Two-dimensional DOA estimation method and device based on MUSIC and OMP and medium

Abstract

The application provides a two-dimensional DOA estimation method and device based on MUSIC and OMP and a medium. After determining echo data of a single snapshot received by each array element in a two-dimensional uniform or non-uniform rectangular array, a multi-dimensional MUSIC algorithm is used. Snapshot data determined according to a direction array element structure is directly used to perform one-dimensional MUSIC operation on echo data of another direction array element, to obtain a middle angle set and a pitch angle set of different targets. Without using a spatial smoothing algorithm, angle accurate estimation is realized without losing array aperture. Then, angles included in the middle angle set and the pitch angle set are matched according to the OMP algorithm. Compared with a two-dimensional spectrum peak search mode, the calculation amount is greatly reduced, and real-time requirements of various application scenarios are met.

Description

TECHNICAL FIELD

[0001] The present application mainly relates to the field of array signal processing, and more particularly to a two-dimensional DOA estimation method and device based on MUSIC and OMP and a medium. BACKGROUND

[0002] In recent years, the Direction Of Arrival (DOA) estimation technology has been widely applied to the research of array signal parameter estimation. In the application of vehicle-mounted millimeter wave radar, for example, a two-dimensional planar array is usually used to estimate the elevation angle and azimuth angle simultaneously, so as to obtain the three-dimensional (3D) spatial coordinates of the target. The two-dimensional planar array can be a rectangular array realized based on the Multiple-Input Multiple-Output (MIMO) technology, which can achieve a high angular resolution with fewer antenna elements and a smaller physical aperture.

[0003] In order to improve the estimation accuracy and stability, as well as the angular resolution, the Multiple Signal Classification (MUSIC) algorithm has been proposed to be applied to the DOA estimation of the rectangular array. However, this two-dimensional DOA estimation scheme needs to generate multiple virtual snapshot data by using spatial smoothing to achieve accurate estimation, which not only reduces the array aperture, but also limits the application to uniform rectangular arrays. In addition, when estimating the two-dimensional angle, the two-dimensional spectrum peak search needs to be performed, which leads to a sharp increase in the computational complexity of the algorithm, and cannot meet the real-time requirements of the application scenarios. SUMMARY

[0004] To solve the above technical problems, the present application provides the following technical solutions:

[0005] The present application provides a two-dimensional DOA estimation method based on MUSIC and OMP, which comprises the following steps:

[0006] determining the echo data of a single snapshot received by each element in the rectangular array;

[0007] performing one-dimensional Multiple Signal Classification (MUSIC) operation on the echo data received by the elements in the same direction to obtain the intermediate angle set and the pitch angle set of different targets; wherein the snapshot data used in the one-dimensional MUSIC operation is determined according to the structure of the elements in another direction, and the intermediate angle of any target is related to the pitch angle and azimuth angle of the target;

[0008] According to the orthogonal matching pursuit (OMP) algorithm, the plurality of intermediate angles in the intermediate angle set and the plurality of elevation angles in the elevation angle set are matched to determine the target intermediate angle and the target elevation angle corresponding to each of the different targets;

[0009] According to the target intermediate angle and the target elevation angle, the target azimuth angle corresponding to the target is obtained.

[0010] Optionally, the one-dimensional multiple signal classification (MUSIC) operation is performed on the echo data received by the same direction element to obtain the intermediate angle set and the elevation angle set of different targets, including:

[0011] According to the spatial order of the rectangular array, the echo data received by each element is arranged to obtain an array echo matrix;

[0012] According to the array echo matrix and the first dimension of the elevation direction element, a first data covariance matrix for the azimuth direction element is obtained;

[0013] According to the array echo matrix and the second dimension of the azimuth direction element, a second data covariance matrix for the elevation direction element is obtained;

[0014] A first steering vector matrix of each element in different directions is obtained;

[0015] According to the first data covariance matrix and the second data covariance matrix, and the first steering vector matrix in the corresponding direction, a power spectrum peak value search is performed to obtain the intermediate angle set and the elevation angle set corresponding to each target in different directions.

[0016] Optionally, according to the first data covariance matrix and the second data covariance matrix, and the first steering vector in the corresponding direction, a power spectrum peak value search is performed to obtain the intermediate angle set and the elevation angle set corresponding to each target in different directions, including:

[0017] The first data covariance matrix and the second data covariance matrix are respectively subjected to eigenvalue decomposition, and according to the number of targets, the array noise eigenvectors corresponding to different directions are obtained;

[0018] According to the array noise eigenvectors and the first steering vector array in the same direction, a peak value search is performed to obtain the same number of maximum values as the number of targets in the direction;

[0019] The angles corresponding to the maximum values in the same direction are used to form the intermediate angle set and the elevation angle set of different targets.

[0020] Optionally, the matching processing of the plurality of intermediate angles in the intermediate angle set and the plurality of elevation angles in the elevation angle set according to the orthogonal matching pursuit (OMP) algorithm to determine the target intermediate angle and the target elevation angle corresponding to each of the different targets comprises:

[0021] performing vectorization processing on the two-dimensional echo data received by the rectangular array to obtain corresponding one-dimensional echo data;

[0022] obtaining a second steering vector matrix corresponding to the one-dimensional echo data according to the first steering vector matrix of each array element in different directions;

[0023] performing sparse vector reconstruction by using the orthogonal matching pursuit (OMP) algorithm according to the second steering vector matrix, the intermediate angle set and the elevation angle set;

[0024] matching each two-dimensional estimated angle in the two-dimensional estimated angle set according to the position of the non-zero element in the reconstructed sparse vector to obtain the target intermediate angle and the target elevation angle corresponding to each of the different targets.

