WO2019047210A1

WO2019047210A1 - Knowledge-based sparse recovery space-time adaptive processing method and system

Info

Publication number: WO2019047210A1
Application number: PCT/CN2017/101216
Authority: WO
Inventors: 阳召成; 汪小叶; 朱轶昂; 黄建军
Original assignee: 深圳大学
Priority date: 2017-09-11
Filing date: 2017-09-11
Publication date: 2019-03-14

Abstract

The present invention is applicable to the field of radar signal processing, and provided thereby is a knowledge-based sparse recovery space-time adaptive processing (STAP) method, comprising: constructing a space-time oriented dictionary according to an airborne radar and an angle-Doppler plane, and determining a clutter angle-Doppler image according to the space-time oriented dictionary (S101); carrying out combined estimation on an array amplitude and phase error of an antenna array of the airborne radar and the clutter angle-Doppler image (S102); calculating a clutter covariance matrix according to the combined estimation (S103); and constructing a space-time filter according to the clutter covariance matrix, and carrying out clutter suppression by means of the space-time filter (S104). The described method solves the problems of performance degradation and high computational complexity of sparse recovery STAP caused by the presence of an array error and improves the clutter suppression level and target detection ability of the radar system.

Description

Knowledge-based sparse recovery space-time adaptive processing method and system

Technical field

The invention belongs to the field of radar signal processing, and in particular relates to a knowledge-based sparse recovery space-time adaptive processing method and system.

Background technique

Space-time adaptive processing is the key technology to improve the performance of airborne radar to detect moving targets, but this technology faces the challenge of limited filter training samples, and the challenge is more severe in non-uniform clutter environments. In the past ten years, the space-time adaptive processing technology has achieved certain developments, such as the proposed reduced dimension STAP (space-time adaptive processing) method, and the reduced rank STAP method. , model-based STAP method, knowledge-aided STAP method, and so on.

With the theory of compressed sensing (CS), the STAP method based on sparse recovery has developed rapidly. The compressed-sensing STAP method can solve the problem of insufficient training samples mentioned above (such methods usually require 4-6 training samples to obtain satisfactory output performance), and thus have attracted extensive attention from scholars at home and abroad.

The power spectrum sparse recovery STAP method that has been disclosed and considered for array error is: [Z.Ma, Y.Liu, H.Meng and X.Wang, "Sparse recovery-based spacetime adaptive processing with array error self-calibration, "ELECTRONICSLETTERS, Vol.50, No. 13, pp. 952-954, June 2014.] proposed method and literature [Y.Zhu, Z.Yang and J.Huang, "Robust sparsity-based space-time adaptive processing Consider array gain/phase errors, "2016 IEEE 13th International Conference on Signal Processing (ICSP), Chengdu, 2016, pp. 1624-1628.] The proposed method (abbreviated as OMP-LS-STAP), but these two Methods need to be at the entire angle when estimating the clutter space-time power spectrum - Searching on the Doppler plane, thus resulting in higher computational complexity.

Summary of the invention

The technical problem to be solved by the present invention is to provide a knowledge-based sparse recovery space-time adaptive processing method and system, which aims to solve the problem of performance degradation and high computational complexity due to array errors in solving the sparse recovery STAP in the prior art. The problem.

The present invention is implemented in this way, a knowledge-based sparse recovery space-time adaptive processing method, comprising:

A space-time oriented dictionary is constructed according to the airborne radar and the angle-Doppler plane, and the clutter angle-Doppler image is determined according to the space-time oriented dictionary;

Performing joint estimation on the amplitude and phase error of the array of the antenna array of the airborne radar and the clutter angle-Doppler image;

Calculating a clutter covariance matrix according to the joint estimate;

A space-time filter is constructed according to the clutter covariance matrix, and the spurious suppression is performed by the space-time filter.

