CN103439691B - Method for broadband networking radar to restrict narrow-band interference - Google Patents

Abstract

The invention discloses a method for a broadband networking radar to restrict narrow-band interference and mainly aims at solving the problem that an existing method is low in calculation efficiency. The method comprises the following steps: (1) initiating a launching signal waveform set S of the networking radar; (2) according to the information of narrow-band interference and reserved frequency bands in working frequency bands of the networking radar, calculating an information matrix R of trapped wave frequency bands so as to obtain a noise subspace matrix B of the information matrix R; (3) calculating an information auxiliary matrix gamma; (4) calculating a frequency domain information auxiliary matrix V; (5) recalculating the launching signal waveform set S' of the networking radar according to the information auxiliary matrix gamma and the frequency domain information auxiliary matrix V; (6) setting a terminating threshold value to be epsilon, if ||S'-S||F<2> >=epsilon, enabling S= S', and returning to the step (3), otherwise S' being the optimum launching signal waveform set, and utilizing the waveform set to conduct trapping waves in frequency bands where the narrow-band interference exists and finishing restriction to the narrow-band interference. The method improves calculation efficiency, reduces calculation complexity, and enables a waveform optimizing method to be effectively applied into the broadband networking radar.

Description

Wide-band networking radar suppresses the method for arrowband interference

Technical field

The present invention relates to Radar Technology field, particularly a kind of method utilizing wide-band networking radar to suppress arrowband interference.The method can be used for radar network system, by optimizing the waveform that transmits, make the transmitted waveform of each node radar while the lower autocorrelation sidelobe of acquisition and cross-correlation interference, the frequency range existed in arrowband interference forms trap, improves the availability of frequency spectrum of wide-band networking radar in congested frequency range.

Background technology

Radar network, refers to the organic radar netting unified allocation of resources process of multi-section radar formed by means of communication.When each portion radar is all operated under broadband mode, be referred to as wide-band networking radar.It is by being optimized cloth station to multi-section radar, and utilize many waveforms, multiband to carry out collaborative detection to target, the radar signal space that closeness is high can be formed, by the fusion treatment technology of multi-layer, the target detection of radar system under sophisticated electronic interference environment and tracking power can be significantly improved.

But, high density signal space is while bringing advantage to wide-band networking radar, also it is made to face the problem of the congested and electromagnetic interference (EMI) of frequency range, especially high frequency HF is operated in, the wide-band networking radar of the frequency range such as very high frequency(VHF) VHF and superfrequency UHF, its working frequency range is often subject to the restriction of the reservation frequency band of the application such as communication, navigation.In addition, some communication apparatus, broadcasting and television network and other radiation devices are operated in the working frequency range of radar network, and its bandwidth transmitted is usually narrow than the signal bandwidth of wide-band networking radar, will form arrowband interference to wide-band networking radar.Meanwhile, the cross-correlation interference problem itself also existed between high auto-correlation distance side lobe and each node radar that transmits of radar network.The higher meeting of autocorrelation sidelobe causes target that in multiple goal scene, energy is strong to be flooded closing on Weak target, and higher cross-correlation interference then can affect the matched filtering performance of each node radar.Arrowband interference and higher autocorrelation sidelobe and cross-correlation interference all can affect target detection and the tracking performance of radar network system.

For the problems referred to above, mainly contain following several method at present:

First method is the wave filter by design Radar signal receiver end, make its frequency range existed in interference form trap, but the method is while suppression interference, can bring the problem of matched filtering mismatch, cause damage to the energy of target echo signal.

Second method, directly design is optimized to transmitting of radar, it is made to form trap at interference band place, and utilize wave filter to reduce its autocorrelation sidelobe at receiving end, but the speed of convergence of the method is slow, and when receiving filter constraint main lobe is constant, the output of wave filter there will be the higher problem of peak sidelobe.

The third method is that consider the performance of its power spectrum and autocorrelation function, can make to transmit obtains sparse spectral characteristic and lower autocorrelation sidelobe simultaneously when designing the transmitted waveform of radar signal.But the computation complexity of the method increases very fast along with the increase of Baud Length and node radar quantity, if the method be applied directly in wide-band networking radar, its operation efficiency is unsatisfactory.

