Far and near field mixed source parameter underdetermined estimation method based on cross correlation
Through the three-dimensional spatial accurate propagation model and cross-correlation calculation of signal sources of symmetric cross-arrays, the problem of low positioning accuracy of near-field source is solved, and high-precision signal source parameter estimation is realized, especially in the case of far-near-field mixed sources, which improves Estimate performance.
Patent Information
- Application Number
- CN202411061794.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-05
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has limited estimation accuracy in near-field source positioning, especially in the case of far-field and near-field mixed sources, and lacks effective parameter estimation methods, especially the under-determined estimation method based on cross-correlations.
The three-dimensional spatial accurate propagation model of signal source based on symmetric cross-arrays is adopted to realize high-precision estimation of signal source parameters through cross-correlation calculation and vectorization operation, including establishing a three-dimensional parameter model of narrowband signal sources and calculating the cross-correlation matrix of received data. Perform vectorization operations and eigenvalue decomposition to obtain spatial spectral functions for peak search.
High-precision signal source parameter estimation is realized, and it can eliminate parameter matching in the case of far-field and near-field mixed sources, distinguish more signal sources, and improve estimation performance.
Smart Images

Figure CN119024267A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a source positioning technology, in particular to an underdetermined estimation method for far-field and near-field mixed source parameters based on cross-correlation. Background Art
[0002] Source localization is an important branch of array signal processing, which has been widely used in many fields such as radar, sonar, mobile communications, etc. According to the distance from the signal source to the array, the signal source can be divided into far-field source and near-field source. The far-field source is modeled as a plane wave, and the phase difference is linearly related to the position of the array element, while the spherical wave phase difference of the near-field source is nonlinear, so it is more complicated to realize the localization of the near-field source.
[0003] Far-field source localization requires the estimation of angle parameters, while near-field source localization requires the estimation of angle and distance parameters at the same time. The far-field source localization algorithm has been developed and mature. In order to use it for the localization of near-field sources, many studies have ignored amplitude attenuation and performed Fresnel approximation on the phase difference. However, the estimation accuracy of the proposed near-field source localization algorithm will be limited, and there is currently no accurate propagation model that can break through this limitation. In three-dimensional scenarios, near-field source localization algorithms are usually based on symmetric arrays and Fresnel approximation, and few algorithms consider amplitude attenuation. In addition, the use of time information of received data can improve the estimation performance of the positioning algorithm. Although many parameter algorithms make full use of space-time information, they are still based on approximate models and cannot achieve high accuracy. In actual scenarios, most of them are mixed far-field sources and near-field sources, and there are few studies on the underdetermined estimation of mixed source parameters, that is, the number of sources is greater than the number of array elements. In the localization of far-field and near-field mixed sources, no underdetermined estimation method of far-field and near-field mixed source parameters based on cross-correlation has been proposed. Summary of the invention
[0004] The technical problem to be solved by the present invention is to provide a method for underdetermined estimation of far-field and near-field mixed source parameters based on cross-correlation, which can realize underdetermined estimation of source parameters without parameter matching and has high estimation accuracy.
