An oam-sk optical communication link demodulation evaluation method

Abstract

The application provides an OAM-SK optical communication link demodulation evaluation method, which comprises the following steps: LDPC encoding and mapping source information into OAM-SK symbols to generate a modulated optical field; simulating the formation of an optical intensity pattern at the receiving end after atmospheric turbulence propagation; inputting the optical intensity pattern into a visual state space model, extracting long-range spatial dependence features through bidirectional selective state space scanning and a gating mechanism, and outputting a symbol decision result; generating a symbol transition probability matrix based on decision statistics, and then simulating a receiving symbol sequence affected by recognition errors, combining LDPC decoding to recover bit information; finally, comparing with the original bit sequence, calculating BER and PSNR, and completing link demodulation evaluation. The application enhances the global structure modeling capability of the turbulence distortion OAM pattern under the condition of maintaining low computational complexity, improves the recognition robustness under strong turbulence conditions, and realizes the effective connection from optical intensity pattern recognition to communication link performance evaluation.

Description

Technical Field

[0001] This application belongs to the fields of free-space optical communication, optical information processing and intelligent communication technology, and specifically relates to an OAM-SK optical communication link demodulation evaluation method. Background Technology

[0002] Free-space optical communication (OSC) has become an important means of realizing future high-speed wireless links due to its significant advantages such as high bandwidth, abundant spectrum resources, strong directivity, and strong resistance to electromagnetic interference. To further improve the spectral efficiency and transmission capacity of OSC, Orbital Angular Momentum (OAM) mode multiplexing and modulation technology has attracted widespread attention. Among them, Orbital Angular Momentum Shift Keying (OAM-SK) can effectively increase the modulation dimension without occupying additional spectrum resources by mapping discrete symbols to different OAM modes or combinations thereof, thereby improving the overall transmission efficiency of the system. However, in actual atmospheric channels, turbulence effects can cause fuzziness in the OAM beam's ring structure, vortex core drift, inter-mode energy crosstalk, and higher-order phase distortion, resulting in significant distortion of the light intensity pattern acquired at the receiver, which severely affects the accuracy of symbol recognition and the reliability of the communication link.

[0003] In existing technologies, methods for receiving and recognizing OAM free-space optical communication under atmospheric turbulence conditions can be mainly classified into three categories. The first category is based on optical front-end compensation or mode sorting schemes, which partially reduce mode distortion by optimizing the transmitting, propagating, or receiving optical structures. The second category is based on image recognition schemes using deep learning models, which take the turbulently distorted OAM light intensity pattern as input and directly output the corresponding OAM mode or symbol category using a network model. Among these, the paper "Orbital angular momentum superimposed mode recognition based on multi-label image classification" published in *Optics Express* by W. Liu, C. Tu, Y. Liu, and Z. Ye studies the problem of OAM superimposed pattern recognition under atmospheric turbulence conditions, employing a deep learning method for multi-label image classification, reflecting a typical application route of this type of technology in the recognition of distorted OAM light intensity patterns. In their paper "Atmospheric turbulence-distorted OAM mode recognition via multi-scale dilated convolution with multi-level feature fusion," X. Cao et al. proposed a deep learning method based on multi-scale dilated convolution and multi-level feature fusion. This method enhances the discriminative performance of distorted OAM modes by improving convolutional feature extraction and multi-scale information fusion. The third type of method attempts to combine the receiver's recognition results with the channel coding and decoding process to evaluate link recovery quality.

[0004] However, the aforementioned existing technical solutions have at least the following shortcomings. First, when multiple distortions such as toroidal structure distortion, vortex core drift, and directional jitter caused by turbulence coexist, the local convolutional structure or windowed modeling method has limited ability to capture long-distance spatial dependencies, making it difficult to stably extract global discriminative features from OAM intensity patterns, resulting in insufficient utilization of long-range structural information in distorted patterns. Second, some models introduce global attention mechanisms with secondary computational complexity or heavy feature interaction modules to improve recognition performance, resulting in excessively high computational costs in high-resolution OAM pattern recognition scenarios, which is not conducive to practical deployment. Third, most existing solutions only focus on improving symbol-level classification accuracy, without further extending classification confusion information to communication link-level error recovery and transmission quality evaluation, thus failing to fully reflect the true contribution of the receiver demodulation method to the end-to-end communication performance of the system. Summary of the Invention

