A signal communication method for multidimensional domains of data links

Abstract

This application provides a signal communication method for a multi-dimensional domain of a data link, comprising: performing cross-domain correlation deconstruction on the data link's spatiotemporal-frequency multi-dimensional signal and error feature data under an electromagnetic environment to obtain the original spatiotemporal-frequency domain feature set; simultaneously integrating signal propagation characteristic data and error source attribute data to generate a multi-source correlation data cluster to generate a spatiotemporal-frequency multi-dimensional fused feature topology; constructing a multi-dimensional domain collaborative error-resistant model based on the spatiotemporal-frequency multi-dimensional fused feature topology; and performing dynamic weight allocation and error suppression strategy iterative optimization on the spatiotemporal-frequency domain error-resistant weight parameters and error suppression strategy parameters in conjunction with spatiotemporal-frequency domain resource adaptation rules to generate a multi-dimensional domain error-resistant control command; performing spatiotemporal-frequency collaborative optimization processing on the data link transmission signal according to the multi-dimensional domain error-resistant control command to obtain an error-resistant optimized signal; simultaneously collecting transmission quality data of the link feedback signal; and fine-tuning the multi-dimensional domain error-resistant control command parameters in reverse by the deviation between the transmission quality data and a preset quality threshold.

Description

Technical Field

[0001] This application relates to the field of error-resistant data link communication technology, and more specifically, to a signal communication method for a multi-dimensional domain of a data link. Background Technology

[0002] In complex electromagnetic environments such as modern battlefields and civilian communications, data links, as the core carriers of information transmission, face multi-source, heterogeneous, and dynamically changing error threats, placing stringent demands on their error resilience, signal transmission stability, and spectrum utilization efficiency. Data links must ensure effective signal transmission under high-error scenarios while simultaneously considering error suppression and signal integrity to meet the real-time communication needs of various services.

[0003] Existing data link error mitigation schemes mostly employ single-domain error mitigation techniques, achieving error suppression through independent spatial beamforming, temporal filtering, or frequency hopping mechanisms. This approach first acquires signal or error data in a single domain, generating error mitigation parameters based on a preset fixed strategy; then, it optimizes the data link signal in a single dimension according to these parameters; finally, it directly injects the processed signal into the transmission link, without a dynamic adjustment mechanism.

[0004] However, this single-domain error-resistance scheme has significant technical drawbacks. Because it fails to integrate the spatiotemporal multidimensional signals and error characteristics across domains, it struggles to fully capture the dynamic correlation between errors and signals under complex electromagnetic environments, resulting in low adaptability of the error-resistance strategy. Furthermore, the lack of dynamic optimization and closed-loop adjustment mechanisms for weight parameters and suppression strategies prevents real-time correction of error-resistance commands based on changes in the electromagnetic environment and transmission quality feedback, limiting the error suppression effect and failing to meet the high-stability transmission requirements of data links under complex electromagnetic environments. Summary of the Invention

[0005] This application provides a signal communication method for a multi-dimensional domain of a data link to at least alleviate the aforementioned technical problems.

[0006] A signal communication method for a multi-dimensional domain of a data link, comprising:

[0007] Step 1: Perform cross-domain correlation deconstruction on the data link space-time-frequency multidimensional signal and error feature data under electromagnetic environment to obtain the original feature set in the space-time-frequency domain. Simultaneously integrate the signal propagation characteristic data and error source attribute data to generate a multi-source correlation data cluster, so as to generate a space-time-frequency multidimensional fusion feature topology.

[0008] Step 2: Construct a multi-dimensional domain collaborative error-resistant model based on the spatiotemporal-frequency multi-dimensional fusion feature topology, and combine the spatiotemporal-frequency domain resource adaptation rules to dynamically allocate the error-resistant weight parameters and error suppression strategy parameters in the spatiotemporal-frequency domain and iteratively optimize the error suppression strategy, thereby generating multi-dimensional domain error-resistant control instructions;

[0009] Step 3: Based on the multi-dimensional domain error-resistant control command, perform spatiotemporal-frequency collaborative optimization processing on the data link transmission signal to obtain the error-resistant optimized signal. Simultaneously collect the transmission quality data of the link feedback signal. By the deviation between the transmission quality data and the preset quality threshold, fine-tune the parameters of the multi-dimensional domain error-resistant control command in reverse to control the error-resistant transmission of the error-resistant optimized signal in the electromagnetic environment.

[0010] Optionally, step 1 includes:

[0011] Step 11: Collect the spatiotemporal frequency multidimensional signal, error feature data, signal propagation characteristic data, and error source attribute data of the data chain; perform format compliance verification and missing value reconstruction and repair on various types of data to generate a complete original data stack.

[0012] Step 12: Perform cross-domain correlation deconstruction on the data chain space-time-frequency multidimensional signal and error feature data in the complete original data stack to obtain the original feature set in the space-time-frequency domain. Simultaneously fuse the signal propagation characteristic data and error source attribute data in the complete original data stack to generate a multi-source correlation data cluster.

[0013] Optionally, step 12 includes:

[0014] Step 121: Perform spatial array feature extraction, time-domain pulse feature parsing, and frequency-domain spectral feature decomposition on the spatial-temporal-frequency multidimensional signal of the data chain in the complete original data stack to obtain the spatial-temporal-frequency signal feature extraction results. Perform error polarization parameter extraction, error intensity quantization, and error incident direction localization on the error feature data in the complete original data stack to obtain the error feature extraction results. Perform cross-domain feature association mapping processing on the spatial-temporal-frequency signal feature extraction results and the error feature extraction results to obtain the original feature set in the spatial-temporal-frequency domain.

[0015] Step 122: Extract the propagation attenuation law and propagation path characteristics of the signal propagation characteristics data from the complete original data stack, and fuse them with the error type information and error emission parameters in the error source attribute data, as well as perform multi-source feature association mapping to generate a multi-source associated data cluster.

[0016] Optionally, step 12 also includes:

[0017] Step 123: Establish the feature correspondence between the original feature set in the space-time-frequency domain and the multi-source associated data cluster, and mark the cross-redundant features and conflicting features.

[0018] Optionally, in the spatial array feature extraction in step 121, spatial array feature parameters are obtained by analyzing the array manifold characteristics, beam pointing characteristics, and spatial diversity gain of the received signal of the data link spatial array; time-domain pulse feature analysis is performed by extracting the time-domain pulse signal from the data link spatial-time-frequency multidimensional signal and extracting the pulse width, repetition period, and rising edge slope to form a time-domain feature subset; frequency-domain spectral feature decomposition is performed by extracting the frequency-domain spectral signal from the data link spatial-time-frequency multidimensional signal and extracting the spectral peak value, bandwidth, and frequency distribution to obtain a frequency-domain feature subset. The above spatial array feature parameters, time-domain feature subset, and frequency-domain feature subset together constitute the spatial-time-frequency signal feature extraction result.

[0019] Optionally, step 1 includes:

[0020] Step 13: Align the original spatiotemporal-frequency domain feature set with the multi-source associated data clusters by time axis homogeneity and dimension normalization to obtain a temporal collaborative fusion dataset. Based on the spatiotemporal-frequency domain association criteria, perform feature clustering divide-and-conquer and cross-domain association modeling on the temporal collaborative fusion dataset to generate an initial cross-domain association feature tensor. Use the mutual information entropy quantization mechanism to remove redundant features from the initial cross-domain association feature tensor to obtain a redundant clean feature tensor. Perform core feature enhancement aggregation on the redundant clean feature tensor through the attention mechanism to generate a spatiotemporal-frequency multidimensional fusion feature topology.

[0021] Optionally, step 13 includes:

[0022] Step 131: Using the timestamp of the data link receiving terminal as the reference source, the original spatiotemporal frequency domain feature set and the multi-source associated data cluster are aligned with the same source on the time axis and then interpolated and reconstructed to integrate them into a dataset with a unified time granularity. Numerical interval standardization mapping is then performed on the dataset with the unified time granularity to generate a time-series collaborative fusion dataset.

[0023] Step 132: Based on the differences in technical attributes in the space-time-frequency domain, the feature data in the time-series collaborative fusion dataset is divided into static attribute data and dynamic response data to form a space-time-frequency feature classification spectrum. The signal features and error features in the space-time-frequency feature classification spectrum are modeled by pairwise association matching through a cross-domain association rule engine to generate the initial cross-domain association feature tensor.

[0024] Optionally, step 13 includes:

[0025] Step 133: Validate and filter the initial cross-domain correlation feature tensor, remove tensor elements without actual physical correlation, calculate the information redundancy of the remaining tensor elements using the mutual information entropy quantization mechanism to mark redundant feature columns, and perform dimension pruning and purification to obtain a redundant purified feature tensor. Sort the feature columns in the redundant purified feature tensor by importance weight, extract several core feature dimensions with the highest ranking, and assign inter-domain influence weight coefficients to different core features through an attention mechanism to generate a spatiotemporal frequency multidimensional fusion feature topology.

[0026] Optionally, step 2 includes:

[0027] Step 21: Extract error polarization feature values, signal propagation attenuation coefficient, and frequency domain occupancy parameters from the spatiotemporal multidimensional fusion feature topology. Based on these three types of parameters, construct a multidimensional domain error-resistant objective functional with the goals of maximizing error suppression, minimizing signal distortion, and optimizing spectrum utilization. Combine the power constraints, bandwidth constraints, and array aperture constraints of data link transmission to establish constraint boundary conditions.

[0028] Step 22: Based on the multidimensional domain robust objective functional and constraint boundary conditions, construct a cognitive-driven multidimensional domain collaborative robustness model based on dynamic environmental perception. The collaborative robustness model has built-in spatiotemporal frequency domain resource adaptation rules and dynamic weight adaptation mechanism.

[0029] Step 23: Iteratively solve the multi-dimensional domain collaborative error-resistant model, determine the spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters through multiple rounds of optimization, and generate multi-dimensional domain error-resistant control instructions after verifying the effectiveness of the two types of parameters.

[0030] Optionally, step 22 includes:

[0031] Step 221: Based on the multidimensional domain robust objective functional and constraint boundary conditions, construct an initial multidimensional domain cooperative robust model;

[0032] Step 222: Real-time acquisition of electromagnetic environment change data and data link transmission status data, and correlation mapping and rule extraction processing of the two types of data to generate spatiotemporal frequency domain resource adaptation rules and embed them into the initial multidimensional domain collaborative error-resistant model;

[0033] Step 223: Analyze the differences in error and signal features in the spatiotemporal multidimensional fusion feature topology, and assign differentiated dynamic weights to the spatial beamforming weights, temporal filtering weights, and frequency resource allocation weights respectively, thereby constructing a dynamic weight adaptation mechanism and embedding it into the initial multidimensional domain collaborative error-resistant model to complete the construction of the multidimensional domain collaborative error-resistant model based on dynamic environmental perception.

[0034] Optionally, step 23 includes:

[0035] Step 231: Perform cognitive optimization calculation on the multi-dimensional domain collaborative error-resistant model based on the initial population, where the initial population is the combined solution vector of error-resistant weights and suppression strategy parameters;

[0036] Step 232: Calculate the fitness value of each combination solution vector in the initial population, and perform selection, crossover, and mutation evolution operations based on the fitness value to generate a new generation of population. Repeat the iterative evolution process until the set termination criterion is met to obtain the final population.

[0037] Step 233: Select the Pareto optimal solution from the final population, analyze the optimal solution to obtain the spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters, verify the effectiveness of the two types of parameters, and generate multidimensional domain error-resistant control instructions after eliminating invalid parameter combinations.

[0038] Optionally, step 3 includes:

[0039] Step 31: Analyze the error resistance weight matrix and error suppression strategy parameters in the multi-dimensional domain error resistance control instruction, retrieve the data link transmission characteristic parameter library, and generate a signal optimization processing strategy based on the two types of parameters and the parameter library data.

