Broadband source weighted direct positioning method based on subspace focusing
By decomposing J-point FFT and covariance matrix at each observation station, building a focus matrix and calculating weight factors, combining the cost function search of grid-based positioning areas, high-precision positioning of broadband radiation sources is achieved, and the existing solutions are solved The problem of limitations of narrowband signal processing in the technology significantly improves positioning accuracy and robustness.
Patent Information
- Application Number
- CN202410984519.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-12
AI Technical Summary
The existing direct positioning algorithm is mainly designed for narrowband signals, and cannot effectively process complex broadband signals, resulting in low positioning accuracy and poor robustness, making it difficult to adapt to the complexity of modern electromagnetic environments.
Using a broadband source weighted direct positioning method based on subspace focus, the J-point FFT, covariance matrix eigenvalue decomposition, focus matrix construction and weight factor calculation are performed in each observation station, combined with the cost function search of the grid positioning area. Achieve high-precision positioning of broadband radiation sources.
The positioning accuracy of broadband radiation sources is significantly improved, and compared with the traditional method, the positioning accuracy and robustness are significantly improved when the calculation complexity is not increased.
Smart Images

Figure CN118938122A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of wireless positioning, and in particular to a broadband source weighted direct positioning method based on subspace focusing. Background Art
[0002] The radiation source positioning technology based on antenna arrays has important applications in many fields such as radar, Internet of Vehicles, Internet of Things, etc. The traditional two-step positioning method divides the positioning process into two independent steps, resulting in information loss, so the positioning accuracy is low and the robustness is poor. Compared with the two-step positioning method, the direct positioning technology directly estimates the radiation source position based on the received data in a single step, which significantly improves the positioning accuracy and robustness.
[0003] Existing direct positioning algorithms are generally designed for narrowband sources. However, with the development of signal processing technology, the electromagnetic environment in space has become more and more complex, and the frequency distribution range of signals has gradually expanded. The limitations of narrowband array signal processing have become increasingly obvious, making it unsuitable for modern environments. Compared with narrowband signals, broadband signals contain more information, have stronger anti-interference capabilities and higher resolution. Therefore, studying broadband source positioning methods has important theoretical significance and practical application value. Summary of the invention
[0004] The purpose of the present invention is to propose a broadband source weighted direct positioning method based on subspace focusing, which can directly locate broadband radiation sources with high precision, improve the positioning accuracy, and is easy to apply in practice.
[0005] In order to achieve the above technical objectives, the technical solution adopted by the present invention is:
[0006] A broadband source weighted direct positioning method based on subspace focusing, the broadband source weighted direct positioning method comprising the following steps:
[0007] Step 1: At each distributed observation station, perform J-point FFT on its uniform linear array received data;
[0008] Step 2: At each observation station, calculate the covariance matrix corresponding to each frequency point and perform eigenvalue decomposition;
[0009] Step 3: At each observation station, construct a focusing matrix to obtain the final covariance matrix;
[0010] Step 4: At each observation station, perform eigenvalue decomposition on the final covariance matrix to calculate the weight factor;
[0011] Step 5: At the central station, grid the positioning area and calculate the cost function value of each grid point;
[0012] Step 6: Find the K peaks with the largest cost function and use their corresponding grid points as the positioning results.
