Basis pursuit denoising sparse reconstruction DOA estimation method based on noise pre-estimation

Abstract

The invention discloses a basis pursuit denoising sparse reconstruction DOA estimation method based on noise pre-estimation, which is applied to a signal receiving device based on a coprime array: a receiving signal on a physical array element of the coprime array is obtained, and the receiving signal is converted to obtain a corresponding first coprime virtual domain receiving signal; a second coprime virtual domain receiving signal is determined according to the first coprime virtual domain receiving signal, and a covariance matrix of the second coprime virtual domain receiving signal is generated; a CAPON spectrum is generated according to the covariance matrix, and an actual noise power is calculated; with the actual noise power as a known quantity, a power spectrum of a search angle is determined by basis pursuit denoising sparse reconstruction; the invention selects the CAPON spectrum to solve the actual noise power size, and makes the noise power a priori information in the sparse reconstruction algorithm, and at the same time considers important parameters such as the angle region range of the incident signal to improve the sparse reconstruction accuracy and enhance the robustness of the algorithm.

Description

Technical Field

[0001] The invention belongs to the technical field of target recognition and positioning, and in particular relates to a basis pursuit denoising sparse reconstruction DOA estimation method based on noise pre-estimation. Background Art

[0002] Estimating the direction of the incoming signal received by the array antenna is the basis for target identification and positioning. An important development trend in the current market is to make the equipment as small as possible while integrating it. If the number of array sensors can be reduced while ensuring the resolution, the equipment can be miniaturized. Since uniform linear arrays meet the Nyquist sampling theorem, when it is necessary to identify multiple signal source directions, it can only be done by continuously adding more array elements. Therefore, in practice, the array structure will be huge, resulting in excessive hardware costs and power consumption.

[0003] As a sparse array, the spacing between the elements of the coprime array is no longer limited by the half-wavelength of the incident signal, and a small number of elements can achieve high-resolution DOA estimation. Compared with uniform linear arrays, coprime arrays have the advantages of higher estimation accuracy, stronger spatial resolution, and the ability to process more signal sources. In addition, by constructing a virtual array, the degree of freedom and aperture of the array are enhanced, and the array has high-resolution array performance.

[0004] In fact, since the structure of coprime array was proposed in 2011, many coprime array signal processing methods have emerged. Since the coprime array is more complex than the uniform linear array structure, and there are some holes in the virtual array, it is a non-uniform continuous array. Therefore, how to reasonably and fully utilize coprime virtual array elements to obtain more accurate DOA estimation is an important research topic in the current array signal processing field, which has certain practical application value.

[0005] Existing coprime array signal processing methods can be divided into two categories: methods based on coprime physical arrays and methods based on coprime virtual arrays. The method based on coprime virtual arrays can obtain a virtual array by jointly processing the differences between physical sensors, which improves the degree of freedom of the array and can effectively realize the underdetermined estimation of the signal source on the virtual array. It is also the main coprime array processing technology. At present, the methods based on coprime virtual arrays mainly include spatial smoothing algorithms based on coprime virtual arrays, DOA estimation methods based on the least absolute shrinkage and selection operator (LASSO: Least Absolute Shrinkage and Selection Operator), and sparse reconstruction DOA estimation methods based on the basis pursuit denoising (BPDN) of the covariance matrix of virtual uniform linear subarrays. However, the parameter estimation performance of these methods is seriously degraded under small snapshots, and the computational complexity is extremely large when the number of signal sources is large. Summary of the invention

[0006] The purpose of the present invention is to provide a basis pursuit denoising and sparse reconstruction DOA estimation method based on noise pre-estimation, so as to estimate the direction of arrival of the signal source under the conditions of less sampling data, coherent source and incoherent source, and improve the estimation accuracy under small snapshot parameters.

[0007] The present invention adopts the following technical solution: a basis pursuit denoising sparse reconstruction DOA estimation method based on noise pre-estimation is applied to a signal receiving device based on a coprime array, comprising the following steps:

[0008] Acquire a received signal on a physical array element of a coprime array, and convert the received signal to obtain a corresponding first coprime virtual domain received signal;

[0009] Determine a second mutually prime virtual domain received signal according to the first mutually prime virtual domain received signal, and generate a covariance matrix of the second mutually prime virtual domain received signal;

[0010] Generate CAPON spectrum according to the covariance matrix and calculate the actual noise power;

[0011] Taking the actual noise power as a known quantity, basis pursuit denoising and sparse reconstruction are used to determine the power spectrum of the search angle, which includes the distance and angle of the target.