Summary of the invention

The object of the invention is to for above-mentioned existing suppression arrowband interference method Problems existing, a kind of method utilizing wide-band networking radar to suppress arrowband interference is proposed, to improve operation efficiency, reduce computation complexity, waveform optimization can be applied in wide-band networking radar effectively.

The technical scheme realizing the object of the invention to transmit waveform from following two aspect optimizations: on the one hand, retrain at frequency domain the energy transmitted, and minimizes the power spectral density PSD that the transmits energy in trap frequency range; On the other hand, the integration sidelobe level of radar network transmitted waveform collection is optimized, with the autocorrelation sidelobe transmitted reduced and cross-correlation interference; Finally, two optimization aim are weighted, are translated into a single-object problem.Its specific implementation step is as follows:

(1) utilize the encoding phase being uniformly distributed each node radar emission signal in random generation radar network, obtain the initial transmissions signal waveform collection S=[s of whole radar network ₁..., s _m..., s _m], wherein, s _mrepresent transmitting of m node radar, m=1,2 ..., M, M are the number of radar network interior joint radar;

(2) according to arrowband interference in radar network working frequency range and the information retaining frequency range, calculate trap band information matrix R, obtain its noise subspace matrix B:

Arrowband interference and reservation frequency range in radar network working frequency range 2a) is established to have N _sindividual, the scope of a kth frequency band is [f _k1, f _k2], weight is w _k> 0, then the dimension of trap band information matrix R is the individual element R of N × N, its (p, q) _pqfor:

$R_{\mathrm{pq}}=\underset{k=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{w}_k\left\{\begin{array}{cc}\frac{e^{-j2\mathrm{\&pi;}{f}_{k2}(q-p)}-e^{-j2\mathrm{\&pi;}{f}_{k1}(q-p)}}{j2\mathrm{\&pi;}(q-p)}& ,q\&NotEqual;p\\ f_{k2}-f_{k1}& ,q=p\end{array}\right.,$

Wherein, N represents the Baud Length of each radar emission signal, and k is the numbering of trap frequency range, k=1 ..., N _s, p represents the line number of trap band information matrix R, p=1 ..., N, q represent the row number of trap band information matrix R, q=1 ..., N, j represent imaginary symbols,

2b) feature decomposition is carried out to trap band information matrix R, and be λ by ascending for eigenwert sequence successively ₁..., λ _i..., λ _n, characteristic of correspondence vector is respectively x ₁..., x _i..., x _n, wherein, i is the label of eigenwert and characteristic quantity, i=1 ..., N;

2c) utilize front Z the proper vector of trap band information matrix R, form its noise subspace matrix B=[x ₁..., x _z], wherein, the order of rank () representing matrix;

(3) according to noise subspace matrix B and each radar emission signal s _m, computing information auxiliary vector and configuration information companion matrix γ=[γ ₁..., γ _m], wherein, () ^hthe conjugate transpose of representing matrix;

(4) according to the initial transmissions signal waveform collection S of radar network, frequency domain information auxiliary vector v is calculated _l, and form frequency domain information companion matrix V=[v ₁..., v _2N] ^h, wherein, l represents the label of frequency domain information auxiliary vector, l=1 ..., 2N;

(5) the information companion matrix γ obtained according to step (3) and step (4) and frequency domain information companion matrix V, recalculates the waveform collection S ' that transmits of radar network;

(6) setting stops threshold value is ε, judges to recalculate whether the waveform collection S ' that transmits obtained is the optimum waveform collection that transmits: if then judge that S ' is not the optimum waveform collection that transmits, make S=S ', be back to step (3); Otherwise S ' is the optimum waveform collection that transmits, the frequency range utilizing this waveform collection to exist in arrowband interference carries out trap, curbs arrowband interference, wherein, || S '-S|| _ffor the Frobenius norm of transmit waveform collection S ' and initial transmissions signal waveform collection S again calculated.

