WO2018045566A1

WO2018045566A1 - Random pulse doppler radar angle-doppler imaging method based on compressed sensing

Info

Publication number: WO2018045566A1
Application number: PCT/CN2016/098598
Authority: WO
Inventors: 阳召成; 全桂华; 黄建军; 黄敬雄
Original assignee: 深圳大学
Priority date: 2016-09-09
Filing date: 2016-09-09
Publication date: 2018-03-15

Abstract

A random pulse Doppler radar angle-Doppler imaging method based on compressed sensing comprises the following steps: S1, obtaining space-time compressed sampling data according to a radar system; S2, respectively performing spatial discretization processing on a Doppler frequency space and a spatial frequency in the space-time compressed sampling data, and obtaining a space-time oriented dictionary; and S3, estimating a space-time power spectrum according to the space-time oriented dictionary, and obtaining an angle-Doppler image containing a clutter and a target. The problem of target and clutter Doppler ambiguity is effectively resolved, the clutter suppression and target detection performance is improved; the number of pulses transmitted by a radar system is reduced, and other radar waveforms can be transmitted or multiple angles can be observed at the same time at a same pulse coherent processing interval, thereby effectively improving the time-dimension multiplexing capability of a radar; at the same number of pulses, the Doppler resolution capability can be effectively improved; the radar waveform low interception and capture capabilities and the interference resistance capability are stronger.

Description

Random Pulse Doppler Radar Angle-Doppler Imaging Method Based on Compressed Sensing

Technical field

The invention belongs to the field of radar signal processing, and more particularly to a random pulse Doppler radar angle-Doppler imaging method based on compressed sensing.

Background technique

Space-Time Adaptive Processing (STAP) is an important technique for clutter suppression and moving target detection in airborne radar systems. The traditional STAP method requires a high number of training samples, and it is difficult to obtain a large number of samples in a non-uniform, non-stationary clutter environment for designing a space-time filter. Such as Auxiliary Channel Processor (ACP), Principal Components (PC), Joint Domain Localized (JDL), Multistage Winer Filter (MWF) algorithm, auxiliary features A series of dimensionality reduction or descending rank algorithms, such as the Auxiliary Eigen Vector Processor (AEP), reduces the number of samples required to 2 times the dimension after dimensionality reduction or the number of wavelet ranks after 2 times the reduced rank. However, the number of training samples is still high relative to non-uniform clutter environments.

Recently, the sparse-based STAP algorithm proposed to improve the convergence of the system is based on the sparse property of the clutter power spectrum in the angle-Doppler domain, and the sparse recovery algorithm is used to reconstruct the clutter of the Doppler. The image is thus estimated to estimate the clutter covariance matrix, and then the space-time filter is designed to achieve the purpose of clutter suppression and target detection. Most existing sparse-based STAP algorithms are implemented in spatial and Doppler dimensions in a uniform sampling manner. However, uniform pulse repetition frequency (PRF) and array element array modes are inevitable. Problems such as Doppler or distance blur will eventually lead to system clutter suppression and target detection performance degradation. At the same time, the uniform pulse transmission mode Doppler radar emission waveform is easy to be intercepted, and the anti-interference ability is poor.

Summary of the invention

Aiming at the defects of the prior art, the object of the present invention is to provide a method for random pulse Doppler radar angle-Doppler imaging based on compressed sensing, aiming at solving the uniform pulse transmitting Doppler radar in the prior art. The transmit waveform is easily intercepted and has poor anti-interference ability, while improving the target processing capability in a single coherent processing time interval.

The invention provides a method for random pulse Doppler radar angle-Doppler imaging based on compressed sensing, comprising the following steps:

S1: obtaining space-time compressed sampling data based on a radar system;

S2: spatially discretizing the Doppler frequency space and the spatial frequency in the space-time compressed sample data, and obtaining a space-time oriented dictionary;

S3: Estimating the space-time power spectrum according to the space-time guiding dictionary, and obtaining an angle-Doppler image including the clutter and the target.

Further, in step S1, K complete extractions are randomly selected from M non-periodic pulse sequences in the radar system, and M non-periodic pulse moments satisfy the relationship T _m+1 =mT _r +εt _m , where K ≤ M, 0 ≤ ε < 1, t _m ～ B (0, T _r ), M is the number of pulse sequences in a single coherent processing time, and K is the total number of pulse pulses in a random pulse repetition period radar system The number, T _m+1 is the pulse emission time of the index m+1, m is the pulse index, T _r is the uniform pulse repetition period, ε is the offset factor, t _m is the random factor of the index m, B(0,T _r ) obeys the Bernoulli distribution in the range of (0,T _r ).

