An adaptive signal beamforming method, device, apparatus and storage medium

Abstract

The application provides a self-adaptive signal beam forming method, device, equipment and storage medium. The method is based on a 2N+1 element uniform linear array receiving signal. The method comprises the following steps: a multiple Toeplitz matrix is constructed, and a covariance matrix is obtained by merging; a signal direction vector is calculated according to an eigenvalue and an eigenvector, and an expected signal direction vector is obtained by optimization; a noise power is estimated by using a minimum eigenvalue, a corresponding relationship between a signal power and the eigenvector is established, and an expected signal power is determined; a covariance matrix of mutually independent noise is reconstructed according to the noise power, the corresponding relationship and the expected signal power; and the expected signal direction vector and the reconstructed covariance matrix are brought into a Cap on beam former, and an optimal weight vector is calculated to control beam forming. The method improves the performance and reliability of the beam forming system in a complex signal environment by constructing a multiple Toeplitz matrix and using the characteristics of the multiple Toeplitz matrix for signal decorrelation and covariance matrix reconstruction.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of signal beam control, and particularly relates to an adaptive signal beam forming method, device, equipment and storage medium. BACKGROUND

[0002] In the field of signal processing, beam forming is an important technique that dynamically adjusts the direction of the beam according to the signal environment, thereby enhancing the desired signal and suppressing interference. However, traditional beam forming algorithms have difficulties in processing coherent signals, especially when the subarray length is insufficient to distinguish coherent signals or the signal-to-noise ratio (SNR) is low, and their performance will be significantly affected. The presence of coherent signals will enhance the correlation of array received signals, making it difficult for traditional algorithms to accurately distinguish and process multiple signal sources.

[0003] To overcome this difficulty, researchers have proposed various improved methods. Among them, the adaptive beam forming method based on Toeplitz matrix reconstruction is an effective solution. However, the existing beam forming method based on Toeplitz matrix still has certain limitations, such as the need for preprocessing of received array noise or insufficient processing accuracy in low SNR environments. SUMMARY

[0004] The purpose of the present application is to overcome the above-mentioned prior art, and provide an adaptive signal beam forming method, device, equipment and storage medium.

[0005] The present application provides an adaptive signal beam forming method, comprising:

[0006] Constructing a multiple Toeplitz matrix based on the signals received by a ULA array of 2N+1 array elements;

[0007] Combining the multiple Toeplitz matrix through a square sum operation to obtain a covariance matrix, the covariance matrix including eigenvalues and eigenvectors;

[0008] Calculating the steering vector of the signal according to the eigenvalues and eigenvectors;

[0009] Optimizing the steering vector using a preset optimization function to obtain a desired signal steering vector;

[0010] Estimating the noise power using the smallest eigenvalue of the covariance matrix;

[0011] Establishing a corresponding relationship between the signal power and the eigenvectors according to the correlation between the eigenvectors and the desired signal steering vector;

[0012] Determining the desired signal power according to the corresponding relationship;

[0013] reconstructing a covariance matrix of the mutually uncorrelated noises according to the noise power, the correspondence and the expected signal power;

[0014] inputting the expected signal steering vector and the covariance matrix of the mutually uncorrelated noises into a Capon beamformer to calculate an optimal weight vector;

[0015] controlling the beamforming of the signals according to the optimal weight vector.

[0016] Optionally, a multiple Toeplitz matrix is constructed based on the signals received by the ULA array of 2N+1 elements, comprising:

[0017] taking each element in the ULA array of 2N+1 elements as a reference element and taking the signal received by each element as a reference signal;

[0018] sequentially obtaining the received signals of all elements and repeatedly constructing multiple Toeplitz matrices under all reference elements.

[0019] Optionally, the multiple Toeplitz matrices are combined through a sum-of-squares operation to obtain a covariance matrix, and the expression is as follows:

[0020]

[0021] wherein, R m is a Toeplitz matrix, and the N is an element serial number.

[0022] Optionally, the steering vector of the signals is calculated according to the eigenvalue and the eigenvector, comprising:

[0023] the eigenvalue and the eigenvector of the covariance matrix are obtained through eigenvalue decomposition, and the direction of arrival (DOA) of the signals is estimated through a multiple signal classification (MUSIC) algorithm to reconstruct the steering vector.

[0024] Optionally, the expected signal power is determined according to the correspondence, comprising:

[0025] after all steering vectors are corresponded to signal powers according to the correspondence, the expected signal power is placed in the first column, and the expected signal power is obtained by calculating the non-zero elements in the first column of the diagonal matrix.

[0026] Optionally, the covariance matrix of the mutually uncorrelated noises is reconstructed according to the noise power, the correspondence and the expected signal power, comprising:

[0027] the expected signal power is deleted to obtain a diagonal matrix without the expected signal power;

[0028] Inverse operation is performed on the diagonal matrix to obtain a reconstructed uncorrelated noise covariance matrix.

[0029] Optionally, the method is applied to control of a communication beam or a radar beam.