[0025] Optionally, the sparse vector reconstruction by using the orthogonal matching pursuit (OMP) algorithm according to the second steering vector matrix, the intermediate angle set and the elevation angle set comprises:

[0026] combining the plurality of intermediate angles in the intermediate angle set with the plurality of elevation angles in the elevation angle set to obtain a two-dimensional estimated angle set of different targets;

[0027] obtaining a steering matrix corresponding to the two-dimensional estimated angle set according to the second steering vector matrix;

[0028] constructing a compressed sensing model according to the steering array and the sparse vector;

[0029] reconstructing the sparse vector in the compressed sensing model by using the orthogonal matching pursuit (OMP) algorithm.

[0030] Optionally, the method further comprises:

[0031] determining the signal amplitude corresponding to the target according to the numerical value of the non-zero element in the reconstructed sparse vector.

[0032] Optionally, the determination of the echo data of a single shot received by each array element in the rectangular array comprises:

[0033] obtaining the relative coordinates of each array element in the rectangular array in different directions and the signal wavelength of each target incident to the rectangular array;

[0034] modeling at least according to the relative coordinates and the signal wavelength to obtain an echo signal model;

[0035] Utilize the echo signal model, obtain the echo data of single snapshot received by each array element in the rectangular array.

[0036] The application also provides a two-dimensional DOA estimation device based on MUSIC and OMP, which comprises:

[0037] An echo data determination module is configured to determine the echo data of single snapshot received by each array element in the rectangular array.

[0038] A one-dimensional MUSIC operation module is configured to perform one-dimensional multiple signal classification (MUSIC) operation on the echo data received by the array elements in the same direction to obtain the intermediate angle set and the pitch angle set of different targets; wherein the snapshot data used in the one-dimensional MUSIC operation is determined according to the structure of the array elements in another direction, and the intermediate angle of any target is associated with the pitch angle and the azimuth angle of the target.

[0039] A matching processing module is configured to perform matching processing on the multiple intermediate angles in the intermediate angle set and the multiple pitch angles in the pitch angle set according to the orthogonal matching pursuit (OMP) algorithm to determine the target intermediate angle and the target pitch angle corresponding to each of the different targets.

[0040] A target azimuth angle obtaining module is configured to obtain the target azimuth angle corresponding to the target according to the target intermediate angle and the target pitch angle.

[0041] Optionally, the one-dimensional MUSIC operation module comprises:

[0042] An array echo matrix obtaining unit is configured to arrange the echo data received by each array element according to the spatial order of the rectangular array to obtain array echo matrices of different structures.

[0043] A first data covariance matrix obtaining unit is configured to obtain a first data covariance matrix for the azimuth direction array elements according to the array echo matrix corresponding to the azimuth direction and the first dimension of the pitch direction array elements.

[0044] A second data covariance matrix obtaining unit is configured to obtain a second data covariance matrix for the pitch direction array elements according to the array echo matrix corresponding to the pitch direction and the second dimension of the azimuth direction array elements.

[0045] A first steering vector matrix obtaining unit is configured to obtain a first steering vector matrix of each array element in different directions.

[0046] A peak searching unit is configured to search for power spectrum peaks according to the first data covariance matrix and the second data covariance matrix and the first steering vector matrix in the corresponding direction, and obtain a middle angle set and a pitch angle set corresponding to each target in different directions.

[0047] The application further provides a computer readable storage medium, which stores a computer program, and the computer program is loaded and executed by a processor to realize the two-dimensional DOA estimation method based on MUSIC and OMP.

[0048] Therefore, the two-dimensional DOA estimation method based on MUSIC and OMP, the device and the medium provided by the application can be used to determine the echo data of a single fast shot received by each array element in a two-dimensional uniform or divided uniform rectangular array, and then a divided dimension MUSIC algorithm is used to directly use the fast shot data determined according to the result of a direction array element to perform one-dimensional MUSIC operation on the echo data of another direction array, so as to obtain the middle angle set and the pitch angle set of different targets, without using a spatial smoothing algorithm, the angle precision estimation is realized without losing the array aperture, and then the angles included in the middle angle set and the pitch angle set are matched according to the OMP algorithm, compared with the two-dimensional spectrum peak search mode, the calculation amount is greatly reduced, and the real-time requirement of various application scenarios is met. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute the embodiments of the application, and for those skilled in the art, other drawings can be obtained without creative labor based on the provided drawings.

[0050] Figure 1 A flowchart of an optional example of the two-dimensional DOA estimation method based on MUSIC and OMP provided by the application;

[0051] Figure 2 A structural diagram of a two-dimensional rectangular array suitable for the two-dimensional DOA estimation method based on MUSIC and OMP provided by the application;

[0052] Figure 3 A flowchart of another optional example of the two-dimensional DOA estimation method based on MUSIC and OMP provided by the application;

[0053] Figure 4 A simulation parameter diagram for simulating and verifying the two-dimensional DOA estimation method based on MUSIC and OMP provided by the application;

[0054] Figure 5 A RMSE curve diagram of the two-dimensional DOA estimation method based on MUSIC and OMP proposed in the present application and the two-dimensional DOA estimation method based on two-dimensional smoothing MUSIC;

[0055] Figure 6 A structure diagram of an optional example of the two-dimensional DOA estimation device based on MUSIC and OMP proposed in the present application;

[0056] Figure 7 A hardware structure diagram of an optional example of the computer device suitable for the two-dimensional DOA estimation method based on MUSIC and OMP proposed in the present application. DETAILED DESCRIPTION

[0057] The technical problem described in the background section is that the accurate estimation of the multiple signal classification (MUSIC) algorithm depends on the accurate estimation of the array covariance matrix, and the accurate estimation of the array covariance matrix requires multiple snapshot data, which is difficult to obtain in a highly dynamic application environment. Therefore, multiple virtual snapshot data are generated by spatial smoothing, which reduces the array aperture of the two-dimensional smoothing MUSIC algorithm and is only suitable for average rectangular arrays. Moreover, the calculation amount is too large, which makes it difficult to meet the actual application.

