Multiple-input multiple-output synthetic aperture radar system based on MIMO-SAR architecture
By using a MIMO-SAR architecture multi-input multi-output synthetic aperture radar system, and leveraging feature stripping and geographic information matching from the BeiDou navigation system, the problem of correspondence between signal partitions and geographic partitions is solved, achieving efficient imaging strategy adaptation and resource optimization.
Patent Information
- Application Number
- CN202610011221.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-01-06
- Publication Date
- 2026-04-24
AI Technical Summary
Existing synthetic aperture radar systems fail to dynamically select key areas based on signal characteristics and geographic information during signal processing, resulting in low data utilization efficiency, a lack of effective correlation between signal partitions and geographic partitions, inability to adapt to complex scene changes, a lack of signal collaborative analysis mechanisms, and fixed imaging strategy generation that cannot be optimized in real time.
A multi-input multi-output synthetic aperture radar system based on MIMO-SAR architecture is adopted. The signal partition identifier is separated by the feature stripping module. Combined with the geographical partition contour of the Beidou navigation system, the signal and geography are matched, the coordination index is calculated, the mapping link is dynamically corrected, and an imaging strategy library is generated.
It achieves precise correspondence between signal zones and geographical zones, improves data consistency and the adaptability of imaging strategies, optimizes resource allocation and processing efficiency, and supports real-time scene changes.
Smart Images
Figure CN121918124A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of radar remote sensing technology, specifically to a multi-input multi-output synthetic aperture radar system based on the MIMO-SAR architecture. Background Technology
[0002] Current synthetic aperture radar (SAR) systems primarily employ fixed imaging areas and pre-defined processing procedures. Existing technologies often use a uniform processing strategy across the entire radar area, failing to dynamically select key areas based on signal characteristics and geographic information. Data utilization efficiency is low, and valuable information from multi-channel signal data is not effectively extracted. Area division methods are simplistic, lacking effective correlation between signal and geographic zones. Core area selection strategies are empirical and cannot adapt to the changing needs of complex scenarios. Signal collaborative analysis mechanisms are lacking, and the complementarity between different channel signals is not fully utilized. Data consistency verification is insufficient, leading to discrepancies between radar observations and geographic information. Imaging strategies are rigidly generated and cannot be dynamically optimized based on real-time data characteristics. Existing methods need to address key technical issues such as signal feature analysis, geographic information fusion, dynamic area selection, and adaptive imaging.
[0003] Traditional SAR systems suffer from significant shortcomings in intelligent processing and adaptive imaging. Signal feature extraction is limited to a single dimension, failing to fully exploit the differences in characteristics across multiple signal channels. The method for generating zoning identifiers is crude, resulting in inaccurate correspondences between signals and geographic space. BeiDou positioning data is underutilized, and real-time geographic location information fails to effectively guide signal processing. The association matching algorithm is highly complex, leading to low efficiency in large-scale data processing. The core zoning selection criteria are subjective, resulting in insufficient accuracy in identifying important areas. The linear calculation of the coordination index fails to capture complex correlations between signals. The consistency audit method is simplistic, making it difficult to detect subtle differences. The mapping link adjustment mechanism is rigid, unable to quickly respond to data changes. The imaging strategy library is updated laggingly, failing to meet real-time processing requirements. Summary of the Invention
[0004] The purpose of this invention is to provide a multi-input multi-output synthetic aperture radar system based on the MIMO-SAR architecture to solve the problems mentioned in the background art.
[0005] To achieve the above objectives, the present invention provides a multi-input multi-output synthetic aperture radar system based on a MIMO-SAR architecture, the system comprising: The data acquisition module is used to acquire multi-channel signal data streams generated by the radar during the transmission and reception phases; The feature stripping module is used to perform feature stripping processing on multi-channel signal data streams to separate signal partition identifiers involved in imaging processing; The geographic information processing module is used to call the real-time positioning information provided by the Beidou navigation system and combine it with the geographic database to generate the geographic partition outline corresponding to the signal data. The association matching module is used to associate and match signal partition identifiers with geographic partition outlines, generate an initial mapping table, and select high-priority partitions from the initial mapping table as the core partition set. The collaborative analysis module is used to calculate the coordination index between signal units within the core partition set, generate a preferred signal combination sequence based on this, and perform data consistency audit on the geographic partition outline based on the preferred signal combination sequence to identify the differences in radar observation data in the geographic partition outline. The strategy generation module is used to dynamically correct the mapping link between the signal partition identifier and the geographic partition outline based on the difference points, and use the corrected mapping link to derive the imaging target list and imaging scheduling sequence in the signal data stream to build an imaging strategy library.
[0006] Preferably, the feature stripping module performs feature stripping processing on the multi-channel signal data stream in the following ways: performing time-frequency transformation on the multi-channel signal data stream and extracting the transformed spectral feature vector; inputting the spectral feature vector into a pre-trained feature classification network to output multiple feature clusters; each feature cluster corresponds to a signal attribute category; for each feature cluster, calculating the Doppler offset, polarization scattering matrix, and coherence coefficient distribution of its internal signal samples; defining the Doppler domain based on the Doppler offset, the polarization domain based on the polarization scattering matrix, and the coherence domain based on the coherence coefficient distribution; and using the Doppler domain, polarization domain, and coherence domain together as signal partition identifiers.
