CN103728595B - Networking radar based on subspace projection suppresses pressing type major lobe suppression method - Google Patents

Abstract

The invention discloses a kind of networking radar based on subspace projection and suppress pressing type major lobe suppression method, mainly solve the problem that monostatic radar can only suppress a kind of interference type.Implementation step is: 1. calculate the baseband receiving signals of each node radar, and it to be as the criterion alignment with undesired signal in time domain, obtains node array radar signal; 2. estimate the covariance matrix of node array radar signal, feature decomposition is carried out to it, and according to feature decomposition result structure noise subspace, calculate the projection matrix to its projection; 3. node array radar signal is projected to noise subspace, obtain the projection vector for target detection; 4. according to the projection vector being used for target detection, structure Generalized Likelihood Ratio function; 5. set detection threshold, and the value of each for Generalized Likelihood Ratio function moment point is compared with detection threshold, obtain the Output rusults of target detection.The present invention can effectively suppress dissimilar undesired signal, can be used for networking radar system.

Description

Networking radar based on subspace projection suppresses pressing type major lobe suppression method

Technical field

The present invention relates to Radar Technology field, in particular to a kind of method of resistance to compression standard major lobe suppression, can be used for networking radar system, under pressing type major lobe suppression condition, the undesired signal entered from networking radar each node radar antenna main lobe is effectively suppressed, improves the target detection performance of networking radar under pressing type major lobe suppression condition.

Background technology

Suppress interfere is a kind of common radar jamming pattern, by frequency directing, make the centre frequency of the centre frequency of jamming transmitter and radar close, high-power undesired signal is entered radar receiver, thus flood target echo, make radar be difficult to target information be detected from reception echo.Suppress interfere is divided into active suppressing formula to disturb and passive suppress interfere.The interference of active suppressing formula is divided into again spot jamming and barrage jamming; Passive suppress interfere is a large amount of chaff delivery mainly, causes strong interference, covers echo signal.

For the suppress interfere being positioned at target proximity, undesired signal can enter from radar receiving antenna main lobe usually; When radar detection distant object, even if jammer range target is comparatively far away, undesired signal also can fall within the main lobe of radar receiving antenna, thus forms pressing type major lobe suppression.Existing adaptive backstepping method technology, Adaptive beamformer technology etc. can only suppress the interference entered from radar receiving antenna secondary lobe, major lobe suppression can cause serious wave beam distortion or main lobe peak bias, effectively cannot detect while suppression interference to target.

Networking radar, refers to the organic radar netting be made up of the node radar that multi-section system is identical or different, and it has mode of operation and collaborative detection mode flexibly, can provide abundant sky, time, frequently resource.

Along with the array of potato masher antenna, jammer has the ability of Multibeam synthesis and system resource scheduling, interference can be implemented to multi-section radar simultaneously, networking radar is faced with the threat of major lobe suppression, in network, each node radar all can be subject to the impact of major lobe suppression, and the detection perform of networking radar can decline rapidly.

For pressing type major lobe suppression, have already been proposed the method that some monostatic radars suppress pressing type major lobe suppression, can be divided three classes by its thinking: (1) utilizes parametric method to recover undesired signal, completes interference cancellation in time domain; (2) frequency domain trap or Time-frequency Filter filtering interfering is utilized; (3) subspace projection based on time domain data vector is utilized to realize AF panel.But said method is all that time domain, frequency domain or the time-frequency domain architectural feature according to undesired signal designs, and has stronger interference type specific aim.When interference type mismatch, its interference rejection capability will reduce and even lost efficacy.

Summary of the invention

The object of the invention is to for above-mentioned suppression pressing type major lobe suppression Problems existing, propose a kind of networking radar based on subspace projection and suppress pressing type main lobe method, to realize the suppression to networking radar pressing type major lobe suppression, improve the target detection performance of networking radar.

