A method for locating radio frequency interference sources based on generalized augmented covariance matrix reconstruction
Through the reconstruction method based on the generalized augmented covariance matrix, the problems of insufficient spatial resolution and low positioning accuracy in RF interference source positioning are solved, and efficient positioning of complex scenarios and accurate positioning of weak sources are achieved.
Patent Information
- Application Number
- CN202111353176.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-11-16
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2041-11-16
AI Technical Summary
The existing RF interference source positioning methods have insufficient spatial resolution and poor positioning accuracy performance, especially in complex pollution scenarios and weak source positioning.
The method based on generalized augmented covariance matrix reconstruction is adopted, and the generalized extended array is obtained through the array expansion model and relaxed constraints, and the generalized augmented covariance matrix is constructed, and the matrix reconstruction is carried out using the generalized reweighted kernel norm minimization algorithm. Combined with subspace algorithms, the arrival angle estimation is performed to obtain the location information of the RF interference source.
The spatial resolution and positioning accuracy of RF interference sources are improved, complex pollution scenarios are effectively handled, and the problems of difficulty in discovering and positioning of weak sources are solved.
Smart Images

Figure CN114200394B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of source positioning, and more specifically, relates to a method for positioning a radio frequency interference source based on generalized augmented covariance matrix reconstruction. Background Art
[0002] Radio frequency interference is an increasingly serious and challenging problem in microwave radiometer measurements. It has become an important limiting factor in the application and development of remote sensing such as geophysical parameter inversion, target detection, and microwave imaging. In recent years, many studies have been devoted to solving the problem of radio frequency interference. At present, there are two main methods for dealing with radio frequency interference. One is to locate the source of radio frequency interference and shut down the illegal source of radio frequency interference by coordinating with relevant management departments; the other is to directly process the contaminated remote sensing data. Both methods usually require the use of the location information of the radio frequency interference source, so the location of the radio frequency interference source is particularly important.
[0003] However, the existing methods for locating RF interference sources have many shortcomings, such as insufficient spatial resolution, poor positioning accuracy, a sharp drop in performance when dealing with complex pollution scenes, and difficulty in discovering and locating weak sources. Therefore, how to achieve high-precision positioning of RF interference sources is an urgent problem to be solved. Summary of the invention
[0004] In view of the above defects or improvement needs of the prior art, the present invention provides a method for locating a radio frequency interference source based on generalized augmented covariance matrix reconstruction, the purpose of which is to relax the constraints of the array expansion model of a multi-sensor system to obtain a generalized extended array, and then transform the reconstruction problem of the corresponding generalized augmented covariance matrix into a generalized reweighted nuclear norm minimization problem, which is solved to reconstruct the generalized augmented covariance matrix; the arrival angle of the reconstructed generalized augmented covariance matrix is estimated to obtain the radio frequency interference source positioning information, thereby solving the technical problems of insufficient spatial resolution and poor positioning accuracy performance in the existing radio frequency interference source positioning.
[0005] To achieve the above object, according to one aspect of the present invention, a method for locating a radio frequency interference source based on generalized augmented covariance matrix reconstruction is provided, comprising:
[0006] S1: Expanding the original sparse array corresponding to the multi-sensor system to obtain the corresponding extended virtual array, and obtaining the generalized extended array by relaxing the constraints of the array extension model on the basis of considering the size of the extended virtual array; constructing the generalized augmented covariance matrix based on the position correspondence between the differential array of the original sparse array and the differential array of the generalized extended array and the covariance matrix of the original sparse array;
[0007] S2: establishing a reconstruction model of the generalized augmented covariance matrix; solving the minimum optimization model corresponding to the reconstruction model using a generalized reweighted nuclear norm minimization algorithm to reconstruct the generalized augmented covariance matrix;
[0008] S3: Use a subspace algorithm to estimate the arrival angle of the reconstructed generalized augmented covariance matrix to obtain the location information of the RF interference source.