[0007] Preferably, the implementation method for calculating the Doppler shift, polarization scattering matrix, and coherence coefficient distribution of signal samples within each feature cluster further includes: normalizing the signal samples in each feature cluster to generate a normalized signal sequence; extracting historical Doppler data from the normalized signal sequence and calculating the average Doppler shift using a sliding window; simultaneously analyzing the polarization state change trajectory of the normalized signal sequence to deduce the evolution law of the polarization scattering matrix; furthermore, calculating the cross-correlation coefficients of adjacent signal segments in the normalized signal sequence to form a coherence coefficient distribution map; determining the boundary of the Doppler domain based on the average Doppler shift, determining the division of the polarization domain based on the evolution law of the polarization scattering matrix, and determining the coverage of the coherence domain based on the coherence coefficient distribution map.
[0008] Preferably, the geographic information processing module calls the real-time positioning information provided by the BeiDou Navigation Satellite System and combines it with a geographic database to generate the geographic partition outline corresponding to the signal data. The implementation method includes: receiving latitude and longitude coordinates and timestamp data sent by the BeiDou Navigation Satellite System; mapping the latitude and longitude coordinates to a digital elevation model to obtain terrain elevation information; performing time alignment processing on the terrain elevation information in combination with the timestamp data to generate a spatiotemporally consistent geographic grid; dividing the geographic grid into regions and dividing it into multiple geographic blocks according to the radar coverage probability and terrain features within the grid; and aggregating the multiple geographic blocks into a geographic partition outline.
[0009] Preferably, the method by which the association matching module associates and matches signal partition identifiers with geographic partition contours includes: establishing feature descriptors for the Doppler, polarization, and coherence domains in the signal partition identifier; simultaneously extracting the terrain type and BeiDou navigation positioning accuracy index of each geographic block in the geographic partition contour; calculating the similarity matrix between the feature descriptors and the terrain type and positioning accuracy index; and linking the items in the similarity matrix that are higher than a threshold using a clustering algorithm to form an initial mapping table.
[0010] Preferably, the implementation method of the association matching module to select high-priority partitions as the core partition set from the initial mapping table includes: analyzing the signal activity and geographical coverage density of each partition in the initial mapping table; calculating the product of signal activity and geographical coverage density as the priority score; selecting the partition with the highest priority score to be included in the core partition set; and simultaneously considering the positioning stability data provided by the Beidou navigation system to perform weighted adjustment on the priority score to ensure the spatial reliability of the core partition set.
[0011] Preferably, the collaborative analysis module calculates the collaboration index between signal units within the core partition set in the following ways: extracting the temporal and spatial features of each signal unit in the core partition set; combining the temporal and spatial features into feature vector pairs; calculating the cosine similarity and Euclidean distance between each pair of feature vectors, fusing these two measures to generate a collaboration index; sorting the signal units according to the magnitude of the collaboration index, and selecting the top-ranked combinations to form a preferred signal combination sequence.
[0012] Preferably, the implementation method of the collaborative analysis module performing data consistency audit on the geographic partition contour based on the signal preferred combination sequence includes: mapping the signal preferred combination sequence back to the geographic partition contour to obtain the geographic block corresponding to each combination; comparing the current radar observation data of the geographic block corresponding to each combination with the records in the historical data warehouse to calculate the data deviation value; verifying the spatiotemporal consistency of the data deviation value using the positioning trajectory provided by the Beidou navigation system; if the data deviation value exceeds the tolerance and the Beidou navigation trajectory supports changes, it is marked as a difference point.
[0013] Preferably, the strategy generation module dynamically corrects the mapping link between the signal partition identifier and the geographic partition outline based on the difference points by: parsing the type and influence range of the difference points to generate a difference description vector; inputting the difference description vector into the mapping adjustment model, and the model outputs the correction coefficient of the mapping link; applying the correction coefficient to adjust the association weight of the corresponding partition in the initial mapping table; and simultaneously integrating fresh positioning data from the BeiDou Navigation Satellite System to perform spatiotemporal calibration on the corrected mapping link to ensure that the mapping link is synchronized with the current geographic partition outline.
[0014] Preferably, the strategy generation module derives the imaging target list and imaging scheduling sequence in the signal data stream using the corrected mapping links in the following ways: scanning the corrected mapping links to extract all associated imaging candidate targets; calculating the signal strength priority and geographical coverage integrity score for each imaging candidate target; sorting the imaging candidate targets by score and assigning imaging time slots in conjunction with the satellite transit schedule provided by the BeiDou Navigation Satellite System; and finally generating the imaging target list and imaging scheduling sequence and storing them in the imaging strategy library.