[0030] The application further provides a self-adaptive signal beam forming device, comprising:

[0031] a matrix module configured to construct multiple Toeplitz matrices based on signals received by a ULA array with 2N+1 array elements;

[0032] a merging module configured to merge the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix, wherein the covariance matrix comprises eigenvalues and eigenvectors;

[0033] a calculation module configured to calculate a steering vector of the signals based on the eigenvalues and the eigenvectors;

[0034] an optimization module configured to optimize the steering vector using a preset optimization function to obtain a desired signal steering vector;

[0035] an estimation module configured to estimate noise power using a minimum eigenvalue of the covariance matrix;

[0036] a relationship module configured to establish a corresponding relationship between signal power and the eigenvectors based on correlation between the eigenvectors and the desired signal steering vector;

[0037] a determination module configured to determine desired signal power based on the corresponding relationship;

[0038] a reconstruction module configured to reconstruct a covariance matrix of uncorrelated noise based on the noise power, the corresponding relationship and the desired signal power;

[0039] a weight vector module configured to input the desired signal steering vector and the covariance matrix of uncorrelated noise into a Capon beam former to calculate an optimal weight vector;

[0040] a control module configured to control beam forming of the signals based on the optimal weight vector.

[0041] Optionally, the method of constructing multiple Toeplitz matrices based on signals received by a ULA array with 2N+1 array elements comprises:

[0042] each array element in the ULA array with 2N+1 array elements is taken as a reference array element, and signals received by each array element are taken as reference signals;

[0043] signals received by all array elements are sequentially obtained, and multiple Toeplitz matrices under all reference array elements are repeatedly constructed.

[0044] Optionally, the multiple Toeplitz matrices are combined by a square sum operation to obtain a covariance matrix, expressed as follows:

[0045]

[0046] wherein R m is a Toeplitz matrix, and the N is an array element serial number.

[0047] Optionally, the steering vector of the signal is calculated according to the eigenvalue and the eigenvector, comprising:

[0048] Eigenvalues and eigenvectors of the covariance matrix are obtained by eigenvalue decomposition, and a direction of arrival (DOA) of the signal is estimated by a multiple signal classification (MUSIC) algorithm to reconstruct the steering vector.

[0049] Optionally, the expected signal power is determined according to the corresponding relationship, comprising:

[0050] According to the corresponding relationship, all steering vectors are corresponded to signal powers, and the expected signal power is placed in the first column, and the expected signal power is obtained by calculating the first column non-zero elements of the diagonal matrix.

[0051] Optionally, the covariance matrix of the mutually uncorrelated noise is reconstructed according to the noise power, the corresponding relationship and the expected signal power, comprising:

[0052] The expected signal power is deleted to obtain a diagonal matrix without the expected signal power.

[0053] The diagonal matrix is inversely operated to obtain the reconstructed covariance matrix of the mutually uncorrelated noise.

[0054] Optionally, the application is applied to control of a communication beam or a radar beam.

[0055] The application provides an adaptive signal beam forming device, comprising:

[0056] A memory is configured to store a computer executable program of the adaptive signal beam forming method.

[0057] The processor is used for calling the computer executable program from the memory to perform: constructing a multiple Toeplitz matrix based on a signal received by a 2N+1 element ULA array; combining the multiple Toeplitz matrix through a square sum operation to obtain a covariance matrix, the covariance matrix comprising eigenvalues and eigenvectors; calculating a steering vector of the signal according to the eigenvalues and the eigenvectors; optimizing the steering vector using a preset optimization function to obtain an expected signal steering vector; estimating a noise power using a minimum eigenvalue of the covariance matrix; establishing a corresponding relationship between signal power and the eigenvectors according to a correlation between the eigenvectors and the expected signal steering vector; determining an expected signal power according to the corresponding relationship; reconstructing a covariance matrix of mutually uncorrelated noises according to the noise power, the corresponding relationship and the expected signal power; inputting the expected signal steering vector and the covariance matrix of the mutually uncorrelated noises into a Capon beamformer to calculate an optimal weight vector; and controlling beamforming of the signal according to the optimal weight vector.

[0058] The application further provides a storage medium comprising a computer executable program stored therein, the computer executable program being used for being called by a processor to perform steps of the adaptive signal beamforming method.

[0059] The application has the following beneficial effects:

[0060] The application provides an adaptive signal beamforming method, comprising: constructing a multiple Toeplitz matrix based on a signal received by a 2N+1 element ULA array; combining the multiple Toeplitz matrix through a square sum operation to obtain a covariance matrix, the covariance matrix comprising eigenvalues and eigenvectors; calculating a steering vector of the signal according to the eigenvalues and the eigenvectors; optimizing the steering vector using a preset optimization function to obtain an expected signal steering vector; estimating a noise power using a minimum eigenvalue of the covariance matrix; establishing a corresponding relationship between signal power and the eigenvectors according to a correlation between the eigenvectors and the expected signal steering vector; determining an expected signal power according to the corresponding relationship; reconstructing a covariance matrix of mutually uncorrelated noises according to the noise power, the corresponding relationship and the expected signal power; inputting the expected signal steering vector and the covariance matrix of the mutually uncorrelated noises into a Capon beamformer to calculate an optimal weight vector; and controlling beamforming of the signal according to the optimal weight vector. The application is based on a signal received by a 2N+1 element uniform linear array (ULA), and the signal decorrelation and covariance matrix reconstruction are performed by constructing a multiple Toeplitz matrix and using its characteristics, so that the performance and reliability of a beamforming system in a complex signal environment are improved. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 is a schematic diagram of an adaptive signal beamforming process in the present application;