[0009] In one embodiment, the S1 includes:
[0010] S11: Using an array expansion model, the original sparse array corresponding to the multi-sensor data is expanded to obtain an extended virtual array. The array expansion model is:
[0011]
[0012] Among them, X o is the original sparse array, n o Equal to X o Number of array elements |X o |;X v is the extended virtual array, n v Equal to X v Number of array elements |X v |DCA v Represents X v Differential array, DCA o = {u k,l =x k -x l :k=1,…,n o ,l=1,…,n o},u k,l For DCA o The element at position (k,l) in ; DCA o For X o The difference array, For DCA v Middle Elements of position;
[0013] S12: Relax the constraints of the array expansion model: On the basis of considering the size of the extended virtual array, the original sparse array is expanded using the relaxed array expansion model to obtain the generalized extended array X e , and obtain the generalized extended array X e Differential Array DCA e ; Represents X e Middle The position vector of the array element, n e =|X e |>|X v |=n v , n e For X e Number of array elements |X e |; For DCA e Middle Elements of position;
[0014] S13: Differential array DCA based on the original sparse array o Differential array DCA with the generalized extended array e The generalized augmented covariance matrix of the generalized extended array is constructed based on the position correspondence between the covariance matrix of the original sparse array.
[0015] In one embodiment, the S13 includes:
[0016] Using the formula Construct the generalized extended array X e The generalized augmented covariance matrix of The generalized extended array X e The size of the covariance matrix R is n o ×n o , the generalized augmented covariance matrix The size is n e ×n e :
[0017] In the formula, represents the generalized augmented covariance matrix The Row, No. The element at column position, [R] k,l Represents the element at the kth row and lth column in the covariance matrix R, k, l = 1, ..., n o .
[0018] In one embodiment, the S2 includes:
[0019] S21: Building the generalized augmented covariance matrix The reconstructed model is expressed as follows:
[0020]
[0021] in, represents the reconstructed generalized augmented covariance matrix, Representation Matrix rank;||.||F represents the F-norm of the matrix, δ F >0 represents the tolerance parameter for controlling the fitting error; H represents the index set composed of the observed element identifiers; represents an observation operator, which only gets the elements indexed by H;
[0022] S22: Solve the minimum optimization model corresponding to the reconstructed model using a generalized reweighted nuclear norm minimization algorithm to obtain a reconstructed generalized augmented covariance matrix
[0023] In one embodiment, the S22 includes:
[0024] S22-1: Convert the reconstructed model into a generalized reweighted nuclear norm minimization model, which is expressed as: ||.|| w,* represents the generalized reweighted nuclear norm of the matrix; The expression is: Representation Matrix The i-th singular value of i ≥0 indicates matrix The generalized reweighted value of the i-th singular value;
[0025] S22-2: Convert the generalized reweighted nuclear norm minimization model into an unconstrained optimization model, expressed as: Where μ represents the regularization parameter; A: represents a linear mapping, M represents the cardinality of the index set H;
[0026] S22-3: Solve the unconstrained optimization model using the generalized reweighted nuclear norm minimization algorithm to obtain the reconstructed generalized augmented covariance matrix
[0027] Among them, the generalized reweighted value expression is: w (k) represents the generalized reweighted vector of the kth iteration in the generalized reweighted nuclear norm minimization algorithm; c is a constant, ε is a positive number, 0≤P<1; represents the generalized augmented covariance matrix obtained at the kth iteration in the generalized reweighted nuclear norm minimization algorithm.
[0028] In one embodiment, the S3 includes:
[0029] S31: For the reconstructed generalized augmented covariance matrix Decompose and divide the corresponding noise subspace U according to the estimated number of RF interference sources n and signal subspace U s ;
[0030] S32: According to the noise subspace U n and the signal subspace U s The orthogonal relationship of is used to obtain the array space spectrum function P(ξ,η), and the arrival angle of the array space spectrum function P(ξ,η) is estimated to obtain the positioning result of the radio frequency interference source;
[0031] Among them, (ξ,η) is expressed as the direction cosine vector, ξ=sinθcosφ, η=sinθsinφ, θ represents the observation elevation angle, and φ represents the observation direction angle.
[0032] According to another aspect of the present invention, there is provided a device for locating a radio frequency interference source based on generalized augmented covariance matrix reconstruction, comprising:
[0033] A construction module is used to expand the original sparse array corresponding to the multi-sensor system to obtain the corresponding extended virtual array, and obtain the generalized extended array by relaxing the constraints of the array extension model on the basis of considering the size of the extended virtual array; based on the position correspondence between the differential array of the original sparse array and the differential array of the generalized extended array and the covariance matrix of the original sparse array, a generalized augmented covariance matrix is constructed;
[0034] A reconstruction module, used to establish a reconstruction model of the generalized augmented covariance matrix; solve the minimum optimization model corresponding to the reconstruction model using a generalized reweighted nuclear norm minimization algorithm to reconstruct the generalized augmented covariance matrix;
[0035] The positioning module is used to estimate the arrival angle of the reconstructed generalized augmented covariance matrix using a subspace algorithm to obtain the positioning information of the radio frequency interference source.