[0012] Further, the first mutually prime virtual domain received signal is:

[0013]

[0014] Among them, X D receiving a signal for a first mutually prime virtual domain, is the covariance matrix of the received signal, vec(·) represents vectorization, K is the number of signal sources, is the total power of K signal sources, a(θ k ) is the steering vector, θ k is the direction angle of the incident signal reaching the array, j is the imaginary sign, λ is the wavelength of the incident signal, P is the set of physical sensor locations in the coprime array, a * (θ k ) is the value of a(θ k ) for conjugation treatment, is the noise power, I D =vec(I P ), I P is the unit matrix of (2M+N-1)×(2M+N-1), where M and N are the number of elements of the coprime array, and the symbol represents the Kronecker product.

[0015] Further, determining the second mutually prime virtual domain received signal includes:

[0016] According to the maximum uniform continuous array element position distribution of the coprime array, the second coprime virtual domain received signal is determined; wherein the second coprime virtual domain received signal is:

[0017]

[0018] In the formula, x R is the second coprime virtual domain receiving signal, A R is the steering vector of the received signal in the second coprime virtual domain, b is the power corresponding to each source angle, I R According to I D The obtained identity matrix corresponds to the maximum uniform continuous array element position in the coprime array virtual array.

[0019] Further, generating a covariance matrix of the second coprime virtual domain received signal includes:

[0020] The spatial smoothing algorithm is used to remove the correlation between the second coprime virtual domain received signals to obtain the covariance matrix;

[0021] Among them, the covariance matrix Specifically:

[0022]

[0023] In the formula, x Ri The signal is received for the second coprime virtual domain of the i-th subspace.

[0024] Furthermore, calculating the actual noise power includes:

[0025] According to the angle corresponding to the CAPON spectrum peak, the noise area range is obtained after a rough estimate

[0026] Will As discrete points, the corresponding noise power is obtained by accumulating and averaging.

[0027] According to the noise power Calculate the actual noise power.

[0028] Furthermore, the noise power Specifically:

[0029]

[0030] Where T is the number of sample points, is the steering vector of the array composed of the largest uniform continuous array elements, θ ni for The discrete sample points in .

[0031] Furthermore, according to the noise power Calculating the actual noise power involves:

[0032]

[0033] in, is the actual noise power,

[0034] Furthermore, the power spectrum of the search angle is determined by using basis pursuit denoising and sparse reconstruction, including:

[0035] Constructing the angular area grid according to the incident source sparsity;

[0036] The power spectrum of the search angle is determined according to the objective function.

[0037] Furthermore, the objective function is:

[0038]

[0039] st diag(B′)≥0

[0040] Among them, A′ R is the steering vector for each search angle, η is the regularization parameter, and B′ is the diagonal matrix of the power for each search angle.

[0041] Another technical solution of the present invention: a basis pursuit denoising and sparse reconstruction DOA estimation device based on noise pre-estimation, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned basis pursuit denoising and sparse reconstruction DOA estimation method based on noise pre-estimation is implemented.

[0042] The beneficial effects of the present invention are as follows: the present invention adopts a spatial smoothing method to obtain a full-rank received signal covariance matrix during the preprocessing of the signal, thereby overcoming the defect of a large correlation between signals in a DOA estimation method based on LASSO, and requiring fewer snapshots than a LASSO method; the CAPON spectrum is used to solve the actual noise power size, and the noise power is made the prior information in the sparse reconstruction algorithm, while considering important parameters such as the angle area range of the incident signal to improve the sparse reconstruction accuracy and enhance the robustness of the algorithm; and the BPDN algorithm is used in the sparse reconstruction problem to obtain a source power estimation through fitting between covariance matrices, thereby effectively improving the accuracy of the source estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a schematic diagram of a coprime array structure in an embodiment of the present invention;

[0044] Figure 2 is a schematic diagram of the distribution of mutually prime virtual domains in an embodiment of the present invention;

[0045] Figure 3 It is a graph of DOA estimation results corresponding to simulations using different methods when the number of snapshots is 100 and the signal-to-noise ratio is 5 in the verification embodiment;

[0046] Figure 4 It is a diagram of DOA estimation results corresponding to simulations using different methods when the number of snapshots is 100 and the signal-to-noise ratio is 20 in the verification embodiment;

