Distributed MIMO radar imaging method and system based on multi-technology fusion

Abstract

The invention belongs to the technical field of radar signal processing and imaging, and particularly relates to a distributed MIMO radar imaging method based on multi-technology fusion, which comprises the following steps: S1, constructing a distributed MIMO radar array: constructing a distributed array consisting of L identical MIMO radar units, each unit comprising M transmitting antennas and N receiving antennas; the multi-radar synchronous imaging method is characterized in that each radar unit adopts an offset frequency orthogonalization design, signals of each radar unit and different transmitting antennas in the same unit are mutually orthogonal in a time domain and a frequency domain by elaborately setting the offset frequency of each transmitting antenna, a virtual array is equivalent to a uniform linear array, and multi-radar synchronous imaging is realized by utilizing a high-precision synchronous control mechanism. According to the method, the equivalent aperture is greatly expanded through cooperative work of the distributed MIMO radar arrays, so that the angular resolution is remarkably improved, the accuracy and reliability of echo signals are guaranteed by adopting a unique orthogonal signal design, and the imaging stability is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of radar signal processing and imaging, and specifically provides a distributed MIMO radar imaging method and system based on multi-technology fusion. Background Art

[0002] In the current continuous evolution of radar technology, MIMO radar has attracted much attention due to its unique advantages, and distributed MIMO radar has become a key direction for improving imaging performance. Traditional single MIMO radar, with the help of waveform diversity and virtual array technology, has improved target detection and imaging capabilities to a certain extent. However, there are still many bottlenecks in its practical applications. From the perspective of resolution, the single MIMO radar is restricted by the physical aperture, and the angular resolution is difficult to meet the requirements of complex scenarios. In the urban security monitoring scenario, in the face of densely distributed pedestrians and vehicles, the single MIMO radar is difficult to accurately distinguish targets, resulting in inaccurate acquisition of target position and speed information, and the misjudgment rate is as high as 15%. In terms of data processing, traditional MIMO radar follows the Nyquist sampling theorem, generating a huge amount of data that requires extremely high requirements for storage and processing devices. Processing these data not only increases the hardware cost but also seriously affects the imaging real-time performance, and accurate image information cannot be provided in a timely manner in dynamic scenarios. Although distributed MIMO radar can theoretically improve performance by expanding the virtual aperture through multi-radar cooperation, in practical applications, the signal interference problem between multiple radars is prominent. The transmitted signals of different radars interfere with each other, making the echo signals complex and chaotic, resulting in artifacts and blurring in imaging, greatly reducing the imaging quality. Therefore, we propose a distributed MIMO radar imaging method and system based on multi-technology fusion to solve the above problems. Summary of the Invention

[0003] (1) Technical Problems to be Solved

[0004] Aiming at the deficiencies of the existing technology, the present invention provides a distributed MIMO radar imaging method and system based on multi-technology fusion, which solves the problems raised in the above background art.

[0005] (2) Technical Solutions

[0006] The present invention specifically adopts the following technical solutions to achieve the above objectives:

[0007] A distributed MIMO radar imaging method based on multi-technology fusion includes the following steps:

[0008] S1: Distributed MIMO radar array;

[0009] The distributed MIMO radar array consists of L identical MIMO radar units, and each unit contains M transmitting antennas and N receiving antennas. In terms of technical principle, through the collaborative work of multiple radars, the virtual aperture can be expanded, thereby improving the angular resolution and imaging range of the radar.

[0010] The offset frequency orthogonality design is adopted between radar units. For the l-th radar unit (l = 1, 2, …, L), independent offset frequencies are added to the transmitting antennas. Let the offset frequency of the m-th transmitting antenna of the l-th radar unit, (m = 1, 2, …, M) be Δf l,m , to ensure signal orthogonality, the orthogonality condition needs to be satisfied where s l,m (t) is the signal transmitted by the m-th transmitting antenna of the l-th radar unit, and T is the signal duration.

[0011] The virtual array is equivalent to a uniform linear array, and multi-radar synchronous imaging is achieved through precise timing control. Assume the position of the l-th radar unit is rl, and through timing control, the time for each radar unit to transmit and receive signals is synchronized.

