Parameter estimation method for arbitrary array element structure EMVS (Empirical Mode Voltage Suppressor) array under near-field multipath

Abstract

The invention discloses a near-field multi-path EMVS (Empirical Mode Voltage Suppressor) array parameter estimation method with any array element structure, which comprises the following specific steps of: firstly, performing discrete Fourier transform on receiving data of the EMVS array with any array element structure, and reconstructing a new receiving data model by using space-time-polarization information in the receiving data; performing trilinear decomposition on the reconstructed data model to obtain estimated values of an array manifold matrix and a delay response matrix; secondly, obtaining estimated values of 2D-DOA and TD at each EMVS by using a rotation invariant relation of a matrix, and obtaining an estimated value of a polarization parameter by using a least square method; and finally, constructing a linear equation to obtain estimated values of near-field multipath 2D-DOA and distance parameters relative to a phase reference point. The method provided by the invention can realize accurate multi-dimensional parameter estimation of near-field multipath, has the advantages of no need of spectral peak search and no need of additional parameter pairing calculation, and is suitable for an EMVS array with any array element interval and geometric shape.

Description

Technical Field

[0001] The invention belongs to the field of array signal processing, and in particular to a parameter estimation method of an EMVS array with arbitrary array element structure under near-field multipath. Background Art

[0002] Source localization is an important research topic in the field of array signal processing, and is widely used in wireless communications, radar, sonar and other fields. However, due to the existence of multipath propagation, especially in urban and indoor environments, accurate positioning under multipath effects has always been a challenging problem. In multipath scenarios, each path reaching the receiving array can be described by parameters such as direction of arrival (DOA), time delay (TD), amplitude (or fading) and polarization state. Compared with the traditional method of estimating DOA only, the combined DOA and TD estimation algorithm makes full use of spatial diversity and time diversity, and has significant advantages in estimation performance.

[0003] Electromagnetic vector sensor (EMVS) has the advantage of polarization diversity. Polarization information, as an inherent feature of wireless electromagnetic signals, is usually combined with space-time diversity to improve the performance of wireless communication. Compared with scalar sensors, EMVS arrays provide additional degrees of freedom for closely spaced paths in space and help detect weak signals. The multi-dimensional parameter estimation method of joint polarization-angle-delay has triggered extensive research and discussion by scholars at home and abroad. For example, in 2004, J. PICHERAL et al. estimated multipath channels based on uniform cross-dipole arrays using the Estimating Signal Parameter via Rotational Invariance Techniques (ESPRIT) method; in 2015, S. HAFNER et al. estimated parameters by optimizing the cost function constructed based on the cross relationship of the receiving channel under multipath propagation. However, the above algorithms are limited to linear or uniform linear array element distribution, and in practice, when the space is limited, it may not meet the distribution requirements of the linear array structure. In 2022, J. He et al. studied a joint polarization-angle-delay parameter estimation method for spatially co-located tripole antennas, but the algorithm cannot perform reliable estimation when the elevation angle or polarization angle approaches zero. In addition, all of the aforementioned algorithms are for far-field sources. When the source is in the near field, the transmission model cannot be simply assumed to be a plane wave, and the DOA and distance of the source need to be estimated to locate the source. In view of the shortcomings of the above methods, a new method is needed to overcome the shortcomings of existing methods in terms of applicable scenarios, parameter estimation performance, and array element structure spacing. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide an EMVS array parameter estimation method for arbitrary array element structure for three-dimensional near-field multipath signals. A space-time-polarization information matrix is ​​constructed based on an accurate source-sensor space model. Through trilinear decomposition and the rotational invariant relationship of the matrix, the joint estimation of multipath two-dimensional arrival direction, distance, polarization and delay parameters is realized. The method is suitable for EMVS arrays with arbitrary array element spacing and geometric shapes.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] The parameter estimation method of EMVS array with arbitrary element structure under near-field multipath includes:

[0007] Step 1: Establish a data model of near-field multipath incident on an EMVS array of arbitrary structure in three-dimensional space;