[0040] Step 32: Based on the signal optimization processing strategy, perform spatial beamforming processing on the data link transmission signal to generate a spatial optimized signal. Apply adaptive time-domain filtering to the spatial optimized signal to generate a space-time joint optimized signal. Perform spectral adaptive adjustment on the space-time joint optimized signal based on frequency domain resource allocation parameters to generate an error-resistant optimized signal.

[0041] Optionally, step 32 includes:

[0042] Step 321: Based on the spatial beamforming weights in the signal optimization processing strategy, perform array beamforming dynamic processing on the data link transmission signal to control the beam direction to form nulls in the error direction and high-gain main lobes in the desired signal direction, and generate a spatial optimization signal accordingly.

[0043] Step 322: Adaptive time-domain filtering is used to decompose the spatial optimization signal into a time-domain signal and identify error components to eliminate time-domain superimposed error components and generate a space-time joint optimization signal.

[0044] Step 323: Based on the frequency domain resource allocation parameters in the signal optimization processing strategy, perform frequency domain resource reconstruction and spectrum adaptive adjustment on the spatiotemporal joint optimization signal to avoid error-occupied frequency bands and optimize spectrum resource allocation efficiency, so as to generate an error-resistant optimization signal.

[0045] Optionally, step 3 includes:

[0046] Step 33: Inject the error-resistant optimization signal into the data link transmission link and simultaneously collect signal transmission quality data. Compare the signal transmission quality data with the preset quality threshold to obtain the deviation value. Use the deviation value to fine-tune the multi-dimensional domain error-resistant control command parameters to control the error-resistant transmission of the error-resistant optimization signal in the electromagnetic environment.

[0047] Optionally, step 33 includes:

[0048] Step 331: Inject the error-resistant optimization signal into the data link transmission link through the transmission interface module to enable real-time signal transmission and transmission status monitoring;

[0049] Step 332: Start the link monitoring module and collect signal transmission quality data fed back from the link according to the preset sampling frequency;

[0050] Step 333: Compare the signal transmission quality data with the preset quality threshold. If there is a deviation, adjust the weight parameters and strategy parameters in the multi-dimensional domain error-resistant control command in reverse based on the deviation value, update the signal optimization processing strategy, and repeat steps 32-33.

[0051] This application's multi-dimensional domain signal communication method for data links addresses the technical shortcomings of traditional single-domain error resistance schemes, such as low adaptability of error resistance strategies and limited error suppression effects. Step 1 involves cross-domain correlation deconstruction of spatiotemporal multi-dimensional signals and error feature data, simultaneously integrating multi-source correlation data to generate a spatiotemporal multi-dimensional fused feature topology. This solves the problem that traditional schemes struggle to comprehensively capture the dynamic correlation characteristics between errors and signals. Compared to the limitations of traditional schemes that only collect single-domain data, this application, through cross-domain correlation deconstruction and multi-source data fusion, comprehensively integrates multi-dimensional information such as spatiotemporal frequency domain signals, errors, propagation characteristics, and error source attributes. The generated fused feature topology can fully reflect the correlation relationships of various dimensions of features under complex electromagnetic environments, significantly improving the adaptability of error resistance strategies compared to traditional schemes and providing a comprehensive feature foundation for subsequent model construction.

[0052] Step 2 constructs a multi-dimensional domain collaborative error-resistant model based on feature topology. By dynamically optimizing weight parameters and suppression strategies using resource adaptation rules, multi-dimensional domain error-resistant control commands are generated, solving the problems of fixed error-resistant parameters and rigid strategies in traditional solutions. Traditional solutions employ preset fixed strategies, which cannot adapt to changes in the electromagnetic environment. This application, however, constructs a collaborative error-resistant model with a built-in dynamic weight adaptation mechanism and resource adaptation rules. Based on multi-objective functionals and iterative optimization of constraints, it achieves dynamic allocation and optimization of weight parameters and suppression strategies. The generated control commands are more closely aligned with the real-time electromagnetic environment, significantly improving the targeting and flexibility of error suppression compared to traditional fixed strategies.

[0053] Based on the control instructions in step 3, spatiotemporal-frequency coordinated optimization processing is performed to generate an error-resistant optimized signal, solving the problem of incomplete error suppression in traditional single-domain processing. Traditional solutions optimize the signal from only a single dimension, making it difficult to cope with multi-source heterogeneous errors. However, this application achieves comprehensive suppression of multi-dimensional errors by forming nulls in the error direction, eliminating time-domain error components, and avoiding frequency-domain error bands through coordinated processing of spatial beamforming, temporal filtering, and frequency-domain resource adjustment. Compared with traditional single-domain processing, the error suppression effect is more comprehensive and the signal integrity is better.

[0054] By collecting transmission quality data in step 33 and fine-tuning the control command parameters based on the deviation, a closed-loop optimization mechanism is formed, solving the problem of the lack of dynamic adjustment mechanism in traditional solutions. Traditional solutions lack feedback adjustment links, making it difficult to continuously optimize error resistance. In contrast, this application monitors transmission quality in real time, compares the deviation with preset thresholds to obtain the deviation, and fine-tunes the parameters in reverse to continuously optimize the error resistance strategy. This allows the error resistance to dynamically adapt to changes in the electromagnetic environment and transmission status. Compared with traditional static solutions, the stability and continuous error resistance of data link transmission are significantly improved, better meeting the communication needs in complex electromagnetic environments. Attached Figure Description

[0055] Figure 1 This is a flowchart illustrating a signal communication method for a multi-dimensional domain of a data link according to an embodiment of this application.

[0056] Figure 2 This is a schematic diagram of the signal communication system of the data link multi-dimensional domain according to an embodiment of this application;

[0057] Figure 3 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0058] like Figure 1 As shown, this embodiment of the present application provides a signal communication method for a multi-dimensional domain of a data link, comprising:

[0059] Step 1: Perform cross-domain correlation deconstruction on the data link space-time-frequency multidimensional signal and error feature data under electromagnetic environment to obtain the original feature set in the space-time-frequency domain. Simultaneously integrate the signal propagation characteristic data and error source attribute data to generate a multi-source correlation data cluster, so as to generate a space-time-frequency multidimensional fusion feature topology.

[0060] Step 2: Construct a multi-dimensional domain collaborative error-resistant model based on the spatiotemporal-frequency multi-dimensional fusion feature topology, and combine the spatiotemporal-frequency domain resource adaptation rules to dynamically allocate the error-resistant weight parameters and error suppression strategy parameters in the spatiotemporal-frequency domain and iteratively optimize the error suppression strategy, thereby generating multi-dimensional domain error-resistant control instructions;

[0061] Step 3: Based on the multi-dimensional domain error-resistant control command, perform spatiotemporal-frequency collaborative optimization processing on the data link transmission signal to obtain the error-resistant optimized signal. Simultaneously collect the transmission quality data of the link feedback signal. By the deviation between the transmission quality data and the preset quality threshold, fine-tune the parameters of the multi-dimensional domain error-resistant control command in reverse to control the error-resistant transmission of the error-resistant optimized signal in the electromagnetic environment.

[0062] Optionally, step 1 includes:

[0063] Step 11: Collect the spatiotemporal frequency multidimensional signal, error feature data, signal propagation characteristic data, and error source attribute data of the data chain; perform format compliance verification and missing value reconstruction and repair on various types of data to generate a complete original data stack.

[0064] Step 12: Perform cross-domain correlation deconstruction on the data chain space-time-frequency multidimensional signal and error feature data in the complete original data stack to obtain the original feature set in the space-time-frequency domain. Simultaneously fuse the signal propagation characteristic data and error source attribute data in the complete original data stack to generate a multi-source correlation data cluster.

[0065] Preferably, the specific implementation process of step 11 is as follows: Considering the high requirements for the integrity and consistency of multi-source data in complex electromagnetic environments, a layered acquisition and repair mechanism is specifically designed: First, based on the transmission characteristics of the data link communication scenario, a multi-dimensional data acquisition unit is deployed to simultaneously capture the data link's spatiotemporal-frequency multi-dimensional signals, error feature data, signal propagation characteristic data, and error source attribute data. Specifically, the data link's spatiotemporal-frequency multi-dimensional signals are acquired by the array antenna receiving module at preset sampling intervals; error feature data is captured in real-time by the error monitoring unit; signal propagation characteristic data is recorded by the channel detection module; and error source attribute data is retrieved from the environmental perception database. For the acquired raw data, format compliance verification is first performed. Based on the data format standards specified in the data link communication protocol, the field length, data type, and encoding method of each type of data are checked to ensure they meet the requirements. Data with format errors are filtered out and marked as format-abnormal datasets. For format-abnormal datasets, repair is performed according to data type. For field-missing anomalies, the missing fields are supplemented using feature interpolation of the same type of data. For encoding error anomalies, the encoding deviation is corrected through protocol parsing inverse operations, generating a format-compliant dataset. Next, missing value detection is performed on the format-compliant dataset, and the missing value ratio of each data type is calculated. For data with a missing value ratio below a preset threshold (e.g., 5%), interpolation based on the temporal correlation of data chain signals is used to repair missing values ​​by utilizing the changing trends of similar data at adjacent time points. For data with a missing value ratio above the preset threshold, multi-source data complementary repair is initiated, calling relevant feature data from other associated datasets and reconstructing the missing data through feature mapping relationships to generate a missing value repair dataset. Finally, the format-compliant dataset and the missing value repair dataset are integrated, and duplicate records are removed to form a complete original data stack. The complete original data stack contains all four types of data that have been verified and repaired, providing a high-quality data foundation for subsequent cross-domain association deconstruction.

[0066] Specifically, in this application, the sampling interval design of the data acquisition unit is strongly correlated with the data type. For rapidly changing data such as spatiotemporal-frequency multidimensional signals from the data link, a shorter sampling interval (e.g., 1 microsecond) is used to ensure the capture of the instantaneous characteristics of the signal. For slowly changing data such as error source attribute data, a longer sampling interval (e.g., 10 milliseconds) is used to reduce acquisition overhead while ensuring data validity. During the format compliance verification process, a data format verification rule base is established, which contains protocol standard parameters for various types of data. During verification, the original data is compared with the parameters in the rule base one by one, generating a verification result log. The log records the location, type, and cause of abnormal data, providing a basis for subsequent repair. During the missing value repair process, temporal correlation interpolation is based on the periodicity and continuity characteristics of the data link signal. By fitting the change curves of adjacent data points, the predicted value at the time of the missing value is calculated to ensure that the repaired data conforms to the signal change pattern. Multi-source data complementary repair is based on the inherent correlation between various types of data. For example, path loss information in signal propagation characteristic data is used to assist in reconstructing the intensity attenuation data in error characteristic data, ensuring that the repaired data is consistent with the actual communication scenario.

[0067] Optionally, step 12 includes:

[0068] Step 121: Perform spatial array feature extraction, time-domain pulse feature parsing, and frequency-domain spectral feature decomposition on the spatial-temporal-frequency multidimensional signal of the data chain in the complete original data stack to obtain the spatial-temporal-frequency signal feature extraction results. Perform error polarization parameter extraction, error intensity quantization, and error incident direction localization on the error feature data in the complete original data stack to obtain the error feature extraction results. Perform cross-domain feature association mapping processing on the spatial-temporal-frequency signal feature extraction results and the error feature extraction results to obtain the original feature set in the spatial-temporal-frequency domain.

[0069] Step 122: Extract the propagation attenuation law and propagation path characteristics of the signal propagation characteristics data from the complete original data stack, and fuse them with the error type information and error emission parameters in the error source attribute data, as well as perform multi-source feature association mapping to generate a multi-source associated data cluster.

