A GNSS-R data radio frequency interference detection method, device and medium

Abstract

The application discloses a GNSS-R data radio frequency interference detection method and device and medium, and belongs to the technical field of satellite navigation and remote sensing data processing, the method comprises the following steps: obtaining GNSS-R observation data and preprocessing to obtain an effective delay Doppler map; the delay Doppler map is divided into a plurality of local structure analysis units according to the delay dimension, and is converted into a one-dimensional sequence; the mean and variance of each unit are calculated as statistical characteristics; the discrimination threshold is determined based on the probability density distribution of the labeled sample; the mean and variance of the unit to be detected are compared with the threshold by using the row-by-row discrimination strategy, and when both of them satisfy the condition at the same time, it is determined that there is radio frequency interference; the recognition result is spatially gridded and counted to obtain the interference spatial distribution probability. The application realizes high-sensitivity identification of weak interference and radio frequency interference under complex ground surface, effectively distinguishes natural scattering changes from real interference, has high detection precision and low calculation complexity.

Description

Technical Field

[0001] This invention relates to a method, apparatus, and medium for detecting radio frequency interference in GNSS-R data, belonging to the field of satellite navigation and remote sensing data processing technology. Background Technology

[0002] Global Navigation Satellite System (GNSS) is widely used in positioning, navigation, and timing, serving as a crucial foundational technology for modern space information services. However, due to the low power of GNSS signals during propagation, they are susceptible to external radio frequency interference (RFI), leading to signal quality degradation and even problems such as reduced positioning accuracy, abnormal observation data, or service interruptions, threatening the stability and reliability of the system. Existing RFI monitoring methods primarily rely on ground-based monitoring networks or dedicated receiving equipment. While these methods can achieve high-precision interference detection in localized areas, they generally suffer from high deployment costs, limited spatial coverage, and difficulty in achieving continuous, large-scale observations, making it challenging to meet the continuous monitoring needs on a global scale or in complex environments such as the ocean.

[0003] Building upon this foundation, in recent years, Global Navigation Satellite System-Reflectometry (GNSS-R) has provided a new technical approach for monitoring spaceborne radio frequency interference. This technology simultaneously receives GNSS direct signals and surface reflected signals, constructing a Delay-Doppler Map (DDM) to achieve a two-dimensional characterization of surface scattering characteristics. Leveraging the advantages of space platform observations, GNSS-R offers significant advantages in terms of spatial coverage and observation continuity.

[0004] GNSS-R technology was initially applied primarily to geophysical parameter inversion fields such as ocean wind speed retrieval, sea ice thickness estimation, soil moisture retrieval, and water distribution monitoring. With its expanding applications, GNSS-R has gradually been used for RFI monitoring and identification. Existing research is mostly based on low-Earth orbit (LEO) satellite GNSS-R platforms (such as the Cyclone Global Navigation Satellite System CYGNSS), utilizing intermediate frequency (IF) data or digital direct current (DDM) data for interference detection. Frequency domain analysis methods have been used to mitigate the impact of RFI on GNSS-R data quality, and multi-source LEO satellite data analysis has further validated the feasibility of using space-based platforms for global RFI monitoring. With the development of the BeiDou system and GNSS-R applications, regional GNSS signals, satellite augmentation system (SBAS) reflected signals, and ground-based L-band interference have gradually become the main sources of radio frequency interference. These interferences lead to a decrease in signal-to-noise ratio (SNR), abnormal DDM structure, and further affect the accuracy of parameter inversion. Building upon this foundation, existing research has proposed interference detection methods based on spaceborne GNSS-R data. For example, interference sources can be detected and located using CYGNSS Level-1 noise floor estimation, or continuous wave interference suppression methods can be employed to improve signal quality. Furthermore, some researchers have used burtosis as a discrimination metric to achieve RFI detection within a fixed threshold framework. These methods primarily rely on the structural change characteristics of the DDM under interference conditions for discrimination.

[0005] Although existing research has explored radio frequency interference detection based on GNSS-R, most methods rely on discriminant analysis based on global statistical indices or single numerical features of the DDM (Distributed Dynamic Distance Model). This fails to fully utilize the two-dimensional spatial structure information of the DDM, especially the local variations between different Doppler channels and along the delay dimension. Furthermore, GNSS-R observation data is susceptible to natural surface scattering effects, such as enhanced coherent scattering from the sea surface, echo characteristic changes in complex regions like mixed land-sea areas and deserts. These factors can cause local power distribution anomalies in the DDM, making its statistical characteristics somewhat similar to those caused by radio frequency interference. This reduces the reliability of methods based on single statistical indices, especially under weak interference or complex background conditions, leading to false positives or false negatives. Therefore, existing methods still fall short in characterizing the local structural differences of the DDM, lacking a fine-grained statistical description mechanism and failing to effectively distinguish between natural scattering variations and anomalous features caused by actual radio frequency interference.

