A robust adaptive beamforming method based on subspace orthogonality

Abstract

The present invention relates to a robust adaptive beamforming method based on subspace orthogonality, firstly, estimating the arrival direction of the signal and the interference, generating the initial value of each steering vector; secondly, correcting the desired signal steering vector, estimating a more accurate desired signal steering vector; thirdly, correcting the interference steering vector, estimating a more accurate interference steering vector; fourthly, estimating the noise power by using the average value of the eigenvalue corresponding to the noise subspace; fifthly, estimating each interference power together with the signal subspace and its eigenvalue matrix; sixthly, reconstructing the interference plus noise covariance matrix; seventhly, calculating the optimal weight vector, forming a robust adaptive beam output for array received data. The present invention can obtain a more accurate interference plus noise covariance matrix and the steering vector of the desired signal, the optimal weight vector is more accurate, and the robustness of the adaptive beamformer is improved.

Description

Technical Field

[0001] The present invention relates to the research field of beamforming in the field of array signal processing, and in particular, in non-ideal situations where various errors may exist, the robustness of adaptive beamforming can be better improved by more accurately reconstructing a desired signal steering vector and an interference plus noise covariance matrix. Background Art

[0002] Among the existing robust adaptive beamforming methods, the more representative methods are: linear constrained minimum variance method, diagonal loading method, characteristic subspace method and uncertainty set method. However, considering the uncertainty of parameter selection and the constraints of the inherent properties of the method itself, the performance of these beamforming methods will be significantly attenuated in the presence of array errors and cannot achieve ideal results.

[0003] In recent years, an adaptive beamforming method based on the reconstruction of the interference plus noise covariance matrix has been studied, which is more robust. This method mainly uses the Capon spatial power spectrum and uses its integral in the non-desired signal angle region as the estimated value of the interference plus noise covariance matrix, effectively eliminating the desired signal component. However, this reconstruction method only directly uses the Capon spatial power spectrum to integrate the angle variable in the non-desired signal angle region, and the final reconstructed interference plus noise covariance matrix is ​​not accurate enough, resulting in this method only having a certain robustness to the direction of arrival error. When there are other types of steering vector errors, the performance of the method cannot be guaranteed. Subsequently, a method for reconstructing the interference plus noise covariance matrix for any type of array error was proposed. A robust adaptive beamforming method (ZL 2017 1 1417222.3) improves the reconstruction accuracy of the interference plus noise covariance matrix, and a multiple robust adaptive beamforming method (ZL 2020 10360284.0) introduces multiple robust technologies to further obtain a more accurate interference plus noise covariance matrix and the steering vector of the desired signal, thereby improving the robustness of the adaptive beamformer. However, most methods reconstruct the interference plus noise covariance matrix based on the Capon spatial power spectrum. Due to the limitations of the Capon spatial power spectrum estimation method, there are still certain errors in the reconstruction process.

[0004] In view of the above analysis, it is necessary to study new robust methods to improve the robustness of the beamformer. Summary of the invention

[0005] The technology of the present invention solves the problem: overcomes the shortcomings of the prior art, makes full use of the orthogonal properties of the signal subspace and the noise subspace and the information contained in their corresponding eigenvalue matrices, and provides a robust adaptive beamforming method based on subspace orthogonality. By more accurately reconstructing the desired signal steering vector and the interference plus noise covariance matrix, the robustness of the beamformer to any type of array error is greatly improved.

[0006] The objective of the present invention is achieved through the following technical solutions:

[0007] The present invention provides a robust adaptive beamforming method based on subspace orthogonality, which first pre-processes array received data, including: estimating the array covariance matrix, performing eigendecomposition on the matrix, and obtaining the signal subspace and noise subspace and their corresponding eigenvalue matrices. It includes the following steps:

[0008] Step 1: Based on the noise subspace obtained by preprocessing, the nominal steering vector is used in the multiple signal classification method to estimate the arrival direction of the signal and the interference, and generate the initial value of each steering vector;

[0009] Step 2: Using the orthogonal property of the signal subspace and the noise subspace, the desired signal steering vector is corrected to estimate a more accurate desired signal steering vector;

