KR101768587B1 - Covariance matrix estimation method for reducing nonstationary clutter and heterogeneity clutter - Google Patents

Abstract

The present invention relates to a covariance matrix estimation method for a radar system, capable of improving the target detection performance by efficiently removing a nonstationary clutter and a heterogeneous clutter in a radar receiving signal. The covariance matrix estimation method includes: a step of calculating a first covariance matrix by using operation information of a transmitter and a receiver corresponding to a test cell of the receiving signal; a step of estimating a second covariance matrix from small amounts of secondary data except for the test cell in the receiving signal; a step of obtaining a final covariance matrix including both a nonstationary clutter element and a heterogeneous clutter element by multiplying a weighted value to each of the first covariance matrix and the second covariance matrix and composing the same; and a step of removing the nonstationary clutter element and the heterogeneous clutter element by applying a filter to the final covariance matrix.

Description

BACKGROUND OF THE INVENTION 1. Field of the Invention [0001] The present invention relates to a covariance matrix estimation method for suppressing abnormal clutter and heterogeneous clutter,

The present invention relates to a target detection of a radar system, and more particularly, to a covariance matrix estimation method capable of improving target detection performance by eliminating nonstationary clutter and heterogeneous clutter in a radar received signal .

The radar received signal consists of the target signal scattered by the target and the clutter and noise scattered by the surrounding environment. Therefore, the clutter components interfere with the target signal, and when the clutter signal is strong, masking the target signal makes it difficult to detect the target. Therefore, in order to increase the target (target) detection rate, a technology for efficiently suppressing the clutter signal from the received signal is required.

로 알려져 있다. 여기서 μ는 상수이고 R는 클러터 성분과 잡음성분의 공분산 행렬이고 s는 표적신호이다. 최적의 가중치(

)로부터 클러터 성분들을 적절히 제거하기 위해서는 클러터와 잡음으로 구성된 간섭신호의 공분산 행렬(R)을 알아야 한다. The optimal weight to maximize the signal-to-interference-plus-noise ratio (SINR)

. Where μ is a constant, R is the covariance matrix of the clutter component and the noise component, and s is the target signal. Optimum weighting (

), We need to know the covariance matrix (R) of the interference signal consisting of clutter and noise.

However, since the covariance matrix R of the interference signal can not be determined substantially, it is necessary to estimate the covariance matrix R from the secondary data excluding the test cell in the received signal. A typical technique for estimating the covariance matrix is a maximum likelihood (ML) estimation technique, which can be obtained by taking an average under an IID (Gaussian Independent and Identically Distributed) assumption which is repeated independently under the same conditions. The ML estimation scheme has an advantage of converging to an actual covariance matrix as the quantity (number) of the secondary data increases.

)에 적용하면 이상적인(Ideal) 경우에 비해 상당한 오차가 발생함을 예측할 수 있다. Reed, Mallet, and Brennan (RMB) rule showed that 2NM second order data is needed to have within 3dB loss (Loss) average for ML estimation technique. Where N and M represent the number of antenna elements and the number of pulses, respectively. Since it is practically impossible to collect 2NM data from the RMB rule in the radar environment and since the IID assumption is not appropriate for the environment in which the clutter changes, we simply add the covariance matrix obtained from the ML estimation method to the weights

), It can be predicted that a considerable error occurs as compared with the ideal case.

In a real environment, clutter is largely influenced by two factors. The first is the heterogeneity of clutter, which is caused by reflectivity change, shadowing, and discrete clutter. Second, the geometry-induced The nonstationary clutter in which the Angle-Doppler change occurs is the main factor. In a monostatic radar using a complete sidelooking array, it can be assumed that the stationary clutter is obtained if the radar system travels less than the CPI.

However, in a monostatic radar environment, a bistatic radar, and a multi-static radar environment, the non-linearity of the Angle-Doppler causes a nonstationary clutter. Therefore, it is not suitable to use a large number of secondary data in a nonstationary environment. Therefore, techniques for estimating a covariance matrix R using a small number of secondary data are required. These techniques have been studied in various fields and various estimation techniques have been proposed to reduce the number of data. Also, in order to suppress the nonstationary clutter caused by the geometry, techniques have been studied to compensate the geometrical variation between the radar system and the clutter and to have a stationary clutter signal.

However, the technique of estimating the covariance matrix obtained from a small amount of secondary data has heterogeneous clutter information but does not have geometry-induced Angle-Doppler clutter information of the test cell. Also, since the geometry variation compensation method does not include heterogeneous clutter information, the clutter suppression performance is degraded in a discrete clutter or clutter amplitude environment.

