DOA estimation method, medium and equipment for distributed millimeter-wave radar networks

Abstract

The present invention provides a DOA estimation method, medium and device for distributed millimeter-wave radar networks, which belongs to the field of radar signal processing technology. After forming a large-aperture MIMO radar system based on multiple small millimeter-wave radars, the present invention uses the steering vector of the two-dimensional MIMO equivalent virtual plane array to construct a two-dimensional single-snapshot sparse array signal model, and by constructing a block Hankel matrix and applying matrix completion technology, the aperture missing in the sparse array is interpolated and extrapolated. On this basis, a fast two-dimensional angle estimation algorithm based on phase coherence factor weighting is adopted to achieve efficient joint estimation of the target azimuth and pitch angle. The present invention can be flexibly applied to different types of distributed millimeter-wave radar networks, effectively overcoming the DOA estimation sidelobe and grating lobe interference effects in the prior art, while improving the signal-to-noise ratio level of the array response.

Description

Technical Field

[0001] The present invention belongs to the field of radar signal processing, and in particular relates to a DOA estimation method, medium and device for a distributed millimeter-wave radar network. Background Art

[0002] Distributed millimeter-wave radar networks, as an emerging detection technology, demonstrate tremendous potential for applications in areas such as autonomous driving and intelligent security, thanks to their high angular resolution, strong anti-interference capabilities, and flexible networked deployment. Direction of Arrival (DOA) estimation, a core component of radar signal processing, directly impacts the accuracy of target positioning and tracking. However, DOA estimation in distributed millimeter-wave radar networks faces numerous challenges. First, due to the sparse spatial distribution of radar units in the network, traditional array signal processing methods struggle to directly construct a complete virtual aperture, resulting in limited angular resolution. Second, sparse array structures are prone to introducing grating and sidelobe interference, reducing the robustness of target angle estimation. Existing methods, most of which are based on centralized arrays or fixed topology designs, struggle to adapt to the dynamic characteristics and high real-time requirements of distributed networks, severely limiting the performance of millimeter-wave radar networks in multi-target scenarios. Summary of the Invention

[0003] This paper addresses the issues of missing aperture, severe interference between grating lobes and side lobes, and insufficient angle estimation accuracy in multi-target scenarios caused by sparse array structures in distributed millimeter-wave radar networks. By innovating sparse array signal reconstruction and efficient angle estimation algorithms, this paper overcomes the bottlenecks of missing aperture and interference suppression in existing technologies, while improving the algorithm's flexibility and adaptability. This provides key technical support for achieving high-precision, low-complexity target angle estimation in distributed millimeter-wave radar networks.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] In a first aspect, the present invention provides a DOA estimation method for a distributed millimeter-wave radar network, comprising the following steps:

[0006] S1: Based on a Multiple-Input Multiple-Output (MIMO) radar system formed by multiple millimeter-wave radars, the steering vector of the two-dimensional MIMO equivalent virtual planar array is obtained, and then a two-dimensional single-snapshot sparse array signal model is constructed;

[0007] S2: Introduce a uniform two-dimensional array with the same aperture size as the two-dimensional single-snapshot sparse array signal model and construct a block Hankel matrix;

[0008] S3: Based on the block Hankel matrix, a matrix completion method is constructed and used to complete the aperture missing in the two-dimensional single-snapshot sparse array signal model. After completion, the full array echo signal is obtained.

[0009] S4: For the full array echo signal, a two-dimensional angle estimation algorithm based on phase coherence factor weighting is used to jointly estimate the target azimuth and elevation angles.

[0010] Optionally, in step S1, the process of obtaining the steering vector of the two-dimensional MIMO equivalent virtual plane array is:

[0011] Based on the MIMO radar system, according to the far-field assumption, the steering vectors of the transmitting array and the receiving array are derived according to the echo signal. and , and then get the azimuth and pitch angle Steering vectors to a 2D MIMO equivalent virtual plane array :

[0012] ;

[0013] in, represents the Kronecker product.

