A signal detection method and system based on iterative estimation in composite Gaussian clutter

Abstract

The invention discloses a signal detection method and system based on iterative estimation in composite Gaussian clutter, belonging to the field of radar target detection. First, it is assumed that clutter from different distance units has the same covariance matrix structure, but has unknown power mismatch, and the mismatch amount is a non-random non-negative number. Then, the clutter covariance matrix estimation value is constructed using the power mismatch amount obtained by the iterative estimation and the data to be detected, the training sample, and the iterative estimation. Then, the test statistic is calculated, and the detection threshold is determined according to the false alarm probability. Finally, the size between the test statistic and the detection threshold is compared. If the test statistic is greater than the detection threshold, it is determined that the target exists, otherwise it is determined that the target does not exist. The detection method and system of the invention take into account the problem of power mismatch in the training sample in the composite Gaussian clutter, obtain all the mismatch amounts through iterative estimation, and realize clutter suppression, signal accumulation and target judgment in an integrated manner, simplify the process, and improve the detection performance.

Description

Technical Field

[0001] The invention belongs to the field of radar target detection, and more specifically, relates to a signal detection method and system based on iterative estimation in composite Gaussian clutter. Background Art

[0002] For multi-channel detection methods, the clutter covariance matrix is ​​usually unknown, and training samples are needed to estimate the covariance matrix in the unit to be detected. The traditional assumption model assumes that the training samples and the data to be detected have the same statistical characteristics and obey the complex Gaussian distribution. This assumption is often not valid in complex electromagnetic environments.

[0003] Due to factors such as clutter fluctuations, the environment often exhibits non-uniform characteristics. In 1999, Kraut et al. proposed a generalized likelihood ratio detector in a partially uniform environment, namely an adaptive coherence estimator. In the partially uniform environment model, the training samples have the same covariance matrix, but there is an unknown power mismatch between the training samples and the data to be detected. In the complex and changeable real environment, this assumption is also too ideal. With the improvement of radar resolution, the number of scatterers in each resolution unit decreases, which makes the clutter in different distance resolution units have different local roughness and makes the amplitude fluctuation larger. In this case, the clutter exhibits non-Gaussian characteristics. The composite Gaussian model is a widely accepted model for modeling non-Gaussian clutter. Under the composite Gaussian model, it is often necessary to assign a priori statistical distribution to the texture component. However, the real clutter environment does not necessarily conform to the given prior. At this time, the detector designed according to the established statistical distribution of the texture component is bound to cause performance loss.

[0004] Therefore, how to improve the detection performance of radar targets in composite Gaussian clutter is a problem that technical personnel in this field need to solve urgently. Summary of the invention

[0005] In view of the above problems or improvement needs of the prior art, the present invention provides a signal detection method and system based on iterative estimation in composite Gaussian clutter, which aims to improve the detection performance of radar targets in composite Gaussian clutter.

[0006] To achieve the above object, according to one aspect of the present invention, a signal detection method based on iterative estimation in composite Gaussian clutter is provided, comprising the following steps:

[0007] Step 1: Assume that the signal is in a known subspace Data vector to be tested The radar receives L training samples z near the unit to be detected l , l = 1, 2, ..., L, where N is the number of radar system channels, p represents the dimension of the signal subspace, and the symbol represents a complex domain;

[0008] Step 2: According to the data to be tested z and training sample z l , l=1,2,…L, the joint probability density function, obtains the temporary estimate of the clutter covariance matrix, denoted as

[0009] Step 3: Use the data to be tested z and training samples z l , l = 1, 2, ... L, and the signal matrix H to derive the signal coordinate vector θ and the power mismatch γ between the training sample and the data to be detected l , l=1,2,…,L, the iterative estimation expression and l=1,2,…,L, where the superscript t+1 indicates the t+1th iteration and the superscript ^ indicates an estimate;

[0010] Step 4: Given an initial value of the power mismatch, denoted as l=1,2,…,L, iterative estimation is performed to obtain and l=1,2,…,L, where t max Indicates the maximum number of iterations, set to 15;

[0011] Step 5: Using the final estimate of the signal coordinate vector The final estimate of the power mismatch is l = 1, 2, ..., L, and a temporary estimate of the clutter covariance matrix Get the iterative estimate of the clutter covariance matrix

[0012] Step 6: Using the iterative estimate of the clutter covariance matrix The data to be tested z and the signal matrix H construct the test statistic t iter ;

[0013] Step 7: Use the test statistic t iter The detection threshold η is determined by the false alarm probability iter ;

[0014] Step 8: Compare the test statistic t iter and the detection threshold η iter The size between them is used to determine whether the target exists.

[0015] Furthermore, the temporary estimate of the clutter covariance matrix in step 2 is The expression is:

[0016]

[0017] Among them, the superscript (·) H represents the conjugate transpose.

