A two-stage clutter suppression method based on space-time domain

Abstract

This invention belongs to the field of integrated radar signal processing technology, specifically involving a two-stage clutter suppression method based on the space-time domain, which solves the problem of clutter suppression in complex low-altitude environments. First, the first stage, based on CLAM (Clear-and-Field Analysis), eliminates the dependence on the radial velocity difference between the target and clutter, directly and accurately shielding clutter from the physical space dimension, efficiently eliminating strong static ground object interference while preserving the energy of low-speed UAVs without loss. Second, the second stage, with its candidate angle traversal and circle fitting algorithm, accurately estimates and reduces the DC bias caused by residual static clutter, successfully overcoming the blind spot problem when the target and clutter spatial angles are close, enabling accurate extraction of weak, low-speed UAV targets at close range.

Description

Technical Field

[0001] This invention belongs to the field of integrated sensing and radar signal processing technology, specifically relating to a two-stage clutter suppression method based on the space-time domain. Background Technology

[0002] In the field of low-altitude security, target detection based on integrated sensor-interaction (ISAC) is a prerequisite for subsequent spatial positioning and trajectory tracking. In actual low-altitude detection environments, there are numerous buildings, towers, and other strongly reflective backgrounds. These static objects create extremely strong environmental clutter, severely masking the weak echoes of UAV targets. Furthermore, non-cooperative UAVs are often characterized by being "low, slow, and small," resulting in slow translational flight speeds and a Doppler frequency shift close to zero.

[0003] Traditional radar systems often employ Moving Target Indication (MTI) or Moving Target Detection (MTD) for clutter suppression. Since the radial velocity of static ground clutter such as buildings is approximately zero, traditional high-pass filters exhibit a significant stopband notch near the zero point in the Doppler frequency domain. When UAVs fly at low speeds, traditional Doppler filters, while filtering out zero-frequency clutter, are highly susceptible to mistakenly detecting UAVs with slow radial translation, leading to severe low-speed detection blind spots and missed target detection.

[0004] To overcome the damage to low-speed targets caused by frequency domain filtering, spatial domain processing can be introduced. However, a simple global spatial zero-forcing mechanism is highly dependent on the spatial separability of the target and clutter. When the UAV target and static ground objects are close in spatial angle, the global zero-forcing matrix will inevitably eliminate the weak target signal along with the clutter. Therefore, there is an urgent need for a novel clutter suppression method that can effectively extract the translational signal of low-speed UAVs in a strong clutter background and overcome the blind zone of close-range angles. To this end, this invention provides a two-stage clutter suppression method based on the space-time domain. Summary of the Invention

[0005] The purpose of this invention is to provide a two-stage clutter suppression method based on the space-time domain. This method addresses the problems of traditional Doppler filtering failing in low-speed scenarios and the suppression blind zone of single-space filtering when the target and clutter angles are similar. The aim of this method is to accurately filter out strong static interference in complex backgrounds and fully extract the translational echo signal of "low, slow and small" UAVs.

[0006] The specific technical solution adopted by this invention is as follows: A two-stage clutter suppression method based on the space-time domain includes the following steps: Step 1: Construct the prior spatial clutter map (CLAM) offline; In the offline phase where no UAV targets enter, the base station receives static background clutter, extracts the slow time series of strong clutter range cells to construct a spatial covariance matrix, and uses the two-dimensional MUSIC algorithm to obtain the prior angle of arrival set of the main static clutter in the base station coverage area to construct a spatial clutter map. Step 2: Perform the first stage of airspace zeroing clutter suppression and initial target estimation online; During the detection phase, a mixed signal containing UAVs and clutter is acquired. An a priori clutter spatial manifold matrix is ​​constructed by calling the offline clutter map. Then, a strictly orthogonal zero-forcing (ZF) matrix is ​​designed to perform spatial projection filtering of strong static clutter. Subsequently, the parameters of the suppressed signal are estimated using the two-dimensional MUSIC algorithm to obtain a preliminary set of angle estimates for UAV targets. Step 3: Construct a joint angle set and perform a point-by-point traversal to force zero; To prevent the target from being forced to zero due to its proximity to clutter angles, the prior clutter map and the estimated target angle are merged into a joint angle set; then, single candidate angles in the set are locked in turn, and the remaining angles are extracted to construct a local interference manifold matrix and a local orthogonal projection operator to filter out the relative spatial interference from the current viewpoint. Step 4: Perform the second stage of slow time-domain DC filtering based on circle fitting (CFA); The slow time sequence of a specific subcarrier after traversing to zero is mapped to the complex baseband I / Q plane. Since the UAV has translational velocity, its slow time phase angle rotates continuously with time, which is represented as a standard circular arc around a huge static clutter DC bias point on the complex plane. The Taubin circle fitting algorithm is used to accurately solve for and subtract this fixed clutter bias. The Doppler rotation component of the UAV is translated and reset to the origin. Finally, the constant false alarm rate (CFAR) decision is combined to output a clean UAV target.