[0005] To address the aforementioned problems in the existing technology, this application provides an OAM-SK optical communication link demodulation evaluation method. The technical problem to be solved by this application is achieved through the following technical solution: An OAM-SK optical communication link demodulation evaluation method includes: S100: Perform LDPC encoding on the source information to be transmitted to obtain an encoded bit stream, and map the encoded bit stream to OAM-SK symbols; S200: Generate an OAM modulated optical field at the transmitting end according to the OAM-SK symbol; S300: After applying propagation processing that simulates the effects of atmospheric turbulence to the OAM modulated light field, a received light intensity pattern is formed on the focal plane of the CCD at the receiving end. S400: The received light intensity pattern is preprocessed, and the preprocessed received light intensity pattern is input into the trained visual state space model to extract long-range spatial dependency features from it, and then the decision result of the OAM-SK symbol is output. S500: Calculate the correspondence between the actual transmitted symbols and the decision results, and generate a symbol transition probability matrix; S600: The symbol transition probability matrix is ​​used to simulate the symbol decision process at the receiving end to obtain the received symbol sequence affected by the recognition error, and the recovered bit information is obtained by decoding and recovering the received symbol sequence by combining the LDPC parity check matrix. S700: Compare the bit information obtained from decoding with the original bit sequence corresponding to the source information to be transmitted to obtain system-level communication performance indicators and complete the link demodulation evaluation.

[0006] Optionally, S100 includes: S110: Convert the source information to be transmitted into a raw bit sequence; S120: Perform LDPC encoding on the original bit sequence to generate an encoded bit stream; S130: Divide the encoded bitstream into several bit groups according to the symbol length according to the preset mapping rules. Each group is mapped to an OAM-SK symbol. Each OAM-SK symbol corresponds to a predefined combination of multiple OAM modes.

[0007] Optionally, the S200 includes: S210: Construct a Laguerre-Gaussian mode with zero radial exponent in the input plane at the transmitting end; S220: Based on the predefined topological charge set corresponding to the OAM-SK symbol, assign complex weights to each Laguerre-Gaussian mode and perform coherent superposition to obtain the OAM modulated optical field.

[0008] Optional, the S300 includes: S310: The OAM modulated optical field is input into the free space channel and simulated through multi-segment propagation. In each propagation segment, the optical field is propagated using the angular spectrum propagation operator, and a random phase screen generated based on the Von Kármán spectrum is superimposed to characterize the turbulence disturbance. S320: The light field after propagation through multiple segments is focused by the receiving end lens, and the light intensity distribution is recorded on the CCD focal plane to form the received light intensity pattern.

[0009] Optionally, the trained visual state space model is obtained in the following way: Construct a dataset containing received light intensity patterns corresponding to various OAM-SK symbols under different turbulence intensities; The dataset is divided into a training set, a validation set, and a test set; A pre-built visual state space model is trained using a training set, and the hyperparameters of the visual state space model are adjusted using a validation set. The performance of the visual state space model is then evaluated using a test set. The visual state space model that meets the performance requirements is used as the trained visual state space model.

[0010] Optional, the S400 includes: S410, the received light intensity pattern is subjected to size unification, cropping and normalization processing so that the image size meets the input requirements of the trained visual state space model, and the preprocessed received light intensity pattern is obtained. S420: Input the preprocessed received light intensity pattern into the trained visual state space model so that it can extract the radial and angular long-range spatial dependence features of the received light intensity pattern with linear complexity through bidirectional selective state space scanning, and suppress invalid features of turbulent disturbance regions through gating mechanism, retain the discriminative features related to ring structure, radial stripes and vortex core, and then output the decision result of the OAM-SK symbol.

[0011] Optionally, the processing of the preprocessed received light intensity pattern by the visual state space model includes: The preprocessed received light intensity pattern is divided into non-overlapping image blocks, and each image block is linearly mapped into a token sequence and a position code is added. The token sequence is input into a network model consisting of multiple stacked visual state space encoders. Each encoder adopts a bidirectional selective state space scanning structure to perform forward and backward scanning on the token sequence, and extracts long-range spatial dependency features through a gating mechanism. The features output by the encoder are globally pooled, and the decision result of the OAM-SK symbol is output through the classification head.