[0018] Furthermore, the coordinate vector θ of the signal in step 3 and the power mismatch γ between the training sample and the data to be detected l , l=1,2,…,L, the iterative estimation expressions are:

[0019]

[0020] and

[0021]

[0022] in,

[0023] Furthermore, the initial value expression of the power mismatch amount in step 4 is:

[0024]

[0025] in,(·) - Represents the generalized inverse; the final and l=1,2,…,L, respectively:

[0026]

[0027] and

[0028]

[0029] in,

[0030]

[0031] Furthermore, the iterative estimate of the clutter covariance matrix in step 5 is The expression is:

[0032]

[0033] Furthermore, the test statistic t in step 6 iter The expression is:

[0034]

[0035] Furthermore, the detection threshold n in step 7 is iter The method to determine is:

[0036] η iter =t(n * ).

[0037] in, M is the number of Monte Carlo simulations, ρ is the false alarm probability value set by the system, For rounding operation, t(n * ) is a sequence Sort n from largest to smallest * maximum values, z(k) is the kth experiment of the data to be detected containing only clutter components, where k = 1, 2, …M.

[0038] Furthermore, in step 8, the determination of whether the target exists is performed according to the following results: if the test statistic t iter Greater than the detection threshold η iter , then the target is judged to exist, otherwise the target is judged not to exist.

[0039] On the other hand, the present invention provides a signal detection system based on iterative estimation in composite Gaussian clutter, comprising:

[0040] Temporary covariance matrix construction module: used to construct a temporary covariance matrix using the data to be detected and the training samples received by the radar;

[0041] Iterative estimation algorithm module: used to solve the iterative estimation value of the signal coordinate vector and the power mismatch amount by using the joint probability density function of the data to be detected and the training sample;

[0042] Iterative covariance matrix construction module: used to reconstruct the iterative covariance matrix using the estimated value of the signal coordinate vector, the estimated value of the power mismatch amount and the temporary covariance matrix;

[0043] Test statistic calculation module: used to construct test statistics using the data to be tested, signal subspace and iterative covariance matrix;

[0044] Detection threshold determination module: used to determine the detection threshold according to the false alarm probability;

[0045] Target judgment module: used to compare the test statistic and the detection threshold and judge whether the target exists. If the test statistic is greater than the detection threshold, the target is judged to exist; otherwise, the target is judged not to exist.

[0046] In general, the above technical solutions conceived by the present invention have the following advantages compared with the prior art:

[0047] Beneficial effects:

[0048] (1) Calculate the power mismatch between each training sample and the clutter covariance matrix of the unit to be detected through iterative estimation l=1,2,…L, which provides a guarantee for improving the estimation of the covariance matrix of the unit to be detected.

[0049] (2) By constructing an iterative covariance matrix estimate The effective estimation of the covariance matrix of the detected unit is achieved without relying on the statistical distribution of clutter.

[0050] (3) By constructing the test statistic t iter , clutter suppression, signal accumulation and target detection are achieved in an integrated manner.

[0051] (4) The present invention can solve the problem of detecting radar targets in composite Gaussian clutter and improve the radar target detection performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 It is a schematic diagram of the principle of a signal detection method based on iterative estimation in composite Gaussian clutter according to the present invention;

[0053] Figure 2 This is a structural framework diagram of a signal detection system based on iterative estimation in composite Gaussian clutter according to the present invention;

[0054] Figure 3 This is a comparison chart of detection probabilities between the method of the present invention and existing methods under different signal-to-noise ratios. DETAILED DESCRIPTION

[0055] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0056] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the protection scope of the present invention.

[0057] According to the composite Gaussian model, the clutter vector of the unit to be detected can be expressed as the product of the slowly varying texture component and the rapidly varying speckle component, namely:

[0058]

[0059] Where τ represents the slowly varying texture component, which is generally a non-negative random variable; x represents the rapidly varying speckle component, which obeys a circularly symmetric complex Gaussian distribution, and the conditional covariance matrix E[xx H |τ]=τR. The training sample can be expressed as l=1,2,…L, conditional covariance matrix The texture component of the clutter of the detection unit is normalized, that is, the covariance matrix is ​​R, and the training sample can be expressed as make Then the conditional covariance matrix of the normalized training sample is

[0060] Assuming that the number of channels of the radar system is N, if the signal exists, it can be represented by an N×1-dimensional vector s. For the subspace signal, assuming that s is located in the subspace H, it can be expressed as s=Hθ. Therefore, the detection problem in a composite Gaussian clutter environment can be expressed by the following binary hypothesis test:

[0061]

[0062] in, is the data to be tested, is the training sample, is a known column full rank signal matrix, is the unknown signal coordinate vector, is the unknown clutter covariance matrix, γ1,γ2,...,γ l >0 is an unknown parameter, which indicates the different power mismatch between each training sample and the data to be detected.