[0005] The technical solution adopted by the present invention to solve the above technical problems is: a method for underdetermined estimation of far-field and near-field mixed source parameters based on cross-correlation, characterized by comprising the following steps:
[0006] Step 1: In a near-field and far-field mixed source positioning scenario, a three-dimensional accurate propagation model of a signal source based on a symmetrical cross array is established, wherein the model includes K narrowband signal sources and a symmetrical cross array;
[0007] The K narrowband signal sources include K1 near-field sources and K2 far-field sources. The three-dimensional parameters of the kth narrowband signal source are recorded as (α k ,β k ,r k); where K1>1, K2>1, K=K1+K2, k=1,2,…,K, α k represents the angle between the incident direction of the kth narrowband signal source and the positive direction of the X-axis in three-dimensional space, α k ∈[0,π],β k represents the angle between the incident direction of the kth narrowband signal source and the positive direction of the Y axis in three-dimensional space, β k ∈[0,π],r k represents the distance between the kth narrowband signal source and the origin of the three-dimensional space. When r k ∈(0,2D 2 / λ], the kth narrowband signal source is a near-field source, and when r k ∈(2D 2 / λ,+∞), the kth narrowband signal source is a far-field source, D represents the array aperture of the symmetrical cross array, and λ is the signal wavelength;
[0008] The symmetric cross array is established on the XOY plane of the three-dimensional space. The symmetric cross array consists of array x located on the X axis of the three-dimensional space and array y located on the Y axis of the three-dimensional space. Both array x and array y are symmetric uniform linear arrays. The total number of array elements in array x is 2M. x +1, the number of single-sided array elements of array x is M x , the distance between adjacent array elements in array x is d, the coordinates of the array element with index m in array x are (md, 0), and the distance between the array element with index m in array x and the kth narrowband signal source is r m,k , The total number of array elements in array y is 2M y +1, the number of single-sided array elements of array y is M y , the distance between adjacent array elements in array y is d, the coordinates of the array element with index n in array y are (0,nd), and the distance between the array element with index n in array y and the kth narrowband signal source is r n,k , Array x and array y share the same array element located at the origin of the three-dimensional space; where M x ≥1, d=λ / 4, m=(-M x ,-M x +1,…,0,1,…,M x ), My ≥ 1, n = (-M y ,-M y +1,…,0,1,…,M y ), 2M x +2M y +1<K;
[0009] K narrowband signal sources are incident on the symmetrical cross array at the same time, and the received data x of the array element with index m in the array x m(t) is expressed as Received data vector of array x Expressed as The received data y of the array element with index n in array y n (t) is expressed as Received data vector of array y Expressed as Among them, t represents the time, a m,k represents the amplitude phase factor of the array element with index m in array x for the kth narrowband signal source, e is a natural constant, j is an imaginary unit, s k (t) represents the incident signal from the kth narrowband signal source, w m (t) represents the additive Gaussian white noise added to the element with index m in the array x. The dimension of x(t) is (2M x +1)×1, (·) T represents the transpose operation, A represents the manifold matrix of array x, and the dimension of A is (2M x +1)×K, A=[a(α1,r1),…,a(α k ,r k ),…,a(α K ,r K )],a(α k ,r k ) represents the steering vector associated with the array x and the kth narrowband signal source, a(α k ,r k ) has a dimension of (2M x +1)×1, s(t) represents the incident signal vector, the dimension of s(t) is K×1, s(t)=[s1(t),…,s K (t)] T , w x (t) represents the noise vector added to the array x, w x The dimension of (t) is (2M x +1)×1, b n,k represents the amplitude phase factor of the array element with index n in array y for the kth narrowband signal source, w n (t) represents the additive Gaussian white noise added to the element with index n in the array y. The dimension of y(t) is (2M y +1)×1, B represents the manifold matrix of array y, and the dimension of B is (2M y +1)×K, B=[b(β1,r1),…,b(β k ,r k ),…,b(β K ,rK )],b(β k ,r k ) represents the steering vector associated with the k-th narrowband signal source, b(β k ,r k ) has a dimension of (2M y +1)×1, w y (t) represents the noise vector added to the array y, w y The dimension of (t) is (2M y +1)×1,
