CN103728607A - Space time code three-dimensional self-adaptation clutter cancelling method for onboard multiple input multiple output (MIMO) radar - Google Patents

Abstract

The invention discloses a space time code three-dimensional self-adaptation clutter cancelling method for onboard multiple input multiple output (MIMO) radar. The method includes first building an onboard MIMO radar echo signal model, calculating residue echo signal energy, minimizing the echo signal energy to obtain a self-adaptation clutter canceller coefficient compensation device, acquiring a space time code three-dimensional self-adaptation clutter canceller through sliding window arrangement, utilizing the canceller to cancel echo signals and cancelling most clutter through preprocessing of the space time code three-dimensional self-adaptation clutter canceller. The method is used before a common space time code processing method and a dimension reducing STAP method to conduct space time code three-dimensional canceling on the clutter in the onboard MIMO radar echo signals, the clutter can be effectively restrained in the first stage of radar signal processing, and detection performance of a moving target is improved. The method is applicable to both positive side looking radars and non-positive side looking radars, and has good moving target detection capability.

Description

The empty time-code three-dimensional adaptive of airborne MIMO radar clutter offsets method

Technical field

The invention belongs to Radar Signal Processing Technology field, relate generally to sky time-code three-dimensional adaptive clutter cancellation (Space-Time Adaptive Clutter Canceller, STACC), be that the empty time-code three-dimensional adaptive of a kind of airborne MIMO radar clutter offsets method specifically.Can be applicable to airborne MIMO radar system, be not only applicable to positive side-looking MIMO radar, be also applicable to non-working side MIMO radar.

Background technology

Clutter cancellation is to occur the earliest it being also the most frequently used mti filter, it is applied to ground radar the earliest, the land clutter signal receiving due to ground radar focuses mostly near 0 frequency of Doppler frequency, and clutter cancellation can be regarded as a Hi-pass filter that effectively suppresses low frequency clutter.But, for airborne MIMO (multiple-input multiple-output) Texas tower, because platform moves, land clutter will can not be concentrated and be distributed near 0 frequency, but expand, be distributed in whole Doppler domain, clutter will present strong coupled characteristic when empty, and original clutter cancellation will be no longer applicable.Because the land clutter of airborne MIMO radar is rendered as space-time two-dimensional coupled characteristic, clutter energy will mainly be distributed on space-Doppler two dimensional surface, and this has just determined that its clutter suppresses to belong to space-time two-dimensional filtering problem.On this basis, space-time two-dimensional adaptive processing method STAP (space-time adaptive processing) has obtained further development.Full dimension STAP method, as adopt optimal processor can be effectively when the sky bidimensional clutter is carried out to filtering, although performance is superior, due to its very high computation complexity and the huge demand to sample size, make it in Practical Project, be difficult to realization.In order to overcome the shortcoming of full dimension STAP, dimensionality reduction STAP technology has obtained very large development.Classical dimensionality reduction STAP method has accessory channel method (ACR), factorization space-time adaptive processes FA method and spreading factor space-time adaptive is processed EFA method, all can obtain good performance in the ideal case, but array error is inevitable in actual conditions, spatial domain error makes space-time two-dimensional clutter spectrum scatter along spatial domain dimension, Clutter Degrees of Freedom increases greatly, and during self-adaptive processing, only have and when self-adaptation degree of freedom is greater than Clutter Degrees of Freedom, just can obtain good clutter rejection, visible, in order to improve clutter rejection, need to increase self-adaptation degree of freedom, but when degree of freedom increases, can bring the problem that sample number increases and operand increases to said method, this makes said method in practical engineering application, cannot obtain good clutter rejection.

Summary of the invention

The object of the invention is to overcome the defect of above-mentioned prior art, provide one to reduce Clutter Degrees of Freedom, be applicable to positive side-looking MIMO radar, the empty time-code three-dimensional adaptive of the airborne MIMO radar clutter that is also applicable to non-working side MIMO radar offsets method.

The present invention is that the empty time-code three-dimensional adaptive of a kind of airborne MIMO radar clutter offsets method, and self-adapting clutter cancellation process of the present invention comprises the steps:

Step 1. is set airborne MIMO radar antenna and within a relevant processing time, is launched K pulse, builds the echo of k pulse of airborne MIMO radar reception, and this echo comprises clutter and noise, and signal model is:

$x_k={\mathrm{AD}}_k{\mathrm{\ρ}}_k+x_k^{\left(w\right)},$

In formula, A is the matrix that spatial domain phase term forms, D _kfor the diagonal matrix that time domain doppler phase item forms, ρ _kbe the random magnitude vector of k pulse moment U clutter scattering unit, U is the number of a range unit inscattering unit,

it is the white Gaussian noise vector in k pulse moment.

Step 2., according to echo signal model, calculates the residual echo signal energy with k adjacent two pulses of pulse || ε _k|| ²for:

|| ε _k|| ²=|| Qx _k-x _k+1|| ², adjacent two pulses are k pulse and k+1 pulse

In formula, Q is clutter cancellation matrix of coefficients, || || be vectorial 2-norm;

Step 3. minimizes residual echo signal energy || ε _k|| ², obtain clutter cancellation matrix of coefficients Q and be:

$Q=R_{k+1,k}{R}_{k,k}^{-1},$

In formula, R _{k+1, k}be clutter and the noise correlation matrix of k pulse and k+1 pulse, R _k,kbe clutter and the noise autocorrelation matrix of k pulse, [] ^-1expression is to matrix inversion.

Step 4. is according to the clutter cancellation matrix of coefficients Q that tries to achieve, chooses in all radar signals one and relevantly processes in the time interval every adjacent two pulses and offset the matrix of coefficients needing and slide window and arrange, and obtains self-adapting clutter canceller, this canceller

be expressed as:

I in formula _nM∈ C ^{nM × NM}be unit matrix, M is aerial array transmitting element number of array, and N is antenna reception element number of array;

The self-adapting clutter canceller that step 5. utilization obtains

the relevant all echo x that process in the time interval in radar signal are carried out to clutter and offset, the radar signal matrix that obtains containing noise signal and remaining noise signal after offseting is

$\tilde{x}=\stackrel{\‾}{Q}x,$

In formula, x is the long vector that tie up NMK × 1 that relevant data of processing the pulse of K altogether that in interval, CPI receives are aligned to, $x={\left[\begin{array}{cccc}{x}_1^T,& x_2^T,& \·\·\·,& x_K^T\end{array}\right]}^T,$

the input signal of processing during for cascade sky.

The present invention is extended to airborne MIMO motion platform by original ground radar clutter canceller, propose a kind of empty time-code three-dimensional adaptive clutter for airborne MIMO radar and offseted method, in the present invention, the empty time-code three-dimensional adaptive clutter cancellation of airborne MIMO radar (STACC) of structure adopts pulse between two to offset, total umber of pulse will reduce one, the time domain dimensionality reduction matrix of cascade dimensionality reduction STAP method is also by time term of corresponding minimizing, reduced the degree of freedom of clutter, and all there is good clutter cancellation performance in positive side-looking radar and non-working side radar.