Further, in step S2, the space-time guidance dictionary is represented by the matrix Φ; wherein

Clutter space-time snapshot of the complex amplitude on the angle-Doppler domain

The target component x _t = Φ γ _t , γ _t is the complex amplitude of the target space-time snapshot, v(f _d,i , f _s,j ) is the space-time steering vector, f _d,i is the Doppler frequency, f _{s , j} is the spatial frequency, γ _{c; i, j} is the complex amplitude of the clutter in the angle-Doppler plane grid, i=1,..., N _d is the Doppler frequency partition index, j=1 , ..., N _s is the spatial frequency index, N _d is the number of Doppler frequency divisions, and N _s is the number of spatial frequency divisions.

Further, in step S2, the spatial discretization process is specifically: normalizing the spatial frequency range to f _s ∈ [-0.5, 0.5], and uniformly dividing the spatial frequency range into N _s = ρ _s N equal division; normalize the Doppler frequency range to f _d ∈[-f _r /2,f _r /2], and uniformly divide the Doppler frequency range into N _d =ρ _d M equal parts; The whole angle-Doppler plane is divided into N _s N _d grids; where ρ _d , ρ _s represent the resolution scale of the Doppler frequency and the spatial frequency, respectively, f _s is the normalized spatial frequency range, N is The number of receiving arrays, f _d is the Doppler frequency range, and f _r is the maximum unambiguous Doppler frequency.

Further, in step S3, the angle-Doppler image

among them,

The energy corresponding to the clutter in the angle-Doppler plane grid, i=0,1,...,N _d ,j=0,1,...N _s ,

To estimate the resulting clutter angle - Doppler image.

Compared with the existing radar system based on the sparse STAP algorithm, the method for realizing angle-Doppler imaging based on the radar system has the following advantages: (1) effectively solving the target, the clutter Doppler blur problem, and improving the miscellaneous Wave suppression and target detection performance; (2) reduce the number of pulses emitted by the radar system, can transmit other radar waveforms in the same pulse coherent processing interval (CPI) or simultaneously observe multiple angles, thereby effectively improving the radar time dimension The multiplexing capability; (3) can effectively improve the Doppler resolution capability under the same number of pulses; (4) the low interception capability of the radar waveform, and the stronger anti-interference ability.

DRAWINGS

1 is a flow chart of a method for implementing a compressed pulse-based random pulse Doppler radar angle-Doppler imaging according to the present invention;

Figure 2 is a plot of the degree of defuzzification versus the parameter factor ε;

3 is an angle-Doppler imaging diagram obtained by different compression ratios in a pulse dimension; wherein (a) is a compression ratio of 1:2, (b) is a compression ratio of 1:3, and (c) is a compression ratio of 1: 4, (d) is a compression ratio of 1:6;

Figure 4 is a schematic diagram of angle-Doppler imaging obtained by Doppler radar, where (a) is an angle-Doppler imaging diagram obtained by a uniform PRF radar, and (b) is an angle-Doppler imaging obtained by a random PRF radar. schematic diagram.

detailed description

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.

The invention proposes a stochastic PRF pulse Doppler radar system based on compressed sensing from the perspective of solving the problem of pulse Doppler ambiguity, improving the practical value of the system and the anti-interference ability of the system. Due to the introduction of random pulses, the Doppler ambiguity can be effectively eliminated and the interception capability of the radar waveform can be reduced. Since the number of transmitted pulses is reduced, the practical value of the system can be effectively improved, and multiple groups can be transmitted in one CPI time. Radar waveforms or simultaneous observation of multiple angles can improve the multiplexing ability of the radar time dimension.

The method for compressive sensing based random pulse Doppler radar angle-Doppler imaging provided by the invention comprises the following steps:

S1: obtaining space-time compressed sampling data based on a random pulse Doppler radar system under compressed sampling;

S2: designing a space-time oriented dictionary after spatially discretizing the Doppler frequency space and the spatial frequency;

S3: Estimate the angle-Doppler image of the clutter or target using the existing sparse recovery algorithm to complete the imaging of the clutter and the target.

The specific implementation process is as follows:

Process 1: based on a random pulsed Doppler frequency radar system under compressed sampling, obtaining space-time compressed sampling data;

The traditional pulse Doppler radar system transmits M equally spaced pulse sequences (period T _r ) in one CPI period. The first pulse transmission time is T ₀ =0, then the m+1 (m=0, ..., M-1) The transmission timing of the pulses is T _m+1 = mT _r . In the radar system of the present invention, K (K ≤ M) completed transmissions are randomly extracted from M non-periodic pulse sequences, and M non-periodic pulse moments satisfy the relationship T _m+1 = mT _r + εt _m (0 ≤ ε < 1), and t _m ～ B (0, T _r ).