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

1, the present invention due to combined optimization the power spectrum density of wide-band networking radar emission signal waveform collection, and integration sidelobe level, compared to existing method three, the optimization obtained transmits the trap better effects if of waveform collection, integration sidelobe level is lower, can more effectively be applied to wide-band networking radar, disturb to suppress arrowband;

2, used in the present invention is a kind of loop iteration method of decomposing based on discrete Fourier transformation sum of subspace, and compared to the method for steepest descent used in existing method three, operation efficiency significantly improves.

Accompanying drawing explanation

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

Fig. 2 is the normalized power spectral density figure of each node radar emission signal obtained by the inventive method;

Fig. 3 is the autocorrelation function curve map of each node radar emission signal obtained by the inventive method;

Fig. 4 is the cross correlation function curve map between each node radar emission signal of obtaining by the inventive method;

Fig. 5 is the curve map changed with Baud Length the operation time of the present invention and existing method three;

Fig. 6 is the curve map of operation time with node radar number change of the present invention and existing method three.

Embodiment

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

The waveform collection S that transmits of step 1, initialization radar network.

1a) set the number of radar network interior joint radar as M, and all transmitting of node radar are phase-coded signal, Baud Length is N, then the s that transmits of m node radar _mfor:

s _m＝[s _m1,…,s _mN] ^H，

Wherein, s _mnrepresent the signal of the n-th code element in m node radar emission signal, represent its encoding phase, j represents imaginary symbols, m represents radar label, m=1,2 ..., M, n represent the code element label of phase-coded signal, n=1,2 ..., N;

1b) utilize the encoding phase being uniformly distributed each node radar emission signal of random generation obtain the initial transmissions signal waveform collection S=[s of whole radar network ₁..., s _m..., s _m].

Step 2, according to arrowband interference in radar network working frequency range with retain the information of frequency range, calculate trap band information matrix R, obtain its noise subspace matrix B.

Arrowband interference and reservation frequency range in radar network working frequency range 2a) is established to have N _sindividual, the scope of a kth frequency band is [f _k1, f _k2], weight is w _k> 0, then the dimension of trap band information matrix R is the individual element R of N × N, its (p, q) _pqfor:

$R_{\mathrm{pq}}=\underset{k=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{w}_k\left\{\begin{array}{cc}\frac{e^{-j2\mathrm{\&pi;}{f}_{k2}(q-p)}-e^{-j2\mathrm{\&pi;}{f}_{k1}(q-p)}}{j2\mathrm{\&pi;}(q-p)}& ,q\&NotEqual;p\\ f_{k2}-f_{k1}& ,q=p\end{array}\right.,$

Wherein, N represents the Baud Length of each radar emission signal, and k is the numbering of trap frequency range, k=1 ..., N _s, p represents the line number of trap band information matrix R, p=1 ..., N, q represent the row number of trap band information matrix R, q=1 ..., N, j represent imaginary symbols,

2b) feature decomposition is carried out to trap band information matrix R, and be λ by ascending for eigenwert sequence successively ₁..., λ _i..., λ _n, characteristic of correspondence vector is respectively x ₁..., x _i..., x _n, wherein, i is the label of eigenwert and characteristic quantity, i=1 ..., N;

2c) utilize front Z the proper vector of trap band information matrix R, form its noise subspace matrix B=[x ₁..., x _z], wherein, the order of rank () representing matrix.

Step 3, computing information companion matrix γ.

3a) according to projection matrix B and each radar emission signal s _m, calculate m information auxiliary vector γ _m=B ^hs _m, wherein, m=1,2 ..., M, M are the number of radar network interior joint radar;

3b) by step 3a) M information auxiliary vector γ obtaining _m, configuration information companion matrix γ=[γ ₁..., γ _m].

Step 4, calculating frequency domain information companion matrix V.

