CN102279390A - Intra-pulse modulation and recognition method of low signal-to-noise radar radiation source signal - Google Patents
Intra-pulse modulation and recognition method of low signal-to-noise radar radiation source signal Download PDFInfo
- Publication number
- CN102279390A CN102279390A CN2011101160748A CN201110116074A CN102279390A CN 102279390 A CN102279390 A CN 102279390A CN 2011101160748 A CN2011101160748 A CN 2011101160748A CN 201110116074 A CN201110116074 A CN 201110116074A CN 102279390 A CN102279390 A CN 102279390A
- Authority
- CN
- China
- Prior art keywords
- signal
- frequency
- formula
- time
- radar emitter
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Landscapes
- Radar Systems Or Details Thereof (AREA)
Abstract
The invention discloses a intra-pulse modulation and recognition method of a low signal-to-noise radar radiation source signal, which comprises the following steps of: receiving a radar radiation source pulse signal by an electronic reconnaissance receiver, after dimension reduction and A/D sampling process from radio frequency to intermediate frequency, obtaining a radar radiation source signal S(t) having different intra-pulse modulation modes, and processing the signal S(t) in a signal processing module, identifying and outputting intra-pulse modulation mode of the radar radiation source signal. With the method, the intra-pulse modulation mode of multiple radar radiation source signals can be correctly identified when the signal-to-noise is as low as -6dB; and compared with the conventional method for identifying multiple radar radiation source signals, the computational complex O (n2) or (n3) is lower.
Description
Affiliated technical field
The present invention relates to radar emitter signal recognition technology field, especially radar emitter signal arteries and veins internal modulation recognition technology field.
Background technology
Radar emitter signal identification is the crucial processing procedure in electronic intelligence reconnaissance, electronic support scouting and the threat warning system, and its identification level is directly connected to the advanced technology degree of radar countermeasure set.Fierceness antagonism along with modern electronic warfare, novel complicated system radar constantly comes into operation and occupies leading position gradually, the signal density of electromagnetism threatening environment is up to 1,200,000 of per seconds more than the pulse, radar frequency of operation coverage has reached 0.1~20GHz, and expanding to 0.05~140GHz, the radar signal waveform changes in a plurality of territories such as time-frequency simultaneously, and stealthy and antijamming capability strengthens greatly.The recognition methods of tradition Wucan number (carrier frequency, pulse arrival time, pulse height, pulse width and pulse arrival direction) is difficult to adapt to intensive like this, complicated and changeable signal environment, and radar emitter signal identification is faced with unprecedented challenge.Lot of research about radar emitter signal identification shows in recent years, and arteries and veins internal modulation identification will be expected to obtain important breakthrough aspect novel Advanced Radar Emitter Signals source signal recognition technology and equipment.Existing radar emitter signal arteries and veins internal modulation recognition methods mainly contains small echo and wavelet package transforms method, wavelet ridge method, alike Y-factor method Y, empirical mode decomposition method, nothing is blured phase reconstruction method, entropy characteristic method, the two spectrometries of contour integral, complexity characteristics method, fractional Fourier-envelope method, fractal dimension method, ambiguity function backbone tangent plane method, instantaneous frequency derived character method and ambiguity function backbone tangent plane characteristic method etc. relatively.These methods or do not consider The noise, perhaps only just effective when signal to noise ratio (S/N ratio) is higher, bluring the phase reconstruction method as relative nothing is that 8dB can effectively discern linear frequency modulation and codiphase radar emitter Signals when above in signal to noise ratio (S/N ratio); Fractional Fourier-envelope method can effectively be discerned different classes of radar emitter signal arteries and veins internal modulation in signal to noise ratio (S/N ratio) during greater than 3dB; Instantaneous frequency derived character method can obtain correct recognition rata more than 90% during greater than 12dB in signal to noise ratio (S/N ratio); Ambiguity function backbone tangent plane characteristic method is effective identification that 2dB can realize the internal modulation of different radar emitter signal arteries and veins when above in signal to noise ratio (S/N ratio); The two spectrometries of alike Y-factor method Y, entropy characteristic method, complexity characteristics method, fractal dimension method, empirical mode decomposition method, contour integral etc. can effectively be discerned different types of arteries and veins internal modulation mode in signal to noise ratio (S/N ratio) during greater than 5dB.In actual conditions, influence owing to receiver internal noise in interference that is subjected to much noise in the communication process and the intercepting and capturing process, usually contain much noise in the radar emitter signal, signal to noise ratio (S/N ratio) is lower, cause useful signal to be submerged in the noise, increased the difficulty of identification greatly, and existing method can't realize effective identification of radar emitter signal arteries and veins internal modulation during less than 0dB in signal to noise ratio (S/N ratio).Therefore, invention is a kind of is applicable to that radar emitter signal arteries and veins internal modulation recognition methods has important application value under the low signal-to-noise ratio condition.
Summary of the invention
In view of the above shortcoming of prior art, the purpose of this invention is to provide a kind of low signal-to-noise ratio radar emitter signal arteries and veins internal modulation recognition methods.The outstanding advantage of this method is, be low to moderate in signal to noise ratio (S/N ratio)-still can correctly discern multiple radar emitter signal arteries and veins internal modulation mode under the 6dB condition, and computational complexity is low, can be used for various electronic intelligence reconnaissances, electronic support is scouted and the threat warning system in.