Realization of the present invention is also: build the echo signal model of k pulse of airborne MIMO radar reception, comprise the steps

1.1 carrier aircraft flying heights are H, and speed is v, radar transmit and receive battle array for the uniform line-array of parallel placement.Aerial array transmitting array number is M, and transmitting array element distance is d _m, reception array element is N, reception array element distance is d _n.In Fig. 1, radar antenna is placed as non-working side radar, between the rectilinear direction at aerial array place and the velocity reversal of carrier aircraft, has an angle it is crab angle.If crab angle

it is positive side-looking radar that radar antenna is placed.Radar operation wavelength λ, pulse repetition rate f _r, relevant a processing in interval has K radar pulse to do coherent accumulation.Suppose relevant a processing in interval, each transmitting array element is launched the train of impulses waveform being comprised of K pulse simultaneously, and M transmitted waveform is mutually orthogonal, at each reception array element place, by M reference transmitted signal, the echo data of K recurrence interval is carried out to matched filtering respectively, so, for one, apart from carrier aircraft position angle, be

the angle of pitch is the ground scatter point of θ, and at n array element place of receiving end, the echoed signal obtaining under corresponding m transmitting array element, a k pulse is

N=0 in formula ..., N-1, m=0 ..., M-1, k=0 ..., K-1,

for the echo amplitude of this clutter scattering point, obey the multiple Gaussian distribution of zero-mean;

for receiving spatial domain frequency item,

for receiving spatial domain frequency,

for transmitting spatial domain frequency item,

for transmitting spatial domain frequency,

for Doppler frequency item, for Doppler frequency.

1.2 hypothesis radar detection points are R apart from carrier aircraft radial distance, and all noise signals that in a pulse, radar receives are the echo sum of all scattering points in Range resolution unit, one, ground,

In formula,

for clutter item,

for noise item, be to obey 0 average, variance is

white Gaussian noise, the clutter c of scattering point _{n, m, k}and the Gaussian noise in above formula (2)

it is independent that relation is added up each other.

1.3 by a range unit along position angle

from 0 to π, the echoed signal that the discrete U of a being divided into scattering unit obtains a pulse is:

In formula be the position angle of i scattering unit,

for the clutter echo amplitude of corresponding i scattering unit,

be comprise transmitting, the spatial domain item in being received in,

for spatial domain frequency,

for Doppler's item,

for Doppler frequency.

The echo of 1.4 k the pulses that radar is received is arranged in the column vector of NM × 1

X _k=[x _{0,0, k}, x _{0,0, k}..., x _{0,0, k}] ^t,

$x_k={\mathrm{AD}}_k{\mathrm{\ρ}}_k+x_k^{\left(w\right)},---\left(4\right)$

In formula

$x_k^{\left(w\right)}={\left[\begin{array}{ccc}{x}_{\mathrm{0,0},k}^{\left(w\right)},& \·\·\·,& x_{N-1,M-1,k}^{\left(w\right)}\end{array}\right]}^T,$

Wherein A is the matrix that spatial domain phase term forms, D _kfor the diagonal matrix that time domain doppler phase item forms, diag () represents vector to be become to diagonal matrix, ρ _kbe the random magnitude vector of k pulse moment U clutter scattering unit,

be in the white Gaussian noise vector mode in k pulse moment, [] ^trepresent vectorial transposition;

Therefore, the echo in k+1 moment, this echo comprises clutter and noise, is written as

$x_{k+1}=A\stackrel{\‾}{D}{D}_k{\mathrm{\ρ}}_{k+1}+w_{k+1}^{\left(w\right)}---\left(5\right)$

Wherein

for the doppler phase item producing in individual pulse, for doppler phase.

The present invention has not only considered spatial domain azimuth information and the doppler information of clutter, has also considered clutter echo amplitude information ρ _k, therefore the target detection performance of the inventive method is more excellent.

Realization of the present invention is also: according to echo signal model, calculate the residual echo signal energy of adjacent two pulses || ε _k|| ², comprise the steps:

2.1 according in formula (3) about k pulse echo signal of airborne MIMO radar, the echoed signal that obtains k+1 pulse is

x _k+1＝[x _0,0,k+1, x _0,0,k+1, …, x _0,0,k+1] ^T (6)

2.2 carry out obtaining after clutter offsets the energy of residual echo signal by the echo of the echo of k+1 pulse and k pulse || ε _k|| ²for

||ε _k|| ²＝||Qx _k-x _k+1|| ² (7)

In formula, Q is clutter cancellation matrix of coefficients, || || be vectorial 2-norm.

The clutter of space-time adaptive clutter cancellation offsets processing target: optimize canceller matrix of coefficients Q, make remaining clutter energy || ε _k|| ²minimum.

Realization of the present invention is also: minimize residual echo signal energy || ε _k|| ², obtain matrix of coefficients Q, comprise the steps:

3.1 by residual echo signal energy || ε _k|| ²minimize, ask

in formula, min represents minimization function:

$\begin{array}{c}\underset{Q}{\min }E\left({\left|\right|{\mathrm{\ϵ}}_k\left|\right|}^2\right)=\underset{Q}{\min }E\left({\left|\right|Q{x}_k-x_{k+1}\left|\right|}^2\right)\\ =\underset{Q}{\min }E\left[{(Q{x}_k-x_{k+1})}^H(Q{x}_k-x_{k+1})\right]\\ =\underset{Q}{\min }E\left\{\mathrm{tr}\right[(Q{x}_k-x_{k+1}){({\mathrm{Qx}}_k-x_{k+1})}^H\left]\right\}\\ =\underset{Q}{\min }\mathrm{tr}(Q{R}_{k,k}{Q}^H-{\mathrm{QR}}_{k,k+1}\\ -R_{k+1,k}{Q}^H+R_{k+1,k+1}\left)\right]\\ =\underset{Q}{\min }J\left(Q\right)\end{array}---\left(8\right)$

E () tabular form mathematical expectation in formula, $R_{k,k}=E\left(x_k{x}_k^H\right),$ $R_{k,k+1}=E\left(x_k{x}_{k+1}^H\right),$ $R_{k+1,k}=E\left(x_{k+1}{x}_k^H\right)$ With $R_{k+1,k+1}=E\left(x_{k+1}{x}_{k+1}^H\right)$ For clutter and noise correlation matrix, () ^hthe complex-conjugate transpose of representing matrix or vector;

3.2 ask conjugation derivative by cost function about clutter cancellation matrix of coefficients, and make derivative $\∂J\left(Q\right)/\∂Q*=0,$

$\begin{array}{c}\∂J\left(Q\right)/\∂Q*=\∂[\mathrm{tr}(Q{R}_{k,k}{Q}^H-Q{R}_{k,k+1}\\ -R_{k+1,k}{Q}^H+R_{k+1,k+1}\left)\right]/\∂Q*\\ ={\mathrm{QR}}_{k,k}-R_{k+1,k}=0\end{array}---\left(9\right)$

Solution is

$Q=R_{k+1,k}{R}_{k,k}^{-1}---\left(10\right)$

Clutter in formula and noise correlation matrix R _{k+1, k}, R _k,kand matrix of coefficients Q, respectively by its maximal possibility estimation

with

replace.Can be estimated to obtain by following formula:

$\left\{\begin{array}{c}{\hat{R}}_{k,k}=\frac{1}{L(K-1)}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}\underset{k=1}{\overset{K-1}{\mathrm{\Σ}}}{x}_k\left(l\right)x_k^H\left(l\right)\\ {\hat{R}}_{k+1,k}=\frac{1}{L(K-1)}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}\underset{k=1}{\overset{K-1}{\mathrm{\Σ}}}{x}_{k+1}\left(l\right)x_k^H\left(l\right)\\ \hat{Q}={\hat{R}}_{k+1,k}{\hat{R}}_{k,k}^{-1}\end{array}\right.---\left(11\right)$

X in formula _k(l) echo of k the pulse that l range unit of expression receives.

From formula (11), can find out,

with can on average estimate by the sliding window of time domain in adjacent sample range unit.Formula (11) formula has larger calculated amount and computation complexity, and in order to reduce calculated amount, the present invention is estimating with sample

or

time, the range unit of choosing symmetrically range unit to be detected both sides is sample, therefore when the adjacent detection range unit of the next one is detected, have the range unit sample of many repetitions, auto-correlation item and the simple crosscorrelation item of the range unit sample of these repetitions are identical, and this calculating needn't re-start, and directly continues to use, double counting can be avoided, calculated amount and computation complexity can be reduced like this.In addition, when next one detection range unit is detected, suppose to have m sample and upgrade, also can calculate above

with

basis on, carry out order m renewal.