It is an object of the present invention to provide a covariance matrix estimation method including both a nonstationary clutter and a heterogeneous clutter component.

It is another object of the present invention to provide a covariance matrix estimation method capable of improving clutter suppression performance by removing an abnormal clutter and a heterogeneous clutter component in a received signal.

According to an aspect of the present invention, there is provided a method of estimating a covariance matrix of a radar system, comprising: calculating a first covariance matrix using operation information of a transmitter and a receiver corresponding to a test cell of a received signal; Estimating a second covariance matrix from a small amount of second order data excluding a test cell from a received signal; Multiplying the first and second covariance matrices by weights, respectively, to obtain a final covariance matrix including both a nonstationary clutter component and a heterogeneous clutter component; And applying a filter to the final covariance matrix to remove both the abnormal clutter component and the heterogeneous clutter component at the same time.

According to an embodiment of the present invention, the operation information may include a route and a speed of a transmitter and a receiver as information.

According to an embodiment of the present invention, the combining step includes eigen-decomposing a final covariance matrix; Obtaining first and second weights that maximize the variance of eigen-values of the final covariance matrix; And multiplying the first and second weights by the first and second covariance matrices, respectively, and combining the first and second weights.

The present invention relates to space-time adaptive processing for suppressing clutter, and is a method for improving the performance of a test cell by using a covariance matrix obtained from a training data and prior information on a moving path and a velocity of a transmitting end and a receiving end of a radar, We construct a covariance matrix to remove abnormal (nonstationary clutter) and heterogeneous clutter. Therefore, the covariance matrices synthesized according to the present invention include both abnormal clutter and heterogeneous clutter components information, which can be applied to the optimal weight, and can improve the cluster suppression performance and improve the detection performance of the target There is an effect that can be.

Fig. 1 is a schematic diagram of a radar system to which the present invention is applied.
2 is a block diagram of a receiving apparatus of a radar system according to the present invention.
3 shows an Angle-Doppler spectrum of a received signal including a target signal and a clutter signal;
FIG. 4 shows the Angle-Doppler spectrum after the clutter signal is removed in FIG. 3; FIG.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings, in which like or similar elements are denoted by the same or similar reference numerals, and redundant description thereof will be omitted. The suffix "module" and " part "for the components used in the following description are given or mixed in consideration of ease of specification, and do not have their own meaning or role. In the following description of the embodiments of the present invention, a detailed description of related arts will be omitted when it is determined that the gist of the embodiments disclosed herein may be blurred. It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory and are intended to provide further explanation of the invention as claimed. , ≪ / RTI > equivalents, and alternatives.

The terminology used in this application is used only to describe a specific embodiment and is not intended to limit the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In the present application, the terms "comprises" or "having" and the like are used to specify that there is a feature, a number, a step, an operation, an element, a component or a combination thereof described in the specification, But do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, or combinations thereof.

1 is a geometry of a radar system to which the present invention is applied.

In a monostatic environment, a bistatic radar structure is assumed, which is a comprehensive environment, because the radar of the monostatic environment is the same as the movement of the transmitter and the receiver in the same direction in the same position in a bistatic radar environment. Array (Uniform Linear Array (ULA) structure.

의 거리에 존재하는 신호는 송신단과 수신단을 초점으로 갖고 거리가

인 타원에 존재하는 클러터로부터 간섭을 받게 된다. 송신단과 수신단의 방향벡터가 각각

,

이고, 거리가

인 타원위에 존재하는 i번째 클러터 셀(cell)에 의해 발생하는 도플러 주파수

는 수학식 1과 같이 표현 할 수 있다.Referring to Figure 1,

The distance between the transmitter and the receiver is the focus.

Interference from clutter in the ellipse. The direction vectors of the transmitting and receiving ends are

,

And the distance is

The Doppler frequency generated by the ith clutter cell on the ellipse

Can be expressed by Equation (1).

는 내적을 의미하고,

와

는 송신단과 수신단이 i번째 클러터 셀과 이루는 단위 벡터를 각각 나타내며,

는 파장을 나타낸다. 한편 모노스태틱 (환경) 레이더가 측방 관측 어레이(Sidelooking array)구조를 갖는 경우 도플러 주파수(

)는 다음 수학식 2와 같다.here

Is the inner product,

Wow

Represents a unit vector formed by the transmitting end and the receiving end with the i-th clutter cell,

Indicates a wavelength. On the other hand, if the mono static (environment) radar has a sidelooking array structure, the Doppler frequency

Is expressed by the following equation (2).

