Robust Adaptive Beamforming Method for Echo Separation in Airborne MIMO SAR

Abstract

The application discloses a kind of robust adaptive beamforming methods for airborne MIMO SAR echo separation, it includes the following steps: 1) using multiple transmitter to emit short time shift orthogonal waveform, and utilize pitch multi-channel system simultaneously receives multiple transmission waveform echo;2) the direction of each signal component is divided into area, and according to pitch dimension space snapshot estimation sample covariance matrix, calculate Capon power spectrum;3) using Capon power spectrum to reconstruct interference noise covariance matrix (Interference Pulse Noise Covariance Matrix, IPNCM), eliminate the expected signal component in covariance matrix;4) the covariance matrix of expected signal is reconstructed, and then the guide vector of the signal is estimated;5) according to IPNCM and expected signal guide vector calculation optimal weight vector, realize echo separation.This method can obtain higher output signal-to-noise ratio, and in low signal-to-noise ratio and guide vector mismatch condition, still can realize echo separation in MIMO SAR system, with good robustness and wide application prospect.

Description

Technical Field

[0001] This invention relates to a robust adaptive beamforming method for echo separation in airborne MIMO SAR. The main idea of ​​this technology is to suppress ambiguity energy in the pitch domain due to the different incident angles of various signal components. It belongs to the fields of airborne MIMO SAR imaging technology and array signal processing technology. Background Technology

[0002] In recent years, MIMO technology has attracted widespread attention in the radar field. Originally applied in communications to overcome channel fading caused by multipath propagation and thus improve channel capacity, MIMO technology in radar, by transmitting multiple sets of orthogonal waveforms at the transmitter and receiving them with multiple antennas at the receiver, effectively increases the number of observation channels using virtual array technology. This enhances the system's degrees of freedom and significantly improves radar detection capabilities. Considering the numerous advantages of MIMO technology, Ender pioneered its application in SAR, overcoming the technical limitations of traditional SAR systems and effectively solving many challenges faced by traditional systems, such as high-resolution wide-swath (HRWS) imaging, ground moving target indication (GMTI), and SAR imaging in complex electromagnetic environments.

[0003] In MIMO SAR, each receiver collects echo signals from all transmitted waveforms. To fully leverage the advantages of MIMO SAR, it is necessary to separate the echoes of each transmitted waveform. Since perfectly orthogonal waveforms at the same frequency do not exist, coupling effects between transmitted waveforms are inevitable. The ambiguity energy caused by waveform mutual coupling accumulates in the imaging scene, significantly increasing the background noise level of the SAR image and severely affecting image quality. Suppressing the ambiguity energy caused by non-perfectly orthogonal waveforms and achieving separation of the echo signals from each transmitted waveform is a major challenge in the field of MIMO SAR.

[0004] Generally, waveform separation mechanisms in the MIMO SAR field mainly include the following: 1) Time-Division Multiplexing (TDM) waveforms; 2) Frequency-Division Multiplexing (FDM) waveforms; and 3) Code-Division Multiplexing (CDM) waveforms. Due to the time division, TDM waveforms significantly increase the total pulse width and reduce the system's Pulse Repetition Frequency (PRF). To avoid Doppler blurring, TDM waveforms require a sufficiently high equivalent PRF, leading to a reduction in the mapping bandwidth. FDM waveforms reduce the correlation between waveforms, thus affecting the coherent processing of subsequent transmitted waveform echo data. Furthermore, the design of FDM waveforms greatly increases system complexity and is not suitable for airborne MIMO SAR systems. In the CDM mechanism, multidimensional coded waveforms, such as Space-Time Coding (STC) waveforms, use Alamouti codes to encode the transmitter and decode the receiver. While this solves the above problems, STC waveforms require a high PRF to achieve echo signal separation. In recent years, the STSO waveform proposed by Krieger has gained widespread popularity. This waveform maintains orthogonality within a certain time delay. However, the cross-correlation energy of the STSO waveform beyond the time delay is not zero, requiring pitch-dimensional DBF (Depth-of-Flight) techniques for suppression. STSO waveforms offer advantages such as simple system implementation and good waveform orthogonality. When using STSO waveforms for wide-span mapping imaging, the processing performance of pitch-dimensional DBF techniques becomes particularly important.