[0008] Step 2: Perform discrete Fourier transform on the array received data and use the space-time-polarization information to reconstruct a new received data model;

[0009] Step 3: Perform trilinear decomposition on the covariance matrix of the rearranged data to obtain the estimated values ​​of the array manifold matrix and the delayed response matrix;

[0010] Step 4: Use the rotation invariance of the array manifold matrix and the vector cross multiplication technique to obtain the 2D-DOA estimate at each EMVS;

[0011] Step 5: Use the rotation invariance of the delay response matrix to obtain the estimated value of TD;

[0012] Step 6: Estimate the polarization parameters using the least squares method;

[0013] Step 7: Construct a linear equation based on the sensor array position information to estimate the position parameters of the near-field multipath.

[0014] In step 1 above, let L represent the number of multipaths, let θ l ,φ l , γ l , η l and r l They represent the elevation angle, azimuth angle, polarization auxiliary angle, polarization phase difference and distance of the lth multipath relative to the phase reference point. The near-field multipath signal received by an N-element arbitrary structure EMVS array in three-dimensional space at time t is expressed as:

[0015]

[0016] in, is the received signal of the ith pole of the nth EMVS; βl is the complex coefficient of the attenuation factor of the lth path; τ l is the transmission time delay of the lth path; s(t) represents the known signal waveform; n(t) is the zero-mean additive white Gaussian noise independent of the signal; a(θ l ,φ l ,γ l ,η l ) represents the spatial polarization steering vector of the array for the lth path.

[0017] The above step 2 includes:

[0018] 1) Perform discrete Fourier transform (DFT) of K frequency points on the received signal of each polar component of the sensor. The data model expression of the kth frequency point is as follows:

[0019] Y(ω k )=ABf(ω k )+n(ω k )(4)

[0020] Among them, Y(ω k ) and n(ω k ) are the DFTs of Y(t) and n(t), respectively. Manifold matrix representing array; represents the path fading factor matrix; represents the delayed response matrix at the kth frequency point;

[0021] 2) After rearranging the data of K frequency points, the model is constructed as follows:

[0022] V=(F1 T ⊙A)E+W(8)

[0023] in represents the delayed response matrix after rearrangement, E can be regarded as the signal source vector of the new received signal matrix V, and W represents the corresponding noise after the data is rearranged.

[0024] Step 3 above includes:

[0025] The covariance matrix of the rearranged data model V is calculated as:

[0026] R=(F1 T ⊙A)R E (F1 T ⊙A) H +σ n 2 I 6NP (9)

[0027] R can be expressed as a third-order PARAFAC model, and the delayed response matrix F1 can be obtained simultaneously through trilinear decomposition. T and an estimate of the array manifold matrix A.

[0028] Step 4 above includes:

[0029] Defining the selection matrix Among them I p and i p,q Represents the p×p identity matrix and the identity matrix I p The qth row of . Define the selection matrix Further we get the following relationship:

[0030]

[0031] in is the rotation invariant factor.

[0032] Using the estimated rotation invariant factor, we calculate it by vector cross product:

[0033]

[0034] in is the estimated rotation invariant factor, and # represents vector cross product.

[0035] From this, the 2D-DOA estimate at the nth EMVS can be calculated:

[0036]

[0037] In step 5 above, define the selection matrix and The following relationship can be obtained:

[0038]

[0039] in Therefore, the estimated value of multipath time delay can be calculated as follows:

[0040]

[0041] The above step 6 includes:

[0042] After obtaining the 2D-DOA estimate at the nth EMVS, the least squares method is used to calculate:

[0043]

[0044] in, Indicates a s (θ n,l,φ n,l ,γ l ,η l ), from which the estimated value of the polarization parameter at the nth EMVS can be calculated as:

[0045]

[0046] After obtaining the estimated polarization parameters at all EMVS, the average value is taken to obtain the final estimated polarization parameter value:

[0047]

[0048] The above step 7 includes:

[0049] According to the geometric relationship of the sensor array position, the parameter equation related to the sensor array position and angle is established as follows:

[0050] Η l Ο l =Γ l (19)

[0051] in,

[0052]

[0053] Using the least squares method, we get l Estimated value of

[0054]

[0055] Therefore, the 2D-DOA and distance calculation formula of the corresponding l-th near-field multipath at the phase reference point are as follows:

[0056]

[0057] The beneficial effects of the present invention are:

[0058] Aiming at the parameter estimation problem under the near-field multipath effect in three-dimensional space, the present invention proposes an algorithm for joint estimation of the wave angle-polarization-delay parameters based on an EMVS array of arbitrary structure. The present invention establishes an accurate source-sensor spatial model; performs discrete Fourier transform on the array received data, and reconstructs a new received data model using the space-time-polarization information therein; performs trilinear decomposition on the reconstructed data model to obtain the estimated values ​​of the array manifold matrix and the delay response matrix; then uses the rotational invariant relationship of the matrix to obtain the estimated values ​​of the 2D-DOA and TD at each EMVS, and uses the least squares method to obtain the estimated values ​​of the polarization parameters; finally, a linear equation is constructed to obtain the estimated values ​​of the 2D-DOA and distance parameters of the near-field multipath relative to the phase reference point. Numerical and simulation results show that the method proposed by the present invention can realize accurate multi-dimensional parameter estimation of near-field multipath, and does not require spectral peak search, does not require additional parameter pairing calculation, has no restrictions on the array element spacing and geometric shape of the EMVS array, and has better parameter estimation performance than existing methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 Schematic diagram of near-field multipath positioning by an EMVS array with arbitrary array element structure;

[0060] Figure 2 2D-DOA and TD estimation scatter plot of the method proposed in the present invention;

[0061] Figure 3 It is a scatter plot of polarization parameter estimation of the method proposed in the present invention;

[0062] Figure 4 Schematic diagram of the comparison of RMSE of 2D-DOA estimation as the signal-to-noise ratio changes.

[0063] Figure 5 Schematic diagram of the comparison of RMSE of polarization parameter estimation as the signal-to-noise ratio changes.

[0064] Figure 6 Schematic diagram of the comparison of RMSE of distance estimation as the signal-to-noise ratio changes.

[0065] Figure 7 Schematic diagram of the comparison of RMSE of delay parameter estimation as the signal-to-noise ratio changes.

[0066] Figure 8 Schematic diagram of the comparison of the RMSE of 2D-DOA estimation as the number of sampling points changes.

[0067] Fig. 9 Schematic diagram of the comparison of the RMSE of polarization parameter estimation as the number of sampling points changes.

[0068] Fig.10A comparison diagram of the RMSE of distance estimation as the number of sampling points changes.

[0069] Fig.11 A comparison diagram of the RMSE of delay parameter estimation as the number of sampling points changes.

[0070] Fig.12 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0071] The embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings.

[0072] The present invention establishes a precise spatial model of a near-field multipath-arbitrary structure EMVS array in three-dimensional space, performs discrete Fourier transform on array received data, reconstructs a new received data model using the space-time-polarization information therein, performs trilinear decomposition on the reconstructed data to obtain estimated values ​​of the array manifold matrix and the delay response matrix; then, the estimated values ​​of 2D-DOA and TD at each EMVS are obtained using the rotational invariant relationship of the matrix, and the estimated values ​​of the polarization parameters are obtained using the least squares method; finally, a linear equation is constructed to obtain the estimated values ​​of the 2D-DOA and distance parameters of the near-field multipath relative to a phase reference point.