The present invention is for solving its technical matters, and the technical scheme that is adopted is:
A kind of low signal-to-noise ratio radar emitter signal arteries and veins internal modulation recognition methods, its step comprises ferret receiver receiving radar Emitter pulse signal, via radio frequency after the frequency reducing and A/D sampling processing of intermediate frequency, obtain the radar emitter signal S (t) with different arteries and veins internal modulation modes to be identified, in signal processing module, signal S (t) is handled again, identify the arteries and veins internal modulation mode and the output of radar emitter signal, it is characterized in that: the described concrete practice that radar emitter signal S (t) is handled is two steps of A and B
A. Radar emitter pulse signal time-frequency atomic features extracts
Comprise time domain time-frequency atomic features and frequency domain time-frequency atomic features two parts.Time domain time-frequency atomic features is meant signal S (t) the time-frequency atomic features that directly extracts in time domain.Frequency domain time-frequency atomic features is meant the time-frequency atomic features that extracts in the frequency domain that signal S (t) obtains through Fourier transform.The latter has only been Duoed Fourier transform than the former, and all the other characteristic extraction procedures are identical.The time-frequency atom that the present invention adopts is described by following elder generation, provides the detailed step of feature extraction again.
The present invention adopts the described real Gabor atom g with best time frequency resolution of formula (1)
γ(t) the time-frequency atom of discerning as the internal modulation of radar emitter signal arteries and veins, promptly
In the formula, q is the energy normalized coefficient,
Be the atomic parameter collection, wherein, s is the flexible yardstick of atom, and u is an atom translation yardstick, and ω is the atom angular frequency,
Be the atom initial phase, g () is a Gaussian function, promptly
The present invention adopts differential evolution algorithm to realize that the time-frequency atomic features of radar emitter signal S (t) extracts, process flow diagram as shown in Figure 1, concrete steps are:
(1) initialization algorithm parameter;
(2) initial value of decomposition number of times h is set to 1;
(3) determine signal R to be decomposed
h, decompose if this is decomposed into for the first time, i.e. h=1, then R
h=S (t) (annotates: R when frequency domain decomposes for the first time
h=S ' (t), wherein, S ' is the Fourier transform of S (t) (t)); If not the decomposition first time, then R
hDetermine by formula (3);
R
h=R
h-1-|<R
h-1,g
h-1>|·R
h-1 (3)
In the formula, R
H-1And g
H-1Be respectively signal and best time-frequency atom when decomposing for the h-1 time;
(4) adopt differential evolution algorithm search time-frequency atom, comprising:
(i) the population P (k) of initialization differential evolution algorithm,
Wherein N is the population size,
Be i individual and
K represents the iterations of differential evolution algorithm, k=1 here,
4 variable s in the difference representation formula (1), u, ω,
And adopt mode shown in the formula (4) to carry out initialization, promptly
In the formula, rand (0,1) is the random number between 0 and 1, a
jAnd b
jBe respectively
The lower bound and the upper bound;
(ii) to each individuality
I=1,2, L, N carries out mutation operation according to formula (5), and the structure variation is individual
, promptly
In the formula:
With
Be respectively three individualities selecting at random in the population, satisfy r1 ≠ r2 ≠ r3 and r1, r2, r3 ∈ 1,2, and L, N}, F is a constant, is called the variation probability.If the variation individuality that produces
A certain component surpass the feasible zone of this component, then adopt formula (4) that this component is carried out initialization again;
(iii) to each individuality
I=1,2, L, N carries out interlace operation according to formula (6), and the structure test is individual
Promptly
In the formula,
For testing individuality
J component, C is a constant, is called crossover probability, rand (0,1) is the random number between 0 and 1, K is an integer between 0 to 4, is determined by random fashion;
(iv) carry out selection operation according to formula (7), the individuality of selecting to have optimal adaptation degree functional value enters population of future generation, promptly
In the formula, i=1,2, L, N, H (X) they are the fitness function value of individual X, are calculated by formula (8)
H(X)=|<R
h,g
X>| (8)
In the formula,<,>be that inner product operation accords with g
XBe the time-frequency atom of individual X according to formula (1) structure;
(v) as if cycle index k 〉=T, best atom search finishes in the then current decomposable process, preserves the fitness function value λ of best time-frequency atom in the current decomposable process simultaneously
T(h) (for the ease of distinguishing, the fitness function value of the best time-frequency atom of frequency domain is designated as λ
F(h)), otherwise cycle index k increases by 1, and algorithm jumps to the continuation operation of (5) step;
(5) as if decomposing number of times h=H, then decompose and finish, otherwise h=h+1 jumps to the continuation operation of (3) step then;
(6) repeating step (2)-step (9) is extracted the best time-frequency atom of frequency domain and is preserved corresponding fitness function value λ
F(h);
(7) according to formula (9) structure h group time-frequency atomic features vector ψ (h), h=1,2, L, H-1, promptly
ψ(h)=[ψ
T(h)ψ
F(h)] (9)
In the formula, ψ
T(h) and ψ
F(h) be respectively h time domain time-frequency atomic features and h frequency domain time-frequency atomic features, its definition is described by formula (10) and formula (11) respectively:
B. adopt directed acyclic graph support vector machine classifier (Directed Acyclic Graph Support VectorMachine, DAGSVM) Discrimination Radar emitter Signals arteries and veins internal modulation mode.
Support vector machine is a kind of sorting technique based on structural risk minimization, nonlinear transformation by kernel function, to treat that classification samples is mapped to high-dimensional feature space from sample space, in feature space, maximize the class interval then, determine the optimal classification lineoid.The SVM of Ti Chuing only is applicable to two class classification problems the earliest, when handling the multicategory classification problem, needs the suitable multicategory classification device of structure.DAGSVM is the strong and high multicategory classification device of nicety of grading of a kind of fault-tolerance.Therefore, the present invention adopts DAGSVM to realize that multiple radar emitter signal arteries and veins internal modulation mode discerns, and its concrete steps are as follows:
(a) design DAGSVM sorter.