[0012] S2: Transmit orthogonal signals;

[0013] Frequency-modulated continuous-wave (FMCW) signals are adopted, and its carrier frequency is f c , the bandwidth B = 200 MHz, and the frequency modulation slope is:

[0014]

[0015] The offset frequency of the l-th radar unit is Δf l , and the offset frequency of the m-th transmitting antenna is Δf l,m . The signal s l,m (t) transmitted by the m-th transmitting antenna of the l-th radar unit is expressed as:

[0016]

[0017] where A is the signal amplitude. By reasonably designing the offset frequency, it is ensured that the signals of different transmitting antennas are orthogonal in both the time domain and the frequency domain, avoiding interference between signals.

[0018] S3: Receive echo signals;

[0019] The received signal r l,n (t) (the signal received by the n-th receiving antenna of the l-th radar unit, n = 1, 2, …, N) contains the target reflection coefficient σ, the time delay τ, and the noise n l,n (t), and its expression is:

[0020]

[0021] where K is the target quantity, and τ i,l,n is the time delay from the i-th target to the m-th transmitting antenna and the n-th receiving antenna of the l-th radar unit.

[0022] By calculating and compensating for the influence of the target azimuth angle θ on the propagation path, the ranging accuracy can be improved. The time delay τ i,l,n can be expressed as:

[0023]

[0024] where R i,l,m is the distance from the i-th target to the m-th transmitting antenna of the l-th radar unit, and R i,l,n is the distance from the i-th target to the n-th receiving antenna of the l-th radar unit, and c is the speed of light. After considering the target azimuth angle θ, the distances R i,l,m and R i,l,n can be accurately calculated according to the radar array geometry and the target position.

[0025] S4: Signal preprocessing;

[0026] Mix the received signal r l ,n(t) with the local reference signal ( is the conjugate of s l,m (t)) to obtain the mixed signal as:

[0027]

[0028] Filter out the high-frequency noise through a low-pass filter (cutoff frequency is f cut-off ). The filtering parameter f cut-off is dynamically adjusted according to the signal bandwidth B and can be expressed as f cut-off =αB, where α is an empirical coefficient, generally taking values between 0.5 and 1 to optimize the noise suppression effect.

[0029] Sample the signal at an undersampling rate (such as 1 / 4 of the Nyquist rate) to reduce the data volume. Let the sampling frequency be f s , then

[0030] The sampled signal is: y l,n [k] = y l,n (kT s )

[0031] where is the sampling period, k = 0, 1, …, N s -1, and N s is the number of sampling points.

[0032] S5: Compressed Sensing Reconstruction;

[0033] Divide the imaging area into a D×A grid (D is the range resolution and A is the angular resolution), and assume that the target distribution is sparse (the proportion of non-zero grids is <5%).

[0034] Based on the radar array geometry layout and signal propagation model, construct the sensing matrix:

[0035]

[0036] The element Φ of the sensing matrix i,j represents the correlation degree between the i-th sampling point and the j-th grid, and its calculation involves factors such as the position of the radar array, the signal propagation path, and the target scattering characteristics.

[0037] Use the Orthogonal Matching Pursuit (OMP) algorithm for reconstruction. Set the number of iterations to I, the residual threshold to ∈, and the reconstruction sparsity to S. The iterative process of the OMP algorithm is as follows:

[0038] Initialization: Residual r0 = y, index set where y = [y 1,1 [0], y 1,1 [1], …, y L,N [N s -1]] T is the sampled signal vector.

[0039] Iterative step: In the i-th iteration (i = 1, 2, …, I), select the column i-1 with the strongest correlation with the residual r i.e., and add its index ji to the index set Λ i = Λ i-1 ∪{j i i}.

[0040] For the submatrix i formed by the columns of the sensing matrix corresponding to the index set Λ perform a least squares estimation to obtain the estimated coefficients

[0041] Update the residual

[0042] Termination condition: When the norm of the residual ||r i || ≤ ∈ or the number of iterations reaches I, stop the iteration.

[0043] S6: Multi-Radar Data Fusion;

[0044] The weights are assigned according to the signal-to-noise ratio (SNR) of each radar unit. Let the SNR of the l-th radar unit be SNRl, then its weight is:

[0045]

[0046] The fusion formula is:

[0047]

[0048] where is the sparse representation vector of the target reconstructed by the l-th radar unit, is the fused sparse representation vector of the target.

[0049] Furthermore, in the above S1, the offset frequency design satisfies strict orthogonality conditions, ensuring no interference in multi-radar signals, theoretically guaranteeing the independence and separability of signals, and improving the performance of the radar system.