[0005] Currently, some scholars have proposed a DBF (Broadcast Frame Formation) technique based on the Least Square (LS) criterion. This technique calculates the theoretical weight vector to achieve echo separation in the STSO (Site-to-Site Array) system, and is simple to implement with low computational complexity. Similarly, based on the known Direction of Arrival (DOA) of the desired and interfering signals, a beamformer based on the Linear Constrained Minimum Variance (LCMV) criterion can be used to suppress interfering signals in the elevation domain. Furthermore, considering the influence of various factors in the echo (such as low signal-to-noise ratio (SNR)), imposing excessive constraints on the beamformer may affect the performance of DBF processing. Therefore, the weight vector in DBF processing can be obtained through Sample Covariance Inversion (SMI). However, the above DBF techniques based on deterministic information may lead to the loss of important information in practical processing. For example, when the imaging scene has topographical undulations, using deterministic information for DBF processing can lead to errors in the pointing of the main lobe and nulls, resulting in problems such as decreased desired signal gain and insufficient suppression of interference signals, and even signal 'self-cancellation', as seen in LCMV and SMI beamformers. To overcome the DOA mismatch problem, existing literature has proposed applying first-order derivative constraints to the pointing of the main beam and interference beams to broaden the main beam and nulls, thereby improving the robustness of DBF. However, DBF techniques based on first-order derivative constraints require a large number of system degrees of freedom. Due to the limited pitch dimension of airborne MIMO SAR, this technique is not suitable for STSO waveform separation based on airborne platforms. In addition, existing literature introduces a robust DBF method based on Eigenspace-Based ESB, which uses MUSIC to determine the accurate DOA of various components, and then performs DBF processing on the echo signal based on the LCMV criterion. On the one hand, the method does not remove the desired signal component when estimating the covariance matrix, which leads to a decrease in the output SINR of the beamformer; on the other hand, the DOA estimation performance deteriorates when the SNR is low, and the high sidelobe gain that the beamformer may produce can lead to a decrease in image quality. Summary of the Invention

[0006] Under the influence of various factors, the DBF performance in airborne MIMO SAR echo separation degrades, thereby affecting the final echo separation effect and leading to image quality degradation. To address this issue, this invention proposes a robust adaptive beamforming method for airborne MIMO SAR echo separation.

[0007] This invention discloses a robust adaptive beamforming method for echo separation in airborne MIMO SAR, comprising the following steps:

[0008] 1) Use multiple transmitters to transmit STSO waveforms and utilize a MIMO SAR elevation multi-channel system to simultaneously receive echoes of multiple transmitted waveforms;

[0009] 2) Divide the regions of origin for each signal component, estimate the sample covariance matrix based on the pitch dimension spatial snapshot, and calculate the Capon power spectrum;

[0010] 3) Reconstruct the IPNCM using Capon power spectrum and remove the desired signal component from the covariance matrix;

[0011] 4) Reconstruct the covariance matrix of the desired signal, and then estimate the guidance vector of the desired signal;

[0012] 5) Calculate the optimal weight vector based on IPNCM and the desired signal guidance vector, and use the optimal weight vector to process the spatial snapshot signal to achieve echo separation.

[0013] Furthermore, in step 1), the process of using multiple transmitters to transmit STSO waveforms and simultaneously receiving echoes of multiple transmitted waveforms using a MIMO SAR elevation multi-channel system is as follows: Multiple partially orthogonal waveforms are generated using the STSO mechanism, ensuring that these transmitted waveforms are orthogonal to each other within a certain time delay. These orthogonal waveforms are transmitted through multiple transmitters, and the echoes of all transmitted waveforms are simultaneously recorded using the elevation multi-channel receiving system. In the STSO mechanism, since the transmitted waveforms are not completely orthogonal, digital beamforming (DBF) technology is needed to further suppress interference components in the echoes.