[0047] Figure 5 It is a graph of DOA estimation results corresponding to the simulation using different methods when the number of snapshots is 5 and the signal-to-noise ratio is 5 in the verification embodiment;

[0048] Figure 6 It is a graph of DOA estimation results corresponding to the simulation using different methods when the number of snapshots is 5 and the signal-to-noise ratio is 20 in the verification embodiment;

[0049] Figure 7 It is a graph of DOA underdetermined estimation results corresponding to simulations using different methods when the number of snapshots is 100 and the signal-to-noise ratio is 5 in the verification embodiment;

[0050] Figure 8 It is a graph of DOA underdetermined estimation results corresponding to simulations using different methods when the number of snapshots is 100 and the signal-to-noise ratio is 20 in the verification embodiment;

[0051] Fig. 9 It is a curve diagram of the operation time and signal-to-noise ratio variation of the simulation using different methods in the verification embodiment;

[0052] Fig.10 It is a curve diagram of the change of the operation time with the number of snapshots in verifying the embodiment using different simulation methods. DETAILED DESCRIPTION

[0053] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0054] The invention takes wireless communication, radar, sonar and other fields as practical application background, can improve the estimation accuracy of the angle of arrival of the signal source under the condition of less sampling data, and provides an effective processing method for target positioning.

[0055] Specifically, the present invention discloses a basis pursuit denoising sparse reconstruction DOA estimation method based on noise pre-estimation, which is applied to a signal receiving device based on a coprime array, and includes the following steps: obtaining a received signal on a physical array element of the coprime array, and converting the received signal to obtain a corresponding first coprime virtual domain received signal; determining a second coprime virtual domain received signal based on the first coprime virtual domain received signal, and generating a covariance matrix of the second coprime virtual domain received signal; generating a CAPON spectrum based on the covariance matrix, and calculating the actual noise power; taking the actual noise power as a known quantity, using basis pursuit denoising sparse reconstruction to determine the power spectrum of the search angle, the power spectrum including the distance and angle of the target.

[0056] The present invention adopts a spatial smoothing method to obtain a full-rank covariance matrix of a received signal during signal preprocessing, thereby overcoming the defect of a DOA estimation method based on LASSO that the correlation between signals is large and requiring fewer snapshots than the LASSO method. The CAPON spectrum is used to solve the actual noise power, and the noise power is made the prior information in the sparse reconstruction algorithm. Meanwhile, important parameters such as the angle region of the incident signal are considered to improve the accuracy of the sparse reconstruction and enhance the robustness of the algorithm. In addition, the BPDN algorithm is used in the sparse reconstruction problem to obtain a source power estimation by fitting between covariance matrices, thereby effectively improving the accuracy of the source estimation.

[0057] In the embodiment of the present invention, firstly, a specific expression of the virtual domain reception signal is given through the corresponding relationship between the coprime array and the virtual array element, and the second coprime virtual domain reception signal corresponding to the maximum uniform continuous array element is obtained according to the distribution of the virtual array element. Specifically, the structural schematic diagram of the coprime array is as follows: Figure 1 As shown in the figure, the virtual domain distribution diagram is as follows Figure 2 shown.

[0058] The covariance matrix of the received signal on the physical array element is converted into the received signal on the virtual array element through vector processing, and the first coprime virtual domain received signal is obtained, that is:

[0059]

[0060] Among them, X D A first mutually prime virtual domain receives a signal, is the covariance matrix of the received signal, vec(·) represents vectorization, K is the number of signal sources, is the total power of K signal sources, a(θ k ) is the steering vector, θ k is the direction angle of the incident signal reaching the array, j is the imaginary sign, λ is the wavelength of the incident signal, P is the set of physical sensor locations in the coprime array, a * (θ k ) is the value of a(θ k ) for conjugation treatment, is the noise power, I D =vec(I P ), I P is the unit matrix of (2M+N-1)×(2M+N-1), where M and N are the number of elements of the coprime array, and the symbol represents the Kronecker product.

[0061] When the array structure is a 2M+N-1 coprime array, P = {mNd, 0≤m≤(2M-1)}∪{nMd, 0≤n≤(N-1)}, is the unit spacing.