For n class radar emitter signal arteries and veins internal modulation mode identification problem, select 1 two classification of any two classes design SVM, like this, can design n (n-1)/2 two classification SVM altogether; Then this n (n-1)/2 two classification SVM are constituted an oriented no circular chart, as shown in Figure 2, this figure has n (n-1)/2 a non-leaf node and n leaf node, each non-leaf node is represented one two class svm classifier device, and link to each other with two nodes of following one deck, each leaf node is represented an output.
(b) training DAGSVM sorter.
From per two kinds of radar emitter signal features to be identified, extract the sample of some, be input in the two corresponding classification svm classifier devices, by calculating, obtain the parameter value of this SVM, so repeatedly, obtain the parameter of whole n (n-1)/2 two classification svm classifier devices,, finish the training process of DAGSVM sorter by this process.
Discrimination Radar emitter Signals arteries and veins internal modulation mode and output.When unknown radar emitter signal is carried out Classification and Identification, earlier its feature samples is input to the root node SVM of DAGSVM, classification results according to root node SVM, adopting down, the left sibling SVM or the right node SVM of one deck continue classification, till reaching certain leaf node SVM of bottom, the represented classification of this leaf node is the classification of unknown radar emitter signal, promptly obtains the output result.
Compared with prior art, the invention has the beneficial effects as follows:
1. can correctly discern signal to noise ratio (S/N ratio) to be low to moderate-multiple radar emitter signal arteries and veins internal modulation mode during 6dB.The present invention adopted complete redundant time-frequency atom to replace Wigner-Ville, the Cohen class, orthogonal basis function in traditional conversion such as Wavelet, utilize the physical feature of the redundancy properties seizure radar emitter signal of former word bank, the local key message that effectively reflects different radar emitter signal by each best time-frequency atomic energy of differential evolution algorithm optimizing acquisition, again by calculating adjacent time-frequency atom and the alike coefficient ratio of matched signal (shown in formula (10) and (11)) separately, farthest to have suppressed the influence of noise to extracted feature, make different low signal-to-noise ratio radar emitter signal in feature space, also present separability preferably, as shown in Figure 3.And the present invention adopts the DAGSVM sorter that has strong fault tolerance and high-class precision in the current machine learning field, has guaranteed the accuracy of Classification and Identification.Therefore, the present invention has better noise inhibiting ability than the recognition methods of existing radar emitter signal arteries and veins internal modulation mode.
2. computational complexity is low.Time domain among the present invention and frequency domain time-frequency atomic features all only need two time-frequency atoms, promptly only need twice iterative computation, and its computational complexity is O (1); Differential evolution algorithm adopts real coding, and its computational complexity and radar emitter signal length are the linear growth relation, and promptly computational complexity is O (n); DAGSVM sorter and radar emitter signal length are irrelevant, and its computational complexity is O (1); Thus, can analyze that to draw computational complexity of the present invention be O (n).This is than the computational complexity O (n of existing multiple radar emitter signal recognition methods
2) or O (n
3) much lower.
Description of drawings
Fig. 1 is that time-frequency atomic features of the present invention extracts process flow diagram.
Fig. 2 is the DAGSVM sorter synoptic diagram that adopts when discerning four kinds of radar emitter signal arteries and veins internal modulation modes in the embodiment of the invention.
Fig. 3 is the characteristic distribution figure of four kinds of radar emitter signal in the embodiment of the invention, and signal to noise ratio (S/N ratio) is changed to-6dB from 4dB.
Fig. 4 is the curve map that four kinds of radar emitter signal arteries and veins internal modulation modes are discerned in the embodiment of the invention accuracy changes with signal to noise ratio (S/N ratio).
In Fig. 2, indicate 1,2,3 and 4 leaf node and represent normal radar emitter Signals (CON), linear FM radar emitter Signals (LFM), biphase coding radar emitter signal (BPSK) and two-phase frequency coding radar emitter signal (BFSK) respectively, indicate the two classification SVM that 1~4,2~4,3~4,1~3,2~3 and 1~2 node represents to divide CON and BFSK, LFM and BFSK, BPSK and BFSK, CON and BPSK, LFM and BPSK, CON and LFM respectively.In Fig. 3, horizontal ordinate is a time domain time-frequency atomic features, and ordinate is a frequency domain time-frequency atomic features, all do not have unit, mark " zero " is the feature samples of CON, and mark "+" is the feature samples of LFM, mark " ● " is the feature samples of BPSK, and mark " ◇ " is the feature samples of BFSK.In Fig. 4, horizontal ordinate is a signal to noise ratio (S/N ratio), and unit is a decibel (dB), and ordinate is a correct recognition rata, no unit.
Embodiment
Below in conjunction with embodiment the present invention is described in further detail.