The clutter cancellation that the present invention utilizes step 4 to obtain

a relevant echo x who processes in the time interval is carried out to clutter and offset, the radar signal matrix that contains echo signal, noise signal and remaining noise signal after offseting is in formula, x is the long vector that tie up NMK × 1 that relevant data of processing the pulse of K altogether receiving in interval are aligned to, $x={\left[\begin{array}{cccc}{x}_1^T,& x_2^T,& \·\·\·,& x_K^T\end{array}\right]}^T,$ The present invention offsets and can suppress most clutter energy.

The present invention offset finish dealing with after during cascade Radar Signal Processing follow-up empty disposal route to residual echo signal

process, further complete the inhibition of radar clutter.

The present invention compared with prior art has following characteristics:

1) the present invention has reduced Clutter Degrees of Freedom, has less calculated amount and computation complexity.The present invention estimates with sample

or

time, the range unit of choosing symmetrically range unit to be detected both sides is sample, therefore when the adjacent detection range unit of the next one is detected, have the range unit sample of many repetitions, auto-correlation item and the simple crosscorrelation item of the range unit sample of these repetitions are identical, and this calculating needn't re-start, and directly continues to use, can avoid double counting, the inventive method has less calculated amount and computation complexity like this.By Fig. 4 and Fig. 5, can find out that the inventive method has reduced the degree of freedom of clutter.

2), compared to existing method TDPC, the present invention has better target detection performance performance.The present invention has utilized the azimuth information of clutter scattering point in Range resolution unit, and doppler information and echo amplitude information, compared to existing method, have been utilized clutter echo amplitude information more, thereby promotes to some extent than existing methodical target detection performance.By Fig. 7 and Figure 10, can find out that the performance of the inventive method has a certain upgrade.

3) the present invention has good clutter rejection, is not only applicable to positive side-looking radar, and is applicable to non-working side radar.In the present invention, the empty time-code three-dimensional adaptive of the airborne MIMO radar clutter cancellation of structure adopts pulse between two to offset, total umber of pulse will reduce one, the time domain dimensionality reduction matrix of cascade dimensionality reduction STAP method is also by time term of corresponding minimizing, reduced the degree of freedom of clutter, make self-adaptation degree of freedom be greater than Clutter Degrees of Freedom, thereby the present invention is clutter reduction energy effectively, the conventional space-time adaptive processing method of cascade of the present invention, more independent space-time adaptive processing method, can improve the detection performance of target better; And by Fig. 3,6,7,8,9 and 10, can find out that the present invention is applicable equally for positive side-looking radar and non-working side radar, can improve well the detection performance of target.

Accompanying drawing explanation

Fig. 1 is the geometry schematic diagram of airborne MIMO radar;

Fig. 2 is process flow diagram of the present invention;

Fig. 3 is clutter power spectrogram (crab angle before and after pre-filtering is processed

), wherein Fig. 3 (a) processes front clutter power spectrogram for pre-filtering, and Fig. 3 (b) is clutter power spectrogram after pre-filtering is processed;

Clutter feature spectrogram (crab angle when Fig. 4 is sky before and after pre-filtering

);

Clutter feature spectrogram (crab angle when Fig. 5 is sky before and after pre-filtering

);

Fig. 6 is the positive side-looking radar improvement factor comparison diagram before and after MTI, FA and EFA method cascade the present invention (STACC);

Fig. 7 is the positive side-looking radar improvement factor comparison diagram of the present invention (STACC) and TDPC difference cascade MTI, FA and EFA method;

Fig. 8 is clutter power spectrogram (crab angle before and after pre-filtering is processed ), wherein Fig. 8 (a) processes front clutter power spectrum for pre-filtering, and Fig. 8 (b) is clutter power spectrum after pre-filtering is processed;

Fig. 9 is the non-working side radar improvement factor comparison diagram before and after MTI, FA and EFA method cascade the present invention (STACC);

Figure 10 is the non-working side radar improvement factor comparison diagram of the present invention (STACC) and TDPC difference cascade MTI, FA and EFA method.

Embodiment

Method implementation process of the present invention is described with reference to the accompanying drawings.

Embodiment 1:

The present invention is directed to the situations such as problem that the Clutter Degrees of Freedom being caused by array error that clutter in prior art exists in suppressing increases and computation complexity height, innovation and research have been launched, the empty time-code three-dimensional adaptive of airborne MIMO radar provided by the invention clutter offsets method, be applied in Radar Signal Processing, airborne early warn ing radar, owing to being erected on very high platform, much far away than ground radar to the visual range of low flyer, thereby in widespread attention.Airborne radar does down while looking work, not only scope is wide for ground clutter, intensity is large, and the clutter of different directions is different with respect to the speed of carrier aircraft, thereby make clutter spectrum broadening greatly, and again due to the cause of MIMO radar emission waveform diversity, cause the clutter amount of airborne radar much larger than ordinary radar, continue to use traditional clutter suppression method make airborne radar cannot be from strong clutter background detection and Identification target, target detection ability is had a strong impact on.This the present invention has been launched to research to clutter, by the empty time-code three-dimensional adaptive of airborne MIMO radar clutter is offseted, pre-filtering processing to radar signal in other words conj.or perhaps, while clutter being carried out to sky before cascade disposal route and traditional adaptive processing method when conventional sky, bidimensional offsets, in the first stage of Radar Signal Processing, just suppressed most of clutter energy, effectively clutter reduction, improve the detection performance of moving-target, see figures.1.and.2, it is special in airborne MIMO radar that clutter of the present invention offsets method, self-adapting clutter of the present invention offsets flow process and comprises:

Step 1. is set airborne MIMO radar antenna and within a relevant processing time, is launched K pulse, builds the echo signal model of airborne MIMO radar received pulse, in K pulse, take k as variable, the echo signal model of k pulse is expressed as, and this echo comprises clutter and noise

$x_k={\mathrm{AD}}_k{\mathrm{\ρ}}_k+x_k^{\left(w\right)},$

In formula, A is the matrix that spatial domain phase term forms, D _kbe the diagonal matrix that k pulse time domain doppler phase item forms, ρ _kbe the random magnitude vector of k pulse moment U clutter scattering unit, it is the white Gaussian noise vector in k pulse moment.The geometry of airborne MIMO radar is referring to Fig. 1:

1.1 build the individually clutter echoed signal of area scattering point; As shown in Figure 1, carrier aircraft flying height is H, and speed is v, radar transmit and receive battle array for the uniform line-array of parallel placement.Aerial array transmitting array number is M, and transmitting array element distance is d _m, reception array number is N, reception array element distance is d _n.In Fig. 1, radar antenna is placed as non-working side radar, between the rectilinear direction at aerial array place and the velocity reversal of carrier aircraft, has an angle

it is crab angle.If crab angle

it is positive side-looking radar that radar antenna is placed.Radar operation wavelength λ, pulse repetition rate f _r, relevant a processing in interval has K radar pulse to do coherent accumulation.Suppose relevant a processing in interval, each transmitting array element is launched the train of impulses waveform being comprised of K pulse simultaneously, and M transmitted waveform is mutually orthogonal, at each reception array element place, by M reference transmitted signal, the echo data of K recurrence interval is carried out to matched filtering respectively, for one, apart from carrier aircraft position angle, be

the angle of pitch is the ground scatter point of θ, and at n array element place of receiving end, the noise signal obtaining under corresponding m transmitting array element, a k pulse is

N=0 in formula ..., N-1, m=0 ..., M-1, k=0 ..., K-1,

for the random echo amplitude of this clutter scattering point, obey the multiple Gaussian distribution of zero-mean, relevant with ground scene,

for receiving spatial domain frequency item, for receiving spatial domain frequency,

for transmitting spatial domain frequency item,

for transmitting spatial domain frequency, for Doppler frequency item,

for Doppler frequency; Wherein K is the pulse sum of transmitting in the relevant processing time.