[0058] To solve the above technical problem, the present application aims to accelerate the operation speed of the two-dimensional direction of arrival (DOA) estimation based on MUSIC, so that it can also be applied to real-time two-dimensional DOA estimation of non-uniform rectangular arrays. Specifically, to solve the problem of large calculation amount proposed above, the present application proposes to perform one-dimensional MUSIC operation in two dimensions, and then perform matching operation on the estimated intermediate angle (intermediate angle variable composed of azimuth angle and elevation angle) and elevation angle, such as orthogonal matching pursuit (OMP) algorithm, to realize accurate angle matching, thereby avoiding two-dimensional spectrum peak search and greatly reducing the calculation amount.

[0059] To solve the problem of obtaining multiple snapshot data, the present application proposes to use another data as multiple snapshot data for operation when performing one-dimensional MUSIC angle estimation in one dimension of the rectangular array, without using spatial smoothing algorithm. In this way, multiple snapshot data can be obtained without losing the array aperture, and accurate MUSIC angle estimation can be realized, so that the two-dimensional DOA estimation method proposed in the present application is not only suitable for uniform rectangular arrays, but also suitable for non-uniform rectangular arrays, thereby increasing the application scenario.

[0060] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort belong to the scope of protection of the present application.

[0061] Referring to Figure 1 An optional flowchart of an example of a two-dimensional DOA estimation method based on MUSIC and OMP proposed in the present application can be executed by a computer device, which includes a terminal device with certain data calculation capability, such as a smart phone, a notebook computer, a vehicle-mounted terminal, etc., or a server, such as a physical server or a cloud server, etc. The present application does not limit the product type of the computer device. As shown in Figure 1 The method can include the following steps.

[0062] In step S11, single-shot echo data received by each array element in the rectangular array is determined.

[0063] In actual applications, a two-dimensional surface array usually includes a matrix array, an L-shaped array and a circular array, etc. Compared with the L-shaped array and the circular array, the array elements of the rectangular array are more densely arranged, and accurate estimation of the two-dimensional angle can still be achieved in the case of single-shot or few-shot, therefore, the present application selects to use the rectangular array for two-dimensional DOA estimation.

[0064] The rectangular array can be a two-dimensional uniform / non-uniform array, and the present application takes a two-dimensional uniform surface array as an example for illustration. The implementation process of two-dimensional DOA estimation for a non-uniform rectangular array is similar, which is not described in detail. Referring to Figure 2 The array structure diagram of the two-dimensional uniform surface array is shown in M-1 ] M×1 The relative coordinates of the M array elements in the x-axis direction (which can be referred to as the azimuth direction) are X=[0,x1,x2,……,x N-1 ] N×1 .

[0065] In actual applications, if K far-field signals (referred to as targets in the present application) are incident to the rectangular array, the azimuth angle (i.e. the incident azimuth angle) of the kth target is θ k and the elevation angle is φ k . The echo can be modeled using at least these parameters by ignoring the influence of noise, and the implementation method of echo modeling is not limited in the present application. It should be noted that the values of the parameters listed above are not limited, and can be determined as appropriate.

[0066] Because the median angle α of the k-th target is directly obtained by the two-dimensional DOA estimation technique k and pitch angle φ k , where the intermediate angle α k It is the azimuth angle θ of the target. k and pitch angle φ k The intermediate angle variable is defined by combination, and the three variables have a sinα relationship. k =sinθ k cosφ k Therefore, after obtaining the two-dimensional DOA estimation results, this application can determine the azimuth angle of the corresponding target based on this trigonometric function relationship. The following will explain how to obtain the median angle α of each target. k and pitch angle The implementation process is described.

[0067] Step S12: Perform a one-dimensional MUSIC operation on the echo data received by the array elements in the same direction to obtain the intermediate angle set and elevation angle set of different targets.

[0068] In this application embodiment, to address the problem of high computational cost of the two-dimensional MUSIC (Multiple Signal Classification) algorithm, a multi-dimensional MUSIC algorithm is proposed, that is, MUSIC is performed in two dimensions (e.g., ...). Figure 2 A one-dimensional MUSIC operation is performed along the x-axis and z-axis directions (as shown) to estimate the intermediate angle and elevation angle of each of the K targets, resulting in an intermediate angle set A and an elevation angle set Φ for different targets. The intermediate angle set A = {α1, α2, ..., α...} K ,} K×1 The set of elevation angles Φ = {φ1, φ2, ..., φ K} K×1 The value of K can be determined as needed. For any target, the intermediate angle α is related to the target's elevation angle φ and azimuth angle θ, and can be determined according to, but is not limited to, the trigonometric function transformation relationship described above.

[0069] It should be noted that during the execution of the above-mentioned one-dimensional MUSIC operation, the characteristic of the antenna array being a rectangular array will be fully utilized to determine the snapshot data using another dimension. That is, the snapshot data used in the above-mentioned one-dimensional MUSIC operation is determined based on the array element structure in another direction.

[0070] Exemplarily, in the one-dimensional MUSIC operation in the azimuth direction, the number N of the array elements in the elevation direction can be used as the snapshot data; in the one-dimensional MUSIC operation in the elevation direction, the number M of the array elements in the azimuth direction can be used as the snapshot data. It can be seen that the multi-dimensional MUSIC algorithm proposed in the application does not need to use the spatial smoothing algorithm to obtain the virtual snapshot data, thereby overcoming a series of problems caused by using the spatial smoothing algorithm, realizing the accurate angle estimation under different types of rectangular arrays, and improving the practicability. The operation principle of the MUSIC algorithm is not described in detail.

[0071] In step S13, the plurality of intermediate angles in the intermediate angle set and the plurality of elevation angles in the elevation angle set are matched according to the orthogonal matching pursuit (OMP) algorithm, to determine the target intermediate angle and the target elevation angle corresponding to each of the different targets.

