MIMO radar tensor decomposition four-dimensional parameter joint estimation method and device

Abstract

The invention provides an MIMO radar tensor decomposition four-dimensional parameter joint estimation method and device, and belongs to the technical field of radar signal processing. The method comprises the following steps: acquiring a radar echo signal received by an L-shaped receiving array after a space-time coding transmitting array of the MIMO radar transmits the radar echo signal, and reconstructing the radar echo signal into a third-order tensor; performing parallel factorization on the third-order tensor to obtain a corresponding factor matrix; and based on the factor matrix, extracting four-dimensional parameters of a target corresponding to the signal, including a two-dimensional direction of arrival, a wave departure direction and an unambiguous Doppler frequency. According to the method, through single tensor decomposition, the technical problems of complex parameter pairing, Doppler ambiguity and large combined estimation calculation amount in a traditional method are solved, and combined, unambiguous and automatic pairing high-precision estimation of four-dimensional parameters is realized.

Description

Technical Field

[0001] This invention belongs to the field of radar signal processing and array signal processing, and specifically relates to a method and apparatus for joint estimation of four-dimensional parameters of tensor decomposition in MIMO radar. Background Technology

[0002] With the increasing complexity of modern radar applications, such as Advanced Driver Assistance Systems (ADAS), autonomous driving, UAV surveillance and defense, and intelligent transportation systems, unprecedented demands are placed on the ability to accurately perceive target states. Simply acquiring target position information is far from sufficient; its motion state must be estimated simultaneously and accurately. Therefore, jointly estimating the target's two-dimensional direction of arrival (DOA, i.e., azimuth and elevation angles), direction of departure (DOD), and radial velocity has become a core research direction for improving radar system performance. The parameter set composed of the two-dimensional DOA (including azimuth and elevation angles), the one-dimensional DOD, and the one-dimensional radial velocity is defined as the target's four-dimensional parameters.

[0003] Multiple-input multiple-output (MIMO) radar effectively improves angular resolution and parameter estimation capabilities through transmit and receive virtual aperture extension. However, existing methods face numerous challenges in solving the four-dimensional parameter estimation problem. For example, while L-shaped receiver arrays can effectively estimate two-dimensional DOA, as passive arrays, they cannot acquire DOD information. Spatiotemporally coded MIMO (ST-MIMO) radar technology can significantly improve angular resolution by assigning orthogonal Doppler frequency shifts to different transmit antennas, but it is usually limited to one-dimensional angle estimation and introduces the problem of reduced Doppler ambiguity range.

[0004] More importantly, when multiple parameters need to be jointly estimated, how to correctly associate parameters obtained from different dimensions and through different algorithms (such as DOA, DOD, Doppler) with the same physical target, i.e., the "parameter pairing" problem, is a major computational obstacle. Traditional methods often require complex spectral peak search and matching, which is computationally intensive and prone to errors.

[0005] Tensor decomposition has become a powerful mathematical tool for solving multidimensional data analysis problems. However, how to construct a radar signal model that can elegantly transform the four-dimensional parameter estimation problem into a single tensor decomposition task, and solve the key technical challenges (such as Doppler blurring), remains a problem that current technologies have not been able to effectively address. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a method and apparatus for joint estimation of four-dimensional parameters in MIMO radar tensor decomposition. This invention combines high accuracy with high computational efficiency, and can robustly achieve automatic pairing and unambiguous estimation of four-dimensional parameters for multiple targets. This invention can be widely applied to scenarios with extremely high requirements for real-time performance and reliability, such as autonomous driving and intelligent transportation, and has significant engineering application value.

[0007] The first aspect of this invention proposes a joint estimation method for four-dimensional parameters of MIMO radar tensor decomposition, comprising:

[0008] The radar echo signal received by the L-type receiving array after the spatiotemporal encoded transmitting array of the MIMO radar is obtained and reconstructed into a third-order tensor.

[0009] Parallel factorization is performed on the third-order tensor to obtain the corresponding factor matrix;

[0010] Based on the factor matrix, the four-dimensional parameters of the target corresponding to the signal are extracted, including: two-dimensional direction of arrival, direction of departure, and unambiguous Doppler frequency.