[0012] Optional, the S500 includes: S510: Based on the symbol decision results of the visual state space model on the test set, statistically analyze the correspondence between the actual transmitted symbols and the received decision symbols; S520: Generate a symbol transition probability matrix based on the statistical results. The symbol transition probability matrix describes the conditional probability distribution of each received decision symbol under the given real transmitted symbol conditions.

[0013] Optional, the S600 includes: S610: Generate the corresponding transmission symbol sequence according to the original bit sequence and preset data units; S620: For each real transmitted symbol in the transmitted symbol sequence, random sampling is performed using the conditional probability distribution corresponding to the real transmitted symbol in the symbol transition probability matrix, and the sampled symbol is used as the corresponding reception decision symbol, thereby obtaining the entire received symbol sequence affected by the recognition error. S630: Combine the bit information corresponding to the received symbol sequence with the LDPC parity-check matrix, and use the sum-product decoding algorithm to perform iterative decoding recovery to obtain the recovered bit information.

[0014] Optional, the S700 includes: The recovered bit information is compared bit by bit with the original bit sequence to calculate the bit error rate (BER), which serves as the system-level communication performance indicator. If the transmitted object is an image, the recovered bit information needs to be reconstructed into a received image and compared with the original image to calculate the peak signal-to-noise ratio (PSNR), which serves as a supplement to the system-level communication performance indicator.

[0015] Beneficial effects: (1) This application uses a visual state space model to demodulate the turbulent distortion OAM light intensity pattern. Through a bidirectional selective state space scanning mechanism, it completes the modeling of the radial and angular long-range dependencies in the light intensity pattern with linear complexity. While maintaining low computational complexity, it enhances the ability to extract the global spatial structure, thereby improving the robustness and stability of symbol recognition under medium and high intensity turbulence conditions.

[0016] (2) This application uses a gating mechanism to adaptively adjust the state update intensity based on the local reliability of the input features, which can effectively suppress invalid features in areas of severe turbulence disturbance, while retaining key discrimination information such as annular texture, radial stripes and vortex core drift, thus improving the characterization ability and classification accuracy of complex distorted patterns.

[0017] (3) This application organically combines the symbol demodulation results with the link recovery assessment. By constructing the symbol transition probability matrix, the classification confusion information output by the identifier is transmitted to the LDPC decoding stage, and further communication performance indicators such as bit error rate (BER) and peak signal-to-noise ratio (PSNR) are obtained. This realizes end-to-end integrated processing from light intensity pattern recognition to system-level communication link performance assessment, and more comprehensively reflects the true contribution of the demodulation method to the overall performance of the free space optical communication system.

[0018] The present application will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating an OAM-SK optical communication link demodulation evaluation method provided in this application; Figure 2 This is a comparison chart of the simulation results of the link performance of this application and the comparative method under different turbulence intensities. Detailed Implementation

[0020] The present application will be described in further detail below with reference to specific embodiments, but the implementation of the present application is not limited thereto.

[0021] like Figure 1 As shown, this application provides an OAM-SK optical communication link demodulation evaluation method, including: S100: Perform LDPC encoding on the source information to be transmitted to obtain an encoded bit stream, and map the encoded bit stream to OAM-SK symbols; The source information to be transmitted is the image or data source to be transmitted.

[0022] S200: Generate an OAM modulated optical field at the transmitting end according to the OAM-SK symbol; S300: After applying propagation processing that simulates the effects of atmospheric turbulence to the OAM modulated light field, a received light intensity pattern is formed on the focal plane of the CCD at the receiving end. S400: The received light intensity pattern is preprocessed, and the preprocessed received light intensity pattern is input into the trained visual state space model to extract long-range spatial dependency features from it, and then the decision result of the OAM-SK symbol is output. S500: Calculate the correspondence between the actual transmitted symbols and the decision results, and generate a symbol transition probability matrix; S600: The symbol transition probability matrix is ​​used to simulate the symbol decision process at the receiving end to obtain the received symbol sequence affected by the recognition error, and the recovered bit information is obtained by decoding and recovering the received symbol sequence by combining the LDPC parity check matrix. S700: Compare the bit information obtained from decoding with the original bit sequence corresponding to the source information to be transmitted to obtain system-level communication performance indicators and complete the link demodulation evaluation.

