A fast DOA estimation method for millimeter wave MIMO radar

Through the rapid convolution and coordinate selection descent criteria, the high complexity problem in the millimeter wave MIMO radar DOA estimation method is solved, which significantly improves real-time performance, realizes high-precision angular resolution estimation, and meets the real-time requirements.

CN115079086BActive Publication Date: 2024-12-06UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210777648.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-04
Publication Date
2024-12-06
Estimated Expiration
2042-07-04

AI Technical Summary

Technical Problem

The DOA estimation method of existing millimeter wave MIMO radar has high computational complexity and cannot meet the real-time requirements, especially the high-complexity problems caused by large-scale matrix inversion and multiplication operations during iteration and the traversal of large-scale sampling grid points. .

Method used

Fast convolution is used to reduce the complexity of matrix inversion and multiplication operations, and iteratively terminate through coordinate selection descent criteria, which significantly improves the real-time performance of the DOA estimation method.

Benefits of technology

It significantly reduces the computational complexity, improves real-time performance, and realizes high-precision angular resolution estimation, meeting the real-time requirements in practical applications.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115079086B_ABST
    Figure CN115079086B_ABST
Patent Text Reader

Abstract

The present invention discloses a method for fast DOA estimation of millimeter wave MIMO radar, and the specific steps are as follows: step 1, modeling the echo of co-located millimeter wave MIMO radar; step 2, initializing the target scattering distribution estimation; step 3, calculating the current iteration target power vector; step 4, calculating the autocorrelation matrix; step 5, calculating the iterative estimation numerator; step 6, calculating the iterative estimation denominator; step 7, calculating the estimation result; step 8, judging the algorithm iteration convergence. In view of the problem of high computational complexity of the existing millimeter wave MIMO radar DOA estimation method, the present invention uses fast convolution to reduce the complexity of matrix inversion and matrix multiplication operations, and adopts the coordinate selection descent criterion for iteration termination, which significantly improves the real-time performance of the existing DOA estimation method and realizes the fast DOA estimation of millimeter wave MIMO radar.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of radar detection, and in particular relates to a millimeter wave MIMO radar fast DOA estimation method. Background Art

[0002] Compared with traditional array antenna radars, the advantages of multiple transmitting antennas and multiple receiving antennas enable millimeter-wave MIMO radars to use fewer array elements to obtain more observation channel data of the target. However, the traditional millimeter-wave MIMO radar beamforming method based on matched filtering has the characteristics of wide main lobe and high side lobe, which cannot achieve high angular resolution and is difficult to meet the needs of practical applications.

[0003] In order to improve the angular resolution of millimeter-wave MIMO radar, many researchers have proposed DOA estimation methods. The literature "Duofang C, Baixiao C, Guodong Q. Angle estimation using ESPRIT in millimeter-wave MIMO radar [J]. Electronics Letters, 2008, 44 (12): 770-771." uses the invariant characteristics of the millimeter-wave MIMO radar transmitting array and receiving array to achieve direction estimation, thereby improving the resolution to a certain extent. The literature "Zhang X, Huang Y, Chen C, et al. Reduced-complexity Capon for direction of arrival estimation in a monostatic multiple-input multiple-output radar [J]. IET Radar, Sonar & Navigation, 2012, 6 (8): 796-801." uses the Capon beamforming algorithm to estimate the direction of arrival of a monostatic millimeter-wave MIMO radar. It reduces the subaperture dimension of the millimeter-wave MIMO radar through dimensionality reduction transformation, thereby reducing the complexity and improving the resolution. In order to further improve the resolution, the paper "Roberts W, Stoica P, Li J, et al. Iterative adaptive approaches to millimeter wave MIMO radar imaging [J]. IEEE Journal of Selected Topics in Signal Processing, 2010, 4 (1): 5-20.", a regularized iterative adaptive (IAA) method was proposed to realize millimeter wave MIMO radar imaging, which effectively improved the azimuth resolution of millimeter wave MIMO radar. This method is based on the linear transformation of echo data with respect to the transformation matrix and scattering coefficient, and uses the estimation result obtained in the previous iteration to construct the autocorrelation matrix, and uses the autocorrelation matrix as the weighting coefficient for weighted least squares estimation. The estimation performance is improved through cyclic iteration, and it has the characteristics of low sidelobe, high resolution, few snapshots and few parameters. However, this method requires large-scale matrix inversion and multiplication operations during the iteration process, and needs to traverse a large number of sampling grid points. The computational complexity is very high and cannot meet the real-time requirements in practical applications. Summary of the invention