4a) defining Discrete Fourier transform A is:

$A=\frac{1}{\sqrt{2N}}[{\mathrm{\&alpha;}}_1,\&CenterDot;\&CenterDot;\&CenterDot;,{\mathrm{\&alpha;}}_{2N}{]}_{2N\&times;2N},$

Wherein, intermediate variable ${\mathrm{\&alpha;}}_l=[e^{j{w}_l},\&CenterDot;\&CenterDot;\&CenterDot;,e^{j2N{w}_l}{]}^H,$ l=1 ..., 2N, N represent the Baud Length of each radar emission signal;

4b) according to Discrete Fourier transform A and initial transmissions signal waveform collection S, calculate the frequency domain matrix that transmits $G=A^H\left[\begin{array}{c}S\\ 0_{N\&times;M}\end{array}\right],$ The dimension of this matrix is 2N × M;

L row element 4c) getting the frequency domain matrix G that transmits forms vector C _l, and calculate the frequency domain information auxiliary vector of its correspondence wherein, || || represent 2 norms of vector;

4d) by step 4c) 2N frequency domain information auxiliary vector v obtaining _l, form frequency domain information companion matrix V=[v ₁..., v _l..., v _2N] ^h.

Step 5, recalculate the waveform collection S ' that transmits of radar network.

5a) according to noise subspace matrix B and information companion matrix γ, rated output spectral density matrix d ₁=B γ, the dimension of this matrix is N × M;

5b) according to Discrete Fourier transform A and frequency domain information companion matrix V, the front N row element getting these two matrix product AV forms integration sidelobe level matrix d ₂;

5c) according to power spectral density matrix d ₁with integration sidelobe level matrix d ₂, calculate the waveform collection S ' that transmits of radar network:

$S^{\&prime;}=e^{j\arg \{{\mathrm{\&lambda;d}}_1+(1-\mathrm{\&lambda;})d_2\},}$

Wherein, arg{} represents argument of complex number, and λ is the weight of optimizing power spectral density in waveform optimization method for designing, and 1-λ is the weight optimizing integration sidelobe level in waveform optimization method for designing, 0≤λ≤1.

It is ε that step 6, setting stop threshold value, judges to recalculate whether the waveform collection S ' that transmits obtained is the optimum waveform collection that transmits: if then judge that S ' is not the optimum waveform collection that transmits, make S=S ', be back to step (3); Otherwise S ' is the optimum waveform collection that transmits, the frequency range utilizing this waveform collection to exist in arrowband interference carries out trap, curbs arrowband interference, wherein, || S '-S|| _ffor the Frobenius norm of transmit waveform collection S ' and initial transmissions signal waveform collection S again calculated.

The ability that the present invention suppresses arrowband to disturb is verified further by following emulation.

1. experiment scene:

Wide-band networking radar system is made up of M=3 portion node radar, and each node radar is all operated in the high frequency band of 4.11MHz ~ 4.82MHz.Having 3 frequency sub-band to there is arrowband interference in this band limits, is 4.15MHz ~ 4.25MHz, 4.31MHz ~ 4.37MHz, 4.48MHz ~ 4.51MHz respectively.Wide τ=200 μ s, sample frequency f during radar emission waveform _s=2.5MHz, then can obtain the Baud Length N=τ f transmitted _s=500.The weight w of each interference band _kall be set to 1, loop termination condition is set to ε=5e-3.Setting λ=0.9 makes to apply more weight to interference band trap in objective function.

2. experiment content and result:

Test 1, utilize the transmit waveform collection of the inventive method to the wide-band networking radar in experiment scene to be optimized, disturb to suppress arrowband, the normalized power spectral density figure that the optimization drawing each node radar transmits, as shown in Figure 2, wherein Fig. 2 (a) is the normalized power spectral density figure that the optimization of first node radar transmits, the normalized power spectral density figure that the optimization of Fig. 2 (b) to be the normalized power spectral density figure that the optimization of second node radar transmits, Fig. 2 (c) be the 3rd node radar transmits.

From Fig. 2, can find out that the optimization of three the node radars utilizing the inventive method to obtain transmits and all can form darker frequency trap at 3 interference bands, average notch depth is about 22dB.