1.2 build all echoed signals that in individual pulse, radar receives; Suppose that radar detection point is R apart from carrier aircraft radial distance, all echoed signals that in a pulse, airborne MIMO radar receives are the echo sum of all scattering points in Range resolution unit, one, ground, referring to Fig. 1,

In formula,

for clutter item, be the clutter c of all scattering points in Range resolution unit, one, ground _{n, m, k}sum,

for noise item, be to obey 0 average, variance is

white Gaussian noise, the clutter c of scattering point _{n, m, k}with the Gaussian noise in above formula

it is independent that relation is added up each other.

All echoed signal discrete processes that in 1.3 pairs of individual pulses, radar receives; By a range unit along position angle

from 0 to π, the echoed signal that the discrete U of a being divided into scattering unit obtains a pulse is:

In formula

be the position angle of i scattering unit,

for the clutter echo amplitude of corresponding i scattering unit,

be comprise transmitting, the spatial domain item in being received in, for spatial domain frequency,

for Doppler's item, for Doppler frequency.U is the number of a range unit inscattering unit, and the value of the number U of discrete division unit should be large as far as possible, but is convenient to emulation experiment, and in this example, U gets 300.

1.4 build the echo signal model of k pulse; The echo of k the pulse that radar is received is arranged in the column vector of NM × 1 dimension, x _k=[x _{0,0, k}, x _{1,0, k}..., x _{n-1, M-1, k}] ^t,

$x_k={\mathrm{AD}}_k{\mathrm{\ρ}}_k+x_k^{\left(w\right)},$

In formula

$x_k^{\left(w\right)}={\left[\begin{array}{ccc}{x}_{\mathrm{0,0},k}^{\left(w\right)},& \·\·\·,& x_{N-1,M-1,k}^{\left(w\right)}\end{array}\right]}^T,$

Wherein A is the matrix that spatial domain phase term forms, its each classify the column vector of length N M × 1 dimension as, be that the column vector that receives dimension N × 1 carries out with the column vector of transmitting dimension M × 1 that Kronecker is long-pending to be obtained, D _kbe the diagonal matrix that k pulse time domain doppler phase item forms, diag () represents vector to be become to diagonal matrix, ρ _kbe the random magnitude vector of k pulse moment U clutter scattering unit,

it is the white Gaussian noise vector in k pulse moment.Obtain thus the echo signal model of k pulse.

Therefore, the echo in k+1 moment, this echo comprises clutter and noise, is written as

$x_{k+1}=A\stackrel{\‾}{D}{D}_k{\mathrm{\ρ}}_{k+1}+x_{k+1}^{\left(w\right)},$

Wherein

for the doppler phase item producing in individual pulse,

for doppler phase.

The constructed echo signal model of the present invention has been utilized the azimuth information of clutter scattering point in Range resolution unit, doppler information and clutter echo amplitude information ρ _k, it is estimated to obtain by maximum likelihood by sample, compared to existing method, has utilized clutter echo amplitude information more, thereby more excellent to the rejection of clutter than existing method.

Step 2., according to echo signal model, calculates the residual echo signal energy with k adjacent two pulses of pulse || ε _k|| ²for:

|| ε _k|| ²=|| Qx _k-x _k+1|| ², adjacent two pulses are k pulse and k+1 pulse

In formula, Q is clutter cancellation matrix of coefficients, || || be vectorial 2-norm;

Step 3. is according to minimizing residual echo signal energy || ε _k|| ², obtain clutter cancellation matrix of coefficients Q and be:

$Q=R_{k+1,k}{R}_{k,k}^{-1},$

In formula, R _{k+1, k}be clutter and the noise correlation matrix of k pulse and k+1 pulse, R _k,kbe clutter and the noise autocorrelation matrix of k pulse, [] ^-1expression is to matrix inversion.

The clutter cancellation matrix of coefficients Q that step 4. is tried to achieve according to step 3, chooses every adjacent two pulses and offsets the matrix of coefficients needing and slide window and arrange, and obtains self-adapting clutter canceller, this canceller

be expressed as:

I in formula _nM∈ C ^{nM × NM}be unit matrix, M and N are respectively aerial array transmitting array number and receive element number of array;

it is the clutter pre-filtering matrix being formed by clutter cancellation matrix of coefficients.From canceller structure, obtain suitable canceller matrix of coefficients Q and can make the clutter of every adjacent two pulses effectively offset.

The self-adapting clutter canceller that step 5. utilization obtains

the relevant all echo x that process in the time interval in radar signal are carried out to clutter and offset, the radar signal matrix that obtains containing echo signal, noise signal and remaining noise signal after offseting is

$\tilde{x}=\stackrel{\‾}{Q}x,$

In formula, x is the long vector that tie up NMK × 1 that relevant data of processing the pulse of K altogether receiving in interval are aligned to, the correlation matrix of x is R=E (xx ^h) ∈ C ^{nMK × NMK}, the input signal of processing during for cascade sky.In the present invention, carry out clutter offset operation after total umber of pulse will reduce one, from K, be reduced to K-1, so the data after offseting

for dimension is the column vector of NM (K-1) × 1, the correlation matrix of its residual echo data is $\tilde{R}=E\left({\tilde{x}\tilde{x}}^H\right)=\stackrel{\‾}{Q}R{\stackrel{\‾}{Q}}^H\∈C^{\mathrm{NM}(K-1)\×\mathrm{NM}(K-1)},$ Whole Radar Signal Processing further completes radar signal clutter and suppresses after cascade is processed.

The inventive method is as the clutter preconditioning technique of airborne MIMO Texas tower, belong to adaptive clutter cancellation, therefore comparatively flexible, in the first stage, clutter is carried out to pre-filtering, effectively clutter reduction signal, be not only applicable to positive side-looking radar, for non-working side radar, still can obtain good performance performance simultaneously.

Embodiment 2:

The empty time-code three-dimensional adaptive of airborne MIMO radar clutter offsets method with embodiment 1, and wherein step 2., according to echo vector model, is calculated the residual echo signal energy of adjacent two pulses || ε _k|| ², detailed process comprises as follows:

2.1 according in formula (3) about k pulse echo signal model of airborne MIMO radar, the echo that obtains k+1 pulse is

x _k+1＝[x _0,0,k+1, x _1,0,k+1, …, x _N-1,M-1,k+1] ^T，

2.2 carry out obtaining after clutter offsets the energy of residual echo signal by the echo of the echo of k+1 pulse and k pulse || ε _k|| ²for

||ε _k|| ²＝||Qx _k-x _k+1|| ²，

In formula, Q is clutter cancellation matrix of coefficients, || || be vectorial 2-norm.

The clutter of space-time adaptive clutter cancellation offsets processing target: optimize canceller matrix of coefficients Q, make remaining clutter energy || ε _k|| ²minimum, even the clutter after offseting and noise residual amount of energy minimum.