[0007] The technical effects achieved by this invention are as follows: This invention combines spatial projection filtering with slow-time circular fitting sequence cancellation to solve the clutter suppression problem in complex low-altitude environments. First, the first stage, based on CLAM, eliminates the dependence on the radial velocity difference between the target and clutter, directly and accurately shielding clutter from the physical space dimension, efficiently eliminating strong static ground object interference while preserving the energy of low-speed UAVs without loss. Second, the second stage, with its candidate angle traversal and circular fitting algorithm, accurately estimates and reduces the DC bias caused by residual static clutter, successfully overcoming the blind spot problem when the target and clutter spatial angles are close, enabling accurate extraction of weak, low-speed UAV targets at close angles. Attached Figure Description

[0008] Figure 1 This is a model diagram of a complex low-altitude environment scene according to the present invention; Figure 2This is a schematic diagram of the circle fitting suppression algorithm based on the I / Q plane in this invention; Figure 3 This is an overall flowchart of the two-stage clutter suppression method based on the space-time domain of this invention; Figure 4 shows the angle measurement results of this invention in a scenario where the clutter and target angles are close. Figure 4a It is an unprocessed two-dimensional angular spectrum; Figure 4b This is the two-dimensional angular spectrum after processing by the algorithm in this paper; Figure 4c This is a comparison of the azimuth spectrum before and after processing by the algorithm in this paper; Figure 4d This is a comparison of the elevation angle spectrum before and after processing by the algorithm in this paper; Figure 5 shows the RMSE diagram of the angle measurement of the present invention in a scenario where the clutter and target angles are close. Figure 5a It is the azimuth RMSE; Figure 5b It is the pitch angle RMSE. Detailed Implementation

[0009] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0010] like Figure 1 As shown in Figure 5, a two-stage clutter suppression method based on the space-time domain is presented. The scenario considered in this invention is the detection of low-speed non-cooperative UAV targets flying in the airspace by an ISAC base station, and includes the following steps: Step 1: Construct the prior spatial clutter map (CLAM) offline; In the offline phase where no UAV targets enter, the base station receives static background clutter, extracts the slow time series of strong clutter range cells to construct a spatial covariance matrix, and uses the two-dimensional MUSIC algorithm to obtain the prior angle of arrival set of the main static clutter in the base station coverage area to construct a spatial clutter map. Step 2: Perform the first stage of airspace zeroing clutter suppression and initial target estimation online; During the detection phase, a mixed signal containing UAVs and clutter is acquired. An a priori clutter spatial manifold matrix is ​​constructed by calling the offline clutter map. Then, a strictly orthogonal zero-forcing (ZF) matrix is ​​designed to perform spatial projection filtering of strong static clutter. Subsequently, the parameters of the suppressed signal are estimated using the two-dimensional MUSIC algorithm to obtain a preliminary set of angle estimates for UAV targets. Step 3: Construct a joint angle set and perform a point-by-point traversal to force zero; To prevent the target from being forced to zero due to its proximity to clutter angles, the prior clutter map and the estimated target angle are merged into a joint angle set; then, single candidate angles in the set are locked in turn, and the remaining angles are extracted to construct a local interference manifold matrix and a local orthogonal projection operator to filter out the relative spatial interference from the current viewpoint. Step 4: Perform the second stage of slow time-domain DC filtering based on circle fitting (CFA); The slow time sequence of a specific subcarrier after traversing to zero is mapped to the complex baseband I / Q plane. Since the UAV has translational velocity, its slow time phase angle rotates continuously with time, which is represented as a standard circular arc around a huge static clutter DC bias point on the complex plane. The Taubin circle fitting algorithm is used to accurately solve for and subtract this fixed clutter bias. The Doppler rotation component of the UAV is translated and reset to the origin. Finally, the constant false alarm rate (CFAR) decision is combined to output a clean UAV target.