Further, a plurality of clutter angle-Doppler units are distributed along the clutter ridge line on the angle-Doppler plane, and the space-time oriented dictionary is constructed according to the airborne radar and the clutter angle-Doppler unit. And determining the clutter angle based on the - space time oriented dictionary - the Doppler image includes:

Calculating a folding coefficient of the clutter ridge by a prior knowledge of the airborne radar, the prior knowledge including a moving speed of the airborne radar, a moving direction, and a pointing of the antenna array;

Determining a distribution of the clutter angle-Doppler unit according to the folding coefficient, and selecting the clutter angle-Doppler unit and its adjacent angle from the angle-Doppler plane according to the distribution - Doppler unit;

Constructing an abstract matrix according to the position of the clutter angle-Doppler unit and its adjacent angle-Doppler unit;

Performing dimension reduction on the space-time oriented dictionary according to the abstract matrix, and according to the space-time guide after dimension reduction Determine the clutter angle - Doppler image to the dictionary.

Further, determining, according to the folding coefficient, a distribution of the clutter angle-Doppler unit, and selecting the clutter angle-Dopp from the angle-Doppler plane according to the distribution condition The Le unit and its adjacent angle-Doppler units include:

Determining a preset size rectangular window centering on each of the clutter ridges on the chaotic ridge line on the angle-Doppler plane;

The angle-Doppler unit selected by the rectangular window is a clutter angle-Doppler unit to be determined;

And constructing an abstract matrix according to the position of the clutter angle-Doppler unit and its adjacent angle-Doppler unit includes:

The abstract matrix is constructed according to the position of the clutter angle to be determined, where the Doppler unit is located.

Further, the calculating the clutter covariance matrix according to the joint estimation comprises:

Calculating a space-time power spectrum of the clutter according to the joint estimate;

The clutter covariance matrix is calculated according to the space time power spectrum.

The invention also provides a knowledge-based sparse recovery space-time adaptive processing system, comprising:

a dictionary building unit, configured to construct a space-time guiding dictionary according to the airborne radar and the angle-Doppler plane, and determine a clutter angle-Doppler image according to the space-time guiding dictionary;

a joint estimating unit, configured to perform joint estimation on an array amplitude and phase error of the antenna array of the airborne radar and the clutter angle-Doppler image;

a matrix calculation unit, configured to calculate a clutter covariance matrix according to the joint estimate;

And a filter construction unit configured to construct a space-time filter according to the clutter covariance matrix, and perform clutter suppression by the space-time filter.

Further, the dictionary construction unit is specifically configured to:

Determining a distribution of the clutter angle-Doppler unit according to the folding coefficient, and selecting the clutter angle-Doppler unit and its adjacent angle from the angle-Doppler plane according to the distribution - Dopp Le unit

The space-time oriented dictionary is dimension-reduced according to the abstract matrix, and the clutter angle-Doppler image is determined according to the reduced-dimensional space-time oriented dictionary.

Further, the dictionary construction unit is further configured to:

Further, the matrix calculation unit is configured to:

Compared with the prior art, the present invention has the beneficial effects that the embodiment of the present invention constructs a space-time guiding dictionary according to the airborne radar and the angle-Doppler plane and determines the clutter angle-Doppler image, and the clutter angle - Amplitude and phase error of the array of Doppler and airborne radar antenna arrays are jointly estimated, the clutter covariance matrix is calculated according to the joint estimation, and the space-time filter is designed according to this, and the space-time filter is designed to perform clutter suppression. . The embodiment of the invention solves the problem that the performance of the sparse recovery STAP is degraded due to the array error and the computational complexity is high, and the radar system suppression level and the target detection capability are improved.

DRAWINGS

1 is a flowchart of a knowledge-based sparse recovery space-time adaptive processing method according to an embodiment of the present invention;

2 is a schematic diagram of selection of a clutter angle-Doppler unit according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of the conventional power spectrum sparse recovery STAP algorithm and KA-OMP-LS based on OMP when different rectangular amplitude and phase errors are used when the prior knowledge is accurately known and the rectangular window is ρ _w = 3×3. - STAP output SINR and target Doppler frequency relationship graph;

4 is an embodiment of the present invention, when the prior knowledge is precisely known, the rectangular window is ρ _w = 3 × 3, and the array phase error is |G / P | _max = 50% / 45 °, OMP-LS- The relationship between the STAP algorithm and the output SINR of the KA-OMP-LS-STAP and the target Doppler frequency;