[0010] Step 2: Based on the accurate three-dimensional propagation model of the signal source based on the symmetric cross array, calculate the received data x of the array element with index m in the array x m (t) and the received data y of the array element with index n in array y n The cross-correlation of (t) is r(m,n,τ), Then take m=-M x ,…,0,…,M x and n=-M y ,…,0,…,M y , r(m,n,τ),m=-M x ,…,0,…,M x ,n=-M y ,…,0,…,M y Arrange and get the cross-correlation matrix R(τ), According to the expression of r(m,n,τ) and the arrangement position of each element in R(τ), we can get R(τ)=AR s (τ)B H +R ω (τ); then R(τ)=AR s (τ)B H +R ω (τ) is vectorized to obtain the vectorized result r(τ) of R(τ), r(τ)=vec{R(τ)}=(B * ⊙A)r s (τ)+vec{R ω (τ)}; finally, take τ=T s ,2T s ,…,LT s , get the virtual receiving data matrix Among them, τ is the delay variable, E[·] represents the mathematical expectation, and x m (t) After a delay of τ, we get x m (t+τ), (·) * is the complex conjugate operation, sk (t) After a delay of τ, we get s k (t+τ), δ(·) is the Dirichlet function, represents the noise power, r sk (τ) represents the autocorrelation of the k-th narrowband signal source, The dimension of R(τ) is (2M x +1)×(2M y +1), R s (τ) represents the autocorrelation matrix of the narrowband signal source, R s The dimension of (τ) is K×K, R s (τ) = diag{r s1 (τ),r s2 (τ),…,r sK (τ)}, diag{·} represents the construction of a diagonal matrix, (·) H is the conjugate transpose operation, R ω (τ) represents the noise covariance matrix, R ω The dimension of (τ) is (2M x +1)×(2M y +1), R ω The subscript position in (τ) is (M x +1,M y +1) is The remaining elements are all 0, vec{·} represents vectorized operation, ⊙ represents Khatri-Rao product operation, r s (τ) = vec{R s (τ)}=[r s1 (τ),r s2 (τ),…,r sK (τ)] T , T s represents the virtual sampling period, L represents the number of pseudo snapshots,
[0011] Step 3: Calculation The covariance matrix of Among them, B * The kth column vector of ⊙A is is the Kronecker product operation;
[0012] Step 4: Right Perform eigenvalue decomposition and take the smallest (2M x +1)×(2M y +1)-K eigenvalues; then the eigenvectors corresponding to these eigenvalues are used to form the noise subspace U N , the steering vector belongs to the signal subspace; then use the noise subspace U N The orthogonality of the signal subspace is taken into account, and according to the range of α α∈[0,π], the range of β β∈[0,π] and the range of r r∈(0,+∞), the spatial spectrum function P(α,β,r) about α, β and r is obtained. Finally, by performing a three-dimensional spectrum peak search on P(α, β, r), K peaks are obtained, and the parameter corresponding to the position of the kth peak is the estimated value of α Estimation of β Estimated value of r like Then the kth narrowband signal source is a near-field source, if Then the kth narrowband signal source is a far-field source; where α is the angle variable between the incident direction of the narrowband signal source and the positive direction of the X-axis in the three-dimensional space, β is the angle variable between the incident direction of the narrowband signal source and the positive direction of the Y-axis in the three-dimensional space, and r is the distance variable between the narrowband signal source and the origin of the three-dimensional space.
[0013] Compared with the prior art, the advantages of the present invention are:
[0014] 1) The method of the present invention is based on the precise propagation model of the signal source in three-dimensional space, making full use of the spatial and temporal information of the received data. Compared with the algorithm of the approximate model, it can achieve higher signal source parameter estimation accuracy without the need for parameter matching.
[0015] 2) Compared with the existing algorithms, the method of the present invention adopts cross-correlation calculation and vectorization operation, which further improves the degree of freedom of the symmetrical cross array, can realize the underdetermined estimation of the far-field and near-field mixed source parameters, and can distinguish more signal sources while having better estimation performance. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 It is a general implementation block diagram of the method of the present invention;
[0017] Figure 2 A schematic diagram of a three-dimensional accurate propagation model of a signal source based on a symmetrical cross array established in the method of the present invention;
[0018] Figure 3 This is a diagram of the three-dimensional parameter estimation result of the near-field source obtained by using the method of the present invention in a simulation experiment;
[0019] Figure 4 This is a diagram of the estimation results of the angle parameters of the far-field and near-field mixed sources obtained by using the method of the present invention in a simulation experiment. DETAILED DESCRIPTION
[0020] The present invention is further described in detail below with reference to the accompanying drawings.