[0010] Step 3: Using the orthogonal property of the signal subspace and the noise subspace, the interference steering vector is corrected to estimate a more accurate interference steering vector;

[0011] Step 4: Estimate the noise power using the average value of the eigenvalues ​​corresponding to the noise subspace;

[0012] Step 5: using the estimated more accurate steering vector of each interference, together with the signal subspace and its eigenvalue matrix, estimate each interference power;

[0013] Step 6: reconstruct the noise covariance matrix according to the estimated noise power, reconstruct the interference covariance matrix according to the estimated more accurate steering vectors and powers of each interference, and reconstruct the interference plus noise covariance matrix;

[0014] Step 7: Calculate the optimal weight vector based on the reconstructed interference plus noise covariance matrix and the more accurate desired signal steering vector to form a robust adaptive beam output for the array receiving data.

[0015] Furthermore, the robust adaptive beamforming method based on subspace orthogonality first pre-processes the M-element array received data, including: estimating the array covariance matrix Perform eigendecomposition on the matrix, E s and Λs are the diagonal matrices composed of the signal (including interference) subspace and its corresponding eigenvalues, E n and Λ n are the diagonal matrices consisting of the noise subspace and its corresponding eigenvalues, E s and E n are orthogonal. The subscripts s and n are symbols for distinguishing signal from noise, respectively, and the superscript H is the conjugate transpose operator of the matrix.

[0016] The step 1 comprises the following steps:

[0017] Step 11: Generate a nominal steering vector with a corresponding direction angle of θ according to the array structure || ||2 is the vector l2 norm operator;

[0018] Step 12: Based on the noise subspace E obtained by preprocessing n , the guide vector For the multiple signal classification method (MUSIC), the MUSIC spatial spectrum is calculated as follows:

[0019]

[0020] First estimate the direction of arrival of the desired signal The number of interferences L (M>L+1) and the direction of arrival of all interferences Then generate the initial value of the steering vector of the desired signal according to the array structure Initial value of each interference steering vector

[0021] The step 2 comprises the following steps:

[0022] Step 21: Give a steering vector perpendicular to the desired signal Tiny neighborhoods on the plane, including two-dimensional rectangular neighborhoods, square neighborhoods, elliptical neighborhoods, circular neighborhoods, cross neighborhoods, and one-dimensional neighborhoods. is the center of the neighborhood, and discretizes R points in the neighborhood. The steering vector of the rth point is Vector The error vector introduced at the rth point; the tiny neighborhood is

[0023] Step 22: Initial value of the desired signal steering vector R discrete points in the tiny neighborhood of The MUSIC spatial spectrum is calculated as follows:

[0024]

[0025] According to the orthogonal property of the signal subspace and the noise subspace, by finding R discrete points in a small neighborhood The maximum value of the desired signal is obtained, and finally the corrected steering vector of the desired signal is obtained as:

[0026]

[0027] If there are multiple maximum values, take their vector average and record it as

[0028] The step 3 comprises the following steps:

[0029] Step 31: Give the initial value of the steering vector perpendicular to each disturbance Tiny neighborhoods on the plane, including two-dimensional rectangular neighborhoods, square neighborhoods, elliptical neighborhoods, circular neighborhoods, cross neighborhoods, and one-dimensional neighborhoods. is the center of the neighborhood, and discretizes R points in the neighborhood. The steering vector of the rth point is Vector The error vector introduced at the rth point; the tiny neighborhood is

[0030] Step 32: For each interfering steering vector R discrete points in the small-dimensional neighborhood of The MUSIC spatial spectrum is calculated as follows:

[0031]

[0032] By using the orthogonal property of the signal subspace and the noise subspace, we can find The maximum value of the interference is finally obtained, which is:

[0033]

[0034] If there are multiple maximum values, take their vector average and record it as

[0035] The step 4 comprises:

[0036] Using the eigenvalue matrix Λ corresponding to the noise subspace n , take the average value of the diagonal elements to estimate the noise power, that is:

[0037]

[0038] Among them, tr{} is the trace operator of the matrix.