FIG. 5 is a schematic diagram showing the output SINR of the KA-OMP-LS-STAP under different rectangular window sizes when the prior art knowledge is precisely known, the amplitude and phase error of the array is |G/P| _max =25%/3.5°. a Doppler frequency relationship graph different from the target;

FIG. 6 is a schematic diagram of the output SINR of the KA-OMP-LS-STAP under different rectangular window sizes when the prior art knowledge is precisely known, the amplitude and phase error of the array is |G/P| _max =50%/45°. a Doppler frequency relationship graph different from the target;

FIG. 7 is a diagram showing the output SINR of the KA-OMP-LS-STAP is different from the target when the amplitude and phase error of the array is |G/P| _max =25%/3.5° and the yaw angle has a measurement error according to an embodiment of the present invention. Puller frequency relationship curve;

FIG. 8 is a diagram showing the output SINR of the KA-OMP-LS-STAP is different from the target when the amplitude error of the array is |G/P| _max =50%/45° and the yaw angle has measurement error according to the embodiment of the present invention. Puller frequency relationship curve;

FIG. 9 is a diagram showing the output SINR of the KA-OMP-LS-STAP and the target when the amplitude error of the array is |G/P| _max =25%/3.5° and the measurement error exists in the platform speed according to the embodiment of the present invention. Le frequency relationship curve;

FIG. 10 is a diagram showing the output SINR of the KA-OMP-LS-STAP and the target when the amplitude error of the array is |G/P| _max =50%/45° and the measurement error exists in the platform speed according to the embodiment of the present invention. Le frequency relationship curve;

11 is an output SINR of a KA-OMP-LS-STAP when the amplitude error of the array is |G/P| _max =25%/3.5° and the measurement error exists in the platform velocity and the yaw angle provided by the embodiment of the present invention. Target different Doppler frequency relationship graph;

FIG. 12 is a diagram showing the output SINR of the KA-OMP-LS-STAP when the amplitude error of the array is |G/P| _max =50%/45° and the measurement error exists in the platform velocity and the yaw angle provided by the embodiment of the present invention. Target different Doppler frequency relationship graph;

13 is a schematic diagram of output power of different algorithms along a distance unit under MountainTop data according to an embodiment of the present invention;

FIG. 14 is a schematic diagram of calculations of three algorithms for comparing KA-OMP-STAP, KA-OMP-LS-STAP, and OMP-LS-STAP by computer simulation experiments according to an embodiment of the present invention.

FIG. 15 is a schematic structural diagram of a knowledge-based sparse recovery space-time adaptive processing system according to an embodiment of the present invention.

Detailed ways

The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

FIG. 1 is a schematic diagram of a knowledge-based sparse recovery space-time adaptive processing method according to an embodiment of the present invention, including:

S101, constructing a space-time oriented dictionary according to the airborne radar and the angle-Doppler plane, and determining a clutter angle-Doppler image according to the space-time oriented dictionary;

S102. Perform joint estimation on an array amplitude phase error of the antenna array of the airborne radar and the clutter angle-Doppler image. In this step, the values obtained after the joint estimation are the array amplitude and phase error and the clutter angle-Doppler image.

S103. Calculate a clutter covariance matrix according to the joint estimation.

S104. Construct a space-time filter according to the clutter covariance matrix, and perform clutter suppression by the space-time filter.

Specifically, the embodiment of the present invention calculates the folding coefficient of the clutter ridge by using the prior knowledge of the motion speed, the moving direction of the airborne radar, and the orientation of the antenna array, and determines the true clutter angle - the Doppler unit is at the entire angle. - the distribution in the Doppler plane, according to which the true clutter angle - Doppler unit and The space-time steering vector corresponding to the adjacent unit is selected from the complete space-time guiding dictionary corresponding to the entire angle-Doppler plane, thereby obtaining a reduced space-time guiding dictionary, and then performing in the reduced space-time guiding dictionary. Sparse recovery of clutter and correction of array amplitude and phase error. For convenience of the following description, the processing method provided by the embodiment of the present invention is simply referred to as KA-OMP-LS-STAP.