은 수신단의 속도이고,

와

은 레이더 수신단의 보어사이트(boresight)와 i번째 클러터 셀이 이루는 고도(elevation)각과 방위(azimuth) 각을 각각 나타낸다. 이에 따른 수신단의 공간(Spatial) 주파수

는 수학식 3과 같다.here

Is the speed of the receiving end,

Wow

Represents the elevation angle and the azimuth angle between the boresight of the radar receiver and the i-th clutter cell, respectively. The spatial frequency of the receiving end

Is expressed by Equation (3).

)는 다음 수학식 4와 같이 표현할 수 있다. Where d is the spacing between the elements of the array. From Equations (2) and (3), it can be seen that a nonstationary clutter having a non-linear structure of Angle-Doppler occurs due to the movement of the transmitting end. In this case, the clutter signal (

) Can be expressed by the following equation (4).

는 Iso-range를 이루는 클러터의 셀의 개수이고

는 클러터의 크기(amplitude)이며

와

는 i번째 클러터 셀에 대한 N차원 공간(Spatial) 조향 벡터와 M차원 도플러 조향 벡터를 각각 나타내고, 아래의 수학식 5 및 수학식 6과 같이 표현된다.here

Is the number of clutter cells in the Iso-range

Is the amplitude of the clutter

Wow

Represents an N-dimensional space steering vector and an M-dimensional Doppler steering vector for an i-th clutter cell, and is expressed by Equation (5) and Equation (6) below.

와

는 i번째 클러터 셀에 대한 정규화된 공간 주파수와 정규화된 도플러 주파수이다. 상기 수학식 4로부터 수신단에 최종적으로 수신되는 클러터 신호(

)는

개의 클러터 셀들에 의해 합쳐진 신호가 된다. 비록 클러터의 크기(

)는 모르지만 송신단과 수신단의 운용정보를 알고 있다면 수학식 4로부터 이론적인 클러터의 구조를 예측 할 수 있다. 특정 range bin(샘플링된 수신신호)에서 클러터에 의해 발생하는 비정상 클러터를 얻기 위해서 모든

개의 클러터 셀들의 크기(amplitude)를

이라고 하면, 이에 해당하는 공분산 행렬

은 수학식 7과 같이 나타낼 수 있다. Where T is a Transpose operation,

Wow

Is the normalized spatial frequency and the normalized Doppler frequency for the ith clutter cell. From Equation (4) above, Clutter signal (

)

Lt; RTI ID = 0.0 > clutter < / RTI > Although the size of the clutter

), But if the operating information of the transmitting end and the receiving end is known, the theoretical clutter structure can be predicted from Equation (4). To obtain the abnormal clutter that is caused by the clutter in a certain range bin (sampled received signal)

The amplitude of the clutter cells is

, The corresponding covariance matrix

Can be expressed by Equation (7).

Where H represents a Hermitian operation.

을 추정할 수 있다. 상기 2차 데이터는 실제로 송신기 및 수신기의 움직임과 속도에 의해 발생하는 데이터이다. 따라서, 실제 환경에서 최대우도 추정기법을 이용할 경우 2차 데이터는 2×어레이 수×펄스 수에 해당하는 것으로 그 양으로 방대하기 때문에, 본 발명은 적은 양의 2차 데이터로부터 차원 저감화 (Rank-Reduced) 기법이나 Forward-Backward smoothing 기법을 이용하여 공분산 행렬을

을 추정할 수 있다. 이렇게 추정한

은 테스트 셀의 2차 데이터들로부터 이산(Discrete) 클러터나 클러터의 크기(amplitude)의 변화 정보를 포함하게 된다. On the other hand, when testing whether there is a target in the test cell while windowing the sampled received signal, the left and right data of the test cell, that is, the covariance matrix from the small amount of secondary data,

Can be estimated. The secondary data is actually data generated by the movement and speed of the transmitter and the receiver. Therefore, when the maximum likelihood estimation technique is used in the actual environment, the secondary data corresponds to the number of 2 × arrays × the number of pulses. Therefore, the present invention is applicable to Rank-Reduced ) Technique or a forward-backward smoothing technique.

Can be estimated. This presumed

From the secondary data of the test cell, information on the amplitude of the discrete clutter or clutter.