Test 2, be the autocorrelation performance checking the optimization utilizing the inventive method to obtain to transmit, the autocorrelation function curve map that the optimization drawing each node radar transmits, as shown in Figure 3, wherein, Fig. 3 (a) is the autocorrelation function curve map that the optimization of first node radar transmits, Fig. 3 (b) is the autocorrelation function curve map that the optimization of second node radar transmits, and Fig. 3 (c) is the autocorrelation function curve map that the optimization of the 3rd node radar transmits.

From Fig. 3, can find out that the optimization utilizing the inventive method to obtain transmits and has the characteristic of low distance side lobe.

Test 3, be the their cross correlation checking the optimization utilizing the inventive method to obtain to transmit, cross correlation function curve map between the optimization drawing each node radar transmits, as shown in Figure 4, wherein, Fig. 4 (a) is the cross correlation function curve map between the optimization of first and second node radar transmits, Fig. 4 (b) is the cross correlation function curve map between the optimization of first and the 3rd node radar transmits, and Fig. 4 (c) is the cross correlation function curve map between the optimization of second and the 3rd node radar transmits.

From Fig. 4, can find out that the optimization utilizing the inventive method to obtain transmits and has the characteristic of low cross-correlation interference.

Experiment 4, the sidelobe performance optimization that the inventive method and existing method three obtain transmitted and frequency spectrum trap characteristic compare, result is as shown in table 1, and wherein, ISL represents the integration sidelobe level transmitted, PSL represents the peak sidelobe transmitted, P _stoprepresent the peak value stopband power transmitted.

The sidelobe performance of the optimization waveform that table 1 the inventive method and existing method three obtain and frequency spectrum trap characteristic

	ISL (dB)	PSL (dB)	Radar 1P stop(dB)	Radar 2P stop(dB)	Radar 3P stop(dB)
						The inventive method	9.55	-17.76	-20.92	-19.33	-20.41
Existing method three	12.68	-16.16	-10.74	-12.75	-11.51

As can be seen from Table 1, the radar network utilizing the inventive method to obtain transmits, what obtain compared to existing method three transmits, integration sidelobe level reduces more than 3dB, peak sidelobe reduces 1.6dB, the simultaneously peak value stopband power reduction at least 6dB of each node radar emission signal, what describe that the performance that transmits that put forward the methods of the present invention produces is better than that existing method three obtains transmits.

Test 5, the operation time of the inventive method and existing method three is compared, the curve that the operation time obtaining the inventive method and existing method three changes with Baud Length as shown in Figure 5, operation time of the inventive method and existing method three with node radar number change curve as shown in Figure 6, wherein, steepest decline iteration step length in existing method three is set to 0.01, testing allocation of computer used is: Inter (R) Pentium (R) G630 CPU processor, frequency 2.7GHz, 4G internal memory, R2011b version Matlab software.

From Fig. 5 and Fig. 6, can see, the inventive method is compared to existing method three, and operation efficiency significantly improves, and especially along with Baud Length and node radar quantity constantly become large, the difference of operation time is increasing.

Claims (1)

1. wide-band networking radar suppresses a method for arrowband interference, and it comprises the steps:

(1) utilize the encoding phase being uniformly distributed each node radar emission signal in random generation radar network, obtain the initial transmissions signal waveform collection S=[s of whole radar network ₁..., s _m..., s _m], wherein, s _mrepresent transmitting of m node radar, m=1,2 ..., M, M are the number of radar network interior joint radar;

(2) according to arrowband interference in radar network working frequency range and the information retaining frequency range, calculate trap band information matrix R, obtain its noise subspace matrix B:

Arrowband interference and reservation frequency range in radar network working frequency range 2a) is established to have N _sindividual, the scope of a kth frequency band is [f _k1, f _k2], weight is w _k>0, then the dimension of trap band information matrix R is the individual element R of N × N, its (p, q) _pqfor:

$R_{\mathrm{pq}}=\underset{k=1}{\overset{N_s}{\mathrm{\&Sigma;}}}{w}_k\left\{\begin{array}{cc}\frac{e^{-j2\mathrm{\&pi;}{f}_{k2}(q-p)}-e^{-j2\mathrm{\&pi;}{f}_{k1}(q-p)}}{j2\mathrm{\&pi;}(q-p)},& q\&NotEqual;p\\ f_{k2-}{f}_{k1},& q=p\end{array}\right.,$