[0073] like Fig.12 As shown, the parameter estimation method of the EMVS array with arbitrary element structure under near-field multipath proposed by the present invention has the following steps:

[0074] Step 1: Establish a data model of near-field multipath incident on an EMVS array of arbitrary structure in three-dimensional space;

[0075] Assume that the near-field narrowband signal s(t) is incident on the array via L near-field multipaths, and the parameters of the lth path (l=1,2,…,L) are given by (θ l ,φ l ,γ l ,η l ,r l ) represents, where θ l ∈[0,π) and φ l ∈[0,2π) represent the elevation angle and azimuth angle respectively, and η l ∈[-π,π) represent the polarization auxiliary angle and polarization phase difference respectively, r l ∈(0.62(D 3 / λ) 1 / 2 ,2D 2 / λ) represents the distance from the phase reference point, where D is the array aperture and λ is the signal wavelength. The electromagnetic steering vector of the nth EMVS on the lth path is expressed as:

[0076]

[0077] Among them, θ n,l and φ n,l represents the elevation angle and azimuth angle of the lth path to the nth EMVS. n,l and h n,l Respectively by a s (θ n,l ,φ n,l ,γ l ,η l ) consists of the first three components and the last three components, representing the electric field components and the magnetic field components.

[0078] The near-field multipath signal received by the N-element arbitrary structure EMVS array at time t is recorded as:

[0079]

[0080] in, represents the received signal of the ith pole of the nth EMVS; β l represents the complex coefficient of the attenuation factor of the lth path; τ l is the transmission time delay of the lth path; s(t) represents the known signal waveform; n(t) is the zero-mean additive white Gaussian noise independent of the signal; a(θ l ,φ l ,γ l ,η l ) represents the spatial polarization steering vector of the array for the lth path, which is expressed as follows:

[0081]

[0082] in, represents the spatial steering response of the nth EMVS, i.e., the phase shift associated with the time delay of the lth path reaching the nth EMVS relative to the phase reference point; is the distance between the lth path and the nth EMVS, where o l = sinθ l cosφ l , p l = sinθ l sinφ l ,u l = cosθ l are the direction cosines along the x-axis, y-axis, and z-axis respectively.

[0083] Step 2: Perform discrete Fourier transform on the array received data and use the space-time-polarization information to reconstruct a new received data model;

[0084] 1) Perform DFT of K frequency points on the received signal of each polar component of the sensor. The data model expression of the kth frequency point is as follows:

[0085]

[0086] Among them, Y(ω k ), s(ω k ) and N(ω k ) are the DFTs of Y(t), s(t) and N(t), respectively. In some applications, such as WIFI and terrestrial mobile communication systems, channel state information (CSI) is usually obtained by deconvolution of known training sequences and pulse shape functions, rather than by time domain sampling. In this case, s(ω)=1 is set in the proposed model, and equation (4) is rewritten in matrix form:

[0087] Y(ω k )=ABf(ω k )+N(ω k ) (5)

[0088] in, Manifold matrix representing array; represents the path fading factor matrix; Represents the delayed response matrix at the kth frequency point.

[0089] For the information of all K frequency points, use express:

[0090] Z=ABF+N (6)

[0091] Where, F = [f(ω1), f(ω2), …, f(ω K )] is an L×K dimensional matrix; N=[n(ω1),n(ω2),…,n(ω K )] is a 6N×K dimensional matrix.

[0092] 2) Reconstruct a new receiving data model using the space-time-polarization information in the receiving data model:

[0093] For K sampling points, we can form Q (Q ≥ 2) groups of sub-matrices, where each sub-matrix has P sampling points (P = K-Q + 1). The qth sub-matrix is ​​expressed as:

[0094] Z q =ABE q F1+N q (7)

[0095] in, diag(·) represents the matrix diagonalization operation, Δω=ω j+1 -ω j Represents the frequency difference between two frequency points (1≤j<K-1); Represents the delayed response matrix after rearrangement.

[0096] In order to exploit the Vandermonde structure in F1, the matrix Z q After vectorization, rearrange the matrix to get the new matrix:

[0097] V=(F1 T ⊙A)E+W(8)

[0098] in represents the delayed response matrix after rearrangement, It can be regarded as the signal source vector of the new received signal matrix V, and W represents the corresponding noise after the data is rearranged.