[0021] The present invention proposes a method for underdetermined estimation of parameters of near-field and far-field mixed sources based on cross-correlation, and its overall implementation block diagram is as follows: Figure 1 As shown, it includes the following steps:
[0022] Step 1: Figure 2 As shown, in a far-field and near-field mixed source positioning scenario, a three-dimensional spatial accurate propagation model of a signal source based on a symmetrical cross array is established, and the model includes K narrowband signal sources and a symmetrical cross array.
[0023] The K narrowband signal sources include K1 near-field sources and K2 far-field sources. The three-dimensional parameters of the kth narrowband signal source are recorded as (α k ,β k ,r k ); where K1>1, K2>1, K=K1+K2, k=1,2,…,K, α k represents the angle between the incident direction of the kth narrowband signal source and the positive direction of the X-axis in three-dimensional space, α k ∈[0,π],β k represents the angle between the incident direction of the kth narrowband signal source and the positive direction of the Y axis in three-dimensional space, β k ∈[0,π],r k represents the distance between the kth narrowband signal source and the origin of the three-dimensional space. When r k ∈(0,2D 2 / λ], the kth narrowband signal source is a near-field source, and when r k ∈(2D 2 / λ,+∞), the kth narrowband signal source is a far-field source, D represents the array aperture of the symmetrical cross array, and λ is the signal wavelength.
[0024] The symmetric cross array is established on the XOY plane of the three-dimensional space. The symmetric cross array consists of array x located on the X axis of the three-dimensional space and array y located on the Y axis of the three-dimensional space. Both array x and array y are symmetric uniform linear arrays. The total number of array elements in array x is 2M. x +1, the number of single-sided array elements of array x is M x , the distance between adjacent array elements in array x is d, the coordinates of the array element with index m in array x are (md, 0), and the distance between the array element with index m in array x and the kth narrowband signal source is r m,k According to the cosine theorem, we can get The total number of array elements in array y is 2M y +1, the number of single-sided array elements of array y is M y , the distance between adjacent array elements in array y is d, the coordinates of the array element with index n in array y are (0,nd), and the distance between the array element with index n in array y and the kth narrowband signal source is r n,k According to the cosine theorem, we can get Array x and array y share the same array element located at the origin of the three-dimensional space; where M x ≥1, d=λ / 4, m=(-M x ,-M x +1,…,0,1,…,M x ), M y ≥1, n=(-M y ,-M y +1,…,0,1,…,M y ), 2M x +2M y +1<K.
[0025] K narrowband signal sources are incident on the symmetrical cross array at the same time, and the received data x of the array element with index m in the array x m (t) is expressed as Received data vector of array x Expressed as The received data y of the array element with index n in array y n (t) is expressed as Received data vector of array y Expressed as Among them, t represents the time, a m,k represents the amplitude phase factor of the array element with index m in array x for the kth narrowband signal source, e is a natural constant, e=2.71…, j is an imaginary unit, s k (t) represents the incident signal from the kth narrowband signal source, w m (t) represents the additive Gaussian white noise added to the element with index m in the array x. The dimension of x(t) is (2M x +1)×1, (·) T represents the transpose operation, Indicates that the index in array x is -M x The received data of the array element of x0(t) represents the received data of the array element with index 0 in array x. Indicates that the index in array x is M x The received data of the array element, A represents the manifold matrix of array x, and the dimension of A is (2M x +1)×K, A=[a(α1,r1),…,a(α k ,r k ),…,a(α K ,r K )],a(α k ,r k ) represents the steering vector associated with the array x and the kth narrowband signal source, a(α k ,r k) has a dimension of (2M x +1)×1, Indicates that the index in array x is -M x The amplitude phase factor of the array element to the kth narrowband signal source, Indicates that the index in array x is M x The amplitude phase factor of the array element to the kth narrowband signal source, s(t) represents the incident signal vector, the dimension of s(t) is K×1, s(t)=[s1(t),…,s K (t)] T , s1(t) represents the signal incident from the first narrowband signal source, s K (t) represents the incident signal from the Kth narrowband signal source, w x (t) represents the noise vector added to the array x, w x The