[0004] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a millimeter-wave MIMO radar fast DOA estimation method that utilizes fast convolution to reduce the complexity of matrix inversion and matrix multiplication operations, while adopting a coordinate selection descent criterion for iteration termination, thereby significantly improving the real-time performance of the existing DOA estimation method.

[0005] The objective of the present invention is achieved through the following technical solution: a millimeter wave MIMO radar fast DOA estimation method, the specific steps are as follows:

[0006] Step 1: Modeling the echo of the co-located millimeter-wave MIMO radar: Assume that the number of transmitting antennas is M and the number of receiving antennas is N; assume that there are multiple unrelated far-field targets at K different directions, and the target scattering intensity is expressed as:

[0007] σ(t)=[σ 1 (t),...,σ k (t),...,σ K (t)] T ,k=1,2,...,K (1)

[0008] Where t represents the distance time variable, σ k (t) represents the target scattering intensity of the kth target grid; M transmitting antennas transmit mutually orthogonal waveforms, and the echoes of MN channels obtained by matched filtering are expressed as:

[0009]

[0010] in represents K azimuth angles, n(t) represents additive noise;

[0011] a R (θ k ) represents the transmitting antenna steering vector, a T (θ k ) represents the receiving antenna steering vector:

[0012] a R (θ k )=[1,exp(-j2πd R sinθ k / λ),...,exp(-j2π(M-1)d R sinθ k / λ)] T (3)

[0013] a T (θ k )=[1,exp(-j2πd T sinθ k / λ),...,exp(-j2π(N-1)d T sinθ k / λ)] T (4)

[0014] λ represents the signal wavelength, d T and d R They represent the spacing between the transmitting linear array element and the receiving linear array element, satisfying d R =d T / N=λ / 2;

[0015] For the nth snapshot, the echo is expressed as:

[0016] y=Aσ+n (5)

[0017] in is a MN×K dimensional matrix, and represents a virtual steering vector of the millimeter-wave MIMO radar;

[0018] Step 2: Initialize target scattering distribution estimation;

[0019] Step 3: Calculate the current iteration target power vector;

[0020] Step 4: Calculate the autocorrelation matrix;

[0021] Step 5: Calculate the iterative estimation numerator;

[0022] Step 6: Calculate the iterative estimation denominator;

[0023] Step 7: Calculate the estimated results;

[0024] Step 8: Determine the convergence of algorithm iteration.

[0025] Furthermore, the specific implementation method of step 2 is: initialize and set the convergence judgment parameter ε, the maximum number of iterations Q, and the reflection coefficient estimation initialization value

[0026]

[0027] Among them, a k represents the k-th column of A, and H represents the conjugate transpose of the matrix or vector.

[0028] Furthermore, the specific implementation method of step three is: assuming that the current number of iterations is q, if q does not reach the maximum value Q, then construct the following target power vector:

[0029]

[0030] in represents the target power in the kth direction, represents the kth element of σ at the qth iteration.

[0031] Furthermore, the specific implementation method of step 4 is: perform a K-point fast Fourier transform on the target power vector, and then intercept the 1st to MNth elements to obtain the column vector r 1 ; Define the autocorrelation matrix R as:

[0032] R=[r 1 ,r 2 ,…,r MN ] (8)

[0033] where r n represents the nth column vector of the autocorrelation matrix; since the autocorrelation matrix R satisfies the Toeplitz property, r 1 Circular shift to get r 2 ,r 3 ,…,r MN .