[0072] Since the MUSIC algorithm is used for angle estimation, a one-to-one correspondence is not established between the plurality of intermediate angles in the intermediate angle set and the plurality of elevation angles in the elevation angle set, and an angle matching algorithm needs to be used to match the angles to determine the target intermediate angle and the target elevation angle corresponding to the same target. In this regard, in order to avoid two-dimensional spectrum peak searching and reduce the calculation amount, the orthogonal matching pursuit (OMP) algorithm can be used to realize accurate angle matching, and accurate estimation of the signal reflection amplitude can also be realized. The operation principle of the OMP algorithm is not described in detail.

[0073] In step S14, the target azimuth angle of the target is obtained according to the target intermediate angle and the target elevation angle of the same target.

[0074] According to the above analysis, the target azimuth angle of the target can be obtained according to sinα k = sinθ k cosφ k , the target intermediate angle α k and the target elevation angle φ k of the same target are substituted into the formula, and the target azimuth angle α k of the target is obtained, but the calculation method is not limited to this.

[0075] In summary, in this embodiment, after determining the echo data of a single snapshot received by each element in a two-dimensional uniform or partially uniform rectangular array, a multidimensional MUSIC algorithm is used. The snapshot data determined based on the results of the array elements in one direction is used to perform a one-dimensional MUSIC operation on the echo data of the array in another direction to obtain the intermediate angle set and elevation angle set of different targets. Without using a spatial smoothing algorithm, accurate angle estimation is achieved without losing the array aperture. Then, the angles contained in the intermediate angle set and elevation angle set are matched according to the OMP algorithm. Compared with the two-dimensional spectral peak search method, the amount of computation is greatly reduced, which meets the real-time requirements of various application scenarios.

[0076] Reference Figure 3 This is a flowchart illustrating another optional example of the two-dimensional DOA estimation method based on MUSIC and OMP proposed in this application. This embodiment can describe an optional refined implementation process of the two-dimensional DOA estimation method based on MUSIC and OMP proposed above, such as... Figure 3 As shown, the method may include:

[0077] Step S31: Determine the echo data of each snapshot received by each element in the rectangular array;

[0078] As described above Figure 2 Taking the echo modeling of a two-dimensional uniform surface array (rectangular array) as an example, and combining it with the relevant description of the two-dimensional rectangular array structure above, we obtain the relative coordinates of each array element in different directions (such as azimuth and elevation directions, i.e., the x-axis and z-axis directions) of the uniform surface array, such as X = [0, x1, x2, ..., x...]. M-1 ] M×1 Z = [0, z1, zx2, ..., z N-1 ] N×1 After determining the signal wavelengths λ of K targets, the influence of noise can be ignored for echo modeling, resulting in the echo data S received by the m-th element in the x-axis direction and the n-th element in the z-axis direction. mn It can be represented as:

[0079]

[0080] In formula (1) above, 1 ≤ m ≤ M, 1 ≤ n ≤ N, and m and n are both positive integers. exp() represents an exponential function with the natural constant e as its base. Since the intermediate angle α of the same target... k Azimuth θ k and pitch angle φ k There is sinα between them k =sinθ k cosφ k Substituting this relationship into formula (1) above, we can obtain:

[0081]

[0082] In combination with the idea of the two-dimensional DOA estimation technique, the echo signal model shown in formula (2) can be used to obtain the echo data received by each array element in the rectangular array, and subsequent steps are performed to determine the intermediate angle a k and the pitch angle f k According to sin a k = sin 0 k cos f k , the azimuth angle 0 k corresponding to the target is obtained.

[0083] In step S32, the echo data received by each array element is arranged according to the spatial order of the rectangular array to obtain an array echo matrix.

[0084] In actual application, the angle estimation of the MUSIC algorithm needs to rely on the accurate estimation of the covariance matrix of the rectangular matrix. In the process of obtaining the data covariance matrix of the rectangular array in different directions, the echo data received by each array element of the single snapshot rectangular matrix can be arranged into an MxN array echo matrix S according to the array spatial order of the rectangular matrix, that is:

[0085]

[0086] As can be seen, each element S mn in the above array echo matrix S can represent the echo data received by the mth element in the x-axis direction and the nth element in the z-axis direction, and its representation can be determined in combination with the above formula (2).

[0087] In step S33, the first data covariance matrix for the azimuth direction array element is obtained according to the array echo matrix and the first dimension of the pitch direction array element.

[0088] In step S34, the second data covariance matrix for the pitch direction array element is obtained according to the array echo matrix and the second dimension of the azimuth direction array element.

[0089] After the array echo matrix S is determined, in the one-dimensional MUSIC operation process in the x-axis direction, the first data covariance matrix R x of the azimuth direction array element can be calculated according to the following formula (4), and similarly, in the one-dimensional MUSIC operation in the z-axis direction, the second data covariance matrix R z of the pitch direction array element can be calculated according to the following formula (5):

[0090]

[0091]

[0092] In the above formula, S H may represent the conjugate transpose matrix of the array echo matrix S T may represent the transpose matrix of the array echo matrix S.

[0093] Step S35, obtaining the first steering vector matrix of each array element in different directions;

[0094] In order to perform spectral peak search on the echo data received by the array of different dimensions, the steering vector of each array element of the x-axis direction array can be obtained to obtain the first steering vector matrix v x (α) in the azimuth direction.

[0095]

[0096] Similarly, the first steering vector matrix v z (φ) of the z-axis direction array.

[0097]

[0098] Step S36, performing eigenvalue decomposition on the first data covariance matrix and the second data covariance matrix respectively, and obtaining the array noise eigenvectors corresponding to different directions according to the number of targets.

[0099] In combination with the related description of the technical solutions of the present application above, the power spectral peak value search can be performed according to the first data covariance matrix and the second data covariance matrix, and the first steering vector matrix in the corresponding direction, to obtain the intermediate angle set and the pitch angle set corresponding to each target in different directions.

