Fast radar single-shot MUSIC angle measurement method

Abstract

The present invention provides a fast radar single-snapshot MUSIC angle measurement method, which relates to the technical field of radar signal processing. The present invention first obtains array frequency domain single-snapshot data; then calculates the covariance matrix of the array frequency domain single-snapshot data, and performs eigenvalue decomposition to obtain a noise subspace; within the viewing angle range where the target is located, the angular resolution is used as the search step, and the spatial spectrum search of the incoming wave directions of multiple targets is performed in combination with the noise subspace to obtain a rough angle estimation value of the multiple targets; finally, the search range interval is determined according to the rough angle estimation values ​​of the multiple targets and the search step, and a golden section algorithm is used for a precise search to obtain accurate angle estimation values ​​of the multiple targets. Under the condition of ensuring the angle estimation accuracy and angular resolution of the algorithm, the present invention minimizes the amount of calculation and the memory space occupied, uses a single snapshot to solve the covariance matrix, and combines the golden section to perform a spatial spectrum search, which greatly improves the real-time performance of the MUSIC algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of radar signal processing, and in particular to a fast radar single-shot MUSIC angle measurement method. Background Art

[0002] Estimation of spatial signal DOA (direction of arrival) is one of the basic problems in array signal processing and one of the important tasks in many fields such as radar, wireless communication and sonar. Using traditional DOA algorithms, in general, large receiving array antennas are required to obtain higher angular resolution. However, due to the limitations of engineering applications and funding, large arrays are difficult to implement. Therefore, DOA estimation methods based on signal subspace, such as MUSIC, ESPRIT, direction estimation method (MODE) and min-norm, are widely known for their ultra-high angular resolution and angular accuracy that break through the Rayleigh limit. However, super-resolution DOA estimation algorithms also generally have the problems of high computational complexity and long computational time, which greatly limits the hardware implementation of these algorithms.

[0003] Among them, when using the MUSIC algorithm for DOA estimation, it is mainly divided into the following steps: (1) Calculate the array covariance matrix according to the number of point snapshots to receive the signal vector; (2) Perform eigenvalue decomposition on the obtained covariance matrix to obtain the signal subspace and noise subspace; (3) Calculate the spatial spectrum function of each angle within the angle search range, and obtain the incident angle by seeking the peak value. The performance and computational complexity of the algorithm are closely related to the following two points: First, the more snapshots of data accumulation, the closer the obtained covariance matrix is ​​to the ideal covariance matrix, and the better the DOA estimation performance, but the corresponding covariance matrix solution The computational complexity is also greater; Second, the number of spatial spectrum peak searches is consistent with the step of the search angle. If the angle estimation accuracy is to be improved, the number of spectrum peak searches must be increased. That is, the existing methods cannot reduce the computational complexity while meeting the accuracy requirements. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the shortcomings of the prior art, the present invention provides a fast radar single-shot MUSIC angle measurement method, which solves the technical problem that the prior method cannot take into account both accuracy and reduction of calculation amount.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0008] The present invention provides a fast radar single-shot MUSIC angle measurement method, which comprises:

[0009] S1. After performing two-dimensional FFT processing on each channel of the radar array echo signal, CFAR detection is used to obtain the array frequency domain single snapshot data;

[0010] S2, calculating the covariance matrix of the single snapshot data in the frequency domain of the array, and performing eigenvalue decomposition to obtain the noise subspace;

[0011] S3, within the viewing angle range of the target, using the angular resolution as the search step, combined with the noise subspace, perform a spatial spectrum search of the incoming wave directions of multiple targets to obtain rough angle estimates of multiple targets;

[0012] S4. Determine the search range based on the rough angle estimates of multiple targets and the search step, and use the golden section algorithm to perform a precise search to obtain accurate angle estimates of multiple targets.

[0013] Preferably, the S1 specifically includes:

[0014] S101, radar array echo signal s m (i,n) performs FFT transformation of N points in the first dimension along the direction of distance dimension n, and then performs FFT transformation of I points in the second dimension along the direction of period i, and records the two-dimensional FFT data as S m (ω i ,ω n ), where ω i and ω n Represent the sequence numbers of the range dimension and Doppler dimension FFT respectively;

[0015] S102, S m (ω i ,ω n ) performs incoherent accumulation of M channels, and obtains Among them, S int is an I×N dimensional incoherent accumulation matrix; for S int Perform two-dimensional CFAR detection and obtain the index number of the target point, recorded as (r p ,d p ), where p = 1, ..., P, P is the total target point detected by CFAR; r p Indicates the distance dimension index of the pth target; d p represents the velocity dimension index of the pth target; M is the number of radar receiving antenna array elements;

[0016] S103, according to the index number detected by CFAR, obtain the frequency domain single snapshot array data S of the pth target P , S P =[S 1 (rp ,d p ),S 2 (r p ,d p ),…,S M (r p ,d p )] T .