For achieving the above object, technical scheme of the present invention comprises the steps:

(1) hypothetical network radar is made up of M node radar, first node radar is operated in sending and receiving state, and all the other node radars are only operated in accepting state, the beam main lobe of each node radar all points to jammer and target region, a suppress interfere is only there is in radar detection area, near suppress interfere, there is Q real goal, according to target, disturb space geometry position relative to networking radar, calculate the baseband receiving signals r of the n-th node radar _n(t):

r _n(t)＝r _nT(t)+r _nJ(t)+w _n(t)，

Wherein, r _nTt () is the baseband receiving signals of the n-th node radar target, $r_{\mathrm{nT}}\left(t\right)=\underset{k=1}{\overset{Q}{\mathrm{\Σ}}}{\mathrm{\α}}_{k,n}\·s_{\mathrm{TX}}(t-\frac{R_{k,1}^{\left(T\right)}+R_{k,n}^{\left(T\right)}}{c})\·e^{-j2\mathrm{\π}(R_{k,1}^{\left(T\right)}+R_{k,n}^{\left(T\right)})/\mathrm{\λ}},$ α _k,nthe echo of kth target complex magnitude when arriving the n-th node radar, k=1,2 ..., Q, Q are the numbers of extraterrestrial target, n=1,2 ..., M, M are the numbers of node radar, M>=2, s _tXt () is base band transmit, the distance that a kth target arrives the 1st node radar, be the distance of a kth target to the n-th node radar, c is the light velocity, and λ is radar operation wavelength, r _nJt () is the baseband receiving signals of the n-th node radar jamming, β _nbe the complex magnitude of undesired signal when arriving the n-th node radar, J (t) is baseband interference signal, the distance that jammer arrives the n-th node radar, w _nt () is the n-th node noise of radar receiver signal, and the noise signal of each node radar receiver is uncorrelated mutually;

(2) baseband receiving signals of each node radar step (1) obtained, time domain is as the criterion with the interference echo signal of first node radar and aligns, the form of signal after alignment with vector is represented, obtains node array radar signal vector r (t):

r(t)＝[r ₁(t),r ₂(t-τ ₁₂),…,r _M(t-τ _1M)] ^T＝S(t)+J(t)+w(t),

Wherein, τ _1nthe time delay of interference echo signal relative to the interference echo signal of first node radar of the n-th node radar, represent variate-value when making objective function get maximal value, represent convolution algorithm, () ^*represent and get conjugation, () ^trepresent and get transposition, S (t) is the target echo signal vector of node radar array, a _tthe steering vector of target relative to node radar array, $a_T={[e^{-j4\mathrm{\π}{R}_{k,1}^{\left(T\right)}/\mathrm{\λ}},e^{-j2\mathrm{\π}(R_{k,1}^{\left(T\right)}+R_{k,2}^T)/\mathrm{\λ}},\·\·\·,e^{-2\mathrm{\π}(R_{k,1}^{\left(T\right)}+R_{k,M}^{\left(T\right)})/\mathrm{\λ}}]}^T,$ α _kthe complex amplitude vector of node radar array target echo, α _k=[α _{k, 1}, α _{k, 2}..., α _k,M] ^t, represent point multiplication operation, J (t) is the interference echo signal phasor of node radar array, a _jthe steering vector of jammer relative to node radar array β is the complex amplitude vector of node radar array interference echo, β=[β ₁..., β _m] ^t, w (t) is the noise signal vector of node radar array;

(3) according to node array radar signal vector r (t) that step (2) obtains, its covariance matrix is estimated

$\hat{R}=\underset{i=1}{\overset{T}{\mathrm{\Σ}}}r\left(i\right)r^H\left(i\right),$

Wherein, i=1,2 ..., T, T are for training covariance matrix time-domain sampling sample number, () ^hrepresent conjugate transpose;

(4) by covariance matrix that following formula obtains step (3) carry out feature decomposition, obtain covariance matrix eigenvalue of maximum characteristic of correspondence vector v ₁with all the other eigenwert characteristic of correspondence vector v ₂~ v _m:

$\hat{R}=({\mathrm{\σ}}_J^2+{\mathrm{\σ}}_w^2)v_1{v}_1^H+\underset{j=2}{\overset{M}{\mathrm{\Σ}}}{\mathrm{\σ}}_w^2{v}_j{v}_j^H,$

Wherein, j=2 ..., M, v _jrepresent a jth proper vector;

(5) the proper vector v that step (4) obtains is utilized ₂~ v _m, structure noise subspace, calculates the projection matrix P to noise subspace projection _nr;

(6) node array radar signal vector r (t) that step (2) obtains is projected in the noise subspace that step (5) constructs, obtain the projection vector z being used for target detection _r(t):

z _r(t)＝P _nrr(t)，

Wherein, P _nrit is the projection matrix to noise subspace projection;

(7) to the projection vector z that step (6) obtains _rt () carries out target detection with generalized likelihood test device, structure Generalized Likelihood Ratio function Λ (t):

$\mathrm{\Λ}\left(t\right)=r{\left(t\right)}^H{P}_{\mathrm{nr}}r\left(t\right);$

(8) detection threshold is set: $\mathrm{\δ}=({\mathrm{\σ}}_w^2/2)F_{{\mathrm{\χ}}_{2(M-1)}^2}^{-1}(1-P_{\mathrm{fa}}),$

Wherein, represent that degree of freedom is card side's distribution of 2 (M-1), for the inverse function of distribution function, P _fait is false-alarm probability;

(9) functional value of each moment point of Generalized Likelihood Ratio function Λ (t) and detection threshold δ are compared, obtain the Output rusults of target detection: if Λ (t) < is δ, represent driftlessness, if Λ (t) > is δ, indicate target.

The present invention compared with prior art tool has the following advantages:

1, compared with the method suppressing pressing type major lobe suppression with monostatic radar, the present invention is owing to have employed networking radar configuration node radar array, the method to subspace projection is used to suppress pressing type major lobe suppression, therefore do not rely on the time-frequency structure feature of undesired signal, dissimilar undesired signal can be applicable to, when interference type mismatch, also there is good interference rejection capability.

2, the present invention is owing to have employed the method for subspace projection, and when applying without the need to knowing the parameters such as amplitude phase error between the geometric parameter of networking radar node array and each node radar, therefore the change of pair array inner structure has stronger adaptive ability.

Accompanying drawing explanation

Fig. 1 is realization flow figure of the present invention;

Fig. 2 is the scene schematic diagram that the present invention uses;

Fig. 3 suppresses FM Noise Jamming Signal by the inventive method, the likelihood ratio function output map of emulation;

Fig. 4 suppresses noise convolution smart munition signal by the inventive method, the likelihood ratio function output map of emulation.

Embodiment

With reference to Fig. 1, specific embodiment of the invention step is as follows:

Step 1: according to target, disturb locus relative to networking radar, calculate the baseband receiving signals r of the n-th node radar _n(t).

1a) according to the locus of target relative to networking radar, calculate the baseband receiving signals r of the n-th node radar target _nT(t):

$r_{\mathrm{nT}}\left(t\right)=\underset{k=1}{\overset{Q}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}_{k,n}\&CenterDot;s_{\mathrm{TX}}(t-\frac{R_{k,1}^{\left(T\right)}+R_{k,n}^{\left(T\right)}}{c})\&CenterDot;e^{-j2\mathrm{\&pi;}(R_{k,1}^{\left(T\right)}+R_{k,n}^{\left(T\right)})/\mathrm{\&lambda;}},$

Wherein, α _k, _nthe echo of kth target complex magnitude when arriving the n-th node radar, α _k, _nobeying average is 0, and variance is σ _α ²multiple Gaussian random process, σ _α ²the average power of target echo, k=1,2 ..., Q, Q are the numbers of extraterrestrial target, n=1,2 ..., M, M are the numbers of node radar, M>=2, s _tXt () is base band transmit, the distance that a kth target arrives the 1st node radar, be the distance of a kth target to the n-th node radar, c is the light velocity, and λ is radar operation wavelength;