[0039] The step 5 comprises:

[0040] Step 51: Subtract the noise from the signal subspace eigenvalue matrix Λ s The influence of , the modified signal subspace eigenvalue matrix is:

[0041]

[0042] Among them, I L+1 is the L+1-dimensional identity matrix;

[0043] Step 52: Using the estimated interference steering vectors Signal subspace E s And its modified eigenvalue matrix The interference power is estimated as follows:

[0044]

[0045] Of course, we can also directly use the eigenvalue matrix Λ s , the interference power is estimated as follows:

[0046]

[0047] Just a slight reduction in accuracy.

[0048] The step 6 comprises the following steps:

[0049] Step 61: Based on the estimated noise power Reconstruct the noise covariance matrix:

[0050]

[0051] Among them, I M is the M-dimensional identity matrix;

[0052] Step 62: Based on the estimated interference steering vectors and power The interference covariance matrix is ​​reconstructed as follows:

[0053]

[0054] Step 63: Based on the reconstructed noise covariance matrix and the interference covariance matrix The interference plus noise covariance matrix is ​​reconstructed as follows:

[0055]

[0056] The step 7 comprises the following steps:

[0057] Step 71: Reconstruct the interference plus noise covariance matrix and a more accurate desired signal steering vector The optimal weight vector is calculated as follows:

[0058]

[0059] Step 72: Weight the kth snapshot data x(k) received by the array with the optimal weight vector w to obtain the output signal y(k) of the beamformer = w H x(k), achieving robust adaptive beamforming.

[0060] The advantages of the present invention compared with the prior art are: it can be seen from the technical solution provided by the present invention that the characteristic decomposition of the covariance matrix of the array received data is performed to obtain the signal subspace and the noise subspace, and the orthogonal properties of the signal subspace and the noise subspace are used to estimate more accurate expected signals and interference steering vectors, and the information contained in the signal subspace eigenvalue matrix and the noise subspace eigenvalue matrix is ​​used to estimate more accurate noise power and interference power, and reconstruct a more accurate interference plus noise covariance matrix, so that the optimal weight vector is more accurate and has better adaptability to any type of array error, thereby significantly improving the robustness of the adaptive beamformer. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0062] Figure 1 A flowchart of a robust adaptive beamforming method based on subspace orthogonality provided by an embodiment of the present invention;

[0063] Figure 2 A schematic diagram of an array signal receiving model provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0064] The embodiment of the present invention performs eigendecomposition of the covariance matrix of the array received data to obtain a signal subspace and a noise subspace, and uses the orthogonal property of the signal subspace and the noise subspace to estimate a more accurate desired signal and interference steering vector, and uses the information contained in the signal subspace eigenvalue matrix and the noise subspace eigenvalue matrix to estimate a more accurate noise power and interference power, reconstruct a more accurate interference plus noise covariance matrix, obtain a more accurate optimal weight vector, and further improve the robustness of the adaptive beamforming method under various array error conditions.

[0065] like Figure 1 As shown, the method first preprocesses the array received data, including: estimating the array covariance matrix, performing eigendecomposition on the matrix, and obtaining the signal subspace and noise subspace and their corresponding eigenvalue matrices.

[0066] The main steps are as follows:

[0067] Step 1: Based on the noise subspace obtained by preprocessing, the nominal steering vector is used in the multiple signal classification method to estimate the arrival direction of the signal and the interference, and generate the initial value of each steering vector;

[0068] Step 2: Using the orthogonal property of the signal subspace and the noise subspace, the desired signal steering vector is corrected to estimate a more accurate desired signal steering vector;

[0069] Step 3: Using the orthogonal property of the signal subspace and the noise subspace, the interference steering vector is corrected to estimate a more accurate interference steering vector;

[0070] Step 4: Estimate the noise power using the average value of the eigenvalues ​​corresponding to the noise subspace;

[0071] Step 5: using the estimated more accurate steering vector of each interference, together with the signal subspace and its eigenvalue matrix, estimate each interference power;

[0072] Step 6: reconstruct the noise covariance matrix according to the estimated noise power, reconstruct the interference covariance matrix according to the estimated more accurate steering vectors and powers of each interference, and reconstruct the interference plus noise covariance matrix;

[0073] Step 7: Calculate the optimal weight vector based on the reconstructed interference plus noise covariance matrix and the more accurate desired signal steering vector to form a robust adaptive beam output for the array receiving data.