[0017] Preferably, S2 specifically includes:

[0018] S201, based on the frequency domain single snapshot array data S obtained in step S1 P Estimated covariance matrix

[0019] S202, perform eigenvalue decomposition on the matrix R to obtain the noise subspace U N , used to calculate the spatial spectrum function.

[0020] Preferably, the S202 includes:

[0021] Perform eigenvalue decomposition on the matrix R to obtain M eigenvalues. In order of the eigenvalue size, the K largest eigenvalues ​​are regarded as the signal subspace U S The remaining MK eigenvalues ​​corresponding to the eigenvectors are regarded as the noise subspace U N ; Wherein, M is the number of radar receiving antenna elements; K is the number of targets.

[0022] Preferably, S3 specifically includes:

[0023] S301, according to the viewing angle range of the actual target, set the angle search range θ min ≤θ≤θ max ; According to the minimum spacing between the two targets and the number of array elements, the required angular resolution Δθ is determined, and this is used as the search step of the spatial spectrum search;

[0024] S302, according to the angle search range [θ min ,θ max ] and search step Δθ, change search value θ i ,θ i =θ min +(l-1)Δθ,l=1,2,…,L, L is the total number of searches, L=(θ max -θ min ) / Δθ, and perform spatial spectrum peak search according to the following spatial spectrum function:

[0025]

[0026] Among them, a(θ i) is the direction-steering vector of the array; P(θ i ) The peak value corresponds to θ i It is a rough estimate of the incoming direction of the target signal; the superscript H represents the conjugate transpose.

[0027] Preferably, the S4 specifically includes:

[0028] S401: Assume the rough angle estimation values ​​of the K targets identified in S3 where k = 1, ..., K, for each target rough angle estimate Based on the search step Δθ in S3, As the search scope for refined search;

[0029] S402, taking the search range in S401 as the golden section interval, and setting the angle accuracy requirement as the minimum value of the golden section three-element point difference as the termination condition for stopping the golden section division, using the golden section algorithm to find the golden section point as the accurate angle estimation value of K targets.

[0030] (III) Beneficial effects

[0031] The present invention provides a fast radar single-shot MUSIC angle measurement method. Compared with the prior art, it has the following beneficial effects:

[0032] The present invention first performs two-dimensional FFT processing on each channel of the radar array echo signal, and then uses CFAR detection to obtain the single snapshot data of the array frequency domain; then calculates the covariance matrix of the single snapshot data of the array frequency domain, and performs eigenvalue decomposition to obtain the noise subspace; within the viewing angle range where the target is located, uses the angular resolution as the search step, and combines the noise subspace to perform spatial spectrum search of the wave directions of multiple targets to obtain rough angle estimates of multiple targets; finally, the search range interval is determined according to the rough angle estimates of multiple targets and the search step, and the golden section algorithm is used for fine search to obtain accurate angle estimates of multiple targets. Under the condition of ensuring the angle estimation accuracy and angular resolution of the algorithm, the present invention minimizes the amount of calculation and the memory space occupied, uses a single snapshot to solve the covariance matrix, and combines the golden section to perform spatial spectrum search, which greatly improves the real-time performance of the MUSIC algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0034] Figure 1 is a block diagram of a fast radar single-shot MUSIC angle measurement method according to an embodiment of the present invention;

[0035] Figure 2 It is the spectrum diagram of the two-dimensional FFT of the radar echo signal;

[0036] Figure 3 It is the MUSIC spatial spectrogram of the original multi-snapshot data;

[0037] Figure 4 It is the MUSIC spatial spectrogram of single snapshot data in frequency domain. DETAILED DESCRIPTION

[0038] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The embodiment of the present application provides a fast radar single-shot MUSIC angle measurement method, which solves the technical problem that the existing method cannot take into account both accuracy and reduction of calculation amount, and achieves that while meeting the real-time and accuracy requirements of the algorithm, the calculation amount of the algorithm is greatly reduced.

[0040] The technical solution in the embodiment of the present application is to solve the above technical problems, and the overall idea is as follows:

[0041] In order to solve the engineering problem that the MUSIC algorithm has high computational complexity and limits its hardware implementation, the embodiment of the present invention adopts single snapshot data and combines efficient spectrum peak search under multiple targets. While meeting the real-time and accuracy requirements of the algorithm, it greatly reduces the computational complexity of the algorithm and has great practical application value.