[0070] Preferably, step 121 is executed first. For the data link space-time-frequency multidimensional signal in the complete original data stack, the array manifold characteristics (reflecting the signal phase and amplitude correlation between array elements), beam pointing characteristics (characterizing the signal arrival direction related parameters), and spatial diversity gain (reflecting the signal enhancement effect of multi-antenna reception) of the array received signal are analyzed by the spatial array feature extraction module. The output is a spatial array feature parameter containing parameters such as array steering vector, beamwidth, and diversity gain value. Then, the time-domain pulse is separated from the data link space-time-frequency multidimensional signal by the time-domain pulse feature analysis module. For the impulse signal, pulse width (signal duration), repetition period (pulse interval), and rise slope (signal amplitude rise rate) are extracted to form a time-domain feature subset. Subsequently, the frequency domain spectral feature decomposition module performs frequency domain transformation on the spatiotemporal multidimensional signal of the data link, extracting spectral peak value (maximum signal strength in the frequency domain), bandwidth (frequency range occupied by the signal), and frequency distribution (intensity distribution of the signal at various frequency points), resulting in a frequency-domain feature subset. The aforementioned spatial array feature parameters, time-domain feature subset, and frequency-domain feature subset together constitute the spatiotemporal signal feature extraction result. For the error feature data in the complete original data stack, the error polarization parameter extraction module obtains the polarization mode and polarization angle of the error signal. The error intensity quantization module converts the error signal amplitude into a standardized intensity value. The error incident direction positioning module determines the azimuth and elevation angles of the error signal. These three factors together constitute the error feature extraction result. The results of the extraction of spatiotemporal signal features and the results of the extraction of error features are input into the cross-domain feature association mapping module. Based on the correspondence between signals and errors in the time, space and frequency dimensions, a feature mapping table is established. For example, the repetition period of the time-domain pulse signal is associated with the time change trend of the error intensity, and the beam pointing in the feature parameters of the spatial array is associated with the error incident direction. Finally, the original feature set of the spatiotemporal frequency domain is generated.

[0071] Next, step 122 is executed to extract the propagation attenuation law (the trend of signal attenuation with transmission distance) and propagation path characteristics (path type of signal transmission, obstacle influence parameters) from the complete original data stack. At the same time, error type information (such as suppression error, deception error, etc.) and error transmission parameters (transmission power of error signal, modulation method) are extracted from the error source attribute data. The propagation attenuation law, propagation path characteristics, error type information, and error transmission parameters are input into the multi-source feature fusion module. First, feature normalization is used to eliminate dimensional differences. Then, an association model is established based on the influence law of propagation path on error signal attenuation. For example, the error intensity quantization result is corrected according to the propagation attenuation law, and the propagation loss of error signal is judged by combining the propagation path characteristics. Finally, a multi-source associated data cluster containing signal-error-propagation environment association information is generated through multi-source feature association mapping.

[0072] Specifically, in this application, the spatial array feature extraction module employs an array signal subspace decomposition method. By analyzing the covariance matrix of the array received signal, it separates the signal subspace and noise subspace, and then solves for the array manifold characteristic parameters. This method differs from traditional single-parameter extraction methods, as it can simultaneously acquire multiple spatial feature correlation parameters. The time-domain pulse feature analysis module, targeting the pulse characteristics of the data link signal, uses an adaptive threshold segmentation method to separate the pulse signal, avoiding pulse extraction deviations caused by traditional fixed thresholds. The frequency-domain spectral feature decomposition module, combined with the bandwidth characteristics of the data link signal, employs a multi-resolution spectral analysis method to achieve a balance between a wide frequency range and high frequency resolution, ensuring the comprehensiveness of spectral feature extraction. During the cross-domain feature correlation mapping process, a three-dimensional spatial-temporal-frequency correlation matrix is ​​established. The rows of the matrix represent time-dimensional features, the columns represent spatial-dimensional features, and the layers represent frequency-dimensional features. The intersection elements represent the correlation strength values ​​of the corresponding three-dimensional features. This matrix enables deep correlation between signals and error features. During the multi-source feature fusion process, based on the obstacle distribution information in the propagation path features, the power attenuation in the error transmission parameters is corrected, so that the fused multi-source associated data clusters are more in line with the actual propagation scenario, providing an accurate multi-source association basis for subsequent feature topology generation.

[0073] Optionally, step 12 also includes:

[0074] Step 123: Establish the feature correspondence between the original feature set in the space-time-frequency domain and the multi-source associated data cluster, and mark the cross-redundant features and conflicting features.

[0075] Preferably, the specific implementation process of step 123 is as follows: Combining the core requirement of avoiding redundancy and conflict in multi-source feature fusion of the data link to improve the accuracy of subsequent models, a feature association and discrimination mechanism is designed: First, a feature association mapping matrix is ​​constructed. The row dimension of the matrix corresponds to all feature items in the original feature set in the spatiotemporal frequency domain, and the column dimension corresponds to all feature items in the multi-source associated data cluster. The element value at the intersection of the matrix represents the association strength between the two corresponding feature items. The association strength is quantified based on the physical meaning correlation of the features in the data link communication scenario and the consistency of data change trends. For example, the "error incident direction feature" in the original feature set in the spatiotemporal frequency domain and the "propagation path obstacle distribution feature" in the multi-source associated data cluster are assigned a higher association strength value due to the existence of a physical causal relationship, thus generating the feature association mapping matrix. Based on this matrix, the feature correspondence between the original feature set in the spatiotemporal frequency domain and the multi-source associated data cluster is established, clarifying the association features and association strength of each feature item in the other set, forming a feature association list.

[0076] Next, for the cross-feature items in the feature association list, a redundant feature identification process is initiated. The information overlap between the cross-feature items is calculated. The information overlap is determined by comprehensively considering the statistical distribution similarity and information entropy difference of the feature data. When the information overlap is higher than a preset threshold (e.g., 80%), the cross-features are marked as cross-redundant features, and the dominant feature item of the redundant feature (i.e., the feature item with more comprehensive information coverage and higher data reliability) is recorded. Subsequently, feature conflict detection is performed. Feature pairs with high association strength but significantly different data values ​​in the feature association list are compared. Conflict determination is made by combining the physical laws of the data link communication scenario and the reliability level of the data acquisition source (e.g., the reliability of data acquired by the array antenna is higher than that of data retrieved from the environmental perception database). If the difference in feature values ​​cannot be explained by physical laws and the feature data from the high-reliability source has no obvious anomalies, the feature pair is marked as conflicting features, and the feature item corresponding to the high-reliability source is locked as a valid feature.

[0077] Finally, a feature identification report is generated, which includes a feature association list, a cross-redundant feature labeling table, and a conflict feature labeling table. The cross-redundant feature labeling table clearly identifies redundant feature groups, dominant feature items, and redundancy removal suggestions. The conflict feature labeling table clearly identifies conflicting feature pairs, effective feature items, and conflict cause analysis. This feature identification report provides a clear basis for feature selection and optimization in the subsequent time-series collaborative fusion dataset generation process, avoiding redundant features from increasing computational overhead and conflicting features from affecting model training performance.

[0078] Specifically, in this application, the quantitative assignment of feature correlation strength adopts a scenario-based weight allocation mechanism. First, feature items are divided into three categories according to the functional attributes of the data link communication scenario: signal features, error features, and propagation features. Basic correlation weights are assigned between features of the same category, while the weights between features of different categories are adjusted based on the closeness of their physical correlation. For example, the correlation weight between signal features and propagation features is higher than the basic weight between signal features and error features. During the calculation of information overlap, the temporal characteristics of the data link signal are introduced. This involves not only comparing static statistical distributions but also analyzing the consistency of feature trends over time. For example, the Pearson correlation coefficient of the feature sequence is calculated using a sliding window to enhance the accuracy of redundancy identification. During feature conflict detection, a reliability rating system for data acquisition sources is established. Each source is rated based on indicators such as the accuracy of the acquisition equipment, the stability of the acquisition environment, and the data verification pass rate. Feature data from high-rated sources dominates conflict determination. Simultaneously, the rationality of the data is verified in conjunction with the physical laws of data link communication. For example, the conflict between the error incident direction feature and the propagation path feature must conform to the reflection and refraction laws of electromagnetic wave propagation. If the conflict violates physical laws, the feature data of the low-reliability source is determined to be abnormal.

[0079] Optionally, in the spatial array feature extraction in step 121, spatial array feature parameters are obtained by analyzing the array manifold characteristics, beam pointing characteristics, and spatial diversity gain of the received signal of the data link spatial array; time-domain pulse feature analysis is performed by extracting the time-domain pulse signal from the data link spatial-time-frequency multidimensional signal and extracting the pulse width, repetition period, and rising edge slope to form a time-domain feature subset; frequency-domain spectral feature decomposition is performed by extracting the frequency-domain spectral signal from the data link spatial-time-frequency multidimensional signal and extracting the spectral peak value, bandwidth, and frequency distribution to obtain a frequency-domain feature subset. The above spatial array feature parameters, time-domain feature subset, and frequency-domain feature subset together constitute the spatial-time-frequency signal feature extraction result.

[0080] Preferably, in the specific technical implementation of the spatial time-frequency signal feature extraction result in step 121, considering the need for the data link to accurately capture the essential features of multi-domain signals in complex electromagnetic environments, a differentiated extraction mechanism is designed for features of different dimensions: First, spatial array feature extraction is performed. For the data link spatial array received signal in the complete original data stack, the covariance matrix of the signal is calculated first through the array signal covariance matrix construction module (the rows and columns of the matrix correspond to each antenna element of the array, and the intersection elements represent the correlation of the received signals of two corresponding antenna elements). Then, the signal subspace is separated through the subspace decomposition module. Based on the signal subspace, the array manifold characteristic parameters (including the correlation between the phase response and amplitude response of each antenna element) are solved in the noise subspace. At the same time, the beam scanning module traverses the preset angle range, records the received signal strength corresponding to each angle, and determines the beam pointing characteristics (including parameters such as main lobe pointing angle, main lobe width, and side lobe suppression ratio). Then, the diversity gain calculation module compares the signal-to-noise ratio difference between the single antenna received signal and the multi-antenna joint received signal to obtain the spatial diversity gain value. The array manifold characteristic parameters, beam pointing characteristics, and spatial diversity gain value together constitute the spatial array characteristic parameters.

[0081] Next, time-domain pulse feature analysis is performed. The time-domain pulse signal is separated from the space-time-frequency multidimensional signal in the data link through an adaptive threshold segmentation module (the threshold is dynamically adjusted according to the statistical distribution of the signal amplitude to avoid pulse loss or misjudgment caused by a fixed threshold). The pulse width detection module calculates the duration from the amplitude of a single pulse signal rising to a preset proportion (e.g., 10%) to falling back to that proportion, thus obtaining the pulse width. The repetition period identification module counts the time interval between the peak values ​​of two adjacent pulse signals, and takes the average of multiple counts as the repetition period. The rising edge slope calculation module fits the amplitude-time curve of the rising edge of the pulse signal, and obtains the slope value of the curve as the rising edge slope. The pulse width, repetition period, and rising edge slope together form a subset of time-domain features.

[0082] Subsequently, frequency domain spectral feature decomposition is performed. The spatiotemporal multidimensional signal of the data link is converted to the frequency domain through a fast Fourier transform module to obtain the frequency domain signal amplitude spectrum. The frequency point with the strongest intensity and its corresponding amplitude in the amplitude spectrum is located by a spectral peak detection module to obtain the spectral peak value. The frequency range in the frequency domain signal with an amplitude higher than the peak value by a preset proportion (e.g., the proportion corresponding to 3dB) is determined by a bandwidth calculation module to obtain the bandwidth. The frequency domain intervals are divided according to a preset frequency interval by a frequency distribution statistics module, and the signal intensity proportion in each interval is calculated to obtain the frequency distribution. The above spectral peak value, bandwidth, and frequency distribution together constitute a frequency domain feature subset.

[0083] Finally, the spatial array feature parameters, time-domain feature subsets, and frequency-domain feature subsets are regularized in terms of feature dimensions, and the data format and sampling dimensions are unified to form the spatial-temporal-frequency signal feature extraction results. These results comprehensively cover the core features of the data link signal in the three dimensions of space, time, and frequency, laying the foundation for subsequent cross-domain correlation with error features.