[0074] Compared with the existing interference plus noise covariance matrix reconstruction type robust adaptive beamforming method, the above-mentioned scheme of the present invention, after the covariance matrix of the array received data is characteristically decomposed to obtain the signal subspace and the noise subspace, uses the orthogonal property of the signal subspace and the noise subspace to estimate more accurate expected signals and interference steering vectors, uses the information contained in the signal subspace eigenvalue matrix and the noise subspace eigenvalue matrix to estimate more accurate noise power and interference power, reconstructs a more accurate interference plus noise covariance matrix, and the optimal weight vector is also more accurate, thereby being able to better improve the robustness of the beamformer to any type of array errors.

[0075] To facilitate understanding, the Capon beamforming method is first introduced, followed by preprocessing, and then the above seven steps are described in detail.

[0076] The examples of the present invention are applicable to any type of array, including linear array, circular array, conformal array, etc., and the applicable wave arrival directions include one-dimensional azimuth, one-dimensional elevation, two-dimensional azimuth and elevation. For the convenience of calculation, only the linear array is discussed here, and the specific array signal model is as follows:

[0077] Consider an array with M elements, receiving narrowband far-field signals from space, then the received data of the array at observation time k (called the kth snapshot data received by the array) is expressed as:

[0078] x(k)=x s (k)+x i (k)+x n (k);

[0079] Among them, x s (k), x i (k) and x n (k) represent the desired signal, interference and noise respectively, and are statistically independent of each other; x s (k) = s(k)a0, s(k) is the waveform of the desired signal, a0 is the true steering vector of the desired signal, represents the interference vector, L is the number of interferences, s l (k) is the waveform of the lth interference, a l is the corresponding real interference steering vector, x n (k) is additive independent and identically distributed Gaussian white noise, signal s(k), each interference s l (k) and noise x n (k) all have zero mean. Figure 2 A schematic diagram of a narrowband far-field signal source in the linear array receiving space is given, where the direction of arrival of the signal (or interference) source is θ, and it is approximately considered to be incident on each array element in the form of a plane wave, d1, d2, …, d M-1 is the distance between each array element and the reference array element.

[0080] In order to enhance the signal in a specific direction, which is equivalent to increasing the gain in that direction, it is necessary to assign specific weighting coefficients to each array element. How to design the optimal weight w = [w1, w2, …, w M ] T This is the main work content of beamforming technology. The array system that can achieve this goal is often called a beamformer. In essence, it is a spatial filter. The output of the beamformer is the weighted sum of the received signals of each array element, that is:

[0081] y(k)=w H x(k);

[0082] For a beamformer with a given weight vector, in order to evaluate its beamforming performance, in addition to using the array pattern to intuitively display it, the output signal-to-interference-noise ratio is often used as a quantitative indicator to measure the overall performance of the beamformer. It is defined as follows:

[0083]

[0084] Among them, x i+n (k) = x i (k)+x n (k) is the interference plus noise component, is the interference plus noise covariance matrix. is the power of the desired signal.

[0085] In order to maximize the output signal-to-noise ratio, Capon et al. proposed to minimize the array output power while ensuring a certain response to the desired signal direction, forming the following optimization problem:

[0086]

[0087] Where R = E{x(k)x H (k)} is the covariance matrix of the array receiving data. Therefore, the weight vector of the beamformer can be obtained as:

[0088]

[0089] This is the Capon beamforming method, which can achieve the maximum output signal-to-interference-noise ratio under ideal conditions. Substituting the obtained weight vector into the objective function of the optimization problem, the output power of the array can be obtained as:

[0090]

[0091] This power is also called Capon power. It is the minimum power output by the array when the desired signal is received without distortion. Therefore, Capon power is regarded as an estimated value of the power of the desired signal. When the steering vector is regarded as a variable, the output power expression is used to represent the signal power in each steering vector direction, which is the spatial power spectrum.