[0014] Optionally, in step S1, the process of constructing the two-dimensional single-snapshot sparse array signal model is:

[0015] Set up a single snapshot array containing Target, The azimuth and elevation angles of the target are , the backscattering coefficient is , , in the absence of noise, the following two-dimensional single-snapshot sparse array signal model is constructed:

[0016] ;

[0017] in, is a two-dimensional single snapshot array response No. elements, Indicates the direction angle and pitch angle To the steering vector of the 2D MIMO equivalent virtual plane array, Indicates the first The position of the element, represents the echo wavelength, is an exponential function.

[0018] Optionally, the process of step S2 is:

[0019] Introduction and Uniform two-dimensional array of the same aperture size , construct a The block Hankel matrix of :

[0020] ;

[0021] ;

[0022] in, , , is a The Hankel matrix of yes Middle elements.

[0023] Optionally, when and hour, The rank of , There exists a Vandermonde factorization as follows:

[0024] ;

[0025] in, is the subarray popularity matrix, is a sparse diagonal matrix representing the target scattering characteristics. The superscript T indicates the transpose.

[0026] Optionally, in step S3, the matrix completion method is expressed as the following matrix completion problem:

[0027] ;

[0028] in, and is the regularization parameter used for balancing, represents the L1 norm, represents the nuclear norm, represents the F norm, is the sampling operator, According to the observation set Extract Matrix Known entries in is the regularization parameter used to control the noise level, Indicates constraints.

[0029] Optionally, the process of step S4 is:

[0030] For full array echo signal , construct a two-dimensional DBF output response for:

[0031] ;

[0032] ;

[0033] in, represents the azimuth-elevation two-dimensional spatial spectrum amplitude distribution, , , the observation space is discretized into a uniform azimuth-elevation grid, Indicates the azimuth and pitch angle The four-dimensional popular matrix of the two-dimensional matrix, the superscript H represents the conjugate transpose, express About Azimuth Hedi Pitch angle The first two-dimensional matrix elements, express Middle elements, is the calculated variable;

[0034] Based on the standard deviation of the phase, the direction Phase coherence factor weighting coefficient for:

[0035] ;

[0036] ;

[0037] in, is the complex signal phase extraction operator, is the standard deviation operation;

[0038] use right Make the following corrections:

[0039] ;

[0040] in, represents the weighted spatial spectrum, is the control factor;

[0041] By weighting the spatial spectrum Perform a two-dimensional peak search to obtain the azimuth and elevation angles of the target.

[0042] In a second aspect, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the DOA estimation method for a distributed millimeter-wave radar network as described in the first aspect.

[0043] In a third aspect, the present invention provides an electronic device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the DOA estimation method for a distributed millimeter-wave radar network as described in the first aspect is implemented.

[0044] The beneficial effects of the present invention are:

[0045] (1) The present invention obtains reconstructed uniform array data through matrix completion, which not only effectively fills the missing data of the sparse array, but also achieves the following three core improvements by restoring the complete array manifold structure: suppressing sidelobe and grating lobe effects; improving the array signal-to-noise ratio; and maintaining the array aperture gain.

[0046] (2) The present invention is based on a fast two-dimensional angle estimation algorithm. While ensuring high-precision DOA estimation, it significantly reduces computational complexity, meets the stringent requirements of distributed millimeter-wave radar networks for low power consumption and real-time performance, and facilitates hardware implementation and engineering deployment.

[0047] (3) The present invention can flexibly adapt to different types of distributed millimeter-wave radar networks, support the collaborative work of heterogeneous radar units (such as those with different numbers of antennas or hardware capabilities), adapt to dynamic changes in network topology, and expand the application scope of DOA estimation technology in complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a flow chart of a DOA estimation method for distributed millimeter-wave radar networks;

[0049] Figure 2 Schematic diagram of a one-dimensional distributed millimeter-wave radar network;

[0050] Figure 3 A location diagram of physical and virtual transmitting array elements in a one-dimensional array distributed millimeter-wave radar network;

[0051] Figure 4 A comparison diagram of the method proposed in the present invention and the traditional digital beam forming (DBF) method;