[0042] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods.

[0043] The embodiment of the present invention provides a fast radar single-shot MUSIC angle measurement method, such as Figure 1 As shown, including:

[0044] S1. After performing two-dimensional FFT processing on each channel of the radar array echo signal, CFAR detection is used to obtain the array frequency domain single snapshot data;

[0045] S2, calculating the covariance matrix of the single snapshot data in the frequency domain of the array, and performing eigenvalue decomposition to obtain the noise subspace;

[0046] S3, within the viewing angle range of the target, using the angular resolution as the search step, combined with the noise subspace, perform a spatial spectrum search of the incoming wave directions of multiple targets to obtain rough angle estimates of multiple targets;

[0047] S4. Determine the search range based on the rough angle estimates of multiple targets and the search step, and use the golden section algorithm to perform a precise search to obtain accurate angle estimates of multiple targets.

[0048] In order to solve the problem of many snapshots and many spectrum peak searches in the MUSIC algorithm, the embodiment of the present invention proposes an accurate and fast MUSIC angle measurement method that is convenient for engineering practice. Under the condition of ensuring the angle estimation accuracy and angular resolution of the algorithm, the embodiment of the present invention minimizes the amount of calculation and the memory space occupied, uses a single snapshot to solve the covariance matrix, and combines the golden section to perform spatial spectrum search, which greatly improves the real-time performance of the MUSIC algorithm.

[0049] The following is a detailed description of each step:

[0050] Assume that there are K targets with the same distance and speed but different wave directions, and the number of radar receiving antenna array elements is M. After down-conversion, intermediate frequency filtering, amplification and other processing, and ADC sampling, the discrete echo signal received by the mth receiving antenna in the i-th cycle can be expressed as s m (i,n), where i=1,2…,I, I is the total number of cycles, m=1,2…,M, n=1,2…,N represents the fast time sampling number of the i-th cycle, and I is the total number of sampling points.

[0051] In step S1, after performing two-dimensional FFT processing on each channel of the radar array echo signal, CFAR detection is used to obtain the array frequency domain single snapshot data. The specific implementation process is as follows:

[0052] S101, radar array echo signal s m (i,n) performs FFT transformation of N points in the first dimension along the direction of distance dimension n, and then performs FFT transformation of I points in the second dimension along the direction of period i, and records the two-dimensional FFT data as S m (ω i ,ω n ), where ω i and ω n They represent the serial numbers of the range dimension and Doppler dimension FFT respectively, and their matrix form is recorded as S m ;

[0053] S102, in order to improve the signal-to-noise ratio of the detection as much as possible, the two-dimensional FFT data S m (ω i ,ω n ) performs incoherent accumulation of M channels, and obtains Among them, S int is an I×N dimensional incoherent accumulation matrix; for S intPerform two-dimensional CFAR detection to detect the range unit and Doppler unit where the target is located, such as Figure 2 As shown, the index number is recorded as (r p ,d p ), where p = 1, ..., P, P is the total target point detected by CFAR; r p Indicates the distance dimension index of the pth target; d p Represents the velocity dimension index of the pth target.

[0054] S103, according to the index number detected by CFAR, the frequency domain single snapshot array data S of the pth target can be obtained. P , S P =[S 1 (r p ,d p ),S 2 (r p ,d p ),…,S M (r p ,d p )] T ,

[0055] In step S2, the covariance matrix of the array frequency domain single snapshot data is calculated, and the eigenvalue decomposition is performed to obtain the noise subspace. The specific implementation process is as follows:

[0056] S201, using single snapshot data S P Estimated covariance matrix

[0057] S202, perform eigenvalue decomposition on the matrix R, and regard the K largest eigenvalues ​​as the signal subspace U in order of the eigenvalue size. S The remaining MK eigenvalues ​​corresponding to the eigenvectors are regarded as the noise subspace U N The noise subspace U N Used to calculate spatial spectrum functions.

[0058] In step S3, within the viewing angle range of the target, the angular resolution is used as the search step, and the spatial spectrum search of the incoming wave directions of multiple targets is performed in combination with the noise subspace to obtain rough angle estimation values ​​of multiple targets. The specific implementation process is as follows:

[0059] S301, in order to reduce the number of searches, set the angle search range θ according to the viewing angle range where the actual target is located min ≤θ≤θ max ; and according to the minimum spacing between the two targets and the number of array elements, the required angular resolution Δθ is determined, and this is used as the search step of the spatial spectrum search;

[0060] S302, according to the angle search range [θ min ,θ max ] and search step Δθ, change search value θ i , where θ i =θ min +(l-1)Δθ,l=1,2,…,L, L is the total number of searches, L=(θ max -θ min ) / Δθ, and perform spatial spectrum peak search according to the following spatial spectrum function:

[0061]

[0062] Among them, a(θ i ) is the direction-steering vector of the array; P(θ i ) The peak value corresponds to θ i It is a rough estimate of the incoming direction of the target signal; the superscript H represents the conjugate transpose.