[0011] In one specific embodiment of the present invention, it further includes:

[0012] Make the transmitter of the MIMO radar system a device containing A uniform linear array of elements, with the receiving end being an L-shaped array; this L-shaped array consists of two array elements sharing a common origin, one along the x-axis and the other along the z-axis. The receiver consists of a uniform linear array of elements, and contains a total of [number] elements. Each array element is independent; the transmitting end adopts the spatiotemporal coding method of Doppler frequency division multiple access;

[0013] Let there be K far-field targets in space. All pulse echo data received by the radar within one coherent processing interval are down-converted, matched-filtered, and sampled, then arranged into a received data matrix X. This data matrix X is then reconstructed into a third-order tensor Y. Where, if the total number of transmitted pulses within one coherent processing interval is Q = P × If there are 1 pulse, then the size of the tensor Y is ( ) × × P, where parameter P represents the number of emitter cycles.

[0014] In a specific embodiment of the present invention, the parallel factorization of the third-order tensor includes:

[0015] The third-order tensor Y is modeled as a rank K PARAFAC model; the tensor Y is decomposed to obtain three factor matrices: A, B, and C; where:

[0016]

[0017]

[0018]

[0019] in, , , These are the factor vectors of the three dimensions of the tensor, which are related to the k-th objective.

[0020] In one specific embodiment of the present invention, the tensor Y is decomposed using the alternating least squares method.

[0021] In one specific embodiment of the present invention, it further includes:

[0022] Two-dimensional direction of arrival (DOA) is estimated using factor matrix A;

[0023] Where each column of factor matrix A It is the steering vector of the k-th target on the L-shaped receiving array; using the L-shaped receiving array, two vectors are respectively along... shaft and A geometric structure consisting of uniform linear subarrays sharing a common origin element, along with an axis, guides the vector. Split into two sub-vectors and These correspond to the responses of target k on the x-axis and z-axis subarrays, respectively. The ESPRIT algorithm is then applied to these two subvectors to estimate the eigenvalues ​​in the two directions related to the target. and Then the cosines of the two directions of the target and The following formula is used to calculate:

[0024]

[0025] in, It is a normalized spatial angular frequency determined by the array parameters. For the spacing between array elements, The signal wavelength and direction cosine are given. and These are the spatial frequency components of the steering vector of target k along the z-axis and x-axis, respectively; based on and The azimuth angle of the k-th target is obtained by calculating using inverse trigonometric functions. and pitch angle This allows for the estimation of the two-dimensional direction of arrival.

[0026] In one specific embodiment of the present invention, it further includes:

[0027] The wave separation direction is estimated using the factor matrix B;

[0028] In which each column of factor matrix B It is a composite factor vector related to the k-th target, coupled with wave direction and coarse Doppler information;

[0029] Solve the following least squares optimization problem:

[0030]

[0031] in, It is the theoretical steering vector of the transmitting array; c is an undetermined normalization constant; | | is a vector The amplitude; This is an estimate of the wave separation direction of the k-th target;

[0032] The least squares optimization problem is solved using a one-dimensional spectral peak search method, yielding the following results. The optimization results.

[0033] In one specific embodiment of the present invention, it further includes:

[0034] The unambiguous Doppler frequencies are estimated using factor matrices B and C, and the specific steps are as follows:

[0035] 1) Extract fine Doppler information from the factor matrix C;

[0036] Where each column of the factor matrix C It is a steering vector corresponding to the slow-time sampling dimension; for The ESPRIT algorithm is applied to obtain the eigenvalues ​​extracted from the factor matrix C. ;

[0037] Fine Doppler estimate of the k-th target The following formula is used to calculate:

[0038]

[0039] in, The pulse repetition interval;

[0040] 2) Extract coarse Doppler information from factor matrix B;

[0041] In which each column of factor matrix B The phase simultaneously contains both wave direction and Doppler information; based on Construct a theoretical compensation vector related to the wave separation direction. :

[0042]

[0043] in, It is The ×1 transmit array steering vector, whose elements characterize the direction of target departure (DOD). The determined spatial phase difference;

[0044] W is a × The transmit coding matrix contains elements of a preset time phase modulation information used to distinguish different transmit antennas;

[0045] right and Element-by-element division is performed to obtain the coarse Doppler steering vector. :

[0046]

[0047] Where ⊘ represents element-wise division;

[0048] For vectors The ESPRIT algorithm is applied to obtain the eigenvalues ​​extracted from the factor matrix B. Then the coarse Doppler estimate of the k-th target The following formula is used to calculate:

[0049]

[0050] 3) By comparing the results of steps 1) and 2), the ambiguity number is determined, thereby obtaining the estimation result of the unambiguous Doppler frequency;

[0051] The expression for calculating the fuzzy number of the k-th target is as follows:

[0052]

[0053] Finally, the unambiguous, high-resolution Doppler estimate of the k-th target is:

[0054] .