[0011] Preferably, step 1 specifically includes: When sensing pure clutter background offline, the received signal is converted to the range domain through time-frequency preprocessing, and a strong clutter range cell is selected. Extract continuous The OFDM symbols constitute a snapshot matrix. , where subscript Represents noise, This indicates the number of antenna channels in the receiving array. Represents the field of complex numbers, that is The dimension is Complex matrices; calculate the spatial covariance matrix. and obtain the noise subspace ; Constructing spatial pseudospectral functions using the two-dimensional MUSIC algorithm: ; in, Indicates the azimuth angle of the receiving array and pitch angle Spatial guidance vector at the location; This is the conjugate transpose of the guiding vector; The noise subspace matrix The conjugate transpose of; After extracting the high-energy angular peak and traversing all strong clutter range cells, a structure containing... The prior angles of arrival of a static clutter, i.e., the spatial clutter map. ;in, Indicates azimuth. Indicates pitch angle; subscript The traversal sequence number of the detected static clutter ( ), and They represent the estimated first and second halves of the series, respectively. The azimuth and elevation angles of a static clutter.

[0012] Preferably, step 2 specifically includes: When a low-altitude UAV target enters the detection airspace, the airspace clutter map is invoked. Extract the corresponding steering vector to construct the a priori clutter space manifold matrix. ;in, It is a matrix composed of the steering vectors corresponding to all prior static clutter angles. This indicates the total number of detected static clutter. Based on the orthogonal projection theorem, a zero-forcing filter matrix for suppressing clutter subspace is constructed. This formula uses the principle of orthogonal projection to calculate the clutter null space. For orthogonal projection operators, The dimension is The identity matrix, clutter space manifold matrix The conjugate transpose of . This represents the matrix inversion operation; Acting on online receiving mixed signal matrix get ,in This represents the clean received signal matrix after filtering to remove static clutter components; the static clutter is projected onto the null space for removal, and a two-dimensional MUSIC algorithm is used to further process it. Estimate the initial set of UAV target angles. ; where subscript Represents the target signal. and These represent the azimuth and pitch angles of the UAV target obtained from the preliminary estimates, respectively.

[0013] Preferably, step 3 specifically includes: Construct a joint angle set ; For the first in the set Candidate angles Extract the guidance vectors from all other viewpoints except that angle to construct the interference manifold matrix. And generate local orthogonal projection operators. ; To this end, Frobenius norm normalization was used to obtain... Perform local spatial filtering on the global receiver matrix This forces the elimination of all spatial interference outside the current perspective.

[0014] Preferably, step 4 specifically includes: To remove the DC bias in the slow time dimension, the Taubin circle fitting algorithm is used to model the scattered points in the complex plane. The specific steps are as follows: Let the first In the nth subcarrier sequence The value of each sampling point is ,in, For subcarrier index, For sampling point index, For clutter indexing, and These represent the real and imaginary parts of the signal sampling point, respectively. The imaginary unit; define intermediate variables. Construct the algebraic residual of the subcarrier circular trajectory: , in, To estimate the algebraic parameters describing the circular trajectory of the subcarrier, the Taubin fitting model is transformed into a generalized eigenvalue solving problem by minimizing the algebraic error and introducing normalization constraints: ; in, Let be a vector consisting of the parameters to be estimated, with superscripts... Indicates vector transpose; It is a second-order moment matrix constructed based on the observed data samples; This is a normalized constraint matrix explicitly constructed using the first-order mean of the samples. These are generalized eigenvalues; For matrix bundle Perform eigenvalue decomposition and select the eigenvector corresponding to the smallest positive eigenvalue. ,in These are the estimated values ​​of the corresponding algebraic parameters; from this, the first... Complex estimate of the static clutter equivalent center on each subcarrier : ; in, Physically, this represents the fixed DC bias component corresponding to the residual static clutter on the subcarrier; after obtaining precise positioning, this DC bias component is subtracted point by point from the original slow-time snapshot sequence: ; in, This indicates that the DC bias has been eliminated, i.e., the clean received signal after removing residual static clutter; Finally, by combining the CFAR energy decision threshold, it is determined whether there is a drone in that angle direction. If there is, the corresponding signal is saved; otherwise, the other angles in the joint angle set are traversed until the entire set is traversed.

