A cognitive adversarial modeling method under incomplete information conditions

Through the dual-matrix Stackelberg game and Bayesian method, the problem of radar confrontation strategy evaluation under incomplete information conditions under complex electromagnetic environments is solved, and the quantitative calculation and confrontation performance evaluation of radar anti-interference ability are realized, and it is extended to radar network and swarm jamming machines adversarial modeling.

CN114779184BActive Publication Date: 2024-12-03XIAN DAHENG TIANCHENG IT CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202210419755.X
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-04-21
Publication Date
2024-12-03
Estimated Expiration
2042-04-21

AI Technical Summary

Technical Problem

In complex electromagnetic environments, existing cognitive adversarial technologies cannot effectively evaluate the effectiveness of adversarial strategies under incomplete information conditions, especially in radar adversarial scenarios, where true cognitive adversarial cannot be achieved.

Method used

Using the method based on the dual-matrix Stackelberg game, the radar anti-jamming strategy allocation matrix and the interference strategy allocation matrix are constructed, combined with the position information of the radar and jammer, the interference power and anti-jamming improvement factors are calculated, and the double-matrix game of the composite strategy is evaluated. The benefits of radar and jammers and the interference strategy is adjusted through Bayesian methods.

Benefits of technology

Quantitative calculation of radar anti-jamming capability and adversarial efficacy evaluation under incomplete information conditions were realized, human factors in traditional methods were overcome, and convergence modeling and evaluation of radar networks and swarm jammers were extended.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN114779184B_ABST
    Figure CN114779184B_ABST
Patent Text Reader

Abstract

A cognitive confrontation modeling method under incomplete information conditions is proposed. On the basis of knowing the main technical parameters of the radar, based on domain knowledge and public information, the radar's anti-interference measures are sorted out and the corresponding empirical formula of anti-interference improvement factors is proposed, including the reasonable setting of the radar's false target recognition rate; after the simulation scene is determined, according to the position information of the jammer and the radar, the traditional radar power suppression coefficient calculation formula is used to calculate the interference-to-signal ratio of the interference power on the radar antenna aperture, and a calculation method for the interference benefit is proposed for the power suppression and deception interference style of the jammer, and then a strategy matrix construction method for constructing a double-matrix Stackelberg game is proposed. The recognition probability of the radar deception of the false target is taken as uncertain information. The confrontation scene under the condition of "one to one" incomplete perception information of the jammer against the radar is simulated and the game effect under a specific confrontation environment is verified.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of radar confrontation modeling and simulation under complex electromagnetic environments, and in particular relates to a cognitive confrontation modeling method under incomplete information conditions. Background Art

[0002] Cognitive confrontation is a new concept of information confrontation that emerged with electromagnetic spectrum warfare, that is, to perceive the opponent's state in real time through the electronic reconnaissance system, adjust one's own confrontation strategy, and maximize the interference / anti-interference efficiency. Existing cognitive confrontations are all based on the assumption that the opponent's state information is fully perceived, and corresponding confrontation measures are formulated in power allocation, working frequency band, confrontation time and airspace, and a benefit evaluation model is established; however, the actual confrontation is an "open-loop" environment, and it is impossible for the two parties to fully grasp the other's state. Under the condition of incomplete perception information, how to evaluate the effectiveness of the confrontation strategy is a scientific problem that has yet to be solved.

[0003] Cognitive confrontation originated from cognitive radio technology. The so-called "cognition" refers to a kind of intellectual activity with consciousness, such as thinking, reasoning, memory, imagination, learning, processing information, applying knowledge, changing priorities, etc. Cognitive confrontation mainly involves the field of radar confrontation. The so-called "cognitive radar" is to give the radar system the ability to perceive the environment in which it is located, and can use artificial intelligence technology to continuously learn and infer the changing laws of the external environment, and adaptively adjust its own working mode, such as adjusting the transmission power, working frequency, modulation mode, transmission waveform, and adopting beam sidelobe blanking and other measures to reduce the possibility of interference and destruction of the radar in a complex confrontation environment. To achieve the above goals, the radar must master the interference power, interference style, networking form and location of the interference party, and also evaluate the electronic reconnaissance capability of the interference party. The core anti-interference capability of the radar is mainly reflected in the recognition rate of multiple false targets. If false targets are to be identified, a large number of interference signal samples that cover all interference styles and strategies and are manually labeled are required, which is an impossible condition in practice. Similarly, it is impossible for the jammer to fully master the radar's anti-jamming measures and signal processing methods, especially the ability to identify false targets. In the actual confrontation process, only part of the radar's working status can be obtained through the electronic support system, and the effectiveness of releasing interference does not completely depend on the radar's change information. For example, the traditional suppression coefficient does not take into account the radar's anti-jamming improvement factor, so it can only represent the signal-to-interference ratio at the radar antenna aperture. Under incomplete information conditions, the existing artificial intelligence-based confrontation models and effectiveness evaluation methods are not in line with the actual confrontation environment, and it is impossible to achieve true cognitive confrontation. Summary of the invention