[0015] Compared with the prior art, the beneficial effects of the present invention are: Feature stripping is performed on the multi-channel signal data stream to separate signal partition identifiers containing parameters such as beam direction and scattering characteristics. Principal component analysis and independent component analysis algorithms are used to extract the spatial distribution features of the signal and establish a correspondence with the geographic coordinate system. Real-time positioning information from the BeiDou Navigation Satellite System is used, combined with a geographic database, to generate geographic partition contours containing features such as terrain type and slope. Image segmentation algorithms are employed to ensure contour accuracy, and coordinate transformation is used to eliminate system bias.
[0016] The signal partition identifiers are associated and matched with the geographic partition outlines to generate an initial mapping table recording the matching confidence level. Through spatial overlay analysis and similarity calculation, the correspondence between signals and geographic space is established. High-priority partitions are selected based on strategic importance and data quality to form a core partition set. For signal units within the core partition set, a synergy index is calculated to evaluate the temporal synchronization and spatial complementarity between signals. A multi-objective optimization algorithm generates an optimal signal combination sequence, which is then sorted in descending order of synergy for processing.
[0017] Data consistency audits are performed on geographic partition contours based on optimal signal combination sequences, employing change detection algorithms to identify discrepancies between radar observation data and expected geographic features. By comparing echo intensity consistency and geometric shape matching, inconsistent areas are located and their severity is assessed. Mapping links are dynamically corrected based on discrepancies, and an adaptive learning algorithm optimizes the correspondence between signals and geospatial data. Correction parameters are weighted according to the spatial distribution of discrepancies, progressively improving registration accuracy.
[0018] An imaging target list and imaging scheduling sequence are derived using corrected mapping links, constructing an imaging strategy library. The target list is generated through importance assessment and task priority ranking, optimizing resource allocation and time scheduling. The strategy library contains parameter configurations for different scenarios, balancing imaging quality and processing efficiency across multiple targets. Strategy recommendations automatically select the optimal solution based on real-time characteristics, supporting online updates and continuous optimization. Attached Figure Description
[0019] Figure 1 This is a schematic diagram illustrating the working principle of the multi-input multi-output synthetic aperture radar system based on the MIMO-SAR architecture described in this invention. Figure 2 This is a flowchart illustrating the working principle of the feature stripping module. Figure 3 A flowchart illustrating the working principle of the geographic information processing module; Figure 4 This is a graph showing the similarity analysis of related matches. Figure 5 For collaborative analysis and consistency auditing diagrams. Detailed Implementation
[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0021] Please see Figure 1 This invention provides a multi-input multi-output synthetic aperture radar system based on a MIMO-SAR architecture, the system comprising: The data acquisition module collects multi-channel signal data streams generated by the radar during the transmission and reception phases. The feature stripping module performs feature stripping processing on the multi-channel signal data streams to separate the signal partition identifiers involved in imaging processing. The geographic information processing module calls the real-time positioning information provided by the BeiDou Navigation Satellite System and combines it with a geographic database to generate geographic partition contours corresponding to the signal data. The association matching module associates and matches the signal partition identifiers with the geographic partition contours and generates an initial mapping table. High-priority partitions are selected from the initial mapping table as the core partition set. The collaborative analysis module calculates the coordination degree index between signal units within the core partition set, and generates a preferred signal combination sequence accordingly. Based on the preferred signal combination sequence, it performs data consistency audit on the geographic partition contours to find the differences in radar observation data within the geographic partition contours. The strategy generation module dynamically corrects the mapping link between the signal partition identifiers and the geographic partition contours based on the differences, and uses the corrected mapping link to derive the imaging target list and imaging scheduling sequence in the signal data stream, constructing an imaging strategy library.
[0022] Example 1: See Figure 2 In specific implementation, the feature stripping module performs feature stripping processing on the multi-channel signal data stream, including performing time-frequency transformation on the multi-channel signal data stream and extracting the transformed spectral feature vector. In specific implementation, the spectral feature vector is input into a pre-trained feature classification network, which outputs multiple feature clusters. Each feature cluster corresponds to a signal attribute category. For each feature cluster, the Doppler offset, polarization scattering matrix, and coherence coefficient distribution of its internal signal samples are calculated. The Doppler domain is defined based on the Doppler offset, the polarization domain is defined based on the polarization scattering matrix, and the coherence domain is defined based on the coherence coefficient distribution. The Doppler domain, polarization domain, and coherence domain are used together as signal partition identifiers. In some embodiments, signal samples in each feature cluster are normalized to generate a normalized signal sequence. Doppler historical data are extracted from the normalized signal sequence, and the average Doppler offset is calculated using a sliding window. Simultaneously, the polarization state change trajectory of the normalized signal sequence is analyzed, and the evolution law of the polarization scattering matrix is derived. In addition, the cross-correlation coefficients of adjacent signal segments in the normalized signal sequence are calculated to form a coherence coefficient distribution map. The boundary of the Doppler domain is determined based on the average Doppler offset, the division of the polarization domain is determined based on the evolution law of the polarization scattering matrix, and the coverage of the coherence domain is determined based on the coherence coefficient distribution map.