[0015] Example Test Description: Simulation verification was performed according to the method described in the specific implementation plan: This embodiment focuses on the simulation analysis of a single ISAC base station detecting non-cooperative UAV targets in complex low-altitude airspace. To fully verify the target extraction and dynamic angle measurement performance of the proposed two-stage clutter suppression method (ZF+CFA) based on the space-time domain in the simulation space, the UAV and clutter are set to have similar angles. The ISAC system parameters and clutter and UAV parameter settings are shown in Tables 1, 2, and 3.

[0016]

[0017]

[0018] Figure 4 shows the MUSIC angle spectrum processed by the proposed two-stage space-time joint algorithm (ZF+CFA) in a scenario where the spatial angles of a non-cooperative UAV and strong ground clutter are very close. The results show that, after the two-stage processing, the weak UAV target signal, originally masked by strong clutter in the airspace, is effectively decoupled. The processed angle spectrum not only accurately obtains the sharp peak value at the target's azimuth but also suppresses the clutter floor in non-target areas to an extremely low level, successfully overcoming the near-angle resolution blind zone of conventional spatial filtering.

[0019] To further quantitatively evaluate the angle measurement performance under near-angle interference scenarios, Figure 5 compares the RMSE performance curves of the proposed algorithm and the traditional MTI algorithm. Analysis shows that when the target is in low-speed translational motion (v=2m / s), the traditional MTI algorithm suffers from false positives due to being trapped in the zero-frequency stopband, while the proposed algorithm achieves angle measurement accuracy close to the ideal angle measurement result without clutter, demonstrating the algorithm's effectiveness under low-speed conditions. When the target speed increases to v=10m / s, the performance of both algorithms tends to be comparable. This comparison intuitively demonstrates the advantage of the proposed algorithm in addressing the low-speed blind zone of "low, slow, and small" targets.

[0020] In summary, the ZF+CFA two-stage architecture proposed in this invention successfully balances clutter suppression and low-speed UAV signal protection in near-angle strong clutter scenarios. This method demonstrates superior clutter suppression performance compared to traditional frequency domain algorithms, fully validating its feasibility in real-world complex low-altitude sensing scenarios.

[0021] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A two-stage clutter suppression method based on the space-time domain, characterized in that: Includes the following steps: Step 1: Construct the prior spatial clutter map (CLAM) offline; In the offline phase where no UAV targets enter, the base station receives static background clutter, extracts the slow time series of strong clutter range cells to construct a spatial covariance matrix, and uses the two-dimensional MUSIC algorithm to obtain the prior angle of arrival set of the main static clutter in the base station coverage area to construct a spatial clutter map. Step 2: Perform the first stage of airspace zeroing clutter suppression and initial target estimation online; During the detection phase, a mixed signal containing UAVs and clutter is acquired. An a priori clutter spatial manifold matrix is ​​constructed by calling the offline clutter map. Then, a strictly orthogonal zero-forcing (ZF) matrix is ​​designed to perform spatial projection filtering of strong static clutter. Subsequently, the two-dimensional MUSIC algorithm is used to estimate the parameters of the suppressed signal to obtain a preliminary set of angle estimates for the UAV target. Step 3: Construct a joint angle set and perform a point-by-point traversal to force zero; The prior clutter map and the initial estimated target angle are merged into a joint angle set; the single candidate angle in the set is locked in turn, and the remaining angles are extracted to construct a local interference manifold matrix and a local orthogonal projection operator to filter out the relative spatial interference of the current viewpoint; Step 4: Perform the second stage of slow time-domain DC filtering based on circle fitting (CFA); The slow time sequence of a specific subcarrier after traversing to zero is mapped to the complex baseband I / Q plane. Since the UAV has translational velocity, its slow time phase angle rotates continuously with time, which is represented as a standard circular arc around a huge static clutter DC bias point on the complex plane. The Taubin circle fitting algorithm is used to accurately solve for and subtract this fixed clutter bias. The Doppler rotation component of the UAV is translated and reset to the origin. Finally, the constant false alarm rate (CFAR) decision is combined to output a clean UAV target.