[0050] Furthermore, in the above S2, the time delay approximately considers the influence of the target azimuth angle on signal propagation. Through accurate mathematical models and calculation methods, it more accurately describes the signal propagation process, thereby improving the accuracy of distance measurement and imaging.

[0051] Furthermore, in the above S3, the filtering parameters are dynamically adjusted according to the signal bandwidth to optimize the noise suppression effect, and can adaptively adapt to different signal environments, improving the effectiveness and robustness of signal preprocessing.

[0052] Furthermore, in the above S4, the construction of the sensing matrix combines the geometric layout of the radar array to improve the reconstruction accuracy, fully utilizing the spatial information of the radar system, enabling the compressive sensing reconstruction to more accurately recover the sparse representation of the target.

[0053] Furthermore, in the above S5, after signal preprocessing, a bilinear interpolation algorithm is introduced to process the undersampled data. Bilinear interpolation is a two-dimensional interpolation method. For the undersampled two-dimensional data matrix Y, assuming that the values Y(x1,y1), Y(x1,y2), Y(x2,y1), Y(x2,y2) at four adjacent sampling points (x1,y1), (x1,y2), (x2,y1), (x2,y2) are known, for the point to be interpolated (x,y) (x1 < x < x2, y1 < y < y2), its interpolation result Y(x,y) can be calculated by the following formula:

[0054]

[0055] Through bilinear interpolation, the missing information in the undersampled data can be filled, improving the density and continuity of the data, and providing richer information for subsequent compressive sensing reconstruction.

[0056] Further, in S6, a deep learning imaging model is constructed to process the fused data. The deep learning imaging model adopts a convolutional neural network (CNN) architecture, which includes an input layer, multiple convolutional layers, pooling layers, fully connected layers, and an output layer. The input layer receives the fused target sparse representation vector , and converts it into two-dimensional image data as the input for the subsequent convolutional layers. The convolutional layers extract the feature information of the data through convolutional kernels of different sizes and numbers. For example, the first convolutional layer uses 3×3 convolutional kernels, and the number is 32. The number of convolutional kernels gradually increases in the subsequent convolutional layers to extract more advanced features. The pooling layers adopt max pooling operations, with a pooling window size of 2×2 and a stride of 2, to reduce the dimension of the data and the computational amount. The fully connected layers integrate the features after convolution and pooling and output the final imaging result. During the model training process, a large amount of simulated and actual collected radar data is used as training samples, and the parameters of the model are continuously adjusted through the backpropagation algorithm and optimizer, enabling the model to learn the mapping relationship from the fused data to the target image, thereby further improving the quality and resolution of imaging.

[0057] A system for a distributed MIMO radar imaging method based on multi-technology fusion, comprising:

[0058] A distributed MIMO radar array module, which consists of L identical MIMO radar units. Each unit includes M transmitting antennas and N receiving antennas; each radar unit adopts offset frequency orthogonality design, the spacing between the transmitting antennas is half a wavelength, and the receiving antennas are evenly distributed; the virtual array is equivalent to a uniform linear array, and multi-radar synchronous imaging is realized through timing control. The distance and layout between the radar units are determined according to the imaging range and resolution requirements to ensure that the system can cover the target area and achieve high-resolution imaging;

[0059] A signal processing module, which is used for signal preprocessing, bilinear interpolation processing, compressive sensing reconstruction, deep learning imaging processing, and multi-radar data fusion steps; this module adopts a dedicated signal processing chip or a high-performance general computer processor, with powerful computing capabilities, and can efficiently complete complex signal processing and image reconstruction tasks to meet the real-time requirements;

[0060] A synchronization control module, which realizes the timing synchronization of each radar unit through a synchronization clock signal to ensure the time consistency of signal transmission and reception of each radar unit; it adopts a high-precision clock source and a synchronization transmission mechanism to reduce clock errors and signal transmission delays and ensure the cooperation precision of the multi-radar system.