[0063] In this step, the angle search is performed with the angular resolution Δθ as a step. The search step is large, which greatly reduces the number of searches. However, this step can only achieve the purpose of distinguishing multiple target angles and obtaining the rough directions of multiple targets. In order to further improve the angle estimation accuracy and reduce the number of searches as much as possible, a refined search in step S4 is performed.

[0064] In step S4, the search range is determined based on the rough angle estimation values ​​of multiple targets and the search step, and the golden section algorithm is used for fine search to obtain the accurate angle estimation values ​​of multiple targets. The specific implementation process is as follows:

[0065] S401: Assume the rough angles of the K targets identified in S3 Where k = 1, ..., K, for each target angle Based on the search step in S3, The search scope for a refined search.

[0066] S402, the search range in S401 is used as the golden section interval, and the angle accuracy requirement is set as the minimum value of the golden section three-element point difference value as the termination condition for the golden section to stop dividing, the golden section algorithm is used to find the golden section point as the search angle, and the spatial spectrum function is used to calculate the spectrum peak value, which reduces the number of searches while ensuring the accuracy of multi-target angle estimation. The principle of the golden section algorithm is as follows:

[0067] Assume that function f(x): has one and only one maximum value on the interval (a,c), where point b is located at the golden section point of the interval (a,c). Select the golden section point x of the interval (b,c) to form a ternary point. When f(b)>f(x), f(b)>f(a) and f(b)>f(x) are satisfied, that is, the maximum point will be located on the interval (a,x) corresponding to the ternary point a,b,x; otherwise, when f(b)<f(x), f(x)>f(b)>f(c) are satisfied, that is, the maximum point will be located on the interval (b,c) corresponding to the ternary point group b,x,c. In this way, continue to calculate and divide the interval corresponding to the ternary point group, the interval range will become smaller and smaller, until the difference of the ternary points in the interval meets the accuracy requirements, then stop dividing and output the peak value.

[0068] In order to verify the effectiveness of the embodiment of the present invention, the embodiment of the present invention conducted the following tests:

[0069] In order to compare the MUSIC angle measurement effect of using multiple snapshots of original echo signals in the time domain and single snapshot data in the frequency domain after coherent accumulation, the present invention uses the original data and the single snapshot data after two-dimensional FFT to solve the covariance matrix respectively and construct the spatial spectrum. The spatial spectrum diagrams are shown as follows: Figure 3 and Figure 4 As shown. Figure 3 and Figure 4 It can be seen that the spatial spectrum obtained using single snapshot data has similar peak sharpness and peak intensity to the original data of multiple snapshots and is not affected. This is because the single snapshot data is obtained after two-dimensional FFT, which has the coherent accumulation of two-dimensional multiple snapshots, reducing the amount of algorithm calculation.

[0070] In order to further test whether the spatial spectrum search can reduce the running time by using coarse search and fine search, the present invention uses the TI1843 DSP environment with a main frequency of 600MHz for comparative tests. In the test, two goals are set, the angle range is ±10°, the angle resolution index is 3°, and the angle accuracy is 0.2°. That is to say, the search range is set to ±10° and the search step is 3° in the coarse search. The golden section iteration stop condition in the fine search is that the interval range is not greater than 0.2°. For the traditional spatial spectrum search, with the traditional 0.2° traversal of the ±10° angle range, about 100 spatial spectrum searches are required, and the processing time required is about 25.2ms after testing; while using the search method of the present invention, only about 25 spatial spectrum searches are required, and the processing time required is about 6.1ms after testing. At the same time, it can be obtained through analysis that the higher the angle accuracy requirement and the wider the viewing angle range, the more obvious the advantage of the present invention in shortening the spatial spectrum search time.

[0071] In summary, compared with the prior art, the present invention has the following beneficial effects:

[0072] 1. The embodiment of the present invention minimizes the amount of calculation and the memory space occupied while ensuring the angle estimation accuracy and angular resolution of the algorithm, adopts a single snapshot to solve the covariance matrix, and combines the golden section to perform spatial spectrum search, thereby greatly improving the real-time performance of the MUSIC algorithm.

