Angle-Doppler two-dimensional combined main lobe interference resisting method based on blind source separation

Abstract

The invention relates to a radar anti-interference technology, and discloses a two-dimensional angle-Doppler combined main lobe interference resisting method based on blind source separation. The method comprises the following steps of: firstly, receiving an echo signal acquired by an array element by using a radar, and rearranging multi-pulse data of each channel according to a PRI period to obtain a multi-channel multi-pulse receiving signal matrix; and then, calculating a covariance matrix of the received signal, carrying out eigenvalue decomposition, and estimating the number of information sources by adopting an MDL method based on an information theory criterion. And taking the estimated number of information sources as prior information, taking the rearranged channel pulse data as input of a JADE blind source separation algorithm, performing pulse compression on separated independent signals, and identifying target signals according to a peak value judgment method. According to the method, under the condition that source signal priori knowledge is not needed, signal separation is achieved by fully utilizing the angle difference and the speed difference between the target and interference, various interference types can be effectively restrained, and the method has wide applicability.

Description

Technical Field

[0001] The invention belongs to the technical field of radar anti-interference, and in particular relates to an angle-Doppler two-dimensional joint anti-main lobe interference method based on blind source separation. Technical Background

[0002] With the rapid development of modern electronic technology, various interference methods have emerged one after another, and the electromagnetic interference environment faced by radar has become increasingly complex. At present, the technology for suppressing sidelobe interference has become relatively mature, and a variety of effective suppression methods have emerged, such as adaptive sidelobe cancellation, ultra-low sidelobe antenna and sidelobe blanking. However, when the interference signal enters the radar antenna mainly through the main lobe or near the main lobe, the sidelobe anti-interference technology is basically ineffective. The blind source separation algorithm does not require accurate prior information of the source signal, so it shows great potential in the engineering application of radar mainlobe anti-interference. The mainlobe anti-interference method of traditional blind source separation usually relies on the phase difference generated by the angle or speed difference between the target and the interference to separate the signal, but when the target signal and the interference are highly overlapped and the dimension is insufficient, the traditional algorithm is difficult to achieve effective separation, which brings severe challenges to the survival of radar in complex electromagnetic environments. Summary of the invention

[0003] The purpose of the present invention is to overcome the shortcomings of the above-mentioned prior art and propose a two-dimensional angle-Doppler joint anti-main lobe interference method based on blind source separation to solve the problem that the traditional blind source separation algorithm has insufficient dimensions and is difficult to effectively separate the target signal, thereby suppressing the main lobe interference. To achieve the above purpose, the present invention adopts the following technical methods to achieve it.

[0004] Step 1: construct a signal model in which the transmitted signal is linear frequency modulation, perform time delay and Doppler frequency shift processing on the signal, and obtain an echo observation signal containing the target and interference.

[0005] Step 2: rearrange and segment the sampling data of the echo observation signal according to the PRI period, and use the angle difference between the target and the interference, as well as the speed difference between the target and the interference between multiple pulses to construct the receiving signal of the angle-Doppler two-dimensional joint array.

[0006] Step 3, calculate the covariance matrix of the angle-Doppler two-dimensional joint received signal matrix, and perform eigendecomposition on it to obtain eigenvalues. Then, the MDL (Minimum Description Length) method based on information theory criteria is used to estimate the number of information sources.

[0007] Step 4: The estimated number of signal sources is used as prior information, and the received signal of the angle-Doppler two-dimensional joint array is input into the JADE (Joint Approximate Diagonalization of Eigenmatrices) blind source separation algorithm to obtain each separated independent signal.

[0008] Step 5: Perform pulse compression processing on each of the separated independent signals, and identify the target signal using a peak determination method.

[0009] Compared with the prior art, the present invention has the following beneficial effects:

[0010] (1) The present invention inputs the multi-channel multi-pulse data as a whole into the JADE blind source separation algorithm, fully utilizes the angle difference and speed difference between the target and the interference, and expands the one-dimensional signal processing into a two-dimensional joint processing, thereby achieving effective separation of the target signal and the interference signal. (2) Even if the source signal does not meet the condition that the number of receiving channels is not less than the number of source signals before separation, it will usually be converted into an overdetermined problem before blind source separation. This shows that the angle-Doppler two-dimensional joint blind source separation algorithm is still applicable under underdetermined conditions, and has a wider applicability than the channel-level and single-channel multi-pulse blind source separation algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0012] Figure 1 It is a technical implementation flow chart of angle-Doppler combined anti-main lobe interference.