In the embodiment of the present invention, it is assumed that the antenna of a pulse-Doppler positive side view airborne radar is a uniform linear array comprising M receiving array elements, and the airborne radar transmits N pulses in a coherent processing unit. In the ideal case where the array does not have amplitude and phase errors. The NM×1 dimension free-time snapshot with no target can be expressed as:

x=x _c +n=Φγ+n (1)

Where x _c is the space-time snapshot corresponding to the clutter, n is the NM × 1 dimension receiver thermal noise, N _d M _s × 1 dimension

For the complex amplitude (or angle-Doppler image) corresponding to the clutter in the space-timed dictionary, matrix

A perfect ideal space-time guidance dictionary for NM×N _d M _s dimension without array error, (·) ^T is transpose operation, NM×1 dimensional vector

For the ideal space-time steering vector, v _d (·) and v _s (·) are the time domain steering vector and the spatial domain steering vector, respectively, (f _d,i , f _s,k ) is the ith time domain grid point. With the kth spatial domain grid point (the entire space-time plane is divided into N _d N _s (N _d N _s >>NM) grids, N _s and N _d are along the spatial frequency axis and time / Dopp respectively The number of grid points of the frequency axis).

Suppose α _s =[α _s,1 ,α _s,2 ,...,α _s,M ] ^T is the amplitude and phase error of the antenna array, and α _s,i is the amplitude and phase error corresponding to the ith array element. The space-time steering vector under the amplitude and phase error of the array can be expressed as

make

Where I _N is an identity matrix of N × N dimensions, and diag(α _s ) is a diagonal matrix after diagonalization of α _s ,

For Kronecker product, which is Hadamard product, the complete space-time guidance dictionary under array error can be expressed as ΓΦ. At this time, the space-free snapshot received without the target under the array error is:

x=ΓΦγ+n (2)

definition

among them

1 _N is a column vector of N x 1 dimension and all elements are all 1. Then formula (2) can be expressed as:

x=Qα _s +n;

The specific implementation process of the embodiment of the present invention is further described below:

Step 1: Construct a space-time oriented dictionary

The angle of the clutter - the Doppler element is distributed along the clutter ridge over the entire angle-Doppler plane. The folding coefficient β of the clutter ridge can be calculated by a priori knowledge of the speed of the airborne radar, the direction of motion and the orientation of the antenna array:

Where f _d represents the clutter Doppler frequency, f _s pulse repetition frequency, v _p is the motion velocity of the airborne radar platform, T _r represents the pulse repetition interval, and d is the antenna array element spacing. The distribution of the clutter angle-Doppler cell is then determined, according to which the angle-Doppler cell in which the clutter is located is selected from the angle-Doppler plane. In this embodiment, considering that there is uncertainty in the prior knowledge, and in order for the selected unit to contain all the real clutter angle-Doppler units, it is necessary to select the units adjacent to the clutter ridges together. come out. The embodiment of the present invention performs the selection by the method shown in FIG. 2, which comprises: determining, in the angle-Doppler plane, a rectangular window of appropriate size centering on the angle-Doppler unit where each clutter ridge is located, Any cell selected by the rectangular window acts as a possible clutter angle-Doppler unit.

Construct a point using the angle chosen above - the location of the Doppler unit

Dimension sampling matrix T:

Where t _i is a column vector of N _d N _s ×1 dimension, only one of all elements in t _i is 1 and the rest are all 0. Then the space-time oriented dictionary after dimension reduction can be obtained by the following formula

Then, the antenna array receives the space-time snapshot without the target, which can be expressed as:

among them,

for

Dimensional clutter angle - Doppler image.

Step 2: Joint Estimation of Array Amplitude and Phase Error and Clutter Angle-Doppler Image

Joint estimation of array amplitude and phase error and clutter angle-Doppler image optimization problems can be constructed as:

Where l is the lth sample of L snapshots,

Representing the real part of the plural,

(

For any constant scalar quantity),

For the angle-Doppler image of the c-th iteration estimation clutter, the optimization problem can be expressed as:

The OMP algorithm based on multiple snapshot samples can be used to solve the optimization problem.