[0073] 2. The embodiment of the present invention uses single-snapshot data in the frequency domain to solve the covariance matrix. After comparative analysis, it has similar angular resolution and angular accuracy as the multi-snapshot in the time domain. At the same time, in terms of spatial spectrum search, the golden section method is used to achieve high-precision angle estimation under multiple targets. The wider the viewing angle range, the higher the angular accuracy requirement, and the more obvious the reduction in the number of searches. After analysis and verification, under the demand for high-precision angle measurement, the number of searches can be reduced to approximately one-third of the traditional search number with the same accuracy. After comparative analysis, the fast MUSIC algorithm proposed in the embodiment of the present invention greatly reduces the computing time and memory space, is convenient for hardware implementation, and has high practical application value.

[0074] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0075] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A fast radar single-shot MUSIC angle measurement method, characterized in that: include: S1. After performing two-dimensional FFT processing on each channel of the radar array echo signal, CFAR detection is used to obtain the array frequency domain single snapshot data; S2, calculating the covariance matrix of the single snapshot data in the frequency domain of the array, and performing eigenvalue decomposition to obtain the noise subspace; S3, within the viewing angle range of the target, using the angular resolution as the search step, combined with the noise subspace, perform a spatial spectrum search of the incoming wave directions of multiple targets to obtain rough angle estimates of multiple targets; S4. Determine the search range based on the rough angle estimates of multiple targets and the search step, and use the golden section algorithm to perform a precise search to obtain accurate angle estimates of multiple targets.

2. The fast radar single-shot MUSIC angle measurement method as claimed in claim 1, characterized in that: The S1 specifically includes: S101, radar array echo signal s m (i,n) performs FFT transformation of N points in the first dimension along the direction of distance dimension n, and then performs FFT transformation of I points in the second dimension along the direction of period i, and records the two-dimensional FFT data as S m (ω i ,ω n ), where ω i and ω n Represent the sequence numbers of the range dimension and Doppler dimension FFT respectively; S102, S m (ω i ,ω n ) performs incoherent accumulation of M channels, and obtains Among them, S int is an I×N dimensional incoherent accumulation matrix; for S int Perform two-dimensional CFAR detection and obtain the index number of the target point, recorded as (r p ,d p ), where p = 1, ..., P, P is the total target point detected by CFAR; r p Indicates the distance dimension index of the pth target; d p represents the velocity dimension index of the pth target; M is the number of radar receiving antenna array elements; S103, according to the index number detected by CFAR, obtain the frequency domain single snapshot array data S of the pth target P , S P =[S1(r p ,d p ),S2(r p ,d p ),…,S M (r p ,d p )] T .

3. The fast radar single-shot MUSIC angle measurement method as claimed in claim 1, characterized in that: The S2 specifically includes: S201, based on frequency domain single snapshot array data S P Estimated covariance matrix S202, perform eigenvalue decomposition on the matrix R to obtain the noise subspace U N , used to calculate the spatial spectrum function.

4. The fast radar single-shot MUSIC angle measurement method as claimed in claim 3, characterized in that: The S202 includes: Perform eigenvalue decomposition on the matrix R to obtain M eigenvalues. In order of the eigenvalue size, the K largest eigenvalues ​​are regarded as the signal subspace U S The remaining MK eigenvalues ​​corresponding to the eigenvectors are regarded as the noise subspace U N ; Wherein, M is the number of radar receiving antenna elements; K is the number of targets.

5. The fast radar single-shot MUSIC angle measurement method as claimed in claim 1, characterized in that: The S3 specifically includes: S301, according to the viewing angle range of the actual target, set the angle search range θ min ≤θ≤θ max ; According to the minimum spacing between the two targets and the number of array elements, the required angular resolution Δθ is determined and used as the search step of the spatial spectrum search; S302, according to the angle search range [θ min ,θ max ] and search step Δθ, change search value θ i ,θ i =θ min +(l-1)Δθ,l=1,2,…,L, L is the total number of searches, L=(θ max -θ min ) / Δθ, and perform spatial spectrum peak search according to the following spatial spectrum function: Among them, a(θ i ) is the direction-steering vector of the array; P(θ i ) The peak value corresponds to θ i It is a rough estimate of the incoming direction of the target signal; the superscript H represents the conjugate transpose.

6. The fast radar single-shot MUSIC angle measurement method as claimed in claim 1, characterized in that: The S4 specifically includes: S401: Assume the rough angle estimation values ​​of the K targets identified in S3 where k = 1, ..., K, for each target rough angle estimate Based on the search step Δθ in S3, As the search scope for refined search; S402, taking the search range in S401 as the golden section interval, and setting the angle accuracy requirement as the minimum value of the golden section three-element point difference as the termination condition for stopping the golden section division, using the golden section algorithm to find the golden section point as the accurate angle estimation value of K targets.

Images