1b) according to the locus of interference relative to networking radar, calculate the baseband receiving signals r of the n-th node radar jamming _nJ(t):

$r_{\mathrm{nJ}}\left(t\right)={\mathrm{\&beta;}}_n\&CenterDot;J(t-\frac{R_n^{\left(J\right)}}{c})\&CenterDot;e^{-j2\mathrm{\&pi;}{R}_n^{\left(J\right)}/\mathrm{\&lambda;}},$

Wherein, β _nbe the complex magnitude of undesired signal when arriving the n-th node radar, J (t) is baseband interference signal, it is the distance that jammer arrives the n-th node radar;

1c) according to the baseband receiving signals r of the n-th node radar target _nTthe baseband receiving signals r of (t) and interference _nJ(t), and the noise signal w of the n-th node radar _nt (), obtains the baseband receiving signals r of the n-th node radar _n(t):

r _n(t)＝r _nT(t)+r _nJ(t)+w _n(t)，

Wherein, noise signal w _nt () obeys average is 0, and variance is σ _w ²multiple Gaussian random process, σ _w ²be noise power, and the noise signal of each node radar is uncorrelated mutually.

Step 2: the baseband receiving signals of each node radar step (1) obtained, time domain is as the criterion with the interference echo signal of first node radar and aligns, obtain node array radar signal vector r (t):

r(t)＝[r ₁(t),r ₂(t-τ ₁₂),…r _n(t-t _1n)…,rM(t-τ _1M)] ^T，

Wherein, τ _1nthe time delay of interference echo signal relative to the interference echo signal of first node radar of the n-th node radar, represent τ when making objective function get maximal value _1n, represent convolution algorithm, () ^*represent and get conjugation, () ^trepresent and get transposition.

Step 3: be made up of signal, interference and noise three part according to node array radar signal vector signal, node array radar signal vector r (t) that step (2) obtains is rewritten as following expression:

r(t)＝S(t)+J(t)+w(t)，

Wherein, S (t) is the target echo signal vector of node radar array, aT is the steering vector of target relative to node radar array, $a_T={[e^{-j4\mathrm{\&pi;}{R}_{k,1}^{\left(T\right)}/\mathrm{\&lambda;}},e^{-j2\mathrm{\&pi;}(R_{k,1}^{\left(T\right)}+R_{k,2}^T)/\mathrm{\&lambda;}},\&CenterDot;\&CenterDot;\&CenterDot;,e^{-2\mathrm{\&pi;}(R_{k,1}^{\left(T\right)}+R_{k,M}^{\left(T\right)})/\mathrm{\&lambda;}}]}^T,$ α _kthe complex amplitude vector of node radar array target echo, α _k=[α _{k, 1}, α _{k, 2}, ... α _k,M] ^t, represent point multiplication operation, J (t) is the interference echo signal phasor of node radar array, a _jthe steering vector of jammer relative to node radar array, β is the complex amplitude vector of node radar array interference echo, β=[β ₁..., β _m] ^t, w (t) is the noise signal vector of node radar array.

Step 4: node array radar signal vector r (t) obtained according to step (3), estimates its covariance matrix

$\hat{R}=\underset{i=1}{\overset{T}{\mathrm{\&Sigma;}}}r\left(i\right)r^H\left(i\right),$

Wherein, i=1,2 ..., T, T are for training covariance matrix time-domain sampling sample number, () ^hrepresent the conjugate transpose to matrix;

Step 5: covariance matrix step (4) obtained by following formula carry out feature decomposition, obtain covariance matrix eigenvalue of maximum characteristic of correspondence vector v ₁with all the other eigenwert characteristic of correspondence vector v ₂~ v _m:

$\hat{R}=({\mathrm{\&sigma;}}_J^2+{\mathrm{\&sigma;}}_w^2)v_1{v}_1^H+\underset{j=2}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{\&sigma;}}_w^2{v}_j{v}_j^H,$

Wherein, j=2 ..., M, v _jrepresent a jth proper vector, represent jamming power;