Wherein, N represents the Baud Length of each radar emission signal, and k is the numbering of trap frequency range, k=1 ..., N _s, p represents the line number of trap band information matrix R, p=1 ..., N, q represent the row number of trap band information matrix R, q=1 ..., N, j represent imaginary symbols,

2b) feature decomposition is carried out to trap band information matrix R, and be λ by ascending for eigenwert sequence successively ₁..., λ _i..., λ _n, characteristic of correspondence vector is respectively x ₁..., x _i..., x _n, wherein, i is the label of eigenwert and characteristic quantity, i=1 ..., N;

2c) utilize front Z the proper vector of trap band information matrix R, form its noise subspace matrix B=[x ₁..., x _z], wherein, the order of rank () representing matrix;

(3) according to noise subspace matrix B and each radar emission signal s _m, computing information auxiliary vector γ _m=B ^hs _m, and configuration information companion matrix γ=[γ ₁..., γ _m], wherein, () ^hthe conjugate transpose of representing matrix;

(4) according to the initial transmissions signal waveform collection S of radar network, frequency domain information auxiliary vector v is calculated _l, and form frequency domain information companion matrix V=[v ₁..., v _2N] ^h, wherein, l represents the label of frequency domain information auxiliary vector, l=1 ..., 2N;

The described initial transmissions signal waveform collection S according to radar network, calculates frequency domain information auxiliary vector v _l, carry out as follows:

4a) define Discrete Fourier transform A:

$A=\frac{1}{\sqrt{2N}}{[{\mathrm{\&alpha;}}_1,...,{\mathrm{\&alpha;}}_{2N}]}_{2N\&times;2N},$

Wherein, intermediate variable ${\mathrm{\&alpha;}}_l={[e^{{\mathrm{jw}}_l},...,e^{{j2\mathrm{Nw}}_l}]}^H,w_l=\frac{2\mathrm{\&pi;}}{2N}l,$ N represents the Baud Length of each radar emission signal;

4b) according to Discrete Fourier transform A and initial transmissions signal waveform collection S, calculate the frequency domain matrix that transmits $G=A^H\left[\begin{array}{c}S\\ 0_{N\&times;M}\end{array}\right],$ The dimension of this matrix is 2N × M;

L row element 4c) getting the frequency domain matrix G that transmits forms vector C _l, and calculate the frequency domain information auxiliary vector of its correspondence wherein, || || represent 2 norms of vector;

(5) the information companion matrix γ obtained according to step (3) and step (4) and frequency domain information companion matrix V, recalculates the waveform collection S ' that transmits of radar network:

5a) according to noise subspace matrix B and information companion matrix γ, rated output spectral density matrix d ₁=B γ, the dimension of this matrix is N × M;

5b) according to Discrete Fourier transform A and frequency domain information companion matrix V, the front N row element getting these two matrix product AV forms integration sidelobe level matrix d ₂;

5c) according to power spectral density matrix d ₁with integration sidelobe level matrix d ₂, calculate the waveform collection S ' that transmits of radar network:

$S^{\&prime;}=e^{j\arg \{{\mathrm{\&lambda;d}}_1+(1-\mathrm{\&lambda;})d_2\}},$

Wherein, arg{} represents argument of complex number, and λ is the weight of optimizing power spectral density in waveform optimization method for designing, and 1-λ is the weight optimizing integration sidelobe level in waveform optimization method for designing, 0≤λ≤1;

(6) setting stops threshold value is ε, judges to recalculate whether the waveform collection S ' that transmits obtained is the optimum waveform collection that transmits: if then judge that S ' is not the optimum waveform collection that transmits, make S=S ', be back to step (3); Otherwise S ' is the optimum waveform collection that transmits, the frequency range utilizing this waveform collection to exist in arrowband interference carries out trap, curbs arrowband interference, wherein, || S '-S|| _ffor the Frobenius norm of transmit waveform collection S ' and initial transmissions signal waveform collection S again calculated.