[0023] In one specific embodiment of this application, S100 includes: S110: Convert the source information to be transmitted into a raw bit sequence; S120: Perform LDPC encoding on the original bit sequence to generate an encoded bit stream; S130: Divide the encoded bitstream into several bit groups according to the symbol length according to the preset mapping rules. Each group is mapped to an OAM-SK symbol. Each OAM-SK symbol corresponds to a predefined combination of multiple OAM modes.

[0024] This embodiment converts the source information to be transmitted into a bit sequence, performs low-density parity-check (LDPC) encoding on the bit sequence to obtain an encoded bit stream, and then groups the encoded bit stream according to symbol length according to a preset mapping rule, mapping it to M-ary OAM-SK symbols. Preferably, M is 16, and each OAM-SK symbol corresponds to a set of predefined OAM mode combinations.

[0025] In one specific embodiment of this application, S200 includes: S210: Construct a Laguerre-Gaussian mode with zero radial exponent in the input plane at the transmitting end; S220: Based on the predefined topological charge set corresponding to the OAM-SK symbol, assign complex weights to each Laguerre-Gaussian mode and perform coherent superposition to obtain the OAM modulated optical field.

[0026] This application generates a corresponding OAM modulated optical field at the transmitting end based on the obtained OAM-SK symbol. Preferably, a Laguerre-Gaussian mode with p=0 can be constructed in the input plane and coherently superimposed according to the topological charge set corresponding to the target symbol to obtain the complex amplitude optical field to be transmitted, which can be expressed as:

[0027]

[0028] in, L This represents the set of OAM patterns corresponding to the target symbol. Indicates the complex weights of each mode. Indicates the beam waist parameters. and These represent the polar radius and azimuth of the input plane, respectively.

[0029] In one specific embodiment of this application, S300 includes: S310: The OAM modulated optical field is input into the free space channel and simulated through multi-segment propagation. In each propagation segment, the optical field is propagated using the angular spectrum propagation operator, and a random phase screen generated based on the Von Kármán spectrum is superimposed to characterize the turbulence disturbance. S320: The light field after propagation through multiple segments is focused by the thin lens at the receiving end, and the light intensity distribution is recorded on the focal plane of the CCD to form the received light intensity pattern.

[0030] The OAM-modulated optical field obtained from S200 is input into a free-space channel, and atmospheric turbulence is simulated using a multi-segment propagation model. In each propagation segment, the optical field is propagated using an angular spectrum propagation operator, and a phase screen is superimposed to characterize turbulent disturbances. After being focused by a thin lens at the receiving end, the optical field forms a received light intensity pattern on the CCD focal plane. Preferably, the propagation process can be represented as follows:

[0031]

[0032]

[0033] in, Indicates the length of adjacent propagation segments. Indicates the operating wavelength. Represents spatial frequency coordinates. This represents the complex amplitude field at the focal plane of the receiving end. This represents the light intensity pattern recorded by the CCD. Preferably, the turbulent phase screen is generated based on the Von Kármán spectrum.

[0034] In one specific embodiment of this application, the trained visual state space model is obtained in the following manner: a. Construct a dataset containing received light intensity patterns corresponding to various OAM-SK symbols under different turbulence intensities; b. Divide the dataset into a training set, a validation set, and a test set; c. Train a pre-built visual state space model using the training set, adjust the hyperparameters of the visual state space model using the validation set, and evaluate the performance of the visual state space model using the test set. The visual state space model of this application can adopt the Vision Mamba architecture, the core of which lies in modeling the long-range radial and angular dependencies in the OAM pattern with linear complexity. Preferably, the bidirectional scanning structure includes: performing forward selective state space scanning and backward selective state space scanning on the same sequence of features, and fusing features through stitching projection or gated summation.

[0035] The gating mechanism is used to adaptively adjust the state update intensity based on the local reliability of the input features, thereby suppressing invalid features in regions of severe turbulence disturbance, while preserving relatively stable annular structures, radial stripes, and vortex core related features.

[0036] d, the visual state space model that meets the performance requirements is taken as the trained visual state space model.