[0052] Figure 5 A location diagram of the physical transmitting array elements in a two-dimensional array distributed millimeter-wave radar network;

[0053] Figure 6 It is a position diagram of virtual transmitting array elements in a two-dimensional array distributed millimeter wave radar network;

[0054] Figure 7 This is the result graph of the traditional DBF method;

[0055] Figure 8 This is a result diagram of the method proposed in the present invention. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0057] In one embodiment, the present invention proposes a DOA estimation method for a distributed millimeter wave radar network, such as Figure 1 As shown, the method includes the following steps:

[0058] S1: After forming a large-aperture MIMO radar system based on multiple small millimeter-wave radars, the steering vector of the two-dimensional MIMO equivalent virtual plane array is obtained, and a two-dimensional single-snapshot sparse array signal model is constructed.

[0059] In this embodiment, after integrating the radar units to form a large aperture MIMO radar system, the steering vectors of the transmitting array and the receiving array are derived according to the echo signal based on the far field assumption. and , and then get the azimuth and pitch angle Steering vectors to a 2D MIMO equivalent virtual plane array :

[0060] ;

[0061] in, represents the Kronecker product.

[0062] So, suppose a single snapshot array contains Target, The azimuth and elevation angles of the target are , the backscatter coefficient is , , in the absence of noise, the 2D single snapshot array responds No. The elements can be represented as:

[0063] ;

[0064] in, Indicates the direction angle and pitch angle To the steering vector of the 2D MIMO equivalent virtual plane array, Indicates the first The position of the element, represents the echo wavelength, is an exponential function.

[0065] S2: Introduce a uniform two-dimensional array with the same aperture size as the two-dimensional single-snapshot sparse array signal model and construct a block Hankel matrix.

[0066] In this embodiment, the introduction Uniform two-dimensional array of the same aperture size , construct a The block Hankel matrix of :

[0067] ;

[0068] ;

[0069] in, , , is a The Hankel matrix of yes Middle elements.

[0070] when and hour, The rank of , There exists a Vandermonde factorization as follows:

[0071] ;

[0072] in, is the subarray popularity matrix, is a sparse diagonal matrix representing the target scattering characteristics. The superscript T indicates the transpose.

[0073] S3: Construct a sparse and low-rank constrained matrix completion method to perform data interpolation and extrapolation on the aperture missing in sparse arrays.

[0074] In this embodiment, since the matrix There are many missing elements (unobserved items are set to zero), so can be considered as In the presence of noise, combining the sparsity of the target diagonal matrix and the block Hankel matrix Due to the low rank property of , the matrix completion problem is formulated as:

[0075] ;

[0076] in, represents the L1 norm, represents the nuclear norm, represents the F norm, is a sampling operator, According to the observation set Extract Matrix The known entries in , the regularization parameter and Plays a balancing role, It is also a regularization parameter used to control the noise level. Indicates constraints.

[0077] S4: For the completed full array echo signal, a fast two-dimensional angle estimation algorithm based on phase coherence factor weighting is used to achieve efficient joint estimation of the target azimuth and pitch angle.

[0078] In this embodiment, when the signal After completing the matrix completion, the estimated full array echo signal can be obtained , construct a two-dimensional DBF output response :

[0079] ;

[0080] ;

[0081] in, represents the azimuth-elevation two-dimensional spatial spectrum amplitude distribution, , , discretize the observation space into A uniform azimuth-elevation grid with grid spacing set according to the angular resolution; Indicates the azimuth and pitch angle The four-dimensional popular matrix of the two-dimensional matrix, the superscript H represents the conjugate transpose, express About Azimuth Hedi Pitch angle The first two-dimensional matrix elements, express Middle elements;

[0082] Define Direction The weight coefficient at , based on the standard deviation of the phase, the phase coherence factor weighting coefficient in the azimuth-elevation direction is generated as:

[0083] ;

[0084] ;

[0085] in, is the complex signal phase extraction operator, is the standard deviation operation;

[0086] use Corrected DBF output:

[0087] ;

[0088] in, is the control factor, which is between 0 and 1.