[0063] The purpose of the present invention is to improve the radar target detection performance in composite Gaussian clutter. To achieve the above purpose, please refer to Figure 1 As shown, this embodiment provides a signal detection method based on iterative estimation in composite Gaussian clutter, including:

[0064] Step 1: Assume that the signal is in a known subspace Data vector to be tested The radar receives L training samples z near the unit to be detected l , l = 1, 2, ..., L, where N is the number of radar system channels, p represents the dimension of the signal subspace, and the symbol represents a complex domain;

[0065] Step 2: According to the data to be tested z and training sample z l , l=1,2,…L, the joint probability density function, obtains the temporary estimate of the clutter covariance matrix, denoted as

[0066] Step 3: Use the data to be tested z and training samples z l , l = 1, 2, ... L, and the signal matrix H to derive the signal coordinate vector θ and the power mismatch γ between the training sample and the data to be detected l , l=1,2,…,L, the iterative estimation expression and l=1,2,...,L, where superscript t+1 indicates the t+1th iteration and superscript ^ indicates an estimate;

[0067] Step 4: Given an initial value of the power mismatch, denoted as l=1,2,...,L, iterative estimation is performed to obtain and l=1,2,...,L, where t max Indicates the maximum number of iterations, set to 15;

[0068] Step 5: Using the final estimate of the signal coordinate vector The final estimate of the power mismatch is l = 1, 2, ..., L, and a temporary estimate of the clutter covariance matrix Get the iterative estimate of the clutter covariance matrix

[0069] Step 6: Using the iterative estimate of the clutter covariance matrix The data to be tested z and the signal matrix H construct the test statistic t iter ;

[0070] Step 7: Use the test statistic t iter The detection threshold η is determined by the false alarm probability iter ;

[0071] Step 8: Compare the test statistic t iter and the detection threshold η iter The size between them is used to determine whether the target exists.

[0072] The temporary estimate of the clutter covariance matrix in step 2 The expression is:

[0073]

[0074] Among them, the superscript (·) H represents the conjugate transpose.

[0075] The coordinate vector θ of the signal in step 3 and the power mismatch γ between the training sample and the data to be detected l , l=1,2,…,L, the iterative estimation expressions are:

[0076]

[0077] and

[0078]

[0079] in,

[0080] The initial value expression of the power mismatch in step 4 is:

[0081]

[0082] in,(·) - Represents the generalized inverse; the final and l=1,2,…,L, respectively:

[0083]

[0084] and

[0085]

[0086] in,

[0087]

[0088] The iterative estimate of the clutter covariance matrix in step 5 is The expression is:

[0089]

[0090] The test statistic t in step 6 iter The expression is:

[0091]

[0092] The detection threshold η in step 7 iter The method to determine is:

[0093] η iter =t(n * );

[0094] in, M is the number of Monte Carlo simulations, ρ is the false alarm probability value set by the system, For rounding operation, t(n * ) is a sequence Sort n from largest to smallest * maximum values, z(k) is the kth experiment of the data to be detected containing only clutter components, where k = 1, 2, …M.

[0095] In step 8, the determination of whether the target exists is performed according to the following results: if the test statistic t iter Greater than the detection threshold η iter , then the target is judged to exist, otherwise the target is judged not to exist.

[0096] See also Figure 2 As shown, the present invention provides a signal detection system based on iterative estimation in composite Gaussian clutter, comprising:

[0097] Temporary covariance matrix construction module: used to construct a temporary covariance matrix using the data to be detected and the training samples received by the radar;

[0098] Iterative estimation algorithm module: used to solve the iterative estimation value of the signal coordinate vector and the power mismatch amount by using the joint probability density function of the data to be detected and the training sample;

[0099] Iterative covariance matrix construction module: used to reconstruct the iterative covariance matrix using the estimated value of the signal coordinate vector, the estimated value of the power mismatch amount and the temporary covariance matrix;

[0100] Test statistic calculation module: used to construct test statistics using the data to be tested, signal subspace and iterative covariance matrix;

[0101] Detection threshold determination module: used to determine the detection threshold according to the false alarm probability;

[0102] Target judgment module: used to compare the test statistic and the detection threshold and judge whether the target exists. If the test statistic is greater than the detection threshold, the target is judged to exist; otherwise, the target is judged not to exist.

[0103] The effect of the present invention is further described below in conjunction with simulation experiments.

[0104] See also Figure 3 As shown in the figure, it is a comparison diagram of the detection probability of the method of the present invention and the existing method under different signal-to-noise ratios. To simplify the calculation, let the number of radar system channels N = 8, the dimension of the signal subspace p = 3, the number of training samples L = 16, and the number of iterative estimations be 15. The speckle component of the composite Gaussian clutter in the simulation experiment is a zero-mean complex Gaussian random process, and its covariance matrix can be modeled as [R] i,j =ρ i-j , 1≤i,j≤N, where ρ=0.95. False alarm probability PFA=10 -3 , the signal-to-noise ratio is defined as θ H H H R -1 Hθ.