Step 6: utilize the proper vector v that step (5) obtains ₂~ v _m, structure noise subspace U _nr:

U _nr＝[v ₂,…,v _M]。

Step 7: utilize noise subspace U _nr, calculate the projection matrix P to noise subspace projection _nr:

$P_{\mathrm{nr}}=U_{\mathrm{nr}}{U}_{\mathrm{nr}}^H.$

Step 8: the projection matrix P obtained according to step (7) _nr, node array signal phasor r (t) that step (3) obtains is projected in the noise subspace that step (6) constructs, obtains the projection vector z for target detection _r(t):

z _r(t)＝P _nrr(t)。

Step 9: the projection vector z that step (8) is obtained _rt () carries out target detection with generalized likelihood test device, structure Generalized Likelihood Ratio function Λ (t).

9a) according to the projection vector z that step (8) obtains _rt (), obtains the data x (t) after dimensionality reduction:

x(t)=Tz _r(t)，

Wherein, T is dimensionality reduction matrix,

9b) according to the data x (t) after dimensionality reduction, target detection is summed up as following Hypothesis Testing Problem:

H ₀:x(t)＝TP _nrw(t)，

H ₁:x(t)＝TP _nrS(t)+TP _nrw(t)，

Wherein, H ₀represent driftlessness, H ₁indicate target;

9c) according to the Hypothesis Testing Problem being used for target detection, structure Generalized Likelihood Ratio function Λ (t):

$\mathrm{\&Lambda;}\left(t\right)=\log \frac{P\left[x\left(t\right)\right|H_1,S\left(t\right)]}{P\left[x\left(t\right)\right|H_0]}=r{\left(t\right)}^H{P}_{\mathrm{nr}}r\left(t\right),$

Wherein, $P\left[x\left(t\right)\right|H_0]=\frac{1}{{\left(2\mathrm{\&pi;}\right)}^{(M-1)/2}|{{\mathrm{\&sigma;}}_w}^2\&CenterDot;T{P}_{\mathrm{nr}}{T}^H|}{e}^{-\frac{x{\left(t\right)}^H{\left(T{P}_{\mathrm{nr}}{T}^H\right)}^{-1}x\left(t\right)}{2{{\mathrm{\&sigma;}}_w}^2}},$ Represent at H ₀under supposing, the probability density function of x (t), () ^-1representing matrix is inverted, and P [x (t) | H ₁, S (t)] represent at H ₁under supposing, the probability density function of x (t), $P\left[x\left(t\right)\right|H_1,S\left(t\right)]=\frac{1}{{\left(2\mathrm{\&pi;}\right)}^{(M-1)/2}|{{\mathrm{\&sigma;}}_w}^2\&CenterDot;T{P}_{\mathrm{nr}}{T}^H|}{e}^{-\frac{\left[x{\left(t\right)-T{P}_{\mathrm{br}}S\left(t\right)]}^H{\left(T{P}_{\mathrm{nr}}{T}^H\right)}^{-1}\right[x\left(t\right)-T{P}_{\mathrm{nr}}S\left(t\right)]}{2{{\mathrm{\&sigma;}}_w}^2}},$ S (t) is the target echo signal vector of node radar array, and when constructing Generalized Likelihood Ratio function, S (t) uses its maximal possibility estimation usually replace, $\hat{S}\left(t\right)=P_{\mathrm{nr}}{U}_{\mathrm{nr}}{\left(T{U}_{\mathrm{nr}}\right)}^{-1}x\left(t\right).$

Step 10: setting detection threshold δ.