[0055] A second aspect of the present invention provides a joint estimation device for four-dimensional parameters of MIMO radar tensor decomposition, comprising:

[0056] The signal third-order tensor construction module is used to obtain the radar echo signal received by the L-type receiving array after the spatiotemporally coded transmitting array of the MIMO radar is transmitted, and reconstruct it into a third-order tensor.

[0057] The factorization module is used to perform parallel factorization on the third-order tensor to obtain the corresponding factor matrix.

[0058] The four-dimensional parameter estimation module is used to extract four-dimensional parameters of the target corresponding to the signal based on the factor matrix, including: two-dimensional direction of arrival, direction of departure, and unambiguous Doppler frequency.

[0059] A third aspect of the present invention provides an electronic device comprising:

[0060] At least one processor; and a memory communicatively connected to said at least one processor;

[0061] The memory stores instructions that can be executed by the at least one processor, and the instructions are configured to perform the above-described MIMO radar tensor decomposition four-dimensional parameter joint estimation method.

[0062] A fourth aspect of the present invention provides a computer-readable storage medium storing computer instructions for causing the computer to execute the above-described MIMO radar tensor decomposition four-dimensional parameter joint estimation method.

[0063] The features and beneficial effects of this invention are as follows:

[0064] This invention transforms the four-dimensional parameter estimation problem into a single tensor decomposition. Utilizing the uniqueness of the PARAFAC decomposition, it achieves automatic pairing of all parameters, avoiding the complex multi-dimensional spectral peak search and matching process, thus significantly reducing computational complexity. It innovatively designs a method to separate coarse and fine Doppler information from different dimensions of the tensor. By combining these two methods, it fundamentally solves the Doppler ambiguity problem introduced by ST-MIMO encoding, expanding the measurement range of the maximum unambiguous velocity. This invention extracts all parameters through a single decomposition and subsequent low-complexity ESPRIT algorithm, eliminating the need for iterative searches and ensuring high efficiency and accuracy in estimation. This makes it highly suitable for applications with high real-time requirements. Attached Figure Description

[0065] Figure 1 This is an overall flowchart of a MIMO radar tensor decomposition four-dimensional parameter joint estimation method according to an embodiment of the present invention.

[0066] Figure 2 This is a visualization of the four-dimensional parameter estimation results in a multi-objective scenario according to a specific embodiment of the present invention.

[0067] Figure 3 This is a performance curve of composite angle error (CAE) versus signal-to-noise ratio (SNR) in a specific embodiment of the present invention.

[0068] Figure 4This is a performance curve of the root mean square error (RMSE) of Doppler frequency as a function of signal-to-noise ratio (SNR) in a specific embodiment of the present invention. Detailed Implementation

[0069] This invention proposes a method and apparatus for joint estimation of four-dimensional parameters of MIMO radar tensor decomposition. The invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0070] The first aspect of this invention proposes a joint estimation method for four-dimensional parameters of MIMO radar tensor decomposition, comprising:

[0071] The radar echo signal received by the L-type receiving array after the spatiotemporal encoded transmitting array of the MIMO radar is obtained and reconstructed into a third-order tensor.

[0072] Parallel factorization is performed on the third-order tensor to obtain the corresponding factor matrix;

[0073] Based on the factor matrix, the four-dimensional parameters of the target corresponding to the signal are extracted, including: two-dimensional direction of arrival, direction of departure, and unambiguous Doppler frequency.

[0074] In a specific embodiment of the present invention, the overall process of the MIMO radar tensor decomposition four-dimensional parameter joint estimation method is as follows: Figure 1 As shown, it includes the following steps:

[0075] 1) Obtain the radar echo signal received by the L-type receiving array after the MIMO radar is transmitted by the spatiotemporally coded transmitting array, and reconstruct it into a third-order tensor.

[0076] In this embodiment, a MIMO radar system is assumed, whose transmitter is a device containing... The receiver is an L-shaped array consisting of two array elements sharing a common origin, one along the x-axis and the other along the z-axis. The receiver consists of ULA arrays of 1000 elements, and contains a total of 1000 ULA arrays. The transmitter employs a Doppler frequency division multiple access (DDMA) spatiotemporal coding scheme, meaning that within one coherent processing interval (CPI), it generates the first independent array element. Each transmitting antenna modulates a unique preset Doppler frequency shift.

