Blind source separation-based radar additive composite jamming space-time joint suppression method and system

Abstract

The application discloses a radar additive composite jamming space-time joint suppression method and system based on blind source separation, and the method comprises the following steps: S1, estimating the DOA of the composite jamming signal according to the observation signal; S2, performing coherent accumulation on the radar echo signal to improve the signal-to-noise ratio; S3, constructing a blocking matrix by using the DOA information of the jamming signal to suppress the composite jamming; S4, performing whitening processing on the signal output by S3 to remove the correlation between different channel signals; S5, suppressing the residual jamming signal by using the BCA algorithm on the whitened signal; and S6, constructing a signal similarity measurement basis by using the cross-correlation function on the signal output by S5 to select the real echo signal. The application effectively combines the space domain anti-jamming method and the time domain anti-jamming method to realize the suppression of the radar composite jamming, and the effect is better and the time is shorter than that of the existing anti-jamming method based on blind source separation.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of radar countermeasure, and particularly relates to a radar additive composite jamming space-time joint suppression method and system based on blind source separation (BSS). BACKGROUND

[0002] As an important detection device, radar can discover targets and determine their positions in space by using electromagnetic signals, and is widely used in many fields, especially in the military field. With the development of technology, electronic warfare (EW) is becoming more and more fierce, and radar is facing serious jamming threats. Using strong electronic jamming to suppress the information system of the opponent radar has become the most effective and most commonly used electromagnetic countermeasure, which greatly challenges the real-time and authenticity of information acquisition of the radar system.

[0003] The jamming effect of simply using one jamming pattern is often not good, and the jamming party needs to combine multiple jamming patterns together for use to form composite jamming to the radar. The commonly used composite jamming is additive composite of noise jamming and deception jamming, which can not only achieve the effect of suppressing the true target but also make the radar detect the false target to achieve the purpose of deception. The influence of the composite jamming electromagnetic environment on the radar system is very serious, which seriously affects the sensing ability and information acquisition ability of the radar system on the battlefield. In order to enable the radar to normally play a role in the future battlefield environment and provide information protection for the own side, it is necessary to study the radar anti-composite jamming technology. Under the threat of additive composite jamming faced by the radar, the radar anti-composite jamming method used in the prior art has the problems of high complexity and slow convergence. SUMMARY

[0004] In view of the problems of high complexity and slow convergence of the existing radar anti-composite jamming method in the face of the threat of additive composite jamming faced by the radar, the application provides a radar additive composite jamming space-time joint suppression method and system based on blind source separation (BSS), which improves the convergence speed.

[0005] There is an anti-jamming scene as follows: a single array radar has a uniform linear array (ULA) antenna, M antenna elements are arranged on the same straight line at uniform intervals, the target echo signal and the jamming signal are different in spatial azimuth angle, and the observation signal model can be represented as:

[0006] X=AS+N=[x1,x2,…,x M ] T

[0007] Wherein, X represents M*L dimensional observation signal matrix, L is signal snapshot number. Any column vector a(θ K ) in M*K dimensional steering matrix A=[a(θ k ), a(θ k ), …, a(θ k )] is steering vector of spatial source signal with incident angle θ k (k=1, …, K), and mathematical expression is a(θ T )=[1, …, exp(j2π(M-1)(d / λ)sin(θ K )) T . S=[s1, s2, …, s M ] T is K*L dimensional spatial source signal matrix, N=[n1, n2, …, n

[0008] In order to achieve the above object, the application adopts the following technical scheme:

[0009] A radar additive composite jamming space-time joint suppression method based on BSS, comprising the following steps:

[0010] S1, estimating the direction of arrival (DOA) of the jamming signal according to the observation signal data.

[0011] S2, accumulating the pulse signal to improve the signal to interference and noise ratio (SINR) of the signal;

[0012] S3, using the DOA information of the jamming signal obtained in step S1 to construct a blocking matrix, and eliminating the jamming signal energy in the accumulated signal in step S2 through blocking matrix preprocessing (BMP) to realize the suppression of the jamming signal;

[0013] S4, whitening the output signal after the blocking matrix preprocessing in step S3 to remove the correlation between different channel signals;

[0014] S5, using a blind source separation algorithm to suppress the residual jamming in the whitened signal in step S4;

[0015] S6, selecting the echo signal from the separated signal.