[0062] Next, P is subjected to a difference joint process to obtain the positions of all virtual array elements, and then the second coprime virtual domain received signal is determined according to the maximum uniform continuous array element position distribution of the coprime array; wherein the second coprime virtual domain received signal is:

[0063]

[0064] In the formula, x R is a second mutually prime virtual domain receiving signal, A R is the steering vector of the received signal in the second coprime virtual domain, b is the power corresponding to each source angle, I R According to I D The unit matrix corresponding to the maximum uniform continuous array element position in the obtained coprime array virtual array is [-(MN+M-1)d, (MN+M-1)d].

[0065] In one embodiment, generating a covariance matrix of the second coprime virtual domain received signal includes:

[0066] The spatial smoothing algorithm is used to remove the correlation between the second coprime virtual domain received signals to obtain the covariance matrix; where the covariance matrix Specifically:

[0067]

[0068] In the formula, x Ri is the second coprime virtual domain receiving signal of the i-th subspace, and {(-i+1+n)d,n=0,1,…,MN+M-1} is the set of array element positions corresponding to the i-th subspace.

[0069] Then, according to the received signal covariance matrix The CAPON spectrum is used to obtain the real noise power by pre-estimating the source angle area and noise power.

[0070] Specifically, calculating the actual noise power includes:

[0071] Assuming the angle error range, according to the angle corresponding to the CAPON spectrum peak, we can get the source angle distribution set after a rough estimate. And the noise area Will As discrete points, the corresponding noise power is obtained by accumulating and averaging. Noise Power Specifically:

[0072]

[0073] Where T is the number of sample points, is the steering vector of the array composed of the largest uniform continuous array elements, θ ni for The discrete sample points in .

[0074] Next, according to the noise power Calculate the actual noise power. Specifically, by analyzing the CAPON spectrum, we know that the CAPON power spectrum of a signal arriving at the array is equal to the signal power and times the actual noise power. Calculating the actual noise power involves:

[0075]

[0076] in, is the actual noise power,

[0077] Finally, suppose there are q direction angles θ′1,θ′2,θ′3,…,θ′ q (q>>K), the steering vector and power of the grid search angle are The relationship between them is calculated, and basis pursuit denoising and sparse reconstruction are used to obtain the power spectrum of all search angles.

[0078] More specifically, the power spectrum of the search angle determined by basis pursuit denoising and sparse reconstruction includes:

[0079] The angle region grid is constructed according to the sparsity of the incident source; the power spectrum of the search angle is determined according to the objective function.

[0080] In this embodiment, the objective function is specifically:

[0081]

[0082] Among them, A′ R is the steering vector for each search angle, η is the regularization parameter (ie, the fitting error of the covariance matrix), B′ is the diagonal matrix of the power of each search angle, and the power of each search angle is diag(B′)=b′.

[0083] In order to verify the effectiveness of the method of the present invention, the following simulation verification example was carried out.

[0084] 1. Simulation data:

[0085] Consider a coprime array consisting of M=3, N=5, and 2M+N-1=10 array elements. The position set of all physical array elements is P={0,3d,5d,6d,9d,10d,12d,15d,20d,25d}, and the unit spacing of the array elements is d=λ / 2, where λ represents the wavelength of the incident signal source. The signal sources are independent of each other, and the noise in the space is only Gaussian white noise. The preset complete angle grid is divided at an interval of 0.1°, and the regularization parameter of the method of the present invention is set to 0.2. This simulation uses the LASSO algorithm, the BPDN algorithm, and the method of the present invention to estimate DOA.

[0086] 2. Simulation content and results:

[0087] In simulation 1, it is assumed that there are four signal sources in the space, and the angles of arrival at the array are -30°, 5°, 20°, and 40° respectively. Under different signal-to-noise ratios and different snapshot numbers, the angle estimation performance under different signal-to-noise ratios is compared using the LASSO algorithm, the BPDN algorithm, and the method of the present invention.

[0088] Figure 3The following are the DOA estimation spatial spectrum change curves using the LASSO algorithm, BPDN algorithm and the proposed method when the number of snapshots is 100 and SNR=5dB. Figure 3 It can be seen that the method of the present invention has a significant ability to suppress noise compared with the LASSO algorithm.

[0089] Figure 4 The following are the DOA estimation spatial spectrum change curves using the LASSO algorithm, BPDN algorithm and the method of the present invention when the number of snapshots is 100 and SNR=20dB. Figure 4 It can be seen that with the increase of the signal-to-noise ratio, the performance of the method of the present invention is more significantly improved, and its main lobe width is narrower.