[0013] Figure 2 This is the array receiving signal amplitude diagram simulated on the Matlab platform.

[0014] Figure 3 This is the separation result diagram of the angle-Doppler two-dimensional joint blind source separation algorithm on the Matlab platform when the interference and target have the same speed but different angles.

[0015] Figure 4 This is the separation result diagram of the angle-Doppler two-dimensional joint blind source separation algorithm on the Matlab platform when the interference and target have the same angle but different speeds. DETAILED DESCRIPTION

[0016] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0017] like Figure 1As shown, the angle-Doppler two-dimensional joint anti-main lobe interference method based on blind source separation includes the following steps:

[0018] Step 1: construct a signal model in which the transmitted signal is a linear frequency modulation, perform time delay and Doppler frequency shift processing on the signal, and obtain an echo observation signal containing the target and interference. Specifically, step 1 includes the following sub-steps:

[0019] Sub-step 1.1, assume that the radar transmits a linear frequency modulation signal

[0020]

[0021] In the formula, A i and are the amplitude and initial phase of the ith pulse signal, f0 is the carrier frequency, K = B / T p is the frequency modulation slope, B is the bandwidth, T p It is a long time.

[0022] Assume that the target distance radar is D r km, radial velocity is v r m / s, target angle is θ r , then the target echo signal of the mth (m=1, ..., M)th receiving element at the ith pulse is

[0023]

[0024] Where α is the target echo amplitude, λ is the wavelength, and the reflection delay t i =(D r -v r (i-1) / T) / c, T is the pulse repetition period.

[0025] Sub-step 1.2, assuming that the interference signal carrier frequency is the same as the target transmission signal carrier frequency, and the interference pattern is dense false target deception interference, which can be expressed as

[0026]

[0027] In the formula, τ k represents the delay of the kth false target, f j is the modulated Doppler frequency, Z is the number of false targets, E = 2 e , e is the number of quantization bits, φ j =2πf0t+πKt 2 .

[0028] Assume the interference distance is D j km, radial velocity is v j m / s, and the interference angle is θ j, then the interference echo signal of the mth (m=1, ..., M)th receiving element at the ith pulse is

[0029]

[0030] In the formula, B i and are the amplitude and initial phase of the i-th pulse signal, β is the interference echo amplitude, δ i Interference reflection delay.

[0031] Step 2, rearrange and segment the sampling data of the echo observation signal according to the PRI period, and use the angle difference between the target and the interference, as well as the speed difference between the target and the interference between multiple pulses to construct the received signal of the angle-Doppler two-dimensional joint array. Specifically, step 2 includes the following sub-steps:

[0032] Sub-step 2.1, rearrange the multi-pulse data of a single receiving channel with PPI as the period. Assuming that it is the mth channel, the radar transmits a total of L pulses, then the multi-pulse receiving signal of this channel can be expressed as

[0033]

[0034] Sub-step 2.2, based on sub-step 2.1, consider adding Gaussian white noise N(t), then the rearranged received signal of L pulses of M channels can be expressed as

[0035]

[0036] Step 3, calculate the covariance matrix of the angle-Doppler two-dimensional joint received signal matrix, and perform eigendecomposition on it to obtain eigenvalues. Then, the MDL method based on information theory criterion is used to estimate the number of information sources.

[0037] The multi-channel multi-pulse received signal matrix is ​​obtained from step 2, and its covariance matrix is ​​calculated to obtain

[0038] R X =XX H / L (7)

[0039] The covariance matrix R X Perform eigendecomposition to obtain

[0040] E D =DVD H (8)

[0041] In the formula, V=diag[λ1, λ2,…,λ M ] is R X The eigenvalues ​​of M ] is the corresponding eigenvector.

[0042] Arrange the obtained eigenvalue vector V in order from large to small, and get

[0043] V s = sort(V) = {l1, l2, ...l M} (9)

[0044] The MDL method based on information theory includes likelihood function and penalty function parts, and it is easy to obtain the source estimation number N.