[0062] Figure 2 is a schematic diagram of an array element in the present application;

[0063] Figure 3 is a schematic diagram of an adaptive signal beamforming device in the present application. DETAILED DESCRIPTION

[0064] Exemplary embodiments of the present disclosure will be described in greater detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the drawings, it is understood that various forms of the present disclosure are implemented without being limited by the embodiments set forth herein. Rather, the embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be accurately conveyed to those skilled in the art.

[0065] Referring to Figure 1 , an adaptive signal beamforming method includes the following steps:

[0066] S101, a multiple Toeplitz matrix is constructed based on a signal received by a ULA array with 2N+1 array elements;

[0067] In the field of radar or communication, an array of array elements for receiving signals is a ULA array with 2N+1 array elements, which is completely symmetrical around the center array element, as shown in Figure 2 .

[0068] There are M incoming signals, of which K are coherent signals, and the rest are independent of each other.

[0069] According to the definition of coherent signals, the array received signal is represented as:

[0070]

[0071] The formula is represented as a vector:

[0072] x(t)=As(t)+n(t) (2)

[0073] where β i is the correlation coefficient of the i-th signal and the first signal, x(t) is a (2N+1)×1 received vector, A is a (2N+1)×M array flow matrix, A=[a1,…,a K ,a K+1 ,…,a M ], s(t) is an M×1 signal vector, and n(t) is a (2N+1)×1 noise vector.

[0074] According to the above formula derivation, let the center array element be the reference array element, and the received signal of the kth array element is expressed as:

[0075]

[0076] wherein k = -N, …, 0, …, N.

[0077] When the center array element is the reference array element, the received signal x0(t) thereof is called the reference signal, and the received signal matrix of all other array elements is obtained based on the reference signal as the reference:

[0078]

[0079] wherein represents the received signal of the kth array element relative to the reference array element, and thus, according to the formula (3), it can be derived that:

[0080]

[0081] Since the first K signals are coherent, the last M-K signals are independent, the noises are also independent of each other, and the noises and signals are also independent of each other, thus:

[0082]

[0083] It can be known that In combination with the formula (3), the formula (5) is simplified as:

[0084]

[0085] The last term in the formula (6) has a value only when k = 0, otherwise it is 0, specifically:

[0086]

[0087] wherein δ(k) represents the step function, δ(0) = 1, and the values of the rest are 0. Let represent the sum of all correlation coefficients, and the formula (6) is rewritten as:

[0088]

[0089] wherein represents the signal power.

[0090] The formula (8) is brought into (4) to express Y0 as:

[0091] Y0=A0f+n0 (9)

[0092] wherein A0 is a column full-rank matrix composed of array steering vectors, and the specific form is:

[0093]

[0094] f is a column matrix composed of power and correlation arrays, and has the form:

[0095]

[0096] n0 is a (2N+1) x M matrix with 1 at the position of k=0 and 0 at other positions.

[0097] The Y0 matrix at this time is similar to the matrix generated by single-shot data, and if the covariance of formula (9) is directly solved, the rank is still not full.

[0098] According to the definition of Toeplitz matrix, Toeplitz matrix is constructed by using the value of Y0 to achieve the purpose of reconstructing the covariance matrix to a full rank matrix, which is as follows:

[0099]

[0100] where B0 is a half-array array stream matrix, which is specifically:

[0101] B0=[a1,…,a K ,a K+1 ,…,a M ] (13)

[0102] and a i is the steering vector of the i-th signal, that is:

[0103]

[0104] where F is a diagonal matrix with the elements in f as diagonal elements. (N+1) I is an (N+1) x (N+1) unit matrix.

[0105] Since the matrix B0 is a column full rank Vandermonde matrix, the rank of the signal covariance matrix satisfies:

[0106]

[0107] That is, according to formula (12) using Toeplitz reconstruction can restore the covariance matrix to full rank, and the signal decorrelation is realized.

[0108] S102, combine the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix, wherein the covariance matrix includes eigenvalues and eigenvectors.

[0109] In step S101, only x0(t) is used as the reference signal. In order to make full use of the signal information and noise information in the array, increase the degree of de-coherence, and repeat the above steps with x m (t), m = -N,..., 0,..., N as the reference signal, the above steps are repeated, specifically:

[0110] When the reference signal is x m (t), formula (4) is rewritten as:

[0111]

[0112] At this time, formula (16) is also rewritten as:

[0113]

[0114] wherein, represents the received signal of the kth array when x m (t) is used as the reference signal.