[0034] Furthermore, the specific implementation method of step 5 is: define vectors ζ and ζ′ as:

[0035]

[0036]

[0037] where a 2 ,...,a MN and e represent the autoregressive coefficient and prediction error of the Yule-Walker equation obtained by the autocorrelation matrix R using the Levinson-Durbin algorithm, respectively, and * represents the conjugate operation;

[0038] The iterative estimated numerator is quickly obtained through the following five steps:

[0039] ① Calculation Fast convolution of y and y, and intercepting the MNth to (2MN-1)th elements to form ζ 1 ;

[0040] ② Calculate ζ and ζ 1 Fast convolution, and intercept the 1st to K elements to form u 1 ;

[0041] ③Calculation Fast convolution of y and y, and intercepting the MNth to (2MN-1)th elements to form ζ′ 1 ;

[0042] ④ Calculate ζ′ and ζ′ 1 Fast convolution, and intercept the 1st to K elements to form u 2 ;

[0043] ⑤Calculate x=u 1 -u 2 , then, we get the iterative estimate numerator of the k-th position

[0044] Furthermore, the specific implementation method of step six is: the iterative estimation denominator is quickly obtained by the following five steps:

[0045] ① Calculation and a k Fast convolution, and intercept the MNth to (2MN-1)th elements to form ζ 2 ;

[0046] ② Calculate ζ and ζ 2 Fast convolution of , and intercept the 1st to K elements to form v 1 ;

[0047] ③Calculation and a k Fast convolution, and intercept the MNth to (2MN-1)th elements to form ζ′ 2 ;

[0048] ④ Calculate ζ′ and ζ′ 2 Fast convolution of , and intercept the 1st to K elements to form v 2 ;

[0049] ⑤Calculate z=v 1 -v 2 , get the iterative estimate denominator of the kth direction

[0050] Furthermore, in step seven, the iterative estimation result is:

[0051]

[0052] Furthermore, the specific implementation method of step eight is: calculating the convergence judgment factor:

[0053]

[0054] like If the k-th DOA is estimated, it is considered to have fully converged, the iteration is stopped, and the current iteration estimation result is output as the DOA estimation result of the k-th azimuth; otherwise, return to step 3;

[0055] Traverse k=1,2,…,K and obtain the final DOA estimation result σ.

[0056] The beneficial effects of the present invention are as follows: in order to solve the problem of high computational complexity of existing millimeter-wave MIMO radar DOA estimation methods, the present invention, on the one hand, utilizes fast convolution to reduce the complexity of matrix inversion and matrix multiplication operations, and at the same time adopts the coordinate selection descent criterion for iteration termination, thereby significantly improving the real-time performance of existing DOA estimation methods and realizing fast DOA estimation of millimeter-wave MIMO radar. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flowchart of the fast DOA estimation method of millimeter wave MIMO radar of the present invention;

[0058] Figure 2 The array configuration of the transmitting array element and the receiving array element of the millimeter wave MIMO radar of the present invention;

[0059] Figure 3 For the original goal;

[0060] Figure 4 This is the result of classic matched filtering angle estimation (SNR = 20dB);

[0061] Figure 5 This is the result of traditional iterative adaptive angle estimation under SNR=20dB condition;

[0062] Figure 6 This is the angle estimation result of the method of the present invention under the condition of SNR=20dB;

[0063] Figure 7 The figure shows the comparison of the number of iterative calculations between the method of the present invention and the traditional iterative DOA estimation method. DETAILED DESCRIPTION

[0064] The present invention adopts simulation experiments to demonstrate the effectiveness of the proposed fast radar imaging DOA estimation method. All steps and conclusions of the present invention are verified to be correct on the Matlab2019b simulation platform. In order for relevant personnel in this field to understand the content of the invention, the method of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0065] like Figure 1 As shown, a millimeter wave MIMO radar fast DOA estimation method of the present invention has the following specific steps:

[0066] Step 1: Modeling of co-located millimeter wave MIMO radar echoes: This embodiment adopts Figure 2 The array configuration of the millimeter wave MIMO radar transmitting array element and receiving array element is shown. The millimeter wave MIMO radar system parameters are set according to Table 1.