[0004] In order to overcome the shortcomings of the above-mentioned prior art, the purpose of the present invention is to provide a cognitive confrontation modeling method under incomplete information conditions, which realizes radar cognitive confrontation modeling and simulation based on double-matrix Stackelberg game.

[0005] In order to achieve the above object, the technical solution adopted by the present invention is:

[0006] A cognitive adversarial modeling method under incomplete information conditions includes the following steps:

[0007] Step 1: Construct radar anti-interference improvement factor D i Calculation method, i=1,2…N R , N R is the total number of anti-jamming measures of the radar;

[0008] Step 2: After the simulation confrontation scenario is determined, the interference-to-signal ratio JSR of the jammer power on the radar antenna aperture is calculated according to the location information of the jammer and the radar;

[0009] Step 3: Take the total number of radar anti-interference measures N R Based on this, the radar anti-interference strategy allocation matrix a is constructed, where "0" represents the anti-interference measures not adopted and "1" represents the anti-interference measures adopted; the radar benefit matrix A is generated;

[0010] Step 4: Take the total number of jammer measures N j Based on this, construct the interference strategy allocation matrix b, where "0" represents the interference measures not adopted and "1" represents the interference measures adopted; generate the interference machine benefit matrix B;

[0011] Step 5: With the radar as the leader and the jammer as the follower, the probability of radar identifying false targets is P d For incomplete information, a double-matrix Stackelberg game with a compound strategy is conducted to calculate the benefits of each jamming measure of the jammer and each anti-jamming measure of the radar.

[0012] Step 6, record the number of times the radar working state changes in step 5, and use the Bayesian method to calculate the posterior probability P of the radar being interfered. When P>0.5, the interference is effective and B remains unchanged; when P≤0.5, the interference is invalid, adjust the interference strategy allocation matrix b, repeat step 4, and end the simulation process when the maximum number of simulations is reached.

[0013] In step 1, radar anti-interference improvement factor D i The calculation of needs to obtain the following parameters of the radar:

[0014] Time domain parameters, frequency domain parameters, spatial domain parameters and radar identification false target probability P d; The time domain parameters include radar transmission pulse width τ, repetition period PRI, waveform agility number N w and radar beam dwell time T; frequency domain parameters include center frequency f c , radar receiver bandwidth W r , intra-pulse modulation method M, radar frequency diversity number N F and the number of radar deception signal frequencies N D ; Spatial parameters include radar main lobe half power point width θ 0.5 ;

[0015] Radar anti-interference improvement factor D i The calculation methods include:

[0016] 1), matched filtering and pulse compression technology

[0017] Pulse pressure signals include: linear / nonlinear frequency modulation signals, pulse coding signals, and the corresponding improvement factors are:

[0018] D LFM =BτL

[0019] D pm =NL

[0020] Where D LFM is the anti-interference improvement factor of the linear frequency modulation (LFM) signal, Bτ is the bandwidth-time product of the radar pulse, and L is the pulse compression loss; D pm is the anti-interference improvement factor of the pulse coding signal, N is the phase coding length, when the Barker code is used, the value of N is 7 to 13, when the M sequence is used, the value of N is 16 to 128, and when the 4-phase code is used, the value of N is 16;

[0021] After taking the decibel number for the improvement factor, the anti-interference improvement factor obtained is:

[0022] D0 = pulse pressure ratio or code length - 1.3dB

[0023] 2), Pulse accumulation improvement factor

[0024] Radar pulse accumulation is divided into coherent accumulation and incoherent accumulation; the number of accumulated pulses is calculated as follows:

[0025]

[0026] In the formula, θ 0.5 is the radar main lobe half-power point width, T is the radar beam dwell time, f PRF is the pulse repetition frequency;

[0027] For coherent accumulation, D1 = n 0.8 , for incoherent accumulation, D1 = n V, V∈[0.7,0.9];

[0028] 3), Waveform Agility Factor

[0029] D2=10lgN w

[0030] Where N w The number of waveform agility changes. Radar increases the difficulty of analyzing the jammer through waveform changes, forcing the jammer to increase the jamming bandwidth, thereby reducing the jamming effectiveness.