[0084] Specifically, in this application, during the construction of the array signal covariance matrix, a sliding window method is used to update the matrix. The window length is dynamically adjusted according to the rate of change of the data link signal (e.g., a shorter window is used when the signal changes rapidly) to ensure that the matrix can reflect the correlation of the array signal in real time. The preset angle range of the beam scanning module is set according to the coverage requirements of the data link communication, and the angle step size is adjusted according to the pointing accuracy requirements, taking into account both detection efficiency and accuracy. In the adaptive threshold segmentation module, the initial threshold value is determined based on the statistical mean and variance of the signal amplitude, and is subsequently dynamically corrected according to the pulse detection results to ensure the accuracy of pulse signal separation. In the time-domain pulse feature analysis, to address potential noise errors in the pulse signal, a time-domain smoothing filter module is used for preprocessing before extraction to retain the essential characteristics of the pulse signal while suppressing noise. In the frequency-domain spectral feature decomposition, to avoid spectral leakage affecting the feature extraction accuracy, the time-domain signal is weighted by a window function before the Fourier transform (the window function type is selected according to the time-domain characteristics of the signal, such as a rectangular window or a Hanning window).

[0085] Optionally, step 1 includes:

[0086] Step 13: Align the original spatiotemporal-frequency domain feature set with the multi-source associated data clusters by time axis homogeneity and dimension normalization to obtain a temporal collaborative fusion dataset. Based on the spatiotemporal-frequency domain association criteria, perform feature clustering divide-and-conquer and cross-domain association modeling on the temporal collaborative fusion dataset to generate an initial cross-domain association feature tensor. Use the mutual information entropy quantization mechanism to remove redundant features from the initial cross-domain association feature tensor to obtain a redundant clean feature tensor. Perform core feature enhancement aggregation on the redundant clean feature tensor through the attention mechanism to generate a spatiotemporal-frequency multidimensional fusion feature topology.

[0087] Optionally, step 13 includes:

[0088] Step 131: Using the timestamp of the data link receiving terminal as the reference source, the original spatiotemporal frequency domain feature set and the multi-source associated data cluster are aligned with the same source on the time axis and then interpolated and reconstructed to integrate them into a dataset with a unified time granularity. Numerical interval standardization mapping is then performed on the dataset with the unified time granularity to generate a time-series collaborative fusion dataset.

[0089] Step 132: Based on the differences in technical attributes in the space-time-frequency domain, the feature data in the time-series collaborative fusion dataset is divided into static attribute data and dynamic response data to form a space-time-frequency feature classification spectrum. The signal features and error features in the space-time-frequency feature classification spectrum are modeled by pairwise association matching through a cross-domain association rule engine to generate the initial cross-domain association feature tensor.

[0090] Preferably, step 131 is executed first. Using the timestamp of the data link receiving terminal as the reference source (this timestamp is generated by the clock module of the receiving terminal, recording the precise moment of signal reception), the time stamp information of the original spatiotemporal frequency domain feature set and the multi-source associated data cluster is extracted. The timestamps of the two types of data are matched with the reference timestamp using a time alignment module. Data with non-overlapping timestamps are reconstructed using linear or polynomial interpolation to ensure that both types of data have corresponding values ​​at the same time node, thus integrating them to obtain a dataset with a unified time granularity (the time granularity is set according to the rate of change of the data link signal; for example, a smaller granularity is used when the signal fluctuates rapidly). Numerical interval standardization mapping is performed on the dataset with the unified time granularity, uniformly mapping the values ​​of different feature items to a preset interval, eliminating the influence of differences in dimensions and numerical ranges, and generating a time-series collaborative fusion dataset. This dataset retains all valid information from the original spatiotemporal frequency domain feature set and the multi-source associated data cluster, and possesses temporal consistency and numerical comparability.

[0091] Next, step 132 is executed. Based on the differences in technical attributes in the space-time-frequency domain, the feature data in the time-series collaborative fusion dataset is classified: static attribute data refers to feature items that change slowly or remain basically unchanged during data link communication, such as the physical size parameters of the antenna array and the inherent type information of the error source; dynamic response data refers to feature items that fluctuate in real time with changes in the communication environment and transmission status, such as the instantaneous amplitude of the signal and the real-time change value of the error intensity. Through classification, a space-time-frequency feature classification spectrum is formed to clarify the attributes and change characteristics of each type of feature. The space-time-frequency feature classification spectrum is input into the cross-domain association rule engine. This engine has a built-in feature association rule library for the data link communication scenario (the rule library is built based on the electromagnetic propagation law and signal error mechanism). It performs pairwise association matching modeling on the signal features and error features in the space-time-frequency feature classification spectrum. For example, the repetition period in the time-domain pulse feature is associated with the time change feature of the error intensity, and the beam pointing in the space-domain array feature is associated with the error incident direction feature. The initial cross-domain association feature tensor is generated through association modeling. The dimensions of this tensor correspond to the time dimension, feature type dimension, and correlation strength dimension, respectively. Tensor elements represent the correlation relationship and correlation strength between two types of features at a specific time node, comprehensively reflecting the multi-dimensional and temporal correlation characteristics of signals and errors.

[0092] Specifically, in this application, the time alignment module adopts a dynamic interpolation strategy, using linear interpolation for data with gradual changes and polynomial interpolation for data with drastic fluctuations, ensuring that the reconstructed data accurately reflects the changing trends of the original data. During the numerical interval standardization mapping process, a differentiated mapping strategy is used for static attribute data and dynamic response data. Static data uses fixed interval mapping, while dynamic data uses adaptive interval mapping. The mapping parameters are adjusted according to the statistical distribution characteristics of the data, ensuring that the mapped data retains both the original differences and uniformity. The rule base of the cross-domain association rule engine adopts a dynamic update mechanism. By continuously learning feature association instances in data chain communication scenarios, the accuracy of association rules is constantly optimized. When modeling association matching, not only linear associations between features are considered, but non-linear association relationships are also captured. The generated initial cross-domain association feature tensor is stored in the form of a three-dimensional matrix. The rows of the matrix represent time nodes, the columns represent feature pair combinations, and the layers represent association strength levels, clearly presenting the temporal changes and strength differences of feature associations, laying the foundation for subsequent redundant feature removal and core feature extraction.

[0093] Optionally, step 13 includes:

[0094] Step 133: Validate and filter the initial cross-domain correlation feature tensor, remove tensor elements without actual physical correlation, calculate the information redundancy of the remaining tensor elements using the mutual information entropy quantization mechanism to mark redundant feature columns, and perform dimension pruning and purification to obtain a redundant purified feature tensor. Sort the feature columns in the redundant purified feature tensor by importance weight, extract several core feature dimensions with the highest ranking, and assign inter-domain influence weight coefficients to different core features through an attention mechanism to generate a spatiotemporal frequency multidimensional fusion feature topology.

[0095] Preferably, the specific implementation process of step 133 is as follows: combining the need for highlighting core associations and eliminating invalid information in the data link feature topology, a multi-stage feature optimization mechanism is constructed: First, the initial cross-domain association feature tensor is validated and screened for validity. Based on the electromagnetic propagation law of data link communication and the physical interaction mechanism of signals and errors, a feature association validity judgment rule base is established. Each feature pair corresponding to each element in the tensor is checked one by one to see if there is an actual physical association. For example, "signal frequency distribution features and error source manufacturing process features" are judged as invalid associations because there is no direct physical association. Tensor elements with no actual physical associations are marked and removed to obtain the feature tensor after physical association screening.

[0096] Next, the designed mutual information entropy quantization mechanism is used to calculate the information redundancy of the remaining elements in the feature tensor after physical association filtering. This mechanism quantifies the redundancy level between features by calculating the degree of overlap of information entropy between pairs of feature columns. When the information redundancy is higher than a preset judgment threshold (e.g., 75%), the corresponding feature column is marked as a redundant feature column. Based on the marking results, the feature tensor is pruned and purified, retaining feature columns with information redundancy below the threshold, removing redundant feature columns and related associated elements, and generating a redundant purified feature tensor. This tensor removes duplicate information, simplifies the feature dimensions, and retains key associated features.

[0097] Then, the feature columns in the redundant cleanup feature tensor are ranked by importance weight. The ranking is based on the contribution of the feature to the error resistance decision in the data chain and the close correlation between the feature and the error suppression effect. A hierarchical weighting method is adopted. First, the feature columns are divided into three categories according to their functional attributes: signal core features, error core features, and propagation environment features. Basic weights are assigned to them. Then, the weight values ​​are further adjusted according to the dynamic response sensitivity and data reliability of the features to obtain the final importance weight of each feature column. Several core feature dimensions with high ranking are extracted (the number of core feature dimensions is determined according to the computational complexity and accuracy requirements of the error resistance model) to ensure that the features most critical to error suppression and signal optimization are retained.

[0098] Finally, the feature data corresponding to the core feature dimensions are input into the attention mechanism module. This module assigns inter-domain influence weight coefficients to different core features based on the degree of correlation between the core features and error resistance strategies in different domains (space, time, and frequency). For example, the "error incident direction feature," which is closely related to spatial beamforming, is given a higher spatial influence weight coefficient, and the "error repetition period feature," which is closely related to temporal filtering, is given a higher temporal influence weight coefficient. By weighting and aggregating the correlation information of the core feature dimensions through weight coefficients, a multi-dimensional fusion feature topology of space, time, and frequency is generated. This topology clearly presents the correlation strength and influence priority of the core features in multiple domains, providing an accurate feature foundation for the subsequent construction of a multi-dimensional domain collaborative error resistance model.

[0099] Specifically, in this application, the feature association validity judgment rule base adopts a dynamic update mechanism, continuously optimizing the accuracy of the judgment rules by learning effective feature association cases in actual data chain communication scenarios. In the mutual information entropy quantification mechanism, information entropy calculation combines the temporal characteristics of data chain features, considering not only the redundancy of static values ​​but also the degree of overlap of feature trends over time, improving the comprehensiveness of redundancy identification. In the hierarchical weight assignment method, the dynamic response sensitivity of features is determined by analyzing the response speed and change magnitude of features when errors occur, and data reliability is comprehensively evaluated based on the accuracy level and verification pass rate of the data acquisition source. In the attention mechanism module, the assignment of inter-domain influence weight coefficients is determined through training sample learning. The training samples cover feature association effect data under different electromagnetic environments and different error types, ensuring that the weight coefficients can accurately reflect the actual impact of features on multi-domain error resistance strategies.

[0100] Optionally, step 2 includes:

[0101] Step 21: Extract error polarization feature values, signal propagation attenuation coefficient, and frequency domain occupancy parameters from the spatiotemporal multidimensional fusion feature topology. Based on these three types of parameters, construct a multidimensional domain error-resistant objective functional with the goals of maximizing error suppression, minimizing signal distortion, and optimizing spectrum utilization. Combine the power constraints, bandwidth constraints, and array aperture constraints of data link transmission to establish constraint boundary conditions.

[0102] Step 22: Based on the multidimensional domain robust objective functional and constraint boundary conditions, construct a cognitive-driven multidimensional domain collaborative robustness model based on dynamic environmental perception. The collaborative robustness model has built-in spatiotemporal frequency domain resource adaptation rules and dynamic weight adaptation mechanism.

[0103] Step 23: Iteratively solve the multi-dimensional domain collaborative error-resistant model, determine the spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters through multiple rounds of optimization, and generate multi-dimensional domain error-resistant control instructions after verifying the effectiveness of the two types of parameters.

[0104] Preferably, the specific implementation process of step 21 is as follows, which deeply aligns with the core requirements of data links in complex electromagnetic environments to balance error resistance, signal integrity, and spectrum utilization efficiency, and constructs a multi-objective coupled optimization system and hardware adaptation constraint mechanism: First, three types of core parameters are accurately extracted from the spatiotemporal multi-dimensional fusion feature topology. Error polarization feature values ​​are obtained through the polarization mode analysis module. This parameter quantifies the polarization state (such as the azimuth angle of linear polarization, the rotation coefficient of circular polarization, etc.) by analyzing the amplitude ratio and phase difference of the error signal on the orthogonal polarization channel, directly providing the polarization matching basis for error suppression in spatial beamforming; the signal propagation attenuation coefficient is generated by the path loss feature extraction module. Based on the signal transmission distance, the dielectric constant of the environmental medium, and the distribution of obstacles, the energy attenuation law is fitted, which is a key compensation parameter for avoiding signal distortion in time-domain filtering; the frequency domain occupancy parameter is calculated by the frequency band occupancy statistics module, and the energy proportion distribution of the error signal is statistically analyzed according to the preset frequency band interval, providing quantitative support for the differentiated allocation of frequency domain resources.