[0067] Table 1

[0068] parameter symbol Numeric Working frequency band <![CDATA[f c ]]> 77GHz Transmit signal bandwidth B 30MHz Number of transmitting elements M 2 Number of receiving elements N 4 Signal-to-Noise Ratio SNR 20dB Scan range - -90°~90° Target grid number K 512 Convergence Threshold ε <![CDATA[10 -3 ]]> Maximum number of iterations Q 20

[0069] According to the array configuration of the transmitting and receiving elements of the millimeter-wave MIMO radar and the parameters in Table 1, the number of transmitting antennas is M = 2, and the number of receiving antennas is N = 4; assuming that there are multiple unrelated far-field targets at K = 512 different directions, the target scattering intensity is expressed as:

[0070] σ(t)=[σ 1 (t),...,σ k (t),...,σ K (t)] T ,k=1,2,...,K (13)

[0071] Where t represents the distance time variable, σ k (t) represents the target scattering intensity of the kth target grid; M transmitting antennas transmit mutually orthogonal waveforms, and the echoes of 8 channels obtained by matched filtering are expressed as:

[0072]

[0073] in represents K azimuth angles, n(t) represents additive noise;

[0074] a R (θ k ) represents the transmitting antenna steering vector, a T (θ k ) represents the receiving antenna steering vector:

[0075] a R (θ k )=[1,exp(-j2πd R sinθ k / λ),...,exp(-j2π(M-1)d R sinθ k / λ)] T (15)

[0076] a T (θ k )=[1,exp(-j2πd T sinθ k / λ),...,exp(-j2π(N-1)d T sinθ k / λ)] T (16)

[0077] λ represents the signal wavelength, λ=0.004m, d T and d RThey represent the spacing between the transmit line array (ULA) elements and the receive line array (ULA) elements, satisfying d R =d T / 4=0.004 / 2;

[0078] For the nth snapshot, the echo is expressed as:

[0079] y=Aσ+n (17)

[0080] in is a MN×K dimensional matrix, and represents a virtual steering vector of the millimeter-wave MIMO radar;

[0081] In the simulation scene, two adjacent point targets are set at 20° and 30°, such as Figure 3 shown.

[0082] Step 2: Initialize the target scattering distribution estimation; the specific implementation method is: initialize and set the convergence judgment parameter ε = 10 -3 , the maximum number of iterations Q = 20, and the initialization value of the reflection coefficient estimate

[0083]

[0084] Among them, a k represents the k-th column of A, and H represents the conjugate transpose of the matrix or vector.

[0085] Step 3: Calculate the target power vector of the current iteration. The specific implementation method is: let the current iteration number be q. If q does not reach the maximum value of 20, then construct the following target power vector:

[0086]

[0087] in represents the target power in the kth direction, represents the kth element of σ at the qth iteration.

[0088] Step 4: Calculate the autocorrelation matrix; perform a 512-point fast Fourier transform on the target power vector, and then intercept the 1st to 8th elements to obtain the column vector r 1 ; Define the autocorrelation matrix R as:

[0089] R=[r 1 ,r 2 ,…,r 8 ] (20)

[0090] where r n represents the nth column vector of the autocorrelation matrix; since the autocorrelation matrix R satisfies the Toeplitz property, r 1Circular shift to get r 2 ,r 3 ,…,r 8 .

[0091] Step 5: Calculate the iterative estimation numerator; the specific implementation method is: define vectors ζ and ζ′ as:

[0092]

[0093]

[0094] where a 2 ,...,a MN and e represent the autoregressive coefficient and prediction error of the Yule-Walker equation obtained by the autocorrelation matrix R using the Levinson-Durbin algorithm, respectively, and * represents the conjugate operation;

[0095] The iterative estimated numerator is quickly obtained through the following five steps:

[0096] ① Calculation The fast convolution of y and y, and the 8th to 15th elements are truncated to form ζ 1 ;

[0097] ② Calculate ζ and ζ 1 Fast convolution, and truncate the 1st to 512th elements to form u 1 ;