[0099] Step 3: Perform trilinear decomposition on the covariance matrix of the rearranged data to obtain the estimated values ​​of the array manifold matrix and the delayed response matrix;

[0100] 1) Calculate the covariance matrix of the rearranged data model V as:

[0101]

[0102] 2) R can be expressed as a third-order PARAFAC model, and the matrix F1 can be obtained simultaneously through trilinear decomposition T and A. Applying complex parallel factor analysis algorithm to speed up the iteration speed, we get F1 T The estimated values ​​of and A are expressed as and

[0103] Step 4: Use the rotation invariance of the array manifold matrix and the vector cross multiplication technique to obtain the 2D-DOA estimate at each EMVS;

[0104] Get the estimation matrix Then, define the selection matrix Among them I p and i p,q Represents the p×p identity matrix and the identity matrix I p The qth row of . Define the selection matrix Further we get the following relationship:

[0105]

[0106] in is the rotation invariant factor. Let Due to m n,i is a row vector, let Therefore, by Perform eigenvalue decomposition to obtain The estimated value of .

[0107] The normalized Poynting vector of the lth multipath estimated by the nth EMVS is expressed as:

[0108]

[0109] in, is the estimated rotation invariant factor, and # represents the vector cross product. From this, the 2D-DOA estimate at the nth EMVS can be calculated:

[0110]

[0111] Step 5: Use the rotation invariance of the delay response matrix to obtain the estimated value of TD;

[0112] Defining the selection matrix and Where 0 p is a p×1 zero vector, and the following relationship can be obtained:

[0113]

[0114] in, express The lth column of Therefore, the estimated value of multipath time delay can be calculated as follows:

[0115]

[0116] Step 6: Estimate the polarization parameters using the least squares method;

[0117] The 2D-DOA estimate at the nth EMVS obtained by equation (12) is and Substituting into formula (1) we get

[0118]

[0119] in, Indicates a s (θ n,l ,φ n,l ,γ l ,η l ) can be estimated by get.

[0120] From this, the estimated polarization parameter at the nth EMVS can be calculated as

[0121]

[0122] Since the polarization parameters obtained at the same EMVS correspond to the elevation angle and azimuth angle one by one, after obtaining the polarization parameter estimates at all EMVS, the average value is taken to obtain the final polarization parameter estimate:

[0123]

[0124] Step 7: Construct a linear equation based on the sensor array position information to estimate the position parameters of the near-field multipath.

[0125] According to the geometric relationship of the sensor array position, the parameter equation related to the sensor array position and angle is established as follows:

[0126]

[0127] Converting equation (18) into matrix form yields

[0128] Η l Ο l =Γ l (19)

[0129] in,

[0130]

[0131] Using the least squares method, we get l Estimated value of

[0132]

[0133] Therefore, the 2D-DOA and distance calculation formula of the lth near-field multipath relative to the phase reference point are as follows:

[0134]

[0135] The present invention has completed MATLAB software simulation experiments to verify the effectiveness of the present invention. In the simulation experiment, the simulation model and simulation parameters are set as follows: assuming an irregularly spaced 3-D EMVS array, the five sensors are located at (-2.5λ, -2.5λ, -2.5λ), (λ, 0.5λ, 2λ), (1.5λ, -λ, -λ), (-λ, 0.5λ, 0.5λ), and (2.5λ, 2.5λ, 2.5λ). In a multipath environment, it is assumed that the phase of the complex path fading factor is uniformly distributed within [0, 2π]. In the method of the present invention, Q = 50 groups of sub-matrices are formed for all sampling points. All results are based on 500 independent Monte Carlo experiments, and the root mean square error (RMSE) is used to evaluate the performance of the algorithm. The RMSE of parameter estimation is defined as

[0136]

[0137] in represents the parameter θ in the cth Monte Carlo simulation l ,φ l 、r l , γ l , η l and τ l The estimated value of α l Represents the true value.