[0077] Assuming there are K far-field targets in space, in this embodiment, all pulse echo data received by the radar within one CPI are down-converted, matched-filtered, and sampled, and then arranged into a received data matrix X. Subsequently, this data matrix X is reconstructed into a third-order tensor Y. If the total number of transmissions within one CPI is Q = P × If there are 1 pulse, then the size of the tensor Y is ( ) × × P. Parameter P represents the number of emitter cycles, and its physical meaning is: within the entire coherent processing interval, by The number of times a transmission sub-cycle is repeated by each transmitting antenna to complete a full spatiotemporally encoded transmission. This parameter P constitutes the slow time dimension of the tensor, which is used to subsequently extract high-resolution fine Doppler information about the target.

[0078] In this embodiment, the tensor quantization operation gives the three dimensions of the tensor a clear physical meaning: the first dimension has a size of ( This corresponds to the spatial response of an L-shaped receiver array, whose structure contains the target's two-dimensional DOA information (i.e., azimuth and elevation angles); the second dimension, with a size of... The first dimension corresponds to the spatiotemporal coupling response of the transmission array, and its structure couples the target's DOD information and coarse Doppler frequency shift information; the third dimension, with a size of P, corresponds to slow-time sampling, and its phase change rate contains the target's high-resolution fine Doppler information.

[0079] 2) Perform parallel factorization (PARAFAC) on the third-order tensor obtained in step 1) to obtain the corresponding factor matrix.

[0080] In this embodiment, the third-order tensor Y constructed in step 1) can be modeled as a PARAFAC model with rank K. By decomposing the tensor Y using algorithms such as Alternating Least Squares (ALS), three factor matrices can be obtained: A, B, and C.

[0081]

[0082]

[0083]

[0084] in, , , These are the factor vectors of the three dimensions of the tensor, which are related to the k-th objective.

[0085] This embodiment leverages the key advantage of the uniqueness of the PARAFAC decomposition under mild conditions. This means that the column vector combination of the factor matrix obtained by the decomposition ( , , This naturally corresponds one-to-one with the k-th physical target. Therefore, all parameters (DOA, DOD, Doppler) subsequently extracted from this set of vectors are automatically paired to the same target, eliminating the need for complex parameter association processing, which greatly simplifies the algorithm and reduces the computational load.

[0086] 3) Based on the factor matrix obtained in step 2), extract the four-dimensional parameters of the target, including: two-dimensional direction of arrival (DOA), direction of departure (DOD), and unambiguous Doppler frequency.

[0087] In this embodiment, using the Vandermonde structure contained in each factor matrix obtained in step 2), all four-dimensional parameters are extracted through a multi-level ESPRIT algorithm and numerical optimization; the specific steps are as follows:

[0088] 3-1) Estimate the two-dimensional DOA using the factor matrix A obtained in step 2). In this embodiment, each column of factor matrix A... This is the steering vector of the k-th target on the L-shaped receiver array. Using the L-shaped receiver array, two vectors are respectively guided along... shaft and A geometric structure consisting of uniform linear subarrays (ULAs) sharing a common origin element can be used to guide the vector. Lossless split into two sub-vectors and These correspond to the responses of target k on the x-axis and z-axis subarrays, respectively. Applying the ESPRIT algorithm to these two subvectors allows us to estimate the eigenvalues ​​in the two directions related to the target, utilizing the rotation invariance between the subarrays. and According to the principle of the ESPRIT algorithm, the cosines of the two directions of the target... and It can be calculated using the following formula:

[0089] Among them, among them, It is a normalized spatial angular frequency determined by the array parameters, and is defined as follows: , For the spacing between array elements, The signal wavelength and direction cosine are given. and These are the spatial frequency components of the steering vector of target k along the z-axis and x-axis, respectively. Finally, the azimuth angle of the k-th target can be obtained by inverse trigonometric function calculation using standard spatial geometric relationships. and pitch angle This completes the estimation of two-dimensional DOA.

[0090] 3-2) Estimate the DOD using the factor matrix B obtained in step 2). In this embodiment, each column of the factor matrix B... The structure is determined by the target's DOD steering vector and coarse Doppler term. Its amplitude | | Only related to the DOD steering vector.

[0091] In this embodiment, each column of the factor matrix B It is a composite factor vector associated with the k-th target, coupled with DOD and coarse Doppler information. One of the key steps in the method described in this embodiment is to utilize the magnitude of this vector | |To separate DOD information. Since Doppler information is only reflected in the phase, its amplitude| The influence of the Doppler term has been eliminated, and the result is only related to the target's steering vector on the launch array. Proportional relationship.