[0016] As a preferred scheme, in step S1, the DOA of the jamming signal is estimated, comprising the following steps:

[0017] S11, the radar first calculates the covariance matrix C of the observation signal X according to the observation signal X XX That is:

[0018]

[0019] In the formula, (·) H is the conjugate transpose, and L is the number of signal snapshots.

[0020] S12, the observation signal covariance matrix C XX is subjected to eigenvalue decomposition, that is:

[0021] C XX = QΛQ H

[0022] In the formula, Q is an eigenmatrix, and Λ is an eigenvalue matrix. The eigenvalue matrix Λ = diag(λ1, …, λ M ) is arranged in descending order of elements, that is, λ1≥ λ2≥ …≥ λ M The eigenvectors in the eigenmatrix Q are rearranged in the order of corresponding eigenvalues as Q'; the first K+1 large eigenvalues in the eigenvalue matrix Λ are retained, and the average of the remaining eigenvalues is:

[0023]

[0024] The reconstructed eigenvalue matrix Λ' = diag(λ1, …, λ K+1 , σ 2 , …, σ 2 ), and the reconstructed covariance matrix C' XX That is:

[0025] C' XX = Q'Λ'Q' H

[0026] In the formula, Λ' is the reconstructed eigenvalue matrix, and Q' is the rearranged eigenmatrix.

[0027] S13, in order to improve the accuracy of DOA estimation, the reconstructed covariance matrix C' XX is used to estimate the DOA of the interference signal in combination with a high-resolution angle of arrival estimation algorithm (such as the Root-MUSIC algorithm).

[0028] As a preferred solution, in the step S2, coherent accumulation is used for signal accumulation, and the coherent accumulation output signal is X C , and the process is represented as:

[0029]

[0030] In the formula, x C is one-way signal output by coherent accumulation, and x kFor the echo signal of the kth pulse, N is the number of coherent pulses, e is the natural constant, j is the imaginary unit, and φ is the phase difference.

[0031] As a preferred solution, the step S3 of blocking matrix preprocessing comprises the following steps:

[0032] S31, calculate the blocking matrix B of the interference signal according to the DOA of the interference signal; let the number of interference sources be N, and calculate the weight of the interference according to the DOA of the interference signal:

[0033]

[0034] In the formula, u i is the weight of the ith interference source, θ i is the DOA of the ith interference source; let μ be a row vector containing N+1 elements, where the element is represented as:

[0035] μ i = (-1) i-1 sum{e sum[nchoosek(u,i-1)]}, i = 1, …, N+1

[0036] In the formula, sum(·) is a cumulative sum function, and nchoosek(·) is a combination function; the blocking matrix B is initialized as a zero matrix of (M-N) × M, and the element is filled in it:

[0037] B(i, i: i+N) = μ, i = 1, …, M-N

[0038] S32, the accumulated signal is preprocessed by the blocking matrix, that is:

[0039] X B = BX C

[0040] In the formula, B is the B blocking matrix, and X B is the signal obtained by preprocessing.

[0041] As a preferred solution, the step S4 of whitening processing comprises the following steps:

[0042] S41, calculate the whitening matrix W of the preprocessed signal X B ;

[0043] The covariance matrix C B of the signal X B is Shur decomposed as:

[0044]

[0045] Wherein, Q and Q ⊥ are high semi-orthogonal matrices, and the column vectors of the two matrices respectively span the signal subspace and the noise subspace. is a diagonal matrix containing n nonzero eigenvalues; signal X B The whitening matrix W is expressed as:

[0046] W = Σ -1 Q H

[0047] S42, using the whitening matrix W to whiten the preprocessed signal X B , that is:

[0048] Z = WX B

[0049] In the formula, Z is the signal matrix after whitening processing.

[0050] As a preferred solution, in step S5, the blind source separation algorithm uses a bounded component analysis algorithm (BCA) to suppress residual interference. The initial parameters of the algorithm are set as: 100 rounds of iteration, initial accuracy of the algorithm is 0.1, and final accuracy is 10 -5 .