[0036] According to another aspect of the present invention, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of the method when executing the computer program.
[0037] According to another aspect of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method described above are implemented.
[0038] In general, the above technical solutions conceived by the present invention can achieve the following beneficial effects compared with the prior art:
[0039] The method provided by the present invention first expands the original sparse array of the multi-sensor system according to the array expansion model to obtain an extended virtual array, obtains a larger-scale generalized extended array by relaxing the constraints of the array expansion model on the basis of considering the size of the extended virtual array, and constructs a generalized augmented covariance matrix according to the corresponding relationship between the differential array of the original sparse array and the differential array of the generalized extended sparse array; considering the different importance of the components of the generalized augmented covariance matrix, the reconstruction problem of the generalized augmented covariance matrix is converted into a generalized reweighted nuclear norm minimization problem, and the generalized reweighted nuclear norm minimization algorithm is used to complete the reconstruction of the generalized augmented covariance matrix; based on the reconstructed generalized augmented covariance matrix, a subspace-type algorithm is used to estimate the arrival angle and locate the radio frequency interference source. The present invention can improve the spatial resolution and positioning accuracy performance, effectively handle complex pollution scenes, and solve the problem of weak source discovery and positioning difficulties. BRIEF DESCRIPTION OF THE DRAWINGS
[0040] Figure 1 is a flow chart of a method for locating a radio frequency interference source based on generalized augmented covariance matrix reconstruction in one embodiment of the present invention;
[0041] Figure 2a is a simulation diagram of the position relationship of array elements of an original sparse array in one embodiment of the present invention;
[0042] Figure 2b is a differential array simulation diagram of an original sparse array in one embodiment of the present invention;
[0043] Figure 3a is a simulation diagram of the position relationship of elements of an extended virtual array in one embodiment of the present invention;
[0044] Figure 3b is a differential array simulation diagram of an extended virtual array in one embodiment of the present invention;
[0045] Figure 4a is a simulation diagram of the position relationship of array elements of a generalized extended array in one embodiment of the present invention;
[0046] Figure 4b is a differential array simulation diagram of a generalized extended array in one embodiment of the present invention;
[0047] Figure 5a is a simulation diagram of the amplitude of the covariance matrix elements in one embodiment of the present invention;
[0048] Figure 5b is a phase simulation diagram of a covariance matrix element in one embodiment of the present invention;
[0049] Figure 6a is a simulation diagram of the amplitude of the generalized augmented covariance matrix elements in one embodiment of the present invention;
[0050] Figure 6b is a phase simulation diagram of a generalized augmented covariance matrix element in one embodiment of the present invention;
[0051] Figure 7a is a simulation diagram of the amplitude of the reconstructed generalized augmented covariance matrix elements in one embodiment of the present invention;
[0052] Figure 7b is a phase simulation diagram of the reconstructed generalized augmented covariance matrix elements in one embodiment of the present invention;
[0053] Figure 8 is a simulation diagram of the locating result of a radio frequency interference source in one embodiment of the present invention;
[0054] Figure 9a is an array space spectrum corresponding to the radio frequency interference source positioning based on the direct covariance matrix in one embodiment of the present invention;
[0055] Figure 9b is an array space spectrum corresponding to the radio frequency interference source positioning based on the augmented covariance matrix in one embodiment of the present invention;
[0056] Fig.9c It is an array space spectrum corresponding to the radio frequency interference source positioning based on matrix completion in one embodiment of the present invention;
[0057] Figure 9d It is an array space spectrum corresponding to the RF interference source positioning based on the generalized augmented covariance matrix reconstruction in one embodiment of the present invention. DETAILED DESCRIPTION
[0058] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.
[0059] like Figure 1 As shown, the present invention provides a method for locating a radio frequency interference source based on generalized augmented covariance matrix reconstruction, comprising:
[0060] S1: Expand the original sparse array corresponding to the multi-sensor system to obtain the corresponding extended virtual array, and obtain the generalized extended array by relaxing the constraints of the array extension model on the basis of considering the size of the extended virtual array; construct the generalized augmented covariance matrix based on the position correspondence between the differential array of the original sparse array and the differential array of the generalized extended array and the covariance matrix of the original sparse array;
[0061] S2: Establish a reconstruction model of the generalized augmented covariance matrix; use the generalized reweighted nuclear norm minimization algorithm to solve the minimum optimization model corresponding to the reconstruction model to reconstruct the generalized augmented covariance matrix;
[0062] S3: Use a subspace algorithm to estimate the arrival angle of the reconstructed generalized augmented covariance matrix to obtain the location information of the RF interference source.