[0023] In practical implementation, the time-frequency transformation operation uses the short-time Fourier transform method to process the multi-channel signal data stream, generating spectral feature vectors. These spectral feature vectors contain amplitude and phase information. A pre-trained feature classification network, based on a convolutional neural network structure, classifies the spectral feature vectors, outputting multiple feature clusters, each representing a unique signal attribute pattern. In the specific implementation, for the Doppler shift calculation of signal samples within each feature cluster, a sliding window method is used to process historical Doppler data, calculating the average Doppler shift. The average Doppler shift can be obtained using the following formula: in: This represents the average Doppler shift. This represents the total number of samples within the sliding window. This represents the Doppler offset value of the k-th sample. In some embodiments, the evolution of the polarization scattering matrix is derived by analyzing the polarization state change trajectory of the normalized signal sequence. The polarization state change trajectory includes the temporal changes of the horizontal and vertical polarization components. The coherence coefficient distribution map is generated by calculating the cross-correlation coefficients of adjacent signal segments in the normalized signal sequence, and the cross-correlation coefficients reflect the degree of similarity between signal segments. Optionally, the boundary of the Doppler domain is set according to the statistical range of the average Doppler offset, the division of the polarization domain is determined based on the eigenvalue distribution of the polarization scattering matrix, and the coverage of the coherence domain is defined according to the threshold region in the coherence coefficient distribution map. It can be understood that normalization processing scales the signal samples to a uniform scale to generate a normalized signal sequence, which facilitates subsequent extraction of Doppler historical data and calculation of cross-correlation coefficients. Optionally, the size of the sliding window is adjusted according to the signal sampling rate to ensure the accuracy of the average Doppler offset calculation. In some embodiments, the pre-trained feature classification network uses labeled signal data during the training phase to optimize network parameters to improve the discriminative ability of feature clusters. Example 2: See Figure 3 In specific implementation, the geographic information processing module calls the real-time positioning information provided by the BeiDou Navigation Satellite System and combines it with the geographic database to generate the geographic partition outline corresponding to the signal data. This includes receiving latitude and longitude coordinates and timestamp data sent by the BeiDou Navigation Satellite System, mapping the latitude and longitude coordinates to the digital elevation model, obtaining terrain elevation information, performing time alignment processing on the terrain elevation information in combination with the timestamp data, generating a spatiotemporally consistent geographic grid, dividing the geographic grid into regions, dividing multiple geographic blocks according to the radar coverage probability and terrain features within the grid, and aggregating multiple geographic blocks into a geographic partition outline.
[0024] In specific implementation, receiving latitude and longitude coordinates and timestamp data sent by the BeiDou Navigation Satellite System is accomplished through a serial communication interface. The latitude and longitude coordinates include longitude values, latitude values, and elevation auxiliary parameters. The timestamp data records the precise time of signal acquisition. The digital elevation model (DEM) is stored in a geographic database, containing elevation data in a regular grid format. The process of mapping latitude and longitude coordinates to the DEM is achieved through coordinate transformation and interpolation algorithms, obtaining the elevation value corresponding to each coordinate point as terrain elevation information. In some embodiments, the terrain elevation information is time-aligned using the timestamp data. A time synchronization algorithm is used to compensate for system errors between different timestamps. The time synchronization algorithm uses the time reference provided by the BeiDou Navigation Satellite System to calibrate the terrain elevation information, generating a spatiotemporally consistent geographic grid. The geographic grid is a three-dimensional data structure containing planar coordinates and elevation information. The grid cell index of the geographic grid is calculated using the following formula: in: The grid cell index representing the geographic grid. Represents latitude coordinates. Represents longitude coordinates. Indicates the minimum latitude value of the region. Indicates the minimum longitude value of the region. Indicates the grid resolution in the latitudinal direction. Indicates the grid resolution along the longitude direction. This indicates the total number of grid cells in the latitudinal direction.
[0025] In practice, the geographic grid is divided into regions, and multiple geographic blocks are divided according to the radar coverage probability and terrain features within the grid. The radar coverage probability is calculated based on the radar equation and reflects the possibility that each geographic grid unit is effectively illuminated by the radar beam. The terrain features include slope, aspect and roughness, which are derived from the digital elevation model. The criterion for dividing geographic blocks is that the radar coverage probability is higher than a set threshold and the terrain features are similar. Topographic features, including slope, aspect, and roughness, are derived from the digital elevation model. Slope is calculated by measuring the elevation change rate of grid cells, reflecting the degree of terrain tilt. Aspect is determined by analyzing the direction of the elevation gradient, indicating the orientation of the terrain surface. Roughness quantifies the degree of terrain undulation by statistically analyzing the local elevation variation coefficient. These topographic features are combined with radar coverage probability for geographic grid segmentation. Radar coverage probability is calculated based on radar equations, taking into account radar parameters and propagation loss. When dividing geographic blocks, a radar coverage probability threshold is set, and only grid cells with probabilities higher than the threshold are retained. At the same time, the similarity of topographic features of adjacent grid cells is checked. If the slope difference is less than the tolerance value and the slope aspect is consistent, they are merged into the same geographic block. Finally, multiple geographic blocks are formed and aggregated into a geographic partition outline.