[0051] As a preferred solution, in the step S6, the signal output by the step S5 is used to construct a signal similarity measure basis by using a cross-correlation function, and a real echo signal is selected therefrom, specifically including the following steps:

[0052] S61, calculating the cross-correlation function between the separated signal and the original pulse signal, that is:

[0053]

[0054] In the formula, R i is the cross-correlation function of the ith signal, s is a reference pulse signal, is the separated ith signal;

[0055] S62, taking the maximum value of the absolute value of the cross-correlation function as a signal similarity measure basis, that is:

[0056] p i = max(R i |)

[0057] In the formula, p i is the signal similarity measure basis;

[0058] S63, selecting the signal with the largest signal similarity as the echo signal, that is:

[0059] i0 = argmax(p i )

[0060] Therefore, the echo signal

[0061] The application also discloses a radar additive composite jamming space-time joint suppression system based on blind source separation, which is used for executing the method and comprises the following modules.

[0062] A DOA estimation module is configured to estimate the direction of arrival (DOA) of the jamming signal according to the observation signal data.

[0063] A pulse accumulation module is configured to accumulate the pulse signal and improve the signal-to-noise ratio.

[0064] A blocking matrix preprocessing module is configured to construct a blocking matrix by using the DOA information of the jamming signal, eliminate the energy of the jamming signal in the accumulated signal by blocking matrix preprocessing, and realize suppression of the jamming signal.

[0065] A whitening processing module is configured to perform whitening processing on the output signal after the blocking matrix preprocessing.

[0066] A blind source separation module is configured to suppress the residual jamming in the whitened signal by using a blind source separation algorithm.

[0067] A signal selection module is configured to select the echo signal from the separated signal.

[0068] Compared with the prior art, the application has the following advantages.

[0069] (1) The BCA algorithm is applied to the field of radar anti-jamming for the first time, and compared with the radar anti-jamming method using the blind source separation algorithm, the application has low complexity and short running time.

[0070] (2) The application proposes a space domain and time domain joint radar composite jamming suppression scheme, and by combining the space domain anti-jamming method and the time domain anti-jamming method, better anti-composite jamming effect can be obtained.

[0071] (3) The application proposes a blocking matrix derivation method under multiple jamming sources, and the blocking matrix under multiple jamming sources can be derived, so that the jamming can be effectively suppressed in the composite jamming scene generated by multiple jamming sources.

[0072] (4) The application proposes a signal selection method based on the cross-correlation function, and the echo signal can be correctly extracted from the multiple signals after the anti-jamming processing by the experiment.

[0073] In view of the composite jamming threat faced by the radar and the low operation efficiency and poor anti-composite jamming effect of the existing method, the radar additive composite jamming space-time joint suppression method and system based on blind source separation of the present application introduces the BCA algorithm into the radar anti-jamming problem, solves the problem that the active deceptive jamming is difficult to suppress by combining the DOA estimation and the blocking matrix preprocessing, and reduces the complexity of the anti-composite jamming method and the short operation time through the BCA algorithm. In addition, the present application also provides a method for selecting echo signals from multiple signals, which can effectively select the correct echo signals. BRIEF DESCRIPTION OF DRAWINGS

[0074] Figure 1 is a flow chart of a radar additive composite jamming space-time joint suppression method based on BSS provided by the preferred embodiment of the present application;

[0075] Figure 2 is a signal processing flow chart of a radar additive composite jamming space-time joint suppression method based on BSS provided by the preferred embodiment of the present application;

[0076] Figure 3 is a comparison chart of noise amplitude modulation and dense false target jamming additive composite jamming suppression before and after;

[0077] Figure 4 is a comparison chart of noise amplitude modulation, dense false target jamming and intermittent sampling and forwarding additive composite jamming suppression before and after;

[0078] Figure 5 is a performance comparison chart of the preferred method and other jamming suppression methods of the present application;

[0079] Figure 6 is a system block diagram of a radar additive composite jamming space-time joint suppression system based on BSS provided by the preferred embodiment of the present application. DETAILED DESCRIPTION

[0080] The technical solutions of the present application are further explained and described below through embodiments.

[0081] For reference Figure 1 and Figure 2 , Figure 1 is a method flow chart provided by a preferred embodiment of the present application, Figure 2 is a signal processing overall block diagram provided by the preferred embodiment of the present application. Specifically, the radar additive composite jamming space-time joint suppression method based on BSS includes the following steps:

[0082] S1, estimate the direction of arrival (DOA) of the interference signal according to the observation signal data. This step is specifically as follows:

[0083] S11、Radar first calculates the covariance matrix C of observation signal X according to the observation signal X XX That is:

[0084]

[0085] In the formula, (·) H is the conjugate transpose, and L is the number of signal snapshots.