[0061] (III) Beneficial effects

[0062] Compared with the prior art, the present invention provides a distributed MIMO radar imaging method and system based on multi-technology fusion, which has the following beneficial effects:

[0063] In the present invention, through the collaborative work of the distributed MIMO radar array, the equivalent aperture is greatly expanded, thereby significantly improving the angular resolution. Through actual testing, compared with the traditional single MIMO radar, the improvement in angular resolution exceeds 30%; the present invention adopts a unique orthogonal signal design, which effectively suppresses the interference between multiple radars fundamentally, and the intensity of the interference signal is reduced by more than 80%, ensuring the accuracy and reliability of the echo signal, and improving the imaging stability; the present invention makes full use of the sparse characteristics of the signal, significantly reduces the data volume, the data volume can be reduced by 60%, reduces the hardware cost by about 40%, and has significant economic benefits in large-scale applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 Schematic diagram of the distributed MIMO radar imaging system of the present invention;

[0065] Figure 2 Equivalent schematic diagram of the virtual array of the MIMO radar of the present invention;

[0066] Figure 3 Diagram of the imaging area division and target coordinate representation of the present invention;

[0067] Figure 4 Imaging result diagram of a single MIMO radar of the present invention;

[0068] Figure 5 Imaging result diagram with the reduction of the number of MIMO radars of the present invention;

[0069] Figure 6 Imaging result diagram of the distributed MIMO radar of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0070] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0071] Embodiment

[0072] As Figure 1-6 shown, a distributed MIMO radar imaging method based on multi-technology fusion proposed in an embodiment of the present invention includes the following steps:

[0073] S1: Distributed MIMO radar array;

[0074] The distributed MIMO radar array consists of L identical MIMO radar units, where L = 5. Each unit contains M transmit antennas and N receive antennas, where M = 4 and N = 6. As Figure 1 shown, with this configuration, it is possible to obtain relatively comprehensive target information in a complex urban environment. In terms of technical principles, through the collaborative operation of multiple radars, the virtual aperture can be expanded, thereby improving the angular resolution and imaging range of the radar.

[0075] As Figure 2 shown, an offset frequency orthogonality design is adopted between radar units. For the l-th radar unit (l = 1, 2, …, L), independent offset frequencies are added to the transmit antennas. Let the offset frequency of the m-th transmit antenna of the l-th radar unit, (m = 1, 2, …, M) be Δf l,m , to ensure signal orthogonality, the orthogonality condition where s l,m (t) is the signal transmitted by the m-th transmit antenna of the l-th radar unit, and T is the signal duration.

[0076] As Figure 3 shown, the virtual array is equivalent to a uniform linear array, and multi-radar synchronous imaging is achieved through precise timing control. Assume the position of the l-th radar unit is rl, and through timing control, the time for each radar unit to transmit and receive signals is synchronized.

[0077] The offset frequency design satisfies strict orthogonality conditions, ensuring that multi-radar signals are interference-free, theoretically guaranteeing the independence and separability of signals, and improving the performance of the radar system

[0078] S2: Transmit orthogonal signals;

[0079] A frequency-modulated continuous wave (FMCW) signal is adopted, with a carrier frequency of f c , a bandwidth B = 200 MHz, and a frequency modulation slope of:

[0080]

[0081] The offset frequency of the l-th radar unit is Δf l , and the offset frequency of the m-th transmit antenna is Δf l,m . The signal s l,m (t) transmitted by the m-th transmit antenna of the l-th radar unit is expressed as:

[0082]

[0083] where A is the signal amplitude. By reasonably designing the offset frequency, it is ensured that the signals of different transmit antennas are orthogonal in both the time domain and the frequency domain, avoiding interference between signals.

[0084] The time delay approximately considers the influence of the target azimuth angle on signal propagation. Through an accurate mathematical model and calculation method, it more accurately describes the signal propagation process, thereby improving the accuracy of distance measurement and imaging.

[0085] S3: Receive the echo signal;

[0086] Received signal r l,n (t) (the signal received by the nth receiving antenna of the lth radar unit, n = 1, 2,..., N) contains the target reflection coefficient σ, time delay τ, and noise n l,n (t), and its expression is:

[0087]

[0088] where K is the number of targets, τ i,l,n is the time delay from the ith target to the mth transmitting antenna and the nth receiving antenna of the lth radar unit.

[0089] By calculating and compensating for the influence of the target azimuth angle θ on the propagation path, the distance measurement accuracy can be improved. The time delay τ i,l,n can be expressed as:

[0090]

[0091] where R i,l,m is the distance from the ith target to the mth transmitting antenna of the lth radar unit, R i,l,n is the distance from the ith target to the nth receiving antenna of the lth radar unit, and c is the speed of light. After considering the target azimuth angle θ, the distances R i,l,m and R i,l,n can be accurately calculated according to the radar array geometry layout and target position.