[0115] Formula (17) is expressed as formula (8):

[0116]

[0117] wherein, δ(k-m) is a step function, that is:

[0118]

[0119] From formula (18), it can be seen that the signal and noise power are independent of the value of m. And Compared with Only the natural logarithm value and the value of the step function change. Similarly, formula (18) is expressed in matrix form:

[0120] Y m = A m f + n m (20)

[0121] wherein,

[0122]

[0123] wherein, D is a diagonal matrix, that is

[0124]

[0125] wherein, f is consistent with the result of formula (21), and n m represents is a (2N+1) x M matrix, only when k = m is 1, and the rest is 0.

[0126] Y mReconstructing into Toeplitz matrix, that is:

[0127]

[0128] Where, I (N+1),m represents an (N+1) x (N+1) matrix whose mth diagonal line is 1 and the rest is 0, when m is less than 0, 1 is located above the main diagonal line, otherwise below the main diagonal line, in particular, when m = 0, it represents the unit matrix.

[0129] According to formula (23), all reconstructed Toeplitz matrices are full rank, and the rank value is the number of signals, so the signal decorrelation is completed at the same time when they are used to reconstruct the covariance matrix. After obtaining all the Toeplitz matrices under the reference elements, square sum them to obtain the final covariance matrix, specifically:

[0130]

[0131] The covariance matrix obtained by expanding the multiple Toeplitz matrices uses the correlation information of all signals in the spatial array. Not only that, the covariance matrix obtained by the square sum operation has better decorrelation effect, and does not need to preprocess the noise, and the decorrelation effect is more obvious in the low signal-to-noise ratio environment.

[0132] S103, calculating the steering vector of the signal according to the eigenvalue and the eigenvector;

[0133] After calculating the final covariance matrix by formula (24), the eigenvalues and eigenvectors of the covariance matrix are obtained by eigenvalue decomposition, specifically:

[0134]

[0135] Where, λ k is the eigenvalue arranged from large to small, e k is the eigenvector corresponding to the kth eigenvalue, represents the power of the kth signal, a k is the steering vector of the kth signal, is the noise power.

[0136] S104, optimizing the steering vector using a preset optimization function to obtain an expected signal steering vector;

[0137] After eigenvalue decomposition, the signal DOA is estimated by the MUSIC algorithm, and the steering vector of the signal is calculated by formula (26), the expression is as follows:

[0138]

[0139] But in practical application, the steering vector is affected by many error factors, and it is easy to mismatch. Therefore, the steering vector needs to be corrected as follows. According to the knowledge of matrix space, the desired signal steering vector and its corresponding eigenvector are in the same subspace, and their inner product is maximum when the steering vector is not mismatched. Therefore, the following optimization function is constructed to correct the steering vector, and the expression of the optimization function is as follows:

[0140]

[0141] Where, <·> represents the inner product operation, a1 represents the desired signal steering vector, and ||·||2 represents the-2 norm. The N+1 eigenvectors obtained by the characteristic decomposition of formula (25) are brought into formula (27) to obtain N+1 values about e k The inner product operation value. Only the maximum value among these values represents the maximum correlation between the two, and at this time, the eigenvector and the desired signal steering vector are in the corresponding state.

[0142] Because the steering vectors of different signals are orthogonal to each other, there is:

[0143]

[0144] Similarly, the desired signal steering vector and the corresponding eigenvector also satisfy the above equation, that is:

[0145]

[0146] Finally, the corrected desired signal steering vector is:

[0147]

[0148] The corrected steering vector of formula (30) is the steering vector located at the 0th array element to the Nth array element. If the steering vector from the -Nth array element to the 0th array element is to be obtained, only the conjugate inversion of the above result is needed. Therefore, the superposition of the two can obtain the final desired signal steering vector

[0149] S05, estimating the noise power using the smallest eigenvalue of the covariance matrix;

[0150] According to the characteristic decomposition of the covariance matrix, we have:

[0151]

[0152] Since the coherent signal does not change the noise space, the smallest N+1-M eigenvalue is still used to estimate the noise power, that is:

[0153]

[0154] S106, establishing a corresponding relationship between signal power and the eigenvector according to the correlation between the eigenvector and the desired signal steering vector;

[0155] Because the expansion of the multiple Topelitz matrix may cause the change of the signal space, the steering vector and its power do not have a one-to-one correspondence, and therefore the following correlation function is defined:

[0156]

[0157] wherein E s represents the signal subspace, i.e. the linear subspace spanned by the first M eigenvectors. The M eigenvectors e k are respectively brought into the signal subspace, and if the eigenvector corresponds to the steering vector, the correlation function value of the eigenvector is maximum. Through this step, the steering vector and the eigenvector are associated, and the eigenvalue and the eigenvector are one-to-one corresponding, so as to be equivalent to the one-to-one correspondence between the steering vector and the power.

[0158] S107, determining the desired signal power according to the corresponding relationship;

[0159] After all the steering vectors are associated with the signal power, the desired signal power is placed in the first column, and the decomposition result of formula (31) is rearranged as follows:

[0160]

[0161] wherein Σ represents a diagonal matrix composed of all the signal powers, B is composed of steering vectors with a length of N+1, and the first column is the desired signal steering vector.