[0013] Step 1 further comprises:
[0014] Based on L observation stations with precisely known positions, each equipped with an M-element uniform linear array, K far-field radiation sources are located; the positions of the observation stations are u l =[x l ,y l ], l = 1, 2, ..., L, the positions of the far-field radiation sources are p k =[x k ,y k ], k = 1, 2, ..., K;
[0015] Assume that the radiation source emits uncorrelated broadband stationary signals, the starting frequency and cutoff frequency of the signal are f0 and f1 respectively, and the received signal x of the lth uniform linear array is l (t) is expressed as:
[0016]
[0017] Among them, ρ l,k is the unknown channel attenuation, n l (t) is additive noise, s k (t) represents the signal emitted by the kth source, α l (p k ) represents the array steering vector;
[0018] By performing FFT of the received signal at point J, the frequency domain model x of the received signal is obtained. l (f j ):
[0019]
[0020] According to the general model of uniform linear array, the array steering vector α of the base station is obtained l (f j , p k ) is as follows:
[0021]
[0022]
[0023] where τ l (p k ) is the sine value of the angle between the line connecting the lth base station and the kth information source and the normal line, d is the array element spacing, and c is the speed of light, i.e., the propagation speed of electromagnetic waves in space; then:
[0024] X l(f j )=A l (f j )Φ l S(f j )+N l (f j ), j = 1, 2, ..., J;
[0025] Among them A l (f j )=[α l (f j ,p1),…,α l (f j , p K )] T is in f j The steering vector of the array at the frequency, Φ l =diag(ρ l,1 ,…,ρ l,K ) is composed of the channel attenuation of K signal sources, and the frequency domain model S(f j )=[s1(f j ), …s K (f j )] T , J is the number of frequency points N l (f j ) is the additive white Gaussian noise in the j-th frequency signal received by the th base station.
[0026] Step 2 further includes:
[0027] The sampling covariance matrix composed of N snapshots of sampled data is used to replace the actual covariance matrix. The sampling covariance matrix of the received signal is It is expressed as:
[0028]
[0029] Perform eigenvalue decomposition on the sampling covariance matrix to obtain:
[0030]
[0031] Among them, U l is a matrix composed of eigenvectors, ∑ l is the eigenvalue λ i The diagonal matrix composed of i is the covariance matrix The i-th eigenvalue of , λ1≥λ2≥...≥λ K ≥λ K+1 =λ K+2 =... =λ M =σ 2, K is the number of sources, and u i is the corresponding i eigenvector; L is the total number of observation stations, M is the total number of uniform linear array elements, σ 2 is the variance of additive Gaussian white noise; the signal subspace U is obtained l,s (f j ) and the noise subspace U l,n (f j )as follows:
[0032] U l,s (f j )=[u1,…,u K ];
[0033] U l,n (f j )=[u K+1 ,…,u ML ].
[0034] Step 3 further includes:
[0035] Construct a focusing matrix, which converts several narrowband signals of different frequencies obtained after FFT of the broadband signal to the same assumed reference frequency f0; the focusing matrix T l (f j ) is selected to satisfy the following formula:
[0036] U l,s (f0) = T l (f j ) l,s (f j );
[0037] Where U l,s (f0) and U l,s (f j ) are the final signal subspace and f j Signal subspace in frequency;
[0038] Use the F-norm to minimize the focus error:
[0039]
[0040] Focusing Matrix T l (f j ) is a Hermitian matrix, and the constraints are:
[0041]
[0042] The signal subspace corresponding to all j frequency points is obtained, and the corresponding focusing matrix is obtained. The covariance matrix corresponding to j frequency points is obtained by the following formula:
[0043]
[0044] The covariance matrices corresponding to the j frequency points are averaged to obtain the final covariance matrix of the lth receiving array:
[0045]
[0046] Step 4 further includes:
[0047] The final covariance matrix obtained by focusing is decomposed by eigenvalue to obtain:
[0048]
[0049] Get the final noise subspace U of each array l,n (f0) is as follows:
[0050] U l,n (f0)=[u l,K+1 ,…,u l,M ];
[0051] Calculate the noise power of the lth receiving array separately and signal power
[0052]
[0053]
[0054] Calculate the weight factor w based on the calculated noise power and signal power l :
[0055] Step 5 further includes:
[0056] Divide the area where the target source may be into several grids, substitute the p corresponding to each grid point into the cost function, and calculate the corresponding f(p); the cost function is:
[0057]
[0058] Where l = 1, 2, ..., L, L is the total number of observation stations, α l (f0, p) is the final array steering vector of the base station; U l,n (f0) is the final noise subspace after focusing of each array.