[0063] In one embodiment, S1 includes:
[0064] S11: Use the array expansion model to expand the original sparse array corresponding to the multi-sensor data to obtain an extended virtual array. The array expansion model is:
[0065]
[0066] Among them, X o is the original sparse array, n o Equal to X o Number of array elements |X o |;X v To expand the virtual array, n v Equal to X v Number of array elements |X v |DCA v Represents X v Differential array, DCA o = {u k,l =x k -x l :k=1,…,n o ,l=1,…,n o},u k,l For DCA o The element at position (k,l) in ; DCA o For X o The difference array, For DCA v Middle Elements of position;
[0067] S12: Relax the constraints of the array expansion model: Based on the consideration of the size of the extended virtual array, the original sparse array is expanded using the relaxed array expansion model to obtain the generalized extended array X e , and obtain the generalized extended array X e Differential Array DCA e ; Represents X e Middle The position vector of the array element, n e =|Xe |>|X v |=n v , n e For X e Number of array elements |X e |; For DCA e Middle Elements of position;
[0068] S13: Differential Array DCA Based on Original Sparse Array o Differential Array DCA with Generalized Extended Array e The generalized augmented covariance matrix of the generalized extended array is constructed based on the position correspondence between them and the covariance matrix of the original sparse array.
[0069] In one embodiment, S13 includes:
[0070] Using the formula Construct a generalized extended array X e The generalized augmented covariance matrix of Generalized Extended Array X e The size of the covariance matrix R is n o ×n o , the generalized augmented covariance matrix The size is n e ×n e :
[0071] In the formula, represents the generalized augmented covariance matrix The Row, No. The element at column position, [R] k,l Represents the element at the kth row and lth column in the covariance matrix R, k, l = 1, ..., n o .
[0072] Specifically, (1) according to the array expansion model, the original sparse array of the multi-sensor system is expanded to obtain the corresponding extended virtual array, and then its expansion constraints are relaxed to obtain a larger-scale generalized extended array, and the generalized augmented covariance matrix is constructed using the covariance matrix of the original sparse array according to the corresponding relationship between the difference correlation array Difference Coarray of the original sparse array and the Difference Coarray of the generalized extended array. The specific steps are as follows:
[0073] (1-1) Establish an array expansion model that expands the original sparse array into an extended virtual array. The expression is as follows:
[0074]
[0075] In the formula, X v is the design variable, representing the extended virtual array, which is a set of element position vectors of the extended virtual array. v Equal to extended virtual array X v Number of array elements |X v |;
[0076] X o represents the original sparse array, which is a set of element position vectors of the original sparse array, n o Equal to X o The number of original sparse array elements |X o |;
[0077] DCA v With DCA o Respectively represent X v With X o Difference Coarray;
[0078] For the original sparse array where x k Represents X o The position vector of the kth array element in, k = 1, ..., n o , then the original sparse array X o The Difference Coarray is expressed as:
[0079] DCA o = {u k,l =x k -x l :k=1,…,n o ,l=1,…,n o};
[0080] In the formula, u k,l For DCA o The element at position (k,l) in ;
[0081] For extended virtual arrays in Represents X v Middle The position vector of the array element, Then expand the virtual array X v The Difference Coarray is expressed as:
[0082]
[0083] In the formula, For DCA v Middle Elements of position;
[0084] (1-2) Use the intelligent optimization algorithm to solve the model established in step (1-1) to obtain X v The optimal solution is X o The corresponding extended virtual array;
[0085] (1-3) Constraints on the array expansion model Relax and consider the size of the extended virtual array to obtain the generalized extended array X e satisfy:
[0086] n e =|X e |>|X v |=n v
[0087] Where n e represents the generalized extended array X e Number of array elements |X e |;
[0088] For generalized extended arrays in Represents X e Middle The position vector of the array element, Then the generalized extended array X e The Difference Coarray is expressed as:
[0089]
[0090] In the formula, For DCA e Middle Elements of position;
[0091] (1-4) According to the original sparse array X o Difference Coarray with generalized extended array X e The position correspondence of the Difference Coarray is used to construct the generalized augmented covariance matrix of the generalized extended array using the covariance matrix R of the original sparse array The size of the covariance matrix R is n o ×n o (The covariance matrix contains the interference source signal data), generalized augmented covariance matrix The size is n e ×n e ; Its construction method is as follows:
[0092]
[0093] In the formula, represents the generalized augmented covariance matrix The Row, No. The element at column position, [R] k,l Represents the element at the kth row and lth column in the covariance matrix R, k, l = 1, ..., n o .