[0162] All the steering vectors are expanded to 2N+1 dimensions by formula (35):

[0163]

[0164] The covariance matrix is reconstructed by using the expanded steering vectors:

[0165]

[0166] wherein A is an array stream matrix composed of the steering vectors in formula (35). Therefore, the diagonal matrix is obtained as follows:

[0167]

[0168] In formula (37), the non-zero elements in the first column of the diagonal matrix Σ are the desired signal powers, and only the desired signal powers are deleted to obtain the diagonal matrix without the desired signal powers.

[0169] S108, reconstruct the covariance matrix of mutually uncorrelated noise according to the noise power, the correspondence and the expected signal power;

[0170] The inverse operation of reuse (37) can obtain the reconstructed covariance matrix of mutually uncorrelated noise (IPN CM), that is:

[0171]

[0172] S109, the expected signal steering vector and the covariance matrix of mutually uncorrelated noise are brought into Capon beamformer to calculate the optimal weight vector;

[0173] Finally, the expected signal steering vector and the reconstructed IPNCM are brought into Capon beamformer, and the optimal weight vector can be obtained, which is expressed as follows:

[0174]

[0175] S110, control the beamforming of the signal according to the optimal weight vector.

[0176] Multiply the signal received by each antenna element (or the signal to be transmitted) by the corresponding weight to adjust the phase and / or amplitude of the signal.

[0177] By applying the optimal weight vector, the shape and direction of the beam can be adjusted to achieve specific communication or radar targets.

[0178] In wireless communication, directional beams are formed to point to specific user equipment to enhance signal strength and reduce interference.

[0179] In radar systems, narrow beams are formed to accurately locate targets, or wide beams are formed to cover a wider search area.

[0180] Further, in practical applications, it is necessary to continuously monitor the effect of beamforming and adjust the optimal weight vector as needed.

[0181] Application scenarios:

[0182] Wireless communication: used to enhance signal transmission between base stations and user equipment, reduce interference, and improve communication quality and capacity.

[0183] Radar system: used for accurate detection and positioning of targets, to improve the resolution and anti-interference ability of the radar system.

[0184] Sonar system: used for underwater detection and positioning, and the detection range and accuracy of the sonar system are improved through beamforming technology.

[0185] In the present application, the Toeplitz-MR algorithm considers a symmetrical ULA, and obtains the receiving signals of other array elements with each array element as a reference array element, and reconstructs a Toeplitz matrix to achieve the purpose of signal decorrelation. Compared with the traditional single Toeplitz matrix, the algorithm uses the method of multiple Toeplitz matrix expansion, and increases the sample quantity by constructing 2N+1 Toeplitz matrices. The algorithm avoids the performance degradation problem caused by using only part of the information, and can realize the decorrelation of the signal without preprocessing the noise by diagonalizing the noise term.

[0186] Please refer to Figure 3 The present application also provides an adaptive signal beam forming device for realizing each step of the adaptive signal beam forming method, which comprises:

[0187] A matrix module 301 constructs a multiple Toeplitz matrix based on the signals received by the ULA array of 2N+1 array elements.

[0188] In the field of radar or communication, a uniform linear array (ULA) completely symmetrical around the center array element is used to receive signals. The array is composed of multiple array elements and receives M incident signals including K coherent signals and the remaining independent signals. Through mathematical derivation, the vector representation of the array receiving signal is obtained, which considers factors such as signal correlation, array flow matrix, signal vector and noise vector.

[0189] In particular, when the center array element is used as the reference array element, the relationship between the receiving signals of other array elements and the reference signal is derived. Since the first K signals are coherent, the last M-K signals are independent, and the noise is independent of the signal, the formula can be simplified to obtain an expression about the receiving signal matrix.

[0190] In order to handle the problem of rank deficiency of the covariance matrix caused by coherent signals, a Toeplitz matrix is constructed using the values of the receiving signals. Through this method, the covariance matrix can be reconstructed into a full rank matrix, thereby realizing the decorrelation of the signal.

[0191] A merging module 302 merges the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix, which includes eigenvalues and eigenvectors.

[0192] In the process of improving the coherence of the signal, in order to make full use of the signal and noise information in the array, the last module is expanded to repeat the execution with each element in the array as a reference signal. When a specific element is used as a reference, the formula of the received signal is rewritten accordingly and converted into a similar form to the original formula, but contains the received signal and correlation coefficient based on the current element.

[0193] Through mathematical transformation, these formulas are further transformed into matrix form, which contains matrices composed of array steering vectors and correlation coefficients. In order to deal with the problem of rank deficiency of the covariance matrix caused by coherent signals, Toeplitz matrices are constructed using the values of the received signals, and these matrices are proved to be full rank, with a rank equal to the number of signals.