For the p-th iteration estimation array amplitude and phase error, the optimization problem can be expressed as:

Solving is available:

among them

y _l,m and x _l,m are vectors respectively

And the mth element in x _l .

Step 3: Estimating the clutter covariance matrix

Through the joint estimation described above, the clutter angle of the L snapshot samples - the Doppler image A and the array amplitude phase error α _s (ie,

After calculating the space-time power spectrum of the clutter:

among them

Calculate the clutter covariance matrix:

Step 4: Design a space-time filter

The space-time filter weight vector is designed to:

among them

The beneficial effects of the embodiments of the present invention are further illustrated by several simulation experiments:

The radar simulation and scene parameters involved in the simulation experiment of the embodiment of the present invention are shown in Table 1. It is assumed that 361 clutter scattering points are equally spaced on the distance ring elements of interest, and their amplitudes are subject to a Gaussian distribution. In this experimental example, the amplitude and phase errors of the array are subject to random uniform distribution, and the internal noise of the receiver obeys a Gaussian distribution, and its power is 1 for a single pulse and a single channel. Experiments were carried out to simulate the amplitude and phase errors of six different arrays: |G/P| _max =0/0°, |G/P| _max =1.5%/0.5°, |G/P| _max = 5% / 1.5 °, |G / P | _max = 15% / 2.5 °, |G / P | _max = 25% / 3.5 °, | G / P | _max = 50% / 45 °. The number of snapshot samples used in this simulation experiment is L=20, the space-time oriented dictionary N _d N _s =7N×7M, the initial value of the maximum number of iterations is set to k=90, and the maximum number of OMP and LS alternate iterations is 70. The threshold of iterative decision is ζ=10 ^-2 , and all results are subjected to 100 Monte Carlo computer simulation experiments.

Table 1

The first experiment examines the performance of the output SINR (Signal to Interference plus Noise Ratio) of the embodiment of the present invention when the rectangular window is ρ _w = 3×3, as shown in the figure. 3. Figure 4 shows. In order to better demonstrate the effectiveness of the embodiments of the present invention, this experiment compares its performance with the performance of OMP-LS-STAP and the OMP based traditional power spectrum sparse recovery STAP algorithm (abbreviated as: OMP-STAP). . The experimental results of FIG. 3 show that the embodiment of the present invention not only has good robustness when the phase amplitude error of the array is small, but also has good robustness when the phase amplitude error of the array is large. When the amplitude and phase error of the array is |G/P| _max =50%/45°, the performance of the embodiment of the present invention is at least 40 dB better than that of the conventional power spectrum sparse recovery STAP, and almost reaches a small array amplitude and phase error. The equivalent output performance. The simulation results of FIG. 4 show that the processing method provided by the embodiment of the present invention is more robust than the OMP-LS-STAP algorithm when the amplitude error of the array is large. The performance of the embodiment of the present invention is at least 25 dB better than the OMP-LS-STAP algorithm when the array amplitude error is |G/P| _max = 50%/45°. It can be seen that the embodiment of the present invention reduces the dimension of the space-time oriented dictionary by using prior knowledge, and reduces the search space of the sparse recovery algorithm, thereby improving the accuracy of joint estimation under larger array amplitude and phase errors.

The second experiment examines that when the prior knowledge is precisely known, the array amplitude error is |G/P| _max = 25%/3.5°, |G/P| _max = 50%/45°, the embodiment of the present invention The output performance under different rectangular window sizes is shown in Figure 5 and Figure 6. The experimental results show that when the rectangular window is ρ _w = 3×3, the output performance curve of the embodiment of the present invention has the smallest notch, and the slow moving target detection performance is optimal. It is worth noting that the performance of the embodiment of the present invention under the rectangular window ρ _w =1×1 is not discussed in the experiment because the prior knowledge obtained is not completely accurate, if only the clutter ridge line is selected. Corresponding angle-Doppler unit, the performance of the algorithm in the actual application will inevitably be seriously reduced. Therefore, in the practical application of the embodiment of the present invention, a rectangular window of ρ _w = 3 × 3 or larger can be selected to optimize the output performance.