[0092] In actual situations, ideal signal statistics are difficult to obtain, and are usually achieved using the sample matrix inversion algorithm. The main idea is to use the sample covariance matrix Instead of the ideal covariance matrix R, we can solve the weight vector, which is defined as:

[0093]

[0094] Where K is the number of snapshots of data received by the array. Since limited snapshots will introduce random errors to R, and considering that the true steering vector is difficult to obtain accurately, it is necessary to use the steering vector obtained according to the known array structure for calculation, and the corresponding Capon space power spectrum can be expressed as:

[0095]

[0096] in, That is, it is the nominal steering vector corresponding to the direction angle θ assumed according to the array structure.

[0097] The performance of the Capon beamforming method will degrade significantly when there are various errors in the array.

[0098] The purpose of the present invention is to estimate more accurate desired signal and interference steering vector by using the orthogonal property of signal subspace and noise subspace, estimate more accurate noise power and interference power by using the information contained in the signal subspace eigenvalue matrix and the noise subspace eigenvalue matrix, estimate more accurate desired signal steering vector and noise plus interference covariance matrix, obtain more accurate weight vector, further improve the performance of beamforming method, and have robustness. It is implemented in the following seven steps after preprocessing.

[0099] Preprocessing:

[0100] Estimate the covariance matrix of the array using the M-element array received data right Perform feature decomposition, E s and Λ s are the diagonal matrices composed of the signal (including interference) subspace and its corresponding eigenvalues, E n and Λ n are the diagonal matrices consisting of the noise subspace and its corresponding eigenvalues, E s and E n are orthogonal. Where K is the number of snapshots of data received by the array, x(k) is the kth snapshot data received by the array, the subscripts s and n are symbols for distinguishing signals from noise, and the superscript H is the conjugate transpose operator of the matrix.

[0101] Step 1:

[0102] Step 11: Generate a nominal steering vector with a corresponding direction angle of θ according to the array structure ||||2 is the vector l2 norm operator;

[0103] Step 12: Based on the noise subspace E obtained by preprocessing n , the guide vector For the multiple signal classification method (MUSIC), the MUSIC spatial spectrum is calculated as follows:

[0104]

[0105] First estimate the direction of arrival of the desired signal The number of interferences L (M>L+1) and the direction of arrival of all interferences Then generate the initial value of the steering vector of the desired signal according to the array structure Initial value of each interference steering vector

[0106] Step 2:

[0107] Step 21: Give a steering vector perpendicular to the desired signal Tiny neighborhoods on the plane, including two-dimensional rectangular neighborhoods, square neighborhoods, elliptical neighborhoods, circular neighborhoods, cross neighborhoods, and one-dimensional neighborhoods. is the center of the neighborhood, and discretizes R points in the neighborhood. The steering vector of the rth point is Vector The error vector introduced at the rth point; the tiny neighborhood is

[0108] Step 22: Initial value of the desired signal steering vector R discrete points in the tiny neighborhood of The MUSIC spatial spectrum is calculated as follows:

[0109]

[0110] According to the orthogonal property of the signal subspace and the noise subspace, by finding R discrete points in a small neighborhood The maximum value of , and finally the steering vector of the corrected desired signal is obtained as:

[0111]

[0112] If there are multiple maximum values, take their vector average and record it as

[0113] Step 3:

[0114] Step 31: Give the initial value of the steering vector perpendicular to each disturbance Tiny neighborhoods on the plane, including two-dimensional rectangular neighborhoods, square neighborhoods, elliptical neighborhoods, circular neighborhoods, cross neighborhoods, and one-dimensional neighborhoods. is the center of the neighborhood, and discretizes R points in the neighborhood. The steering vector of the rth point is Vector The error vector introduced at the rth point; the tiny neighborhood is

[0115] Step 32: For each interfering steering vector R discrete points in the tiny neighborhood of The MUSIC spatial spectrum is calculated as follows:

[0116]

[0117] By using the orthogonal property of the signal subspace and the noise subspace, we can find The maximum value of the interference is finally obtained as follows:

[0118]

[0119] If there are multiple maximum values, take their vector average and record it as

[0120] Step 4:

[0121] Using the eigenvalue matrix Λ corresponding to the noise subspace n , take the average value of the diagonal elements to estimate the noise power, that is:

[0122]

[0123] Among them, tr{} is the trace operator of the matrix.