[0014] Furthermore, in step 2), the incoming regions of each signal component are divided, and the Capon power spectrum is calculated based on the sample covariance matrix estimated using elevation-dimensional spatial snapshots. The process is as follows: To achieve subsequent IPNCM reconstruction and desired signal guidance vector estimation, it is necessary to determine the approximate incoming regions of various signal components and divide them into non-overlapping regions. In addition, the elevation-dimensional covariance matrix carries spatial information of various incident signals, which can generally be obtained through adjacent distance-pulse unit sample estimation. The Capon power spectrum is then calculated based on the divided regions and the sample covariance matrix.

[0015] Furthermore, in step 3), the process of reconstructing the IPNCM using the Capon power spectrum and removing the desired signal component from the covariance matrix is ​​as follows: In the region from which the interference signal originates, the Capon power spectrum corresponding to each incident angle is calculated, and this spectrum is integrated within that region to obtain the covariance matrix corresponding to a certain interference signal. The covariance matrices of all the reconstructed interference signals are then summed to obtain the final IPNCM.

[0016] Furthermore, in step 4), the process of reconstructing the covariance matrix of the desired signal and then estimating the guidance vector of the desired signal is as follows: the covariance matrix of the desired signal is obtained in a manner similar to that of IPNCM reconstruction. The guidance vector of the desired signal can be estimated using the covariance matrix of the desired signal. Generally, steps 3) and 4) can further improve the output signal-to-interference-plus-noise ratio (SINR) of the beamformer, while avoiding the impact of guidance vector mismatch on the beamformer performance, thereby ensuring the effectiveness of echo separation.

[0017] Furthermore, in step 5), the optimal weight vector is calculated based on the IPNCM and the desired signal guidance vector. This optimal weight vector is then used to process the spatial snapshot signal to achieve echo separation. The process involves calculating the optimal weight vector using the IPNCM and the desired signal guidance vector based on the Minimum Variance Distortionless Response (MVDR) criterion. By processing the spatial snapshot signal, interference components in the echo can be suppressed, thereby achieving the goal of MIMO SAR echo separation.

[0018] Compared with existing technologies, the beneficial effects of the invented robust adaptive beamforming method for airborne MIMO SAR echo separation are:

[0019] This method, by reconstructing the IPNCM, removes the desired signal component from the sample covariance matrix, further improving the output SINR of the beamformer. At low SNR, the beamformer sacrifices some interference suppression capability to obtain a lower sidelobe level, resulting in relatively good echo separation performance for MIMO SAR. Furthermore, when interference and desired signal guidance vectors mismatch, this method avoids signal 'self-cancellation' through IPNCM reconstruction and desired signal guidance vector estimation, thus ensuring the beamformer's suppression performance. In summary, the proposed method exhibits good robustness and broad application prospects in practical processing. Attached Figure Description

[0020] Figure 1It is the transmit and receive geometric model of an airborne MIMO SAR system;

[0021] Figure 2 This is a schematic diagram of the mismatch between the guidance vectors caused by the terrain undulations.

[0022] Figure 3 It is the division of the incoming regions of each signal component in the unit to be processed;

[0023] Figure 4 These are the truth map and the imaging result map, where (a) is the truth map of the scene illuminated by transmitted waveform 1; (b) is the truth map of the scene illuminated by transmitted waveform 2; (c) is the imaging result of the mixed signal after pulse compression of transmitted waveform 2; (d) is the echo imaging result of waveform 1 after DBF processing; and (e) is the echo imaging result of waveform 2 after DBF processing.

[0024] Figure 5 These are the truth map and the imaging result map, where (a) is the truth map; (b) is the imaging result after robust DBF processing when SNR=30dB; (c) is the imaging result after robust DBF processing when SNR=15dB; and (d) is the imaging result after robust DBF processing when SNR=5dB.

[0025] Figure 6 It is a terrain relief model;

[0026] Figure 7 The images are the imaging results, where (a) is the imaging result of a single transmitted waveform under terrain undulation; (b) is the imaging result of mixed signals under terrain undulation; and (c) is the imaging result after processing with robust DBF under terrain undulation. Detailed Implementation

[0027] The robust adaptive beamforming method for airborne MIMO SAR echo separation proposed in this invention will be described in detail below with reference to the accompanying drawings.