[0037] In one specific embodiment of this application, S400 includes: S410, the received light intensity pattern is subjected to size unification, cropping and normalization processing so that the image size meets the input requirements of the trained visual state space model, and the preprocessed received light intensity pattern is obtained. This application performs size unification, cropping, or normalization processing on the received light intensity images, and constructs training, validation, and test sets. Preferably, the image input size is uniformly 224×224, and different candidate demodulation models adopt a consistent data preprocessing strategy to ensure comparison fairness.

[0038] S420: Input the preprocessed received light intensity pattern into the trained visual state space model so that it can extract the radial and angular long-range spatial dependence features of the received light intensity pattern with linear complexity through bidirectional selective state space scanning, and suppress invalid features of turbulent disturbance regions through gating mechanism, retain the discriminative features related to ring structure, radial stripes and vortex core, and then output the decision result of the OAM-SK symbol.

[0039] In one specific embodiment of this application, the processing procedure of the visual state space model for the preprocessed received light intensity pattern includes: The preprocessed received light intensity pattern is divided into non-overlapping image blocks, and each image block is linearly mapped into a token sequence and a position code is added. The token sequence is input into a network model consisting of multiple stacked visual state space encoders. Each encoder adopts a bidirectional selective state space scanning structure to perform forward and backward scanning on the token sequence, and extracts long-range spatial dependency features through a gating mechanism. The features output by the encoder are globally pooled, and the decision result of the OAM-SK symbol is output through the classification head.

[0040] This application divides the preprocessed received light intensity pattern into several non-overlapping image blocks, linearly maps each image block into a token sequence, adds position encoding, and inputs it into a visual state space model encoder. The encoder adopts a bidirectional selective state space scanning structure, performs forward and backward scanning on the token sequence respectively, and extracts long-range spatial dependency features through gating mechanism, adaptive feature mixing and fusion operations. Then, it outputs the corresponding OAM-SK symbol class probability or decision result through global pooling and classification head.

[0041] In one specific embodiment of this application, S500 includes: S510: Based on the symbol decision results of the visual state space model on the test set, statistically analyze the correspondence between the actual transmitted symbols and the received decision symbols; S520: Generate a symbol transition probability matrix based on the statistical results. The symbol transition probability matrix describes the conditional probability distribution of each received decision symbol under the given real transmitted symbol conditions.

[0042] Based on the symbol decision results, this application statistically analyzes the correspondence between the actual transmitted symbols and the received decision symbols to establish a symbol decision statistical relationship; then, based on the statistical relationship, it generates a symbol transition probability matrix to describe the conditional probability distribution of each received decision symbol under the given actual transmitted symbol condition.

[0043] The symbol decision statistics of this application can not only be used for demodulation performance analysis, but can also be further converted into a link-level symbol transfer model, so that the output of the recognizer can be directly connected with the decoding process of the communication system, thereby establishing an end-to-end performance analysis path.

[0044] In one specific embodiment of this application, S600 includes: S610: Generate the corresponding transmission symbol sequence according to the original bit sequence and preset data units; S620: For each real transmitted symbol in the transmitted symbol sequence, random sampling is performed using the conditional probability distribution corresponding to the real transmitted symbol in the symbol transition probability matrix, and the sampled symbol is used as the corresponding reception decision symbol, thereby obtaining the entire received symbol sequence affected by the recognition error. S630: Combine the bit information corresponding to the received symbol sequence with the LDPC parity-check matrix, and use the sum-product decoding algorithm to perform iterative decoding recovery to obtain the recovered bit information.

[0045] This application generates a transmitted symbol sequence based on several OAM-SK symbols corresponding to each pixel or data unit; then, based on the obtained symbol transition probability matrix, it simulates or performs equivalent sampling on the symbol decision process at the receiving end to obtain a received symbol sequence affected by recognition errors; further, it combines the LDPC parity check matrix and the sum-product decoding algorithm to decode and recover the bit information corresponding to the received symbol sequence.

[0046] In one specific embodiment of this application, S700 includes: The recovered bit information is compared bit by bit with the original bit sequence to calculate the bit error rate (BER), which serves as the system-level communication performance indicator. If the transmitted object is an image, the recovered bit information needs to be reconstructed into a received image and compared with the original image to calculate the peak signal-to-noise ratio (PSNR), which serves as a supplement to the system-level communication performance indicator.