[0105] As can be seen from the figure, the detection performance of the method proposed in the present invention is significantly improved compared with the adaptive normalized matched filter (ANMF) using the normalized sample covariance matrix estimation.

[0106] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings, but it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to the specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0107] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A signal detection method based on iterative estimation in composite Gaussian clutter, characterized in that: include: Step 1: Assume that the signal is located in the known signal matrix Zhang Cheng's subspace, the data vector to be detected The radar receives L training samples z near the unit to be detected l , l = 1, 2, ..., L, where N is the number of radar system channels, p represents the dimension of the signal subspace, and the symbol represents a complex domain; Step 2: According to the data to be tested z and training sample z l , l=1,2,…L, the joint probability density function, obtains the temporary estimate of the clutter covariance matrix, denoted as Step 3: Use the data to be tested z and training samples z l , l = 1, 2, ... L, and the signal matrix H to derive the signal coordinate vector θ and the power mismatch γ between the training sample and the data to be detected l , l=1,2,…,L, the iterative estimation expression and Where, the superscript t+1 indicates the t+1th iteration, and the superscript ^ indicates an estimate; Step 4: Given an initial value of the power mismatch, denoted as Iterate and estimate and Among them, t max Indicates the maximum number of iterations, set to 15; Step 5: Using the final estimate of the signal coordinate vector and the final estimate of the power mismatch and a temporary estimate of the clutter covariance matrix Get the iterative estimate of the clutter covariance matrix Step 6: Using the iterative estimate of the clutter covariance matrix The data to be tested z and the signal matrix H construct the test statistic t iter ; Step 7: Use the test statistic t iter The detection threshold η is determined by the false alarm probability iter ; Step 8: Compare the test statistic t iter and the detection threshold η iter The size between them is used to determine whether the target exists.

2. The signal detection method based on iterative estimation in composite Gaussian clutter according to claim 1, characterized in that: The temporary estimate of the clutter covariance matrix in step 2 The expression is: Among them, the superscript (·) H represents the conjugate transpose.

3. The signal detection method based on iterative estimation in composite Gaussian clutter according to claim 2, characterized in that: The coordinate vector θ of the signal in step 3 and the power mismatch γ between the training sample and the data to be detected l , l=1,2,…,L, the iterative estimation expressions are: and in, 4. The signal detection method based on iterative estimation in composite Gaussian clutter according to claim 3 is characterized in that: The initial value expression of the power mismatch in step 4 is: in,(·) - Represents the generalized inverse; the final and They are: and in, 5. The signal detection method based on iterative estimation in composite Gaussian clutter according to claim 4, characterized in that: The iterative estimate of the clutter covariance matrix in step 5 is The expression is:

6. The signal detection method based on iterative estimation in composite Gaussian clutter according to claim 5, characterized in that: The test statistic t in step 6 iter The expression is:

7. The signal detection method based on iterative estimation in composite Gaussian clutter according to claim 6, characterized in that: The detection threshold η in step 7 iter The method for determining is: or iter =t(n * ); in, M is the number of Monte Carlo simulations, ρ is the false alarm probability value set by the system, For rounding operation, t(n * ) is a sequence Sort n from largest to smallest * maximum values, z(k) is the kth experiment of the data to be detected containing only clutter components, where k = 1, 2, …M.

8. The signal detection method based on iterative estimation in composite Gaussian clutter according to claim 7, characterized in that: In step 8, the determination of whether the target exists is performed according to the following results: if the test statistic t iter Greater than the detection threshold η iter , then the target is judged to exist, otherwise the target is judged not to exist.

9. A signal detection system based on iterative estimation in composite Gaussian clutter, characterized in that: A signal detection method based on iterative estimation in composite Gaussian clutter according to any one of claims 1 to 8, include: Temporary covariance matrix construction module: used to construct a temporary covariance matrix using the data to be detected and the training samples received by the radar; Iterative estimation algorithm module: used to solve the iterative estimation value of the signal coordinate vector and the power mismatch amount by using the joint probability density function of the data to be detected and the training sample; Iterative covariance matrix construction module: used to reconstruct the iterative covariance matrix using the estimated value of the signal coordinate vector, the estimated value of the power mismatch amount and the temporary covariance matrix; Test statistic calculation module: used to construct test statistics using the data to be tested, signal subspace and iterative covariance matrix; Detection threshold determination module: used to determine the detection threshold according to the false alarm probability; Target judgment module: used to compare the test statistic and the detection threshold and judge whether the target exists. If the test statistic is greater than the detection threshold, the target is judged to exist; otherwise, the target is judged not to exist.

Images