At this time, the echo of a certain distance unit can be expressed as an NM×1 dimensional space-time sample x=x _t +x _c +x _n , where x _t , x _c , x _n are targets, clutter, and receiver, respectively. The thermal noise component, N is the number of receiving antenna elements.

The target component can be expressed as

Where σ _t , v _dt (f _dt ), v _st (f _st ) respectively represent the complex amplitude, time domain steering vector, and spatial guidance vector of the target, f _dt =2(v _p cosφ+v _t )/λ _c ,f _St = d cosφ / λ _c are the Doppler frequency and the normalized spatial frequency of the target; v _p , v _t , φ, λ _c are the airborne velocity, the target radial velocity, the beam arrival angle and the wavelength, respectively. And have:

v _dt (f _dt )=[1,exp(j2πf _dt T ₁ )...,exp(j2πf _dt T _M-1 )] ^T ,

v _st (f _st )=[1,exp(j2πf _st ),...,exp(j2π(N-1)f _st )] ^T ,

Ignoring the effect of distance blur, the clutter component can be expressed as:

among them

They are the complex amplitude, time and space steering vectors of the i-th clutter block on the interest distance loop, and N _c is the number of independent clutter blocks in the interest distance clutter loop.

Under the radar system, only a small number of pulses are emitted, and the echo received by a certain distance unit is represented as an empty time snapshot of NK×1 (NK<NM) dimension.

At this point, the compressed sampling of the original signal is completed. The target component after sampling is:

Where H, Ψ are the sampling matrix and the measurement matrix, respectively, and I _N is an N×N unit matrix. Then the compressed received echo can be expressed as

Process 2: Design a space-time oriented dictionary to complete the sparse representation of the signal.

Due to the limited Doppler frequency and spatial frequency in the space-time snapshot of clutter, the normalized spatial frequency range is f _s ∈ [-0.5, 0.5], and the Doppler frequency is f _d ∈ [-f _r /2,f _r /2], the spatial frequency range and the Doppler frequency range are uniformly divided into N _s = ρ _s N, N _d = ρ _d M aliquots (ρ _d , ρ _s respectively represent Doppler frequencies And the spatial frequency resolution scale, which can control the size of the quantization error), divides the entire angle-Doppler plane into N _s N _d grids, thereby discretizing the Doppler frequency space and the spatial frequency space. The discretized Doppler frequency and spatial frequency approximation represent clutter echoes. However, this inevitably leads to quantization error. However, it is found by simulation test data and measured data that the clutter model established by a large number of discretized clutter scatterers can better reflect the actual clutter characteristics.

use

Representing a set of discretized Doppler frequencies and spatial frequencies, the clutter space-time snap signal can be expressed as:

Where v(f _d,k ,f _s,i ), f _d,k ,f _s,i are the space-time guided vectors of the ki discrete points of the entire angle-Doppler plane, and the discretized Doppler Frequency, normalized spatial frequency.

The matrix Φ represents the NM×N _s N _d dimensional space-time oriented dictionary (overcomplete basis) whose expression is:

And

For the clutter space, snap the complex amplitude on the angle-Doppler domain, ie the angle-Doppler image. Meanwhile, the power in the entire angular spectrum of the target - Doppler plane also sparse characteristic, therefore, the target component can be expressed as x _{_{_t}} = Φγ _t, γ _t certain complex amplitude space when snapped.

Under the radar system, the received echo can be expressed as:

Where Θ=ΨΦ is the perceptual matrix of NK×N _s N _d dimension, and γ is the angle-Doppler image containing the clutter and the target of N _s N _d ×1 dimension.

Process 3: Estimating the space-time power spectrum to achieve angle-Doppler imaging.

The angle containing the clutter and the target - the Doppler image γ can be estimated by solving the optimization problem of the following formula

Subject to

among them,

Indicates the l _p norm, ξ _e is the noise error allowed.

Thus, the angle-Doppler image can be estimated

Express

And

The beneficial effects of the newly proposed radar system in the present invention will be analyzed from the following two points: First, the Doppler ambiguity (including target Doppler ambiguity and clutter Doppler ambiguity) under the radar system is analyzed. Secondly, the angle-Doppler image recovery performance based on Doppler-dimensional compression sampling is analyzed.

(1) Analysis of the Doppler ambiguity under the radar system

Set the degree of defuzzification to

Where P _b and P _r represent the sum of the energy powers of all clutter blocks in the clutter fuzzy region and the clutter ridge region on the angle-Doppler plane, respectively.

As shown in Fig. 2, the graph reflects the relationship between the degree of defuzzification and the different parameters ε. When selecting the appropriate parameter selection range (such as 0.3 < ε < 0.7), the sum of the powers of all the clutter blocks in the fuzzy region is better. Low, it can be approximated that the clutter Doppler blur (as well as the target Doppler blur) is basically eliminated. In fact, due to the limited number of pulses emitted by the pulse Doppler radar, complete de-fuzzing of Doppler is not possible.