Embodiment 3:

The empty time-code three-dimensional adaptive of airborne MIMO radar clutter offsets method with embodiment 1-2, and wherein step 3. minimizes residual echo signal energy || ε _k|| ², obtaining clutter cancellation matrix of coefficients Q, its detailed process comprises as follows:

3.1 by residual echo signal energy || ε _k|| ²minimize, ask

in formula, min represents minimization function:

$\begin{array}{c}\underset{Q}{\min }E\left({\left|\right|{\mathrm{\ϵ}}_k\left|\right|}^2\right)=\underset{Q}{\min }E\left({\left|\right|Q{x}_k-x_{k+1}\left|\right|}^2\right)\\ =\underset{Q}{\min }E\left[{(Q{x}_k-x_{k+1})}^H(Q{x}_k-x_{k+1})\right]\\ =\underset{Q}{\min }E\left\{\mathrm{tr}\right[(Q{x}_k-x_{k+1}){({\mathrm{Qx}}_k-x_{k+1})}^H\left]\right\}\\ =\underset{Q}{\min }\mathrm{tr}(Q{R}_{k,k}{Q}^H-{\mathrm{QR}}_{k,k+1}\\ -R_{k+1,k}{Q}^H+R_{k+1,k+1}\left)\right]\\ =\underset{Q}{\min }J\left(Q\right)\end{array},$

E () tabular form mathematical expectation in formula, $R_{k,k}=E\left(x_k{x}_k^H\right),$ $R_{k,k+1}=E\left(x_k{x}_{k+1}^H\right),$ $R_{k+1,k}=E\left(x_{k+1}{x}_k^H\right)$ With $R_{k+1,k+1}=E\left(x_{k+1}{x}_{k+1}^H\right)$ For clutter and noise correlation matrix.J (Q) is for offseting the cost function of matrix of coefficients about clutter.Residual echo signal energy a hour corresponding clutter to offset effect best.

3.2 ask conjugation derivative by cost function about clutter cancellation matrix of coefficients, and make derivative $\∂J\left(Q\right)/\∂Q*=0,$

$\begin{array}{c}\∂J\left(Q\right)/\∂Q*=\∂[\mathrm{tr}(Q{R}_{k,k}{Q}^H-Q{R}_{k,k+1}\\ -R_{k+1,k}{Q}^H+R_{k+1,k+1}\left)\right]/\∂Q*\\ ={\mathrm{QR}}_{k,k}-R_{k+1,k}=0\end{array},$

Solution is

$Q=R_{k+1,k}{R}_{k,k}^{-1}\text{,}$

From above formula, clutter cancellation matrix of coefficients Q is tried to achieve by clutter and noise correlation matrix, and clutter and noise correlation matrix can be estimated by sample, so the clutter cancellation that the present invention proposes is adaptive canceler.In reality, the clutter in formula (10) and noise correlation matrix R _{k+1, k}, R _k,kand matrix of coefficients Q, respectively by its maximal possibility estimation

with

replace, can be estimated to obtain by following formula:

$\left\{\begin{array}{c}{\hat{R}}_{k,k}=\frac{1}{L(K-1)}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}\underset{k=1}{\overset{K-1}{\mathrm{\Σ}}}{x}_k\left(l\right)x_k^H\left(l\right)\\ {\hat{R}}_{k+1,k}=\frac{1}{L(K-1)}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}\underset{k=1}{\overset{K-1}{\mathrm{\Σ}}}{x}_{k+1}\left(l\right)x_k^H\left(l\right)\\ \hat{Q}={\hat{R}}_{k+1,k}{\hat{R}}_{k,k}^{-1}\end{array}\right.,$

In formula, x _k(l) echo of k the pulse that l range unit of expression receives.

R _k,kmaximal possibility estimation

and R _{k+1, k}maximal possibility estimation

conventionally all can on average estimate by the sliding window of time domain in adjacent sample range unit.(11) formula can be found out, has larger calculated amount and computation complexity, and in order to reduce calculated amount, the present invention is estimating with sample

or

time, the range unit of choosing symmetrically range unit to be detected both sides is sample, therefore when the adjacent detection range unit of the next one is detected, have the range unit sample of many repetitions, auto-correlation item and the simple crosscorrelation item of the range unit sample of these repetitions are identical, and this calculating needn't re-start, and directly continues to use, can avoid double counting, make the inventive method there is less calculated amount and computation complexity.In addition, when next one detection range unit is detected, suppose to have m sample and upgrade, also can calculate above

with

basis on, carry out order m renewal, further reduce calculated amount.

Embodiment 4:

The empty time-code three-dimensional adaptive of airborne MIMO radar clutter offsets method with embodiment 1-3, because pre-filtering processing of the present invention has suppressed most of clutter component, reduced Clutter Degrees of Freedom, when pre-filtering output data are carried out to self-adaptive processing, to there is more degree of freedom in system to suppress residual spur noise component, the adaptive performance of raising method, thereby the dimensionality reduction STAP of less degree of freedom also can obtain comparatively desirable performance, in order to embody better the performance of the inventive method, the empty time-code three-dimensional adaptive of airborne MIMO radar clutter is offseted to the method conventional MTI method of cascade and cascade dimensionality reduction STAP method respectively, associating clutter suppresses, specific implementation process is as follows:

1) the conventional MTI method of cascade

Conventional MTI(Conv.MTI) method is traditional moving target detection method comparatively conventional in airborne early warn ing radar, it is by the matched filtering of Space-Time bidimensional, realizes the detection to moving-target.

Hypothetical target steering vector is

$s=s_t\⊗s_s,$

Wherein s _tfor target time domain steering vector, s _sfor target spatial domain steering vector,

for Kronecker amasss.The conventional MTI of bidimensional coupling while adopting sky, its weight vector is

Wherein

long-pending for Hadamard, it is that vectorial corresponding element is long-pending.H _tand h _sbe respectively the static weight vector in time domain and spatial domain, for example Chebyshev weighs, and is mainly used in reducing the secondary lobe of wave filter.

Carry out clutter and offset after operation, goal orientation vector will be

$\tilde{s}=\stackrel{\‾}{Q}s,$

The weight vector of corresponding conventional MTI is

${\tilde{w}}_{\mathrm{Conv}.\mathrm{MTI}}=\stackrel{\‾}{Q}{w}_{\mathrm{Conv}.\mathrm{MTI}},$

As shown in Fig. 6 and Fig. 9, conventional MTI performance after cascade self-adapting clutter canceller of the present invention has obtained obvious improvement.

2) cascade dimensionality reduction STAP method

Take comparatively ripe FA method as example, first FA method carries out time domain doppler filtering to each spatial domain passage, then each Doppler's passage is carried out to space domain self-adapted processing.FA method is as the representational method of one of dimensionality reduction STAP, and choosing of its dimensionality reduction matrix is that Doppler filter by Doppler's passage to be detected forms.Suppose that passage to be detected is k Doppler's passage, its filter vector is that a dimension is the column vector t of K × 1 _k=[1, exp (j2 π f _k) ..., exp (j2 π (K-1) f _k)] ^t, wherein f _kfor passage Doppler frequency to be detected.The form of dimensionality reduction matrix is $T_{\mathrm{FA}}=t_k\⊗I_{\mathrm{NM}}\∈C^{\mathrm{NMK}\×\mathrm{NM}}.$

For the situation that does not adopt clutter to offset, to the data x ∈ C receiving ^{nMK × 1}with signal guide vector s ∈ C ^{nMK × 1}directly carry out obtaining after dimension-reduction treatment

$\left\{\begin{array}{c}{x}_{\mathrm{FA}}=T_{\mathrm{FA}}^{H}x\∈C^{\mathrm{NM}\×1}\\ s_{\mathrm{FA}}=T_{\mathrm{FA}}^{H}s\∈C^{\mathrm{NM}\×1}\end{array}\right.,$

Wherein x _fA, s _fAfor reception data and the signal guide vector of dimensionality reduction, it obtains after treating sense channel doppler filtering.Data after dimensionality reduction and signal guide vector are carried out to self-adaptive processing, can obtain the weight vector of FA

$w_{\mathrm{FA}}=R_{\mathrm{FA}}^{-1}{s}_{\mathrm{FA}}\∈C^{\mathrm{NM}\×1},$

Wherein

for dimensionality reduction data x _fA∈ C ^{nM × 1}correlation matrix.