10a) according to Generalized Likelihood Ratio function Λ (t) that step (9) constructs, calculate Generalized Likelihood Ratio function Λ in aimless situation ₁(t):

${\mathrm{\&Lambda;}}_1\left(t\right)=w{\left(t\right)}^H{U}_{\mathrm{nr}}{U}_{\mathrm{nr}}^{H}w\left(t\right),$

10b) according to step 10a) Generalized Likelihood Ratio function Λ in the driftlessness situation that obtains ₁t noise signal vector w (t) of () and node radar array obeys multiple Gaussian distribution, obtain Λ ₁t () obeys card side's distribution that degree of freedom is 2 (M-1);

10c) according to step 10b) Λ that obtains ₁t () obeys card side's distribution that degree of freedom is 2 (M-1), obtain false-alarm probability P _faexpression formula as follows:

$\begin{array}{c}{P}_{\mathrm{fa}}=P[{\mathrm{\&Lambda;}}_1\left(t\right)>\mathrm{\&delta;}|H_0]\\ =P[{\mathrm{\&chi;}}_{2(M-1)}^2>(2\mathrm{\&delta;}/{\mathrm{\&sigma;}}_w^2)]\end{array},$

Wherein, represent that degree of freedom is card side's distribution of 2 (M-1);

10d) according to step 10c) the false-alarm probability P that obtains _faexpression formula, calculate detection threshold δ:

$\mathrm{\&delta;}=({\mathrm{\&sigma;}}_w^2/2)F_{{\mathrm{\&chi;}}_{2(M-1)}^2}^{-1}(1-P_{\mathrm{fa}}),$

Wherein, for the inverse function of distribution function.

Step 11: the functional value of each moment point of Generalized Likelihood Ratio function Λ (t) and detection threshold δ are compared, obtain the Output rusults of target detection: if Λ (t) < is δ, represent driftlessness, if Λ (t) > is δ, indicate target.

The present invention verifies by following emulation further to the rejection of pressing type major lobe suppression.

1. experiment scene:

As shown in Figure 2, networking radar is made up of six node radars, first node radar is operated in receipts, hair-like state, and all the other node radars are only operated in accepting state, each node radar is isomorphism radar, running parameter is identical, signal waveform is all linear FM signal, frequency of operation 1.4286GHz, sampling rate 12MHz, signal bandwidth 10MHz, pulse repetition rate 20KHz, pulsewidth 50us, beam angle 3.5294 °, the coordinate of six node radars is respectively: (0, 0) km, (2.21, 3.90) km, (-5.13, 4.47) km, (0,-4.24) km, (4.13,-1.26) km and (-2.37,-0.97) km, signal interference ratio is respectively :-40dB, 40.99dB,-40.07dB,-41.91dB,-41.32dB and-41.22dB, signal to noise ratio (S/N ratio) is 15dB, the coordinate of jammer is (0, 38) km, there are two real goals, target 1 coordinate is (0, 38.9) km, target 2 coordinate is (0, 37.1) km.

2. experiment content:

Experiment 1, when undesired signal is niose-modulating-frenquency jamming, adopt the inventive method, node array radar signal vector is projected to noise subspace, obtain the projection vector for target detection, generalized likelihood test is carried out to the projection vector of target detection, the likelihood ratio function output map of emulation, as shown in Figure 3.

Experiment 2, when undesired signal is noise convolution smart munition, adopt the inventive method, node array radar signal vector is projected to noise subspace, obtain the projection vector for target detection, generalized likelihood test is carried out to the projection vector of target detection, the likelihood ratio function output map of emulation, as shown in Figure 4.

3. interpretation:

Can see from Fig. 3 and Fig. 4, the echoed signal of the echoed signal of two real goals and jammer itself all clearly shows, show the present invention not only to the suppression that FM Noise Jamming Signal has been carried out effectively, and effective suppression has also been carried out to noise convolution smart munition, make target detection become possibility.

In sum, method of the present invention can effectively suppress dissimilar undesired signal, improves the detection perform of target.