[0092] Therefore, this embodiment estimates DOD by solving the following least-squares optimization problem:

[0093]

[0094] in, is the theoretical steering vector of the transmitting array, which is a known function that varies with angle θ; c is an undetermined normalization constant.

[0095] In practice, this optimization problem can be solved efficiently using a one-dimensional peak search method. Specifically, the angle to be estimated, θ, is discretized within its physical range (e.g., [0, π]). Then, for each discrete angle, the error norm (i.e., the degree of matching between the model and the data) in the above formula is calculated. Finally, the angle that minimizes the error norm (i.e., has the highest degree of matching) is selected as the estimated value of the direction of departure (DOD) of the k-th target. .

[0096] 3-3) Use the factor matrices B and C obtained in step 2) to estimate the unambiguous Doppler frequencies.

[0097] This step is crucial in solving the Doppler blur problem introduced by spatiotemporal coding in this embodiment. Its core idea is to fuse high-resolution, fine Doppler information with coarse Doppler information covering a large unambiguous range. The specific steps are as follows:

[0098] 3-3-1) Extract fine Doppler information from factor matrix C.

[0099] Each column of factor matrix C This is a steering vector corresponding to the slow-time sampling dimension, and its phase change rate directly reflects the Doppler frequency shift of the target. Because the slow-time sampling interval is relatively large (for...),... Therefore, this vector has high frequency resolution. This embodiment... By applying the ESPRIT algorithm and utilizing its internal rotation invariance, an eigenvalue can be obtained. To distinguish it from other eigenvalues ​​obtained in subsequent steps, the subscript 'c' here indicates that the eigenvalue was extracted from the factor matrix C. This represents a high-precision but somewhat ambiguous fine Doppler estimate of the k-th target. It can be calculated using the following formula:

[0100]

[0101] in, The pulse repetition interval is given. This estimate is highly accurate, but its unambiguous range is limited by the Nyquist sampling rate and is typically small.

[0102] 3-3-2) Extract coarse Doppler information from factor matrix B.

[0103] Each column of factor matrix B The phase contains both DOD and Doppler information. The estimated DOD value has been obtained in step 3-2). Then, a theoretical compensation vector related only to DOD can be constructed. The theoretical compensation vector This represents the theoretical spatiotemporal steering vector contributed solely by DOD. It is based on known radar system parameters (such as the transmit coding matrix W) and the DOD estimate obtained in step 3-2). It is constructed using this method. Specifically, its expression is:

[0104]

[0105] It is The ×1 "transmission array steering vector" has elements that characterize the direction of target deflection (DOD). The spatial phase difference is determined.

[0106] W is a × The "transmission coding matrix" contains elements of preset time phase modulation information used to distinguish different transmission antennas.

[0107] Therefore, product Coupling spatial phase with temporal coding yields the following results. This represents the theoretical "spatiotemporal response fingerprint" generated by the radar system itself for the DOD, without the target's own Doppler effect. After obtaining the theoretical compensation vector... Then, by comparing it with the actual factor vector By performing element-wise division (i.e., Hadamard division), the influence of DOD can be compensated for, thus obtaining a pure coarse Doppler steering vector. :

[0108]

[0109] Here, ⊘ represents element-wise division.

[0110] For this vector By applying the ESPRIT algorithm again, a feature value can be obtained. Here, the subscript 'b' indicates that the eigenvalue is extracted from the factor matrix B. The coarse Doppler estimate of the k-th target, which is fuzzy but has no large fuzzy range. It can be calculated using the following formula:

[0111]

[0112] This estimate is based on a shorter time interval. Its unambiguous range is much larger than that of fine Doppler estimation, but its resolution is lower.

[0113] 3-3-3) By comparing the results of steps 3-3-1) and 3-3-2), the ambiguity number is determined, thereby obtaining the estimation result of the unambiguous Doppler frequency.

[0114] The expression for calculating the fuzzy number of the k-th target is as follows:

[0115]

[0116] in, This is the pulse repetition interval.

[0117] Finally, the unambiguous, high-resolution Doppler estimate of the k-th target is:

[0118]

[0119] Through the above steps, this embodiment achieves { for each target} , , The estimation of four-dimensional parameters in a joint, unambiguous and automatically paired manner.

[0120] Figure 2 This is a visualization of the four-dimensional parameter estimation results in a multi-objective scenario according to a specific embodiment of the present invention.