When the clutter that adopts the present invention to propose offsets method, data are carried out to pre-filtering processing, the reception data and the signal guide vector that obtain after pre-filtering are

with

it is to be noted, the Doppler filter of the passage to be detected in FA method will change, that is that umber of pulse will reduce one because carrying out clutter offsets after pre-filtering, and therefore the filter vector of k Doppler's passage will become the column vector of a dimension for (K-1) × 1 ${\tilde{t}}_k={[1,\exp \left(j2\mathrm{\π}{f}_k\right),\·\·\·,\exp \left(j2\mathrm{\π}(K-2)f_k\right)]}^T$ The form of dimensionality reduction matrix is

to the reception data after pre-filtering with signal guide vector

carry out dimension-reduction treatment, obtain

$\left\{\begin{array}{c}{\tilde{x}}_{\mathrm{FA}}={\tilde{T}}_{\mathrm{FA}}^H\tilde{x}\∈C^{\mathrm{NM}\×1}\\ {\tilde{s}}_{\mathrm{FA}}={\tilde{T}}_{\mathrm{FA}}^H\tilde{s}\∈C^{\mathrm{NM}\×1}\end{array}\right.,$

For above dimensionality reduction data and signal guide vector, carry out self-adaptive processing, the weight vector that obtains FA method is

${\tilde{w}}_{\mathrm{FA}}={\tilde{R}}_{\mathrm{FA}}^{-1}{\tilde{s}}_{\mathrm{FA}}\∈C^{\mathrm{NM}\×1},$

Wherein ${\tilde{R}}_{\mathrm{FA}}\∈C^{\mathrm{NM}\×\mathrm{NM}}$ Data ${\tilde{x}}_{\mathrm{FA}}\∈C^{\mathrm{NM}\×1}$ Correlation matrix.

${\tilde{R}}_{\mathrm{FA}}=E\left({\tilde{x}}_{\mathrm{FA}}{\tilde{x}}_{\mathrm{FA}}^H\right)={\tilde{T}}_{\mathrm{FA}}^H\tilde{R}{\tilde{T}}_{\mathrm{FA}}={\tilde{T}}_{\mathrm{FA}}^H\hat{Q}R{\hat{Q}}^H{\tilde{T}}_{\mathrm{FA}},$

As can be seen from the above equation, clutter cancellation cascade FA method is the operation of two-stage dimensionality reduction.First order dimensionality reduction is by clutter pre-filtering matrix of the present invention

realize, it has suppressed the energy of clutter; The dimensionality reduction of the second level is the dimensionality reduction matrix by FA method

realize.Due to the first order dimensionality reduction of the present invention's proposition

suppress most clutter energy, made the dimensionality reduction STAP method of the second level can better suppress remaining clutter, thereby improved the detection performance to target.

Embodiment 5:

The empty time-code three-dimensional adaptive of airborne MIMO radar clutter offsets method with embodiment 1-4, below by emulation, the present invention is explained again given following simulated conditions:

Airborne MIMO transmitting radar antenna is counted M=5, transmitting array element distance d _m=0.8m, receives array number N=8, receives array element distance d _n=0.1m, radar wavelength λ=0.2m, pulse repetition rate f _r=2400Hz, once the relevant umber of pulse of processing in the time interval (CPI) is K=16.Carrier aircraft height H=6000m, flying speed v=120m/s, sensing point is apart from carrier aircraft radial distance R=100Km.Noise power

miscellaneous noise ratio is CNR=60dB.There are a moving target, position angle in the positive side of carrier aircraft

normalization Doppler frequency f _d0=0.25, signal to noise ratio snr=0dB, by 0 position angle to π scope

be divided into 300 scattering units.

Definition p (f _s, f _d) be the power of clutter in space-time two-dimensional plane, the computing formula of Space-Time two dimension power (Capon) spectrum is

wherein s (f _s, f _d) be space-time two-dimensional search steering vector,

refer to the correlation matrix of clutter and noise, and in l-G simulation test

in also comprised the correlation matrix of the echo signal of inserting.Before clutter pre-filtering

wherein R is clutter and noise covariance matrix, R=E (xx ^h) ∈ C ^{nMK × NMK}, R _sfor echo signal correlation matrix.After clutter pre-filtering, as Fig. 3 and Fig. 8, by the comparison to clutter power spectrum before and after clutter pre-filtering, after can seeing intuitively and offseting by clutter cancellation of the present invention, clutter power is effectively suppressed, and echo signal highlights.

Be commonly used to describing method or method performance quality.This improvement factor refers to output letter miscellaneous noise ratio and the ratio of inputting letter miscellaneous noise ratio, in l-G simulation test, and the normalization spatial domain frequency of given improvement factor curve the computing formula of improvement factor is $\mathrm{IF}=\frac{{\mathrm{SCNR}}_{\mathrm{out}}}{{\mathrm{SCNR}}_{\mathrm{in}}}=\frac{\left|w^{H}s({\hat{f}}_s,f_d)\right|(\mathrm{CNR}+1){\mathrm{\σ}}_w^2}{w^H{R}^{-1}w},$ The weight vector of disposal route when wherein w is sky.In experiment by more conventional MTI method, the conventional MTI of FA method and EFA method and the STACC+ after cascade clutter cancellation of the present invention, the performance of these 6 kinds of methods of STACC+FA and STACC+EFA.Conventional MTI method spatial domain and time domain all adopt Chebyshev window weighting, to reduce secondary lobe impact.Wherein spatial domain is 40dB Chebyshev power, and time domain is 60dB Chebyshev power.Doppler's passage that FA method is chosen target place carries out space domain self-adapted processing.EFA method, except choosing the Doppler's passage of target place, is also combined 2 adjacent Doppler's passages and is carried out 3 channel combined self-adaptive processing.The range unit that FA method and EFA method are all chosen range unit to be detected both sides is symmetrically sample, and selected sample number is the twice of its processor degree of freedom, the sample number that wherein FA method is chosen is 2NM=80, and the sample number that EFA method is chosen is 6NM=240.By the inventive method at positive side-looking radar in noise signal is processed, the result before and after processing is contrasted, and result after treatment and prior art are compared.

1, first with Matlab, draw in the airborne MIMO radar of positive side-looking motion platform situation the variation diagram that uses the space-time two-dimensional clutter power before and after clutter piece canceller clutter reduction when empty, as shown in Figure 3.From scheming, can see that clutter power distributes along the diagonal line of Space-Time two dimensional surface, clutter offsets rear clutter energy and has obtained inhibition.Before clutter offsets, echo signal seems very little compared to noise signal, is not easy to be detected.After clutter offsets, clutter energy is effectively suppressed, and makes echo signal more obvious compared with noise signal, more detected.

Clutter feature spectrogram when 2, Fig. 4 is sky before and after filtering; Large eigenwert part in figure represents the power of clutter, and little eigenwert represents the power of noise, as can be seen from the figure, during through sky after filtering, large eigenwert declines clearly, although noise power has a bit, improves, and clutter power has obtained great inhibition on the whole.In addition, the minimizing of the large eigenwert of clutter also shows the reduction of clutter order, and Clutter Degrees of Freedom reduces, and therefore the adaptive approach of subsequent cascaded has more degree of freedom to carry out clutter reduction.

3, Fig. 6 is the positive side-looking radar improvement factor comparison diagram before and after MTI, FA and EFA method cascade the present invention (STACC); As can be seen from the figure conventional MTI method, the performance after FA method and EFA method cascade the present invention (self-adapting clutter canceller) all in various degree be better than there is no cascade situation of the present invention.At normalization Doppler frequency f _d=0.1 main clutter district, MTI, FA and EFA have respectively 19.5dB, the improvement of 40.85dB and 2.24dB after cascade the present invention.At normalization Doppler frequency f _d=0.4 sidelobe clutter district, MTI, FA and EFA have respectively 7.63dB, the improvement of 2.45dB and 1.2dB after cascade the present invention.In these three kinds of methods, to the improvement of EFA performance, be minimum, this is because EFA method has been the method that approaches accurate suboptimum.In addition, three kinds of methods are all obvious than the improvement in sidelobe clutter district at main clutter district performance improvement, and this is by the detection being more conducive to slower-velocity target.