Embodiment
A kind of low signal-to-noise ratio radar emitter signal arteries and veins internal modulation recognition methods, its step comprises ferret receiver receiving radar Emitter pulse signal, via radio frequency after the frequency reducing and A/D sampling processing of intermediate frequency, obtain radar emitter signal S to be identified (t), present embodiment is considered 4 kinds of modulation systems, be CON, LFM, BPSK and BFSK, signal to noise ratio (S/N ratio) is changed to-10dB from 4dB, the parameter of each radar emitter signal is set to: pulse width is 4us, sample frequency is 500MHz, the carrier frequency of CON is 150MHz, the frequency range of LFM is 50 to 200MHz, and the carrier frequency of BPSK is 150MHz, adopts 13 Barker codes, two carrier frequency of BFSK are respectively 50MHz and 200MHz, also adopt 13 Barker codes; In signal processing module signal S (t) is handled, identify the arteries and veins internal modulation mode and the output of radar emitter signal, it is characterized in that: the described concrete practice that radar emitter signal S (t) is handled is:
1. Radar emitter pulse signal time-frequency atomic features extracts, and comprises time domain time-frequency atomic features and frequency domain time-frequency atomic features, and its detailed step is:
Step 1: the initialization algorithm parameter is as follows: N=50, H=2, F=0.6, C=0.2, T=800;
Step 2: the initial value that decomposes number of times h is set to 1;
Step 3: if h=1, then R
1=S (t); If h=2, then R
2=R
1-|<R
1, g
1>| R
1, g wherein
1The best time-frequency atom of coupling when decomposing for the first time;
Step 4: adopt differential evolution algorithm search time-frequency atom, comprising:
(i) mode initialization population shown in the employing formula (3)
K=1, wherein
I=1,2, L, 50,
Represent 4 variable s in the formula (1) respectively, u, ω,
Its feasible zone be respectively s ∈ (0, N], u ∈ [0, N), ω ∈ (π, π],
In the formula, rand (0,1) is the random number between 0 and 1, a
jAnd b
jBe respectively
The lower bound and the upper bound;
(ii) to each individuality
I=1,2, L, 50, individual according to the variation of formula (5) structure
, promptly
In the formula,
With
Be respectively three individualities selecting at random in the population, satisfy r1 ≠ r2 ≠ r3 and r1, r2, r3 ∈ 1,2 ..., 50}.If the variation individuality that produces
A certain component surpass the feasible zone of this component, then adopt formula (4) that this component is carried out initialization again;
(iii) to each individuality
I=1,2, L, 50, individual according to the test of formula (6) structure
Promptly
In the formula,
For testing individuality
J component, rand (0,1) is the random number between 0 and 1, K is an integer between 0 to 4, is determined by random fashion;
(iv) the individuality of selecting to have optimal adaptation degree value according to formula (7) enters population of future generation, promptly
In the formula, i=1,2, L, 50, H (X) is the fitness function value of the defined individual X of formula (8), promptly
H(X)=|<R
h,g
X>|
In the formula,<,>be that inner product operation accords with g
XBe the time-frequency atom of individual X according to formula (1) structure;
(v) as if cycle index k 〉=800, best atom search finishes in the then current decomposable process, preserves the fitness function value λ of best time-frequency atom in the current decomposable process simultaneously
T(h), otherwise cycle index k increases by 1, and algorithm jumps to the continuation operation of (5) step;
Step 5: if decompose number of times h=2, then decompose and finish, otherwise jumped to for (3) step;
Step 6: repeating step (2)-step (9), extract the best time-frequency atom of frequency domain and preserve corresponding fitness function value λ
F(h);
Step 7: according to one group of two dimension of formula (9) structure time-frequency atomic features vector ψ=[ψ
Tψ
F], ψ wherein
TAnd ψ
FBe respectively time domain time-frequency atomic features and frequency domain time-frequency atomic features, its definition is described by formula (10) and formula (11) respectively:
The present invention adopts above-mentioned steps that 4 kinds of radar emitter signal are carried out feature extraction respectively, for every kind of signal, signal to noise ratio (S/N ratio) is spaced apart 1dB, under 15 kinds of different signal to noise ratio (S/N ratio) conditions, all extract 50 samples, therefore, for every kind of state of signal-to-noise, 4 kinds of radar emitter signal have 200 feature samples, thereby 4 kinds of radar emitter signal always have 3000 samples under all signal to noise ratio (S/N ratio) conditions.Fig. 3 has provided signal to noise ratio (S/N ratio) and has changed to-4 kinds of radar emitter signal characteristic distribution figure during 6dB from 4dB, has 2200 feature samples.
2. adopt DAGSVM sorter Discrimination Radar emitter Signals arteries and veins internal modulation mode, its concrete steps are as follows:
Step 1: design DAGSVM sorter.Present embodiment is 4 class radar emitter signal arteries and veins internal modulation mode identification problems, selects 1 two classification of any two classes design SVM, like this, can design 6 two classification SVM altogether; Then these 6 two classification SVM are constituted an oriented no circular chart, as shown in Figure 2, this figure has 6 non-leaf nodes and 4 leaf nodes, each non-leaf node is represented one two class svm classifier device, and link to each other with two nodes of following one deck, each leaf node is represented an output, promptly a kind of radar emitter signal.
Step 2: training DAGSVM sorter.From per two kinds of radar emitter signal features to be identified, extract 120 samples (be changed to from-1dB-6 kinds of situations of 6dB respectively extract 20 samples) in signal to noise ratio (S/N ratio), be input in the two corresponding classification svm classifier devices, by calculating, obtain the parameter value of this SVM, so repeatedly, obtain the parameter of whole 6 two classification svm classifier devices,, finish the training process of DAGSVM sorter by this process.
Discrimination Radar emitter Signals arteries and veins internal modulation mode.When unknown radar emitter signal is carried out Classification and Identification, earlier its feature samples is input to DAGSVM root node SVM as shown in Figure 2, classification results according to root node SVM, adopting down, the left sibling SVM or the right node SVM of one deck continue classification, till reaching certain leaf node SVM of bottom, the represented classification of this leaf node is the classification of unknown radar emitter signal, promptly obtains the output result.Present embodiment 200 samples to 4 kinds of radar emitter signal under each signal to noise ratio (S/N ratio) condition have carried out Classification and Identification, and the curve that the average correct recognition rata of these 4 kinds of radar emitter signal changes with signal to noise ratio (S/N ratio) as shown in Figure 4.
As shown in Figure 4, the present invention signal to noise ratio (S/N ratio) from 4dB be reduced to-during 6dB, average correct recognition rata remains on 100%, in signal to noise ratio (S/N ratio) be-during 7dB, average correct recognition rata is 90%, shows that fully the present invention is low to moderate-still can correctly discerns during 6dB 4 kinds of radar emitter signal arteries and veins internal modulation modes in signal to noise ratio (S/N ratio).