[0006] The above problems indicate that existing DDM-based RFI detection methods are essentially still at the level of global feature or single statistical analysis, lacking the ability to characterize local statistical differences in the two-dimensional structure of DDM, and therefore it is difficult to achieve effective decoupling of weak interference and background scattering changes at the structural level.

[0007] In summary, there is an urgent need for a radio frequency interference detection method that can fully utilize DDM structural information, improve the ability to identify local interference, and possess good adaptability and stability. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method, device and medium for detecting radio frequency interference in GNSS-R data, so as to improve the detection accuracy and stability under weak interference and complex surface conditions, and solve the problem of false detection or missed detection under complex surface scattering conditions.

[0009] To achieve the above objectives, the present invention is implemented using the following technical solution:

[0010] In a first aspect, the present invention provides a method for detecting radio frequency interference in GNSS-R data, comprising:

[0011] Obtain the bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value from GNSS-R observation data;

[0012] The bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value are preprocessed to obtain the preprocessed effective delay Doppler image;

[0013] The preprocessed effective delay Doppler image is structurally divided according to the delay dimension. The delay power distribution sequence corresponding to each fixed delay channel is defined as a local structural analysis unit, thereby transforming the two-dimensional delay Doppler image into multiple one-dimensional local statistical analysis sequences.

[0014] For each local structural analysis unit, a row-by-row statistical analysis is performed on the delay dimension to calculate the statistical characteristics of the one-dimensional local statistical analysis sequence, thereby obtaining the mean and variance of each local structural analysis unit.

[0015] Based on the probability density distribution of the statistical characteristics of labeled radio frequency interference samples and non-radio frequency interference samples, a discrimination threshold is determined, wherein the labeled radio frequency interference samples are derived from the preprocessed effective delay Doppler map or historical observation data.

[0016] The mean and variance of each local structural analysis unit in the delayed Doppler image to be detected are compared with the discrimination threshold. When the mean and variance of any local structural analysis unit simultaneously meet the preset discrimination condition, it is determined that there is radio frequency interference in the delayed Doppler image, and the radio frequency interference identification result is obtained.

[0017] Furthermore, the data preprocessing includes land mask screening, NaN removal, and signal-to-noise ratio threshold screening; the data after land mask screening, NaN removal, and signal-to-noise ratio threshold screening is used as the effective delayed Doppler image after preprocessing.

[0018] Furthermore, the delayed Doppler image is a two-dimensional matrix data structure containing multiple Doppler channels and multiple delay sampling units, wherein each row of data corresponds to a delay power distribution sequence under a fixed Doppler channel; after the preprocessed effective delayed Doppler image is structurally divided according to the delay dimension, each local structural analysis unit corresponds to a row in the two-dimensional matrix.

[0019] Furthermore, the formulas for calculating the mean and variance of each local structural analysis unit are as follows: ;

[0020] ;

[0021] in, Indicates the first in the delayed Doppler graph line, number The power values ​​of the column; Indicates the number of sampling points in the delay dimension; Indicates the first The mean of the row data; Indicates the first Variance of row data.

[0022] Furthermore, the discrimination threshold is determined based on the probability density distribution of the statistical characteristics of the labeled radio frequency interference samples and non-radio frequency interference samples, wherein the statistical characteristics include mean and variance; the probability density distribution of the statistical characteristics is obtained by a probability density estimation method based on histograms or a probability density function construction method based on kernel density estimation; the discrimination threshold is obtained by the intersection of the probability density distributions of the statistical characteristics of the radio frequency interference samples and non-radio frequency interference samples; wherein the labeled radio frequency interference samples and non-radio frequency interference samples are obtained by manual annotation through visual interpretation or by screening historical observation data, and are used to construct a probability density distribution model of the statistical characteristics, and the discrimination threshold is obtained by data-driven modeling of the probability density functions of different categories of samples.

[0023] Furthermore, a row-by-row discrimination strategy is adopted to perform threshold determination on the statistical features of each local structural analysis unit. When the mean and variance of the statistical features of at least one local structural analysis unit simultaneously meet the corresponding discrimination threshold conditions, it is determined that the delayed Doppler image has radio frequency interference; otherwise, it is determined that there is no significant radio frequency interference. The delayed Doppler image determined to have radio frequency interference and its corresponding location information are taken as the radio frequency interference identification result.

[0024] Furthermore, the mean and variance of each local structural analysis unit are combined and calculated to construct a differential statistical characteristic index, which is used to characterize the statistical distribution differences between different Doppler channels. Based on the differential statistical characteristic index, local structural anomalies are judged to identify single structural interference and mixed structural interference, wherein the mixed structural interference includes a delayed Doppler image anomaly mode in which horseshoe scattering structure and strip scattering structure coexist.