dimension of (t) is (2M x +1)×1, Indicates appending to array x with index -M x The additive Gaussian white noise on the array element, w0(t) represents the additive Gaussian white noise added to the array element with index 0 in the array x, Indicates appending to array x with index M x Additive Gaussian white noise on the array element, b n,k represents the amplitude phase factor of the array element with index n in array y for the kth narrowband signal source, w n (t) represents the additive Gaussian white noise added to the element with index n in the array y. The dimension of y(t) is (2M y +1)×1, Indicates that the index in array y is -M y y0(t) represents the received data of the array element with index 0 in array y. Indicates that the index in array y is M y The received data of the array element, B represents the manifold matrix of array y, and the dimension of B is (2M y +1)×K, B=[b(β1,r1),…,b(β k ,r k ),…,b(β K ,r K )],b(β k ,r k ) represents the steering vector associated with the k-th narrowband signal source, b(β k ,r k ) has a dimension of (2M y +1)×1, Indicates that the index in array y is -M y The amplitude phase factor of the array element to the kth narrowband signal source, Indicates that the index in array y is M y The amplitude phase factor of the array element to the kth narrowband signal source, w y (t) represents the noise vector added to the array y, w y The dimension of (t) is (2M y +1)×1, represents the additive Gaussian white noise attached to the element with index -My in the array y, w0(t) represents the additive Gaussian white noise attached to the element with index 0 in the array y, Indicates that the index in array y is M. y Additive white Gaussian noise on the array elements.
[0026] Step 2: Based on the accurate three-dimensional propagation model of the signal source based on the symmetric cross array, calculate the received data x of the array element with index m in the array x m (t) and the received data y of the array element with index n in array y n The cross-correlation of (t) is r(m,n,τ), Then take m=-M x ,…,0,…,M x and n=-M y ,…,0,…,M y , r(m,n,τ),m=-M x ,…,0,…,M x ,n=-M y ,…,0,…,M y Arrange and get the cross-correlation matrix R(τ), According to the expression of r(m,n,τ) and the arrangement position of each element in R(τ), we can get R(τ)=AR s (τ)B H +R ω (τ); then R(τ)=AR s (τ)B H +R ω (τ) is vectorized to obtain the vectorized result r(τ) of R(τ), r(τ)=vec{R(τ)}=(B * ⊙A)r s (τ)+vec{R ω (τ)}; finally, take τ=T s ,2T s ,…,LT s , get the virtual receiving data matrix Among them, τ is the delay variable, E[·] represents the mathematical expectation, and x m (t) After a delay of τ, we get x m (t+τ), (·) * is the complex conjugate operation, s k (t) After a delay of τ, we get s k (t+τ), δ(·) is the Dirichlet function, represents the noise power, r sk (τ) represents the autocorrelation of the k-th narrowband signal source, The dimension of R(τ) is (2M x +1)×(2M y +1), r(-M x ,-M y ,τ) indicates that the index in array x is -M x The received data of the array element with the array y indexed as -M y The received data of the array element The cross-correlation of r(-M x ,M y ,τ) indicates that the index in array x is -M x The received data of the array element with the index in array y being M y The received data of the array element The cross-correlation of x ,-M y ,τ) indicates that the index in array x is M x The received data of the array element with the array y indexed as -M y The received data of the array element The cross-correlation of x ,M y ,τ) indicates that the index in array x is M x The received data of the array element with the index in array y being M y The received data of the array element The cross-correlation, R s (τ) represents the autocorrelation matrix of the narrowband signal source, R s The dimension of (τ) is K×K, R s (τ) = diag{r s1 (τ),r s2 (τ),…,r sK (τ)}, diag{·} represents the construction of a diagonal matrix, r s1 (τ) represents the autocorrelation of the first narrowband signal source, r s2 (τ) represents the autocorrelation of the second narrowband signal source, rsK (τ) represents the autocorrelation of the Kth narrowband signal source, (·) H is the conjugate transpose operation, R ω (τ) represents the noise covariance matrix, R ω The dimension of (τ) is (2M x +1)×(2M y +1), R ω The subscript position in (τ) is (M x +1,M y +1) is The remaining elements are all 0, vec{·} represents vectorized operation, ⊙ represents Khatri-Rao product operation, r s (τ) = vec{R s (τ)}=[r s1 (τ),r s2 (τ),…,r sK (τ)] T , T s represents the virtual sampling period, L represents the number of pseudo snapshots, and r(τ) contains τ=T s When r(T s ), where τ = 2T s When r(2T s ), where τ = LT s When r(LT s ), r s (τ) where τ = T s When r s (T s ), r s (τ) where τ = 2T s When r s (2T s ), r s (τ) where τ = LT s When r s (LT s ).