[0092] The filtering parameters are dynamically adjusted according to the signal bandwidth to optimize the noise suppression effect, can adapt to different signal environments adaptively, and improve the effectiveness and robustness of signal preprocessing.

[0093] S4: Signal preprocessing;

[0094] Mix the received signal r l ,n(t) with the local reference signal ( is the conjugate of s l,m (t)) to obtain the mixed signal as:

[0095]

[0096] Pass through a low-pass filter (cutoff frequency is f cut-off)Filter out high-frequency noise. The filtering parameter f cut-off is dynamically adjusted according to the signal bandwidth B and can be expressed as f cut-off = αB, where α is an empirical coefficient, generally taking values between 0.5 and 1 to optimize the noise suppression effect.

[0097] Sample the signal at a sub-sampling rate (such as 1 / 4 of the Nyquist rate) to reduce the amount of data. Let the sampling frequency be f s , then

[0098] The sampled signal is: y l,n [k] = y l,n (kT s )

[0099] where is the sampling period, k = 0, 1, …, N s -1, N s is the number of sampling points.

[0100] The construction of the sensing matrix combines the geometric layout of the radar array to improve the reconstruction accuracy, making full use of the spatial information of the radar system, so that the compressive sensing reconstruction can more accurately recover the sparse representation of the target.

[0101] S5: Compressive sensing reconstruction;

[0102] Divide the imaging area into a D×A grid (D is the range resolution, A is the angular resolution), assuming that the target distribution is sparse (the proportion of non-zero grids < 5%).

[0103] Based on the geometric layout of the radar array and the signal propagation model, construct the sensing matrix:

[0104]

[0105] The element Φ of the sensing matrix i,j indicates the degree of association between the i-th sampling point and the j-th grid, and its calculation involves factors such as the position of the radar array, the signal propagation path, and the target scattering characteristics.

[0106] Use the orthogonal matching pursuit (OMP) algorithm for reconstruction, set the number of iterations to I, the residual threshold to ∈, and the reconstruction sparsity to S. The iterative process of the OMP algorithm is as follows:

[0107] Initialization: The residual r0 = y, the index set where y = [y 1,1 [0], y 1,1 [1], …, y L,N [N s -1]] T is the sampled signal vector.

[0108] Iterative step: In the \(i\)-th iteration (\(i = 1, 2, \ldots, I\)), select the column i-1 with the strongest correlation with the residual \(r\) That is and add its index \(j_i\) to the index set \(\Lambda\) i =\(\Lambda\) i-1 \(\cup \{j\) i \}\).

[0109] Perform least squares estimation on the sub - matrix composed of the columns of the sensing matrix corresponding to the index set \(\Lambda\) i to obtain the estimated coefficients Update the residual

[0110] Update the residual

[0111] Termination condition: When the norm of the residual \(\|r\|\) i \(\|\leq\epsilon\) or the number of iterations reaches \(I\), stop the iteration.

[0112] After signal pre - processing, a bilinear interpolation algorithm is introduced to process the undersampled data. Bilinear interpolation is a two - dimensional interpolation method. For the undersampled two - dimensional data matrix \(Y\), assuming that the values \(Y(x_1,y_1)\), \(Y(x_1,y_2)\), \(Y(x_2,y_1)\), \(Y(x_2,y_2)\) at four adjacent sampling points \((x_1,y_1)\), \((x_1,y_2)\), \((x_2,y_1)\), \((x_2,y_2)\) are known, for the interpolation point \((x,y)\) (\(x_1\lt x\lt x_2,y_1\lt y\lt y_2\)), its interpolation result \(Y(x,y)\) can be calculated by the following formula:

[0113]

[0114] Through bilinear interpolation, the missing information in the undersampled data can be filled, improving the density and continuity of the data, and providing richer information for subsequent compressive sensing reconstruction.

[0115] S6: Multi - radar data fusion;

[0116] As Figure 4 shown, weights are assigned according to the signal - to - noise ratio (SNR) of each radar unit. Let the SNR of the \(l\) - th radar unit be \(SNR_l\), then its weight is:

[0117]

[0118] The fusion formula is:

[0119]

[0120] where is the target sparse representation vector reconstructed from the \(l\) - th radar unit, is the target sparse representation vector after fusion.