[0121] In this embodiment, two scenarios were simulated using Monte Carlo independent experiments: a "large-angle separation" scenario and a "small-angle separation" scenario, both with a signal-to-noise ratio set to 0dB. "Large-angle separation" refers to two targets with significant differences in spatial angles (azimuth, pitch, DOD), used to verify the basic accuracy of the algorithm. "Small-angle separation" refers to two targets being very close in spatial angles, a highly challenging scenario used to verify the high-resolution capability of the invention.

[0122] Figure 2 In the diagram, four wireframe cubes represent the true four-dimensional parameters (azimuth, pitch, DOD, and Doppler frequency) of two targets in two different scenes, while the solid point clouds surrounding the true values ​​represent the estimation results of 500 independent Monte Carlo experiments. Darker solid point clouds represent targets with higher Doppler frequencies (i.e., high speed), while lighter solid point clouds represent targets with lower Doppler frequencies (i.e., low speed). The results show that even when the parameters of two targets are very close, the estimation results of the method described in this invention can still closely cluster around their respective true values, and the two target clusters are clearly distinguishable without confusion or misjudgment. This demonstrates that the method described in this embodiment has high accuracy and excellent resolution.

[0123] Figure 3 This is a performance curve of the composite angle error (CAE) versus signal-to-noise ratio (SNR) in a specific embodiment of the present invention. This embodiment uses statistical analysis of 500 Monte Carlo tests to reduce the impact of random noise. Figure 3 This paper presents a performance comparison of the Composite Angle Error (CAE) under two different target spacing scenarios. In this embodiment, two dual-target scenarios were simulated: a "far-separated" scenario, where the angles (azimuth, elevation, DOD) of target 1 are {90°, 110°, 35°}, and the angles of target 2 are {50°, 70°, 85°}; and a "close-separated" scenario, where the angles of target 1 are {60°, 85°, 55°}, and the angles of target 2 are {50°, 95°, 65°}. The horizontal axis in the figure represents the signal-to-noise ratio (SNR), ranging from -20 dB to 20 dB; the vertical axis represents the composite angle error. The curves in the figure show the errors of the two targets (large-angle separation of targets 1 and 2 and small-angle separation of targets 1 and 2) under these two scenarios, and are compared with the theoretical performance limit, the Cramer-Rhodes boundary (CRB). The results show that as the SNR increases, the angle error in all scenarios steadily decreases. The performance of the "far-spaced" scenario is better than that of the "close-spaced" scenario, which is consistent with the physical expectation of reduced interference between targets. Importantly, all curves gradually approach CRB at high signal-to-noise ratios, which fully demonstrates the effectiveness and asymptotic optimality of the method described in this embodiment for joint angle estimation.

[0124] Figure 4This is a performance curve of the root mean square error (RMSE) of Doppler frequency versus signal-to-noise ratio (SNR) in a specific embodiment of the present invention. In this embodiment, a low-speed target T1 (true Doppler is 300Hz) and a high-speed target T2 (true Doppler is 2400Hz) are simulated. Under the current system parameters, the unambiguous range of conventional fine estimation is only [-312.5Hz, 312.5 Hz]. Figure 4 In the figure, the horizontal axis represents the signal-to-noise ratio, and the vertical axis represents the root mean square error of Doppler. Figure 4 The results of traditional methods (Fine Estimation) using only fine Doppler information are compared with the final results of the method described in this embodiment after deblurring (Final Estimation).

[0125] The simulation results are as follows:

[0126] For the low-speed target T1, its 300 Hz Doppler frequency falls within the unambiguous range. Therefore, both the method using only fine Doppler information (hereinafter referred to as "fine estimation") and the final estimation method proposed in this invention, which combines coarse and fine information (hereinafter referred to as "final estimation"), can accurately estimate its frequency. The RMSE results of the two methods are almost identical, and both decrease significantly with the increase of signal-to-noise ratio, demonstrating excellent estimation performance.

[0127] For a high-speed target T2: its true Doppler frequency of 2400 Hz is far beyond the unambiguous range. In this case:

[0128] The “fine estimation” method fails completely due to severe Doppler blur. Its estimation result is an incorrect, constant high error value (RMSE of 100 Hz) that is unrelated to the improvement of the signal-to-noise ratio.

[0129] The method described in this embodiment perfectly solves this problem by successfully fusing coarse Doppler information to resolve ambiguity numbers. Its final estimated RMSE almost completely overlaps with the result for low-velocity target T1, and also significantly decreases to an extremely low level (below 0.01 Hz at high SNR) as the signal-to-noise ratio increases.