[0094] In one embodiment, S2 includes:
[0095] S21: Building the generalized augmented covariance matrix The reconstructed model is expressed as follows:
[0096]
[0097] in, represents the reconstructed generalized augmented covariance matrix, Representation Matrix rank;
[0098] ||.|| F represents the F-norm of the matrix, δ F >0 represents the tolerance parameter for controlling the fitting error; H represents the index set composed of the observed element identifiers; represents an observation operator, which only gets the elements indexed by H;
[0099] S22: Use the generalized reweighted nuclear norm minimization algorithm to solve the minimum optimization model corresponding to the reconstructed model and obtain the reconstructed generalized augmented covariance matrix
[0100] In one embodiment, S22 includes:
[0101] S22-1: Convert the reconstruction model into a generalized reweighted nuclear norm minimization model, which is expressed as: ||.|| w,* represents the generalized reweighted nuclear norm of the matrix; The expression is: Representation Matrix The i-th singular value of i ≥0 indicates matrix The generalized reweighted value of the i-th singular value;
[0102] S22-2: The generalized reweighted nuclear norm minimization model is transformed into an unconstrained optimization model, expressed as: Where μ represents the regularization parameter; A: represents a linear mapping, M represents the cardinality of the index set H;
[0103] S22-3: Solve the unconstrained optimization model using the generalized reweighted nuclear norm minimization algorithm, and finally obtain the reconstructed generalized augmented covariance matrix
[0104] Among them, the generalized reweighted value expression is: w (k) represents the generalized reweighted vector of the kth iteration in the generalized reweighted nuclear norm minimization algorithm; c is a constant, ε is a positive number, 0≤P<1; represents the generalized augmented covariance matrix obtained at the kth iteration in the generalized reweighted nuclear norm minimization algorithm.
[0105] Specifically, (2) considering the different importance of the components of the generalized augmented covariance matrix, the reconstruction problem of the generalized augmented covariance matrix is transformed into a generalized reweighted nuclear norm minimization problem, and the generalized augmented covariance matrix obtained in step (1) is reconstructed using the generalized reweighted nuclear norm minimization algorithm to obtain the reconstructed generalized augmented covariance matrix. The specific steps are as follows:
[0106] (2-1) Establish the generalized augmented covariance matrix The reconstructed model is expressed as follows:
[0107]
[0108] In the formula, represents the reconstructed generalized augmented covariance matrix, Representation Matrix rank;
[0109] ||.|| F represents the Frobenius norm of the matrix, δ F >0 represents the tolerance parameter for controlling the fitting error; H represents the index set composed of the observed element identifiers; represents an observation operator, which only gets the elements indexed by H;
[0110] (2-2) Solve the reconstruction model established in step (2-1) to obtain the reconstructed generalized augmented covariance matrix The specific steps are as follows:
[0111] (2-2-1) Considering the different importance of the components of the generalized augmented covariance matrix, the reconstruction model established in step (2-1) is transformed into a generalized reweighted nuclear norm minimization model, which is expressed as follows: In the formula, ||.|| w,*represents the generalized reweighted nuclear norm of the matrix, that is, the sum of the generalized reweighted singular values of the matrix; therefore It can be further expressed as:
[0112]
[0113] Representation Matrix The i-th singular value of i ≥0 indicates matrix The generalized reweighted value of the i-th singular value;
[0114] (2-2-2) The generalized reweighted nuclear norm minimization model of (2-2-1) can be further transformed into an unconstrained optimization model, expressed as follows:
[0115]
[0116] Where μ represents the regularization parameter; A: represents a linear mapping, M represents the cardinality of the index set H;
[0117] (2-2-3) The model established in (2-2-2) is solved using the generalized reweighted nuclear norm minimization algorithm. In the present invention, the generalized reweighted value is realized by solving the (P-1)th power of the matrix singular value vector, and the expression is as follows:
[0118]
[0119] In the formula, w (k) represents the generalized reweighted vector of the kth iteration in the generalized reweighted nuclear norm minimization algorithm; c represents a constant that is always greater than zero, ε represents a sufficiently small positive number, 0≤P<1; It represents the generalized augmented covariance matrix obtained in the kth iteration of the generalized reweighted nuclear norm minimization algorithm; finally, the model established by (2-2-2) is solved to obtain the reconstructed generalized augmented covariance matrix
[0120] In one embodiment, S3 includes:
[0121] S31: For the reconstructed generalized augmented covariance matrix Decompose and divide the corresponding noise subspace U according to the estimated number of RF interference sources n and signal subspace U s ;
[0122] S32: According to the noise subspace U n and signal subspace U sThe orthogonal relationship of is used to obtain the array space spectrum function P(ξ,η), and the arrival angle of the array space spectrum function P(ξ,η) is estimated to obtain the positioning result of the radio frequency interference source;
[0123] Among them, (ξ,η) is expressed as the direction cosine vector, ξ=sinθcosφ, η=sinθsinφ, θ represents the observation elevation angle, and φ represents the observation direction angle.