[0138] In order to verify the effectiveness of the method proposed in the present invention, the inventors designed the following simulation experiment:

[0139] Experiment 1: Assume that a near-field narrowband signal source has L = 7 near-field multipaths, and the parameters are set as θ = (55°, 25°, 72°, 60°, 30°, 40°, 45°), φ = (50°, 60°, 40°, 15°, 65°, 30°, 75°), r = (15λ, 12λ, 10λ, 20λ, 25λ, 22λ, 30λ), γ = (28°, 35°, 45°, 50°, 66°, 70°, 80°), η = (-90°, -80°, -50°, -20°, 30°, 32°, 90°), τ = (85, 30, 10, 100, 50, 20, 120) ns; the signal-to-noise ratio is set to 15 dB, and the number of sampling points K is set to 256.

[0140] The scatter plots of 2D-DOA and TD estimation of the method proposed in the present invention are as follows: Figure 2 As shown, the scatter plot of polarization parameter estimation is as follows Figure 3 As shown. Combined Figure 2 , Figure 3 It can be seen that the elevation angle, azimuth angle, TD and polarization parameters of the seven paths can be accurately identified and correctly paired.

[0141] Experiment 2: Assume that a narrowband signal source has L = 2 near-field multipaths, and the parameters are set to θ = (60°, 30°), r=(10λ,12λ)、γ=(45°,45°)、η=(-90°,90°)、τ=(50,80)ns, the number of sampling points K is set to 256, and the signal-to-noise ratio SNR varies from 0dB to 40dB. The method of the present invention is compared with the ESPRIT-like method (WEN FQ, SHI JP, ZHANG Z Z. Closed-form estimation algorithm for EMVS-MIMOradar with arbitrary sensor geometry[J]. Signal Processing, 2021, 186(8):108–117.) for constructing a normalized Poynting vector and the Cramer-Rao Bound (CRB). The comparison results of the RMSE of 2D-DOA, polarization parameters, distance and TD estimation with the change of signal-to-noise ratio are shown as follows. Figure 4 , Figure 5 , Figure 6 and Figure 7 The results show that when the signal-to-noise ratio is larger, the parameter estimation performance of the two methods is better. Figure 4 , Figure 5 and Figure 6 It can be seen that the estimation performance of the proposed method and the ESPRIT method for 2D-DOA, polarization parameters and distance is almost the same under the current snapshot number. Figure 7 It can be seen that the estimation performance of the method proposed in the present invention for TD is significantly better than that of the ESPRIT method, and is closer to the CRB.

[0142] Experiment 3: Assuming that the signal-to-noise ratio (SNR) is fixed at 15 dB, the number of sampling points varies from 64 to 320, and other parameters are set the same as in Experiment 2. The parameter estimation performance of the proposed method and the ESPRIT method under different sampling point numbers is compared. The RMSE of 2D-DOA, polarization parameters, distance and TD estimation with the change of sampling point number are shown in Figure 2. Figure 8 , Fig. 9 , Fig.10 and Fig.11 As shown. Figure 8 , Fig. 9 and Fig.10It can be seen that for the estimation of 2D-DOA, distance and polarization parameters, the estimation accuracy of the method proposed in the present invention is significantly better than that of the ESPRIT method when the number of sampling points is small; for the estimation of TD, the accuracy of the method proposed in the present invention is better than that of the ESPRIT method when the number of sampling points is large. Therefore, combined with Experiment 2, it can be shown that the parameter estimation performance of the method proposed in the present invention is better than that of the ESPRIT-like method.

[0143] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A parameter estimation method for an EMVS array with arbitrary element structure under near-field multipath, characterized in that: include: Step 1: Establish a data model of near-field multipath incident on an EMVS array of arbitrary structure in three-dimensional space; Step 2: Perform discrete Fourier transform on the array received data and use the space-time-polarization information to reconstruct a new received data model; Step 3: Perform trilinear decomposition on the covariance matrix of the rearranged data to obtain the estimated values ​​of the array manifold matrix and the delayed response matrix; Step 4: Use the rotation invariance of the array manifold matrix and the vector cross multiplication technique to obtain the 2D-DOA estimate at each EMVS; Step 5: Use the rotation invariance of the delay response matrix to obtain the estimated value of TD; Step 6: Estimate the polarization parameters using the least squares method; Step 7: Construct a linear equation based on the sensor array position information to estimate the position parameters of the near-field multipath.