[0026] Optionally, multiple geographic blocks are aggregated into a geographic partition outline. A region growing algorithm is used to merge adjacent geographic blocks. The algorithm, centered on each geographic block, checks the differences in radar coverage probability and terrain features between adjacent blocks. If the difference is less than a tolerance value, the blocks are merged, ultimately forming a continuous geographic partition outline. It can be understood that a spatiotemporally consistent geographic grid ensures a strict correspondence between radar observation data and geospatial location, and time alignment processing eliminates positioning deviations caused by asynchronous data acquisition times. In some embodiments, the resolution of the digital elevation model matches the size of the geographic grid, ensuring that terrain elevation information accurately reflects actual terrain changes. Optionally, the geographic block division results are stored in a vector polygon data format for subsequent processing by the association and matching module.
[0027] Example 3: In specific implementation, the association matching module associates and matches signal partition identifiers with geographic partition contours. This includes establishing feature descriptors for the Doppler, polarization, and coherence domains of the signal partition identifier. The Doppler domain feature descriptor is constructed based on Doppler offset statistics, the polarization domain feature descriptor is derived based on the eigenvalues of the polarization scattering matrix, and the coherence domain feature descriptor is generated based on the histogram of coherence coefficient distribution. Simultaneously, it extracts the terrain type and BeiDou navigation positioning accuracy index for each geographic block in the geographic partition contour. The terrain type is derived from the digital elevation model, including flat land, mountains, and water bodies. The BeiDou navigation positioning accuracy index reflects the positioning error range. A similarity matrix is calculated between the feature descriptors, terrain type, and positioning accuracy index. The distance between feature vectors is calculated using a cosine similarity function. A clustering algorithm links items in the similarity matrix that exceed a threshold. The clustering algorithm uses the K-means method to group high-similarity items, forming an initial mapping table. This initial mapping table records the correspondence between signal partition identifiers and geographic partition outlines. In practice, high-priority partitions are selected from the initial mapping table as the core partition set. This includes analyzing the signal activity and geographic coverage density of each partition in the initial mapping table. Signal activity is obtained by multiplying the number of signal units within the partition by the average signal strength. Geographic coverage density is calculated by the ratio of partition area to radar illumination frequency. The product of signal activity and geographic coverage density is calculated as the priority score, which is calculated using the following formula: in: Indicates priority score, Indicates signal activity. Representing geographic coverage density, the partition with the highest priority score is selected and included in the core partition set. Simultaneously, positioning stability data provided by the BeiDou Navigation Satellite System is considered, and the priority scores are weighted and adjusted. Positioning stability data includes positioning variance and historical trajectory consistency indicators. The weighting adjustment is achieved by multiplying the priority score by a stability weight factor to ensure the spatial reliability of the core partition set. In some embodiments, the Doppler domain feature descriptor includes the Doppler center frequency and bandwidth parameters, the polarization domain feature descriptor includes polarization entropy and anisotropy coefficients, and the coherence domain feature descriptor includes the mean and variance of the coherence coefficients. In some embodiments, the similarity matrix calculation uses the standardized Euclidean distance method to ensure consistency in the dimensions of different feature dimensions. Optionally, the threshold in the clustering algorithm is dynamically adjusted based on historical mapping data to optimize link accuracy. Optionally, a time decay factor is introduced in the signal activity calculation, giving greater emphasis to recent signal data. It can be understood that the BeiDou navigation positioning accuracy directly affects the reliability of geographic partition contours; high positioning accuracy ensures the accuracy of matching feature descriptors with terrain types.
[0028] See Figure 4 This paper presents the similarity analysis results of the association and matching between signal partition identifiers and geographic partition outlines. The chart uses a heatmap format to visually display the matching degree between different signal partitions and geographic blocks, with colors ranging from light to dark indicating high similarity. Blue circles mark high-similarity groups identified by the K-means clustering algorithm; these groups form the basis of the initial mapping table. The chart includes a comprehensive similarity calculation of Doppler domain feature descriptors, polarization domain feature descriptors, and coherence domain feature descriptors with terrain type and BeiDou navigation and positioning accuracy indicators. The X-axis represents the geographic block index, covering the distribution of different terrain types (flatlands, mountains, and water bodies); the Y-axis represents the signal partition index, reflecting the feature combination of Doppler offset statistics, polarization scattering matrix characteristics, and coherence coefficient distribution. Through the visualization analysis of the similarity matrix, regions with high matching between signal features and geographic features can be clearly identified; these regions are selected as important candidates for the core partition set. The boundaries of the clustering results are clearly visible, indicating that the algorithm can effectively distinguish matching patterns of different feature combinations. This visualization method provides intuitive data support for subsequent priority selection and spatial reliability assessment.