[0028] Figure 1 This represents the transmit and receive geometric model of an airborne MIMO SAR system. The radar system operates in frontal side-looking mode, and the radar platform moves along the X-axis. The aircraft's velocity is v. aTo simplify the model, the remaining channels used for receiving echo signals in the elevation direction are not labeled. In a MIMO SAR system, all transmitters simultaneously transmit STSO waveforms, and each receiver receives echoes from all transmitted waveforms. If the echoes from all transmitted waveforms in each receiving channel can be separated, more equivalent phase centers can be obtained in the azimuth direction. This means that compared to traditional single-input multiple-output (SIMO) systems, MIMO SAR systems can have lower PRF and wider mapping swathes.

[0029] Taking a point target as an example, after the received echo signals of M transmitted waveforms are down-converted and pulse-compressed, the amplitude information of each component can be expressed as:

[0030]

[0031] In the formula, k r T p B r σ0 and t0 represent the frequency modulation slope, pulse width, bandwidth, scattering coefficient, and echo delay, respectively. sinc(x) = sin(πx) / (πx). Considering the far-field assumption, the echo delays corresponding to different transmitted waveforms are approximately equal, and the effect of this difference on sinc(x) is negligible. Therefore, the echo delay is always t0 in the equation. Observing the above equation, it can be seen that for M transmitted waveforms, 2M-1 signal components can be generated after pulse compression, and the echo delay between two adjacent components is T0. p / M. When the imaging scene is large enough, a certain processing unit can simultaneously contain a maximum of 2M-1 signal components. Assume the angle of arrival of each signal component in this unit is θ = {θ...} -M+1 , …, θ -1 θ s ,θ1,…,θ M-1 The corresponding spatial guidance vector can be expressed as:

[0032]

[0033] In the formula, K represents the number of elevation receiving channels, and w k (θ)=2π / λ·kdsinθ. The elevation spatial snapshot signal used for subsequent DBF processing can be expressed as:

[0034]

[0035] In the formula, α=[α i,-M+1 , …, α i,-1 α s α i,1 , …, α i,M-1 ] TThis represents the amplitude information of each incident signal, and T denotes the transpose operation. q v s Let and represent the interference signal guidance vector, the desired signal guidance vector, and the noise component, respectively. Since the various components have different angles of arrival, echo signal separation can be achieved using DBF (Digital Frame Filtering) technology. By weighting the snapshot signal in the above equation, a single signal component can be obtained. The theoretical weighting matrix ω LS It can be calculated as:

[0036] ω LS =[ω -M+1 ,ω -M+2 ,…,ω M-1 ]=V·(V H V) -1

[0037] In the formula, H and (·) -1 These represent the conjugate transpose operation and the matrix inversion operation, respectively. By choosing different weight vectors ω... q Different interference components α can then be separated. i,q Or signal component α s When the selected weight vector is ω0, the desired separation result can be obtained.

[0038] However, when there are undulating terrain elements in the scene, such as Figure 2 As shown, θ0, θ s and θ s '' represents the downward viewing angle from the beam center, and the theoretical and actual incident angles of the desired signal, respectively. The target point's height fluctuation is Δh, and the incident angle error caused by the height fluctuation is Δθ=arcsin(H / R)-arcsin((H-ΔH) / R), where arcsin(·) represents the arcsine value. Furthermore, the incident angles of each interference signal component will also fluctuate within a small range. Therefore, the actual weighted matrix should be:

[0039] ω′ LS = (V+e)·((V+e) H (V+e)) -1

[0040] In the formula, e represents the steering vector error matrix, and V+e represents the steering vector matrix corresponding to the actual incoming signal. When prior information contains errors, directly using prior information for DBF processing of the echo signal will degrade the separation performance of each transmitted waveform. Therefore, this invention proposes a robust adaptive beamforming method for echo separation in airborne MIMO SAR. To address the DOA error caused by terrain undulations during DBF processing and improve the robustness of the beamformer, two aspects are typically considered: firstly, removing the desired signal component from the sample covariance matrix and reconstructing the IPNCM; secondly, re-estimating the steering vector of the desired signal. These processes can improve the output SINR of the beamformer and ensure that constraints are accurately applied to the direction of the desired signal to avoid 'self-cancellation'. The main steps are as follows:

[0041] 1) Use multiple transmitters to transmit STSO waveforms and utilize a MIMO SAR elevation multi-channel system to simultaneously receive echoes of multiple transmitted waveforms;

[0042] In this MIMO SAR system, the STSO transmission mechanism is selected. To obtain the best performance, the generated M transmission waveforms can be represented as follows:

[0043]

[0044] In the formula, rect(·) and exp{·} represent the normalized window function and exponential function, respectively. Considering that each receiving channel simultaneously receives the echoes of all transmitted waveforms, the mixed signal collected by the nth receiving channel can be expressed as:

[0045]

[0046] In the formula, Let δ(t) and f represent the single-channel response corresponding to transmitter m and receiver n. c and t n,m These represent the impulse response function, carrier frequency, and echo delay, respectively.

[0047] 2) Divide the incoming regions of each signal component and estimate the sample covariance matrix based on the elevation-dimensional spatial snapshot.

[0048] Calculate the Capon power spectrum;

[0049] For 2M-1 incident signals in a certain unit to be processed, the sample covariance matrix can be estimated as follows:

[0050]

[0051] In the formula, K, L and v n,JThese represent the number of pitch channels, the number of samples, and the steering vector of the noise signal, respectively. and These represent the desired signal power, interference component power, and noise power, respectively. S+I = [u1, u2, ..., u 2M-1 ] and U n =[u 2M u 2M+1 ,…,u K ]. Λ S+I and Λ n This is a diagonal matrix. It's important to note that when there are no noise components, the above equation is generally a singular matrix; noise components increase the dependencies between elements of the covariance matrix. Furthermore, the eigenvalues ​​obtained from the above decomposition mainly consist of two parts: the principal eigenvalue λ... i The values ​​of i = 1, 2, ..., 2M-1 are related to the power of each signal and are not zero. The smallest eigenvalue is λ. i Let i = 2M, 2M+1, ..., K represent the noise power, and its value tends to 0. i The principal eigenvectors u corresponding to i = 1, 2, ..., 2M-1 i Let i = 1, 2, ..., 2M-1 span the signal subspace. Similarly, the noise subspace can be represented by u. i Let i = 2M, 2M+1, ..., K represent the values.

[0052] IPNCM is reconstructed using the Capon power spectrum, which describes the power distribution of various signals in space. Due to the sparsity of the incident signals in space, the covariance matrix can be obtained by integrating the Capon power spectrum over each interference region. Figure 3 As shown, the originating region of the q-th interference signal is Θ. q Furthermore, the regions from which each interference signal originates do not overlap. Generally, the Capon power spectrum is:

[0053]

[0054] Where a(θ) represents the guiding vector corresponding to a certain incident direction.

[0055] 3) Reconstruct the IPNCM using Capon power spectrum and remove the desired signal component from the covariance matrix;

[0056] Based on the incoming range defined in step 2) and the calculated Capon power spectrum, the covariance matrix of a specific interference signal can be reconstructed.

[0057]

[0058] IPNCM reconstruction requires obtaining and summing the covariance matrices of all interfering signals. Generally, the estimation result of IPNCM can be expressed as:

[0059]

[0060] IPNCM reconstruction is crucial to beamformer performance. Traditional beamformer methods, such as Sample Covariance Matrix Inversion (SMI) beamformers, utilize... replace The optimal weight vector is calculated. On the one hand, this operation will cause the output SINR of the beamformer to decrease; on the other hand, when the signal's guide vector is mismatched, the SMI method will exhibit a signal 'self-cancellation' phenomenon.