[0025] Furthermore, the method also includes: performing spatial gridding statistical processing on the radio frequency interference identification results to obtain the spatial distribution probability of the radio frequency interference, specifically:

[0026] The study area is divided into regular spatial grid cells, and the distribution characteristics of radio frequency interference are quantitatively expressed based on these grid cells. Specifically, according to the latitude and longitude information of the observation points, the study area is divided into regular grids according to a preset spatial resolution, and the grid cell index of each observation point is calculated. The grid cell index is determined as follows: ;

[0027] ;

[0028] in, These represent the row and column indices of the grid, respectively. Indicates the latitude and longitude of the observation point. Indicates the starting coordinates of the study area. Indicates grid resolution;

[0029] Spatial statistical processing is performed on the radio frequency interference discrimination results. The study area is divided into a preset spatial grid, and the spatial grid is used as the statistical unit. The distribution characteristics of radio frequency interference within each grid unit are calculated. The probability of radio frequency interference occurring within each grid unit is defined as follows:

[0030] ;

[0031] in This indicates the number of observation points within a grid cell that are identified as radio frequency interference. This indicates the total number of valid latitude and longitude observation points contained within a grid cell; the... As a quantitative characterization index of the spatial distribution of radio frequency interference, it is used to describe the probability of radio frequency interference occurring at a given spatial scale.

[0032] Secondly, the present invention provides a radio frequency interference detection device for GNSS-R data, used to implement the radio frequency interference detection method for GNSS-R data as described in any one of the preceding claims, comprising:

[0033] The data acquisition module is used to acquire bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value from GNSS-R observation data;

[0034] The preprocessing module is used to preprocess the bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value to obtain the preprocessed effective delay Doppler image.

[0035] The structure partitioning module is used to partition the preprocessed effective delay Doppler image according to the delay dimension, and define the delay power distribution sequence corresponding to each fixed delay channel as a local structure analysis unit, thereby transforming the two-dimensional delay Doppler image into multiple one-dimensional local statistical analysis sequences.

[0036] The statistical analysis module is used to perform row-by-row statistical analysis on the delay dimension for each local structural analysis unit, calculate the statistical characteristics of the one-dimensional local statistical analysis sequence, and obtain the mean and variance of each local structural analysis unit.

[0037] The threshold determination module is used to determine the discrimination threshold based on the probability density distribution of the statistical characteristics of the labeled radio frequency interference samples and the non-radio frequency interference samples, wherein the labeled radio frequency interference samples are derived from the preprocessed effective delay Doppler map or historical observation data.

[0038] The judgment module is used to compare the mean and variance of each local structure analysis unit in the delayed Doppler image to be detected with the discrimination threshold. When the mean and variance of any local structure analysis unit simultaneously meet the preset discrimination conditions, it is determined that there is radio frequency interference in the delayed Doppler image, and the radio frequency interference identification result is obtained.

[0039] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0040] Fourthly, the present invention provides an electronic device, comprising:

[0041] Memory, used to store computer programs / instructions;

[0042] A processor for executing the computer program / instructions to implement the steps of any of the methods described above.

[0043] Fifthly, the present invention provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of any of the methods described above.

[0044] Compared with the prior art, the beneficial effects achieved by the present invention are as follows:

[0045] 1. This invention provides a method, device, and medium for detecting radio frequency interference in GNSS-R data. By dividing a two-dimensional delayed Doppler image into multiple one-dimensional local structure analysis units along the delay dimension, the mean and variance of each unit are calculated as statistical features. The discrimination threshold is adaptively determined based on the probability density distribution of labeled samples. Then, the mean and variance of each local structure unit are jointly judged using a row-by-row discrimination strategy. This achieves high-sensitivity identification of radio frequency interference under weak interference and complex surface conditions. At the same time, it effectively distinguishes between structural anomalies caused by natural scattering changes and real interference, improves detection accuracy and robustness, and reduces computational complexity.

[0046] 2. Based on the statistical characteristic probability distribution difference between radio frequency interference samples and non-radio frequency interference samples, this invention achieves adaptive selection of the discrimination threshold by determining the intersection of the two distributions, thus avoiding the problem of insufficient adaptability of the fixed threshold method to different surface scattering conditions, thereby improving the stability and adaptability of the method in complex scattering backgrounds.

[0047] 3. The present invention adopts a line-by-line discrimination mechanism to independently determine each local structural unit, so that the RFI identification process is transformed from global statistical discrimination to local structural discrimination, thereby enhancing the detection capability of local abnormal features under weak interference and mixed interference conditions;

[0048] 4. This invention constructs a discriminative model based on low-dimensional statistical features, eliminating the need for high-dimensional features or complex machine learning models. This reduces computational complexity and data processing overhead, making it suitable for rapid processing and batch application of large-scale GNSS-R observation data. Compared to existing methods based on global statistical features, this invention effectively identifies different types of radio frequency interference structural features by performing row-by-row statistical feature analysis on local structural units. It can not only detect typical elongated strong interference structures but also further identify weak mixed interference scenarios where horseshoe-shaped and elongated structures coexist, thus significantly improving the detection coverage and robustness of the method under complex interference conditions. Attached Figure Description

[0049] Figure 1 This is a flowchart of a radio frequency interference detection method based on statistical characteristics of GNSS-R data provided in an embodiment of the present invention;

[0050] Figure 2These are schematic diagrams of a DDM under three conditions provided in the embodiments of the present invention, wherein (a) is a schematic diagram of the DDM under no-interference conditions; (b) is a schematic diagram of the DDM under complete interference conditions; and (c) is a schematic diagram of the DDM under partial interference conditions.