[0029] Example 4: In specific implementation, the collaborative analysis module calculates the collaboration index between signal units within the core partition set. This includes extracting the temporal and spatial features of each signal unit in the core partition set. The temporal features include the signal amplitude mean, signal amplitude variance, and signal frequency components. The spatial features include the signal source latitude and longitude coordinates and the signal propagation direction vector. The temporal and spatial features are combined into feature vector pairs, which are formed by concatenating the temporal and spatial features of two signal units. The cosine similarity and Euclidean distance between each pair of feature vectors are calculated. The cosine similarity reflects the consistency of the feature vector direction, and the Euclidean distance reflects the spatial difference of the feature vectors. These two measures are fused to generate the collaboration index, which is calculated using the following formula: in: Indicators of synergy Represents cosine similarity. Represents Euclidean distance. This represents the cosine similarity weight coefficient. The Euclidean distance weighting coefficient is used to rank signal units according to the magnitude of the coordination index. The top-ranked combinations are selected to form a preferred signal combination sequence. In specific implementations, data consistency audits are performed on geographic partition contours based on the preferred signal combination sequence. This includes mapping the preferred signal combination sequence back to the geographic partition contour, obtaining the geographic block corresponding to each combination (the geographic block is retrieved from the geographic partition contour through coordinate matching), comparing the current radar observation data of the geographic block corresponding to each combination with records in the historical data warehouse. The current radar observation data includes radar echo intensity and phase information, while the historical data warehouse records include archived radar observation data. Data deviation values are calculated using the absolute difference method. The spatiotemporal consistency of the data deviation values is verified using the positioning trajectory provided by the BeiDou Navigation Satellite System (BDS). The positioning trajectory provided by BDS includes time-series position coordinates. If the data deviation value exceeds the tolerance and the BDS trajectory supports modification, it is marked as a difference point. In some embodiments, the temporal features also include signal duration windows and signal energy distribution, while the spatial features also include signal elevation information and signal coverage. In some embodiments, the historical data warehouse uses a distributed database to store historical radar observation data; see Table 1 for the data deviation value calculation table.
[0030] Table 1: Calculation Table of Data Deviation Values Optionally, the cosine similarity weight coefficient and the Euclidean distance weight coefficient are set to fixed values during offline training. Optionally, the tolerance value is set to a constant threshold based on the measurement accuracy of the radar system. It can be understood that the feature vector pair combination encompasses information in both the temporal and spatial domains, and the synergy index integrates cosine similarity and Euclidean distance to provide a comprehensive similarity measure.
[0031] See Figure 5 This chart presents a comprehensive analysis of the signal unit coordination index and data consistency audit results within the core partition set. Using a dual Y-axis bar chart, the blue bars on the left represent the coordination index of signal unit combinations, while the bars on the right display the data deviation values for the corresponding geographic blocks. Red bars indicate discrepancies exceeding the tolerance limit, while green bars represent data deviations within the normal range. The coordination index combines cosine similarity and Euclidean distance metrics, reflecting the degree of matching between signal units in both temporal and spatial domains. Temporal features include parameters such as signal amplitude statistics, frequency components, and energy distribution; spatial features include spatial attributes such as signal source coordinates, propagation direction, and elevation information. Data deviation values are calculated by comparing current radar observation data with historical data warehouse records, reflecting changes in radar echo intensity and phase information within geographic blocks. The orange dashed line in the chart indicates the tolerance threshold for data consistency audits; discrepancies exceeding this threshold require further verification using positioning trajectories provided by the BeiDou Navigation Satellite System. This visualization method can quickly identify geographic blocks requiring focused attention, providing data support for dynamically correcting mapping links and optimizing imaging strategies. By combining collaborative analysis with consistency auditing, the system can ensure the spatiotemporal consistency of geographic partition outlines and data reliability.
[0032] Example 5: In specific implementation, the strategy generation module dynamically corrects the mapping link between the signal partition identifier and the geographic partition outline based on the difference points. This includes parsing the type and influence range of the difference points. The types of difference points include radar echo intensity differences and phase continuity differences. The influence range is quantified by the number of geographic blocks surrounding the difference points, generating a difference description vector. The difference description vector contains the difference type code and the influence range value. The difference description vector is input into the mapping adjustment model, which is based on a neural network structure. The model outputs the correction coefficient of the mapping link. The correction coefficient is a numerical adjustment factor. The correction coefficient is applied to adjust the association weight of the corresponding partition in the initial mapping table. The initial value of the association weight is the value calculated by the similarity matrix. The adjustment process is achieved through multiplication operations. At the same time, fresh positioning data from the BeiDou Navigation Satellite System is integrated. The fresh positioning data includes the latest collected latitude and longitude coordinates and positioning accuracy indicators. The corrected mapping link is spatiotemporally calibrated. The spatiotemporal calibration is completed through coordinate transformation and time synchronization processing to ensure that the mapping link is synchronized with the current geographic partition outline. In practical implementation, the strategy generation module uses the corrected mapping links to derive the imaging target list and imaging scheduling sequence in the signal data stream. This includes scanning the corrected mapping links, extracting all associated imaging candidate targets (signal partitions with valid mapping links to geographic partition contours), calculating the signal strength priority and geographic coverage integrity score for each imaging candidate target (signal strength priority based on average signal power, geographic coverage integrity score based on the coverage ratio of the target area in the radar beam), and combining the satellite transit timetable provided by the BeiDou Navigation Satellite System (which includes satellite observable time windows). The imaging candidate targets are sorted by score and assigned imaging time slots, with the imaging time slot being the smallest unit of radar system resource allocation. Finally, the imaging target list and imaging scheduling sequence are generated and stored in the imaging strategy library. The comprehensive score of the imaging candidate targets is calculated using the following formula: in: This represents the overall score of the imaging candidate targets. Indicates signal strength priority. Indicates the geographic coverage integrity score. This represents the signal strength weighting coefficient. This represents the weighting coefficient for geographic coverage integrity.