[0124] For example, in the method provided by the present invention, the "Y" type array carried by the European Space Agency (SMOS, Soil Moisture and Ocean Salinity) satellite can be selected as the original sparse array, and the number of array elements is n o =69, Figure 2a is the original sparse array element position relationship, Figure 2b A Difference Coarray representing the original sparse array.
[0125] The array expansion model is solved using an intelligent optimization algorithm to obtain an extended virtual array with n array elements. v =931, the array element position relationship and Difference Coarray are as follows Figure 3a and Figure 3b As shown. Among them, Figure 3a represents the position relationship of the extended virtual array elements, Figure 3b A Difference Coarray representing an extended virtual array.
[0126] Relax the constraints of the array expansion model to obtain a larger generalized extended array with n array elements. e =1333, the array element position relationship and Difference Coarray are as follows Figure 4a and Figure 4b As shown. Among them, Figure 4a Represents the position relationship of the elements of the generalized extended array (generalized extended virtual array structure), Figure 4b Difference Coarray representing a generalized extended array (generalized extended virtual array structure).
[0127] In this embodiment, a shot of SMOS data with a timestamp of 2011-03-06T04:29:59 is selected. The size of the covariance matrix of the original sparse array is 69×69, and its element distribution is as follows: Figure 5a and Figure 5b As shown. Among them, Figure 5a represents the magnitude of the covariance matrix elements, Figure 5b Represents the phase of the covariance matrix elements.
[0128] According to the element correspondence between the Difference Coarray of the original sparse array and the Difference Coarray of the generalized extended array, the covariance matrix of the original sparse array is used to construct the generalized augmented covariance matrix of the generalized extended array. Its size is 1333×1333, and its element distribution is as follows Figure 6a and Figure 6b As shown. Among them, Figure 6a represents the magnitude of the generalized augmented covariance matrix elements, Figure 6b represents the phase of the generalized augmented covariance matrix elements.
[0129] (2) Considering the different importance of the components of the generalized augmented covariance matrix, the reconstruction problem of the generalized augmented covariance matrix is transformed into a generalized reweighted nuclear norm minimization problem. The generalized reweighted nuclear norm minimization algorithm is used to reconstruct the generalized augmented covariance matrix obtained in step (1) to obtain the reconstructed generalized augmented covariance matrix.
[0130] By constructing a reconstruction model of the generalized augmented covariance matrix, the reconstruction model of the generalized augmented covariance matrix is transformed into a generalized reweighted nuclear norm minimization model, and then further transformed into an unconstrained optimization model.
[0131] The optimization model is solved by the generalized reweighted nuclear norm minimization algorithm, so as to reconstruct the generalized augmented covariance matrix and obtain the reconstructed generalized augmented covariance matrix.
[0132] In this embodiment, the generalized reweighted nuclear norm minimization algorithm based on fixed point iteration (FPI) is used to reconstruct the generalized augmented covariance matrix. Finally, the reconstructed generalized augmented covariance matrix is obtained, and its size is 1333×1333, and its element distribution is as follows: Figure 7a and Figure 7b As shown. Among them, Figure 7a represents the magnitude of the reconstructed generalized augmented covariance matrix elements, Figure 7b represents the phase of the reconstructed generalized augmented covariance matrix element.
[0133] Note that in specific implementations, other generalized reweighted nuclear norm minimization methods may also be used to achieve reconstruction of the generalized augmented covariance matrix.
[0134] (3) Based on the reconstructed generalized augmented covariance matrix obtained in step (2), a subspace algorithm is used to estimate the arrival angle to obtain the positioning result of the RF interference source.
[0135] The reconstructed generalized augmented covariance matrix is decomposed to obtain the corresponding noise subspace and signal subspace.