Claims (2)
1. low signal-to-noise ratio radar emitter signal arteries and veins internal modulation recognition methods, its step comprises ferret receiver receiving radar Emitter pulse signal, via radio frequency after the frequency reducing and A/D sampling processing of intermediate frequency, obtain the radar emitter signal S (t) with different arteries and veins internal modulation modes to be identified, in signal processing module, signal S (t) is handled again, identify the arteries and veins internal modulation mode and the output of radar emitter signal, the described concrete practice that radar emitter signal S (t) is handled is:
A) Radar emitter pulse signal time-frequency atomic features extracts, and obtains time domain time-frequency atomic features and frequency domain time-frequency atomic features:
Adopt the described real Gabor atom g of formula (1) with best time frequency resolution
γ(t) the time-frequency atom of discerning as the internal modulation of radar emitter signal arteries and veins, promptly
The time-frequency atomic features of realizing radar emitter signal S (t) with differential evolution algorithm extracts, and concrete steps are:
(1) initialization algorithm parameter;
(2) initial value of decomposition number of times h is set to 1;
(3) determine signal R to be decomposed
h, decompose if this is decomposed into for the first time, i.e. h=1, then R
h=S (t) (annotates: R when frequency domain decomposes for the first time
h=S ' (t), wherein, S ' is the Fourier transform of S (t) (t)); If not the decomposition first time, then R
hDetermine by formula (2);
R
h=R
h-1-|<R
h-1,g
h-1>|·R
h-1 (2)
In the formula, R
H-1And g
H-1Be respectively signal and best time-frequency atom when decomposing for the h-1 time;
(4) adopt differential evolution algorithm search time-frequency atom, comprising:
(i) the population P (k) of initialization differential evolution algorithm,
Wherein N is the population size,
Be i individual and
K represents the iterations of differential evolution algorithm, k=1 here,
4 variable s in the difference representation formula (1), u, ω,
And adopt mode shown in the formula (3) to carry out initialization, promptly
In the formula, rand (0,1) is the random number between 0 and 1, a
jAnd b
jBe respectively
The lower bound and the upper bound;
(ii) to each individuality
I=1,2, L, N carries out mutation operation according to formula (4), and the structure variation is individual
, promptly
In the formula:
With
Be respectively three individualities selecting at random in the population, satisfy r1 ≠ r2 ≠ r3 and r1, r2, r3 ∈ 1,2, and L, N}, F is a constant, is called the variation probability; If the variation individuality that produces
A certain component surpass the feasible zone of this component, then adopt formula (3) that this component is carried out initialization again;
(iii) to each individuality
I=1,2, L, N carries out interlace operation according to formula (5), and the structure test is individual
Promptly
In the formula,
For testing individuality
J component, C is a constant, is called crossover probability, rand (0,1) is the random number between 0 and 1, K is an integer between 0 to 4, is determined by random fashion;
(iv) carry out selection operation according to formula (6), the individuality of selecting to have optimal adaptation degree functional value enters population of future generation, promptly
In the formula, i=1,2, L, N, H (X) they are the fitness function value of individual X, are calculated by formula (7)
H(X)=|<R
h,g
X>| (7)
In the formula,<,>be that inner product operation accords with g
XBe the time-frequency atom of individual X according to formula (1) structure;
(v) as if cycle index k 〉=T, best atom search finishes in the then current decomposable process, preserves the fitness function value λ of best time-frequency atom in the current decomposable process simultaneously
T(h) (for the ease of distinguishing, the fitness function value of the best time-frequency atom of frequency domain is designated as λ
F(h)), otherwise cycle index k increases by 1, and algorithm jumps to the continuation operation of (5) step;
(5) as if decomposing number of times h=H, then decompose and finish, otherwise h=h+1 jumps to the continuation operation of (3) step then;
(6) repeating step (2)-step (9) is extracted the best time-frequency atom of frequency domain and is preserved corresponding fitness function value λ
F(h);
(7) according to formula (8) structure h group time-frequency atomic features vector ψ (h), h=1,2, L, H-1, promptly
ψ(h)=[ψ
T(h)ψ
F(h)] (8)
In the formula, ψ
T(h) and ψ
F(h) be respectively h time domain time-frequency atomic features and h frequency domain time-frequency atomic features, its definition is described by formula (9) and formula (10) respectively:
B) with A) gained time-frequency atomic features finishes radar emitter signal arteries and veins internal modulation mode and discerns in the directed acyclic graph support vector machine classifier DAGSVM that has set up, and the building mode of described directed acyclic graph support vector machine classifier DAGSVM is as follows:
(a) design DAGSVM sorter
Discern for n class radar emitter signal arteries and veins internal modulation mode, select 1 two classification of any two classes design SVM, design n (n-1)/2 two classification SVM altogether; This n (n-1)/2 two classification SVM constitute an oriented no circular chart, n (n-1)/2 a non-leaf node and n leaf node are arranged, each non-leaf node is represented one two class svm classifier device, and links to each other with two nodes of following one deck, and each leaf node is represented an output;
(b) training DAGSVM sorter
From per two kinds of radar emitter signal features to be identified, extract the sample of some, be input in the two corresponding classification svm classifier devices and calculate the parameter value that obtains this SVM, so repeatedly, obtain the parameter of whole n (n-1)/2 two classification svm classifier devices, by this process, finish the training process of DAGSVM sorter;
Described A) gained time-frequency atomic features is input to the root node SVM of DAGSVM, classification results according to root node SVM, adopting down, the left sibling SVM or the right node SVM of one deck continue classification, till reaching certain leaf node SVM of bottom, the represented classification of this leaf node is the classification of unknown radar emitter signal, promptly obtains the output result.