[0033] In some embodiments, the mapping adjustment model employs a fully connected neural network structure, with the number of input layer nodes matching the dimension of the difference description vector, and the number of output layer nodes matching the number of correction coefficients. In some embodiments, the imaging time slot allocation uses a greedy algorithm, prioritizing the allocation of time resources to imaging candidate targets with high comprehensive scores. Optionally, upper and lower thresholds are set for the association weight adjustment to prevent excessive weight drift. Optionally, the imaging strategy library uses a relational database to store the imaging target list and imaging scheduling sequence. It is understood that the fresh positioning data provided by the BeiDou Navigation Satellite System provides a spatial reference and time reference for the spatiotemporal calibration of the mapping links. It is understood that signal strength priority reflects the echo signal quality of the imaging candidate targets, and the geographic coverage integrity score reflects the observation completeness of the target area. The correction coefficient is applied to the numerical association weights in the initial mapping table. The weights are updated through scalar multiplication. During the integration of fresh positioning data from the BeiDou Navigation Satellite System, the fresh positioning data is matched with the geographic partition contours to ensure that the mapping links reflect the latest geospatial relationships. The scanned and corrected mapping links use a traversal algorithm to access all valid links and extract imaging candidate targets to form a temporary set. The signal strength priority calculation uses the logarithm of the average signal power. The geographic coverage integrity score is determined by the ratio of the number of grids covered by the radar beam in the target area to the total number of grids. The satellite transit timetable comes from the forecast data of the BeiDou Navigation Satellite System. The imaging time slot allocation takes into account the radar system resource constraints and the overlap of transit time windows. The imaging target list records the identification and attribute information of the imaging candidate targets. The imaging scheduling sequence defines the time arrangement and resource allocation scheme of the imaging tasks.
[0034] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0035] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.
Claims
1. A multi-input multi-output synthetic aperture radar system based on a MIMO-SAR architecture, characterized in that, The system includes: The data acquisition module is used to acquire multi-channel signal data streams generated by the radar during the transmission and reception phases; The feature stripping module is used to perform feature stripping processing on multi-channel signal data streams to separate signal partition identifiers involved in imaging processing; The geographic information processing module is used to call the real-time positioning information provided by the Beidou navigation system and combine it with the geographic database to generate the geographic partition outline corresponding to the signal data. The association matching module is used to associate and match signal partition identifiers with geographic partition outlines, generate an initial mapping table, and select high-priority partitions from the initial mapping table as the core partition set. The collaborative analysis module is used to calculate the coordination index between signal units within the core partition set, generate a preferred signal combination sequence based on this, and perform data consistency audit on the geographic partition outline based on the preferred signal combination sequence to identify the differences in radar observation data in the geographic partition outline. The strategy generation module is used to dynamically correct the mapping link between the signal partition identifier and the geographic partition outline based on the difference points, and use the corrected mapping link to derive the imaging target list and imaging scheduling sequence in the signal data stream to build an imaging strategy library.
2. The multi-input multi-output synthetic aperture radar system based on MIMO-SAR architecture according to claim 1, characterized in that, The feature stripping module performs feature stripping processing on multi-channel signal data streams in the following ways: performing time-frequency transformation on the multi-channel signal data streams and extracting the transformed spectral feature vectors; inputting the spectral feature vectors into a pre-trained feature classification network to output multiple feature clusters; each feature cluster corresponds to a signal attribute category; for each feature cluster, calculating the Doppler offset, polarization scattering matrix, and coherence coefficient distribution of its internal signal samples; defining the Doppler domain based on the Doppler offset, the polarization domain based on the polarization scattering matrix, and the coherence domain based on the coherence coefficient distribution; and using the Doppler domain, polarization domain, and coherence domain together as signal partition identifiers.