[0051] Figure 3 This is a probability density distribution diagram of the variance of radio frequency interference samples and non-radio frequency interference samples provided in the embodiments of the present invention;

[0052] Figure 4 This is a probability density distribution diagram of the mean values ​​of radio frequency interference samples and non-radio frequency interference samples provided in the embodiments of the present invention;

[0053] Figure 5 This diagram illustrates the accuracy comparison between Kurtosis and the DDM statistical feature method, based on a sample set and using the visually interpreted DDM as the RFI ground truth.

[0054] Figure 6 This diagram illustrates the use of binary labels to annotate visual interpretation, Kurtosis, and DDM statistical methods; where (a), (b), and (c) represent visual interpretation, Kurtosis, and DDM statistical methods, respectively.

[0055] Figure 7 The results are geographic scatter plots; where (a), (b), and (c) represent the comparison of the identification results of visual interpretation, Kurtosis, and DDM statistical feature method using geographic scatter plots, respectively.

[0056] Figure 8 The diagram illustrates the comparison of classification performance between the two methods, where (a) shows the number of detected samples and classification accuracy of the two methods on the dataset, (b) shows the comparison of ROC curves of the two methods, and (c) shows the comparison of precision, recall, F1 score and MCC of the two methods.

[0057] Figure 9 This is a schematic diagram of the case study results conducted in the study area based on GNSS-R observation data from 2020 to 2025; the annual spatial distribution map of RFI identified by the DDM statistical feature method is used to reflect the spatial distribution differences and trends of RFI in different years. Detailed Implementation

[0058] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0059] Example 1: This example describes a method for detecting radio frequency interference in GNSS-R data, including:

[0060] Obtain the bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value from GNSS-R observation data;

[0061] The bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value are preprocessed to obtain the preprocessed effective delay Doppler image;

[0062] The preprocessed effective delay Doppler image is structurally divided according to the delay dimension. The delay power distribution sequence corresponding to each fixed delay channel is defined as a local structural analysis unit, thereby transforming the two-dimensional delay Doppler image into multiple one-dimensional local statistical analysis sequences.

[0063] For each local structural analysis unit, a row-by-row statistical analysis is performed on the delay dimension to calculate the statistical characteristics of the one-dimensional local statistical analysis sequence, thereby obtaining the mean and variance of each local structural analysis unit.

[0064] Based on the probability density distribution of the statistical characteristics of labeled radio frequency interference samples and non-radio frequency interference samples, a discrimination threshold is determined, wherein the labeled radio frequency interference samples are derived from the preprocessed effective delay Doppler map or historical observation data.

[0065] The mean and variance of each local structural analysis unit in the delayed Doppler image to be detected are compared with the discrimination threshold. When the mean and variance of any local structural analysis unit simultaneously meet the preset discrimination condition, it is determined that there is radio frequency interference in the delayed Doppler image, and the radio frequency interference identification result is obtained.

[0066] The radio frequency interference detection method for GNSS-R data provided in this embodiment involves the following steps in its application:

[0067] Step 1, as follows Figure 1 As shown, the bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value are obtained from GNSS-R observation data. This embodiment uses GNSS-R observation data acquired by the CYGNSS satellite platform, and the data parameters used include bistatic radar cross section, delayed Doppler peak value, longitude, latitude, and signal-to-noise ratio.

[0068] Step 2: Preprocess the bistatic radar cross section, longitude, latitude, signal-to-noise ratio (SNR), and peak value to obtain the preprocessed effective delayed Doppler image. The data preprocessing includes land masking, NaN removal, and SNR thresholding, where the SNR threshold is set to be greater than 3 dB to remove low SNR observation data. In this embodiment, firstly, missing data and outlier observations are removed; then, a land masking method is used to retain only observation data from the land area; finally, samples with an SNR greater than 3 dB are selected to obtain the preprocessed effective delayed Doppler image.

[0069] Step 3: The preprocessed effective delayed Doppler image is structurally divided according to the delay dimension. The delay power distribution sequence corresponding to each fixed delay channel is defined as a local structural analysis unit, thereby transforming the two-dimensional delayed Doppler image into multiple one-dimensional local statistical analysis sequences. In this embodiment, the delayed Doppler image is a two-dimensional matrix data structure containing 11 Doppler channels and 17 delay sampling units. Each row of data corresponds to the delay power distribution sequence under a fixed Doppler channel, and each local structural analysis unit corresponds to a row in the two-dimensional matrix. For example... Figure 2 As shown, (a) is a schematic diagram of the delayed Doppler image structure under no interference conditions, which is a single, smooth-out scattering peak; (b) is a schematic diagram of the delayed Doppler image structure under complete interference conditions, which is a thin, bright strip distributed along the delay dimension; (c) is a schematic diagram of the delayed Doppler image structure under local interference conditions, which is a mixed anomalous mode in which horseshoe-shaped scattering structure and long strip-shaped scattering structure coexist.