[0105] Supported by three core parameters, a multi-dimensional error-resistant objective functional is constructed. This functional forms a unified optimization criterion by coupling and fusing three core optimization objectives: maximizing error suppression is achieved by quantifying the amplitude attenuation of the error signal after spatiotemporal-frequency co-processing, with the polarization matching degree between the error polarization eigenvalue and the spatial weight, and the fit between the frequency domain occupancy parameter and the frequency band avoidance strategy as core calculation factors; minimizing signal distortion is achieved by comparing the waveform similarity, phase consistency, and amplitude deviation of the data link signal before and after processing, combined with the dynamic compensation of the signal propagation attenuation coefficient, constraining the damage amplitude of error-resistant processing to the useful signal; and optimizing spectrum utilization is achieved by optimizing the compactness of frequency domain resource allocation and the utilization rate of idle frequency bands, dividing the frequency band boundaries between the signal and the error based on the frequency domain occupancy parameter to maximize the utilization efficiency of the effective frequency band. The weight coefficients of the three objectives are dynamically adjusted according to the priority of the data link communication scenario. For example, in military tactical communication scenarios, the weight of error suppression is higher than other objectives, while in civilian broadband communication scenarios, the weight of spectrum utilization is prioritized.

[0106] Subsequently, considering the hardware physical limitations and communication protocol specifications of the data link transmission, constraint boundary conditions were established: power constraints were set based on the hardware power rating of the transmitting terminal, limiting the signal transmission power after error-resistant processing to not exceed a preset threshold, avoiding hardware overload damage and unnecessary energy loss; bandwidth constraints were determined based on the legal spectrum range specified by the data link communication protocol, clarifying the effective range of frequency domain resource allocation, preventing signal transmission failure or illegal errors caused by exceeding compliant frequency bands; array aperture constraints were determined based on the physical dimensions, number of array elements, and element spacing of the spatial array antenna, limiting the adjustment range of beamforming weights, ensuring that the algorithm design matches the hardware physical implementation capabilities, and avoiding unachievable beam pointing or gain indicators. These three types of constraints work together to constitute the feasible solution space boundary of the objective functional, ensuring that the optimization results have both excellent error-resistant performance and meet the practical requirements of engineering implementation.

[0107] Specifically, in this application, the polarization mode analysis module synchronously acquires the received signals from the orthogonal polarized antenna, calculates the amplitude ratio and phase difference of the two signals, and generates error polarization characteristic values. This process, combined with the polarization response characteristics of the array antenna, ensures that the parameters can truly reflect the polarization nature of the error. The path loss feature extraction module integrates multiple sets of measured data under different transmission scenarios to establish a mapping relationship between the signal propagation attenuation coefficient and the transmission distance and environmental parameters, thereby improving the scenario adaptability of the coefficient. The frequency band occupancy statistics module divides the entire working frequency domain according to uniform frequency band intervals, calculates the energy proportion of the error signal in each interval, and generates a continuous frequency domain occupancy parameter curve. In the coupling and fusion process of the objective functional, the weight coefficients are determined by combining offline training and online feedback. Offline training optimizes the initial weights based on a large amount of typical scenario data, while online feedback dynamically corrects them based on real-time transmission quality data. The threshold setting of the constraint boundary conditions adopts the "general standard + scenario fine-tuning" mode. The general standard strictly follows the requirements of industry hardware specifications and communication protocols, while the scenario fine-tuning adjusts the constraint leniency according to specific application scenarios (such as close-range tactical collaborative communication and long-range cross-sea communication). This avoids performance redundancy or hardware over-limitation caused by excessively loose constraints, while also preventing excessively strict constraints from limiting the optimization potential.

[0108] In the logical association between the three core parameters, the target functional, and the constraints: the error polarization eigenvalue is directly input into the calculation of the error suppression maximization objective, while simultaneously limiting the polarization matching constraint range of the spatial domain weights; the signal propagation attenuation coefficient is used in the calculation of the signal distortion minimization objective, and its magnitude determines the dynamic compensation intensity of the time-domain filtering; the frequency domain occupancy parameter supports the realization of the optimal spectrum utilization objective, while defining the constraint boundary of frequency domain resource allocation. Together, these three ensure the strong coupling and adaptability between the target functional and the constraints, providing a precise optimization benchmark for the subsequent construction of a multi-dimensional domain collaborative error-resistant model.

[0109] Optionally, step 22 includes:

[0110] Step 221: Based on the multidimensional domain robust objective functional and constraint boundary conditions, construct an initial multidimensional domain cooperative robust model;

[0111] Step 222: Real-time acquisition of electromagnetic environment change data and data link transmission status data, and correlation mapping and rule extraction processing of the two types of data to generate spatiotemporal frequency domain resource adaptation rules and embed them into the initial multidimensional domain collaborative error-resistant model;

[0112] Step 223: Analyze the differences in error and signal features in the spatiotemporal multidimensional fusion feature topology, and assign differentiated dynamic weights to the spatial beamforming weights, temporal filtering weights, and frequency resource allocation weights respectively, thereby constructing a dynamic weight adaptation mechanism and embedding it into the initial multidimensional domain collaborative error-resistant model to complete the construction of the multidimensional domain collaborative error-resistant model based on dynamic environmental perception.

[0113] Preferably, step 221 is executed first, based on the multidimensional domain error-resistant objective functional and constraint boundary conditions, to build the core framework of the initial multidimensional domain collaborative error-resistant model. This framework includes a space-time-frequency three-dimensional error-resistant module, a parameter optimization module, and a result output module. The space-time-frequency three-dimensional error-resistant module corresponds to three major error-resistant strategies: spatial beamforming, temporal filtering, and frequency resource allocation. The parameter optimization module is responsible for iteratively optimizing various error-resistant weights and strategy parameters. The result output module is used to generate the final multidimensional domain error-resistant control instructions. The framework internally realizes parameter transfer and data interaction between modules through data interfaces to ensure the coherence of the basic operation logic of the model.

[0114] Next, step 222 is executed to activate the environment and status data acquisition module, which collects electromagnetic environment change data (including error type changes, error intensity fluctuations, and changes in propagation environment medium) and data link transmission status data (including signal-to-noise ratio, transmission bit error rate, and spectrum occupancy changes) in real time. The two types of data are input into the correlation mapping module, which analyzes the causal relationship between electromagnetic environment changes and transmission status changes to extract correlation rules (e.g., the correlation law of signal-to-noise ratio decrease when error intensity increases suddenly). The rule extraction module then transforms the correlation relationship into executable spatiotemporal frequency domain resource adaptation rules (e.g., automatically increasing the error suppression weight of spatial beamforming and adjusting the frequency band avoidance range of frequency domain resource allocation when the error intensity exceeds a preset threshold). The generated spatiotemporal frequency domain resource adaptation rules are embedded into the initial multidimensional domain collaborative error resistance model through the rule embedding interface, enabling the model to dynamically adjust the resource allocation strategy according to changes in environment and status.

[0115] Then, step 223 is executed to analyze the error features and signal features in the spatiotemporal multidimensional fusion feature topology, and extract the difference parameters between the two (including polarization mode differences, time-domain waveform differences, frequency-domain distribution differences, etc.). Based on the difference parameters, a dynamic weight allocation model is constructed, and differentiated dynamic weights are assigned to the spatial beamforming weights, time-domain filtering weights, and frequency-domain resource allocation weights respectively. For example, when the polarization modes of the error and the signal are significantly different, the weight ratio of the spatial beamforming weight is increased; when the time-domain waveforms of the two are significantly different, the influence weight of the time-domain filtering weight is increased. Through the weight adaptation interface, the dynamic weight adaptation mechanism is embedded into the initial multidimensional domain collaborative error-resistant model, and the construction of the multidimensional domain collaborative error-resistant model based on dynamic environmental perception is completed. This model can dynamically adjust the error-resistant strategy and weight allocation according to changes in the electromagnetic environment, transmission state fluctuations, and the characteristic differences between the error and the signal, so as to achieve adaptive optimization of the error-resistant effect.

[0116] Specifically, in this application, the core framework of the initial multi-dimensional domain collaborative error-resistant model adopts a modular design, with clear functional boundaries for each module, facilitating the subsequent embedding of resource adaptation rules and dynamic weight adaptation mechanisms. The sampling frequency of the environmental and state data acquisition module is dynamically adjusted according to the rate of change of the data link signal, increasing the sampling frequency when signal fluctuations are severe to ensure that the acquired data can reflect real changes in a timely manner. The association mapping module adopts a scenario-based association analysis method, combining historical data under different electromagnetic environment scenarios to improve the accuracy of association rules. The rule embedding interface adopts a standardized data format to ensure that the spatiotemporal frequency domain resource adaptation rules can be seamlessly integrated into the model's parameter optimization process. In the dynamic weight allocation model, the calculation of differentiated dynamic weights is based on the quantized value of feature difference parameters. By establishing a mapping relationship table between difference parameters and weight coefficients, rapid allocation and adjustment of weights are achieved. The weight adaptation interface supports real-time updates of weight coefficients, ensuring that the dynamic weight adaptation mechanism can respond promptly to changes in errors and signal characteristics, so that the model's error-resistant strategy is always optimally adapted to the current scenario.

[0117] After the model is built, the spatiotemporal frequency domain resource adaptation rules and the dynamic weight adaptation mechanism work together: when the electromagnetic environment or transmission state changes, the spatiotemporal frequency domain resource adaptation rules trigger macroscopic adjustments to the error resistance strategy, and the dynamic weight adaptation mechanism performs microscopic optimization of the weights based on the differences in the characteristics of errors and signals. The combination of the two enables the model to have both the flexibility to cope with environmental changes and the accuracy to address the differences in characteristics, effectively improving the error resistance stability of the data link in complex dynamic electromagnetic environments.

[0118] Optionally, step 23 includes:

[0119] Step 231: Perform cognitive optimization calculation on the multi-dimensional domain collaborative error-resistant model based on the initial population, where the initial population is the combined solution vector of error-resistant weights and suppression strategy parameters;

[0120] Step 232: Calculate the fitness value of each combination solution vector in the initial population, and perform selection, crossover, and mutation evolution operations based on the fitness value to generate a new generation of population. Repeat the iterative evolution process until the set termination criterion is met to obtain the final population.

[0121] Step 233: Select the Pareto optimal solution from the final population, analyze the optimal solution to obtain the spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters, verify the effectiveness of the two types of parameters, and generate multidimensional domain error-resistant control instructions after eliminating invalid parameter combinations.

[0122] Preferably, the specific implementation process of step 23 is as follows: Combining the core requirement of data link error-resistant parameter optimization to take into account both multi-objective balance and real-time performance, a cognitive optimization solution and multi-round iterative optimization mechanism is designed: First, step 231 is executed to start the cognitive optimization solution process for the multi-dimensional domain collaborative error-resistant model, and the constituent dimensions of the initial population are determined. Each combined solution vector contains three types of error-resistant weight parameters: spatial beamforming weight, temporal filtering weight, and frequency resource allocation weight, as well as three types of error suppression strategy parameters: beam pointing strategy, filter type selection, and frequency band allocation strategy. Each element of the combined solution vector corresponds to a specific physical meaning and value range (for example, the value range of spatial beamforming weight matches the gain adjustment capability of the array antenna). Based on the core feature parameters in the spatiotemporal multi-dimensional fusion feature topology, the initial population is generated by combining uniform distribution sampling and elite initialization. This ensures the diversity of the population and improves the overall quality of the initial population by introducing historical optimal solution fragments, ensuring that the initial population can cover a better solution space range.