[0136] According to the orthogonal relationship between the noise subspace and the signal subspace, the subspace algorithm is used to obtain the array space spectrum function, and the arrival angle is estimated to obtain the location of the RF interference source.
[0137] In this embodiment, the reconstructed generalized augmented covariance matrix is analyzed by singular value decomposition and MUSIC algorithm, and the RF interference source positioning result obtained is as follows: Figure 8 shown. Figure 8 Represents the obtained array spatial spectrum.
[0138] Note that other subspace algorithms may also be used to locate the RF interference source during specific implementation.
[0139] In order to further reflect the advantages of the present invention, a comparison between the present invention and other RF interference source locating methods in terms of positioning accuracy performance, spatial resolution and array space spectrum is specially set up. That is, the data of this embodiment is uniformly adopted, and the MUSIC algorithm is used to locate a known RF interference source in the embodiment (located at (0.2973, 0.0597)) and give a complete array space spectrum. In addition to the positioning method of the present invention (GACMR), other RF interference source locating methods are based on the direct covariance matrix (DC), based on the augmented covariance matrix (AC), and based on matrix completion (MC). The comparison of positioning accuracy performance and spatial resolution is shown in Table 1, and the array space spectrum comparison diagrams corresponding to DC, AC, MC and GACMR are shown as follows. Figure 9a , Figure 9b , Fig.9c and Figure 9d shown.
[0140] Positioning method DC AC MC GACMR Positioning accuracy 0.00225 0.00182 0.00153 0.00058 Spatial resolution / 0.00345 0.00195 0.00285
[0141] Table 1
[0142] In the embodiment provided by the present invention, the positioning results obtained illustrate that the above-mentioned radio frequency interference source positioning method based on generalized augmented covariance matrix reconstruction has high spatial resolution and positioning accuracy performance, and can effectively process complex pollution scenes, effectively solving the problem of difficulty in discovering and locating weak sources.
[0143] According to another aspect of the present invention, there is provided a device for locating a radio frequency interference source based on generalized augmented covariance matrix reconstruction, comprising:
[0144] A construction module is used to expand the original sparse array corresponding to the multi-sensor system to obtain the corresponding extended virtual array, and obtain the generalized extended array by relaxing the constraints of the array extension model on the basis of considering the size of the extended virtual array; a generalized augmented covariance matrix is constructed based on the position correspondence between the differential array of the original sparse array and the differential array of the generalized extended array and the covariance matrix of the original sparse array;
[0145] A reconstruction module is used to establish a reconstruction model of the generalized augmented covariance matrix; a generalized reweighted nuclear norm minimization algorithm is used to solve the minimum optimization model corresponding to the reconstruction model to reconstruct the generalized augmented covariance matrix;
[0146] The positioning module is used to estimate the arrival angle of the reconstructed generalized augmented covariance matrix using a subspace algorithm to obtain the positioning information of the radio frequency interference source.
[0147] According to another aspect of the present invention, an electronic device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and the steps of the method are implemented when the processor executes the computer program.
[0148] According to another aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, the steps of the method being implemented when the computer program is executed by a processor.
[0149] It will be easily understood by those skilled in the art that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for locating a radio frequency interference source based on generalized augmented covariance matrix reconstruction, characterized in that: include: S1: Expanding the original sparse array corresponding to the multi-sensor system to obtain the corresponding extended virtual array, and obtaining the generalized extended array by relaxing the constraints of the array extension model on the basis of considering the size of the extended virtual array; constructing the generalized augmented covariance matrix based on the position correspondence between the differential array of the original sparse array and the differential array of the generalized extended array and the covariance matrix of the original sparse array; S2: establishing a reconstruction model of the generalized augmented covariance matrix; solving the minimum optimization model corresponding to the reconstruction model using a generalized reweighted nuclear norm minimization algorithm to reconstruct the generalized augmented covariance matrix; S3: Use the subspace algorithm to estimate the arrival angle of the reconstructed generalized augmented covariance matrix to obtain the location information of the RF interference source; The S1 includes: S11: using an array expansion model to expand the original sparse array corresponding to the multi-sensor data to obtain an extended virtual array, and the array expansion model is: X o is the original sparse array, n o Equal to X o Number of array elements |X o |;X v is the extended virtual array, n v Equal to X v Number of array elements |X v |DCA v Represents X v Differential array, DCA o = {u k,l =x k -x l :k=1,…,n o ,l=1,…,n o },u k,l For DCA o The element at position (k,l) in ; DCA o For X o The difference array, For DCA v Middle Elements of position; S12: Relax the constraints of the array expansion model: On the basis of considering the size of the extended virtual array, the original sparse array is expanded using the relaxed array expansion model to obtain the generalized extended array X e , and obtain the generalized extended array X e Differential Array DCA e ; Represents X e Middle The position vector of the array element, n e =|X e |>|X v |=n v , n e For X e Number of array elements |X e |; For DCA e Middle Elements of position; S13: Differential array DCA based on the original sparse array o Differential array DCA with the generalized extended array e The generalized augmented covariance matrix of the generalized extended array is constructed based on the position correspondence between the covariance matrix of the original sparse array.