[0059] Compared with the prior art, the present invention has the following beneficial effects:
[0060] The broadband source weighted direct positioning method based on subspace focusing of the present invention, in the scenario of positioning broadband radiation sources, compared with the traditional ISM-DPD method and FSS-DPD method, significantly improves the positioning accuracy while ensuring that the calculation complexity is not increased. BRIEF DESCRIPTION OF THE DRAWINGS
[0061] Figure 1 It is a flow chart of the broadband source weighted direct positioning method based on subspace focusing of the present invention;
[0062] Figure 2 Positioning scene graph for distributed uniform linear array;
[0063] Figure 3 It is a schematic diagram of the two-dimensional spatial spectrum of the method of the present invention;
[0064] Figure 4 A scatter plot of the results of locating multiple radiation sources by the method of the present invention;
[0065] Figure 5 Schematic diagram of root mean square error performance of positioning results of the present invention and the traditional positioning method under different signal-to-noise ratios;
[0066] Figure 6 Schematic diagram of root mean square error performance of positioning results of the present invention and the traditional positioning method under different snapshot numbers. DETAILED DESCRIPTION
[0067] The embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings.
[0068] Symbols: In this paper, bold uppercase letters, bold lowercase letters, and italic letters, such as A, a, and a, represent matrices, vectors, and scalars, respectively. T , (·) H and(·) -1 denote the matrix transpose, conjugate transpose and inverse operations, E(·), ||·|| F , tr(·), Re(·), diag(·) represent the expectation, F-norm, trace, taking the real part, and converting the vector to a diagonal matrix, respectively, with triangular superscript letters represents the estimated value of R.
[0069] The present invention provides a broadband source weighted direct positioning method based on subspace focusing, such as Figure 1 As shown, the specific steps include:
[0070] Step 1: At each distributed observation station, perform J-point FFT on its uniform linear array received data.
[0071] Consider the location area as follows Figure 2Based on L observation stations with precisely known positions, each equipped with an M-element uniform linear array, their positions are u l =[x l ,y l ], (l = 1, 2, ..., L), locate K far-field radiation sources, their positions are p k =[x k ,y k ], (k = 1, 2, ..., K). Assume that the radiation source emits uncorrelated broadband stationary signals, and the starting frequency and cutoff frequency of the signal are f0 and f1 respectively. The received signal x of the lth uniform linear array l (t) can be expressed as:
[0072]
[0073] where ρ l,k is the unknown channel attenuation, n l (t) is additive noise, which is generally assumed to be additive white noise that obeys Gaussian distribution. k (t) represents the signal emitted by the kth source, α l (p k ) represents the array steering vector.
[0074] Since the array steering vector is related to the frequency of the signal, the frequency is a constant in narrowband signal processing, so the array steering vector is only related to the array and the source position. When the source transmits a broadband signal, the array steering vector is also related to the frequency of the signal at different frequency points, so the frequency domain representation is used for subsequent signal processing. By performing J-point FFT on the received signal, the frequency domain model of the received signal is obtained.
[0075]
[0076] According to the general model of uniform linear array, the array steering vector expression of the base station can be obtained as follows:
[0077]
[0078]
[0079] where τ l (p k ) is the sine value of the angle between the line connecting the first base station and the kth information source and the normal line; then:
[0080] X l (f j )=A l (f j )Φ l S(f j )+Nl (f j ), j = 1, 2, ..., J;
[0081] Among them A l (f j )=[α l (f j ,p1),…,α l (f j , p K )] T is in f j The steering vector of the array at the frequency, Φ l =diag(ρ l,1 ,…,ρ l,K ) consists of the channel attenuation of K signal sources, S(f j )=[s1(f j ), …s K (f j )] T The above formula is the frequency domain model of the broadband signal output by the first base station. It can be seen that it is similar to the composition of j narrowband signals.