[0027] Step 3: Calculation The covariance matrix of Among them, B * The kth column vector of ⊙A is Also the steering vector, is the Kronecker product operation.
[0028] Step 4: Right Perform eigenvalue decomposition and take the smallest (2M x +1)×(2M y+1)-K eigenvalues; then the eigenvectors corresponding to these eigenvalues are used to form the noise subspace U N , the steering vector belongs to the signal subspace; then use the noise subspace U N The orthogonality of the signal subspace is taken into account, and according to the range of α α∈[0,π], the range of β β∈[0,π] and the range of r r∈(0,+∞), the spatial spectrum function P(α,β,r) about α, β and r is obtained. Finally, by performing a three-dimensional spectrum peak search on P(α, β, r), K peaks are obtained, and the parameter corresponding to the position of the kth peak is the estimated value of α Estimation of β Estimated value of r like Then the kth narrowband signal source is a near-field source, if Then the kth narrowband signal source is a far-field source; where α is the angle variable between the incident direction of the narrowband signal source and the positive direction of the X-axis in the three-dimensional space, β is the angle variable between the incident direction of the narrowband signal source and the positive direction of the Y-axis in the three-dimensional space, r is the distance variable between the narrowband signal source and the origin of the three-dimensional space, and a(α k ,r k ) k ,r k Substituting α and r, we get a(α, r), b(β k ,r k ) k ,r k Substituting β,r gives b(β,r).
[0029] In order to verify the effectiveness and accuracy of the method of the present invention, a simulation test is carried out on the method of the present invention, as follows:
[0030] Set the number of single-sided array elements of array x to be equal to the number of single-sided array elements of array y, which is M. x =M y =2, the total number of array elements of array x is 2M x +1=5, the total number of array elements in array y is 2M y +1=5, the total number of array elements of the symmetrical cross array is 2M x +2M y+1=9, the spacing between adjacent array elements is d=λ / 4, the array aperture is D=λ, the distance range between the near-field source and the origin of the three-dimensional space is (0,2λ], the distance range between the far-field source and the origin of the three-dimensional space is (2λ,+∞), and the signal wavelength is λ=1. Assume that there are K=10 narrow-band signal sources, and their actual three-dimensional parameters (where the angle parameter values are expressed as angle values) are (45°,90°,1.6λ), (35°,155°,1.3λ), (50°,135°,1.2λ), (65 °,115°,1.5λ), (80°,85°,1.4λ), (130°,45°,1100λ), (150°,65°,1500λ), (165°,25°,1900λ), (10°,170°,1300λ), (60°,60°,1700λ), including K1=5 near-field sources and K2=5 far-field sources, incident on the symmetrical cross array. Set the signal-to-noise ratio SNR to 30dB, the number of snapshots to 2000, and the number of pseudo snapshots to L=50.