[0086] S12, the observation signal covariance matrix C XX is subjected to eigenvalue decomposition, that is:

[0087] C XX = QΛQ H

[0088] In the formula, Q is the characteristic matrix, and Λ is the eigenvalue matrix. The eigenvalue matrix Λ = diag(λ1,…, λ M ) is arranged in descending order of elements, that is, λ1≥ λ2≥…≥ λ M The eigenvectors in the characteristic matrix Q are rearranged in the order of corresponding eigenvalues as Q'. The first K+1 large eigenvalues of the eigenvalue matrix Λ are retained, and the average of the remaining eigenvalues is:

[0089]

[0090] The reconstructed eigenvalue matrix Λ' = diag(λ1,…, λ K+1 ,σ 2 ,…,σ 2 ), and the reconstructed covariance matrix C' XX is:

[0091] C' XX = Q'Λ'Q' H

[0092] S13, in order to improve the accuracy of DOA estimation, the reconstructed covariance matrix C' XX is used in combination with a high-resolution angle of arrival estimation algorithm (such as Root-MUSIC algorithm) to estimate the DOA of the interference signal.

[0093] S2, the coherent accumulation of the pulse signal is performed to improve the signal-to-interference-and-noise ratio of the signal, and the coherent accumulation signal is X C , and the process can be represented as:

[0094]

[0095] In the formula, x C is one-way signal output by coherent accumulation, x i is the echo signal of the kth pulse, N is the number of coherent pulses, and φ is the phase difference.

[0096] S3, using the DOA information of the interference signal obtained in step S1 to construct a blocking matrix, and using the blocking matrix to pre-process the accumulated signal to eliminate the energy of the interference signal in the accumulated signal, so as to suppress the interference signal. This step is specifically as follows:

[0097] S31, calculating the blocking matrix B of the interference signal according to the DOA of the interference signal. Assuming that the number of interference sources is N, the weight of the interference is calculated according to the DOA of the interference signal:

[0098]

[0099] In the formula, u i is the weight of the i th interference source, θ i is the DOA of the i th interference source. Assuming that μ is a row vector containing N+1 elements, and the elements can be represented as:

[0100]

[0101] In the formula, sum(·) is a cumulative sum function, and nchoosek(·) is a combination function. The blocking matrix B is initialized as a zero matrix of (M-N)×M, and the elements are filled in as:

[0102] B(i,i:i+N)=μ,i=1,…,M-N

[0103] S32, pre-processing the accumulated signal using the blocking matrix, that is:

[0104] X B =BX C

[0105] In the formula, B is the B blocking matrix, and X B is the signal obtained after the pre-processing.

[0106] S4, performing whitening processing on the output signal after the blocking matrix pre-processing in step S3. This step is specifically as follows:

[0107] S41, calculating the whitening matrix W of the pre-processed signal X B . The covariance matrix C B of the signal X B is Schur decomposed as:

[0108]

[0109] In the formula, Q and Q ⊥ are high semi-orthogonal matrices, and the column vectors of the two matrices respectively span the signal subspace and the noise subspace. is a diagonal matrix containing n non-zero eigenvalues. In order to reduce the dimension of the signal, the signal X BThe whitening matrix W can be expressed as:

[0110] W = Σ -1 Q H

[0111] S42, using the whitening matrix W to pre-process the signal X B whitening processing, that is:

[0112] Z = WX B

[0113] In the formula, Z is the signal matrix after whitening processing.

[0114] S5, using a blind source separation algorithm to suppress residual interference in the signal whitened in step S4. The process of suppressing interference signals from the whitened signal using the BCA algorithm can be expressed as:

[0115]

[0116] In the formula, u m is the mth column vector of the source signal extraction matrix U. Projecting the signal y = [y1, …, y L ] onto the hyperplane can be recorded as the point set:

[0117] Y = {(Re(y l ), Im(y l ))l = 1, …, L}

[0118] In the formula, Re(·) is the real part of a complex number, Im(·) is the imaginary part of a complex number, and L is the number of signal snapshots. Obviously, the point set Y is a convex set, and its boundary can be defined as a convex hull. The mathematical expression of the convex hull is:

[0119]

[0120] In the formula, is the convex hull of the convex set Y, records the elements that make up the convex hull, and conv(·) is the convex hull calculation.