[0082] Step 2: At each observation station, calculate the covariance matrix corresponding to each frequency point and perform eigenvalue decomposition.
[0083] Since the sampling length of the actual received signal is limited, in practice, a sampling covariance matrix composed of N snapshots of sampled data is used to replace the actual covariance matrix, and the sampling covariance matrix of the received signal is expressed as:
[0084]
[0085] We perform eigenvalue decomposition and obtain:
[0086]
[0087] Among them, U l is a matrix composed of eigenvectors, ∑ l is the eigenvalue λ i The diagonal matrix composed of i is the covariance matrix The i-th eigenvalue of . Note that λ1≥λ2≥...≥λ K ≥λ K+1 =λ K+2 =... =λ M =σ 2 , K is the number of sources, and u i is the corresponding i The eigenvector of . The signal subspace and noise subspace can be obtained as follows:
[0088]
[0089] Step 3: At each observation station, construct a focusing matrix to obtain the final covariance matrix.
[0090] After FFT, a broadband signal will get many narrowband signals of different frequencies. The focusing subspace converts these narrowband signals of different frequencies to the same assumed reference frequency f0 by constructing a focusing matrix, and then averages the power spectrum density of all the focused signals, and then the narrowband signal processing algorithm can be used for processing. The selection of the focusing matrix T needs to satisfy the following formula:
[0091] U l,s (f0) = T l (f j ) l,s (f j );
[0092] We use the F-norm to minimize the focusing error, namely:
[0093]
[0094] We note that the focusing matrix T l (f j ) is a Hermitian matrix, so the constraints can be taken as:
[0095]
[0096] To construct the focusing matrix, we derive as follows:
[0097]
[0098] The traces of the first two matrices in the above polynomial are both constants. In order to minimize the cost function, we need the third term, that is, to maximize the third term. Let the matrix C be expressed as follows:
[0099]
[0100] Performing singular value decomposition on C yields:
[0101] C=UΣV H ;
[0102] Where ∑=diag{σ 11 , σ 22 , …, σ MM}, assuming that the rank of C is r, then:
[0103] σ 11 ≥σ 22 ≥…σ rr ≥σr+1,r+1 =…=σ MM =0;
[0104] So the last term can be written as:
[0105]
[0106] make Since Z is the product of three unitary matrices, Z is also a unitary matrix, so |Z ii |≤1, i=1, 2..., M, the above formula can be simplified as follows
[0107]
[0108] If and only if |Z ii When |=1, the above equality holds, and equation (17) takes the maximum value, so the solution of the focusing matrix can be considered as follows:
[0109] T l (f j )=UV H ;
[0110] Therefore, we only need to obtain the signal subspace corresponding to all j frequency points to get the corresponding focusing matrix. The covariance matrix corresponding to j frequency points can be obtained by the following formula:
[0111]
[0112] As shown in the following formula, the covariance matrix corresponding to these j frequency points is averaged to obtain the final covariance matrix of the lth receiving array:
[0113]
[0114] Step 4: At each observation station, perform eigenvalue decomposition on the final covariance matrix to calculate the weight factor;
[0115] Performing eigenvalue decomposition on the final covariance matrix obtained by focusing can be obtained as follows
[0116]
[0117] The final noise subspace of each array is obtained as follows
[0118] U l,n (f0)=[u l,K+1 ,…,u l,M ];
[0119] Calculate the noise power and signal power of the lth receiving array according to the following formula:
[0120] Noise power:
[0121] Signal Power:
[0122] Finally, the weight factor is calculated based on the calculated noise power and signal power.
[0123] Step 5: At the central station, grid the positioning area and calculate the cost function value of each grid point; at the same time, the cost function for searching the radiation source position is
[0124]
[0125] The area where the target source may be located, generally a two-dimensional space, is divided into several grids, and then the p corresponding to each grid point is substituted into the above cost equation to calculate the corresponding f(p).