2. The method for locating a radio frequency interference source according to claim 1, wherein: The S13 includes: Using the formula Construct the generalized extended array X e The generalized augmented covariance matrix of The generalized extended array X e The size of the covariance matrix R is n o ×n o , the generalized augmented covariance matrix The size is n e ×n e : In the formula, represents the generalized augmented covariance matrix The Row, No. The element at column position, [R] k,l Represents the element at the kth row and lth column in the covariance matrix R, k, l = 1, ..., n o .
3. The method for locating a radio frequency interference source according to claim 1, wherein: The S2 includes: S21: Building the generalized augmented covariance matrix The reconstructed model is expressed as follows: in, represents the reconstructed generalized augmented covariance matrix, Representation Matrix rank;||.|| F represents the F-norm of the matrix, δ F >0 represents the tolerance parameter for controlling the fitting error; H represents the index set composed of the observed element identifiers; represents an observation operator, which only gets the elements indexed by H; S22: Solve the minimum optimization model corresponding to the reconstructed model using a generalized reweighted nuclear norm minimization algorithm to obtain a reconstructed generalized augmented covariance matrix 4. The method for locating a radio frequency interference source according to claim 3, wherein: The S22 includes: S22-1: Convert the reconstructed model into a generalized reweighted nuclear norm minimization model, which is expressed as: ||.|| w,* represents the generalized reweighted nuclear norm of the matrix; The expression is: Representation Matrix The i-th singular value of i ≥0 indicates matrix The generalized reweighted value of the i-th singular value; S22-2: Convert the generalized reweighted nuclear norm minimization model into an unconstrained optimization model, expressed as: Where μ represents the regularization parameter; A: represents a linear mapping, M represents the cardinality of the index set H; S22-3: Solve the unconstrained optimization model using the generalized reweighted nuclear norm minimization algorithm to obtain the reconstructed generalized augmented covariance matrix Among them, the generalized reweighted value expression is: w (k) represents the generalized reweighted vector of the kth iteration in the generalized reweighted nuclear norm minimization algorithm; c is a constant, ε is a positive number, 0≤P<1; represents the generalized augmented covariance matrix obtained at the kth iteration in the generalized reweighted nuclear norm minimization algorithm.
5. The method for locating a radio frequency interference source according to any one of claims 1 to 4, characterized in that: The S3 includes: S31: For the reconstructed generalized augmented covariance matrix Decompose and divide the corresponding noise subspace U according to the estimated number of RF interference sources n and signal subspace U s ; S32: According to the noise subspace U n and the signal subspace U s The orthogonal relationship of is used to obtain the array space spectrum function P(ξ,η), and the arrival angle of the array space spectrum function P(ξ,η) is estimated to obtain the positioning result of the radio frequency interference source; Among them, (ξ,η) is expressed as the direction cosine vector, ξ=sinθcosφ, η=sinθsinφ, θ represents the observation elevation angle, and φ represents the observation direction angle.
6. A radio frequency interference source location device based on generalized augmented covariance matrix reconstruction, characterized in that: The method for locating a radio frequency interference source according to claim 1 comprises: A construction module is used to expand the original sparse array corresponding to the multi-sensor system to obtain the corresponding extended virtual array, and obtain the generalized extended array by relaxing the constraints of the array extension model on the basis of considering the size of the extended virtual array; based on the position correspondence between the differential array of the original sparse array and the differential array of the generalized extended array and the covariance matrix of the original sparse array, a generalized augmented covariance matrix is constructed; A reconstruction module, used to establish a reconstruction model of the generalized augmented covariance matrix; solve the minimum optimization model corresponding to the reconstruction model using a generalized reweighted nuclear norm minimization algorithm to reconstruct the generalized augmented covariance matrix; The positioning module is used to estimate the arrival angle of the reconstructed generalized augmented covariance matrix using a subspace algorithm to obtain the positioning information of the radio frequency interference source.
7. An electronic device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.
8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.