[0121] The perimeter of the convex hull can be recorded as

[0122]

[0123] The contrast function of the nth iteration constructed by the perimeter of the convex hull can be expressed as:

[0124]

[0125] In the formula, is the extraction vector ​the L2 norm of u Thus Δy i = u m Δz i , i = 1,..., V, The gradient is:

[0126]

[0127] The step size of gradient descent is set as:

[0128]

[0129] where eta is the algorithm accuracy. The updated vector u m can be expressed as:

[0130]

[0131] The new extracted vector u is normalized. Let the extracted signal be expressed as:

[0132]

[0133] S6, selecting the echo signal from the separated signals. The specific process is as follows:

[0134] S61, calculating the cross-correlation function between the separated signals and the original pulse signal, i.e.

[0135]

[0136] where R i is the cross-correlation function of the ith signal, s is the reference pulse signal, is the separated ith signal;

[0137] S62, taking the maximum value of the absolute value of the cross-correlation function as the signal similarity measure, i.e.

[0138] ρ i = max(R i |)

[0139] where ρ i is the signal similarity measure;

[0140] S63, selecting the signal with the maximum signal similarity as the echo signal, i.e.

[0141] i0= argmax(ρ i )

[0142] Thus, the echo signal

[0143] According to the above-mentioned embodiment, the test is carried out, wherein the number of space targets is 1, and the number of space sources is known. The signal-to-noise ratio (SNR) is 10 dB, the real target is set at a distance of 2000 range bins from the radar, and the incident angle of the echo signal is 0°; the composite jamming is noise amplitude modulation (NAM) jamming and dense false target jamming (DFTJ) jamming generated by 2 jamming sources at different spatial positions and added at the radar end, wherein the jamming-to-signal ratio (JSR) of the NAM is 20 dB, and the incident angle is 1°; the JSR of the DFTJ is 10 dB, and the incident angle is 0.1°. The composite jamming suppression test is carried out, and see Figure 3 . Figure 3 The composite jamming suppression effect comparison diagram provided by the preferred embodiment of the present application is shown in FIG. 6, wherein (a), (b), and (c) are respectively the time-domain diagram, the time-frequency diagram, and the pulse compression diagram of the observed signal before jamming suppression; (d), (e), and (f) are respectively the time-domain diagram, the time-frequency diagram, and the pulse compression diagram of the selected echo signal after jamming suppression. It can be seen that the method provided by the present application can effectively resist the additive composite situation of two jamming sources, the time-frequency domain image of the selected echo signal is clean and clear, the signal time-frequency characteristics can be clearly seen, and the target peak value is clear after pulse compression. It is calculated that the signal-to-jamming-and-noise ratio gain of the output signal compared with the observed signal after taking the anti-jamming measures is 34.60 dB.

[0144] In the SNR is 10 dB, the real target is set at a distance of 2000 range bins from the radar, and the incident angle of the echo signal is 0°; the composite jamming is noise amplitude modulation (NAM) jamming, dense false target jamming (DFTJ) jamming, and interrupted-sampling repeater jamming (ISRJ) jamming generated by 3 jamming sources at different spatial positions and added at the radar end, wherein the JSR of the NAM is 20 dB, and the incident angle is -1°; the JSR of the DFTJ is 10 dB, and the incident angle is 0.2°; the JSR of the ISRJ is 15 dB, and the incident angle is 0.1°. The composite jamming suppression test is carried out, and see Figure 4 . Figure 4is a composite interference suppression effect comparison chart provided by the preferred embodiment of the present application, wherein (a), (b), (c) are respectively the time-domain chart, time-frequency chart and pulse pressure chart of the observation signal before interference suppression; (d), (e), (f) are respectively the time-domain chart, time-frequency chart and pulse pressure chart of the selected echo signal after interference suppression. After anti-interference using the method of the present application, the recovered real echo signal is relatively clear, which is in sharp contrast with that before anti-interference, and (f) and (c) show that the false target peak is obviously suppressed and does not affect the detection of the real target peak. Through calculation, compared with the observation signal, the signal-to-interference-and-noise ratio gain of the recovered real echo signal is 40.14 dB.