[0090] Figure 5 The following are the DOA estimation spatial spectrum change curves using the LASSO algorithm, BPDN algorithm and the method of the present invention when the number of snapshots is 5 and SNR=5dB. Figure 5 It can be seen that the method of the present invention can correctly estimate the signal arrival angle under low signal-to-noise ratio and small snapshot number.

[0091] Figure 6 The figure is a graph showing the spatial spectrum variation of DOA estimation using the LASSO algorithm, BPDN algorithm and the method of the present invention when the number of snapshots is 5 and SNR is 20 dB, wherein Proposed is the method of the present invention. Figure 6 It can be seen that under high signal-to-noise ratio and small snapshot number, compared with the LASSO algorithm and the BPDN algorithm, the method of the present invention has a stronger ability to suppress non-information source directions.

[0092] In simulation 2, the number of signal snapshots is 100, and there are 12 signal sources in the space, whose incident angles are -60°, -50°, -40°, -30°-20°, -10°, 0°, 10°, 20°, 30°, 40°, and 50°, respectively. The angle estimation performance under different signal-to-noise ratios is compared using the LASSO algorithm, the BPDN algorithm, and the method of the present invention.

[0093] Figure 7 The following are the DOA estimation spatial spectrum change curves using the LASSO algorithm, BPDN algorithm and the method of the present invention when the number of snapshots is 100 and SNR=5dB. Figure 7 It can be seen that the LASSO algorithm and the BPDN algorithm have the problem of angle deviation when the signal-to-noise ratio is low, and the method of the present invention correctly estimates all signal sources.

[0094] Figure 8 The following are the DOA estimation spatial spectrum change curves using the LASSO algorithm, BPDN algorithm and the method of the present invention when the number of snapshots is 100 and SNR=20dB. Figure 8It can be seen that in the case of multiple information sources, the method of the present invention can clearly distinguish the directions of various information sources and has stable estimation performance.

[0095] Simulation 3, assuming that there are 14 signal sources, whose incident azimuths are evenly distributed between [-60°, 60°]. The operation time of the LASSO algorithm, the BPDN algorithm and the method of the present invention are compared respectively.

[0096] Fig. 9 is the curve of the operation time changing with the signal-to-noise ratio when the number of snapshots is 50, Fig. 9 It can be seen that the LASSO algorithm has the shortest operation time, but the performance of the algorithm is poor and there is a problem of false target estimation. The method of the present invention has accurate estimation and short operation time.

[0097] Fig.10 This is the curve of the operation time changing with the number of snapshots when the signal-to-noise ratio is 5dB. Fig.10 It can be seen that in the case of large snapshots, the operation time of the BPDN algorithm and the method of the present invention both increases, and the method of the present invention has lower complexity and shorter operation time.

[0098] In summary, Figures 3 to 10 It is further shown that under the condition of a small number of snapshots, due to the large estimation error of the covariance matrix, the presence of correlated signals in the LASSO algorithm, and the large complexity of the BPDN algorithm, the correct angle estimation result cannot be obtained; the method proposed in the present invention can improve the DOA estimation performance of coprime virtual arrays under the condition of small snapshots, by pre-estimating the noise power in the BPDN algorithm, selecting the CAPON spectrum to solve the actual noise power size, making the noise power the prior information in the sparse reconstruction algorithm, and considering important parameters such as the angle range of the incident signal to improve the sparse reconstruction accuracy, thereby enhancing the robustness of the DOA estimation algorithm and providing an effective solution for practical applications.

[0099] Another technical solution of the present invention: a basis pursuit denoising and sparse reconstruction DOA estimation device based on noise pre-estimation, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned basis pursuit denoising and sparse reconstruction DOA estimation method based on noise pre-estimation is implemented.

[0100] It should be noted that the information interaction, execution process and other contents between the modules of the above-mentioned device are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0101] Those skilled in the art can clearly understand that, for the convenience and simplicity of description, only the division of the above-mentioned functional modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. The functional modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0102] The device may be a computing device such as a desktop computer, a notebook, a PDA, or a cloud server. The device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the device may include more or fewer components, or a combination of certain components, or different components, such as an input / output device, a network access device, etc.