[0194] Finally, by summing the square of the Toeplitz matrix under all reference elements, the final covariance matrix is obtained. This covariance matrix not only utilizes the correlation information of all signals in the spatial array, but also achieves better decorrelation effect through multiple Toeplitz matrix expansion and sum of squares operation, especially in low SNR environment. This method does not need to pre-process the noise, which improves the efficiency and accuracy of signal processing.

[0195] The calculation module 303 calculates the steering vector of the signal according to the eigenvalue and the eigenvector;

[0196] After calculating the final covariance matrix, by performing eigenvalue decomposition, the eigenvalues arranged in size and the corresponding eigenvectors can be obtained. These eigenvalues represent the power of each signal, and the eigenvectors are related to the steering vector of the signal.

[0197] The optimization module 304 optimizes the steering vector using a preset optimization function to obtain a desired signal steering vector;

[0198] After eigenvalue decomposition, the direction of arrival (DOA) of the signal can be estimated using the MUSIC algorithm, and the steering vector of the signal can be calculated by formula.

[0199] However, in practical applications, the steering vector is easily affected by various error factors and mismatched. In order to correct the steering vector, an optimization function is constructed, which is based on the knowledge of matrix space and uses the principle that the desired signal steering vector and its corresponding eigenvector are in the same subspace and have the maximum inner product. The eigenvector obtained by eigenvalue decomposition is substituted into the optimization function, and the eigenvector with the maximum inner product with the desired signal steering vector is found, that is, they are considered to have the maximum correlation. At this time, the eigenvector and the desired signal steering vector correspond to each other. Since the steering vectors of different signals are orthogonal to each other, this property can be used to further verify and correct the steering vector. Finally, through the corrected steering vector (including the steering vectors of the 0th array element to the Nth array element and the -Nth array element to the 0th array element, which are obtained by superimposing after flipping and taking the conjugate), a more accurate desired signal steering vector can be obtained.

[0200] The estimation module 305 estimates the noise power using the smallest eigenvalue of the covariance matrix;

[0201] According to the eigenvalue decomposition principle of the covariance matrix, the eigenvalue and eigenvector can be used to analyze the characteristics of the signal and noise. Among them, the coherent signal will not affect the noise space, so the noise power can still be estimated by the smallest eigenvalue. This conclusion provides an important basis for signal processing and parameter estimation.

[0202] The relationship module 306 establishes the corresponding relationship between the signal power and the eigenvector according to the correlation between the eigenvector and the desired signal steering vector;

[0203] Due to the expansion of the multiple Topelitz matrix, it may cause changes in the signal space, so the steering vector and its power do not have a one-to-one correspondence. Therefore, the following correlation function is defined:

[0204]

[0205] where E s represents the signal subspace, that is, the linear subspace spanned by the first M eigenvectors. The M eigenvectors e k are respectively brought into the signal subspace, and if the eigenvector corresponds to the steering vector, the correlation function value is maximum. Through this step, the steering vector and the eigenvector are connected, and the eigenvalue and the eigenvector are one-to-one corresponding, so it is equivalent to the one-to-one correspondence between the steering vector and the power.

[0206] The determination module 307 determines the desired signal power according to the corresponding relationship;

[0207] After corresponding all steering vectors to the signal power, the desired signal power is placed in the first column, and the decomposition result of formula (31) is rearranged as follows:

[0208]

[0209] wherein ∑ denotes a diagonal matrix consisting of all signal powers, B consists of steering vectors with array element length N+1, and the first column is the desired signal steering vector.

[0210] All steering vectors are expanded in dimension to 2N+1 using formula (35):

[0211]

[0212] The covariance matrix is reconstructed using the expanded steering vectors:

[0213]

[0214] wherein A is an array flow matrix consisting of steering vectors in (35). Therefore, the diagonal matrix is obtained as:

[0215]

[0216] In formula (37), the first column non-zero element of the diagonal matrix ∑ is the desired signal power, and only by deleting it can the diagonal matrix without the desired signal power be obtained.

[0217] The reconstruction module 308 reconstructs the covariance matrix of mutually uncorrelated noises according to the noise power, the corresponding relationship and the desired signal power;

[0218] The inverse operation of (37) is used to obtain the reconstructed covariance matrix of mutually uncorrelated noises (IPNC M), that is:

[0219]

[0220] The weight vector module 309 brings the desired signal steering vector and the covariance matrix of mutually uncorrelated noises into the Capon beamformer to calculate the optimal weight vector.

[0221] Finally, the desired signal steering vector and the reconstructed IPNCM are brought into the Capon beamformer to obtain the optimal weight vector, and the expression is as follows:

[0222]

[0223] The control module 310 controls the beamforming of the signal according to the optimal weight vector.

[0224] Further, a multiple Toeplitz matrix is constructed based on the signals received by the 2N+1 element ULA array, including:

[0225] Each array element in the 2N+1 array is taken as a reference array element, and the signal received by each array element is taken as a reference signal.

[0226] The received signals of all array elements are sequentially obtained, and multiple Toeplitz matrices under all reference array elements are repeatedly constructed.