[0070] Step 4: For each local structural analysis unit, perform row-by-row statistical analysis along the delay dimension to calculate the statistical characteristics of the one-dimensional local statistical analysis sequence. These statistical characteristics include the mean and variance, which characterize the central tendency and dispersion of the signal intensity distribution in the delayed Doppler image. The formulas for calculating the mean and variance are as follows:

[0071] ;

[0072] ;

[0073] in, Indicates the first in the delayed Doppler graph line, number The power values ​​of the column; Indicates the number of sampling points in the delay dimension; Indicates the first The mean of the row data; Indicates the first Variance of row data.

[0074] Step 5: Determine the discrimination threshold based on the probability distribution of the statistical characteristics of the labeled RF interference samples and non-RF interference samples. The labeled RF interference samples originate from the preprocessed effective delayed Doppler images or historical observation data. In this embodiment, the delayed Doppler image samples are manually labeled through visual interpretation, resulting in 2843 effective delayed Doppler image samples, including 886 RF interference samples and the remainder labeled as non-RF interference samples. The discrimination threshold is determined based on the probability density distribution of the statistical characteristics of the labeled RF interference samples and non-RF interference samples. The probability density distribution of the statistical characteristics is obtained through a histogram-based probability density estimation method or a kernel density estimation-based probability density function construction method. The discrimination threshold is obtained by finding the intersection of the probability density distributions of the statistical characteristics of the RF interference samples and non-RF interference samples. In this embodiment, the kernel density estimation method is used to fit the probability density of the mean and variance of the RF interference samples and non-RF interference samples, and the intersection of the two distributions is taken as the discrimination threshold. The discrimination threshold is determined based on the probability density distribution results of the statistical characteristics of labeled radio frequency interference samples and non-radio frequency interference samples. It is obtained through data-driven modeling of the probability density functions of different categories of samples, rather than using a fixed empirical threshold. Figure 3 The figure shows the variance probability density distribution of samples with and without radio frequency interference. Figure 4 The figure shows the mean probability density distribution of radio frequency interference and non-radio frequency interference samples. The intersection of the two curves in the figure is the corresponding discrimination threshold.

[0075] Step 6: Compare the mean and variance of each local structural analysis unit in the delayed Doppler image to be detected with the discrimination threshold. When the mean and variance of any local structural analysis unit simultaneously meet the preset discrimination condition, it is determined that radio frequency interference exists in the delayed Doppler image, and the radio frequency interference identification result is obtained. This embodiment adopts a row-by-row discrimination strategy to perform threshold judgment on the statistical features of each local structural analysis unit. The row-by-row discrimination strategy is based on row-by-row statistical feature extraction under a fixed Doppler channel. By reducing the two-dimensional aliasing effect, it realizes the conversion of two-dimensional data to one-dimensional statistical sequence. When the mean and variance of the statistical features of at least one local structural analysis unit simultaneously meet the corresponding discrimination threshold condition, it is determined that radio frequency interference exists in the delayed Doppler image; otherwise, it is determined that there is no significant radio frequency interference. The delayed Doppler image determined to have radio frequency interference and its corresponding location information are used as the radio frequency interference identification result.

[0076] like Figure 5 As shown, based on the sample set constructed in this embodiment, the accuracy of the peak-based fixed threshold method and this method is compared using the visually interpreted delayed Doppler image as the true value of radio frequency interference. Figure 5The results show that the accuracy of this method is significantly higher than that of the traditional Kurtosis method. Figure 6 As shown, (a) is the binary labeling result of visual interpretation, (b) is the binary labeling result based on the traditional Kurtosis method, and (c) is the binary labeling result of this method. It can be seen that this method is more consistent with the visual interpretation results, with fewer missed detections and false detections. Figure 7 The geographic scatter plot shown includes (a) the spatial distribution of radio frequency interference as interpreted visually, (b) the identification result of the traditional Kurtosis method, and (c) the identification result of this method. The spatial distribution identified by this method is more consistent with the visual interpretation result. Figure 8 As shown, (a) compares the number of detected samples and classification accuracy of the two methods on the dataset, (b) compares the ROC curves of the two methods, and (c) compares the precision, recall, F1 score and Matthews correlation coefficient of the two methods. The present method is superior to the traditional Kurtosis method in all metrics.