[0031] The method of the present invention is used to perform underdetermined estimation of far-field and near-field mixed source parameters based on cross-correlation. Figure 3 The three-dimensional parameter estimation results of five near-field sources are given. Figure 3 It can be seen that the three-dimensional parameters of the five near-field sources can be estimated correctly. Figure 4 The angle parameter estimation results of 5 near-field sources and 5 far-field sources are given. Figure 4 It can be seen from the figure that when the number of signal sources is greater than the number of array elements, the angle parameters of the 10 signal sources can be estimated correctly. The method of the present invention can achieve underdetermined estimation of near-field and far-field mixed source parameters, and the parameters are automatically paired.
Claims
1. A method for underdetermined estimation of near-field and far-field mixed source parameters based on cross-correlation, characterized in that The steps include: Step 1: In a near-field and far-field mixed source positioning scenario, a three-dimensional accurate propagation model of a signal source based on a symmetrical cross array is established, wherein the model includes K narrowband signal sources and a symmetrical cross array; The K narrowband signal sources include K1 near-field sources and K2 far-field sources. The three-dimensional parameters of the kth narrowband signal source are recorded as (α k ,β k ,r k ); where K1>1, K2>1, K=K1+K2, k=1,2,…,K, α k represents the angle between the incident direction of the kth narrowband signal source and the positive direction of the X-axis in three-dimensional space, α k ∈[0,π],β k represents the angle between the incident direction of the kth narrowband signal source and the positive direction of the Y axis in three-dimensional space, β k ∈[0,π],r k represents the distance between the kth narrowband signal source and the origin of the three-dimensional space. When r k ∈(0,2D 2 / λ], the kth narrowband signal source is a near-field source, and when r k ∈(2D 2 / λ,+∞), the kth narrowband signal source is a far-field source, D represents the array aperture of the symmetrical cross array, and λ represents the signal wavelength; The symmetric cross array is established on the XOY plane of the three-dimensional space. The symmetric cross array consists of array x located on the X axis of the three-dimensional space and array y located on the Y axis of the three-dimensional space. Both array x and array y are symmetric uniform linear arrays. The total number of array elements in array x is 2M. x +1, the number of single-sided array elements of array x is M x , the distance between adjacent array elements in array x is d, the coordinates of the array element with index m in array x are (md, 0), and the distance between the array element with index m in array x and the kth narrowband signal source is r m,k , The total number of array elements in array y is 2M y +1, the number of single-sided array elements of array y is M y , the distance between adjacent array elements in array y is d, the coordinates of the array element with index n in array y are (0,nd), and the distance between the array element with index n in array y and the kth narrowband signal source is r n,k , Array x and array y share the same array element located at the origin of the three-dimensional space; where M x ≥1, d=λ / 4, m=(-M x ,-M x +1,…,0,1,…,M x ), M y ≥1, n=(-M y ,-M y +1,…,0,1,…,M y ), 2M x +2M y +1<K; K narrowband signal sources are incident on the symmetrical cross array at the same time, and the received data x of the array element with index m in the array x m (t) is expressed as Received data vector of array x Expressed as The received data y of the array element with index n in array y n (t) is expressed as Received data vector of array y Expressed as Among them, t represents the time, a m,k represents the amplitude phase factor of the array element with index m in array x for the kth narrowband signal source, e is a natural constant, j is an imaginary unit, s k (t) represents the incident signal from the kth narrowband signal source, w m (t) represents the additive Gaussian white noise added to the element with index m in the array x. The dimension of x(t) is (2M x +1)×1, (·) T represents the transpose operation, A represents the manifold matrix of the array