[0089] By weighting the spatial spectrum Perform two-dimensional peak search to accurately obtain the azimuth and elevation angles of the target.

[0090] Next, the effectiveness of the method proposed in the present invention is illustrated by combining specific experiments.

[0091] Figure 2 A one-dimensional distributed millimeter-wave radar network structure is demonstrated, which consists of two independent radar units, which are coherently cascaded through local oscillator and trigger signal synchronization to form a coherent radar system, and each radar is equipped with two transmitters and four receivers. Figure 3 Shown Figure 2 The positions of the physical transmitting array elements and virtual transmitting array elements in the radar network structure. It can be seen that 32 virtual receiving array elements can form a virtual array with an aperture of 74d (d is equal to half a wavelength). Figure 4 The comparison diagram of the proposed method and the traditional DBF method is shown. The proposed method has the highest beam corresponding resolution, with a theoretical limit of 1.1°, and the side lobe is as low as -30dB or less. Figure 4 It is judged by the maximum side lobe amplitude except the main lobe.

[0092] Figure 5 The physical transmitting array element position of a two-dimensional array distributed millimeter wave radar network structure is shown, and the corresponding virtual transmitting array element position is Figure 6 Each sensor is equipped with four transmit elements and four receive elements, forming a total of up to 64 physical channels to synthesize a MIMO two-dimensional virtual array in the area of [0,107](d)×[0,13](d). Figure 7 and Figure 8The spectra of two targets at azimuth and elevation angles of (10°, -10°) and (-20°, 15°) are shown. The sparse arrays used adopt the traditional DBF method and the method proposed in this paper. Figure 7 and Figure 8 The color bar on the right represents the normalized amplitude (dB). Both methods can produce peaks at the corresponding positions, however, the side lobes obtained by the method proposed in this invention are significantly lower.

[0093] Under the same experimental conditions as above, different SNRs are set for two different network structures. The test results are compared with the traditional DBF method as shown in Table 1. It can be seen that the method proposed in this invention can not only improve the resolution but also reduce the maximum sidelobe level, which illustrates the effectiveness and correctness of the method proposed in this invention for DOA estimation of distributed millimeter-wave radar networks.

[0094] Table 1 Comparative test results of the method proposed in this invention and the traditional DBF method

[0095]

[0096] In summary, this invention compensates for the missing virtual aperture by constructing a block Hankel matrix structure and using matrix completion techniques to interpolate and extrapolate sparse array data. It also combines phase coherence factor weighting to suppress grating lobe and sidelobe interference, thereby improving the signal-to-noise ratio of the array response. The proposed method ensures high-precision angle estimation while maintaining low computational complexity, adapting to the real-time processing requirements of distributed millimeter-wave radar networks and providing a reliable solution for efficient and robust DOA estimation of multiple targets in complex environments.

[0097] In another embodiment, the present invention provides a computer-readable storage medium storing a computer program, which enables a computer to execute the DOA estimation method for a distributed millimeter-wave radar network of the aforementioned embodiment.

[0098] In another embodiment, the present invention proposes an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the DOA estimation method for a distributed millimeter-wave radar network of the aforementioned embodiment is implemented.

[0099] In the embodiments disclosed herein, computer storage media may be tangible media that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. Computer storage media may include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media may include electrical connections based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CDROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0100] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0101] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions based on the principles of the present invention are within the scope of protection of the present invention. It should be noted that for those skilled in the art, various improvements and modifications that do not depart from the principles of the present invention should be considered within the scope of protection of the present invention.