Claims (3)

1. the networking radar based on subspace projection suppresses a pressing type major lobe suppression method, comprises the steps:

(1) hypothetical network radar is made up of M node radar, first node radar is operated in sending and receiving state, and all the other node radars are only operated in accepting state, the beam main lobe of each node radar all points to jammer and target region, a suppress interfere is only there is in radar detection area, near suppress interfere, there is Q real goal, according to target, disturb space geometry position relative to networking radar, calculate the baseband receiving signals r of the n-th node radar _n(t):

r _n(t)＝r _nT(t)+r _nJ(t)+w _n(t)，

Wherein, r _nTt () is the baseband receiving signals of the n-th node radar target, α _k,nthe echo of kth target complex magnitude when arriving the n-th node radar, k=1,2 ..., Q, Q are the numbers of extraterrestrial target, n=1,2 ..., M, M are the numbers of node radar, M>=2, s _tXt () is base band transmit, the distance that a kth target arrives the 1st node radar, be the distance of a kth target to the n-th node radar, c is the light velocity, and λ is radar operation wavelength, r _nJt () is the baseband receiving signals of the n-th node radar jamming, β _nbe the complex magnitude of undesired signal when arriving the n-th node radar, J (t) is baseband interference signal, the distance that jammer arrives the n-th node radar, w _nt () is the n-th node noise of radar receiver signal, and the noise signal of each node radar receiver is uncorrelated mutually;

(2) baseband receiving signals of each node radar step (1) obtained, time domain is as the criterion with the interference echo signal of first node radar and aligns, the form of signal after alignment with vector is represented, obtains node array radar signal vector r (t):

r(t)＝[r ₁(t),r ₂(t-τ ₁₂),…,r _M(t-τ _1M)] ^T＝S(t)+J(t)+w(t),

Wherein, τ _1nthe time delay of interference echo signal relative to the interference echo signal of first node radar of the n-th node radar, represent variate-value when making objective function get maximal value, represent convolution algorithm, () ^*represent and get conjugation, () ^trepresent and get transposition, S (t) is the target echo signal vector of node radar array, a _tthe steering vector of target relative to node radar array, α _kthe complex amplitude vector of node radar array target echo, α _k=[α _{k, 1}, α _{k, 2}..., α _k,M] ^t, represent point multiplication operation, J (t) is the interference echo signal phasor of node radar array, a _jthe steering vector of jammer relative to node radar array β is the complex amplitude vector of node radar array interference echo, β=[β ₁..., β _m] ^t, w (t) is the noise signal vector of node radar array;

(3) according to node array radar signal vector r (t) that step (2) obtains, its covariance matrix is estimated

$\hat{R}=\underset{i=1}{\overset{D}{\&Sigma;}}r\left(i\right)r^H\left(i\right),$

Wherein, i=1,2 ..., D, D are for training covariance matrix time-domain sampling sample number, () ^hrepresent conjugate transpose;

(4) by covariance matrix that following formula obtains step (3) carry out feature decomposition, obtain covariance matrix eigenvalue of maximum characteristic of correspondence vector v ₁with all the other eigenwert characteristic of correspondence vector v ₂~ v _m:

$\hat{R}=({\mathrm{\&sigma;}}_J^2+{\mathrm{\&sigma;}}_w^2)v_1{v}_1^H+\underset{j=2}{\overset{M}{\&Sigma;}}{\mathrm{\&sigma;}}_w^2{v}_j{v}_j^H,$

Wherein, j=2 ..., M, v _jrepresent a jth proper vector, represent jamming power, σ _w ²represent noise power;

(5) the proper vector v that step (4) obtains is utilized ₂~ v _m, structure noise subspace U _nr, calculate the projection matrix P to noise subspace projection _nr;

(6) node array radar signal vector r (t) that step (2) obtains is projected in the noise subspace that step (5) constructs, obtain the projection vector z being used for target detection _r(t):

z _r(t)＝P _nrr(t)，

Wherein, P _nrit is the projection matrix to noise subspace projection;

(7) to the projection vector z that step (6) obtains _rt () carries out target detection with generalized likelihood test device, structure Generalized Likelihood Ratio function Λ (t):

Λ(t)＝r(t) ^HP _nrr(t)；

(8) detection threshold is set: $\mathrm{\&delta;}=({\mathrm{\&sigma;}}_w^2/2)F_{{\mathrm{\&chi;}}_{2(M-1)}^2}^{-1}(1-P_{fa}),$

Wherein, represent that degree of freedom is card side's distribution of 2 (M-1), for the inverse function of distribution function, P _fait is false-alarm probability;

(9) functional value of each moment point of Generalized Likelihood Ratio function Λ (t) and detection threshold δ are compared, obtain the Output rusults of target detection: if Λ (t) < is δ, represent driftlessness, if Λ (t) > is δ, indicate target.

