation step and the frequency domain phase compensation step are adjusted by using an adaptive strategy by comparing the symbol error rate with the preset threshold. Then, the polarization state transfer matrix updated by the outer loop is fed back to step S1 for the next round of polarization state decoupling processing. The time domain phase compensation step size and frequency domain phase compensation step size adjusted by the inner loop are fed back to step S2 for the next round of phase noise compensation processing.

8. The integrated optical signal receiving method according to claim 7, characterized in that: Step S5 includes the following contents: Perform confidence propagation-assisted soft decision decoding on the equalized signal to generate a soft information set of the symbol; construct a phase residual model based on the equalized signal and the soft information set of the symbol, calculate the actual phase, estimated phase and phase residual, and optimize the parameters of the phase residual model; use the phase residual model to adjust the soft information set of the symbol, calculate the phase residual prediction value and correct the soft information of the symbol; make a hard decision on the corrected soft information set of the symbol to generate an estimated symbol sequence, convert it into a binary bit stream and perform decoding and deinterleaving operations to output a reconstructed video data stream.

9. An integrated optical signal receiver, used to implement an integrated optical signal receiving method according to any one of claims 1 to 8, characterized in that: include: Polarization decoupling unit, phase compensation unit, constellation equalization unit, loop optimization unit and soft decision decoding unit; Polarization decoupling unit: Separates the two orthogonal polarization state signals transmitted by optical fiber, constructs a dynamic polarization state transfer matrix using the Kalman filter algorithm, eliminates the cross coupling caused by the birefringence effect, and outputs an independent polarization state electrical signal sequence; Phase compensation unit: performs time-frequency domain processing on the independent polarization state electrical signal sequence, compensates for the linear phase noise caused by the laser line width and the nonlinear phase shift caused by the fiber Kerr effect, and outputs a phase compensation signal; Constellation equalization unit: maps the phase compensation signal to the four-dimensional constellation space, uses the fuzzy C-means algorithm to cluster the constellation points, generates an adaptive decision boundary, updates the equalizer tap coefficients, and outputs the equalized signal; Loop optimization unit: updates the polarization state transfer matrix according to the cluster center dispersion and adjusts the phase compensation step size based on the symbol error rate; Soft decision decoding unit: performs confidence propagation-assisted soft decision decoding on the equalized signal, constructs a phase residual model to correct symbol confidence, and outputs a reconstructed video data stream.

Images