[0100] Specifically, since the eigenvectors corresponding to the data covariance matrix are composed of signal eigenvectors and noise eigenvectors, the present application can perform eigenvalue decomposition (Eigen Value Decomposition, EVD) on the first data covariance matrix and the second data covariance matrix respectively, such as performing eigenvalue decomposition on the first data covariance matrix corresponding to the azimuth direction to obtain the corresponding first eigenvectors R x = U x ∑ x U x H Then, the first eigenvectors can be sorted according to the size of the eigenvalues, the first eigenvectors corresponding to the largest K eigenvalues are used to form a signal subspace, and the remaining first eigenvectors are used to form a noise subspace, so that the first eigenvectors can be transformed into R x = U xs ∑ xs U xsH +U xn ∑ xn U xn H , where U xs represents the horizontal array signal subspace, i.e. the array signal eigenvector, U xn represents the horizontal array noise subspace, i.e. the array noise eigenvector. Similarly, the present application can perform eigen decomposition on the second data covariance matrix in this way to obtain the noise eigenvector corresponding to the elevation direction.

[0101] Step S37, peak searching is performed according to the array noise eigenvector of the same direction and the first steering vector array to obtain the same number of maximum values as the target in the direction;

[0102] Step S38, the angles corresponding to the maximum values of the same direction are used to form the intermediate angle set and the elevation angle set of different targets;

[0103] When performing spectrum peak searching in different directions, the power spectrum in the x-axis direction can be represented as:

[0104]

[0105] The power spectrum in the z-axis direction can be represented as:

[0106]

[0107] Peak searching is performed on P x (α) and P z (φ) respectively, K maximum values of the spectrum peak are obtained, the intermediate angle set A corresponding to the K maximum values searched in the x-axis direction can be represented as A = {α1, α2, …, α K ,} K×1 The elevation angle set Φ corresponding to the K maximum values searched in the z-axis direction can be represented as Φ = {φ1, φ2, …, φ K} K×1 The present application does not describe the one-dimensional spectrum peak searching implementation method described above in detail, and the calculation amount is greatly reduced compared with the two-dimensional spectrum peak searching method.

[0108] Step S39, vectorization processing is performed on the two-dimensional echo data received by the rectangular array to obtain corresponding one-dimensional echo data;

[0109] The application can use the OMP algorithm to match the intermediate angle and the pitch angle obtained above to determine the intermediate angle and the pitch angle corresponding to the same target. Therefore, before the OMP algorithm is executed, the two-dimensional echo data received by the rectangular array can be arranged into a one-dimensional echo signal S1, and the vectorization of the array echo matrix S can be used to obtain the one-dimensional echo data S1:

[0110] S1 = vec(S)

[0111] = [s 11 ,s 12 …s 1N ,s 21 ,s 22 ,…s 2N ,…,s M1 ,s M2 ,…s MN ] MN×1 (10)

[0112] In the above formula, vec() can represent a vectorization function, and the vectorization implementation process of the application will not be described in detail.

[0113] Step S310, obtaining a second steering vector matrix corresponding to the one-dimensional echo data according to the first steering vector matrix of each array element in different directions;

[0114] After obtaining the first steering vector matrix v z (φ) and v x (α) of each array element in different directions, the second steering vector matrix v a11 (α, φ) corresponding to the one-dimensional echo data S1 can be obtained according to the first steering vector matrix, which can be represented as:

[0115] v all (α, φ) = [kron(v z (φ), v x (α))] MN×1 (11)

[0116] In the above formula (11), kron() can represent the Kronecker product operation, and the operation process thereof will not be described in detail.

[0117] Step S311, using the orthogonal matching pursuit (OMP) algorithm to reconstruct a sparse vector according to the second steering vector matrix, the intermediate angle set and the pitch angle set;

[0118] Step S312, matching each two-dimensional estimated angle in the two-dimensional estimated angle set according to the position of the non-zero element in the reconstructed sparse vector to obtain the target intermediate angle and the target pitch angle corresponding to different targets respectively;

[0119] In the process of constructing the compressed sensing model, for K targets, K estimated values ​​of the intermediate angle α are obtained respectively according to the above method, that is, A={α1,α2,……,α K ,} K×1 And K estimated values ​​of the pitch angle φ, i.e., Φ = {φ1, φ2, ..., φ K} K×1 Then, the K value of the two-dimensional angle (α, φ) where the target may be located can be obtained. 2 This involves combining multiple intermediate angles from the intermediate angle set with multiple elevation angles from the elevation angle set to obtain a two-dimensional estimated angle set χ for different targets.

[0120] χ={(α1,φ1),(α2,φ1),…,(α K ,φ1),(α1,φ2),(α2,φ2),…(α K ,φ2),…,(α1,φ K ),(α2,φ K ),…,(α K ,φ K (12)

[0121] Based on the aforementioned second guiding vector array v a11 (α,φ), obtain the guiding matrix V corresponding to the two-dimensional estimated angle set χ:

[0122]

[0123] As can be seen, the two-dimensional estimated angle set χ described above can serve as a complete redundant dictionary for the OMP operation. Assume the observed signal is denoted as S1, the sparse vector is denoted as x, and the sparsity of x can be K, and its size can be K. 2 ×1, therefore, based on the above two-dimensional estimated angle set χ and the second guiding vector array v a11 The formulas corresponding to (α,φ) are used to construct a compressed sensing model: S1=Vx, that is, the compressed sensing method is used to construct the echo signal model of a two-dimensional rectangular array, so as to obtain the angle of the target by solving the values ​​of the variables in the compressed sensing. The construction and implementation process of the compressed sensing model is not described in detail in this application.

[0124] Then, the OMP algorithm can be used to reconstruct the sparse vector x in the compressed sensing model. Based on the position of the non-zero element in the sparse vector x, indexing is performed in the two-dimensional estimated angle set χ, that is, the target mid-angle and target elevation angle corresponding to the steering vector at that position are obtained, thus completing the matching of the mid-angle and elevation angle of the same target. At this time, the signal amplitude of the corresponding target can be determined based on the value of the non-zero element in the reconstructed sparse vector x, such as determining the value of the non-zero element as the amplitude of the corresponding target.