[0123] Next, step 232 is executed to calculate the fitness value of each combination solution vector in the initial population. The fitness function is constructed based on a multidimensional error-resistant objective functional, comprehensively considering the weighted sum of three indicators: error suppression effect, signal distortion degree, and spectral efficiency. The weight coefficients are consistent with the weights of the objective functional in step 21. Based on the fitness value, a selection strategy combining roulette wheel selection and elite retention is adopted to retain high-quality individuals with high fitness values. At the same time, to avoid local optima, a certain proportion of randomly selected individuals are introduced. Crossover is performed on the selected individuals. Arithmetic crossover is used for the error-resistant weight parameters to ensure the rationality of the parameters after crossover. For the error suppression strategy parameters... The selection, crossover, and mutation processes are used to effectively integrate different strategy combinations. After crossover, individuals undergo mutation. The weight parameters are mutated using Gaussian mutation, fluctuating within a small range around the current value to explore better solutions. The strategy parameters are mutated using random mutation to switch strategy types with a low probability and avoid population homogenization. A new generation of population is generated through selection, crossover, and mutation operations. The above iterative evolution process is repeated until the set termination criteria are met. The termination criteria include the number of iterations reaching a preset upper limit, the change in the optimal fitness value of the population over multiple generations being lower than a preset threshold, and the average fitness value of the population reaching a preset standard. The iteration stops when any of the criteria is met, and the final population is obtained.

[0124] Then, step 233 is executed to screen Pareto optimal solutions for all combined solution vectors in the final population, construct a Pareto front for multi-objective optimization, and screen out non-dominant solutions that are not dominant in the three objectives of error suppression, signal distortion, and spectrum utilization. From the Pareto optimal solution set, combined with the priority requirements of the current communication scenario of the data link (e.g., in military scenarios, solutions with better error suppression effects are preferred, while in civilian scenarios, solutions with better spectrum utilization are preferred), the final optimal solution is determined. The optimal solution is analyzed to obtain the spatiotemporal frequency domain error resistance weight matrix and error suppression strategy parameters. The rows of the weight matrix correspond to array antenna elements or frequency bands in the frequency domain, the columns correspond to different error resistance objectives, and the matrix elements are the corresponding weight coefficients. The strategy parameters specify the specific execution method of various error resistance strategies. The effectiveness of the two types of parameters is verified by simulating the data link transmission scenario to verify whether the error suppression effect after the parameter application meets expectations, whether the signal distortion is within the allowable range, and whether the power, bandwidth, and other constraint boundary conditions are met. Invalid parameter combinations that do not meet the verification requirements are eliminated, and the verified parameter combinations are encapsulated into multi-dimensional domain error resistance control instructions.

[0125] Specifically, in this application, the initial population size is dynamically adjusted according to the solution space complexity. When the solution space dimension is high, the population size is appropriately increased to ensure solution coverage. In the calculation of the fitness function, a scene adaptation factor is introduced, and the weighting ratio of the three types of indicators is dynamically adjusted according to the error type of the electromagnetic environment and the signal propagation conditions to make the fitness value more in line with the actual scene requirements. The elite retention ratio in the selection strategy is dynamically adjusted according to the iteration stage. In the early stage of iteration, fewer elite individuals are retained to ensure population diversity, and the retention ratio is increased in the later stage of iteration to accelerate convergence. The crossover probability and mutation probability are also dynamically adjusted. In the early stage of iteration, a higher crossover probability and a lower mutation probability are used to promote population evolution, and in the later stage of iteration, the crossover probability is reduced and the mutation probability is increased to avoid getting trapped in local optima. In the Pareto optimal solution selection process, a fast non-dominated sorting algorithm is used to improve selection efficiency. At the same time, a crowding calculation mechanism is introduced to select solutions with higher crowding into the optimal solution set, ensuring a uniform distribution of the solution set. In the validity verification stage, a multi-scenario simulation model is constructed, covering combinations of different error intensities and propagation environments to ensure the robustness of parameter combinations. The verified multi-dimensional domain error-resistant control instructions have clear parameter values ​​and strategy directions, which can directly drive the error-resistant processing module of the data link to perform corresponding operations.

[0126] Throughout the iterative solution process, the cognitive optimization mechanism dynamically adjusts the search direction by memorizing the historical best solution and the current population's evolutionary trend. For example, when the population fitness value does not improve for several consecutive generations, it automatically expands the search range or adjusts the mutation method to avoid the algorithm getting trapped in local optima and ensure that the final spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters can achieve the optimal error-resistant effect under multi-objective balance.

[0127] Optionally, step 3 includes:

[0128] Step 31: Analyze the error resistance weight matrix and error suppression strategy parameters in the multi-dimensional domain error resistance control instruction, retrieve the data link transmission characteristic parameter library, and generate a signal optimization processing strategy based on the two types of parameters and the parameter library data.

[0129] Step 32: Based on the signal optimization processing strategy, perform spatial beamforming processing on the data link transmission signal to generate a spatial optimized signal. Apply adaptive time-domain filtering to the spatial optimized signal to generate a space-time joint optimized signal. Perform spectral adaptive adjustment on the space-time joint optimized signal based on frequency domain resource allocation parameters to generate an error-resistant optimized signal.

[0130] Preferably, the specific implementation process of step 31 is as follows: closely adhering to the core requirement that data link signal optimization needs to adapt to error-resistant parameters and hardware transmission characteristics, a collaborative mechanism for parameter parsing and strategy generation is constructed: First, the transmission interface of the multi-dimensional domain error-resistant control command is connected, and the spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters encapsulated in the command are extracted; the spatiotemporal frequency domain error-resistant weight matrix is ​​dimensionally analyzed, the rows of the matrix correspond to the antenna elements of the data link spatial array or the frequency band division in the frequency domain, the columns correspond to different error-resistant target scenarios (such as strong error scenarios and weak error scenarios), and the element values ​​at the intersections are the weight coefficients under the corresponding scenarios, clarifying the physical meaning and control range of each weight coefficient; the error suppression strategy parameters are type-analyzed, distinguishing three types of parameters: spatial beam pointing strategy, temporal filter type selection, and frequency band allocation strategy, clarifying the specific execution requirements of each type of strategy (such as the angle range of beam pointing, the signal type adapted to the filter type, and the boundary division of the frequency band allocation).

[0131] Next, a pre-established data link transmission characteristic parameter library is retrieved. This library stores the inherent transmission parameters and scenario adaptation parameters of the data link hardware, including the gain adjustment range of the array antenna, the frequency response characteristics of the filter, the bandwidth limitation of the transmission link, and the signal modulation and demodulation method. The parameter library is categorized and stored according to the data link model and communication scenario, supporting fast retrieval and recall. The parsed error-resistant weight matrix and error suppression strategy parameters are then matched with the corresponding parameters in the parameter library. For example, the spatial beamforming weights are matched with the gain adjustment range of the array antenna, and the frequency domain resource allocation parameters are matched with the bandwidth limitation of the transmission link to verify the compatibility between the error-resistant parameters and the hardware characteristics.

[0132] The core framework for signal optimization processing strategy is constructed based on the matching results. This framework comprises three modules: spatial optimization sub-strategy, time-domain optimization sub-strategy, and frequency-domain optimization sub-strategy. The spatial optimization sub-strategy combines spatial beamforming weights and antenna gain characteristics to clarify the beam pointing adjustment steps and gain allocation scheme, ensuring that nulls are formed in the error direction without exceeding the antenna hardware capabilities. The time-domain optimization sub-strategy determines the adjustment scheme of filtering parameters (such as filter cutoff frequency and filter order) based on time-domain filtering weights and filter frequency response characteristics, adapting to different types of time-domain error suppression requirements. The frequency-domain optimization sub-strategy, based on frequency-domain resource allocation parameters and link bandwidth limitations, divides the effective frequency band and avoidance frequency band for signal transmission, clarifying the triggering conditions and execution flow for frequency band switching. The three sub-strategies are then integrated collaboratively to supplement the timing logic of strategy execution (such as spatial beamforming first, then time-domain filtering, and finally frequency-domain adjustment) and anomaly handling mechanisms (such as degradation strategies for parameter mismatch), generating a complete signal optimization processing strategy. This strategy responds to the parameter requirements of error-resistant control commands and adapts to the hardware transmission characteristics of the data link.

[0133] Specifically, in this application, the parsing process of the error-resistant weight matrix adopts a hierarchical parsing method. First, the dimension identifier of the matrix is ​​parsed, then the specific values ​​of each element are extracted, and the rationality of the values ​​is verified, eliminating outliers that exceed the physical control range. The parsing of error suppression strategy parameters introduces scenario adaptation verification to determine the matching degree between the strategy type and the current electromagnetic environment scenario. For example, in a narrowband error scenario, the rationality of the frequency band allocation strategy is verified. The data link transmission characteristic parameter library adopts a dynamic update mechanism. By periodically collecting actual transmission performance data of the data link hardware, the inherent parameters in the parameter library are corrected to improve the accuracy of parameter matching. In the parameter association matching process, a combination of threshold judgment and interpolation adaptation is used. When the error-resistant parameters exceed the inherent range of the hardware, adapted hardware executable parameters are generated through interpolation calculation to avoid the strategy failing to be implemented.

[0134] In the collaborative integration of signal optimization processing strategies, the timing logic design is based on the physical flow of signal processing, ensuring that the execution order of each sub-strategy conforms to the signal transmission rules. The anomaly handling mechanism includes two approaches: parameter degradation and strategy switching. When the conflict between the error-resistant parameters and hardware characteristics is minor, parameter degradation is used; when the conflict is significant, the strategy switches to an alternative error suppression strategy, ensuring the continuity and effectiveness of signal optimization processing. The generated signal optimization processing strategy is output in a standardized format, clearly defining the parameter settings, execution conditions, and expected results for each processing step, providing a clear execution basis for subsequent space-time-frequency collaborative optimization processing.

[0135] Optionally, step 32 includes:

[0136] Step 321: Based on the spatial beamforming weights in the signal optimization processing strategy, perform array beamforming dynamic processing on the data link transmission signal to control the beam direction to form nulls in the error direction and high-gain main lobes in the desired signal direction, and generate a spatial optimization signal accordingly.

[0137] Step 322: Adaptive time-domain filtering is used to decompose the spatial optimization signal into a time-domain signal and identify error components to eliminate time-domain superimposed error components and generate a space-time joint optimization signal.

[0138] Step 323: Based on the frequency domain resource allocation parameters in the signal optimization processing strategy, perform frequency domain resource reconstruction and spectrum adaptive adjustment on the spatiotemporal joint optimization signal to avoid error-occupied frequency bands and optimize spectrum resource allocation efficiency, so as to generate an error-resistant optimization signal.

[0139] Preferably, the specific implementation process of step 321 is as follows: Combining the core requirements of precise pointing to the desired signal and suppression of errors in the spatial domain of the data link, a dynamic beamforming mechanism for the array is designed: First, the spatial beamforming weights in the signal optimization processing strategy are extracted. The rows of this weight matrix correspond to each antenna element of the spatial array, the columns correspond to the beam control targets under different error scenarios, and the intersection elements are the excitation weights of each antenna element, clarifying the correspondence between the weights and the antenna elements and the control targets; The excitation signal amplitude and phase control parameters of each antenna element are generated according to the weight matrix. The amplitude parameter determines the signal transmission strength of the antenna element, and the phase parameter controls the transmission phase difference of the signal.

[0140] Next, the array beamforming module is activated, and the excitation control parameters of each antenna element are input into the beam controller. By adjusting the signal amplitude and phase of each antenna element, the array antenna forms a high-gain main lobe in the direction of the desired signal. The width of the main lobe is set according to the accuracy requirements of the direction of arrival of the desired signal to ensure that the signal energy is concentrated towards the receiving end. At the same time, nulls are formed in the error direction. The depth of the nulls is set according to the error intensity. The higher the error intensity, the greater the depth of the nulls, so as to significantly attenuate the error signal energy.

