An adaptive subspace detection method and system suitable for partially homogeneous environments

Abstract

The present application belongs to the technical field of multi-channel signal detection, and particularly relates to a self-adaptive subspace detection method and system suitable for a partially homogeneous environment, comprising the following steps: Step 1: constructing a target signal subspace matrix and an interference signal subspace matrix; Step 2: constructing a sample covariance matrix by using a training sample; Step 3: constructing a quasi-whitening matrix by using the sample covariance matrix, and then obtaining a whitening matrix by performing quasi-whitening processing; Step 4: constructing an orthogonal projection matrix of the target signal and a corresponding complementary projection matrix; Step 5: constructing an intermediate variable matrix and a non-zero eigenvalue decomposition, and obtaining a non-uniform parameter; Step 6: constructing a detection statistic by using a to-be-detected data matrix, the orthogonal projection matrix of the target signal, the orthogonal projection matrix of the interference signal and the non-uniform parameter; Step 7: determining a detection threshold; and Step 8: determining whether the target exists. The present application improves the detection performance of an airborne radar in a partially homogeneous environment.

Description

Technical Field

[0001] This invention belongs to the field of multi-channel signal detection technology, specifically relating to an adaptive subspace detection method and system suitable for partially uniform environments. Background Technology

[0002] With the continuous improvement of airborne radar technology and the development of pulse coherent processing technology, airborne radar can acquire multi-channel data and achieve efficient detection of target signals through Space-time Adaptive Detection (STAD) technology. Currently, the most widely used model in STAD is the subspace model. In this model, the row and column elements of the signal are located in a known subspace, but the corresponding coordinates are unknown. Both target signals and interference signals can be represented by the subspace model.

[0003] Furthermore, in adaptive detection, the STAD method utilizes the correlation characteristics of signal and noise in different channels (i.e., the spatial information of the target), thus exhibiting superior detection performance compared to traditional single-channel signal detection. However, existing STAD methods typically assume a uniform detection background, meaning the statistical characteristics of noise in the target sample are consistent with those in the training samples. In real-world scenarios, due to environmental variations and limitations of the radar itself, the statistical characteristics of noise between the target sample and the training samples often differ significantly, severely impacting detector performance. A commonly used model to describe environmental non-uniformity is the partially uniform environment model, in which the covariance matrices of the training samples and the target sample differ by an unknown non-uniform parameter.

[0004] Currently, for the adaptive target detection problem in partially homogeneous environments, the main approach is to design detectors based on the Generalized Likelihood Ratio Test (GLRT) criterion. Such GLRT detectors can achieve good detection performance under certain scenarios. However, since the noise covariance matrix, the target signal subspace, and the subspace coordinates are all unknown, GLRT detectors cannot achieve detection advantages in all parameter scenarios, making the development of new subspace detection methods essential. Based on this consideration, this invention employs the Durbin criterion to design a novel adaptive subspace detection method and system suitable for partially homogeneous environments. Its aim is to improve the detection performance of airborne radar in partially homogeneous environments and provide a new avenue for the application of STAD technology. Summary of the Invention

[0005] The purpose of this invention is to provide an adaptive subspace detection method and system suitable for partially homogeneous environments, with the aim of improving the detection performance of airborne radar in partially homogeneous environments.

[0006] The specific technical solution adopted by this invention is as follows:

[0007] An adaptive subspace detection method applicable to partially homogeneous environments includes the following steps:

[0008] Step 1: Construct the target signal subspace matrix and the interference signal subspace matrix, respectively using matrix... and The dimensions of the matrices are respectively and ,in , Indicates the number of system channels;

[0009] Step 2: Construct the sampling covariance matrix using training samples ,in Represents training samples, symbol Indicates conjugate transpose;

[0010] Step 3: Utilize the sampling covariance matrix Construct a quasi-whitening matrix, and then apply it to the target signal subspace matrix. Interference signal subspace matrix and the data matrix to be detected Quasi-whitening processing was performed to obtain whitening matrices. , and The data matrix to be detected The dimension is , Indicates extended dimensions;