[0145] The incident angle of the real echo signal is 0°; the JSR of the NAM is 15 dB, and the incident angle is 1°; the JSR of the DFTJ is 10 dB, and the incident angle is 0.3°; the JSR of the ISRJ is 15 dB, and the incident angle is -0.5°; the SNR is changed from 5 dB to 15 dB, and 50 Monte Carlo experiments are performed under each SNR. Please refer to Figure 5 . Among them, (a), (b), (c) are respectively the signal-to-noise ratio gain curve, the false target suppression rate curve and the algorithm running time curve with the change of the signal-to-noise ratio (SNR). (a) shows that the SINR gain of the method (BMP-BCA) provided by the present application is slightly lower than that of the Music Method and similar to that of the EMP-JADE, compared with the anti-interference method of the feature projection combined with the feature matrix joint similar diagonalization algorithm (EMP-JADE) and the anti-interference method of the blind source separation combined with the multiple signal classification (Music Method).(b) shows that the false target suppression rate of the method provided by the present application and the Music Method increases significantly with the increase of the SNR, the false target suppression rate of the EMP-JADE presents fluctuation, and the false target suppression rate of the method provided by the present application is better than that of the other two methods.(c) shows that the running time of the method of the present application is much smaller than that of the two compared methods, and the algorithm efficiency is higher.

[0146] As shown in Figure 6 , the present application also discloses a radar additive composite interference space-time joint suppression system based on blind source separation, which is used for executing the above method and comprises the following modules:

[0147] The DOA estimation module estimates the DOA of the interference signal according to the observation signal X and provides the DOA information of the interference signal for the blocking matrix preprocessing module;

[0148] The pulse accumulation module improves the signal-to-noise ratio in a coherent accumulation manner according to the observation signal;

[0149] The blocking matrix preprocessing module performs blocking matrix preprocessing on the observation signal X by using the DOA of the interference signal to suppress the composite interference;

[0150] Whitening module: eliminate the correlation between different channel signals, so that the characteristics of each independent signal source are more obvious;

[0151] Blind source separation module: optimize different channel signals, suppress residual interference signals and noise introduced by blocking matrix preprocessing;

[0152] Signal selection module: select echo signals from the multiple signals separated by blind source separation processing as output.

[0153] Other contents of this embodiment can refer to the above method embodiments.

[0154] In summary, the above content is a further detailed description of the present application in combination with specific embodiments, but the specific implementation of the present application cannot be limited to these descriptions. The DOA estimation method based on covariance matrix reconstruction, the blocking matrix derivation method under multiple interference sources, and the signal selection method based on cross-correlation function are the core technical contents of the present application. Any small sample radar composite jamming semi-supervised transfer learning recognition method formed by corresponding modification, replacement, improvement, etc. within the above design principles and implementation points of the present application should be included in the protection scope of the present application.

Claims

1. A radar additive composite jamming space-time joint suppression method based on blind source separation, characterized in that, The method comprises the following steps: S1, estimating the direction of arrival (DOA) of the interference signal according to the observation signal data; S2, accumulating the pulse signals to improve the signal-to-interference-and-noise ratio (SINR); S3, constructing a blocking matrix using the DOA information of the interference signal obtained in step S1, eliminating the interference signal energy in the accumulated signals by blocking matrix preprocessing, and realizing the suppression of the interference signal; S4, whitening the output signals after the blocking matrix preprocessing in step S3; S5, suppressing the residual interference in the whitened signals in step S4 by using a blind source separation algorithm; S6, selecting the echo signal from the separated signals; Step S5 is specifically as follows: the process of suppressing the interference signal y from the whitened signals by using a bounded component analysis algorithm is represented as follows: In the formula, (•) H For the conjugate transpose, Z is the signal matrix after whitening, u m It is the first of the source signal extraction matrix U. m Column vector; to store interference signals The projection onto the hyperplane is denoted as a point set: In the formula, Re(•) is the real part of a complex number, Im(•) is the imaginary part of a complex number, L is the number of signal snapshots, and the point set Y is a convex set, the boundary of which is defined as a convex hull; the mathematical representation of the convex hull is as follows: wherein is the convex hull of the set Y, The elements that make up the convex hull are recorded, and conv(•) computes the convex hull; The perimeter of the convex shell is denoted by : The first n Contrast function of the second iteration is represented as: wherein the L2 norm of L2; since therefore, the gradient is: Step size of gradient descent is set to: In the formula, eta is the algorithm accuracy; The update of the eta is represented as: Here the new Unitized; let The extracted signal Is represented as: where W is a whitening matrix, X B is the pre-processed signal.