[0227] Further, the multiple Toeplitz matrices are combined through a square sum operation to obtain a covariance matrix, and the expression is as follows:

[0228]

[0229] wherein R m is a Toeplitz matrix, and N is an array element number.

[0230] Further, the steering vector of the signal is calculated according to the eigenvalue and the eigenvector, and includes:

[0231] Eigenvalues and eigenvectors of the covariance matrix are obtained through eigenvalue decomposition, and a direction of arrival (DOA) of the signal is estimated through a multiple signal classification (MUSIC) algorithm to reconstruct the steering vector.

[0232] Further, the expected signal power is determined according to the corresponding relationship, and includes:

[0233] After all steering vectors are corresponded to signal powers according to the corresponding relationship, the expected signal power is placed in the first column, and the expected signal power is obtained by calculating the first column non-zero elements of the diagonal matrix.

[0234] Further, the covariance matrix of mutually uncorrelated noise is reconstructed according to the noise power, the corresponding relationship, and the expected signal power, and includes:

[0235] The expected signal power is deleted to obtain a diagonal matrix without the expected signal power.

[0236] The diagonal matrix is inversely operated to obtain the reconstructed covariance matrix of mutually uncorrelated noise.

[0237] Further, the application is applied to control of a communication beam or a radar beam.

[0238] The application provides a self-adaptive signal beam forming device, which includes:

[0239] A memory is configured to store a computer executable program of the self-adaptive signal beam forming method.

[0240] The processor is used to call the computer executable program from the memory and execute: constructing a multiple Toeplitz matrix based on the signals received by the ULA array of 2N+1 array elements; combining the multiple Toeplitz matrix through a square sum operation to obtain a covariance matrix, the covariance matrix including eigenvalues and eigenvectors; calculating a steering vector of the signal according to the eigenvalues and the eigenvectors; optimizing the steering vector using a preset optimization function to obtain an expected signal steering vector; estimating a noise power using the minimum eigenvalue of the covariance matrix; establishing a corresponding relationship between signal power and the eigenvectors according to the correlation between the eigenvectors and the expected signal steering vector; determining an expected signal power according to the corresponding relationship; reconstructing a covariance matrix of mutually independent noises according to the noise power, the corresponding relationship and the expected signal power; bringing the expected signal steering vector and the covariance matrix of mutually independent noises into a Capon beamformer to calculate an optimal weight vector; and controlling the beamforming of the signal according to the optimal weight vector.

[0241] The application further provides a storage medium, comprising: a computer executable program stored therein, the computer executable program being used to be called by a processor to execute the steps of the adaptive signal beamforming method.

[0242] The above description is only preferred embodiments of the present application, and it should be pointed out that, for those skilled in the art, without departing from the technical principles of the present application, a number of improvements and modifications can be made, and these improvements and modifications should be considered as the protection scope of the present application.

Claims

1. A method of adaptive signal beamforming, characterized by, The method comprises the following steps: constructing multiple Toeplitz matrices based on signals received by a ULA array with 2N+1 array elements; combining the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix, wherein the covariance matrix comprises eigenvalues and eigenvectors; calculating a steering vector of the signals according to the eigenvalues and the eigenvectors; optimizing the steering vector using a preset optimization function to obtain an expected signal steering vector; estimating noise power using the minimum eigenvalue of the covariance matrix; establishing a corresponding relationship between signal power and the eigenvectors according to the correlation between the eigenvectors and the expected signal steering vector; determining expected signal power according to the corresponding relationship; reconstructing a covariance matrix of mutually uncorrelated noise according to the noise power, the corresponding relationship and the expected signal power; calculating an optimal weight vector by inputting the expected signal steering vector and the covariance matrix of mutually uncorrelated noise into a Capon beamformer; controlling beamforming of the signals according to the optimal weight vector. The method for determining expected signal power according to the corresponding relationship comprises the following steps: after all steering vectors are corresponded with signal power according to the corresponding relationship, the expected signal power is placed in the first column, and the expected signal power is obtained by calculating the first column non-zero elements of a diagonal matrix. The method for reconstructing a covariance matrix of mutually uncorrelated noise according to the noise power, the corresponding relationship and the expected signal power comprises the following steps: the expected signal power is deleted to obtain a diagonal matrix without the expected signal power; and the diagonal matrix is subjected to inverse operation to obtain the reconstructed covariance matrix of mutually uncorrelated noise.

2. The adaptive signal beamforming method of claim 1, wherein, The method for constructing multiple Toeplitz matrices based on signals received by a ULA array with 2N+1 array elements comprises the following steps: taking each array element in the ULA array with 2N+1 array elements as a reference array element, and taking the signals as reference signals; sequentially obtaining all the reference signals, and repeatedly constructing multiple Toeplitz matrices under all reference array elements.

3. The adaptive signal beamforming method of claim 1, wherein, The method for combining the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix comprises the following steps: ； wherein is a Toeplitz matrix, said N is the array element number.

4. The adaptive signal beamforming method of claim 1, wherein, The method for calculating a steering vector of the signals according to the eigenvalues and the eigenvectors comprises the following steps: obtaining the eigenvalues and the eigenvectors of the covariance matrix by eigen decomposition, and estimating the direction of arrival (DOA) of the signals by MUSIC algorithm to reconstruct the steering vector.