[0031] 4), frequency diversity factor

[0032] D3=10lgN F

[0033] Where N F is the number of radar frequency diversity;

[0034] 5), Deceptive emission factor:

[0035] D4=10lg(N D +1)

[0036] Where N D The number of frequencies of radar deception signals. Deception emission means that the radar transmits low-power deception signals from the auxiliary antenna at multiple frequencies, making it impossible for the jammer to accurately aim at the radar's main operating frequency.

[0037] 6) False target identification technology

[0038]

[0039] Where N r is the number of false target echoes identified, N totle is the total number of echoes;

[0040] Order r1 =D0+D1,s r2 =D2,s r3 =D3,s r4 =D4,s r5 =D5, then A = a.*[s r1 ,s r2 ,sr3,s r4 ,s r5 ], .* is the dot multiplication operator, which means that each element of a and s r1 ~s r5 Multiply, a is M×N R The number of rows M is the total number of anti-interference strategies that the radar can adopt.

[0041] In step 2, the calculation formula of the interference signal ratio JSR of the interference power at the radar antenna aperture is as follows:

[0042]

[0043] Where P j is the interference power, and the distance R from the jammer to the radar antenna j Inversely proportional to the square of P r is the target echo power and the distance R from the target to the radar antenna t Inversely proportional to the fourth power, in the self-defense jamming scenario, there is R j =R t ; G j is the gain of the jammer antenna relative to the radar direction, γ j is the polarization mismatch coefficient of the jammer antenna relative to the radar antenna, P t is the radar transmitting pulse power, G r is the gain of the radar receiving antenna toward the jammer, and σ is the RCS of the protected target.

[0044] In step 4, the interference benefit factor of the jammer includes:

[0045] (1)Broadband noise suppression benefits:

[0046]

[0047] Where W j is the interference bandwidth, W r is the radar receiver bandwidth;

[0048] (2) Narrowband smart noise suppression benefits:

[0049]

[0050] Where, T r =τ is the radar transmission pulse width, k = W r / T r is the radar pulse bandwidth-time width ratio;

[0051] (3) 'C&I' Slice Interference Benefit:

[0052] s j3 =10lg(η 2 W r T r )+12,η∈[0,1]

[0053] Where η is the slice coefficient;

[0054] (4) Benefits of dense false targets:

[0055] sj4 =10lg(N f )(1-P d )

[0056] Where N f is the number of false targets;

[0057] (5) Distance-speed deception benefits:

[0058] s j5 =JSR(1-P d )

[0059] Then the interference strategy matrix B = b.*[s j1 ,s j2 ,s j3 ,s j4 ,s j5 ], after A and B are determined, a double matrix Stackelberg game with a composite strategy is performed, and b is M×N j The matrix whose row number M is the jammer's jamming strategy corresponding to the radar's anti-jamming strategy.

[0060] The double matrix Stackelberg game process of the composite strategy is as follows:

[0061] S1: Input strategy matrix A = A T ,B=B T ;T-matrix transpose;

[0062] S2: Take each column of A as the coefficient of function f and calculate the game result of the corresponding column of B, that is, calculate the linear programming:

[0063]

[0064] stDx<d,D eq x=d eq ,l b <x<u b

[0065] Among them, the matrix D is the coefficient matrix obtained by subtracting the interference strategy in A from other strategies in B, d,D eq and d eq is the constraint condition; b ,u b is the lower and upper bound of x;

[0066] S3: Record all the game results between A and B: fval = f(x j ),j=1,2,…M, select the maximum value of fval as the benefit u of the radar in the game A ;

[0067] S4: Benefits of radar anti-jamming u A , calculate the jammer's benefit u B ;

[0068] S5: Calculate u d =u A -u B , if u d >0, radar outperforms jammer, conversely, jammer outperforms radar.