The third experiment examined the amplitude and phase error of the array as |G/P| _max =25%/3.5°, |G/P| _max =50%/45°, and the output performance of the embodiment of the present invention when the prior knowledge is inaccurate, 7 through 12, assume experimental airborne platform velocity v _up measurement error and measurement error of the yaw angle φ _u satisfy the uniformly distributed random. The experimental results show that the embodiment of the present invention can also exhibit better robust performance under a rectangular window of suitable size when the prior knowledge is inaccurate. When the deviation between the prior knowledge and the true value is larger, the rectangular window is also larger, which can ensure that the rectangular window can select all the real clutter angle-Doppler units, thereby improving the performance of the algorithm. As shown in Figs. 8 and 9, when the prior knowledge has a small error, the performance of the present invention is optimal in the rectangular window ρ _w = 3 × 3. When the error of the prior knowledge increases, as shown in Fig. 9 to Fig. 12, the platform motion velocity measurement error is v _up ~ [-10, 10] m / s, and the yaw angle measurement error φ _u ~ [-3 °, At 3°], the performance of the embodiment of the present invention exhibits poor robustness in the rectangular window ρ _w = 3 × 3, and exhibits better robustness in the rectangular window ρ _w = 7 × 7.

The fourth experiment verified the effectiveness of the embodiments of the present invention using MountainTop data t38pre01v1.mat, CPI6. In this experiment, N=16, M=14, the pulse repetition frequency PRF is 625 Hz, and the instantaneous bandwidth after pulse compression is 500 kHz. The measured data contains 403 independent distance dimension sampling samples. The clutter azimuth is 245° and the target azimuth is 275°, both of which have a Doppler frequency of 156 Hz. In this experiment, the number of samples used to train the space-time filter is 18. In order to prevent target cancellation, 6 samples adjacent to the distance unit to be detected are used as protection units. For the sparse recovery algorithm, the sparsity is set to 20, the number of alternate iterations is 20, the space-time oriented dictionary N _d N _s = 4N × 4M, and the rectangular window size is ρ _w = 3 × 3, and other parameters are the same as in Table 1.

Figure 13 shows the output power of different algorithms over a distance of 145-165 km (a distance unit with a target of 154 km). Table 2 shows the difference between the output power of each algorithm at the target distance unit and the second high output power of the adjacent distance unit. The results of FIG. 13 and Table 2 show that the clutter suppression performance of the processing method provided by the embodiment of the present invention is slightly better than the clutter suppression performance of the OMP-LS-STAP algorithm compared to the OMP-LS-STAP algorithm. Moreover, the difference between the output power at the target distance unit and the second high output power of the adjacent distance unit is larger in the embodiment of the present invention, and thus is more advantageous for subsequent constant false alarm processing.

Table 2

The fifth experiment discusses the computational complexity of the embodiments of the invention (KA-OMP-LS-STAP) and OMP-LS-STAP. Because the two algorithms use the same algorithm for calculating the space-time filter weight vector and estimating the amplitude and phase error of the array, and the maximum number of iterations is the same in the iterative solution, the two algorithms are discussed in estimating the clutter angle-Doppler. The computational complexity of the image time is equivalent to discussing the computational complexity of the two algorithms.

The computational complexity of OMP-LS-STAP in estimating the clutter angle-Doppler image is O(NMN _d N _s ). The computational complexity of KA-OMP-LS-STAP in estimating clutter angle-Doppler images and the knowledge-based OMP power spectrum sparse recovery STAP algorithm (abbreviated as KA-OMP-STAP, the algorithm uncorrected array amplitude The computational complexity of the error)

Since the angle of the clutter-Doppler element is sparse in the angle-Doppler plane, there is

(ρ _w is the size of the rectangular window), so the computational complexity of KA-OMP-LS-STAP is much smaller than the computational complexity of the OMP-LS-STAP algorithm. Since KA-OMP-STAP does not use the LS algorithm to estimate the amplitude and phase error of the array, the computational complexity of the algorithm is smaller than that of KA-OMP-LS-STAP. Therefore, the computational relationship of these three algorithms is: KA-OMP-STAP<KA-OMP-LS-STAP<<OMP-LS-STAP.