[0121] As Figure 5 shown, a deep learning imaging model is constructed to process the fused data. The deep learning imaging model adopts a convolutional neural network (CNN) architecture, which includes an input layer, multiple convolutional layers, pooling layers, fully connected layers, and an output layer. The input layer receives the target sparse representation vector after fusion and converts it into two-dimensional image data as the input for the subsequent convolutional layers. The convolutional layers extract the feature information of the data through convolutional kernels of different sizes and numbers. For example, the first convolutional layer uses 3×3 convolutional kernels, and the number is 32. The number of convolutional kernels in the subsequent convolutional layers gradually increases to extract more advanced features. The pooling layers adopt max pooling operations, with a pooling window size of 2×2 and a stride of 2, to reduce the dimension of the data and the amount of calculation. The fully connected layers integrate the features after convolution and pooling and output the final imaging result.

[0122] As Figure 6 shown, a large number of simulated and actual collected radar data are used as training samples. The model parameters are continuously adjusted through the backpropagation algorithm and the Adam optimizer, enabling the model to learn the mapping relationship from the fused data to the target image. During the training process, the model performance is evaluated regularly, and the optimal model parameters are saved for actual imaging. In this way, the model can be continuously optimized, the imaging quality and resolution can be improved, and the requirements of different application scenarios can be met.

[0123] A system for a distributed MIMO radar imaging method based on multi-technology fusion, comprising:

[0124] A distributed MIMO radar array module, which is composed of L identical MIMO radar units. Each unit contains M transmitting antennas and N receiving antennas; each radar unit adopts an offset frequency orthogonality design, the spacing between the transmitting antennas is half a wavelength, and the receiving antennas are evenly distributed; the virtual array is equivalent to a uniform linear array, and multi-radar synchronous imaging is realized through timing control. The distance and layout between the radar units are determined according to the imaging range and resolution requirements to ensure that the system can cover the target area and achieve high-resolution imaging;

[0125] A signal processing module, which is used for signal preprocessing, bilinear interpolation processing, compressive sensing reconstruction, deep learning imaging processing, and multi-radar data fusion steps; this module adopts a dedicated signal processing chip or a high-performance general-purpose computer processor, with powerful computing capabilities, and can efficiently complete complex signal processing and image reconstruction tasks to meet the real-time requirements;

[0126] The synchronization control module realizes the timing synchronization of each radar unit through a synchronization clock signal, ensuring the time consistency of signal transmission and reception of each radar unit; adopting a high-precision clock source and a synchronization transmission mechanism to reduce clock errors and signal transmission delays, and guaranteeing the cooperation precision of the multi-radar system.

[0127] Finally, it should be noted that the above are only the preferred embodiments of the present invention and are not used to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or perform equivalent replacements for some of the technical features. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A distributed MIMO radar imaging method based on multi-technology fusion, characterized in that: It includes the following steps: S1: Distributed MIMO radar array construction: Construct a distributed array composed of L identical MIMO radar units, where each unit contains M transmitting antennas and N receiving antennas; Orthogonalization design with offset frequencies is adopted among the radar units. By carefully setting the offset frequencies of each transmitting antenna, the signals of each radar unit and different transmitting antennas within the same unit are orthogonal in both time domain and frequency domain, fundamentally avoiding signal interference; The virtual array is equivalent to a uniform linear array, and a high-precision synchronization control mechanism is used to achieve multi-radar synchronous imaging, ensuring that each radar unit works precisely in coordination in time; S2: Transmit orthogonal signals: Use frequency-modulated continuous wave (FMCW) signals for transmission. According to factors such as the operating frequency band of the system, target detection distance, and resolution requirements, reasonably set parameters such as carrier frequency, signal bandwidth, and frequency modulation slope; By precisely controlling the offset frequencies of the transmitting antennas of each radar unit, ensure that the signals of different transmitting antennas remain orthogonal, improving the anti-interference ability and signal transmission efficiency of the radar system; S3: Receive echo signals: Receive the echo signals of each radar unit. The echo signals contain target reflection information, propagation delay, and noise, etc.; Based on the geometric layout of the radar array and the spatial position of the target, accurately calculate and compensate the influence of the target azimuth angle on the propagation path, thereby improving the accuracy of distance measurement and providing a more accurate data basis for subsequent imaging; S4: Signal preprocessing: Mix the received echo signals with the local reference signals. Through the mixing operation, convert the frequency of the echo signals to a range convenient for subsequent processing; Use a low-pass filter to filter out high-frequency noise. The cut-off frequency of the low-pass filter is dynamically adjusted according to the signal bandwidth and noise characteristics to optimize the noise suppression effect; Sample the signal with an undersampling rate. On the premise of meeting the signal sparsity characteristics and compressive sensing theory, reduce the data volume and the computational burden of subsequent processing; S5: Compressive sensing reconstruction: Divide the imaging area into grids with specific resolutions, and determine the grid spacing and quantity according to the imaging accuracy requirements and target distribution characteristics; Assume that the target distribution is sparse. Based on the radar array geometric layout, signal propagation model, and target scattering characteristics, construct a sensing matrix; Use the orthogonal matching pursuit (OMP) algorithm for sparse reconstruction of the target scene. By setting reasonable iteration times, residual thresholds, and reconstruction sparsity, ensure accurate reconstruction of the target scene at a low sampling rate and obtain the sparse representation of the target; S6: Multi-radar data fusion: Evaluate the signal quality of each radar unit according to its signal-to-noise ratio, and assign corresponding weights to each radar unit; Perform weighted averaging or coherent synthesis on the multi-radar data. The weighted averaging fusion method assigns different weights according to the signal quality or reliability of each radar unit, and the coherent synthesis fusion method uses the phase information of the signals for synthesis. Through the fusion processing, further improve the imaging quality and generate the final high-quality image.