As shown in Fig. 4. (a) and (b), angle-Doppler imaging obtained by uniform pulse Doppler radar and random pulse Doppler radar, respectively. Under uniform PRF, Doppler ambiguity (target and clutter simultaneous blurring) will occur, and under random PRF, Doppler ambiguity will be effectively eliminated.

(2) Angle-Doppler imaging recovery performance analysis based on Doppler-dimensional compression sampling

Under the random PRF radar system, K pulse transmissions are randomly extracted from M=96 non-uniform pulse sequences in one CPI, and the angled Doppler images are obtained by using the received echo signals (Doppler compressed sampling signals). It can be seen from Fig. 3 (a) - (d) that K = 48, 32, 24, 16 pulses are respectively emitted (Doppler compression ratio is 1:2, 1:3, 1:4, 1:6, respectively) The angle obtained by using the compressed signal - the recovery performance of the Doppler image can be close to the recovery performance of the original signal. Therefore, the stochastic PRF radar system based on Doppler-dimensional compression sampling under the invention is feasible.

The invention relates to radar signal processing and ground moving target detection, and proposes a random impulse Doppler array radar based on compressed sensing, which constructs a clutter containing according to the sparse characteristics of the clutter or target in the angle-Doppler domain. The angle of the target - the sparse recovery problem of the Doppler image, the angle-Doppler image is obtained by solving the problem, which provides a basis for subsequent target detection. Compared with the traditional pulse Doppler radar, the introduction of random pulses can effectively eliminate the Doppler blur; because the number of transmitted pulses is reduced, the radar waveform interception capability is reduced, and multiple sets of radar waveforms are transmitted in one CPI. Or observing multiple angles at the same time, improving the multiplexing ability of the radar time dimension.

Those skilled in the art will appreciate that the above description is only a 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 scope of the present invention, All should be included in the scope of protection of the present invention.

Claims

A random pulse Doppler radar angle-Doppler imaging method based on compressed sensing, characterized in that it comprises the following steps:

S1: obtaining space-time compressed sampling data based on a radar system;

S2: spatially discretizing the Doppler frequency space and the spatial frequency in the space-time compressed sample data, and obtaining a space-time oriented dictionary;

S3: Estimating the space-time power spectrum according to the space-time guiding dictionary, and obtaining an angle-Doppler image including the clutter and the target.
The method according to claim 1, wherein in step S1, K complete transmissions are randomly selected from M non-periodic pulse sequences in the radar system, and M non-periodic pulse moments satisfy the relationship. T m+1 =mT r +εt m , where K≤M, 0≤ε<1, t m ～B(0,T r ), M is the number of pulse sequences in a single coherent processing time, and K is a random pulse The number of total transmitted pulses in the repetitive cycle radar system, T m+1 is the pulse transmission time of index m+1, m is the pulse index, T r is the uniform pulse repetition period, ε is the offset factor, and t m is the index m The random factor, B(0,T r ), is in the range of (0,T r ) obeying the Bernoulli distribution.
The method according to claim 1 or 2, wherein in step S2, the space-time guidance dictionary is represented by the matrix Φ;

among them,
Clutter space-time snapshot of the complex amplitude on the angle-Doppler domain
The target component x t = Φ γ t , γ t is the complex amplitude of the target space-time snapshot, v(f d,i , f s,j ) is the space-time steering vector, f d,i is the Doppler frequency, f s , j is the spatial frequency, γ c; i, j is the complex amplitude of the clutter in the angle-Doppler plane grid, i=1,..., N d is the Doppler frequency partition index, j=1 , ..., N s is the spatial frequency index, N d is the number of Doppler frequency divisions, and N s is the number of spatial frequency divisions.
The method according to any one of claims 1 to 3, wherein in step S2, the spatial discretization processing is specifically:

The spatial frequency range is normalized to f s ∈ [-0.5, 0.5], and the spatial frequency range is evenly divided into N s = ρ s N equal parts;

The Doppler frequency range is f d ∈[-f r /2,f r /2], and the Doppler frequency range is evenly divided into N d =ρ d M aliquots; achieving the entire angle-Doppler The plane is divided into N s N d grids;

Where ρ d , ρ s represent the resolution scale of Doppler frequency and spatial frequency, respectively, f s is the normalized spatial frequency range, N is the number of receiving arrays, f d is the Doppler frequency range, and f r is the maximum unblurred Doppler frequency.
The method according to claim 1 or 2, wherein in step S3, said angle-Doppler image
among them,
The energy corresponding to the clutter in the angle-Doppler plane grid, i=0,1,...,N d ,j=0,1,...N s ,
To estimate the resulting clutter angle - Doppler image.