와

은 수학식 8와 같이 적절히 가중치

와

이 곱해진 후 합성되어 Nonstationary한 특성과 Heterogeneous 클러터를 모두 포함하는 클러터의 최적 공분산 행렬

을 얻을 수 있다. The covariance matrix of the obtained clutter

Wow

Lt; RTI ID = 0.0 > (8) < / RTI &

Wow

The optimal covariance matrix of the clutter, including both nonstationary characteristics and heterogeneous clusters,

Can be obtained.

와

는 대각행렬(diagonal matrix)로서 수학식 9와 같이 NM개의 diagonal term으로 이루어져 있다.Here,

Wow

Is a diagonal matrix and is composed of NM diagonal terms as shown in Equation (9).

)은 최적 필터(Optimal filter)

에 적용되어 클러터를 수신신호로부터 억제할 수 있다. 상기 가중치

와

를 얻기 위해 평균이 0이고 분산이

인 백색 가우시안 노이즈가 포함된 NM차원을 갖는 공분산 행렬(

)에 대한 최적 필터를 고유 분석(eigen-analysis) 형식으로 표현하면 수학식 10과 같다.The combined optimal covariance matrix (

) Is the optimal filter (Optimal filter)

So that the clutter can be suppressed from the received signal. The weight

Wow

The average is 0 and the variance is

A covariance matrix with NM dimensions with white Gaussian noise (

) Can be expressed by an eigen-analysis format as shown in Equation 10.

는 클러터 공분산 행렬(

)의

번째 고유값(eigen-value)이고,

은

에 해당하는 고유 벡터(eigen-vector)이며,

이다. 수학식 10으로부터 고유값(eigen-value)이

경우에 해당하는 신호는 제거되고

인 신호에 대해서는 거의 제거되지 않는 것을 확인할 수 있다. 그리고 클러터를 이루는 차수(Dimension)가 K일 때 (NM-K)개에 해당되는 공분산 행렬(

)의 고유값은

와 같다. 그러므로 클러터의 공분산 행렬(

)의 고유값(eigen-values) 대비 분산(

)값이 클수록 최적 필터(Optimal filter)

에 의해 클러터 제거 성능이 향상됨을 알 수 있다.here

Is the clutter covariance matrix (

)of

Eigen-value < / RTI >

silver

Eigen-vector < / RTI > corresponding to < RTI ID =

to be. From equation (10), the eigen-value

The corresponding signal is removed

It can be confirmed that almost no signal is removed. And the covariance matrix corresponding to (NM-K) when the dimension constituting the clutter is K (

) Is the unique value of

. Therefore, the covariance matrix of clutter (

Eigen-values versus variance (

) The larger the value, the more optimal filter (Optimal filter)

The clutter removal performance is improved.

)을 고유 분해(eigen-decomposition)하면 다음 수학식 11과 같이 표현할 수 있다.Clutter covariance matrix (

Eigen-decomposition can be expressed by the following equation (11).

와

은 클러터 공분산 행렬

와

의 평균 파워가 같도록 만들기 위한 정규화된 인수(Normalized factor)이다. 따라서 가중치

와

를 얻는 조건은 다음 수학식 12와 같다.here

Wow

Clutter Covariance Matrix

Wow

Is a normalized factor for making the average powers of the two equal. Therefore,

Wow

Is given by the following equation (12).

와

는 수학식 13과 같다.Where W is a constant. The weights satisfying the expression (12)

Wow

Is expressed by Equation (13).

는 수학식 12의 조건에 의한 water level이다. 결과적으로 수신부에서는 클러터의 공분산 행렬

와

을 구하고 수학식 13으로부터 얻은 가중치를 적용시켜 최적의 공분산 행렬(

)를 합성할 수 있고, 이를 최적 필터(Optimal filter)에 적용시킴으로써 비정상 클러터와 이종 클러터 성분들을 동시에 제거할 수 있다. here

Is the water level according to the condition of equation (12). As a result, in the receiver,

Wow

And the weight obtained from the equation (13) is applied to obtain an optimal covariance matrix

) Can be synthesized and applied to an optimal filter to remove both the abnormal clutter and the heterogeneous clutter components simultaneously.

2 is a block diagram of a receiving apparatus of a radar system for processing a signal received at a receiving end.