[0061] 4) Reconstruct the covariance matrix of the desired signal, and then estimate the guidance vector of the desired signal;

[0062] To further improve the robustness of the proposed method and increase the output SINR of the beamformer, it is necessary to estimate the steering vector of the desired signal. To obtain the steering vector of the desired signal, its covariance matrix needs to be reconstructed, similar to the reconstruction method of IPNCM, i.e.:

[0063]

[0064] Generally, for signals from any direction, their covariance matrix has the following properties:

[0065]

[0066] In the formula, and a real These represent the theoretical and actual values ​​of the signal guidance vector, σ and σ', respectively. 2 This represents the power of the signal. Observing the above equation, we can see that the estimated guidance vector differs from the actual guidance vector of the signal only in amplitude information. Therefore, the estimated value of the desired signal guidance vector can be obtained through the above equation. Based on the norm constraint of the guidance vector, the desired signal guidance vector can be expressed as:

[0067]

[0068] In the formula, This represents the theoretical value of the desired signal guidance vector, where ||·|| is the modulo operation.

[0069] 5) Calculate the optimal weight vector based on IPNCM and the desired signal guidance vector, and use the optimal weight vector to process the spatial snapshot signal to achieve echo separation.

[0070] Generally speaking, the output of a beamformer consisting of K receiving units at a certain moment is:

[0071] y out =ω H χ=ω H ·(x s +x i +x n )

[0072] In the formula, ω represents the weight vector calculated by the proposed method. s x i and x n These represent the desired signal component, interference component, and noise component in the spatial snapshot, respectively. Generally, the optimal weight vector ω should ensure that the output SINR of the beamformer is maximized. SINR is defined as:

[0073]

[0074] In the formula, E{·} represents the statistical expectation, and R i+n This represents the theoretical IPNCM. The problem of maximizing the output SINR is mathematically equivalent to the MVDR problem, i.e.:

[0075]

[0076] The optimal weight vector can be calculated using the above formula as follows:

[0077]

[0078] The products generated in steps 3) and 4) and Substituting into the above formula, we can obtain the optimal weight vector for echo separation.

[0079] The robust adaptive beamforming method for echo separation in airborne MIMO SAR, proposed in this invention, was used to process simulation data to verify the robustness of the proposed method. The simulation parameters involved in the experiment are shown in Table 1. The radar operated in front-side-looking mode, with K=4 and M=2. The illumination scenarios for the two transmitted waveforms selected in this experiment are as follows: Figure 4 (a) and Figure 4 As shown in (b), in order to ensure that the imaging results of the mixed signal contain more blur energy, the size of the imaging scene is set to 1.5km*1.5km, and the horizontal and vertical directions of the image represent the range and azimuth directions, respectively. Figure 4 Image (c) represents the imaging result after pulse compression of the transmitted waveform. It can be seen that the image contains a significant amount of blur energy, severely affecting image quality. Therefore, the robust DBF method proposed in this paper needs to be used to process the echo data. The specific processing results are as follows: Figure 4 (d) and Figure 4 As shown in (e), the imaging results processed by the DBF method are basically consistent with the true image, indicating that the method can suppress interference components in the echo and has good suppression performance.

[0080] Table 1. Main parameters involved in the simulation data.

[0081]

[0082] Furthermore, the echo separation performance of the proposed DBF method under different SNRs is as follows: Figure 5 As shown. Figure 5 In Figure (a), the scene is shown as a ground truth image, with a scene size of 1km * 1.5km. The SNR values ​​set for the experiment were 30dB, 15dB, and 5dB. The imaging results after DBF processing are shown below. Figure 5 (b) Figure 5 (c) and Figure 5 As shown in (d). According to Figure 5 It can be seen that the imaging results under different SNRs are basically consistent with the true value map, indicating that the proposed method has good robustness under different SNRs. Generally, when the SNR decreases, the residual interference components in the echo inevitably increase. This is because the beamformer will sacrifice a certain degree of interference suppression capability to ensure the beamformer sidelobe level and prevent more noise components from existing in the separated echo. However, as shown in Figure (5), the residual interference components caused by the decrease in SNR will not have a serious impact on the observation effect of the imaging scene.