5. The adaptive signal beamforming method according to any one of claims 1-4, characterized in that, The method is applied to control of communication beams or radar beams.

6. An adaptive signal beamforming apparatus, characterized by The method comprises the following steps: a matrix module, which constructs multiple Toeplitz matrices based on signals received by a ULA array with 2N+1 array elements; a combining module, which combines the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix, wherein the covariance matrix comprises eigenvalues and eigenvectors; a calculating module, which calculates a steering vector of the signals according to the eigenvalues and the eigenvectors; an optimization module, which optimizes the steering vector using a preset optimization function to obtain an expected signal steering vector; an estimating module, which estimates noise power using the minimum eigenvalue of the covariance matrix; an establishing module, which establishes a corresponding relationship between signal power and the eigenvectors according to the correlation between the eigenvectors and the expected signal steering vector; a relationship module, which establishes a corresponding relationship between signal power and the eigenvectors according to the correlation between the eigenvectors and the desired signal steering vector; a determination module, which determines the desired signal power according to the corresponding relationship; a reconstruction module, which reconstructs the covariance matrix of mutually uncorrelated noise according to the noise power, the corresponding relationship and the desired signal power; a weight vector module, which brings the desired signal steering vector and the covariance matrix of mutually uncorrelated noise into a Capon beamformer to calculate the optimal weight vector; a control module, which controls the beamforming of the signal according to the optimal weight vector; wherein determining the desired signal power according to the corresponding relationship comprises: placing the desired signal power in the first column after corresponding all steering vectors with signal power according to the corresponding relationship, and obtaining the desired signal power by calculating the non-zero elements of the first column of the diagonal matrix; and reconstructing the covariance matrix of mutually uncorrelated noise according to the noise power, the corresponding relationship and the desired signal power comprises: deleting the desired signal power to obtain a diagonal matrix without the desired signal power; and performing inverse operation on the diagonal matrix to obtain the reconstructed covariance matrix of mutually uncorrelated noise.

7. The apparatus of claim 6 wherein, constructing multiple Toeplitz matrices based on the signals received by the ULA array of 2N+1 array elements, comprising: taking each array element in the ULA array of 2N+1 array elements as a reference array element, and taking the signal received by each array element as a reference signal; sequentially obtaining all the reference signals and repeatedly constructing multiple Toeplitz matrices under all reference array elements.

8. The apparatus of claim 6 wherein, combining the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix, and the expression is as follows: ； wherein is a Toeplitz matrix, said N is the array element number.

9. The apparatus of claim 6 wherein, calculating the steering vector of the signal according to the eigenvalue and the eigenvector, comprising: obtaining the eigenvalue and the eigenvector of the covariance matrix by eigenvalue decomposition, and estimating the direction of arrival (DOA) of the signal by MUSIC algorithm to reconstruct the steering vector.

10. The adaptive signal beam forming apparatus according to any one of claims 6 to 9, wherein, application in the control of communication beams or radar beams.

11. An adaptive signal beamforming device, characterized by comprising: a memory for storing the computer executable program of the adaptive signal beamforming method of any one of claims 1-5; a processor for calling the computer executable program from the memory and performing: constructing multiple Toeplitz matrices based on the signals received by the ULA array of 2N+1 array elements; combining the multiple Toeplitz matrices through square sum operation to obtain a covariance matrix, and the covariance matrix comprises an eigenvalue and an eigenvector; calculating the steering vector of the signal according to the eigenvalue and the eigenvector; optimizing the steering vector using a preset optimization function to obtain a desired signal steering vector; estimating the noise power using the minimum eigenvalue of the covariance matrix; establishing a corresponding relationship between signal power and the eigenvectors according to the correlation between the eigenvectors and the desired signal steering vector; determining the desired signal power according to the corresponding relationship; reconstructing the covariance matrix of mutually uncorrelated noise according to the noise power, the corresponding relationship and the desired signal power; The desired signal steering vector and the covariance matrix of the mutually independent noise are input into a Capon beamformer to calculate an optimal weight vector; Beamforming of the signal is controlled according to the optimal weight vector; The method further comprises: determining the desired signal power according to the correspondence relationship, including: corresponding all steering vectors with signal powers according to the correspondence relationship, placing the desired signal power in a first column, and obtaining the desired signal power by calculating non-zero elements in the first column of a diagonal matrix; and reconstructing the covariance matrix of the mutually independent noise according to the noise power, the correspondence relationship and the desired signal power, including: deleting the desired signal power to obtain a diagonal matrix without the desired signal power; and performing an inverse operation on the diagonal matrix to obtain the reconstructed covariance matrix of the mutually independent noise.

12. A storage medium, characterized by The method further comprises: A computer executable program is stored in the computer readable storage medium, and the computer executable program is used to be called by the processor to execute the steps of the adaptive signal beamforming method according to any one of claims 1 to 5.

Images