[0126] Step 6: Find the K peaks with the largest cost function, and the corresponding grid points are the positioning results;
[0127] like Figure 3 The figure shows the two-dimensional spatial spectrum obtained by calculating the cost function in the possible area by the method of the present invention, where the radiation sources are located at (23, 52) m and (68, 38) m. It can be seen that the two-dimensional spatial spectrum has two maximum peaks at the positions corresponding to the two radiation sources, which means that the radiation source can be located by finding the K largest peak positions of the two-dimensional spatial spectrum.
[0128] Figure 4 It is a scatter plot of the positioning results of the method of the invention, where two radiation sources are considered to be located at p1 = [23m, 52m]T and p2 = [68m, 38m] respectively. T q observation stations are located at u1 = [0m, 0m] T 、u2=[0m,100m] T 、u3=[100m,0m] T and u4=[100m,100m] T . Take the number of snapshots J = 500, the number of array elements M = 8, and the signal-to-noise ratio as 10dB. Figure 4 It can be seen that the method of the present invention can locate the radiation source with high precision.
[0129] The performance estimation standard of the present invention is the root mean square error (RMSE), and the RMSE of radiation source positioning and mutual coupling coefficient estimation are respectively defined as:
[0130]
[0131] in, It represents the estimated result of the kth radiation source position in the mcth test, MC represents the number of Monte Carlo simulations, and the number of Monte Carlo simulations in this section is 500.
[0132] Figure 5 The performance curve of the root mean square error of the positioning result obtained by the method of the present invention and the signal-to-noise ratio is shown in FIG. Figure 4 Same. From Figure 5 It can be seen that the present invention achieves higher positioning accuracy.
[0133] Figure 6 The performance curve of the root mean square error of the positioning result obtained by the method of the present invention and the signal-to-noise ratio is shown in FIG. Figure 4 Same. From Figure 6 It can be seen that the present invention achieves higher mutual coupling estimation accuracy.
[0134] Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can adopt the form of complete hardware embodiments, complete software embodiments, or embodiments in combination with software and hardware. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code. The scheme in the embodiments of the present application can be implemented in various computer languages, for example, object-oriented programming language Java and literal scripting language JavaScript, etc.
[0135] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0136] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1A function specified in one or more boxes.
[0137] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0138] Although the preferred embodiments of the present application have been described, those skilled in the art may make other changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.
[0139] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.
Claims
1. A broadband source weighted direct positioning method based on subspace focusing, characterized in that: The broadband source weighted direct positioning method comprises the following steps: Step 1: At each distributed observation station, perform J-point FFT on its uniform linear array received data; Step 2: At each observation station, calculate the covariance matrix corresponding to each frequency point and perform eigenvalue decomposition; Step 3: At each observation station, construct a focusing matrix to obtain the final covariance matrix; Step 4: At each observation station, perform eigenvalue decomposition on the final covariance matrix to calculate the weight factor; Step 5: At the central station, grid the positioning area and calculate the cost function value of each grid point; Step 6: Find the K peaks with the largest cost function and use their corresponding grid points as the positioning results.
2. The broadband source weighted direct positioning method based on subspace focusing according to claim 1 is characterized in that: Step 1 further comprises: Based on L observation stations with precisely known positions, each equipped with an M-element uniform linear array, K far-field radiation sources are located; the positions of the observation stations are u l =[x l ,y l ],l=1,2,…,L, the positions of far-field radiation sources are p k =[x k ,y k ], k = 1, 2, ..., K; Assume that the radiation source emits uncorrelated broadband stationary signals, the starting frequency and cutoff frequency of the signal are f0 and f1 respectively, and the received signal x of the lth uniform linear array is l (t) is expressed as: Among them, ρ l,k is the unknown channel attenuation, n l (t) is additive noise, s k (t) represents the signal emitted by the kth source, α l (p k ) represents the array steering vector; By performing FFT of the received signal at point J, the frequency domain model x of the received signal is obtained. l (f j ): According to the general model of uniform linear array, the array steering vector α of the base station is obtained l (f j ,p k ) is as follows: where τ l (p k ) is the sine value of the angle between the line connecting the lth base station and the kth information source and the normal line, d is the array element spacing, and c is the speed of light, i.e., the propagation speed of electromagnetic waves in space; then: X l (f j )=A l (f j )Φ l S(f j )+N l (f j ),j=1,2,…,J; Among them A l (f j )=[α l (f j ,p1),…,α l (f j ,p K )] T is in f j The steering vector of the array at the frequency, Φ l =diag(ρ l,1 ,…,ρ l,K ) is composed of the channel attenuation of K signal sources, and the frequency domain model S(f j )=[s1(f j ),…s K (f j )] T , J is the number of frequency points, N l (f j ) is the additive white Gaussian noise in the j-th frequency signal received by the l-th base station.