[0011] Step 4: Using the quasi-whitening matrix Construct the orthogonal projection matrix of the target signal and the corresponding complementary projection matrix Simultaneously utilizing the quasi-whitening matrix Construct the orthogonal projection matrix of the interference signal and the corresponding complementary projection matrix ;

[0012] Step 5: Construct intermediate variables and non-zero eigenvalues ​​using the orthogonal projection matrix of the quasi-whitened data matrix to be detected and the interference signal, and obtain non-uniform parameters by combining the system dimension. ;

[0013] Step 6: Construct detection statistics using the quasi-whitened data matrix to be detected, the orthogonal projection matrix of the target signal, the orthogonal projection matrix of the interference signal, and the non-uniform parameters;

[0014] Step 7: Determine the detection threshold using the preset false alarm probability value;

[0015] Step 8: Compare the detection statistic with the detection threshold to determine whether the target exists.

[0016] Preferably, in step 1, the constructed target signal matrix and interference signal matrix All are full-rank unitary matrices, and unitary matrices and Linearly independent.

[0017] Preferably, in step 3, the sampling covariance matrix is ​​used. The constructed quasi-whitening matrix is ​​compared with the target signal subspace matrix. Interference signal subspace matrix and the data matrix to be detected The specific steps for performing a near-whitening process are as follows: , and .

[0018] Preferably, in step 4, the quasi-whitened target signal matrix is ​​used. The constructed orthogonal projection matrix and the complement projection matrix are respectively expressed as: and ,in Represents the inverse of a matrix. For dimension is The identity matrix; utilizing the quasi-whitened interference signal matrix The constructed orthogonal projection matrix and the complement projection matrix are expressed as follows: and .

[0019] Preferably, in step 5, the intermediate variable constructed using the orthogonal projection matrix of the quasi-whitened data matrix to be detected and the interference signal is... Obtain unknown parameters The solution to the following formula: ;in, The number of training samples, for The One non-zero eigenvalue, This represents the total number of non-zero eigenvalues.

[0020] Preferably, in step 6, the detection statistic constructed using the quasi-whitened data matrix to be detected, the orthogonal projection matrix of the target signal, the orthogonal projection matrix of the interference signal, and the non-uniform parameters is expressed as follows:

[0021] ;

[0022] in, For dimension is The identity matrix, Represents the trace of a matrix.

[0023] Preferably, in step 7, the detection threshold determined using the preset false alarm probability value is:

[0024] ;in, Indicates the detection threshold. , For the number of Monte Carlo simulations, This represents the false alarm probability value set by the system. This indicates the rounding operation. For sequence

[0025] ;

[0026] When arranged from largest to smallest The maximum value, for The conjugate transpose of the matrix. , This represents the first data point of the test data containing only interference and thermal noise components. Secondary implementation .

[0027] Preferably, in step 8, the detection statistic is compared with the detection threshold to determine whether the target exists, and the determination is made in the following two cases:

[0028] If the detection statistic Greater than or equal to the detection threshold If so, then the target is determined to exist;

[0029] If the detection statistic Less than the detection threshold If the target does not exist, then it is determined that the target does not exist.

[0030] An adaptive subspace detection system suitable for partially homogeneous environments includes:

[0031] Target and Interference Data Matrix Construction Module: Used to construct the target signal subspace matrix and the interference signal subspace matrix;

[0032] Sampling covariance matrix construction module: Used to construct the sampling covariance matrix from training samples;

[0033] Data quasi-whitening module: used to perform quasi-whitening processing on the target signal subspace matrix, interference signal subspace matrix and the data matrix to be detected, to obtain the quasi-whitened target signal subspace matrix, the quasi-whitened interference signal subspace matrix and the quasi-whitened data matrix to be detected, respectively;

[0034] Orthogonal projection matrix construction module: used to construct the orthogonal projection matrix of the target signal and the orthogonal projection matrix of the interference signal respectively using the quasi-whitened target signal subspace matrix and the quasi-whitened interference signal subspace matrix;

[0035] Non-uniform parameter solving module: used to perform eigenvalue decomposition on intermediate variables constructed by the orthogonal projection matrix of the data matrix to be detected and the interference signal, and then obtain the non-uniform parameters of the environment by combining the system dimension;

[0036] The detection statistics construction module is used to construct detection statistics from the quasi-whitened target signal matrix, the orthogonal projection matrix of the interference signal, the data matrix to be detected, and environmental non-uniform parameters.