2. The parameter estimation method of an EMVS array with arbitrary element structure under near-field multipath according to claim 1 is characterized in that: In step 1, the specific method of establishing the data model of near-field multipath incident on an EMVS array of arbitrary structure in three-dimensional space is as follows: let L represent the number of multipaths, let θ l ,φ l , γ l , η l and r l They represent the elevation angle, azimuth angle, polarization auxiliary angle, polarization phase difference and distance of the lth multipath relative to the phase reference point. The near-field multipath signal received by an N-element EMVS array with arbitrary structure in three-dimensional space at time t is expressed as: in, is the received signal of the ith pole of the nth EMVS; β l is the complex coefficient of the attenuation factor of the lth path; τ l is the transmission time delay of the lth path; s(t) represents the known signal waveform; N(t) is the zero-mean additive white Gaussian noise independent of the signal; a(θ l ,φ l ,γ l ,η l ) represents the spatial polarization steering vector of the array for the lth path.

3. The parameter estimation method of an EMVS array with arbitrary element structure under near-field multipath according to claim 2 is characterized in that: In step 2, the array received data is subjected to discrete Fourier transform, and the specific method of reconstructing a new received data model using the space-time-polarization information therein is as follows: 1) Perform DFT of K frequency points on the received signal of each polar component of the sensor. The data model expression of the kth frequency point is as follows: Y(ω k )=ABf(ω k )+n(ω k (4) Among them, Y(ω k ) and n(ω k ) are the DFTs of Y(t) and n(t), respectively. Manifold matrix representing array; represents the path fading factor matrix; represents the delayed response matrix at the kth frequency point; 2) After rearranging the data of K frequency points, the model is constructed as follows: V=(F1 T ⊙A)E+W(8) in represents the delayed response matrix after rearrangement, E can be regarded as the signal source vector of the new received signal matrix V, and W represents the corresponding noise after the data is rearranged.

4. The parameter estimation method of an EMVS array with arbitrary element structure under near-field multipath according to claim 3 is characterized in that: Step 3 above includes: The covariance matrix of the rearranged data model V is calculated as follows: R can be expressed as a third-order PARAFAC model, and the delayed response matrix F1 can be obtained simultaneously through trilinear decomposition. T and an estimate of the array manifold matrix A.

5. The parameter estimation method of an EMVS array with arbitrary element structure under near-field multipath according to claim 4 is characterized in that: Step 4 includes: Defining the selection matrix Among them I p and i p,q Represents the p×p identity matrix and the identity matrix I p The qth row of the selection matrix Further we get the following relationship: in is the rotation invariant factor, Using the estimated rotation invariant factor, we calculate it by vector cross product: in is the estimated rotation invariant factor, # represents vector cross product, From this, the 2D-DOA estimate at the nth EMVS can be calculated:

6. The parameter estimation method of an EMVS array with arbitrary array element structure under near-field multipath according to claim 5 is characterized in that: In step 5, define the selection matrix and The following relationship can be obtained: in Therefore, the estimated value of multipath time delay can be calculated as follows:

7. The parameter estimation method of an EMVS array with arbitrary element structure under near-field multipath according to claim 6 is characterized in that: Step 6 above includes: After obtaining the 2D-DOA estimate at the nth EMVS, the least squares method is used to calculate: in, Indicates a s (θ n,l ,φ n,l ,γ l ,η l ), from which the estimated value of the polarization parameter at the nth EMVS can be calculated as: After obtaining the estimated polarization parameters at all EMVS, the average value is taken to obtain the final estimated polarization parameter value:

8. The parameter estimation method of an EMVS array with arbitrary element structure under near-field multipath according to claim 7 is characterized in that: The above step 7 includes: According to the geometric relationship of the sensor array position, the parameter equation related to the sensor array position and angle is established as follows: OR l THE l =C l (19) in, Using the least squares method, we get l Estimated value of Therefore, the 2D-DOA and distance calculation formula of the corresponding l-th near-field multipath at the phase reference point are as follows:

Images