[0125] Step S313, obtaining the target azimuth angle of the target according to the target intermediate angle and the target pitch angle of the same target.

[0126] In combination with the above manner, the target intermediate angle α k and the target pitch angle φ k of the same target are obtained. k Then, according to the trigonometric function relationship among the azimuth angle, the pitch angle and the intermediate angle of the same incident signal, such as sinα k = sinθ k cosφ k , the target intermediate angle and the target pitch angle are operated to obtain the target azimuth angle θ .

[0127] The two-dimensional DOA estimation method based on MUSIC and OMP described in the above embodiments is simulated and compared with the two-dimensional DOA estimation method based on the two-dimensional smoothing MUSIC algorithm under the simulation parameters as shown in the following table. Figure 4 Since the two-dimensional DOA estimation method based on the two-dimensional smoothing MUSIC algorithm can only be applied to a uniform rectangular array, the simulation comparison can be performed on the uniform rectangular array, the relative coordinates of each array element in the x-axis direction of the uniform rectangular array can be represented as X = [0, d x , 2d x , …, (M-1)d x ], and the relative coordinates of each array element in the z-axis direction of the uniform rectangular array can be represented as Z = [0, d z , 2d z , …, (N-1)d z ], where d x represents the fixed interval between adjacent array elements in the x-axis direction, and d z represents the fixed interval between adjacent array elements in the z-axis direction. For example, assuming that M = N = 20, d x = d z = 0.5λ, under 5000 times of Monte Carlo simulation, the root mean squares error (RMSE) of the two two-dimensional DOA estimation methods under different signal noise ratios (SNR) can be compared, thereby realizing the effectiveness evaluation of the two-dimensional DOA estimation method based on MUSIC and OMP proposed in the present application, where the RMSE evaluation formula is as follows:

[0128]

[0129] In the above formula (14), N mc represents the number of Monte Carlo simulations, and N mcK can represent the target number, and in the simulation, K=3 is taken as an example, and the angle of the target can be randomly generated in each Monte Carlo simulation.

[0130] Thus, for a given SNR, the RMSE curves of the two-dimensional DOA estimation method based on MUSIC and OMP proposed in the application and the traditional two-dimensional DOA estimation method based on two-dimensional smoothing MUSIC algorithm are as shown in FIG. 6. Under 5000 times of Monte Carlo simulation, the average time consumed by the two two-dimensional DOA estimation methods is 0.0041s and 0.7320s respectively. It can be seen that the two-dimensional DOA estimation method based on MUSIC and OMP proposed in the application greatly reduces the running time of the two-dimensional OMP algorithm and improves the operation efficiency. Figure 5

[0131] Refer to Figure 6 , the structure schematic diagram of an optional example of the two-dimensional DOA estimation device based on MUSIC and OMP proposed in the application, which can include:

[0132] The echo data determination module 61 is configured to determine the echo data of a single snapshot received by each array element in the rectangular array;

[0133] The one-dimensional MUSIC operation module 62 is configured to perform one-dimensional multiple signal classification (MUSIC) operation on the echo data received by the array elements in the same direction to obtain the intermediate angle set and the elevation angle set of different targets;

[0134] The snapshot data used in the one-dimensional MUSIC operation is determined according to the structure of the array elements in another direction, and the intermediate angle of any target is associated with the elevation angle and the azimuth angle of the target;

[0135] The matching processing module 63 is configured to perform matching processing on the multiple intermediate angles in the intermediate angle set and the multiple elevation angles in the elevation angle set according to the orthogonal matching pursuit (OMP) algorithm to determine the target intermediate angle and the target elevation angle corresponding to each of the different targets;

[0136] The target azimuth angle obtaining module 64 is configured to obtain the target azimuth angle corresponding to the target according to the target intermediate angle and the target elevation angle.

[0137] Optionally, the echo data determination module 61 can include:

[0138] The parameter obtaining unit is configured to obtain the relative coordinates of each array element in the rectangular array in different directions and the signal wavelength of each target incident on the rectangular array;

[0139] ​an echo modeling unit, configured to model, according to at least the relative coordinates and the signal wavelength, to obtain an echo signal model;

[0140] an echo data obtaining unit, configured to obtain, by using the echo signal model, echo data of a single snapshot received by each array element in the rectangular array.

[0141] Optionally, the one-dimensional MUSIC operation module 62 can include:

[0142] an array echo matrix obtaining unit, configured to arrange the echo data received by each array element according to a spatial order of the rectangular array, to obtain an array echo matrix;

[0143] a first data covariance matrix obtaining unit, configured to obtain, according to the array echo matrix and a first dimension number of the elevation-azimuth direction array element, a first data covariance matrix for the azimuth direction array element;

[0144] a second data covariance matrix obtaining unit, configured to obtain, according to the array echo matrix and a second dimension number of the azimuth direction array element, a second data covariance matrix for the elevation direction array element;

[0145] a first steering vector matrix obtaining unit, configured to obtain a first steering vector matrix of each array element in different directions;

[0146] a spectrum peak searching unit, configured to perform power spectrum peak searching according to the first data covariance matrix and the second data covariance matrix, and the first steering vector matrix in the corresponding direction, to obtain an intermediate angle set and an elevation angle set corresponding to each target in different directions.

[0147] Optionally, the spectrum peak searching unit can include:

[0148] an array noise eigenvector obtaining unit, configured to perform eigen decomposition on the first data covariance matrix and the second data covariance matrix respectively, and obtain array noise eigenvectors corresponding to different directions according to the number of targets;

[0149] a maximum value obtaining unit, configured to perform peak searching according to the array noise eigenvectors and the first steering vector matrix in the same direction, to obtain the same number of maximum values as the number of targets in the direction;

[0150] an angle set constituting unit, configured to constitute an intermediate angle set and an elevation angle set of different targets by using angles corresponding to the maximum values in the same direction.