[0069] In step 6, the jammer observes the radar state variable x=[A m ,f c ,B w ,M d ,prf] estimates whether the interference is effective, and uses "0" and "1" to indicate whether the radar state variable x has changed significantly after a confrontation. There are x = 2 5 states; let the prior probability density of x be g m Following the Gaussian mixture distribution model, after a confrontation, the posterior probability of the radar state is calculated using the Bayesian method: P = posterior (g m ,x), where g m =fitgmdist(x,5) is to fit the Gaussian mixture distribution model. If P<0.5, it is considered that the current interference strategy B is invalid and B needs to be changed. Otherwise, it is considered that the current interference strategy is effective and B remains unchanged.

[0070] Compared with the prior art, the present invention has the following beneficial effects:

[0071] 1. The present invention quantitatively calculates the anti-interference capability of the radar based on the known technical performance of some radars under the condition of incomplete perception information, takes the unknown probability of false target identification by the radar as the system variable, and realizes a reasonable evaluation of the effectiveness of cognitive confrontation.

[0072] 2. After determining the slant range between the jammer and the target radar, the present invention uses the total number of anti-interference measures of the radar and the total number of jammer interference patterns as the basic basis for the radar anti-interference strategy allocation matrix a and the interference strategy allocation matrix b, and then generates the radar anti-interference strategy matrix A and the interference strategy matrix B, thereby overcoming the human factors existing in the traditional benefit matrix construction method that relies on the experience of field experts.

[0073] 3. From a methodological point of view, the proposed method of the present invention can be extended to the modeling and confrontation effect evaluation of the system confrontation between the radar network and the swarm jammer. As long as the elements of the system confrontation and the types of incomplete information are sorted out, a reasonable evaluation of the confrontation effect under different conditions can be made. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Figure 1The modeling and simulation flow chart of the present invention.

[0075] Figure 2 P is the probability of false target identification by radar d The double matrix Stackelberg game result with is the variable.

[0076] Figure 3 P d =0.5, double matrix Stackelberg game results with the number of false targets as a variable.

[0077] Figure 4 P d =0.6, double matrix Stackelberg game results with the number of false targets as variables. DETAILED DESCRIPTION

[0078] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative work belong to the scope of protection of the present invention.

[0079] like Figure 1 As shown, reference Figure 1 The present invention provides a cognitive adversarial modeling method under incomplete information conditions, which specifically includes the following steps:

[0080] S1) Obtain the main technical parameters of the radar, including:

[0081] Time domain parameters: radar transmit pulse width τ, repetition period PRI, number of waveform agility N w and radar beam dwell time T;

[0082] Frequency domain parameters: center frequency f c , radar receiver bandwidth W r , intra-pulse modulation method M, radar frequency diversity number N F and the number of radar deception signal frequencies N D ;

[0083] Spatial parameters: Radar main lobe half-power point width θ 0.5 ;

[0084] The probability P of radar identifying false targetsd .

[0085] Input jammer jamming pattern S J ={J1,J2…J NJ}.

[0086] S2) According to the known radar technical performance, quantitatively calculate the radar's anti-interference ability, that is, calculate the radar anti-interference improvement factor D i , i=1,2…N R , N R is the total number of anti-interference measures of radar. According to domain knowledge, the anti-interference measures of radar are usually:

[0087] S2.1: Matched filtering and pulse compression technology, the pulse compression signal is usually: linear / nonlinear frequency modulation signal, pulse coding signal. The corresponding improvement factor is:

[0088] D LFM =BτL (1)

[0089] Where D LFM is the anti-interference improvement factor of the linear frequency modulation (LFM) signal, Bτ is the bandwidth-time product of the radar pulse, and L is the pulse compression loss (generally -1.3dB).

[0090] D pm =NL (2)

[0091] Where D pm is the anti-interference improvement factor of the pulse coded signal, N is the phase coding length, and the general value of N is: 7~13 (Barker code), 16~128 (M sequence), 16 (4-phase code).

[0092] After taking the decibel number of the improvement factor, the anti-interference improvement factor is obtained:

[0093] D0 = pulse pressure ratio (or phase encoding length) -1.3dB (3)

[0094] S2.2: Pulse accumulation improvement factor

[0095] Radar pulse integration is divided into coherent integration and incoherent integration; the number of accumulated pulses n is calculated by the following formula:

[0096]

[0097] In the formula, θ 0.5 is the radar main lobe half-power point width, T is the radar beam dwell time, f PRF is the pulse repetition frequency.

[0098] For coherent accumulation, D1 = n 0.8 , for incoherent accumulation, D1 = nV , V∈[0.7,0.9].