[0045]

[0046] Step 4: Use the estimated number of signal sources as prior information and input the received signal of the angle-Doppler two-dimensional joint array into the JADE blind source separation algorithm to obtain each separated independent signal. Specifically, step 4 includes the following sub-steps:

[0047] Sub-step 4.1, the received signal needs to be pre-processed before blind source separation, which specifically includes the following sub-steps:

[0048] Sub-step 4.1.1, zero mean preprocessing, the main function of this processing is to remove the offset (or DC component) in the data so that the mean of the data is zero. You can directly subtract the expected value of the observed signal from its own to make its own expectation zero, which can be expressed as

[0049]

[0050] In the formula, is the observed signal after zero mean preprocessing.

[0051] Sub-step 4.1.2, whitening preprocessing, the main purpose of whitening is to remove the correlation in the data and transform each feature of the data into an independent component with unit variance, thereby reducing the complexity of blind source separation. Let the whitening matrix be C, then the observed signal after whitening preprocessing can be expressed as The key to whitening is to find the whitening matrix C so that

[0052]

[0053] In the formula, R XX is the covariance matrix of the signal before whitening preprocessing, I is the unit matrix, and the whitening matrix C can be obtained

[0054]

[0055] In the formula, Λ k For R XX The diagonal matrix composed of the first k largest eigenvalues ​​of 2 is the noise variance, Ek is a k-order unit vector, V k is a matrix consisting of eigenvectors corresponding to the k eigenvalues.

[0056] Sub-step 4.2, the preprocessed observation signal is used as the input signal of the JADE algorithm. The input signal model is

[0057] X(k)=BS(k)+N(k) (14)

[0058] The observed signal after preprocessing is

[0059]

[0060] Wherein, U is a unitary matrix, and U=CB.

[0061] From equation (14), we can see that we only need to calculate the matrix U and then recover the source signal by inverting it. The JADE algorithm uses the fourth-order cumulant matrix D of the observed signal. z (P) is used to estimate the matrix U, and the fourth-order cumulant is defined as follows:

[0062]

[0063] Where cum(·,·,·,·) is a fourth-order cumulant operation, [·] * It is a conjugate operation, 1≤i,j, h,l≤N.

[0064] Matrix D z The expression of the i-th row and j-th column of (P) is

[0065]

[0066] Where P is an arbitrary matrix, p ij is the element in the i-th row and j-th column of the matrix P.

[0067] According to the multilinear properties of higher-order cumulants, the fourth-order cumulant matrix can be expressed as

[0068]

[0069] In the formula, λ j is the fourth-order accumulator of the signal source, [·] H is the conjugate transpose, u j is the j-th column vector of matrix U, Λ p is a diagonal matrix, and

[0070] For the matrix D z (P) Perform feature decomposition

[0071] D z(P)=VAV H (19)

[0072] In the formula, Λ is a diagonal matrix, matrix V is an estimate of matrix U, and there is

[0073] V=UD E T (20)

[0074] Where D E is a diagonal matrix with diagonal elements of ±1, and T is a permutation matrix.

[0075] Through the matrix V, the expression of the separated signal S(k) can be obtained as

[0076]

[0077] From the above formula, we can get the separated independent signal components.

[0078] Step 5: Pulse compression is performed on the separated independent signals, and the target signal is identified by the peak discrimination method. Since there is uncertainty in the order of the separated independent source signals, pulse compression can help distinguish the target echo from the interference signal. The processing formula for pulse compression is as follows

[0079]

[0080] In the formula, conv(·) represents the convolution operation, (·) * It means to find the conjugate.

[0081] Table 1 Radar simulation parameter settings

[0082]

[0083] The computer simulation is used to analyze the suppression performance of the angle-Doppler two-dimensional joint blind source separation algorithm on the main lobe dense forwarding interference. The simulation analysis is divided into two scenarios:

[0084] (1) The interference and target have the same speed but different angles. Set the target's azimuth angle to 30°, radial velocity to -300m / s, the generated Doppler frequency shift to -240Hz, the target distance to 90km, and the corresponding discrete sampling points to 1201 points. The radial velocity of the interference is the same as the target, and the azimuth angle is 28°, so the relative angle with the target is -2°, the distance between the interference source and the radar is 40km, and the discrete sampling points are 533 points. Assume that after receiving each radar pulse signal, the interference source will forward the modulated signal emitted by the radar, and the forwarding interval is 22 sampling points. The signal-to-noise ratio is set to 10dB, and the signal-to-interference ratio is -30dB. In this simulation scenario, the array received signal amplitude diagram is as follows: Figure 2As shown in Figure 1, the target signal is masked by the interference signal, and the interference signal amplitude is large, which seriously affects the recognition of the target signal. The angle-Doppler two-dimensional joint blind source separation algorithm is used to suppress interference. The results are shown in Figure 1. Figure 3 As shown in Figure 2, the angle-Doppler two-dimensional joint blind source separation algorithm can effectively separate the interference and target signals by using the angle difference between the interference and the target.

[0085] (2) The interference and target have the same angle but different speeds. Under the same simulation conditions, the target's azimuth is 30°, the radial velocity is -300m / s, and the generated Doppler frequency shift is -240Hz; the interference source has the same azimuth as the target, the radial velocity is -400m / s, and the relative velocity between the two is -100m / s. In this scenario, the angle-Doppler two-dimensional joint blind source separation algorithm is used for interference suppression. The results are as follows: Figure 4 By utilizing the phase difference generated by the speed difference between the interference and the target, the algorithm effectively separates the interference and target signals.

[0086] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A two-dimensional angle-Doppler joint anti-main lobe interference method based on blind source separation, characterized in that: The following steps are involved: Step 1: construct a signal model in which the transmitted signal is linear frequency modulation, perform time delay and Doppler frequency shift processing on the signal, and obtain an echo observation signal containing the target and interference. Step 2: rearrange and segment the sampling data of the echo observation signal according to the PRI period, and use the angle difference between the target and the interference, as well as the speed difference between the target and the interference between multiple pulses to construct the receiving signal of the angle-Doppler two-dimensional joint array. Step 3, calculate the covariance matrix of the angle-Doppler two-dimensional joint received signal matrix, and perform eigendecomposition on it to obtain eigenvalues. Then, the MDL (Minimum Description Length) method based on information theory criteria is used to estimate the number of information sources. Step 4: The estimated number of signal sources is used as prior information, and the received signal of the angle-Doppler two-dimensional joint array is input into the JADE (Joint Approximate Diagonalization of Eigenmatrices) blind source separation algorithm to obtain each separated independent signal. Step 5: Perform pulse compression processing on each independent signal obtained by separation, and identify the target signal using the peak determination method.

2. According to the angle-Doppler two-dimensional joint anti-main lobe interference method based on blind source separation in claim 1, step 1 comprises the following sub-steps: Sub-step 1.1, assume that the radar transmits a linear frequency modulation signal In the formula, A i and are the amplitude and initial phase of the ith pulse signal, f0 is the carrier frequency, K = B / T p is the frequency modulation slope, B is the bandwidth, T p It is a long time. Assume that the target distance radar is D r km, radial velocity is v r m / s, target angle is θ r , then the target echo signal of the mth (m=1, ..., M)th receiving element at the ith pulse is Where α is the target echo amplitude, λ is the wavelength, and the reflection delay t i =(D r -v r (i-1) / T) / c, T is the pulse repetition period. Sub-step 1.2, assuming that the interference signal carrier frequency is the same as the target transmission signal carrier frequency, the interference pattern is dense false target deception interference, the deception interference can be expressed as In the formula, τ k represents the delay of the kth false target, f j is the modulated Doppler frequency, Z is the number of false targets, E = 2 e , e is the number of quantization bits, φ j =2πf0t+πKt 2 . Assume the interference distance is D j km, radial velocity is v j m / s, and the interference angle is θ j , then the interference echo signal of the mth (m=1, ..., M)th receiving element at the ith pulse is In the formula, B i and are the amplitude and initial phase of the i-th pulse signal, β is the interference echo amplitude, δ i Interference reflection delay.