2. The two-stage clutter suppression method based on the space-time domain according to claim 1, characterized in that: Specifically, step 1 is as follows: When sensing pure clutter background offline, the received signal is converted to the range domain through time-frequency preprocessing, and a strong clutter range cell is selected. Extract continuous The OFDM symbols constitute a snapshot matrix. , where subscript Represents noise, This indicates the number of antenna channels in the receiving array. Represents the field of complex numbers, that is The dimension is Complex matrices; calculate the spatial covariance matrix. and obtain the noise subspace ; Constructing spatial pseudospectral functions using the two-dimensional MUSIC algorithm: ； in, Indicates the azimuth angle of the receiving array and pitch angle Spatial guidance vector at the location; This is the conjugate transpose of the guiding vector; The noise subspace matrix The conjugate transpose of; After extracting the high-energy angular peak and traversing all strong clutter range cells, a structure containing... The prior angles of arrival of a static clutter, i.e., the spatial clutter map. ;in, Indicates azimuth. Indicates pitch angle; subscript The traversal sequence number of the detected static clutter ( ), and They represent the estimated first and second halves of the series, respectively. The azimuth and elevation angles of a static clutter.

3. The two-stage clutter suppression method based on the space-time domain according to claim 2, characterized in that: Step 2 specifically includes: When a low-altitude UAV target enters the detection airspace, the airspace clutter map is invoked. Extract the corresponding steering vector to construct the a priori clutter space manifold matrix. ;in, It is a matrix composed of the steering vectors corresponding to all prior static clutter angles. This indicates the total number of detected static clutter. Based on the orthogonal projection theorem, a zero-forcing filter matrix for suppressing clutter subspace is constructed. This formula uses the principle of orthogonal projection to calculate the clutter null space. For orthogonal projection operators, The dimension is The identity matrix, clutter space manifold matrix The conjugate transpose of . This represents the matrix inversion operation; Acting on online receiving mixed signal matrix get ,in This represents the clean received signal matrix after filtering to remove static clutter components; the static clutter is projected onto the null space for removal, and a two-dimensional MUSIC algorithm is used to further process it. Estimate the initial set of UAV target angles. ; where subscript Represents the target signal. and These represent the azimuth and pitch angles of the UAV target obtained from the preliminary estimates, respectively.

4. The two-stage clutter suppression method based on the space-time domain according to claim 1, characterized in that: Step 3 specifically includes: Construct a joint angle set ; For the first in the set Candidate angles Extract the guidance vectors from all other viewpoints except that angle to construct the interference manifold matrix. And generate local orthogonal projection operators. ; To this end, Frobenius norm normalization was used to obtain... Perform local spatial filtering on the global receiver matrix This forces the elimination of all spatial interference outside the current perspective.

5. The two-stage clutter suppression method based on the space-time domain according to claim 1, characterized in that: Step 4 specifically includes: To remove the DC bias in the slow time dimension, the Taubin circle fitting algorithm is used to model the scattered points in the complex plane. The specific steps are as follows: Let the first In the nth subcarrier sequence The value of each sampling point is ,in, For subcarrier index, For sampling point index, For clutter indexing, and These represent the real and imaginary parts of the signal sampling point, respectively. The imaginary unit; define intermediate variables. Construct the algebraic residual of the subcarrier circular trajectory: ， in, To estimate the algebraic parameters describing the circular trajectory of the subcarrier, the Taubin fitting model is transformed into a generalized eigenvalue solving problem by minimizing the algebraic error and introducing normalization constraints: ； in, Let be a vector consisting of the parameters to be estimated, with superscripts... Indicates vector transpose; It is a second-order moment matrix constructed based on the observed data samples; This is a normalized constraint matrix explicitly constructed using the first-order mean of the samples. These are generalized eigenvalues; For matrix bundle Perform eigenvalue decomposition and select the eigenvector corresponding to the smallest positive eigenvalue. ,in These are the estimated values ​​of the corresponding algebraic parameters; from this, the first... Complex estimate of the static clutter equivalent center on each subcarrier : ； in, Physically, this represents the fixed DC bias component corresponding to the residual static clutter on the subcarrier; after obtaining precise positioning, this DC bias component is subtracted point by point from the original slow-time snapshot sequence: ； in, This indicates that the DC bias has been eliminated, i.e., the clean received signal after removing residual static clutter; Finally, by combining the CFAR energy decision threshold, it is determined whether there is a drone in that angle direction. If there is, the corresponding signal is saved; otherwise, the other angles in the joint angle set are traversed until the entire set is traversed.

Images