2. the low signal-to-noise ratio radar emitter signal arteries and veins internal modulation recognition methods according to claim 1, it is characterized in that: described n class Radar emitter modulation signal is four class Radar emitter modulation signal, that is: CON, LFM, BPSK and BFSK
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110116074 CN102279390B (en) | 2011-05-06 | 2011-05-06 | Intra-pulse modulation and recognition method of low signal-to-noise radar radiation source signal |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN 201110116074 CN102279390B (en) | 2011-05-06 | 2011-05-06 | Intra-pulse modulation and recognition method of low signal-to-noise radar radiation source signal |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102279390A true CN102279390A (en) | 2011-12-14 |
CN102279390B CN102279390B (en) | 2013-04-10 |
Family
ID=45104942
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN 201110116074 Expired - Fee Related CN102279390B (en) | 2011-05-06 | 2011-05-06 | Intra-pulse modulation and recognition method of low signal-to-noise radar radiation source signal |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102279390B (en) |
Cited By (28)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102590791A (en) * | 2012-01-10 | 2012-07-18 | 哈尔滨工程大学 | Method for sorting signals of radar radiation source by using coverage in complex dense environment |
CN102882819A (en) * | 2012-09-04 | 2013-01-16 | 西安电子科技大学 | Digital demodulation signal identification method under non-gaussian noise |
CN103093244A (en) * | 2013-01-25 | 2013-05-08 | 西南交通大学 | Radar signal identification method based on sparse decomposition |
CN103217676A (en) * | 2013-05-06 | 2013-07-24 | 西安电子科技大学 | Radar target identification method under noise background based on bispectrum de-noising |
CN106529478A (en) * | 2016-11-08 | 2017-03-22 | 中国人民解放军空军工程大学 | Radar radiation source signal identification method according to three-dimensional entropy characteristic |
CN106706551A (en) * | 2016-12-05 | 2017-05-24 | 河南工业大学 | Method and device for selecting histidine terahertz absorption spectrum wavelength based on differential evolution |
CN106778610A (en) * | 2016-12-16 | 2017-05-31 | 哈尔滨工程大学 | A kind of intra-pulse modulation recognition methods based on time-frequency image feature |
CN106769992A (en) * | 2016-12-05 | 2017-05-31 | 河南工业大学 | Glutamine Terahertz spectrum wavelength system of selection and device based on differential evolution |
CN106842141A (en) * | 2017-04-14 | 2017-06-13 | 中国人民解放军武汉军械士官学校 | A kind of high-order is repeatedly conjugated lagged product Intrapulse analysis method |
CN107122738A (en) * | 2017-04-26 | 2017-09-01 | 成都蓝色起源科技有限公司 | Automatic Communication Signals Recognition based on deep learning model and its realize system |
CN107301381A (en) * | 2017-06-01 | 2017-10-27 | 西安电子科技大学昆山创新研究院 | Recognition Method of Radar Emitters based on deep learning and multi-task learning strategy |
CN107748356A (en) * | 2017-11-06 | 2018-03-02 | 西安电子科技大学 | Phase-coded signal deception jamming recognition methods based on multiple resemblance Coefficient |
CN107770108A (en) * | 2017-10-23 | 2018-03-06 | 佛山科学技术学院 | A kind of combined modulation recognition methods of K mean clusters and classification training SVM classifier |
CN107918110A (en) * | 2017-11-06 | 2018-04-17 | 西安电子科技大学 | Linear FM signal deception jamming recognition methods based on multiple resemblance Coefficient |
CN109116345A (en) * | 2018-08-01 | 2019-01-01 | 西南交通大学 | A kind of Radar Signal Recognition method and device |
CN109446910A (en) * | 2018-09-17 | 2019-03-08 | 中国人民解放军空军工程大学 | A kind of communication emitter Signals classifying identification method |
CN109856603A (en) * | 2019-03-26 | 2019-06-07 | 森思泰克河北科技有限公司 | Radar anti-interference method, terminal device and storage medium |
CN110163040A (en) * | 2018-03-28 | 2019-08-23 | 信阳农林学院 | Radar emitter signal identification technology in non-gaussian clutter |
CN110988802A (en) * | 2019-11-11 | 2020-04-10 | 浙江大学 | Radar radiation source identification system based on signal scale decomposition |
CN111401185A (en) * | 2020-03-10 | 2020-07-10 | 北京波尔通信技术股份有限公司 | Electromagnetic signal and equipment correlation analysis method and system based on radio frequency fingerprint SEV |
CN111612130A (en) * | 2020-05-18 | 2020-09-01 | 吉林大学 | Frequency shift keying communication signal modulation mode identification method |
CN111680737A (en) * | 2020-06-03 | 2020-09-18 | 西安电子科技大学 | Radar radiation source individual identification method under differential signal-to-noise ratio condition |
CN112115821A (en) * | 2020-09-04 | 2020-12-22 | 西北工业大学 | Multi-signal intelligent modulation mode identification method based on wavelet approximate coefficient entropy |
CN112462344A (en) * | 2020-11-12 | 2021-03-09 | 中国人民解放军空军预警学院 | Method for extracting radar signal intra-pulse characteristic parameters through time-frequency domain transformation processing |
CN112578360A (en) * | 2020-11-12 | 2021-03-30 | 中国人民解放军空军预警学院 | Method for extracting radar signal intra-pulse characteristic parameters based on transform domain |
CN112578359A (en) * | 2020-11-12 | 2021-03-30 | 中国人民解放军空军预警学院 | Method for extracting radar signal intra-pulse characteristic parameters through bispectrum transformation processing |
CN112859019A (en) * | 2021-01-11 | 2021-05-28 | 北京无线电计量测试研究所 | Intra-pulse modulation type parameter extraction system and using method |
CN117289236A (en) * | 2023-11-27 | 2023-12-26 | 成都立思方信息技术有限公司 | Short-time radar signal intra-pulse modulation type identification method, device, equipment and medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