[0077] By performing differential combination calculations on the mean and variance of each local structural analysis unit, a differential statistical characteristic index is constructed to characterize the statistical distribution differences between different Doppler channels. Based on this differential statistical characteristic index, local structural anomalies are identified, thereby simultaneously identifying single structural interference and mixed structural interference. The mixed structural interference includes delayed Doppler image anomaly patterns where horseshoe scattering structures and strip-shaped scattering structures coexist.

[0078] After constructing the radio frequency interference (RF) identification and occurrence probability prediction model, the performance of the DDM-based statistical feature method was evaluated and compared with the peak method. To quantitatively measure the classification performance of these two methods in the RF interference identification task, the evaluation focused on overall identification correctness, positive class identification ability, and comprehensive performance under class imbalance conditions. Specifically, accuracy, precision, recall, and F1 score were selected. The value is used as an evaluation metric to comprehensively assess the performance of different methods in radio frequency interference identification tasks. The specific calculation formula is as follows:

[0079]

[0080]

[0081]

[0082]

[0083] in, These are samples that the model correctly predicted as "yes". These are samples that the model correctly predicted as "no". These are samples that the model incorrectly predicted as "no". These are samples that the model incorrectly predicted as "no". Represents the overall return for a correct prediction. This represents the overall loss due to incorrect predictions; the larger the difference, the better the model performance.

[0084] Step 7: Perform spatial gridding statistical processing on the radio frequency interference identification results to obtain the spatial distribution probability of radio frequency interference. Specifically, the study area is divided into regular spatial grid units, and the distribution characteristics of radio frequency interference are quantitatively expressed based on the spatial grid units. In this embodiment, the spatial resolution of the spatial grid units is 15 km × 15 km. According to the latitude and longitude information of the observation points, the study area is divided into regular grids according to the preset spatial resolution, and the grid unit index to which each observation point belongs is calculated. The grid unit index is determined in the following way:

[0085] ;

[0086] ;

[0087] in, These represent the row and column indices of the grid, respectively. Indicates the latitude and longitude of the observation point. Indicates the starting coordinates of the study area. This represents the grid resolution. Spatial statistical processing is performed on the radio frequency interference (RF) discrimination results. The study area is divided using a preset spatial grid, and the RF interference distribution characteristics within each grid unit are calculated using this spatial grid as the statistical unit. Within each grid unit, the probability of RF interference occurrence is defined as:

[0088] ;

[0089] in This indicates the number of observation points within a grid cell that are identified as radio frequency interference. This indicates the total number of valid latitude and longitude observation points contained within a grid cell; the... As a quantitative characterization index of the spatial distribution of radio frequency interference, it is used to describe the probability of radio frequency interference occurring at a given spatial scale, and its value ranges from 0 to 1.

[0090] The following description, in conjunction with a preferred embodiment, illustrates the content involved in the above embodiments.

[0091] In a further embodiment, this embodiment selects GNSS-R observation data of the land area in the study area from 2020 to 2025 to verify the applicability of the method of the present invention under long-term series conditions. First, GNSS-R observation data from the CYGNSS satellite is acquired, and the corresponding DDM data and observation geometric information are extracted. The raw data is preprocessed, including removing invalid samples, eliminating NaN data, and filtering low-quality observation data based on the SNR threshold. Subsequently, statistical features are extracted for each DDM sample, including indicators such as mean and variance, and a feature vector is constructed to characterize the scattering structure properties. To unify the expression of RFI identification results, the study area is divided into a 15 km × 15 km spatial grid as the basic statistical unit, and the RFI distribution characteristics within each grid unit are statistically analyzed. Within each grid unit, the probability of RFI occurrence is defined as:

[0092]

[0093] This indicates the number of observations within the grid cell that are identified as RFI by the model. This indicates the total number of valid latitude and longitude observation points contained within the grid cell. It represents the statistical probability of RFI occurring at a given spatial scale, and its value ranges from 0 to 1.

[0094] Based on the DDM statistical characteristic method, land data of the study area were analyzed year by year to obtain the spatial distribution results of RFI, such as Figure 9 As shown, based on GNSS-R observation data from 2020 to 2025, the case study results in the study area, using this method, reveal the annual spatial distribution of radio frequency interference (RF) data. This effectively reflects the spatial distribution differences and trends of RF interference across different years. The figure shows the spatial distribution from 2020 to 2025 from left to right and top to bottom, revealing a year-on-year increasing trend in RF interference activity within the study area. Compared to this method, the traditional Kurtosis method identifies a smaller interference range and less pronounced variation characteristics, indicating that this method has higher sensitivity and stability. Furthermore, this method can simultaneously identify single-structure interference and mixed-structure interference, where mixed-structure interference includes DDM anomaly patterns where horseshoe-shaped and strip-shaped scattering structures coexist.