[0103] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0104] In some embodiments, the memory may be an internal storage unit of the device, such as a hard disk or memory of the device. In other embodiments, the memory may also be an external storage device of the device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the device. Further, the memory may also include both an internal storage unit of the device and an external storage device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program, etc. The memory may also be used to temporarily store data that has been output or is to be output.

[0105] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. Those of ordinary skill in the art can appreciate that the modules and algorithm steps of each example described in conjunction with the embodiments disclosed in the present invention can be implemented with electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

Claims

1. A basis pursuit denoising and sparse reconstruction DOA estimation method based on noise pre-estimation, characterized in that: The invention is applied to a signal receiving device based on a coprime array, comprising the following steps: Acquire a received signal on a physical array element of a coprime array, and convert the received signal to obtain a corresponding first coprime virtual domain received signal; Determine a second mutually prime virtual domain received signal according to the first mutually prime virtual domain received signal, and generate a covariance matrix of the second mutually prime virtual domain received signal; Generate a CAPON spectrum according to the covariance matrix and calculate the actual noise power; Taking the actual noise power as a known quantity, basis pursuit denoising and sparse reconstruction are used to determine the power spectrum of the search angle, wherein the power spectrum includes the distance and angle of the target; Calculating the actual noise power involves: According to the angle corresponding to the CAPON spectrum peak, the noise area range is obtained after a rough estimate Will As discrete points, the corresponding noise power is obtained by accumulating and averaging. According to the noise power Calculate the actual noise power; The noise power Specifically: Where T is the number of sample points, is the steering vector of the array composed of the largest uniform continuous array elements, θ ni for Discrete sample points in ; According to the noise power Calculating the actual noise power involves: in, is the actual noise power, 2. The method for DOA estimation by basis pursuit denoising and sparse reconstruction based on noise pre-estimation according to claim 1, characterized in that: The first mutually prime virtual domain received signal is: Among them, X D receiving a signal for a first mutually prime virtual domain, is the covariance matrix of the received signal, vec(·) represents vectorization, K is the number of signal sources, is the total power of K signal sources, a(θ k ) is the steering vector, θ k is the direction angle of the incident signal reaching the array, j is the imaginary sign, λ is the wavelength of the incident signal, P is the set of physical sensor positions in the coprime array, a * (θ k ) is the value of a(θ k ) for conjugation treatment, is the noise power, I D =vec(I P ), I P is a unit matrix of (2M+N-1)×(2M+N-1), where M and N are the number of elements of the coprime array, and the symbol represents the Kronecker product.

3. The method for DOA estimation by basis pursuit denoising and sparse reconstruction based on noise pre-estimation according to claim 2, characterized in that: Determining the second mutually prime virtual domain received signal includes: According to the maximum uniform continuous array element position distribution of the coprime array, a second coprime virtual domain received signal is determined; wherein the second coprime virtual domain received signal is: In the formula, x R The second mutually prime virtual domain receives the signal, A R is the steering vector of the received signal in the second coprime virtual domain, b is the power corresponding to each source angle, I R According to I D The obtained unit matrix corresponds to the maximum uniform continuous array element position in the coprime array virtual array.

4. The method for DOA estimation by basis pursuit denoising and sparse reconstruction based on noise pre-estimation according to claim 3, characterized in that: Generating a covariance matrix of the second coprime virtual domain received signal includes: Using a spatial smoothing algorithm to remove the correlation between the second coprime virtual domain received signals, to obtain a covariance matrix; Among them, the covariance matrix Specifically: In the formula, x Ri The signal is received for the second coprime virtual domain of the i-th subspace.

5. The method for DOA estimation by basis pursuit denoising and sparse reconstruction based on noise pre-estimation according to claim 4, characterized in that: The power spectrum of the search angle is determined by using basis pursuit denoising and sparse reconstruction, including: Constructing the angular area grid according to the incident source sparsity; The power spectrum of the search angle is determined according to the objective function.

6. The method for DOA estimation by basis pursuit denoising and sparse reconstruction based on noise pre-estimation according to claim 5, characterized in that: The objective function is: Among them, A′ R is the steering vector for each search angle, η is the regularization parameter, and B′ is the diagonal matrix of the power for each search angle.

7. A DOA estimation device based on basis pursuit denoising and sparse reconstruction of noise pre-estimation, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the basis pursuit denoising and sparse reconstruction DOA estimation method based on noise pre-estimation as described in any one of claims 1 to 6 is implemented.

Images