[0151] Optionally, the matching processing module 63 can include:

[0152] vectorization processing unit, configured to perform vectorization processing on the two-dimensional echo data received by the rectangular array to obtain corresponding one-dimensional echo data;

[0153] a second steering vector matrix obtaining unit, configured to obtain a second steering vector matrix corresponding to the one-dimensional echo data according to the first steering vector matrix of each array element in different directions;

[0154] a sparse vector reconstruction unit, configured to perform sparse vector reconstruction by using an orthogonal matching pursuit (OMP) algorithm according to the second steering vector matrix, the intermediate angle set and the elevation angle set;

[0155] a matching unit, configured to match each two-dimensional estimated angle in the two-dimensional estimated angle set according to the position of a non-zero element in the reconstructed sparse vector, to obtain a target intermediate angle and a target elevation angle corresponding to each target.

[0156] Optionally, the sparse vector reconstruction unit can include:

[0157] a two-dimensional estimated angle set obtaining unit, configured to combine a plurality of intermediate angles in the intermediate angle set and a plurality of elevation angles in the elevation angle set to obtain a two-dimensional estimated angle set of different targets;

[0158] a steering matrix obtaining unit, configured to obtain a steering matrix corresponding to the two-dimensional estimated angle set according to the second steering vector matrix;

[0159] a compressed sensing model constructing unit, configured to construct a compressed sensing model according to the steering array and the sparse vector;

[0160] a reconstruction unit, configured to reconstruct the sparse vector in the compressed sensing model by using an orthogonal matching pursuit (OMP) algorithm.

[0161] In some other embodiments, the apparatus can further include:

[0162] a signal amplitude determining module, configured to determine a signal amplitude corresponding to the target according to the value of the non-zero element in the reconstructed sparse vector.

[0163] It should be noted that the various modules and units in the above apparatus embodiments can be stored in the memory as program modules, and the processor executes the above program modules stored in the memory to realize the corresponding functions. The functions realized by the various program modules and their combinations, and the achieved technical effects can be referred to the descriptions of the corresponding parts of the above method embodiments, and the present embodiment will not be described herein.

[0164] The application further provides a computer readable storage medium, which can store a computer program that can be invoked and loaded by a processor to implement the steps of the two-dimensional DOA estimation based on MUSIC and OMP described in the above embodiments.

[0165] With reference to Figure 7 For an optional example of the hardware structure of the computer device suitable for the two-dimensional DOA estimation based on MUSIC and OMP proposed in the application, as shown in Figure 7 The computer device can include at least one memory 71, at least one processor 72 and at least one communication module 73, wherein:

[0166] The memory 71 can be used to store the program of the two-dimensional DOA estimation based on MUSIC and OMP described in the above method embodiments; the processor 72 can load and execute the program stored in the memory to implement the steps of the two-dimensional DOA estimation based on MUSIC and OMP described in the above method embodiments. The device type of the memory 71 and the processor 72 is not limited in the application and can be determined as appropriate.

[0167] The above communication module 73 can include but is not limited to a communication module for realizing data interaction through a wireless communication network, such as a WIFI module, a 5G / 6G (fifth generation mobile communication network / sixth generation mobile communication network) module, a GPRS module, an antenna, etc., to realize data communication between the computer device and other devices.

[0168] It should be understood that the structure of the computer device described in the above embodiments does not constitute a limitation on the computer device in the embodiments of the application, and in actual application, the computer device can include more components than those described above, or some components can be combined, such as a power supply, various sensor modules, etc., which are not enumerated one by one in the application.

[0169] Finally, it should be noted that, in the above embodiments, the relationship terms such as first, second, etc. are only used to distinguish one operation, unit or module from another operation, unit or module, and do not necessarily require or imply any actual relationship or sequence between the units, operations or modules. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method or system including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method or system. Without more limitations, the element defined by the statement "including a" does not exclude the presence of another identical element in the process, method or system including the element.

[0170] The various embodiments described in this specification are presented by way of example, and each embodiment is presented for the purpose of conveying the novelty and inventive aspects of the present application. Each embodiment is presented in a progressive or alternative manner, and each embodiment emphasizes different aspects of the application. The same or similar parts and principles expressed in each embodiment can be combined with each other, and the same or similar parts and principles expressed in each embodiment can be applied to other embodiments. For the devices and computer equipment disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, they are described more simply, and the relevant parts are described in the method part.

[0171] The above description of disclosed embodiments provides enabling disclosure sufficient for one of ordinary skill in the art to practice or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A two-dimensional DOA estimation method based on MUSIC and OMP, characterized in that, The method includes: Determine the echo data of each snapshot received by each element in the rectangular array; A one-dimensional multi-signal classification (MUSIC) operation is performed on the echo data received by array elements in the same direction to obtain the intermediate angle set and elevation angle set of different targets; wherein, the snapshot data used in the one-dimensional MUSIC operation is determined based on the structure of another array element, and the intermediate angle of any target is associated with the elevation angle and azimuth angle of that target. Based on the Orthogonal Matching Pursuit (OMP) algorithm, multiple intermediate angles in the intermediate angle set and multiple elevation angles in the elevation angle set are matched to determine the target intermediate angle and target elevation angle corresponding to each different target. Based on the target's median angle and elevation angle, the target's azimuth angle is obtained. The step of performing a one-dimensional multi-signal classification (MUSIC) operation on the echo data received by array elements in the same direction to obtain the intermediate angle set and elevation angle set of different targets includes: The echo data received by each array element are arranged according to the spatial order of the rectangular array to obtain the array echo matrix; Based on the array echo matrix and the first dimension of the elevation direction array elements, the first data covariance matrix for the azimuth direction array elements is obtained; Based on the array echo matrix and the second dimension of the azimuth array element, a second data covariance matrix for the elevation array element is obtained. Obtain the first guiding vector matrix of each array element in different directions; Based on the first data covariance matrix and the second data covariance matrix, as well as the first steering vector matrix in the corresponding direction, a power spectrum peak search is performed to obtain the intermediate angle set and elevation angle set corresponding to each target in different directions; The process of matching multiple intermediate angles in the intermediate angle set and multiple elevation angles in the elevation angle set according to the Orthogonal Matching Pursuit (OMP) algorithm to determine the target intermediate angle and target elevation angle corresponding to each different target includes: The two-dimensional echo data received by the rectangular array is vectorized to obtain the corresponding one-dimensional echo data; Based on the first steering vector matrix of each array element in different directions, the second steering vector matrix corresponding to the one-dimensional echo data is obtained; Based on the second steering vector matrix, the intermediate angle set, and the elevation angle set, sparse vector reconstruction is performed using the Orthogonal Matching Pursuit (OMP) algorithm; Based on the positions of the non-zero elements in the reconstructed sparse vector, each two-dimensional estimated angle in the two-dimensional estimated angle set is matched to obtain the target intermediate angle and target elevation angle corresponding to each of the different targets.