[0130] As the results above show, the Doppler effect of the low-speed target T1 is within the unambiguous range, and the results from both methods are consistent. However, for the high-speed target T2, its true Doppler effect far exceeds the unambiguous range, causing traditional fine estimation to completely fail and resulting in a large error. The method described in this invention successfully removes the ambiguity by combining coarse and fine information. Its final estimation error is at the same level as that of the low-speed target T1, and it decreases significantly with increasing signal-to-noise ratio. This fully verifies the effectiveness and robustness of this invention in solving the Doppler ambiguity problem.

[0131] To implement the above embodiments, a second aspect of the present invention proposes a joint estimation device for four-dimensional parameters of MIMO radar tensor decomposition, comprising:

[0132] The signal third-order tensor construction module is used to obtain the radar echo signal received by the L-type receiving array after the spatiotemporally coded transmitting array of the MIMO radar is transmitted, and reconstruct it into a third-order tensor.

[0133] The factorization module is used to perform parallel factorization on the third-order tensor to obtain the corresponding factor matrix.

[0134] The four-dimensional parameter estimation module is used to extract four-dimensional parameters of the target corresponding to the signal based on the factor matrix, including: two-dimensional direction of arrival, direction of departure, and unambiguous Doppler frequency.

[0135] It should be noted that the aforementioned explanation of an embodiment of a joint estimation method for four-dimensional parameters of MIMO radar tensor decomposition also applies to a joint estimation device for four-dimensional parameters of MIMO radar tensor decomposition in this embodiment, and will not be repeated here. According to the embodiment of the present invention, a joint estimation device for four-dimensional parameters of MIMO radar tensor decomposition acquires the radar echo signal received by an L-type receiving array after transmission by the spatiotemporally coded transmitting array of the MIMO radar, and reconstructs it into a third-order tensor; performs parallel factor decomposition on the third-order tensor to obtain the corresponding factor matrix; and extracts the four-dimensional parameters of the target corresponding to the signal based on the factor matrix, including: two-dimensional direction of arrival, direction of departure, and unambiguous Doppler frequency. This robustly achieves automatic pairing and unambiguous estimation of four-dimensional parameters of multiple targets, combining high accuracy and high computational efficiency.

[0136] To implement the above embodiments, a third aspect of the present invention provides an electronic device, comprising:

[0137] At least one processor; and a memory communicatively connected to said at least one processor;

[0138] The memory stores instructions that can be executed by the at least one processor, and the instructions are configured to perform the above-described MIMO radar tensor decomposition four-dimensional parameter joint estimation method.

[0139] To implement the above embodiments, a fourth aspect of the present invention provides a computer-readable storage medium storing computer instructions for causing the computer to execute the above-described MIMO radar tensor decomposition four-dimensional parameter joint estimation method.

[0140] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0141] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform a MIMO radar tensor decomposition four-dimensional parameter joint estimation method according to the above embodiments.

[0142] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0143] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0144] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0145] Any process or method described in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the function involved, as will be understood by those skilled in the art to which embodiments of this application pertain.

[0146] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0147] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0148] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0149] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0150] The storage medium mentioned above can be a read-only memory, a disk, or an optical disk, etc. Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application.

Claims

1. A method for MIMO radar tensor decomposition four-dimensional parameter joint estimation, characterized in that, The method comprises: acquiring radar echo signals received by an L-shaped receiving array after being transmitted by a space-time coded transmitting array of a MIMO radar, and reconstructing into a three-order tensor; performing parallel factorization on the three-order tensor to obtain corresponding factor matrices; extracting four-dimensional parameters of a target corresponding to the signals based on the factor matrices, including two-dimensional direction of arrival, wave departure direction, and unambiguous Doppler frequency.

2. The method of claim 1, wherein, The method further comprises: Let the transmitting end of the MIMO radar system be a uniform linear array containing array elements, and the receiving end be an L-shaped array; the L-shaped array is composed of two uniform linear arrays containing array elements along the x-axis and z-axis respectively, sharing an origin array element, and the receiving end contains independent array elements in total; wherein a space-time coding manner of Doppler frequency division multiple access is adopted at a transmitting end; There are K far-field targets in the memory space. All the pulse echo data received by the radar in a coherent processing interval (CPI) is arranged into a received data matrix X after down-conversion, matched filtering and sampling, and then the data matrix X is reconstructed into a third-order tensor Y. If Q = P × pulses are transmitted in a CPI, the size of the tensor Y is ( ) × × P, and the parameter P represents the number of transmission sub-periods.