[0141] During beamforming, the output signal of the array antenna is acquired in real time. The beam direction monitoring module verifies whether the main lobe pointing and null position meet the strategy requirements. If there is a deviation, the excitation weight of the corresponding antenna element is finely adjusted based on the deviation value to ensure that the main lobe is stably pointing in the direction of the desired signal and the null is accurately aligned with the direction of the error. Finally, a spatially optimized signal is generated after spatial beamforming dynamic processing. This signal achieves the dual effect of enhancing the desired signal and suppressing the error signal in the spatial domain.

[0142] Preferably, in the specific technical implementation of step 322, for the time-domain error components that may still remain in the spatial optimization signal, an adaptive time-domain filtering mechanism is designed to process them: First, the spatial optimization signal is input into the time-domain signal decomposition module, and the signal is decomposed into wavelet coefficients of different frequency scales through multi-scale wavelet decomposition, and the wavelet coefficients corresponding to the signal fundamental component and the error component are distinguished—the wavelet coefficients of the signal fundamental component have concentrated amplitudes on the main frequency scale, while the wavelet coefficients of the error component show abnormal peaks on a specific frequency scale.

[0143] Then, the error component identification module is activated. Based on the amplitude and frequency distribution characteristics of wavelet coefficients, combined with the characteristic description of time-domain errors in the signal optimization processing strategy (such as the time-domain pulse width and repetition period of the error), the wavelet coefficients of the corresponding error components are marked. The marked error wavelet coefficients are attenuated through a threshold suppression mechanism, so that the wavelet coefficients of the fundamental components of the signal are not affected, thus avoiding signal distortion.

[0144] Finally, an inverse wavelet transform is performed on the processed wavelet coefficients to reconstruct the time-domain optimized spatial-temporal joint optimized signal. This signal has eliminated the error components superimposed in the time domain, and the integrity and stability of the time domain waveform are significantly improved compared with the spatial domain optimized signal.

[0145] Preferably, the specific implementation process of step 323 is as follows: combining the needs of efficient utilization of data link frequency domain resources and error avoidance, a spectrum adaptive adjustment mechanism is designed: First, extract the frequency domain resource allocation parameters in the signal optimization processing strategy. These parameters clarify the effective frequency band, avoidance frequency band and resource allocation ratio of each frequency band for signal transmission. The effective frequency band is the error-free or low-error frequency band, and the avoidance frequency band is the high-error frequency band occupied by errors.

[0146] Next, the frequency domain resource reconstruction module is activated to convert the spatiotemporally optimized signal to the frequency domain, obtaining the frequency domain spectrum distribution of the signal. Based on the frequency domain resource allocation parameters, the frequency domain components of the signal are migrated from the avoidance frequency band to the effective frequency band. During the migration, the temporal waveform characteristics of the signal remain unchanged, and only the position of the frequency domain distribution is adjusted. At the same time, the spectrum resource allocation within the effective frequency band is optimized so that the signal energy is evenly distributed within the effective frequency band, avoiding energy overload in local frequency bands and improving spectrum utilization.

[0147] Finally, an inverse Fourier transform is performed on the reconstructed frequency domain signal to convert it back to the time domain signal, generating an error-resistant optimized signal. This signal avoids the frequency band occupied by errors, optimizes the efficiency of spectrum resource allocation, and retains the core information of the desired signal, thus possessing strong error-resistant transmission capability.

[0148] Specifically, in this application, the beamforming process in step 321 introduces a dynamic calibration mechanism. By monitoring the directional changes of the desired signal and errors in real time, when the directional deviation exceeds a preset threshold, the spatial beamforming weights are automatically updated, and the positions of the main lobe and null are adjusted to adapt to the dynamically changing electromagnetic environment. In the adaptive time-domain filtering in step 322, the wavelet decomposition scale is set according to the bandwidth characteristics of the data link signal. The error component identification module combines the historical error feature library to improve the accuracy of error identification and avoid false suppression of signal components. In the frequency domain resource reconstruction in step 323, the frequency band migration adopts a smooth transition mechanism to reduce signal distortion during frequency band switching. The spectrum allocation ratio is dynamically adjusted according to the bandwidth of the effective frequency band and the signal transmission rate requirements to ensure that the utilization of spectrum resources is adapted to the signal transmission requirements.

[0149] Optionally, step 3 includes:

[0150] Step 33: Inject the error-resistant optimization signal into the data link transmission link and simultaneously collect signal transmission quality data. Compare the signal transmission quality data with the preset quality threshold to obtain the deviation value. Use the deviation value to fine-tune the multi-dimensional domain error-resistant control command parameters to control the error-resistant transmission of the error-resistant optimization signal in the electromagnetic environment.

[0151] Optionally, step 33 includes:

[0152] Step 331: Inject the error-resistant optimization signal into the data link transmission link through the transmission interface module to enable real-time signal transmission and transmission status monitoring;

[0153] Step 332: Start the link monitoring module and collect signal transmission quality data fed back from the link according to the preset sampling frequency;

[0154] Step 333: Compare the signal transmission quality data with the preset quality threshold. If there is a deviation, adjust the weight parameters and strategy parameters in the multi-dimensional domain error-resistant control command in reverse based on the deviation value, update the signal optimization processing strategy, and repeat steps 32-33.

[0155] Preferably, the specific implementation process of step 331 is as follows: In combination with the requirement that the data link signal transmission needs to ensure stability and monitorability, a signal injection and status monitoring mechanism is designed: First, the error-resistant optimized signal is input to the transmission interface module. This module is adapted to the interface protocol of the data link transmission link (such as serial interface, radio frequency interface, etc.), and performs format conversion and level adaptation of the signal to ensure that the signal meets the link transmission requirements; the converted error-resistant optimized signal is injected into the data link transmission link through the interface driver circuit to start the real-time signal transmission process.

[0156] Simultaneously, a transmission status monitoring unit is activated. This unit is connected in parallel with the transmission link to collect the transmission parameters of the link in real time, including signal transmission amplitude, phase stability, and preliminary statistical values ​​of the link bit error rate. The status analysis module performs real-time analysis on the collected transmission parameters to determine whether the signal is being transmitted normally and whether there are any abnormalities in the link (such as link interruption or abnormal signal attenuation), and generates a transmission status monitoring report, providing a basis for subsequent quality data collection and parameter fine-tuning.

[0157] Preferably, in the specific technical implementation of step 332, a precise acquisition mechanism is designed for the signal quality feedback from the transmission link: First, the link monitoring module is started, and the acquisition parameters of the monitoring module are configured according to the key points of transmission quality in the signal optimization processing strategy (such as the bit error rate, signal-to-noise ratio, and signal distortion of data link communication). The sampling frequency is set according to the signal transmission rate to ensure that the acquired data can fully reflect the changes in signal quality, and the sampling duration covers a complete signal transmission cycle.

[0158] The signal transmission quality data fed back from the link is collected according to the preset sampling frequency, including bit error rate data (the ratio of the number of erroneous symbols to the total number of symbols per unit time), signal-to-noise ratio data (the ratio of signal amplitude to noise amplitude), and signal distortion data (the waveform difference between the actual received signal and the original error-resistant optimized signal). The collected quality data is stored in association with timestamps to form a time-series transmission quality dataset to ensure the traceability and continuity of the data.

[0159] During the data collection process, the data verification module verifies the validity of the quality data, eliminates abnormal data caused by link errors or collection errors, retains valid data that truly reflects the transmission quality, and ensures the accuracy of subsequent comparison and analysis.

[0160] Preferably, the specific implementation process of step 333 is as follows: In combination with the requirement that the error resistance parameters of the data link need to be dynamically adapted to the transmission scenario, a deviation feedback and parameter fine-tuning mechanism is designed: First, a preset quality threshold is retrieved. This threshold is set according to the service requirements of data link communication (such as the bit error rate threshold of military communication being lower than that of civilian communication). The thresholds include bit error rate threshold, signal-to-noise ratio threshold, signal distortion threshold, etc., forming a quality threshold system.

[0161] The time-series transmission quality dataset is compared one by one with the preset quality threshold system, and the deviation value of each quality indicator is calculated. If the transmission bit error rate is higher than the threshold, the deviation value is positive; if the signal-to-noise ratio is lower than the threshold, the deviation value is negative; if the signal distortion exceeds the threshold range, the deviation value is the absolute value of the excess part.

[0162] Based on the magnitude and direction of the deviation, the parameter fine-tuning module is activated: for the spatial-temporal-frequency domain error-resistant weight parameters in the multi-dimensional domain error-resistant control command, the corresponding weights are fine-tuned according to the deviation ratio (e.g., when the bit error rate is too high, the spatial domain beamforming weights are fine-tuned to enhance signal gain, or the temporal domain filtering weights are fine-tuned to optimize error suppression); for the error suppression strategy parameters, if the deviation stems from insufficient strategy adaptability (e.g., the frequency domain avoidance band does not completely avoid errors), the strategy details are adjusted (e.g., the frequency domain avoidance band range is expanded).

[0163] After the parameter fine-tuning is completed, the multi-dimensional domain error-resistant control instruction is updated, and the signal optimization processing strategy is regenerated based on the updated instruction. Then, the space-time-frequency collaborative optimization processing flow of step 32 is repeated to continuously optimize the transmission quality of the error-resistant optimized signal and ensure that the signal can still meet the preset quality requirements when the electromagnetic environment changes or the link status fluctuates.

[0164] Specifically, in this application, the transmission status monitoring unit in step 331 adopts a non-intrusive monitoring method to avoid errors in the transmission link. The transmission status monitoring report is synchronized to subsequent modules in real time to ensure timely response to abnormal situations. The link monitoring module in step 332 supports dynamic adjustment of the acquisition parameters. When the transmission quality fluctuates greatly, it automatically increases the sampling frequency to capture more detailed quality changes. The preset quality threshold system in step 333 supports scenario-based configuration. Different threshold combinations can be called for different communication scenarios (such as short-range high-reliability communication and long-range tolerant communication). The deviation value is calculated using a weighted summation method, and the weights are set according to the importance of the quality indicators (such as the bit error rate weight being higher than the signal distortion weight).

[0165] During parameter fine-tuning, a fine-tuning amplitude limitation mechanism is introduced to avoid signal abrupt changes caused by over-adjustment. The fine-tuning amplitude is adaptively set based on historical fine-tuning results to ensure the smoothness and effectiveness of parameter adjustments. The updated signal optimization processing strategy maintains compatibility with the original strategy, adjusting only the deviation-related optimization steps, balancing optimization efficiency and signal stability. The entire process forms a dynamic adaptation mechanism of "signal injection - quality acquisition - deviation comparison - parameter fine-tuning - strategy update," ensuring that the transmission quality of the error-resistant optimized signal continuously meets the preset requirements.

[0166] like Figure 2 The diagram shows a signal communication system in a multi-dimensional domain of a data link, comprising:

[0167] The feature topology generation module is used to perform cross-domain correlation deconstruction of data link spatiotemporal-frequency multidimensional signals and error feature data under electromagnetic environment to obtain the original feature set in the spatiotemporal-frequency domain. Simultaneously, it integrates signal propagation characteristic data and error source attribute data to generate multi-source correlation data clusters, so as to generate spatiotemporal-frequency multidimensional fused feature topology.

[0168] The control instruction generation module is used to construct a multi-dimensional domain collaborative error-resistant model based on the spatiotemporal multi-dimensional fusion feature topology, and to dynamically allocate weights and iteratively optimize error suppression strategies for spatiotemporal and frequency domain error-resistant weight parameters and error suppression strategy parameters by combining spatiotemporal and frequency domain resource adaptation rules, thereby generating multi-dimensional domain error-resistant control instructions.

[0169] The signal optimization transmission module is used to perform spatiotemporal-frequency collaborative optimization processing on the data link transmission signal according to the multi-dimensional domain error-resistant control command to obtain the error-resistant optimized signal. It simultaneously collects the transmission quality data of the link feedback signal and finely adjusts the multi-dimensional domain error-resistant control command parameters in reverse by the deviation between the transmission quality data and the preset quality threshold, so as to control the error-resistant transmission of the error-resistant optimized signal in the electromagnetic environment.

[0170] like Figure 3 The image shows an electronic device that includes a processor and a memory.