USH1726H (en) * | 1997-03-31 | 1998-05-05 | Chen; Victor C. | CFAR time-frequency processor for signal detection and extraction in noise |
DE10325147A1 (en) * | 2003-05-28 | 2004-12-16 | Friedrich-Schiller-Universität Jena | Signal analysis method for time frequency analysis of signal sequences uses atoms in a dictionary in trials for a signal sequence |
US20050149296A1 (en) * | 2003-12-31 | 2005-07-07 | Sieracki Jeffrey M. | Greedy adaptive signature discrimination system and method |
CN101587186A (en) * | 2008-05-22 | 2009-11-25 | 赵力 | Characteristic extraction method of radar in-pulse modulation signals |
-
2011
- 2011-05-06 CN CN 201110116074 patent/CN102279390B/en not_active Expired - Fee Related
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
USH1726H (en) * | 1997-03-31 | 1998-05-05 | Chen; Victor C. | CFAR time-frequency processor for signal detection and extraction in noise |
DE10325147A1 (en) * | 2003-05-28 | 2004-12-16 | Friedrich-Schiller-Universität Jena | Signal analysis method for time frequency analysis of signal sequences uses atoms in a dictionary in trials for a signal sequence |
US20050149296A1 (en) * | 2003-12-31 | 2005-07-07 | Sieracki Jeffrey M. | Greedy adaptive signature discrimination system and method |
CN101587186A (en) * | 2008-05-22 | 2009-11-25 | 赵力 | Characteristic extraction method of radar in-pulse modulation signals |
Non-Patent Citations (1)
Title |
---|
程吉祥: "复杂体制雷达辐射源信号时频原子特征提取方法", 《西安交通大学学报》 * |
Cited By (42)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102590791A (en) * | 2012-01-10 | 2012-07-18 | 哈尔滨工程大学 | Method for sorting signals of radar radiation source by using coverage in complex dense environment |
CN102882819A (en) * | 2012-09-04 | 2013-01-16 | 西安电子科技大学 | Digital demodulation signal identification method under non-gaussian noise |
CN102882819B (en) * | 2012-09-04 | 2015-02-04 | 西安电子科技大学 | Digital demodulation signal identification method under non-gaussian noise |
CN103093244A (en) * | 2013-01-25 | 2013-05-08 | 西南交通大学 | Radar signal identification method based on sparse decomposition |
CN103217676A (en) * | 2013-05-06 | 2013-07-24 | 西安电子科技大学 | Radar target identification method under noise background based on bispectrum de-noising |
CN103217676B (en) * | 2013-05-06 | 2014-08-20 | 西安电子科技大学 | Radar target identification method under noise background based on bispectrum de-noising |
CN106529478A (en) * | 2016-11-08 | 2017-03-22 | 中国人民解放军空军工程大学 | Radar radiation source signal identification method according to three-dimensional entropy characteristic |
CN106706551B (en) * | 2016-12-05 | 2020-03-24 | 河南工业大学 | Histidine terahertz absorption spectrum wavelength selection method and device based on differential evolution |
CN106706551A (en) * | 2016-12-05 | 2017-05-24 | 河南工业大学 | Method and device for selecting histidine terahertz absorption spectrum wavelength based on differential evolution |
CN106769992A (en) * | 2016-12-05 | 2017-05-31 | 河南工业大学 | Glutamine Terahertz spectrum wavelength system of selection and device based on differential evolution |
CN106778610A (en) * | 2016-12-16 | 2017-05-31 | 哈尔滨工程大学 | A kind of intra-pulse modulation recognition methods based on time-frequency image feature |
CN106778610B (en) * | 2016-12-16 | 2020-04-07 | 哈尔滨工程大学 | Intra-pulse modulation identification method based on time-frequency image characteristics |
CN106842141A (en) * | 2017-04-14 | 2017-06-13 | 中国人民解放军武汉军械士官学校 | A kind of high-order is repeatedly conjugated lagged product Intrapulse analysis method |
CN107122738A (en) * | 2017-04-26 | 2017-09-01 | 成都蓝色起源科技有限公司 | Automatic Communication Signals Recognition based on deep learning model and its realize system |
CN107301381A (en) * | 2017-06-01 | 2017-10-27 | 西安电子科技大学昆山创新研究院 | Recognition Method of Radar Emitters based on deep learning and multi-task learning strategy |
CN107770108A (en) * | 2017-10-23 | 2018-03-06 | 佛山科学技术学院 | A kind of combined modulation recognition methods of K mean clusters and classification training SVM classifier |
CN107748356A (en) * | 2017-11-06 | 2018-03-02 | 西安电子科技大学 | Phase-coded signal deception jamming recognition methods based on multiple resemblance Coefficient |
CN107918110A (en) * | 2017-11-06 | 2018-04-17 | 西安电子科技大学 | Linear FM signal deception jamming recognition methods based on multiple resemblance Coefficient |
CN107748356B (en) * | 2017-11-06 | 2019-07-23 | 西安电子科技大学 | Phase-coded signal deception jamming recognition methods based on multiple resemblance Coefficient |
CN110163040A (en) * | 2018-03-28 | 2019-08-23 | 信阳农林学院 | Radar emitter signal identification technology in non-gaussian clutter |
CN110163040B (en) * | 2018-03-28 | 2022-12-30 | 信阳农林学院 | Radar radiation source signal identification technology in non-Gaussian clutter |
CN109116345A (en) * | 2018-08-01 | 2019-01-01 | 西南交通大学 | A kind of Radar Signal Recognition method and device |
CN109446910B (en) * | 2018-09-17 | 2021-12-07 | 中国人民解放军空军工程大学 | Communication radiation source signal classification and identification method |
CN109446910A (en) * | 2018-09-17 | 2019-03-08 | 中国人民解放军空军工程大学 | A kind of communication emitter Signals classifying identification method |
CN109856603A (en) * | 2019-03-26 | 2019-06-07 | 森思泰克河北科技有限公司 | Radar anti-interference method, terminal device and storage medium |
CN110988802A (en) * | 2019-11-11 | 2020-04-10 | 浙江大学 | Radar radiation source identification system based on signal scale decomposition |