[0037] Detection threshold calculation module: used to determine the detection threshold based on the preset value of the false alarm probability;

[0038] Target decision module: used to compare the detection statistics with the detection threshold and determine whether the target exists.

[0039] The technical effects achieved by this invention are as follows:

[0040] In this invention, a detection statistic is constructed. It integrates interference suppression, signal accumulation, and target detection;

[0041] In this invention, the sampling covariance matrix is ​​utilized. The detector proposed in this invention is made to have constant false alarm rate characteristics by performing quasi-whitening processing on the detection data matrix, target signal and interference signal matrix.

[0042] This invention can solve the problem of adaptive target detection in partially uniform environments and improves radar target detection performance compared with existing detection methods based on the GLRT criterion. Attached Figure Description

[0043] Figure 1 This is a flowchart illustrating an adaptive subspace detection method applicable to partially uniform environments according to the present invention.

[0044] Figure 2 This is a structural framework diagram of an adaptive subspace detection system applicable to partially uniform environments according to the present invention;

[0045] Figure 3 This is a schematic diagram of the detection probability of the extended target under different signal-to-noise ratios in the target detection results of the method proposed in this invention;

[0046] Figure 4 This is a schematic diagram of the detection probability under different degrees of non-uniformity in the target detection results of the method proposed in this invention. Detailed Implementation

[0047] 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.

[0048] Example 1:

[0049] Taking an airborne radar linear array as the research object, it is assumed that its number of array elements is... The number of pulses is The corresponding system dimension Sample to be tested Occupy a The echo cells are independently and identically distributed, therefore The dimension is .exist In the hypothesis, the sample to be tested Includes target signal components Interference signal components and noise components . signal component and They can be represented as:

[0050] (1)

[0051] and (2)

[0052] in, and They are respectively dimensional target signal subspace and The interference signal subspace is a dimensional matrix, and both are column-full rank unitary matrices. and These are the corresponding unknown subspace coordinates, with corresponding dimensions of... and Noise components Each column Satisfying the mean is The covariance is The complex Gaussian distribution, i.e. .

[0053] exist In the hypothesis, the sample to be tested Includes only interference signal components and noise components In summary, the binary hypothesis detection problem for extended target detection can be expressed as:

[0054] (3)

[0055] in, For dimension The training sample set, Equal to the total number of training samples; These are the training samples. The noise components in each column All satisfy the mean is The covariance is The complex Gaussian distribution, i.e. In a partially homogeneous environment, the noise covariance matrix of the sample to be tested... and the noise covariance matrix in the training samples The relationship between the two is as follows:

[0056] (4)

[0057] in, For non-uniform parameters of the environment, in a uniform environment

[0058] like Figure 1 As shown, an adaptive subspace detection method suitable for partially uniform environments includes the following steps:

[0059] Step 1: Construct the target signal subspace matrix and the interference signal subspace matrix, respectively using matrix... and The dimensions of the matrices are respectively and ,in , Indicates the number of system channels;

[0060] In step 1, the target signal matrix is ​​constructed. and interference signal matrix All are full-rank unitary matrices, and unitary matrices and Linear independence;

[0061] Step 2: Construct the sampling covariance matrix using training samples ,in Represents training samples, symbol Indicates conjugate transpose;

[0062] Step 3: Utilize the sampling covariance matrix Construct a quasi-whitening matrix, and then apply it to the target signal subspace matrix. Interference signal subspace matrix and the data matrix to be detected Quasi-whitening processing was performed to obtain whitening matrices. , and The data matrix to be detected The dimension is , Indicates extended dimensions;

[0063] In step 3, the sampling covariance matrix is ​​used. The constructed quasi-whitening matrix is ​​compared with the target signal subspace matrix. Interference signal subspace matrix and the data matrix to be detected The specific steps for performing a near-whitening process are as follows:

[0064] (5)

[0065] Step 4: Using the quasi-whitening matrix Construct the orthogonal projection matrix of the target signal and the corresponding complementary projection matrix Simultaneously utilizing the quasi-whitening matrix Construct the orthogonal projection matrix of the interference signal and the corresponding orthogonal projection matrix ;

[0066] In step 4, the quasi-whitened target signal matrix is ​​used. The constructed orthogonal projection matrix and the complement projection matrix are respectively expressed as: and ,in Represents the inverse of a matrix. For dimension is The identity matrix; utilizing the quasi-whitened interference signal matrix The constructed orthogonal projection matrix and the complement projection matrix are expressed as follows:

[0067] (6)

[0068] Step 5: Construct intermediate variables and non-zero eigenvalues ​​using the orthogonal projection matrix of the quasi-whitened data matrix to be detected and the interference signal, and obtain non-uniform parameters by combining the system dimension. ;

[0069] In step 5, the intermediate variable constructed using the orthogonal projection matrix of the quasi-whitened data matrix to be detected and the interference signal is... Obtain unknown parameters The solution to the following formula:

[0070] (7)

[0071] in, The number of training samples, for The One non-zero eigenvalue, This represents the total number of non-zero eigenvalues.

[0072] Step 6: Construct detection statistics using the quasi-whitened data matrix to be detected, the orthogonal projection matrix of the target signal, the orthogonal projection matrix of the interference signal, and the non-uniform parameters;

[0073] In step 6, the detection statistics constructed using the quasi-whitened data matrix to be detected, the orthogonal projection matrix of the target signal, the orthogonal projection matrix of the interference signal, and the non-uniform parameters are expressed as follows:

[0074] (8)

[0075] in, For dimension is The identity matrix, Represents the trace of a matrix;

[0076] Step 7: Determine the detection threshold using the preset false alarm probability value;

[0077] In step 7, the detection threshold determined using the preset false alarm probability value is:

[0078] (9)

[0079] in, Indicates the detection threshold. , For the number of Monte Carlo simulations, This represents the false alarm probability value set by the system. This indicates the rounding operation. For sequence

[0080] (10)

[0081] When arranged from largest to smallest The maximum value, for The conjugate transpose of the matrix. , This represents the first data point of the test data containing only interference and thermal noise components. Secondary implementation ;

[0082] Step 8: Compare the detection statistic with the detection threshold to determine whether the target exists.

[0083] In step 8, the detection statistic is compared with the detection threshold to determine whether the target exists. The determination is made in the following two cases:

[0084] If the detection statistic Greater than or equal to the detection threshold If so, then the target is determined to exist;

[0085] If the detection statistic Less than the detection threshold If the target does not exist, then it is determined that the target does not exist.

[0086] In this invention, a detection statistic is constructed. It integrates interference suppression, signal accumulation, and target detection;

[0087] In this invention, the sampling covariance matrix is ​​utilized. The detector proposed in this invention is made to have constant false alarm rate characteristics by performing quasi-whitening processing on the detection data matrix, target signal and interference signal matrix.

[0088] This invention can solve the problem of adaptive target detection in partially uniform environments and improves radar target detection performance compared with existing detection methods based on the GLRT criterion.

[0089] Example 2:

[0090] like Figure 2 As shown: An adaptive subspace detection system suitable for partially homogeneous environments, comprising:

[0091] Target and Interference Data Matrix Construction Module: Used to construct the target signal subspace matrix and the interference signal subspace matrix;

[0092] Sampling covariance matrix construction module: Used to construct the sampling covariance matrix from training samples;

[0093] Data quasi-whitening module: used to perform quasi-whitening processing on the target signal subspace matrix, interference signal subspace matrix and the data matrix to be detected, to obtain the quasi-whitened target signal subspace matrix, the quasi-whitened interference signal subspace matrix and the quasi-whitened data matrix to be detected, respectively;

[0094] Orthogonal projection matrix construction module: used to construct the orthogonal projection matrix of the target signal and the orthogonal projection matrix of the interference signal respectively using the quasi-whitened target signal subspace matrix and the quasi-whitened interference signal subspace matrix;

[0095] Non-uniform parameter solving module: used to perform eigenvalue decomposition on intermediate variables constructed by the orthogonal projection matrix of the data matrix to be detected and the interference signal, and then obtain the non-uniform parameters of the environment by combining the system dimension;

[0096] The detection statistics construction module is used to construct detection statistics from the quasi-whitened target signal matrix, the orthogonal projection matrix of the interference signal, the data matrix to be detected, and environmental non-uniform parameters.