In addition, the OMP-LS-STAP algorithm and the literature [Z.Ma, Y. Liu, H. Meng and X. Wang, "Sparse recovery-based spacetime adaptive processing with array error self-calibration," ELECTRONICS LETTERS, Vol. 50, No.13, pp.952-954, June2014.] The proposed algorithm (abbreviated as IAD-SR-STAP) uses the sparse recovery algorithm and LS alternate iteration to jointly estimate the clutter power spectrum and the array amplitude and phase error. The sparse recovery algorithm used by the former is OMP, while the sparse recovery algorithm adopted by the latter is LASSO. The computational complexity of OMP is O(NMN _d N _s ), and the computational complexity of LASSO is O((N _d N _s ) ³ ). The comparison shows that the computational complexity of the OMP-LS-STAP algorithm is lower than the computational complexity of the IAD-SR-STAP algorithm.

In summary: the calculated amount relationship of the above algorithm is KA-OMP-STAP<KA-OMP-LS-STAP<<OMP-LS-STAP<IAD-SR-STAP.

The calculations of the three algorithms KA-OMP-STAP, KA-OMP-LS-STAP and OMP-LS-STAP are compared by computer simulation experiments, as shown in Fig. 14. The algorithm runtime environment is MATLAB (R2014a, 8.3.0.532), Intel(R) Core(TM) i7-4790CPU@3.6GHz, 16.0GB (RAM), Windows10 (64bit). Space-time oriented dictionary N _d N _s =7N×7M, rectangular window is ρ _w =3×3, maximum iteration number in OMP is k=90, maximum number of OMP and LS alternate iterations 70, iterative decision threshold For ζ=10 ^-2 . Other parameters are shown in Table 1. All results were averaged over 100 Monte Carlo simulations. The results in the figure show that the computational complexity of the present invention is much smaller than the computational complexity of the OMP-LS-STAP algorithm, but slightly higher than the computational complexity of the KA-OMP-STAP algorithm. It can be seen that the use of prior knowledge does improve the computational speed of the OMP-LS-STAP algorithm.

FIG. 15 is a diagram showing a knowledge-based sparse recovery space-time adaptive processing system according to an embodiment of the present invention, including:

a dictionary construction unit 201, configured to construct a space-time guidance dictionary according to the airborne radar and the angle-Doppler plane, and determine a clutter angle-Doppler image according to the space-time guidance dictionary;

The joint estimating unit 202 is configured to perform joint estimation on an array amplitude and phase error of the antenna array of the airborne radar and the clutter angle-Doppler image;

a matrix calculation unit 203, configured to calculate a clutter covariance matrix according to the joint estimation;

The filter construction unit 204 is configured to construct a space-time filter according to the clutter covariance matrix, and perform clutter suppression by the space-time filter.

Further, the dictionary construction unit 201 is specifically configured to:

Determining a distribution of the clutter angle-Doppler unit according to the folding coefficient, and selecting the clutter angle-Doppler unit and its neighboring from the angle-Doppler plane according to the distribution condition of Angle-Doppler unit;

Further, the dictionary construction unit 201 is further configured to:

Further, the matrix calculation unit 203 is configured to:

The embodiment of the invention firstly uses the prior knowledge to implement the reduced space-time oriented dictionary, and then adopts Orthogonal Matching Pursuit (OMP, orthogonal matching tracking algorithm) and Least Square (LS, least squares) alternate iterative algorithm to realize the clutter angle - Joint estimation of the phase error between the Doppler image and the array, and then design an adaptive space-time filter to achieve clutter suppression.

The embodiment of the invention can be applied to the radar clutter suppression field of the motion platform to solve the problem that the performance of the sparse recovery STAP is degraded due to the array error and the computational complexity is high, and the radar system suppression level and the target detection capability are improved.

The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection of the present invention. Within the scope.