3. The MIMO-SAR-based multi-input multi-output synthetic aperture radar system according to claim 2, characterized in that, The implementation method for calculating the Doppler shift, polarization scattering matrix, and coherence coefficient distribution of signal samples within each feature cluster also includes: normalizing the signal samples in each feature cluster to generate a normalized signal sequence; extracting historical Doppler data from the normalized signal sequence and calculating the average Doppler shift using a sliding window; simultaneously analyzing the polarization state change trajectory of the normalized signal sequence to deduce the evolution law of the polarization scattering matrix; furthermore, calculating the cross-correlation coefficients of adjacent signal segments in the normalized signal sequence to form a coherence coefficient distribution map; determining the boundary of the Doppler domain based on the average Doppler shift, determining the division of the polarization domain based on the evolution law of the polarization scattering matrix, and determining the coverage of the coherence domain based on the coherence coefficient distribution map.
4. The multi-input multi-output synthetic aperture radar system based on MIMO-SAR architecture according to claim 1, characterized in that, The geographic information processing module calls the real-time positioning information provided by the BeiDou Navigation Satellite System and generates the geographic partition outline corresponding to the signal data by combining it with the geographic database. The implementation method includes: receiving latitude and longitude coordinates and timestamp data sent by the BeiDou Navigation Satellite System; mapping the latitude and longitude coordinates to the digital elevation model to obtain terrain elevation information; performing time alignment processing on the terrain elevation information in combination with the timestamp data to generate a spatiotemporally consistent geographic grid; dividing the geographic grid into regions and dividing it into multiple geographic blocks according to the radar coverage probability and terrain features within the grid; and aggregating the multiple geographic blocks into a geographic partition outline.
5. The MIMO-SAR-based multi-input multi-output synthetic aperture radar system according to claim 1, characterized in that, The method by which the association matching module associates and matches signal partition identifiers with geographic partition contours includes: establishing feature descriptors for the Doppler, polarization, and coherence domains in the signal partition identifier; simultaneously extracting the terrain type and BeiDou navigation and positioning accuracy index of each geographic block in the geographic partition contour; calculating the similarity matrix between the feature descriptors and the terrain type and positioning accuracy index; and linking the items in the similarity matrix that are higher than a threshold through a clustering algorithm to form an initial mapping table.
6. The multi-input multi-output synthetic aperture radar system based on MIMO-SAR architecture according to claim 1, characterized in that, The implementation method of the association matching module to select high-priority partitions as the core partition set from the initial mapping table includes: analyzing the signal activity and geographical coverage density of each partition in the initial mapping table; calculating the product of signal activity and geographical coverage density as the priority score; selecting the partition with the highest priority score to be included in the core partition set; and simultaneously considering the positioning stability data provided by the Beidou navigation system to make weighted adjustments to the priority scores to ensure the spatial reliability of the core partition set.
7. The MIMO-SAR-based multi-input multi-output synthetic aperture radar system according to claim 1, characterized in that, The collaborative analysis module calculates the collaboration index between signal units within the core partition set in the following ways: extracting the temporal and spatial features of each signal unit in the core partition set; combining the temporal and spatial features into feature vector pairs; calculating the cosine similarity and Euclidean distance between each pair of feature vectors, fusing these two measures to generate a collaboration index; sorting the signal units according to the magnitude of the collaboration index, and selecting the top-ranked combinations to form a preferred signal combination sequence.
8. The MIMO-SAR-based multi-input multi-output synthetic aperture radar system according to claim 7, characterized in that, The collaborative analysis module performs data consistency auditing on geographic partition contours based on the preferred signal combination sequence in the following ways: mapping the preferred signal combination sequence back to the geographic partition contour to obtain the geographic block corresponding to each combination; comparing the current radar observation data of the geographic block corresponding to each combination with the records in the historical data warehouse to calculate the data deviation value; verifying the spatiotemporal consistency of the data deviation value using the positioning trajectory provided by the BeiDou Navigation Satellite System; if the data deviation value exceeds the tolerance and the BeiDou Navigation Satellite System supports changes, it is marked as a difference point.
9. The MIMO-SAR-based multi-input multi-output synthetic aperture radar system according to claim 1, characterized in that, The strategy generation module dynamically corrects the mapping link between the signal partition identifier and the geographic partition outline based on the difference points. The implementation method includes: parsing the type and influence range of the difference points to generate a difference description vector; inputting the difference description vector into the mapping adjustment model, and the model outputs the correction coefficient of the mapping link; applying the correction coefficient to adjust the association weight of the corresponding partition in the initial mapping table; and simultaneously integrating fresh positioning data from the BeiDou Navigation Satellite System to perform spatiotemporal calibration on the corrected mapping link to ensure that the mapping link is synchronized with the current geographic partition outline.
10. The multi-input multi-output synthetic aperture radar system based on MIMO-SAR architecture according to claim 1, characterized in that, The strategy generation module derives the imaging target list and imaging scheduling sequence in the signal data stream using the corrected mapping links in the following ways: scanning the corrected mapping links to extract all associated imaging candidate targets; calculating the signal strength priority and geographical coverage integrity score for each imaging candidate target; sorting the imaging candidate targets by score and assigning imaging time slots based on the satellite transit schedule provided by the BeiDou Navigation Satellite System; and finally generating the imaging target list and imaging scheduling sequence and storing them in the imaging strategy library.