[0098] ③Calculation Fast convolution of y and y, and truncate the 8th to 15th elements to form ζ′ 1 ;

[0099] ④ Calculate ζ′ and ζ′ 1 Fast convolution, and truncate the 1st to 512th elements to form u 2 ;

[0100] ⑤Calculate x=u 1 -u 2 , then, we get the iterative estimate numerator of the k-th position

[0101] Step 6: Calculate the iterative estimated denominator; the iterative estimated denominator is quickly obtained through the following five steps:

[0102] ① Calculation and a k The fast convolution of ζ is performed, and the 8th to 15th elements are truncated to form ζ 2 ;

[0103] ② Calculate ζ and ζ 2 Fast convolution of , and truncate the 1st to 512th elements to form v 1;

[0104] ③Calculation and a k Fast convolution, and truncate the 8th to 15th elements to form ζ′ 2 ;

[0105] ④ Calculate ζ′ and ζ′ 2 Fast convolution of , and truncate the 1st to 512th elements to form v 2 ;

[0106] ⑤Calculate z=v 1 -v 2 , get the iterative estimate denominator of the kth direction

[0107] Step 7: Calculate the estimated results:

[0108]

[0109] Step 8: Algorithm iteration convergence judgment; the specific implementation method is: calculate the convergence judgment factor:

[0110]

[0111] like If the k-th DOA is estimated, it is considered to have fully converged, the iteration is stopped, and the current iteration estimation result is output as the DOA estimation result of the k-th azimuth; otherwise, return to step 3;

[0112] Traverse k=1,2,…,K and obtain the final DOA estimation result σ.

[0113] Thus, the method of the present invention can realize the fast DOA estimation of millimeter wave MIMO radar and obtain the high-precision estimation of the target azimuth. In order to compare and prove the effectiveness of the present invention, a control experiment was carried out using the classic matched filtering method, the traditional iterative adaptive method and the method proposed by the present invention under the condition of a signal-to-noise ratio of 20dB. The results are as follows: Figure 4 , Figure 5 , Figure 6 As shown, Figure 4 This is the result of classic matched filtering angle estimation (SNR=20dB). Figure 5 This is the result of traditional iterative adaptive angle estimation under SNR=20dB condition. Figure 6 : This is the angle estimation result of the method of the present invention under the condition of SNR=20dB. In the figure, the horizontal axis (Angle) is the angle, and the vertical axis (Normalized amplitude) is the normalized amplitude. Figure 7 The variation of the number of iteration points of the traditional iterative adaptive method and the method of the present invention with the number of iterations is shown. Figure 4 and Figure 5 , Figure 6It can be seen that the iterative adaptive method has a significantly improved azimuth resolution effect compared to the classical matched filter imaging method. Figure 7 It can be seen that the method of the present invention has lower computational complexity in the iterative process than the traditional iterative adaptive method. The reason for this acceleration effect is that the method proposed by the present invention can judge the iterative convergence situation, and the iteration of the data that has fully converged can be stopped in time, which effectively reduces the scale of redundant calculations. In summary, the effectiveness of the fast DOA estimation method proposed by the present invention is proved.