Claims

1. A DOA estimation method for distributed millimeter-wave radar networks, characterized in that: The steps include: S1: Based on a MIMO radar system formed by multiple millimeter-wave radars, a steering vector of a two-dimensional MIMO equivalent virtual plane array is obtained, and then a two-dimensional single-snapshot sparse array signal model is constructed. In step S1, the process of constructing the two-dimensional single-snapshot sparse array signal model is as follows: Assume a single snapshot array contains P targets, the direction angle and pitch angle of the pth target are The backscatter coefficient is β p , p=1,...,P, in the absence of noise, the following two-dimensional single-snapshot sparse array signal model is constructed: Among them, [M] m,n is the (m,n)th element of the two-dimensional single snapshot array response M, Indicates the direction angle θ p and pitch angle To the steering vector of the 2D MIMO equivalent virtual plane array, represents the position of the (m,n)th element in the two-dimensional MIMO equivalent virtual plane array, λ represents the echo wavelength, and exp is the exponential function; S2: Introduce a uniform two-dimensional array with the same aperture size as the two-dimensional single-snapshot sparse array signal model and construct a block Hankel matrix; the process of step S2 is: Introduce a uniform two-dimensional array with the same aperture size as M Construct an N1×(M1-N1+1) block Hankel matrix X E : in, Y m is a L×(M2-L+1) Hankel matrix, x m,L is the (m+1,L+1)th element in X; When N1≥r and L≥r, X E The rank of X is r, E There exists a Vandermonde factorization as follows: X E =B∑B T ; Where B is the subarray popularity matrix, ∑ is a sparse diagonal matrix representing the target scattering characteristics, and the superscript T indicates the transpose; S3: Based on the block Hankel matrix, a matrix completion method is constructed and used to complete the missing aperture in the two-dimensional single-snapshot sparse array signal model. After completion, the full array echo signal is obtained. In step S3, the matrix completion method is expressed as the following matrix completion problem: Among them, τ p and τ r is the regularization parameter used for balancing, ||·||1 represents the L1 norm, ||·|| * represents the nuclear norm, ||·|| F represents the F norm, P Ω is the sampling operator, P Ω (XM) represents the known entries in the matrix XM extracted according to the observation set Ω, δ is the regularization parameter used to control the noise level, and st represents the constraint condition; S4: For the full array echo signal, a two-dimensional angle estimation algorithm based on phase coherence factor weighting is used to jointly estimate the target azimuth and elevation angles.

2. The DOA estimation method for a distributed millimeter-wave radar network according to claim 1, wherein: In step S1, the process of obtaining the steering vector of the two-dimensional MIMO equivalent virtual plane array is: Based on the MIMO radar system, according to the far-field assumption, the steering vectors of the transmitting array and the receiving array are derived according to the echo signal. and Then we can get the azimuth angle θ and the pitch angle Steering vectors to a 2D MIMO equivalent virtual plane array in, represents the Kronecker product.

3. The DOA estimation method for a distributed millimeter-wave radar network according to claim 1, wherein: The process of step S4 is: For full array echo signal Construct a 2D DBF output response for: Among them, I jk represents the azimuth-elevation two-dimensional spatial spectrum amplitude distribution, j = 1, ..., J, k = 1, ..., K, and the observation space is discretized into J × K uniform azimuth-elevation grids. Indicates the azimuth angle θ and the pitch angle The four-dimensional popular matrix of the two-dimensional matrix, the superscript H represents the conjugate transpose, a j,k (m1,m2) means Regarding the j-th azimuth angle θ j and the kth pitch angle The (m1+1,m2+1)th element of the two-dimensional matrix, express The (m1+1,m2+1)th element in s j,k (m1,m2) are calculation variables; Based on the standard deviation of the phase, the direction Phase coherence factor weighting coefficient for: φ j,k =angle[s j,k (0,0),s j,k (0,1),...,s j,k (0,M1-1),s j,k (1,0),...,s j,k (M1-1,M2-1)]; Among them, angle is the complex signal phase extraction operator, std is the standard deviation operation; use to I jk Make the following corrections: Among them, I′ jk represents the weighted spatial spectrum, u is the control factor; By weighting the spatial spectrum I′ jk Perform a two-dimensional peak search to obtain the azimuth and elevation angles of the target.

4. A computer-readable storage medium storing a computer program, characterized in that: The computer program enables a computer to execute the DOA estimation method for a distributed millimeter-wave radar network as described in any one of claims 1 to 3.

5. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the DOA estimation method for a distributed millimeter-wave radar network is implemented as described in any one of claims 1 to 3.

Images