3. According to the angle-Doppler two-dimensional joint anti-main lobe interference method based on blind source separation in claim 2, step 2 comprises the following sub-steps: Sub-step 2.1, rearrange the multi-pulse data of a single receiving channel with PRI as the period. Assuming that it is the mth channel, the radar transmits a total of L pulses, then the multi-pulse receiving signal of this channel can be expressed as Sub-step 2.2, based on sub-step 2.1, consider adding Gaussian white noise N(t), then the rearranged received signal of L pulses of M channels can be expressed as 4. The angle-Doppler two-dimensional joint anti-main lobe interference method based on blind source separation according to claim 3 is characterized in that: The covariance matrix of the angle-Doppler two-dimensional joint received signal matrix is ​​calculated and eigen-decomposed to obtain eigenvalues. Then, the MDL method based on information theory criterion is used to estimate the number of signal sources. The multi-channel multi-pulse received signal matrix is ​​obtained from step 2, and its covariance matrix is ​​calculated to obtain R X =XX H / L The covariance matrix R X Perform eigendecomposition to obtain E D =DVD H In the formula, V=diag[λ1, λ2,…,λ M ] is R X The eigenvalues ​​of M ] is the corresponding eigenvector. Arrange the obtained eigenvalue vector V from large to small, and we get V s =sort(V)={l1,l2,…l M } The MDL method based on information theory includes likelihood function and penalty function parts, and it is easy to obtain the source estimate number N.

5. According to the angle-Doppler two-dimensional joint anti-main lobe interference method based on blind source separation in claim 4, step 4 specifically comprises the following sub-steps: Sub-step 4.1, the received signal needs to be pre-processed before blind source separation, which specifically includes the following sub-steps: Sub-step 4.1.1: The main purpose of this processing is to remove the offset (or DC component) in the data so that the mean of the data is zero. The expected value of the observed signal can be directly subtracted to make the expected value zero, which can be expressed as In the formula, is the observed signal after zero mean preprocessing. Sub-step 4.1.2, whitening preprocessing, the main purpose of whitening is to remove the correlation in the data and transform each feature of the data into an independent component with unit variance, thereby reducing the complexity of blind source separation. Let the whitening matrix be C, then the observed signal after whitening preprocessing can be expressed as The key to whitening is to find the whitening matrix C so that In the formula, R XX is the covariance matrix of the signal before whitening preprocessing, I is the unit matrix, and the whitening matrix C can be obtained In the formula, Λ k For R XX The diagonal matrix composed of the first k largest eigenvalues ​​of 2 is the noise variance, E k is a k-order unit vector, V k is a matrix consisting of eigenvectors corresponding to the k eigenvalues. Sub-step 4.2: Use the preprocessed observation signal as the input signal of the JADE algorithm. The input signal model is X(k) = BS(k) + N(k) The observed signal after preprocessing is Wherein, U is a unitary matrix, and U=CB. From the input signal model, it can be seen that the source signal can be recovered by calculating the matrix U and inverting it. The JADE algorithm uses the fourth-order cumulant matrix D of the observed signal z (P) is used to estimate the matrix U, and the fourth-order cumulant is defined as follows: Where cum(·,·,·,·) is a fourth-order cumulant operation, [·] * It is the conjugate operation, 1≤i,j,h,l≤N. Matrix D z The expression of the j-th row and j-th column of (P) is Where P is an arbitrary matrix, p ij is the element in the i-th row and j-th column of the matrix P. According to the multilinear properties of higher-order cumulants, the fourth-order cumulant matrix can be expressed as In the formula, λ j is the fourth-order accumulator of the signal source, [·] H is the conjugate transpose, u j is the j-th column vector of matrix U, Λ p is a diagonal matrix, and For the matrix D z (P) Perform feature decomposition D z (P)=VΛV H In the formula, Λ is a diagonal matrix, matrix V is an estimate of matrix U, and there is V=UD E T Where D E is a diagonal matrix with diagonal elements of ±1, and T is a permutation matrix. Through the matrix V, the expression of the separated signal S(k) can be obtained as From the above formula, we can get the separated independent signal components.

6. The angle-Doppler two-dimensional joint anti-main lobe interference method based on blind source separation according to claim 5, characterized in that: Step 5: Pulse compression is performed on the separated independent signals, and the target signal is identified by the peak discrimination method. Since there is uncertainty in the order of the separated independent source signals, pulse compression can help distinguish the target echo from the interference signal. The processing formula for pulse compression is as follows In the formula, conv(·) represents the convolution operation, (·) * It means to find the conjugate.

Images