4, Fig. 7 is the positive side-looking radar improvement factor comparison diagram of the present invention (STACC) and TDPC difference cascade MTI, FA and EFA method; As can be seen from the figure the present invention (STACC) makes performance have a certain upgrade more because having utilized echo amplitude information compared with TDPC.

To sum up, adopt the inventive method to carry out the performance that clutter offsets and be better than adopting existing TDPC method in positive side-looking radar, the inventive method can be than the more effective clutter reduction of existing TDPC method.

Embodiment 6:

The empty time-code three-dimensional adaptive of airborne MIMO radar clutter offsets method with embodiment 1-5, and simulated conditions, with embodiment 5, uses the inventive method at non-working side radar

in noise signal is processed,

Result before and after processing is contrasted, and result after treatment and prior art are compared.

1, first with Matlab, draw the variation diagram that uses the space-time two-dimensional clutter power before and after clutter reduction of the present invention in the airborne MIMO radar of non-working side motion platform situation, as shown in Figure 8.From scheming, can see that clutter power spectrum distribution track is semiellipse at Space-Time two dimensional surface and distributes.After clutter offsets, clutter energy has obtained inhibition along the distribution track of clutter.

2, Fig. 5 is the clutter feature spectrogram that uses filtering of the present invention front and back in the airborne MIMO radar of non-working side motion platform situation; As can be seen from the figure, clutter energy has obtained great inhibition.In addition, clutter order has also reduced a lot, and Clutter Degrees of Freedom reduces, and therefore the adaptive approach of subsequent cascaded has more degree of freedom to carry out clutter reduction.

3, Fig. 9 is the non-working side radar improvement factor comparison diagram before and after MTI, FA and EFA method cascade the present invention (STACC); As can be seen from the figure conventional MTI method, the performance after FA method and EFA method cascade the present invention (STACC) also all in various degree be better than the situation that there is no cascade the present invention (STACC).At normalization Doppler frequency f _d=-0.25 main clutter district, MTI, FA and EFA have respectively 23.49dB after cascade the present invention (STACC), the improvement of 39.2dB and 1.34dB.At normalization Doppler frequency f _d=0.15 sidelobe clutter district, MTI, FA and EFA have respectively 4.78dB after cascade the present invention (STACC), the improvement of 2.59dB and 1dB.Identical with positive side-looking radar situation, in these three kinds of methods, to the improvement of EFA performance, be also minimum.

4, Figure 10 is the non-working side radar improvement factor comparison diagram of the present invention (STACC) and TDPC difference cascade MTI, FA and EFA method; Can find out, the present invention (STACC) is more a little better than the performance of TDPC, and that is that the important information that forms it mainly contains clutter scattering point azimuth information, doppler information and the echo amplitude information in Range resolution unit because for clutter covariance matrix.TDPC has utilized the azimuth information of clutter and doppler information to carry out clutter reduction very fully, the inventive method according to the adaptive principle echo amplitude information of having utilized clutter on this basis more, from the present embodiment, in the contrast of two experiments, can find out, Performance Ratio TDPC has lifting.

To sum up can draw, the inventive method had both been applicable to positive side-looking radar, was applicable to again non-working side radar, and in both, all had good clutter cancellation performance performance.Before using MTI, FA and EFA method, adopt the inventive method to carry out pre-filtering to clutter, can obtain low interference filter data, can improve to a certain extent the detection performance of target, and make subsequent cascaded method performance be improved significantly.

In brief, the empty time-code three-dimensional adaptive of the airborne MIMO radar of one of the present invention clutter cancellation, main contents comprise: the airborne MIMO radar echo signal of model model, calculate residual echo signal energy and minimized, obtain self-adapting clutter canceller matrix of coefficients, by sliding window, arrange and then obtain sky time-code three-dimensional adaptive clutter cancellation design result, utilize empty time-code three-dimensional adaptive clutter cancellation to offset echoed signal, by the pre-service of empty time-code three-dimensional adaptive clutter cancellation, eliminate most of clutter.When the inventive method is empty for routine, before disposal route and dimensionality reduction STAP method, the clutter of airborne MIMO radar echo signal being carried out to sky time-code three-dimensional offsets, can, at just effective clutter reduction of the first stage of Radar Signal Processing, improve the detection performance of moving-target.It has realized the popularization to airborne motion platform by traditional ground radar clutter canceller, and the present invention is not only applicable to positive side-looking radar, for non-working side radar, is suitable for too, all has good moving-target detectability.

Claims (4)

1. the empty time-code three-dimensional adaptive of an airborne MIMO radar clutter offsets method, it is characterized in that: utilize the echoed signal of adjacent two pulses of airborne MIMO radar to subtract each other and offset operation, noise signal after offseting is inhibited, facilitate the follow-up further processing of radar signal, described self-adapting clutter cancellation process comprises the steps:

Step 1. is set airborne MIMO radar antenna and is launched K pulse relevant a processing in the time interval, builds the echo of k reception of impulse of airborne MIMO radar, and this echo comprises clutter and noise, and signal model is:

$x_k={\mathrm{AD}}_k{\mathrm{\ρ}}_k+x_k^{\left(w\right)},$

In formula, A is the matrix that spatial domain phase term forms, D _kfor the diagonal matrix that time domain doppler phase item forms, ρ _kbe the random magnitude vector of k pulse moment U clutter scattering unit, U is the number of a range unit inscattering unit, it is the white Gaussian noise vector in k pulse moment;

Step 2., according to echo signal model, calculates the residual echo signal energy of adjacent two pulses || ε _k|| ²for:

||ε _k|| ²＝||Qx _k-x _k+1|| ²，

In formula, Q is clutter cancellation matrix of coefficients, || || be vectorial 2-norm;

Step 3. is according to minimizing residual echo signal energy || ε _k|| ², obtain clutter cancellation matrix of coefficients Q and be:

$Q=R_{k+1,k}{R}_{k,k}^{-1},$

In formula, R _{k+1, k}be clutter and the noise correlation matrix of k pulse and k+1 pulse, R _k,kbe clutter and the noise autocorrelation matrix of k pulse, [] ^-1expression is to matrix inversion;

Step 4. is according to the clutter cancellation matrix of coefficients Q that tries to achieve, chooses in all radar signals one and relevantly processes in interval every adjacent two pulses and offset needed matrix of coefficients and slide window arrangement, obtains self-adapting clutter canceller, this canceller

be expressed as:

I in formula _nM∈ C ^{nM × NM}be unit matrix, M is aerial array transmitting element number of array, and N is antenna reception element number of array;

The self-adapting clutter canceller that step 5. utilization obtains the relevant all echo x that process in the time interval in radar signal are carried out to clutter and offset, the radar signal matrix that obtains containing noise signal and remaining noise signal after offseting is

$\tilde{x}=\stackrel{\‾}{Q}x,$

In formula, x is the long vector that tie up NMK × 1 that relevant data of processing the pulse of K altogether receiving in interval are aligned to, $x={\left[\begin{array}{cccc}{x}_1^T,& x_2^T,& \·\·\·,& x_K^T\end{array}\right]}^T,$

the input signal of processing during for cascade sky.