[0095] This embodiment divides a two-dimensional delayed Doppler image into multiple one-dimensional local structural analysis units along the delay dimension, calculates the mean and variance of each unit as statistical features, and adaptively determines the discrimination threshold based on the probability density distribution of labeled samples. Then, a row-by-row discrimination strategy is used to jointly determine the mean and variance of each local structural unit, achieving high-sensitivity identification of radio frequency interference under weak interference and complex surface conditions. It also effectively distinguishes between structural anomalies caused by natural scattering variations and real interference, improving detection accuracy and robustness while reducing computational complexity. Compared with existing fixed-threshold methods based on single statistical features, this method, through row-by-row statistical feature analysis of local structural units in the delayed Doppler image, can effectively identify different types of radio frequency interference structural features. It can not only detect typical elongated strong interference structures but also further identify weak mixed interference scenarios where horseshoe-shaped and elongated structures coexist, significantly improving detection coverage and robustness under complex interference conditions.

[0096] Example 2: This example provides a radio frequency interference detection device for GNSS-R data, including: Iran

[0097] The data acquisition module is used to acquire bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value from GNSS-R observation data;

[0098] The preprocessing module is used to preprocess the bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value to obtain the preprocessed effective delay Doppler image.

[0099] The structure partitioning module is used to partition the preprocessed effective delay Doppler image according to the delay dimension, and define the delay power distribution sequence corresponding to each fixed delay channel as a local structure analysis unit, thereby transforming the two-dimensional delay Doppler image into multiple one-dimensional local statistical analysis sequences.

[0100] The statistical analysis module is used to perform row-by-row statistical analysis on the delay dimension for each local structural analysis unit, calculate the statistical characteristics of the one-dimensional local statistical analysis sequence, and obtain the mean and variance of each local structural analysis unit.

[0101] The threshold determination module is used to determine the discrimination threshold based on the probability density distribution of the statistical characteristics of the labeled radio frequency interference samples and the non-radio frequency interference samples, wherein the labeled radio frequency interference samples are derived from the preprocessed effective delay Doppler map or historical observation data.

[0102] The judgment module is used to compare the mean and variance of each local structure analysis unit in the delayed Doppler image to be detected with the discrimination threshold. When the mean and variance of any local structure analysis unit simultaneously meet the preset discrimination conditions, it is determined that there is radio frequency interference in the delayed Doppler image, and the radio frequency interference identification result is obtained.

[0103] The specific functions of each module described above are explained in the relevant content of the method in Embodiment 1, and will not be repeated here.

[0104] Example 3: This example provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described in Example 1.

[0105] Example 4: This example provides an electronic device, including:

[0106] Memory, used to store computer programs / instructions;

[0107] A processor for executing the computer program / instructions to implement the steps of any of the methods described in Embodiment 1.

[0108] Example 5: This example provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the method described in any one of Examples 1.

[0109] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

[0110] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0111] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0112] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0113] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0114] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure and not to limit its protection scope. Although this disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading this disclosure, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the pending claims.

Claims

1. A method for detecting radio frequency interference in GNSS-R data, characterized in that, include: Obtain the bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value from GNSS-R observation data; The bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value are preprocessed to obtain the preprocessed effective delay Doppler image; The preprocessed effective delay Doppler image is structurally divided according to the delay dimension. The delay power distribution sequence corresponding to each fixed delay channel is defined as a local structural analysis unit, thereby transforming the two-dimensional delay Doppler image into multiple one-dimensional local statistical analysis sequences. For each local structural analysis unit, a row-by-row statistical analysis is performed on the delay dimension to calculate the statistical characteristics of the one-dimensional local statistical analysis sequence, thereby obtaining the mean and variance of each local structural analysis unit. Based on the probability density distribution of the statistical characteristics of labeled radio frequency interference samples and non-radio frequency interference samples, a discrimination threshold is determined, wherein the labeled radio frequency interference samples are derived from the preprocessed effective delay Doppler map or historical observation data. The mean and variance of each local structural analysis unit in the delayed Doppler image to be detected are compared with the discrimination threshold. When the mean and variance of any local structural analysis unit simultaneously meet the preset discrimination condition, it is determined that there is radio frequency interference in the delayed Doppler image, and the radio frequency interference identification result is obtained.

2. The method for detecting radio frequency interference in GNSS-R data according to claim 1, characterized in that: The data preprocessing includes land masking, NaN removal, and signal-to-noise ratio thresholding; the data after land masking, NaN removal, and signal-to-noise ratio thresholding are used as the effective delayed Doppler images after preprocessing.

3. The method for detecting radio frequency interference in GNSS-R data according to claim 1, characterized in that: The delayed Doppler image is a two-dimensional matrix data structure containing multiple Doppler channels and multiple delay sampling units, wherein each row of data corresponds to a delay power distribution sequence under a fixed Doppler channel; after the preprocessed effective delayed Doppler image is structurally divided according to the delay dimension, each local structural analysis unit corresponds to a row in the two-dimensional matrix.

4. The method for detecting radio frequency interference in GNSS-R data according to claim 1, characterized in that: The formulas for calculating the mean and variance of each local structural analysis unit are as follows: ; ； in, Indicates the first in the delayed Doppler graph line, number The power values ​​of the column; Indicates the number of sampling points in the delay dimension; Indicates the first The mean of the row data; Indicates the first Variance of row data.