[0047] This application compares the obtained decoding result with the original bit sequence to calculate the bit error rate (BER); when the transmitted object is an image, the recovered bit stream is reassembled into a received image and compared with the original image to calculate the peak signal-to-noise ratio (PSNR), thereby realizing a system-level communication performance evaluation of the demodulation method.

[0048] To verify the effectiveness of this application, the following simulation experiments were conducted.

[0049] 1. Simulation conditions In a preferred embodiment, a set of 16 OAM-SK symbols is constructed, each symbol being represented by a combination of preset OAM modes. The transmitting end converts the source image into a bitstream, encodes it using LDPC, maps it to 16-ary OAM-SK symbols, and generates the corresponding transmitted light field using a Laguerre-Gaussian mode superposition method with p=0. After propagating through an atmospheric turbulence channel, the transmitted light field is captured by the CCD at the receiving end, showing the focal plane intensity pattern.

[0050] During the dataset construction process, the input images at the receiving end are uniformly processed to a size of 224×224 and divided into training set, validation set and test set; to ensure the fairness of the comparison, different candidate demodulation models adopt the same data preprocessing process and input specifications.

[0051] In a specific example, 10 different intensities of atmospheric turbulence can be set, and corresponding samples can be generated for each symbol category at each turbulence level to form a dataset for model training and testing. To verify the effectiveness of the method in this application, comparative experiments can be conducted using CNN models, Swin Transformer models, and the Vision Mamba model used in this application.

[0052] 2. Simulation Content The light intensity pattern is input into the Vision Mamba demodulation model used in this application. This model first divides the image into patches, then forms a token sequence through linear projection, extracts features through a bidirectional selective state-space module, and finally outputs the corresponding OAM-SK symbol decision result. Furthermore, the symbol transition description required for link recovery assessment can be generated based on the symbol decision statistics obtained during the testing phase, and the BER can be calculated in conjunction with LDPC decoding. In image transmission scenarios, the recovered bitstream can also be reconstructed into a received image and the PSNR can be calculated.

[0053] Simulation results are as follows Figure 2 As shown, the visual state space model method used in this application can achieve lower BER and better PSNR under different turbulence intensities, indicating that the method has good recognition robustness and link recovery capability under medium and high intensity turbulence conditions.

[0054] It is worth noting that the terms "first" and "second" in this application are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0055] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of this application and should not be construed as limiting the specific implementation of this application to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of this application, and all such modifications or substitutions should be considered within the scope of protection of this application.

Claims

1. A demodulation evaluation method for OAM-SK optical communication links, characterized in that, include: S100: Perform LDPC encoding on the source information to be transmitted to obtain an encoded bit stream, and map the encoded bit stream to OAM-SK symbols; S200: Generate an OAM modulated optical field at the transmitting end according to the OAM-SK symbol; S300: After applying propagation processing that simulates the effects of atmospheric turbulence to the OAM modulated light field, a received light intensity pattern is formed on the focal plane of the CCD at the receiving end. S400: Preprocess the received light intensity pattern, and input the preprocessed received light intensity pattern into the trained visual state space model to extract the long-range space from it. Based on the dependent features, the decision result of the OAM-SK symbol is then output; S500: Calculate the correspondence between the actual transmitted symbols and the decision results, and generate a symbol transition probability matrix; S600: The symbol transition probability matrix is ​​used to simulate the symbol decision process at the receiving end to obtain the received symbol sequence affected by the recognition error, and the recovered bit information is obtained by decoding and recovering the received symbol sequence by combining the LDPC parity check matrix. S700: Compare the bit information obtained from decoding with the original bit sequence corresponding to the source information to be transmitted to obtain system-level communication performance indicators and complete the link demodulation evaluation.

2. The OAM-SK optical communication link demodulation evaluation method according to claim 1, characterized in that, S100 includes: S110: Convert the source information to be transmitted into a raw bit sequence; S120: Perform LDPC encoding on the original bit sequence to generate an encoded bit stream; S130: Divide the encoded bitstream into several bit groups according to the symbol length according to the preset mapping rules. Each group is mapped to an OAM-SK symbol. Each OAM-SK symbol corresponds to a predefined combination of multiple OAM modes.