3. The broadband source weighted direct positioning method based on subspace focusing according to claim 1 is characterized in that: Step 2 further includes: The sampling covariance matrix composed of N snapshots of sampled data is used to replace the actual covariance matrix. The sampling covariance matrix of the received signal is It is expressed as: Perform eigenvalue decomposition on the sampling covariance matrix to obtain: Among them, U l is a matrix of eigenvectors, Σ l is the eigenvalue λ i The diagonal matrix composed of i is the covariance matrix The i-th eigenvalue of , λ1≥λ2≥...≥λ K ≥λ K+1 =λ K+2 =... =λ M =σ 2 , K is the number of sources, and u i is the corresponding i eigenvector; L is the total number of observation stations, M is the total number of uniform linear array elements, σ 2 is the variance of additive Gaussian white noise; the signal subspace U is obtained l,s (f j ) and the noise subspace U l,n (f j )as follows: YOU l,s (f j )=[u1,…,u K ]; IN l,n (f j )=[in K+1 ,…,in ML ]。 4. The broadband source weighted direct positioning method based on subspace focusing according to claim 3 is characterized in that: Step 3 further includes: Construct a focusing matrix, which converts several narrowband signals of different frequencies obtained after FFT of the broadband signal to the same assumed reference frequency f0; the focusing matrix T l (f j ) is selected to satisfy the following formula: U l,s (f0)=T l (f j )U l,s (f j ); Where U l,s (f0) and U l,s (f j ) are the final signal subspace and f j Signal subspace in frequency; Use the F-norm to minimize the focus error: Focusing Matrix T l (f j ) is a Hermitian matrix, and the constraints are: The signal subspace corresponding to all j frequency points is obtained, and the corresponding focusing matrix is obtained. The covariance matrix corresponding to j frequency points is obtained by the following formula: The covariance matrices corresponding to the j frequency points are averaged to obtain the final covariance matrix of the lth receiving array:
5. The broadband source weighted direct positioning method based on subspace focusing according to claim 4 is characterized in that: Step 4 further includes: The final covariance matrix obtained by focusing is decomposed by eigenvalue to obtain: Get the final noise subspace U of each array l,n (f0) is as follows: IN l,n (f0)=[u l,K+1 ,…,in l,M ]; Calculate the noise power of the lth receiving array separately and signal power Calculate the weight factor w based on the calculated noise power and signal power l :
6. The broadband source weighted direct positioning method based on subspace focusing according to claim 1 is characterized in that: Step 5 further includes: Divide the area where the target source may be into several grids, substitute the p corresponding to each grid point into the cost function, and calculate the corresponding f(p); the cost function is: Where l = 1, 2, …, L, L is the total number of observation stations, α l (f0,p) is the final array steering vector of the base station; U l,n (f0) is the final noise subspace after focusing of each array.
Citation Information
Patent Citations
Direct positioning method of distributed uniform linear array based on unknown mutual coupling
CN115685061A
Satellite navigation multi-interference source direct positioning method, system and equipment
CN116774250A
One-dimensional DOA estimation method based on combined signals at specific frequencies
WO2021139208A1