5. The method for detecting radio frequency interference in GNSS-R data according to claim 1, characterized in that: The discrimination threshold is determined based on the probability density distribution of the statistical characteristics of labeled radio frequency interference samples and non-radio frequency interference samples. The statistical characteristics include mean and variance. The probability density distribution of the statistical characteristics is obtained by a probability density estimation method based on histograms or a probability density function construction method based on kernel density estimation. The discrimination threshold is obtained by the intersection of the probability density distributions of the statistical characteristics of radio frequency interference samples and non-radio frequency interference samples. The labeled radio frequency interference samples and non-radio frequency interference samples are obtained by manual annotation through visual interpretation or by screening historical observation data, and are used to construct a probability density distribution model of the statistical characteristics. The discrimination threshold is obtained by data-driven modeling of the probability density functions of different categories of samples.

6. The method for detecting radio frequency interference in GNSS-R data according to claim 1, characterized in that: A row-by-row discrimination strategy is adopted to determine the threshold of the statistical characteristics of each local structural analysis unit. When the mean and variance of the statistical characteristics of at least one local structural analysis unit simultaneously meet the corresponding discrimination threshold conditions, it is determined that the delayed Doppler image has radio frequency interference. Otherwise, it is determined to be without significant radio frequency interference; The delayed Doppler image and its corresponding location information that indicate the presence of radio frequency interference are used as the radio frequency interference identification result.

7. The method for detecting radio frequency interference in GNSS-R data according to claim 1, characterized in that: The mean and variance of each local structural analysis unit are combined and calculated to construct a differential statistical characteristic index, which is used to characterize the statistical distribution differences between different Doppler channels. Based on the differential statistical characteristic index, local structural anomalies are judged to identify single structural interference and mixed structural interference, wherein the mixed structural interference includes a delayed Doppler image anomaly mode in which horseshoe scattering structure and strip scattering structure coexist.

8. The method for detecting radio frequency interference in GNSS-R data according to claim 1, characterized in that: The method further includes: performing spatial gridded statistical processing on the radio frequency interference identification results to obtain the spatial distribution probability of the radio frequency interference, specifically: The study area is divided into regular spatial grid cells, and the distribution characteristics of radio frequency interference are quantitatively expressed based on these grid cells. Specifically, according to the latitude and longitude information of the observation points, the study area is divided into regular grids according to a preset spatial resolution, and the grid cell index of each observation point is calculated. The grid cell index is determined as follows: ; ； in, These represent the row and column indices of the grid, respectively. Indicates the latitude and longitude of the observation point. Indicates the starting coordinates of the study area. Indicates grid resolution; Spatial statistical processing is performed on the radio frequency interference discrimination results. The study area is divided into a preset spatial grid, and the spatial grid is used as the statistical unit. The distribution characteristics of radio frequency interference within each grid unit are calculated. The probability of radio frequency interference occurring within each grid unit is defined as follows: ； in This indicates the number of observation points within a grid cell that are identified as radio frequency interference. This indicates the total number of valid latitude and longitude observation points contained within a grid cell; the... As a quantitative characterization index of the spatial distribution of radio frequency interference, it is used to describe the probability of radio frequency interference occurring at a given spatial scale.

9. A radio frequency interference detection device for GNSS-R data, used to implement the radio frequency interference detection method for GNSS-R data according to any one of claims 1-8, characterized in that, include: The data acquisition module is used to acquire bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value from GNSS-R observation data; The preprocessing module is used to preprocess the bistatic radar cross section, longitude, latitude, signal-to-noise ratio, and peak value to obtain the preprocessed effective delay Doppler image. The structure partitioning module is used to partition the preprocessed effective delay Doppler image according to the delay dimension, and define the delay power distribution sequence corresponding to each fixed delay channel as a local structure analysis unit, thereby transforming the two-dimensional delay Doppler image into multiple one-dimensional local statistical analysis sequences. The statistical analysis module is used to perform row-by-row statistical analysis on the delay dimension for each local structural analysis unit, calculate the statistical characteristics of the one-dimensional local statistical analysis sequence, and obtain the mean and variance of each local structural analysis unit. The threshold determination module is used to determine the discrimination threshold based on the probability density distribution of the statistical characteristics of the labeled radio frequency interference samples and the non-radio frequency interference samples, wherein the labeled radio frequency interference samples are derived from the preprocessed effective delay Doppler map or historical observation data. The judgment module is used to compare the mean and variance of each local structure analysis unit in the delayed Doppler image to be detected with the discrimination threshold. When the mean and variance of any local structure analysis unit simultaneously meet the preset discrimination conditions, it is determined that there is radio frequency interference in the delayed Doppler image, and the radio frequency interference identification result is obtained.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When executed by a processor, the computer program implements the steps of the method according to any one of claims 1-8.

Images