[0097] Detection threshold calculation module: used to determine the detection threshold based on the preset value of the false alarm probability;

[0098] Target decision module: used to compare the detection statistics with the detection threshold and determine whether the target exists.

[0099] The effects of the present invention will be further explained below with reference to simulation experiments;

[0100] Simulation experiment;

[0101] Assume the radar system has the following number of channels: Let the false alarm probability be Covariance matrix of clutter and noise The The element at position 1 was set to , where the symbol This indicates taking the absolute value. Figure 3 The detection probabilities of the target under different signal-to-noise ratios (SNR) are given by the method proposed in this invention, wherein the subspace dimensions of the target signal and the interference signal are respectively... and , Number of training samples Noise-to-interference ratio INR and SNR are respectively and

[0102] from Figure 3 As can be seen, when the signal-to-noise ratio is higher than 23 dB, the proposed method achieves a detection probability of over 80% for the extended target. Compared with existing GLRT criterion detectors, the proposed detection method exhibits better detection performance. Furthermore, Figure 4 The detection probabilities of the proposed method for targets with different degrees of non-uniformity are given, wherein... , Signal-to-noise ratio Noise-to-interference ratio The results show that the detector performance decreases as the degree of non-uniformity increases, but the Durbin criterion detector proposed in this invention still has a higher detection probability than the traditional GLRT detector.

[0103] In this invention, a detection statistic is constructed. It integrates interference suppression, signal accumulation, and target detection;

[0104] In this invention, the sampling covariance matrix is ​​utilized. The detector proposed in this invention is made to have constant false alarm rate characteristics by performing quasi-whitening processing on the detection data matrix, target signal and interference signal matrix.

[0105] This invention can solve the problem of adaptive target detection in partially uniform environments and improves radar target detection performance compared with existing detection methods based on the GLRT criterion.

[0106] 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. An adaptive subspace detection method applicable to partially homogeneous environments, characterized in that: Includes the following steps: Step 1: Construct the target signal subspace matrix and the interference signal subspace matrix, respectively using matrix... and The dimensions of the matrices are respectively and ,in , Indicates the number of system channels; Step 2: Construct the sampling covariance matrix using training samples ,in Represents training samples, symbol Indicates conjugate transpose; Step 3: Utilize the sampling covariance matrix Construct a quasi-whitening matrix, and then apply it to the target signal subspace matrix. Interference signal subspace matrix and the data matrix to be detected Quasi-whitening processing was performed to obtain whitening matrices. , and The data matrix to be detected The dimension is , Indicates extended dimensions; Step 4: Using the whitening matrix Construct the orthogonal projection matrix of the target signal Simultaneously utilizing the whitening matrix Construct the orthogonal projection matrix of the interference signal and the corresponding complementary projection matrix ; Step 5: Utilize the whitening matrix The intermediate variables and non-zero eigenvalues ​​are constructed by using the complementary projection matrix of the interference signal, and the non-uniform parameters are obtained by combining the system dimension. ; In step 5, the intermediate variable constructed using the whitening matrix and the complementary projection matrix of the interference signal is... Obtaining non-uniform parameters The solution to the following formula: ;in, The number of training samples, for The One non-zero eigenvalue, This represents the total number of non-zero eigenvalues. Step 6: Utilize the whitening matrix The detection statistics, constructed from the orthogonal projection matrix of the target signal, the complement projection matrix of the interference signal, and the non-uniform parameters, are expressed as follows: in, For dimension is The identity matrix, Represents the trace of a matrix. For dimension is The identity matrix; Step 7: Determine the detection threshold using the preset false alarm probability value; Step 8: Compare the detection statistic with the detection threshold to determine whether the target exists.