[0171] The memory is used to store computer programs;

[0172] When the processor executes the program stored in the memory, it implements the functions of each module of the signal communication system of the data link multidimensional domain as described above, or implements the steps of the signal communication method of the data link multidimensional domain.

[0173] The above Figure 2 and Figure 3 For an exemplary explanation, please refer to the above. Figure 1 This will not be elaborated upon here.

Claims

1. A signal communication method for a multi-dimensional domain of a data link, characterized in that, include: Step 1: Perform cross-domain correlation deconstruction on the data link space-time-frequency multidimensional signal and error feature data under electromagnetic environment to obtain the original feature set in the space-time-frequency domain. Simultaneously integrate the signal propagation characteristic data and error source attribute data to generate a multi-source correlation data cluster, so as to generate a space-time-frequency multidimensional fusion feature topology. Step 2: Construct a multi-dimensional domain collaborative error-resistant model based on the spatiotemporal-frequency multi-dimensional fusion feature topology, and combine the spatiotemporal-frequency domain resource adaptation rules to dynamically allocate the error-resistant weight parameters and error suppression strategy parameters in the spatiotemporal-frequency domain and iteratively optimize the error suppression strategy, thereby generating multi-dimensional domain error-resistant control instructions; Step 3: Based on the multi-dimensional domain error-resistant control command, perform spatiotemporal-frequency collaborative optimization processing on the data link transmission signal to obtain the error-resistant optimized signal. Simultaneously collect the transmission quality data of the link feedback signal. Adjust the parameters of the multi-dimensional domain error-resistant control command in reverse by the deviation between the transmission quality data and the preset quality threshold to control the error-resistant transmission of the error-resistant optimized signal in the electromagnetic environment. Step 2 includes: Step 21: Extract error polarization feature values, signal propagation attenuation coefficient, and frequency domain occupancy parameters from the spatiotemporal multidimensional fusion feature topology. Based on these three types of parameters, construct a multidimensional domain error-resistant objective functional with the goals of maximizing error suppression, minimizing signal distortion, and optimizing spectrum utilization. Combine the power constraints, bandwidth constraints, and array aperture constraints of data link transmission to establish constraint boundary conditions. Step 22: Based on the multidimensional domain error-resistant objective functional and constraint boundary conditions, construct an initial multidimensional domain collaborative error-resistant model; collect electromagnetic environment change data and data link transmission status data in real time, and perform correlation mapping and rule extraction processing on the two types of data to generate spatiotemporal frequency domain resource adaptation rules and embed them into the initial multidimensional domain collaborative error-resistant model; analyze the differences in error and signal characteristics in the spatiotemporal frequency multidimensional fusion feature topology, and assign differentiated dynamic weights to the spatial beamforming weights, temporal filtering weights, and frequency domain resource allocation weights respectively, thereby constructing a dynamic weight adaptation mechanism and embedding it into the initial multidimensional domain collaborative error-resistant model, completing the construction of a multidimensional domain collaborative error-resistant model based on dynamic environmental perception. The collaborative error-resistant model has built-in spatiotemporal frequency domain resource adaptation rules and dynamic weight adaptation mechanism. Step 23: Iteratively solve the multi-dimensional domain collaborative error-resistant model, determine the spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters through multiple rounds of optimization, and generate multi-dimensional domain error-resistant control instructions after verifying the effectiveness of the two types of parameters.

2. The signal communication method for a multi-dimensional domain of a data link according to claim 1, characterized in that, Step 1 includes: Step 11: Collect the spatiotemporal frequency multidimensional signal, error feature data, signal propagation characteristic data, and error source attribute data of the data chain; perform format compliance verification and missing value reconstruction and repair on various types of data to generate a complete original data stack. Step 12: Perform cross-domain correlation deconstruction on the data chain space-time-frequency multidimensional signal and error feature data in the complete original data stack to obtain the original feature set in the space-time-frequency domain. Simultaneously fuse the signal propagation characteristic data and error source attribute data in the complete original data stack to generate a multi-source correlation data cluster.

3. The signal communication method for a multi-dimensional domain of a data link according to claim 2, characterized in that, Step 12 includes: Step 121: Perform spatial array feature extraction, time-domain pulse feature analysis, and frequency-domain spectral feature decomposition on the spatial-temporal-frequency multidimensional signal in the data chain of the complete original data stack to obtain the spatial-temporal-frequency signal feature extraction results. Perform error polarization parameter extraction, error intensity quantization, and error incident direction localization on the error feature data in the complete original data stack to obtain the error feature extraction results. Perform cross-domain feature association mapping processing on the spatial-temporal-frequency signal feature extraction results and the error feature extraction results to obtain the original feature set in the spatial-temporal-frequency domain. Specifically, error polarization parameter extraction involves obtaining the polarization mode and polarization angle of the error signal through the error polarization parameter extraction module. Error intensity quantization involves converting the error signal amplitude into a standardized intensity value through the error intensity quantization module. Error incident direction localization involves determining the azimuth and elevation angles of the error signal through the error incident direction localization module. Step 122: Extract the propagation attenuation law and propagation path characteristics of the signal propagation characteristics data from the complete original data stack, and fuse them with the error type information and error emission parameters in the error source attribute data, as well as perform multi-source feature association mapping to generate a multi-source associated data cluster.

4. The signal communication method for a multi-dimensional domain of a data link according to claim 3, characterized in that, Step 12 also includes: Step 123: Establish the feature correspondence between the original feature set in the space-time-frequency domain and the multi-source associated data cluster, and mark the cross-redundant features and conflicting features.

5. The signal communication method for a multi-dimensional domain of a data link according to claim 3, characterized in that, In step 121, the spatial array feature extraction is carried out by analyzing the array manifold characteristics, beam pointing characteristics, and spatial diversity gain of the received signal of the data link spatial array to obtain the spatial array feature parameters. Time-domain pulse feature analysis extracts time-domain pulse signals from the spatiotemporal multidimensional signals of the data link, and extracts pulse width, repetition period, and rise slope to form a time-domain feature subset; frequency domain... Spectral feature decomposition extracts frequency domain spectral signals from the spatiotemporal multidimensional signals in the data link by analyzing the spectral peak, bandwidth, and frequency distribution to obtain frequency domain feature subsets. The aforementioned spatial array feature parameters, temporal feature subsets, and frequency domain feature subsets together constitute the spatiotemporal signal feature extraction results.

6. The signal communication method for a multi-dimensional domain of a data link according to claim 1, characterized in that, Step 1 includes: Step 13: Align the original spatiotemporal-frequency domain feature set with the multi-source associated data clusters by time axis homogeneity and dimension normalization to obtain a temporal collaborative fusion dataset. Based on the spatiotemporal-frequency domain association criteria, perform feature clustering divide-and-conquer and cross-domain association modeling on the temporal collaborative fusion dataset to generate an initial cross-domain association feature tensor. Use the mutual information entropy quantization mechanism to remove redundant features from the initial cross-domain association feature tensor to obtain a redundant clean feature tensor. Perform core feature enhancement aggregation on the redundant clean feature tensor through the attention mechanism to generate a spatiotemporal-frequency multidimensional fusion feature topology.

7. The signal communication method for a multi-dimensional domain of a data link according to claim 6, characterized in that, Step 13 includes: Step 131: Using the timestamp of the data link receiving terminal as the reference source, the original spatiotemporal frequency domain feature set and the multi-source associated data cluster are aligned with the same source on the time axis and then interpolated and reconstructed to integrate them into a dataset with a unified time granularity. Numerical interval standardization mapping is then performed on the dataset with the unified time granularity to generate a time-series collaborative fusion dataset. Step 132: Based on the differences in technical attributes in the space-time-frequency domain, the feature data in the time-series collaborative fusion dataset is divided into static attribute data and dynamic response data to form a space-time-frequency feature classification spectrum. The signal features and error features in the space-time-frequency feature classification spectrum are modeled by pairwise association matching through a cross-domain association rule engine to generate the initial cross-domain association feature tensor.

8. The signal communication method for a multi-dimensional domain of a data link according to claim 6, characterized in that, Step 13 includes: Step 133: Validate and filter the initial cross-domain correlation feature tensor, remove tensor elements without actual physical correlation, calculate the information redundancy of the remaining tensor elements using the mutual information entropy quantization mechanism to mark redundant feature columns, and perform dimensional pruning and purification to obtain a redundant purified feature tensor. Sort the feature columns in the redundant purified feature tensor by importance weight, extract several core feature dimensions with the highest ranking, and assign inter-domain influence weight coefficients to different core features through an attention mechanism to generate a spatiotemporal-frequency multidimensional fusion feature topology.

9. The signal communication method for a multi-dimensional domain of a data link according to claim 1, characterized in that, Step 23 includes: Step 231: Perform cognitive optimization calculation on the multi-dimensional domain collaborative error-resistant model based on the initial population, where the initial population is the combined solution vector of error-resistant weights and suppression strategy parameters; Step 232: Calculate the fitness value of each combination solution vector in the initial population, and perform selection, crossover, and mutation evolution operations based on the fitness value to generate a new generation of population. Repeat the iterative evolution process until the set termination criterion is met to obtain the final population. Step 233: Select the Pareto optimal solution from the final population, analyze the optimal solution to obtain the spatiotemporal frequency domain error-resistant weight matrix and error suppression strategy parameters, verify the effectiveness of the two types of parameters, and generate multidimensional domain error-resistant control instructions after eliminating invalid parameter combinations.

10. The signal communication method for a multi-dimensional domain of a data link according to claim 1, characterized in that, Step 3 includes: Step 31: Analyze the error resistance weight matrix and error suppression strategy parameters in the multi-dimensional domain error resistance control instruction, retrieve the data link transmission characteristic parameter library, and generate a signal optimization processing strategy based on the two types of parameters and the parameter library data. Step 32: Based on the signal optimization processing strategy, perform spatial beamforming processing on the data link transmission signal to generate a spatial optimized signal. Apply adaptive time-domain filtering to the spatial optimized signal to generate a space-time joint optimized signal. Perform spectral adaptive adjustment on the space-time joint optimized signal based on frequency domain resource allocation parameters to generate an error-resistant optimized signal.

11. The signal communication method for a multi-dimensional domain of a data link according to claim 10, characterized in that, Step 32 includes: Step 321: Based on the spatial beamforming weights in the signal optimization processing strategy, perform array beamforming dynamic processing on the data link transmission signal to control the beam direction to form nulls in the error direction and high-gain main lobes in the desired signal direction, and generate a spatial optimization signal accordingly. Step 322: Adaptive time-domain filtering is used to decompose the spatial optimization signal into a time-domain signal and identify error components to eliminate time-domain superimposed error components and generate a space-time joint optimization signal. Step 323: Based on the frequency domain resource allocation parameters in the signal optimization processing strategy, perform frequency domain resource reconstruction and spectrum adaptive adjustment on the spatiotemporal joint optimization signal to avoid error-occupied frequency bands and optimize spectrum resource allocation efficiency, so as to generate an error-resistant optimization signal.

12. The signal communication method for a multi-dimensional domain of a data link according to claim 1, characterized in that, Step 3 includes: Step 33: Inject the error-resistant optimization signal into the data link transmission link and simultaneously collect signal transmission quality data. Compare the signal transmission quality data with the preset quality threshold to obtain the deviation value. Use the deviation value to fine-tune the multi-dimensional domain error-resistant control command parameters to control the error-resistant transmission of the error-resistant optimization signal in the electromagnetic environment.

13. The signal communication method for a multi-dimensional domain of a data link according to claim 12, characterized in that, Step 33 includes: Step 331: Inject the error-resistant optimization signal into the data link transmission link through the transmission interface module to enable real-time signal transmission and transmission status monitoring; Step 332: Start the link monitoring module and collect signal transmission quality data fed back from the link according to the preset sampling frequency; Step 333: Compare the signal transmission quality data with the preset quality threshold. If there is a deviation, adjust the weight parameters and strategy parameters in the multi-dimensional domain error-resistant control command in reverse based on the deviation value, update the signal optimization processing strategy, and repeat steps 32-33.

Images