CN111401185A (en) * | 2020-03-10 | 2020-07-10 | 北京波尔通信技术股份有限公司 | Electromagnetic signal and equipment correlation analysis method and system based on radio frequency fingerprint SEV |
CN111401185B (en) * | 2020-03-10 | 2023-06-20 | 北京波尔通信技术股份有限公司 | Electromagnetic signal and equipment association analysis method and system based on radio frequency fingerprint SEV |
CN111612130A (en) * | 2020-05-18 | 2020-09-01 | 吉林大学 | Frequency shift keying communication signal modulation mode identification method |
CN111680737B (en) * | 2020-06-03 | 2023-03-24 | 西安电子科技大学 | Radar radiation source individual identification method under differential signal-to-noise ratio condition |
CN111680737A (en) * | 2020-06-03 | 2020-09-18 | 西安电子科技大学 | Radar radiation source individual identification method under differential signal-to-noise ratio condition |
CN112115821B (en) * | 2020-09-04 | 2022-03-11 | 西北工业大学 | Multi-signal intelligent modulation mode identification method based on wavelet approximate coefficient entropy |
CN112115821A (en) * | 2020-09-04 | 2020-12-22 | 西北工业大学 | Multi-signal intelligent modulation mode identification method based on wavelet approximate coefficient entropy |
CN112578359A (en) * | 2020-11-12 | 2021-03-30 | 中国人民解放军空军预警学院 | Method for extracting radar signal intra-pulse characteristic parameters through bispectrum transformation processing |
CN112578360A (en) * | 2020-11-12 | 2021-03-30 | 中国人民解放军空军预警学院 | Method for extracting radar signal intra-pulse characteristic parameters based on transform domain |
CN112462344A (en) * | 2020-11-12 | 2021-03-09 | 中国人民解放军空军预警学院 | Method for extracting radar signal intra-pulse characteristic parameters through time-frequency domain transformation processing |
CN112578360B (en) * | 2020-11-12 | 2023-09-19 | 中国人民解放军空军预警学院 | Method for extracting characteristic parameters in radar signal pulse based on transform domain |
CN112578359B (en) * | 2020-11-12 | 2023-09-19 | 中国人民解放军空军预警学院 | Method for extracting characteristic parameters in radar signal pulse through bispectrum conversion processing |
CN112462344B (en) * | 2020-11-12 | 2023-09-19 | 中国人民解放军空军预警学院 | Method for extracting characteristic parameters in radar signal pulse through time-frequency domain transformation processing |
CN112859019A (en) * | 2021-01-11 | 2021-05-28 | 北京无线电计量测试研究所 | Intra-pulse modulation type parameter extraction system and using method |
CN117289236A (en) * | 2023-11-27 | 2023-12-26 | 成都立思方信息技术有限公司 | Short-time radar signal intra-pulse modulation type identification method, device, equipment and medium |
CN117289236B (en) * | 2023-11-27 | 2024-02-09 | 成都立思方信息技术有限公司 | Short-time radar signal intra-pulse modulation type identification method, device, equipment and medium |
Also Published As
Publication number | Publication date |
---|---|
CN102279390B (en) | 2013-04-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102279390B (en) | Intra-pulse modulation and recognition method of low signal-to-noise radar radiation source signal | |
CN107576943B (en) | Adaptive Time and Frequency Synchronization compression method based on Rayleigh entropy | |
Das et al. | Marine vessel classification based on passive sonar data: the cepstrum‐based approach | |
CN107301381A (en) | Recognition Method of Radar Emitters based on deep learning and multi-task learning strategy | |
CN102749615B (en) | A kind of method of Signal analysis | |
CN109379311A (en) | Ultrashort wave signal specific recognition methods based on convolutional neural networks | |
CN110187313B (en) | Radar signal sorting and identifying method and device based on fractional order Fourier transform | |
CN110110738A (en) | A kind of Recognition Method of Radar Emitters based on multi-feature fusion | |
CN116359851A (en) | Radar active interference detection and identification method and device based on converged network | |
Yang et al. | Sparsity‐based space–time adaptive processing using complex‐valued Homotopy technique for airborne radar | |
CN116866129A (en) | Wireless communication signal detection method | |
CN110376553A (en) | Low SNR signal parameter extracting method based on pulse cutting | |
CN101783777A (en) | Digital modulation signal recognizing method | |
CN111680737B (en) | Radar radiation source individual identification method under differential signal-to-noise ratio condition | |
Addabbo et al. | HRR profile estimation using SLIM | |
Li et al. | A deep convolutional network for multitype signal detection and classification in spectrogram | |
CN106357575A (en) | Multi-parameter jointly-estimated interference type identification method | |
Yang et al. | One-dimensional deep attention convolution network (ODACN) for signals classification | |
CN116797846A (en) | Method and device for identifying small sample radar radiation source based on RoAtten-PN network | |
Allabakash et al. | Wavelet transform‐based methods for removal of ground clutter and denoising the radar wind profiler data | |
Guven et al. | Classifying LPI radar waveforms with time-frequency transformations using multi-stage CNN system | |
Han et al. | Smart noise jamming suppression method based on fast fractional filtering | |
Mou et al. | Motion classification for radar moving target via STFT and convolution neural network | |
Huang et al. | Modulation identification method based on time-frequency analysis and support vector machine | |
CN106569182A (en) | Phase coding signal carrier frequency estimation method based on minimum entropy |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20130410 Termination date: 20160506 |