x, and the dimension of A is (2M x +1)×K, A=[a(α1,r1),…,a(α k ,r k ),…,a(α K ,r K )],a(α k ,r k ) represents the steering vector associated with the array x and the kth narrowband signal source, a(α k ,r k ) has a dimension of (2M x +1)×1, 1≤k≤K, s(t) represents the incident signal vector, the dimension of s(t) is K×1, s(t)=[s1(t),…,s K (t)] T , w x (t) represents the noise vector added to the array x, w x The dimension of (t) is (2M x +1)×1, b n,k represents the amplitude phase factor of the array element with index n in array y for the kth narrowband signal source, w n (t) represents the additive Gaussian white noise added to the element with index n in the array y. The dimension of y(t) is (2M y +1)×1, B represents the manifold matrix of array y, and the dimension of B is (2M y +1)×K, B=[b(β1,r1),…,b(β k ,r k ),…,b(β K ,r K )],b(β k ,r k ) represents the steering vector associated with the k-th narrowband signal source, b(β k ,r k ) has a dimension of (2M y +1)×1, 1≤k≤K,w y (t) represents the noise vector added to the array y, w y The dimension of (t) is (2M y +1)×1, Step 2: Based on the accurate three-dimensional propagation model of the signal source based on the symmetric cross array, calculate the received data x of the array element with index m in the array x m (t) and the received data y of the array element with index n in array y n The cross-correlation of (t) is r(m,n,τ), Then take m=-M x ,…,0,…,M x and n=-M y ,…,0,…,M y , r(m,n,τ),m=-M x ,…,0,…,M x ,n=-M y ,…,0,…,M y Arrange and get the cross-correlation matrix R(τ), According to the expression of r(m,n,τ) and the arrangement position of each element in R(τ), we can get R(τ)=AR s (τ)B H +R ω (τ); then R(τ)=AR s (τ)B H +R ω (τ) is vectorized to obtain the vectorized result r(τ) of R(τ), r(τ)=vec{R(τ)}=(B * ⊙A)r s (τ)+vec{R ω (τ)}; finally, take τ=T s ,2T s ,…,LT s , get the virtual receiving data matrix Among them, τ is the delay variable, E[·] represents the mathematical expectation, and x m (t) After a delay of τ, we get x m (t+τ), (·) * is the complex conjugate operation, s k (t) After a delay of τ, we get s k (t+τ), δ(·) is the Dirichlet function, represents the noise power, r sk (τ) represents the autocorrelation of the kth narrowband signal source, The dimension of R(τ) is (2M x +1)×(2M y +1), R s (τ) represents the autocorrelation matrix of the narrowband signal source, R s The dimension of (τ) is K×K, R s (τ) = diag{r s1 (τ),r s2 (τ),…,r sK (τ)}, diag{·} represents the construction of a diagonal matrix, (·) H is the conjugate transpose operation, R ω (τ) represents the noise covariance matrix, R ω The dimension of (τ) is (2M x +1)×(2M y +1), R ω The subscript position in (τ) is (M x +1,M y +1) is The remaining elements are all 0, vec{·} represents vectorized operation, ⊙ represents Khatri-Rao product operation, r s (τ) = vec{R s (τ)}=[r s1 (τ),r s2 (τ),…,r sK (τ)] T , T s represents the virtual sampling period, L represents the number of pseudo snapshots, Step 3: Calculation The covariance matrix of Among them, B * The kth column vector of ⊙A is is the Kronecker product operation; Step 4: Right Perform eigenvalue decomposition and take the smallest (2M x +1)×(2M y +1)-K eigenvalues; then the eigenvectors corresponding to these eigenvalues are used to form the noise subspace U N , the steering vector belongs to the signal subspace; then use the noise subspace U N The orthogonality of the signal subspace is taken into account, and according to the range of α α∈[0,π], the range of β β∈[0,π] and the range of r r∈(0,+∞), the spatial spectrum function P(α,β,r) about α, β and r is obtained. Finally, by performing a three-dimensional spectrum peak search on P(α, β, r), K peaks are obtained, and the parameter corresponding to the position of the kth peak is the estimated value of α Estimation of β Estimated value of r like Then the kth narrowband signal source is a near-field source, if Then the kth narrowband signal source is a far-field source; where α is the angle variable between the incident direction of the narrowband signal source and the positive direction of the X-axis in the three-dimensional space, β is the angle variable between the incident direction of the narrowband signal source and the positive direction of the Y-axis in the three-dimensional space, and r is the distance variable between the narrowband signal source and the origin of the three-dimensional space.