[0124] Step 5:

[0125] Step 51: Subtract the noise from the signal subspace eigenvalue matrix Λ s The influence of , the modified signal subspace eigenvalue matrix is:

[0126]

[0127] Among them, I L+1 is the L+1-dimensional identity matrix;

[0128] Step 52: Using the estimated interference steering vectors Signal subspace E s And its modified eigenvalue matrix The interference power is estimated as follows:

[0129]

[0130] Of course, we can also directly use the eigenvalue matrix Λ s , the interference power is estimated as follows:

[0131]

[0132] Just a slight reduction in accuracy.

[0133] Step 6:

[0134] Step 61: Based on the estimated noise power The noise covariance matrix is ​​reconstructed as follows:

[0135]

[0136] Among them, I M is the M-dimensional identity matrix in the M-element array;

[0137] Step 62: Based on the estimated interference steering vectors and power The interference covariance matrix is ​​reconstructed as follows:

[0138]

[0139] Step 63: Based on the reconstructed noise covariance matrix and the interference covariance matrix The interference plus noise covariance matrix is ​​reconstructed as follows:

[0140]

[0141] Step 7:

[0142] Step 71: Reconstruct the interference plus noise covariance matrix and a more accurate desired signal steering vector The optimal weight vector is calculated as follows:

[0143]

[0144] Step 72: Weight the kth snapshot data x(k) received by the array with the optimal weight vector w to obtain the output signal y(k) of the beamformer = w H x(k), achieving robust adaptive beamforming.

[0145] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above embodiments can be implemented by software, or by means of software plus necessary general hardware platforms. Based on such understanding, the technical solutions of the above embodiments can be embodied in the form of software products, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments of the present invention.

[0146] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A robust adaptive beamforming method based on subspace orthogonality first pre-processes the received data of the M-element array and estimates the covariance matrix of the array The matrix is ​​subjected to eigendecomposition to obtain the signal subspace and noise subspace and their corresponding eigenvalue matrices, where E s and Λ s are the diagonal matrices composed of the signal subspace and its corresponding eigenvalues, E n and Λ n are the diagonal matrices consisting of the noise subspace and its corresponding eigenvalues, E s and E n are orthogonal, where The subscripts s and n are symbols for distinguishing signals from noises, respectively, and the superscript H is a conjugate transpose operator of a matrix, which is characterized in that the method further includes the following steps: Step 1: Based on the noise subspace obtained by preprocessing, the nominal steering vector is used in a multiple signal classification method to estimate the direction of arrival of the desired signal and the interference, and the initial value of the desired signal steering vector and the initial value of each interference steering vector are generated; Step 2: Using the orthogonal property of the signal subspace and the noise subspace, the initial value of the desired signal steering vector in step 1 is corrected to estimate a more accurate desired signal steering vector; Step 3: using the orthogonal property of the signal subspace and the noise subspace, the initial values ​​of the interference steering vectors in step 1 are corrected to estimate more accurate interference steering vectors; Step 4, estimating the noise power by using the average value of the eigenvalues ​​corresponding to the noise subspace obtained by preprocessing; Step 5: using the more accurate interference steering vectors estimated in step 3, together with the signal subspace and its eigenvalue matrix, to estimate the interference powers; Step 6, reconstruct the noise covariance matrix according to the noise power estimated in step 4; reconstruct the interference covariance matrix according to the interference steering vectors estimated in step 3 and the interference powers in step 5; finally, reconstruct the interference plus noise covariance matrix according to the reconstructed noise covariance matrix and the interference covariance matrix; Step 7: Calculate the optimal weight vector based on the interference plus noise covariance matrix reconstructed in step 6 and the more accurate desired signal steering vector in step 2 to form a robust adaptive beam output for array received data.