을 추정하는 제1공분산 추정부(100)와, 수신 신호에서 테스트 셀(Test cell)을 제외한 2차(Secondary) 데이터를 이용하여 클러터의 공분산 행렬

을 추정하는 제2공분산 추정부(200)와, 상기 제1,제2공분산 추정부(100, 200)에서 추정된 공분산 행렬

,

에 가중치

와

를 적용한 후 합성하여 최종 공분산 행렬(

)를 구하는 공분산 합성부(300)와, 상기 공분산 합성부(300)에서 합성된 공분산 행렬(

)에 최적 필터(Optimal filter)를 적용시켜 비정상 클러터와 이종 클러터 성분들을 제거하는 클러터 신호 억제부(400)를 포함한다. Referring to FIG. 2, a receiving apparatus of a radar system receives operation information of a transmitter and a receiver corresponding to a test cell from an operation information providing unit, and receives a covariance matrix

A first covariance estimation unit 100 for estimating a covariance matrix of the clutter using the second data excluding the test cell from the received signal,

A second covariance estimator 200 for estimating a covariance matrix 200 estimated by the first and second covariance estimators 100 and 200,

,

Weight

Wow

And then the final covariance matrix (

) Obtained by the covariance combining unit 300 and a covariance matrix

And a clutter signal suppression unit 400 for removing the abnormal clutter and the heterogeneous clutter components by applying an optimal filter to the clutter signal.

3 is an Angle-Doppler spectrum of a received signal including a target signal and a clutter signal, and shows a 3 dB loss of the target power as a threshold value.

Referring to FIG. 3, the target is located at a point where the normalized angle and the normalized Doppler are zero (indicated by a circle), and the other part shows that the clutter signal exists.

FIG. 4 is an Angle-Doppler spectrum after the clutter signal is removed according to an embodiment of the present invention. FIG. 4 shows that the clutter signal is removed while leaving only the target signal.

,

을 이용하여 테스트 셀의 비정상 클러터와 이종 클러터를 제거하기 위한 합성된 공분산 행렬(

)을 구한다. 따라서, 본 특허에서 제안된 방식으로 합성된 공분산 행렬 (

)은 비정상 클러터와 이종 클러터 성분들의 정보를 모두 포함하고 있어서 최적의 필터(

)에 적용될 수 있으며 클터터 억제 성능을 높일 수 있고 이에 따른 표적의 탐지 성능도 향상 시킬 수 있는 장점이 있다. As described above, according to the present invention, as a space-time adaptive processing for suppressing a clutter, a covariance matrix obtained from preliminary information on a moving path and a velocity of a transmitting end and a receiving end of a radar,

,

And the combined covariance matrix for removing the heterogeneous clutter from the test cell (

). Thus, the covariance matrix (< RTI ID = 0.0 >

) Contains both information of abnormal clutter and heterogeneous clutter components, so that the optimal filter

), It is possible to increase the suppression performance of the clutter and to improve the detection performance of the target.

The present invention described above can be embodied as computer-readable codes on a medium on which a program is recorded. The computer readable medium includes all kinds of recording devices in which data that can be read by a computer system is stored. In addition, the computer may include a control unit. Accordingly, the above description should not be construed in a limiting sense in all respects and should be considered illustrative. The scope of the present invention should be determined by rational interpretation of the appended claims, and all changes within the scope of equivalents of the present invention are included in the scope of the present invention.

100: first covariance estimator 200: second covariance estimator
300: covariance matrix synthesis unit 400:

Claims (5)

Calculating a first covariance matrix using operation information of a transmitter and a receiver corresponding to a test cell of a received signal;
Estimating a second covariance matrix using secondary data that is left and right data of a test cell in a received signal;
Multiplying the first and second covariance matrices by weights, respectively, to obtain a final covariance matrix including both a nonstationary clutter component and a heterogeneous clutter component; And
And applying a filter to the final covariance matrix to remove both the abnormal clutter component and the heterogeneous clutter component at the same time. 2. The method according to claim 1,
And a moving path and a velocity of the transmitter and the receiver. The method of claim 1, wherein the abnormal clutter component
Wherein the Angle-Doppler includes a clutter having a non-linear relationship due to the movement of the transmitter. The method of claim 1, wherein the heterogeneous clutter component
And information on a size change of a discrete clutter or a clutter from the secondary data of the test cell is included in the estimated covariance matrix of the radar system. 2. The method of claim 1, wherein obtaining the final covariance matrix comprises:
Eigen-decomposing the first and second covariance matrices;
Obtaining first and second weights that maximize a variance value of eigen-values of the first and second covariance matrices; And
And multiplying the first and second weights by the first and second covariance matrixes, respectively, and combining the first and second weights.

Images