[0083] Finally, simulation experiments analyzed the impact of guidance vector mismatch on the performance of the proposed robust DBF processing. The terrain undulation model used in this experiment is as follows: Figure 6 As shown, the terrain undulation is 100m, and the projected size of the image scene on the horizontal plane is 600m*2.5km. Figure 7 In the middle (a), the imaging result generated by a single emission waveform under this fluctuation model is the true value map. Figure 7 Figure (b) shows the imaging result of the mixed signal without DBF processing. It can be seen that blur energy severely affects the scene observation effect. Therefore, the proposed robust DBF method is used to suppress interference in the echo, and the suppressed imaging result is shown below. Figure 7 As shown in (c), this method can overcome the impact of DOA mismatch on beamformer performance and achieve better processing results. In summary, the robust adaptive beamforming method for airborne MIMO SAR echo separation proposed in this invention exhibits good separation performance under various conditions, demonstrating the robustness and broad application prospects of the method.

Claims

1. A robust adaptive beamforming method for airborne MIMO SAR echo separation, characterized in that: The method comprises the following steps: 1) transmitting STSO waveforms by multiple transmitters, and simultaneously receiving echoes of multiple transmitted waveforms by using a MIMO SAR elevation multi-channel system; 2) dividing the direction-of-arrival area of each signal component, and estimating a sample covariance matrix according to the elevation dimension space snapshot, and calculating a Capon power spectrum; 3) reconstructing the IPNCM by using the Capon power spectrum, and eliminating the expected signal component in the covariance matrix; 4) reconstructing the covariance matrix of the expected signal, and further estimating the steering vector of the expected signal; 5) calculating an optimal weight vector according to the IPNCM and the expected signal steering vector, processing the space snapshot signal by using the optimal weight vector, and realizing echo separation; In the step 1), the process of transmitting STSO waveforms by multiple transmitters and simultaneously receiving echoes of multiple transmitted waveforms by using a MIMO SAR elevation multi-channel system is as follows: multiple partially orthogonal waveforms are generated by using the STSO mechanism, and the transmitted waveforms can ensure orthogonality within a certain time delay; the orthogonal waveforms are transmitted by multiple transmitters, and the echoes of all transmitted waveforms are simultaneously recorded by using the elevation multi-channel receiving system; in the STSO mechanism, since the transmitted waveforms are not completely orthogonal, the digital beam forming (DBF) technology is needed to further suppress the interference components in the echoes; In the step 2), the process of dividing the direction-of-arrival area of each signal component, and estimating a sample covariance matrix according to the elevation dimension space snapshot, and calculating a Capon power spectrum is as follows: in order to realize the subsequent IPNCM reconstruction and expected signal steering vector estimation, the approximate direction-of-arrival area of various signal components needs to be determined and divided without overlap; in addition, the spatial information of various incident signals is carried in the elevation dimension covariance matrix, which can be obtained by estimating adjacent range-pulse unit samples; the Capon power spectrum is calculated according to the divided area and the sample covariance matrix; In the step 3), the process of reconstructing the IPNCM by using the Capon power spectrum, and eliminating the expected signal component in the covariance matrix is as follows: in the direction-of-arrival area of the interference signal, the Capon power spectrum corresponding to each incident angle is calculated, and the covariance matrix corresponding to a certain interference signal is obtained by integrating the Capon power spectrum in the area; all the covariance matrices of the reconstructed interference signals are added to obtain the final IPNCM; In the step 4), the process of reconstructing the covariance matrix of the expected signal, and further estimating the steering vector of the expected signal is as follows: the covariance matrix of the expected signal is obtained by using a similar method as the IPNCM reconstruction; the expected signal covariance matrix can be used to estimate the steering vector of the expected signal; In the step 5), the process of calculating an optimal weight vector according to the IPNCM and the expected signal steering vector, processing the space snapshot signal by using the optimal weight vector, and realizing echo separation is as follows: the optimal weight vector is calculated according to the IPNCM and the expected signal steering vector, and the space snapshot signal is processed by using the optimal weight vector to realize echo separation. The step 5) is to calculate the optimal weight vector according to the IPNCM and the desired signal steering vector, and to process the spatial snapshot signal by using the optimal weight vector to realize the echo separation process, which is: according to the minimum variance distortionless response (MVDR) criterion, the optimal weight vector is calculated by using the IPNCM and the desired signal steering vector; by processing the spatial snapshot signal, the suppression of the interference component in the echo can be realized, so as to achieve the purpose of MIMO SAR echo separation.

Images