3. The method of claim 2, wherein, the performing parallel factorization on the three-order tensor comprises: modeling the three-order tensor Y as a PARAFAC model with a rank K; and decomposing the tensor Y to obtain three factor matrices A, B, and C; wherein: wherein, , , are the factor vectors of the three dimensions of the corresponding tensor related to the k-th target, respectively.

4. The method of claim 3, wherein, the decomposing the tensor Y adopts an alternating least squares method.

5. The method of claim 3, wherein, The method further comprises: estimating the two-dimensional direction of arrival by using the factor matrix A. Where each column of factor matrix A It is the steering vector of the k-th target on the L-shaped receiving array; using the L-shaped receiving array, two vectors are respectively along... shaft and A geometric structure consisting of uniform linear subarrays sharing a common origin element, along with an axis, guides the vector. Split into two sub-vectors and These correspond to the responses of target k on the x-axis and z-axis subarrays, respectively. The ESPRIT algorithm is then applied to these two subvectors to estimate the eigenvalues ​​in the two directions related to the target. and Then the cosines of the two directions of the target and The following formula is used to calculate: where, is a normalized spatial angular frequency determined by array parameters, is the inter-element spacing, is the signal wavelength, direction cosine and are the spatial frequency components of the steering vector of the target k on the z-axis and the x-axis, respectively; based on and , the azimuth angle and the elevation angle of the kth target are obtained by inverse trigonometric function calculation, thereby completing the estimation of the two-dimensional direction of arrival.

6. The method of claim 3, wherein, The method further comprises: estimating the wave departure direction by using the factor matrix B. where each column of the factor matrix B is a complex factor vector related to the kth target, coupling the wave departure direction and the coarse Doppler information; solving the following least squares optimization problem: wherein is the theoretical steering vector of the transmit array; c is a normalization constant to be determined; is the amplitude of the vector is the estimated value of the wave departure direction of the kth target.​ The least square optimization problem is solved by one-dimensional peak search method to obtain the optimization result of .

7. The method of claim 3, wherein, The method further comprises: estimating the unambiguous Doppler frequency by using the factor matrices B and C, and the specific steps are as follows: 1) extracting fine Doppler information from the factor matrix C; Wherein, each column of the factor matrix C It is a steering vector corresponding to the slow-time sampling dimension; for The ESPRIT algorithm is applied to obtain the eigenvalues ​​extracted from the factor matrix C. ; fine Doppler estimate for the kth target is calculated by the equation: wherein is the pulse repetition interval; 2) extracting coarse Doppler information from the factor matrix B; where each column of the factor matrix B contains phase information that simultaneously contains both the wave departure direction and Doppler information; based on a theoretical compensation vector is constructed that is related to the wave departure direction : wherein is a transmit array steering vector whose elements represent the spatial phase difference determined by the target wave departure direction (DOD) ​ W is a x transmit code matrix whose elements contain preset time phase modulation information for distinguishing different transmit antennas; :​​​ wherein represents element-wise division; on the vector Applying the ESPRIT algorithm, the eigenvalues extracted from the factor matrix B are obtained The coarse Doppler estimate of the kth target is is calculated by 3) determining an ambiguity number by comparing the results of steps 1) and 2), so as to obtain an estimation result of the unambiguous Doppler frequency; wherein the ambiguity number of the kth target is calculated according to the following expression: finally, the unambiguous and high-resolution Doppler estimation of the kth target is: 。 8.A device for joint estimation of four-dimensional parameters by MIMO radar tensor decomposition, characterized in that, The method comprises: a signal three-order tensor construction module, configured to acquire radar echo signals received by an L-shaped receiving array after being transmitted by a space-time coded transmitting array of a MIMO radar, and reconstructing into a three-order tensor; a factorization module, configured to perform parallel factorization on the three-order tensor to obtain corresponding factor matrices; a four-dimensional parameter estimation module, configured to extract four-dimensional parameters of a target corresponding to the signals based on the factor matrices, including two-dimensional direction of arrival, wave departure direction, and unambiguous Doppler frequency.

9. An electronic device, comprising: The method comprises: at least one processor; and a memory connected in communication with the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are configured to execute the method in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer readable storage medium stores computer instructions for causing the computer to execute the method in any one of claims 1-7.

Images