2. The robust adaptive beamforming method based on subspace orthogonality according to claim 1, characterized in that: The step 1 comprises the following steps: Step 11: Generate a nominal steering vector with a corresponding direction angle of θ according to the array structure || ||2 is the vector l2 norm operator; Step 12: Based on the noise subspace E obtained by preprocessing n , the nominal steering vector In the multiple signal classification method MUSIC, the MUSIC spatial spectrum is calculated as follows: First estimate the direction of arrival of the desired signal The number of interferences L and the direction of arrival of all interferences M>L+1; then generate the initial value of the steering vector of the desired signal according to the array structure Initial value of each interference steering vector 3. The robust adaptive beamforming method based on subspace orthogonality according to claim 2, characterized in that: The step 2 comprises the following steps: Step 21: Give the initial value of the steering vector perpendicular to the desired signal A tiny neighborhood on a plane, including a two-dimensional rectangular neighborhood, a square neighborhood, an elliptical neighborhood, a circular neighborhood, a cross neighborhood, or a one-dimensional neighborhood. is the center of the neighborhood, and discretizes R points in the neighborhood. The steering vector of the rth point is Vector The error vector introduced at the rth point; the tiny neighborhood is Step 22: Initial value of the desired signal steering vector R discrete points in the tiny neighborhood of The MUSIC spatial spectrum is calculated as follows: According to the orthogonal property of the signal subspace and the noise subspace, by finding R discrete points in a small neighborhood The maximum value of the desired signal is obtained, and finally the corrected steering vector of the desired signal is obtained as: If there are multiple maximum values, take their vector average and record it as 4. The robust adaptive beamforming method based on subspace orthogonality according to claim 3, characterized in that: The step 3 comprises the following steps: Step 31: Give the initial value of the steering vector perpendicular to each disturbance A tiny neighborhood on a plane, is the center of the neighborhood, and discretizes R points in the neighborhood. The steering vector of the rth point is Vector The error vector introduced at the rth point; the tiny neighborhood is Step 32: For each interfering steering vector R discrete points in the tiny neighborhood of The MUSIC spatial spectrum is calculated as follows: By using the orthogonal property of the signal subspace and the noise subspace, we can find The maximum value of the interference is finally obtained, which is: If there are multiple maximum values, take their vector average and record it as 5. The robust adaptive beamforming method based on subspace orthogonality according to claim 4, characterized in that: The step 4 comprises: Using the eigenvalue matrix Λ corresponding to the noise subspace n , take the average value of the diagonal elements to estimate the noise power, that is: Among them, tr{} is the trace operator of the matrix.

6. The robust adaptive beamforming method based on subspace orthogonality according to claim 5, characterized in that: The step 5 comprises: Step 51: Subtract the noise from the signal subspace eigenvalue matrix Λ s The influence of , the modified signal subspace eigenvalue matrix is: Among them, I L+1 is the L+1-dimensional identity matrix; Step 52: Using the estimated interference steering vectors Signal subspace E s And its modified eigenvalue matrix The interference power is estimated as follows: Or directly use the eigenvalue matrix Λ s , the interference power is estimated as follows:

7. The robust adaptive beamforming method based on subspace orthogonality according to claim 6, characterized in that: The step 6 comprises the following steps: Step 61: Based on the estimated noise power Reconstruct the noise covariance matrix: Among them, I M is the M-dimensional identity matrix; Step 62: Based on the estimated interference steering vectors and power The interference covariance matrix is ​​reconstructed as follows: Step 63: Based on the reconstructed noise covariance matrix and the interference covariance matrix The interference plus noise covariance matrix is ​​reconstructed as follows:

8. The robust adaptive beamforming method based on subspace orthogonality according to claim 7, characterized in that: The step 8 comprises the following steps: Step 71: Reconstruct the interference plus noise covariance matrix and a more accurate desired signal steering vector The optimal weight vector is calculated as follows: Step 72: Weight the kth snapshot data x(k) received by the array with the optimal weight vector w to obtain the output signal y(k) of the beamformer = w H x(k), achieving robust adaptive beamforming.

Images