2. the empty time-code three-dimensional adaptive of airborne MIMO radar according to claim 1 clutter offsets method, it is characterized in that: the echo signal model of k the pulse that the airborne MIMO radar of structure described in step 1 receives, comprises the steps:

1.1 build the individually clutter echoed signal of area scattering point; Suppose that carrier aircraft flying height is H, speed is v, radar transmit and receive battle array for the uniform line-array of parallel placement, aerial array transmitting array number is M, transmitting array element distance is d _m, reception array number is N, reception array element distance is d _n, radar operation wavelength λ, pulse repetition rate f _rrelevant a processing in interval, each transmitting array element is launched the train of impulses waveform being comprised of K pulse simultaneously, and M transmitted waveform is mutually orthogonal, at each reception array element place, by M reference transmitted signal, the echo data of K recurrence interval is carried out to matched filtering respectively, for one, apart from carrier aircraft position angle, be

, the ground scatter point that the angle of pitch is θ, at n array element place of receiving end, the clutter echoed signal obtaining under corresponding m transmitting array element, a k pulse is

N=0 in formula ..., N-1, m=0 ..., M-1, k=0 ..., K-1,

for the random echo amplitude of this clutter scattering point, obey the multiple Gaussian distribution of zero-mean,

for receiving spatial domain frequency item, for receiving spatial domain frequency, for transmitting spatial domain frequency item, for transmitting spatial domain frequency,

for Doppler frequency item,

for Doppler frequency;

1.2 build all echoed signals that in individual pulse, radar receives; Suppose that radar detection point is R apart from carrier aircraft radial distance, all noise signals that in a pulse, radar receives are the echo sum of all scattering points in Range resolution unit, one, ground, along position angle

integration from 0 to π (this range unit is the intersection of the sphere take R as radius and ground level centered by carrier aircraft)

In formula, for clutter item,

for noise item, be to obey 0 average, variance is

white Gaussian noise, the clutter c of scattering point _{n, m, k}with the Gaussian noise in above formula

it is independent that relation is added up each other;

All echoed signal discrete processes that in 1.3 pairs of individual pulses, radar receives; By a range unit along position angle

from 0 to π, the echoed signal that the discrete U of a being divided into scattering unit obtains a pulse is:

In formula

be the position angle of i scattering unit,

for the clutter echo amplitude of corresponding i scattering unit,

be comprise transmitting, the spatial domain item in being received in,

for spatial domain frequency,

for Doppler's item,

for Doppler frequency, U is the number of a range unit inscattering unit;

1.4 build the echo signal model of k pulse; The echo of k the pulse that radar is received is arranged in the column vector of NM × 1 dimension, x _k=[x _{0,0, k}, x _{1,0, k}..., x _{n-1, M-1, k}] ^t,

$x_k={\mathrm{AD}}_k{\mathrm{\ρ}}_k+x_k^{\left(w\right)},$

In formula

$x_k^{\left(w\right)}={\left[\begin{array}{ccc}{x}_{\mathrm{0,0},k}^{\left(w\right)},& \·\·\·,& x_{N-1,M-1,k}^{\left(w\right)}\end{array}\right]}^T,$

Wherein A is the matrix that spatial domain phase term forms, its each classify the column vector of length N M × 1 dimension as, be that the column vector that receives dimension N × 1 carries out with the column vector of transmitting dimension M × 1 that Kronecker is long-pending to be obtained, D _kbe the diagonal matrix that k pulse time domain doppler phase item forms, diag () represents vector to be become to diagonal matrix, ρ _kbe the random magnitude vector of k pulse moment U clutter scattering unit,

be the white Gaussian noise vector in k pulse moment, [] ^trepresent vectorial transposition;

The echo in k+1 moment, this echo comprises clutter and noise, is written as

$x_{k+1}=A\stackrel{\‾}{D}{D}_k{\mathrm{\ρ}}_{k+1}+w_{k+1}^{\left(w\right)},$

Wherein

for the doppler phase item producing in individual pulse, wherein

for doppler phase,

for crab angle.

3. the empty time-code three-dimensional adaptive of airborne MIMO radar according to claim 2 clutter offsets method, it is characterized in that: described in step 2 according to echo signal model, calculate the residual echo signal energy of adjacent two pulses || ε _k|| ², comprise the steps:

2.1 k pulse echo signals that receive according to airborne MIMO radar, the echoed signal that obtains k+1 pulse is

x _k+1＝[x _0,0,k+1, x _1,0,k+1, …, x _N-1,M-1,k+1] ^T，

2.2 carry out obtaining after clutter offsets the energy of residual echo signal by the echo of the echo of k+1 pulse and k pulse || ε _k|| ²for

||ε _k|| ²＝||Qx _k-x _k+1|| ²，

In formula, Q is clutter cancellation matrix of coefficients, || || be vectorial 2-norm.

4. the empty time-code three-dimensional adaptive of airborne MIMO radar according to claim 3 clutter offsets method, it is characterized in that: described in step 3, minimize residual echo signal energy || ε _k|| ², obtain matrix of coefficients Q, comprise the steps:

3.1 by residual echo signal energy || ε _k|| ²minimize, ask

in formula, min represents minimization function:

$\begin{array}{c}\underset{Q}{\min }E\left({\left|\right|{\mathrm{\ϵ}}_k\left|\right|}^2\right)=\underset{Q}{\min }E\left({\left|\right|Q{x}_k-x_{k+1}\left|\right|}^2\right)\\ =\underset{Q}{\min }E\left[{(Q{x}_k-x_{k+1})}^H(Q{x}_k-x_{k+1})\right]\\ =\underset{Q}{\min }E\left\{\mathrm{tr}\right[(Q{x}_k-x_{k+1}){({\mathrm{Qx}}_k-x_{k+1})}^H\left]\right\}\\ =\underset{Q}{\min }\mathrm{tr}(Q{R}_{k,k}{Q}^H-{\mathrm{QR}}_{k,k+1}\\ -R_{k+1,k}{Q}^H+R_{k+1,k+1}\left)\right]\\ =\underset{Q}{\min }J\left(Q\right)\end{array},$

E () tabular form mathematical expectation in formula, $R_{k,k}=E\left(x_k{x}_k^H\right),$ $R_{k,k+1}=E\left(x_k{x}_{k+1}^H\right),$ $R_{k+1,k}=E\left(x_{k+1}{x}_k^H\right)$ With $R_{k+1,k+1}=E\left(x_{k+1}{x}_{k+1}^H\right)$ For clutter and noise correlation matrix, () ^hthe complex-conjugate transpose of representing matrix or vector;

3.2 ask conjugation derivative by cost function about clutter cancellation matrix of coefficients, and make derivative $\∂J\left(Q\right)/\∂Q*=0,$

$\∂J\left(Q\right)/\∂Q*=\∂\left[\mathrm{tr}({\mathrm{QR}}_{k,k}{Q}^H-{\mathrm{QR}}_{k,k+1}-R_{k+1,k}{Q}^H+R_{k+1,k+1})\right]/\∂Q*={\mathrm{QR}}_{k,k}-R_{k+1,k}=0,$

Solution is $Q=R_{k+1,k}{R}_{k,k}^{-1},$

Clutter in formula and noise correlation matrix R _{k+1, k}, R _k,kand matrix of coefficients Q, respectively by its maximal possibility estimation

with

replace, can be estimated to obtain by following formula:

$\left\{\begin{array}{c}{\hat{R}}_{k,k}=\frac{1}{L(K-1)}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}\underset{k=1}{\overset{K-1}{\mathrm{\Σ}}}{x}_k\left(l\right)x_k^H\left(l\right)\\ {\hat{R}}_{k+1,k}=\frac{1}{L(K-1)}\underset{l=1}{\overset{L}{\mathrm{\Σ}}}\underset{k=1}{\overset{K-1}{\mathrm{\Σ}}}{x}_{k+1}\left(l\right)x_k^H\left(l\right)\\ \hat{Q}={\hat{R}}_{k+1,k}{\hat{R}}_{k,k}^{-1}\end{array}\right.,$

In formula, x _k(l) echo of k the pulse that l range unit of expression receives.

Images