2. The adaptive subspace detection method applicable to partially uniform environments according to claim 1, characterized in that: In step 1, the target signal matrix is ​​constructed. and interference signal matrix All are full-rank unitary matrices, and unitary matrices and Linearly independent.

3. The adaptive subspace detection method applicable to partially uniform environments according to claim 1, characterized in that: In step 3, the sampling covariance matrix is ​​used. The constructed quasi-whitening matrix is ​​compared with the target signal subspace matrix. Interference signal subspace matrix and the data matrix to be detected The specific steps for performing a near-whitening process are as follows: , and .

4. The adaptive subspace detection method applicable to partially uniform environments according to claim 1, characterized in that: In step 4, the whitened target signal matrix is ​​used. The constructed orthogonal projection matrix is ​​expressed as: ,in Representing the inverse of a matrix, using the whitening interference signal matrix The constructed orthogonal projection matrix and the complement projection matrix are expressed as follows: and .

5. The adaptive subspace detection method applicable to partially uniform environments according to claim 1, characterized in that: In step 7, the detection threshold determined using the preset false alarm probability value is: ;in, Indicates the detection threshold. , For the number of Monte Carlo simulations, This represents the false alarm probability value set by the system. This indicates the rounding operation. For sequence; Arranged from largest to smallest The maximum value, for The conjugate transpose of the matrix. , This represents the first data point of the test data containing only interference and thermal noise components. Secondary implementation .

6. The adaptive subspace detection method applicable to partially uniform environments according to claim 1, characterized in that: In step 8, the detection statistic is compared with the detection threshold to determine whether the target exists. The determination is made in the following two cases: If the detection statistic Greater than or equal to the detection threshold If so, then the target is determined to exist; If the detection statistic Less than the detection threshold If the target does not exist, then it is determined that the target does not exist.

7. An adaptive subspace detection system suitable for partially uniform environments, characterized in that: include: Target and Interference Data Matrix Construction Module: Used in step 1 to construct the target signal subspace matrix and the interference signal subspace matrix; Using matrices respectively and The dimensions of the matrices are respectively and ,in , Indicates the number of system channels; The sampling covariance matrix construction module is used in step 2 to construct the sampling covariance matrix using training samples. ,in Represents training samples, symbol Indicates conjugate transpose; Data whitening module: used in step 3 to utilize the sampling covariance matrix Construct a quasi-whitening matrix, and then apply it to the target signal subspace matrix. Interference signal subspace matrix and the data matrix to be detected Quasi-whitening processing was performed to obtain whitening matrices. , and The data matrix to be detected The dimension is , Indicates extended dimensions; Orthogonal projection matrix construction module: used in step 4 using the whitening matrix Construct the orthogonal projection matrix of the target signal Simultaneously utilizing the whitening matrix Construct the orthogonal projection matrix of the interference signal and the corresponding complementary projection matrix ; Non-uniform parameter solving module: used in step 5 to solve using the whitening matrix The intermediate variables and non-zero eigenvalues ​​are constructed by using the complementary projection matrix of the interference signal, and the non-uniform parameters are obtained by combining the system dimension. ; In step 5, the intermediate variable constructed using the whitening matrix and the complementary projection matrix of the interference signal is... Obtaining non-uniform parameters The solution to the following formula: ;in, The number of training samples, for The One non-zero eigenvalue, This represents the total number of non-zero eigenvalues. Detection statistic construction module: used in step 6 to utilize the whitening matrix The orthogonal projection matrix of the target signal, the complement projection matrix of the interference signal, and the detection statistics constructed from the non-uniform parameters are expressed as follows: in, For dimension is The identity matrix, Represents the trace of a matrix. For dimension is The identity matrix; Detection threshold calculation module: used in step 7 to determine the detection threshold using a preset false alarm probability value; Target decision module: used in step 8 to compare the detection statistic with the detection threshold and determine whether the target exists.

Images