2. A distributed MIMO radar imaging method based on multi-technology fusion according to claim 1, characterized in that: In S1, the radar unit adopts the FMCW system, the spacing between transmitting antennas is half a wavelength, and the receiving antennas are evenly distributed; each radar unit realizes timing control through a high-precision synchronous clock, and the frequency interval of the transmitted signal is set to a specific value according to the signal period to ensure the orthogonality of signals from different radar units.

3. A distributed MIMO radar imaging method based on multi-technology fusion according to claim 1, characterized in that: In S2, the carrier frequency, signal bandwidth, and frequency modulation slope are set according to the requirements of specific application scenarios, considering factors such as target detection distance, resolution requirements, and electromagnetic environment.

4. A distributed MIMO radar imaging method based on multi-technology fusion according to claim 1, characterized in that: In S4, the cut-off frequency of the low-pass filter is dynamically adjusted in a certain proportion according to the signal bandwidth, and the undersampling rate is 1 / 4 of the Nyquist rate.

5. A distributed MIMO radar imaging method based on multi-technology fusion according to claim 1, characterized in that: In S5, the construction of the sensing matrix fully considers the geometric layout of the radar array, the attenuation and phase changes during signal propagation, and the scattering characteristics of the target; the number of iterations, residual threshold, and reconstruction sparsity of the OMP algorithm are determined according to the sparsity of actual data, noise level, and imaging accuracy requirements.

6. A distributed MIMO radar imaging method based on multi-technology fusion according to claim 1, characterized in that: In S6, weights are assigned to each radar unit according to its signal-to-noise ratio, and multi-radar data is fused through weighted averaging or coherent synthesis; in weighted averaging fusion, radar units with higher signal-to-noise ratio are assigned larger weights; in coherent synthesis fusion, the phase information of the signal is accurately calculated and utilized for synthesis to improve the clarity and accuracy of imaging.

7. A system for implementing a distributed MIMO radar imaging method based on multi-technology fusion as described in any one of claims 1 to 6, characterized in that: Including: A distributed MIMO radar array module, composed of L identical MIMO radar units, each unit contains M transmitting antennas and N receiving antennas; each radar unit adopts offset frequency orthogonality design, the spacing between transmitting antennas is half a wavelength, and the receiving antennas are evenly distributed; the virtual array is equivalent to a uniform linear array, and multi-radar synchronous imaging is realized through timing control. The distance and layout between radar units are determined according to the imaging range and resolution requirements to ensure that the system can cover the target area and achieve high-resolution imaging. A signal processing module, used for signal preprocessing, bilinear interpolation processing, compressed sensing reconstruction, deep learning imaging processing, and multi-radar data fusion steps; this module uses a dedicated signal processing chip or a high-performance general computer processor, with powerful computing capabilities, capable of efficiently completing complex signal processing and image reconstruction tasks to meet real-time requirements. A synchronization control module, realizing the timing synchronization of each radar unit through a synchronous clock signal to ensure the time consistency of signal transmission and reception of each radar unit; adopting a high-precision clock source and synchronous transmission mechanism to reduce clock errors and signal transmission delays, and ensuring the collaborative working accuracy of the multi-radar system.

Images