3. The OAM-SK optical communication link demodulation evaluation method according to claim 1, characterized in that, S200 includes: S210: Construct a Laguerre-Gaussian mode with zero radial exponent in the input plane at the transmitting end; S220: Based on the predefined topological charge set corresponding to the OAM-SK symbol, assign complex weights to each Laguerre-Gaussian mode and perform coherent superposition to obtain the OAM modulated optical field.

4. The OAM-SK optical communication link demodulation evaluation method according to claim 1, characterized in that, The S300 includes: S310: The OAM modulated optical field is input into the free space channel and simulated through multi-segment propagation. In each propagation segment, the optical field is propagated using the angular spectrum propagation operator, and a random phase screen generated based on the Von Kármán spectrum is superimposed to characterize the turbulence disturbance. S320: The light field after propagation through multiple segments is focused by the receiving end lens, and the light intensity distribution is recorded on the CCD focal plane to form the received light intensity pattern.

5. The OAM-SK optical communication link demodulation evaluation method according to claim 1, characterized in that, The trained visual state space model is obtained in the following way: Construct a dataset containing received light intensity patterns corresponding to various OAM-SK symbols under different turbulence intensities; The dataset is divided into a training set, a validation set, and a test set; A pre-built visual state space model is trained using a training set, and the hyperparameters of the visual state space model are adjusted using a validation set. The performance of the visual state space model is then evaluated using a test set. The visual state space model that meets the performance requirements is used as the trained visual state space model.

6. The OAM-SK optical communication link demodulation evaluation method according to claim 1, characterized in that, The S400 includes: S410, the received light intensity pattern is subjected to size unification, cropping and normalization processing so that the image size meets the input requirements of the trained visual state space model, and the preprocessed received light intensity pattern is obtained. S420: Input the preprocessed received light intensity pattern into the trained visual state space model so that it can extract the radial and angular long-range spatial dependence features of the received light intensity pattern with linear complexity through bidirectional selective state space scanning, and suppress invalid features of turbulent disturbance regions through gating mechanism, retain the discriminative features related to ring structure, radial stripes and vortex core, and then output the decision result of the OAM-SK symbol.

7. The OAM-SK optical communication link demodulation evaluation method according to claim 6, characterized in that, The visual state space model's processing of the preprocessed received light intensity pattern includes: The preprocessed received light intensity pattern is divided into non-overlapping image blocks, and each image block is linearly mapped into a token sequence and a position code is added. The token sequence is input into a network model consisting of multiple stacked visual state space encoders. Each encoder adopts a bidirectional selective state space scanning structure to perform forward and backward scanning on the token sequence, and extracts long-range spatial dependency features through a gating mechanism. The features output by the encoder are globally pooled, and the decision result of the OAM-SK symbol is output through the classification head.

8. The OAM-SK optical communication link demodulation evaluation method according to claim 1, characterized in that, The S500 includes: S510: Based on the symbol decision results of the visual state space model on the test set, statistically analyze the correspondence between the actual transmitted symbols and the received decision symbols; S520: Generate a symbol transition probability matrix based on the statistical results. The symbol transition probability matrix describes the conditional probability distribution of each received decision symbol under the given real transmitted symbol conditions.

9. The OAM-SK optical communication link demodulation evaluation method according to claim 1, characterized in that, The S600 includes: S610: Generate the corresponding transmission symbol sequence according to the original bit sequence and preset data units; S620: For each real transmitted symbol in the transmitted symbol sequence, random sampling is performed using the conditional probability distribution corresponding to the real transmitted symbol in the symbol transition probability matrix, and the sampled symbol is used as the corresponding reception decision symbol, thereby obtaining the entire received symbol sequence affected by the recognition error. S630: Combine the bit information corresponding to the received symbol sequence with the LDPC parity-check matrix, and use the sum-product decoding algorithm to perform iterative decoding recovery to obtain the recovered bit information.

10. The OAM-SK optical communication link demodulation evaluation method according to claim 9, characterized in that, The S700 includes: The recovered bit information is compared bit by bit with the original bit sequence to calculate the bit error rate (BER), which serves as the system-level communication performance indicator. If the transmitted object is an image, the recovered bit information needs to be reconstructed into a received image and compared with the original image to calculate the peak signal-to-noise ratio (PSNR), which serves as a supplement to the system-level communication performance indicator.

Images