[0114] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A millimeter wave MIMO radar fast DOA estimation method, characterized in that: The specific steps are as follows: Step 1: Modeling the echo of the co-located millimeter-wave MIMO radar: Assume that the number of transmitting antennas is M and the number of receiving antennas is N; assume that there are multiple unrelated far-field targets at K different directions, and the target scattering intensity is expressed as: σ(t)=[σ1(t),...,σ k (t),…,s K (t)] T ,k=1,2,...,K (1) Where t represents the distance time variable, σ k (t) represents the target scattering intensity of the kth target grid; M transmitting antennas transmit mutually orthogonal waveforms, and the echoes of MN channels obtained by matched filtering are expressed as: in represents K azimuth angles, n(t) represents additive noise; a R (θ k ) represents the transmitting antenna steering vector, a T (θ k ) represents the receiving antenna steering vector: a R (i k )=[1,exp(-j2πd R sinth k / λ),…,exp(-j2π(M-1)d R sinth k / l)] T (3) a T (i k )=[1,exp(-j2πd T sinth k / λ),…,exp(-j2π(N-1)d T sinth k / l)] T (4) λ represents the signal wavelength, d T and d R They represent the spacing between the transmitting linear array element and the receiving linear array element, satisfying d R =d T / N=λ / 2; For the nth snapshot, the echo is expressed as: y=Aσ+n (5) in is a MN×K dimensional matrix, and represents a virtual steering vector of the millimeter-wave MIMO radar; Step 2: Initialize target scattering distribution estimation; Step 3: Calculate the current iteration target power vector; Step 4: Calculate the autocorrelation matrix; Step 5: Calculate the iterative estimation numerator; the specific implementation method is: define vectors ζ and ζ′ as: where a2,...,a MN and e represent the autoregressive coefficient and prediction error of the Yule-Walker equation obtained by the autocorrelation matrix R using the Levinson-Durbin algorithm, respectively, and * represents the conjugate operation; The iterative estimated numerator is quickly obtained through the following five steps: ① Calculation and Fast convolution of y, and truncate the MNth to (2MN-1)th elements to form ζ1; ② Calculate the fast convolution of ζ and ζ1, and intercept the 1st to Kth elements to form u1; ③Calculation Fast convolution of y and truncate the MNth to (2MN-1)th elements to form ζ′1; ④ Calculate the fast convolution of ζ′ and ζ′1, and intercept the 1st to Kth elements to form u2; ⑤ Calculate x = u1-u2, and then get the iterative estimate numerator of the kth direction Step 6: Calculate the iterative estimated denominator. The specific implementation method is: the iterative estimated denominator is quickly obtained through the following five steps: ① Calculation and a k The fast convolution is performed, and the elements from MN to (2MN-1) are intercepted to form ζ2; ② Calculate the fast convolution of ζ and ζ2, and intercept the 1st to Kth elements to form v1; ③Calculation and a k The fast convolution is performed, and the elements from MN to (2MN-1) are truncated to form ζ′2; ④ Calculate the fast convolution of ζ′ and ζ′2, and intercept the 1st to Kth elements to form v2; ⑤ Calculate z = v1-v2 to get the iterative estimate denominator of the kth direction Step 7: Calculate the estimated results; Step 8: Determine the convergence of algorithm iteration.

2. The millimeter wave MIMO radar fast DOA estimation method according to claim 1, characterized in that: The specific implementation method of step 2 is: initialize and set the convergence judgment parameter ε, the maximum number of iterations Q, and the reflection coefficient estimation initialization value Among them, a k represents the k-th column of A, and H represents the conjugate transpose of the matrix or vector.

3. The millimeter wave MIMO radar fast DOA estimation method according to claim 1, characterized in that: The specific implementation method of step 3 is: assuming that the current number of iterations is q, if q does not reach the maximum value Q, then construct the following target power vector: in represents the target power in the kth direction, represents the kth element of σ at the qth iteration.

4. The millimeter wave MIMO radar fast DOA estimation method according to claim 1, characterized in that: The specific implementation method of step 4 is: perform K-point fast Fourier transform on the target power vector, and then intercept the 1st to MNth elements to obtain the column vector r1; define the autocorrelation matrix R as: R=[r1,r2,…,r MN ] (8) where r n represents the nth column vector of the autocorrelation matrix; since the autocorrelation matrix R satisfies the Toeplitz property, r1 is cyclically shifted to obtain r2, r3, …, r MN .

5. The millimeter wave MIMO radar fast DOA estimation method according to claim 1, characterized in that: In step seven, the iterative estimation result is:

6. The millimeter wave MIMO radar fast DOA estimation method according to claim 1, characterized in that: The specific implementation method of step eight is: calculate the convergence judgment factor: like It is considered that sufficient convergence has been achieved, the iteration is stopped, and the current iteration estimation result is output as the DOA estimation result of the kth azimuth; Otherwise, return to step 3; Traverse k=1,2,…,K and obtain the final DOA estimation result σ.