2. the networking radar based on subspace projection according to claim 1 suppresses pressing type major lobe suppression method, the projection matrix P that the calculating wherein described in step (5) projects to noise subspace _nr, be calculated as follows:

$P_{nr}=U_{nr}{U}_{nr}^H,$

Wherein, U _nr=[v ₂..., v _m], v ₂..., v _mrepresent the proper vector of node array radar signal vectors covariance matrices, M is the number of node radar, M>=2, () ^hrepresent the conjugate transpose to matrix.

3. the networking radar based on subspace projection according to claim 1 suppresses pressing type major lobe suppression method, and structure Generalized Likelihood Ratio function Λ (t) wherein described in step (7) carries out as follows:

7a) according to the projection vector z that step (6) obtains _rt (), obtains the data x (t) after dimensionality reduction:

x(t)＝Tz _r(t)，

Wherein, T is dimensionality reduction matrix,

7b) according to the data x (t) after dimensionality reduction, target detection is summed up as following Hypothesis Testing Problem:

H ₀:x(t)＝TP _nrw(t)，

H ₁:x(t)＝TP _nrS(t)+TP _nrw(t)，

Wherein, H ₀represent driftlessness, H ₁indicate target, P _nrbe the projection matrix to noise subspace projection, w (t) is the noise signal vector of node radar array, and S (t) is the target echo signal vector of node radar array;

7c) according to the Hypothesis Testing Problem being used for target detection, structure Generalized Likelihood Ratio function Λ (t):

$\mathrm{\&Lambda;}\left(t\right)=log\frac{P\&lsqb;x\left(t\right)|H_1,S\left(t\right)\&rsqb;}{P\&lsqb;x\left(t\right)|H_0\&rsqb;}=r{\left(t\right)}^H{P}_{nr}r\left(t\right),$

Wherein, P [x (t) | H ₀] represent at H ₀under supposing, the probability density function of x (t), P [x (t) | H ₁, S (t)] represent at H ₁under supposing, the probability density function of x (t), $P\&lsqb;x\left(t\right)|H_0\&rsqb;=\frac{1}{{\left(2\mathrm{\&pi;}\right)}^{(M-1)/2}|{{\mathrm{\&sigma;}}_w}^2\&CenterDot;{\mathrm{TP}}_{nr}{T}^H|}{e}^{-\frac{x{\left(t\right)}^H{\left({\mathrm{TP}}_{nr}{T}^H\right)}^{-1}x\left(t\right)}{2{{\mathrm{\&sigma;}}_w}^2}},$ $P\&lsqb;x\left(t\right)|H_1,S\left(t\right)\&rsqb;=\frac{1}{{\left(2\mathrm{\&pi;}\right)}^{(M-1)/2}|{{\mathrm{\&sigma;}}_w}^2\&CenterDot;{\mathrm{TP}}_{nr}{T}^H|}\&CenterDot;e^{-\frac{{\&lsqb;x\left(t\right)-{\mathrm{TP}}_{nr}S\left(t\right)\&rsqb;}^H{\left({\mathrm{TP}}_{nr}{T}^H\right)}^{-1}\&lsqb;x\left(t\right)-{\mathrm{TP}}_{nr}S\left(t\right)\&rsqb;}{2{{\mathrm{\&sigma;}}_w}^2}},$ () ^-1represent inversion operation, S (t) is the target echo signal vector of node radar array, and when constructing Generalized Likelihood Ratio function, S (t) uses its maximal possibility estimation usually replace, $\hat{S}\left(t\right)=P_{nr}{U}_{nr}{\left({\mathrm{TU}}_{nr}\right)}^{-1}x\left(t\right).$