Claims

A knowledge-based sparse recovery space-time adaptive processing method, which comprises:

A space-time oriented dictionary is constructed according to the airborne radar and the angle-Doppler plane, and the clutter angle-Doppler image is determined according to the space-time oriented dictionary;

Performing joint estimation on the amplitude and phase error of the array of the antenna array of the airborne radar and the clutter angle-Doppler image;

Calculating a clutter covariance matrix according to the joint estimate;

A space-time filter is constructed according to the clutter covariance matrix, and the spurious suppression is performed by the space-time filter.
The processing method according to claim 1, wherein a plurality of clutter angle-Doppler units are distributed along the clutter ridge line on the angle-Doppler plane, and the airborne radar and the The wave angle-Doppler unit constructs a space-time oriented dictionary and determines the clutter angle according to the space-time oriented dictionary-Doppler image including:

Calculating a folding coefficient of the clutter ridge by a prior knowledge of the airborne radar, the prior knowledge including a moving speed of the airborne radar, a moving direction, and a pointing of the antenna array;

Determining a distribution of the clutter angle-Doppler unit according to the folding coefficient, and selecting the clutter angle-Doppler unit and its neighboring from the angle-Doppler plane according to the distribution condition Angle - Doppler unit;

Constructing an abstract matrix according to the position of the clutter angle-Doppler unit and its adjacent angle-Doppler unit;

The space-time oriented dictionary is dimension-reduced according to the abstract matrix, and the clutter angle-Doppler image is determined according to the reduced-dimensional space-time oriented dictionary.
The processing method according to claim 2, wherein said determining a distribution of said clutter angle-Doppler cells according to said folding coefficient, selecting from said angle-Doppler plane according to said distribution The clutter angle-Doppler unit and its adjacent angle-Doppler unit include:

Determining, on the angle-Doppler plane, centering on each clutter ridge on the clutter ridge line a rectangular window of preset size;

The angle-Doppler unit selected by the rectangular window is a clutter angle-Doppler unit to be determined;

And constructing an abstract matrix according to the position of the clutter angle-Doppler unit and its adjacent angle-Doppler unit includes:

The abstract matrix is constructed according to the position of the clutter angle to be determined, where the Doppler unit is located.
The processing method according to claim 1, wherein said calculating a clutter covariance matrix according to said joint estimation comprises:

Calculating a space-time power spectrum of the clutter according to the joint estimate;

The clutter covariance matrix is calculated according to the space time power spectrum.
A knowledge-based sparse recovery space-time adaptive processing system, comprising:

a dictionary building unit, configured to construct a space-time guiding dictionary according to the airborne radar and the angle-Doppler plane, and determine a clutter angle-Doppler image according to the space-time guiding dictionary;

a joint estimating unit, configured to perform joint estimation on an array amplitude and phase error of the antenna array of the airborne radar and the clutter angle-Doppler image;

a matrix calculation unit, configured to calculate a clutter covariance matrix according to the joint estimate;

And a filter construction unit configured to construct a space-time filter according to the clutter covariance matrix, and perform clutter suppression by the space-time filter.
The processing system according to claim 5, wherein the dictionary construction unit is specifically configured to:

Calculating a folding coefficient of the clutter ridge by a prior knowledge of the airborne radar, the prior knowledge including a moving speed of the airborne radar, a moving direction, and a pointing of the antenna array;

Determining a distribution of the clutter angle-Doppler unit according to the folding coefficient, and selecting the clutter angle-Doppler unit and its neighboring from the angle-Doppler plane according to the distribution condition Angle - Doppler unit;

Constructing an abstract matrix according to the position of the clutter angle-Doppler unit and its adjacent angle-Doppler unit;

The space-time oriented dictionary is dimension-reduced according to the abstract matrix, and the clutter angle-Doppler image is determined according to the reduced-dimensional space-time oriented dictionary.
The processing system according to claim 6, wherein the dictionary construction unit is further configured to:

Determining a preset size rectangular window centering on each of the clutter ridges on the chaotic ridge line on the angle-Doppler plane;

The angle-Doppler unit selected by the rectangular window is a clutter angle-Doppler unit to be determined;

The abstract matrix is constructed according to the position of the clutter angle to be determined, where the Doppler unit is located.
The processing system of claim 5 wherein said matrix computing unit is operative to:

Calculating a space-time power spectrum of the clutter according to the joint estimate;

The clutter covariance matrix is calculated according to the space time power spectrum.