2. The method according to claim 1, characterized in that, The step of performing a power spectrum peak search based on the first data covariance matrix and the second data covariance matrix, as well as the first steering vector in the corresponding direction, to obtain the intermediate angle set and elevation angle set corresponding to each target in different directions includes: The first data covariance matrix and the second data covariance matrix are respectively decomposed into features, and array noise feature vectors corresponding to different directions are obtained according to the number of targets. Peak search is performed based on the array noise feature vector and the first steering vector array in the same direction to obtain the maximum value in that direction that is the same as the number of targets; Using the angles corresponding to the maximum values ​​in the same direction, we can construct intermediate angle sets and elevation angle sets for different targets.

3. The method according to claim 1, characterized in that, The step of reconstructing sparse vectors using the Orthogonal Matching Pursuit (OMP) algorithm based on the second steering vector matrix, the intermediate angle set, and the elevation angle set includes: By combining multiple intermediate angles from the intermediate angle set with multiple elevation angles from the elevation angle set, a two-dimensional estimated angle set for different targets is obtained; Based on the second guiding vector matrix, the guiding matrix corresponding to the two-dimensional estimated angle set is obtained; Based on the aforementioned guiding array and sparse vectors, a compressed sensing model is constructed; The sparse vectors in the compressed sensing model are reconstructed using the orthogonal matching pursuit (OMP) algorithm.

4. The method according to claim 1, characterized in that, The method further includes: The signal amplitude corresponding to the target is determined based on the values ​​of the non-zero elements in the reconstructed sparse vector.

5. The method according to any one of claims 1-4, characterized in that, The determination of the echo data of each snapshot received by each element in the rectangular array includes: The relative coordinates of each element in the rectangular array in different directions, as well as the signal wavelength of each target incident on the rectangular array, are obtained. The echo signal model is obtained by modeling based at least on the relative coordinates and the signal wavelength. Using the echo signal model, the echo data of a single snapshot received by each element in the rectangular array is obtained.

6. A two-dimensional DOA estimation device based on MUSIC and OMP, characterized in that, The device includes: The echo data determination module is used to determine the echo data of each snapshot received by each array element in the rectangular array; A one-dimensional MUSIC operation module is used to perform one-dimensional multi-signal classification MUSIC operation on the echo data received by array elements in the same direction to obtain the intermediate angle set and elevation angle set of different targets; wherein, the snapshot data used in the one-dimensional MUSIC operation is determined based on the structure of another array element, and the intermediate angle of any target is associated with the elevation angle and azimuth angle of the target. The matching processing module is used to perform matching processing on multiple intermediate angles in the intermediate angle set and multiple elevation angles in the elevation angle set according to the orthogonal matching pursuit (OMP) algorithm, and determine the target intermediate angle and target elevation angle corresponding to each of the different targets. The target azimuth angle acquisition module is used to obtain the target azimuth angle corresponding to the target based on the target intermediate angle and the target elevation angle; The one-dimensional MUSIC operation module includes: An array echo matrix obtaining unit is used to arrange the echo data received by each array element according to the spatial order of the rectangular array to obtain array echo matrices with different structures. The first data covariance matrix obtaining unit is used to obtain the first data covariance matrix for the azimuth direction array element based on the array echo matrix corresponding to the azimuth direction and the first dimension of the elevation direction array element. The second data covariance matrix acquisition unit is used to obtain the second data covariance matrix for the elevation direction array element based on the array echo matrix corresponding to the elevation direction and the second dimension of the azimuth direction array element. The first guiding vector matrix obtaining unit is used to obtain the first guiding vector matrix of each array element in different directions; The peak search unit is used to perform a power spectrum peak search based on the first data covariance matrix and the second data covariance matrix, as well as the first steering vector matrix in the corresponding direction, to obtain the intermediate angle set and elevation angle set corresponding to each target in different directions. The matching processing module includes: A vectorization processing unit is used to perform vectorization processing on the two-dimensional echo data received by the rectangular array to obtain the corresponding one-dimensional echo data. The second steering vector matrix obtaining unit is used to obtain the second steering vector matrix corresponding to the one-dimensional echo data based on the first steering vector matrix of each array element in different directions. The sparse vector reconstruction unit is used to perform sparse vector reconstruction based on the second steering vector matrix, the intermediate angle set, and the elevation angle set using the orthogonal matching pursuit (OMP) algorithm. The matching unit is used to match each two-dimensional estimated angle in the two-dimensional estimated angle set according to the position of the non-zero element in the reconstructed sparse vector, so as to obtain the target intermediate angle and target elevation angle corresponding to the different targets.

7. A computer-readable storage medium, characterized in that, It stores a computer program, which is loaded and executed by a processor to implement the two-dimensional DOA estimation method based on MUSIC and OMP as described in any one of claims 1-5.

Images