2. The radar additive compound jamming blanking and nulling method based on blind source separation according to claim 1, characterized in that, Step S1 comprises the following steps: S11. The radar calculates an observation signal covariance matrix C from the observation signals X XX : In the formula, L is the number of signal snapshots. S12, on the observation signal covariance matrix C XX Eigenvalue decomposition is performed: In the formula, Q is a characteristic matrix, is an eigenvalue matrix; the eigenvalue matrix is arranged in descending order of element size, that is The eigenvectors in the characteristic matrix Q are rearranged in order of the corresponding eigenvalues, that is ; the eigenvalue matrix The first K+1 large eigenvalues, and the average of the remaining eigenvalues is: reconstructed eigenvalue matrix , reconstructed covariance matrix is: S13, using the reconstructed covariance matrix The DOA of the interference signal is estimated in combination with a high-resolution angle of arrival estimation algorithm.

3. The radar additive compound jamming blanking and nulling method based on blind source separation of claim 2, wherein, In step S2, the pulse signal is accumulated in phase to improve the signal-to-interference-and-noise ratio of the signal. The signal accumulated in phase is X C , and the process is represented as: wherein is a signal of a coherent accumulation output, is an echo signal of the k th pulse, is the number of coherent pulses, e is the natural constant, and j is the imaginary unit, is the phase difference.

4. The radar additive compound jamming blanking and nulling method based on blind source separation according to claim 3, characterized in that, Step S3 comprises the following steps: S31、According to the DOA of the interference signal, the blocking matrix B of the interference signal is calculated; let the number of interference sources be N According to the DOA of the interference signal, the weight of the interference is calculated: wherein is the weight of the i th interferer, is the DOA of the i th interferer; and is a row vector with N+1 elements, where the elements are given by: In the formula, sum(•) is a summation function, and nchoosek(•) is a combination function; the blocking matrix B is initialized as a zero matrix of (M-N) x M, and elements are filled in the zero matrix. S32, performing blocking matrix preprocessing on the accumulated signals: In the formula, B is the blocking matrix.

5. The radar additive compound jamming blanking and nulling method based on blind source separation according to claim 4, characterized in that, Step S4 is specifically as follows: S41, calculate the signal X obtained by preprocessing B whitening matrix W; X B covariance matrix C B Schur decomposition into: wherein is a high semi-orthogonal matrix, the column vectors of the two matrices span the signal subspace and the noise subspace, respectively; is a diagonal matrix containing non-zero eigenvalues; X B The whitening matrix representation of X is: S42, whitening matrix W is used on X B whitening processing is performed: 。 6. The radar additive compound jamming blanking and nulling method based on blind source separation according to claim 5, characterized in that, Step S6 comprises the following steps: S61, calculating the cross-correlation function between the separated signals and the original pulse signals: where s is a reference pulse signal, is the isolated i-th signal; S62, taking the maximum value of the absolute value of the cross-correlation function as the signal similarity measurement basis: In the formula, is a signal similarity measure dependence; S63, selecting the signal with the maximum signal similarity as the echo signal, wherein: Thus, the echo signal .

7. A radar additive composite jamming space-time joint suppression system based on blind source separation, the system being configured to perform the method according to any one of claims 1 to 6, characterized in that, The system comprises the following modules: A DOA estimation module: estimating the direction of arrival (DOA) of the interference signal according to the observation signal data; A pulse accumulation module: accumulating the pulse signals to improve the signal-to-noise ratio (SINR); A blocking matrix preprocessing module: constructing a blocking matrix using the DOA information of the interference signal, eliminating the interference signal energy in the accumulated signals by blocking matrix preprocessing, and realizing the suppression of the interference signal; A whitening processing module: whitening the output signals after the blocking matrix preprocessing; A blind source separation module: suppressing the residual interference in the whitened signals by using a blind source separation algorithm; A signal selection module: selecting the echo signal from the separated signals.

Images