[0099] S2.3: Waveform agility technology, the corresponding waveform agility factor is:

[0100] D2=10log 10 N w (5)

[0101] Where N w The number of radar waveforms. Radar increases the difficulty of analysis for the jammer by changing the waveform, forcing the jammer to increase the jammer bandwidth, thereby reducing the jamming effectiveness.

[0102] S2.4: Frequency diversity factor

[0103] D3=10lgN F (6)

[0104] Where N F is the number of radar frequency diversity.

[0105] S2.5: RF masking technology, the corresponding deceptive emission factor is:

[0106] D4=10lg(N D +1) (7)

[0107] Where N D The number of frequencies of radar deception signals. Deception emission means that the radar transmits low-power deception signals from the auxiliary antenna at multiple frequencies, making it impossible for the jammer to accurately aim at the radar's main operating frequency.

[0108] S2.6: False target identification technology

[0109] Since digital radio frequency memory (DRFM) technology is applied to radar jammers, forwarding multi-target deception jamming can successfully enter the radar terminal through the above four anti-jamming measures. Modern radars have adopted corresponding recognition algorithms for multiple false target echoes, and identify false targets from aspects such as waveform parameters, polarization characteristics, and track association. This technology can quantitatively describe its ability to identify false targets based on the radar signal processing capability. The corresponding interference improvement factor formula for false target recognition technology is:

[0110]

[0111] Where N r is the number of false target echoes identified, N totle is the total number of echoes.

[0112] S3) Let s r1 =D0+D1,s r2 =D2,s r3=D3,s r4 =D4,s r5 =D5, then

[0113] A=a.*[s r1 ,s r2 ,s r3 ,s r4 ,s r5 ] (9)

[0114] Where A is the radar benefit matrix, that is, the radar anti-interference strategy matrix, and .* is the dot multiplication operator, which means that each row of a is multiplied by s. r1 ~s r5 Multiply, a is the radar anti-interference strategy allocation matrix, which is an M×N R A matrix, where "0" indicates that the anti-interference measure is not adopted, and "1" indicates that the anti-interference measure is adopted. The number of matrix rows M is the total number of anti-interference strategies that the radar can adopt, which needs to be determined through domain knowledge and public information. In this embodiment, N is selected R =5, M=8.

[0115] S4) Calculate the jammer's jamming effectiveness, including:

[0116] S4.1: Broadband noise suppression benefits:

[0117]

[0118] Where W j is the interference bandwidth, W r is the radar receiver bandwidth. JSR=P j / P r , is the interference-to-signal ratio of the jammer power at the radar antenna aperture, which is calculated based on the position information of the jammer and the radar after the simulation confrontation scenario is determined. j is the interference power, and the distance R from the jammer to the radar antenna j Inversely proportional to the square of P r is the target echo power and the distance R from the target to the radar antenna t Inversely proportional to the fourth power, in the self-defense jamming scenario, there is R j =R t ,but:

[0119]

[0120] Among them G j is the gain of the jammer antenna relative to the radar direction, γ j is the polarization mismatch coefficient of the jammer antenna relative to the radar antenna, P t is the radar transmitting pulse power, G ris the gain of the radar receiving antenna toward the jammer, and σ is the RCS of the protected target.

[0121] S4.2: Narrowband smart noise suppression benefits:

[0122]

[0123] Where, T r =τ is the radar transmission pulse width, k = W r / T r is the radar pulse bandwidth-time width ratio.

[0124] S4.3: 'C&I' Slice Interference Benefits:

[0125] s j3 =10lg(η 2 W r T r )+12,η∈[0,1] (13)

[0126] Where η is the slice coefficient.

[0127] S4.4: Benefits of dense false targets:

[0128] s j4 =10lg(N f )(1-P d ) (14)

[0129] Where N f is the number of false targets.

[0130] S4.5: Distance-speed deception benefits:

[0131] s j5 =JSR(1-P d ) (15)

[0132] but:

[0133] B=b.*[s j1 ,s j2 ,s j3 ,s j4 ,s j5 ] (16)

[0134] Where B is the jammer benefit matrix, also known as the jammer strategy matrix, and b is the jammer strategy allocation matrix, which is an M×N j A matrix, where "0" indicates that the jamming measure is not adopted, "1" indicates that the jamming measure is adopted, the number of rows in the matrix M is the jamming strategy corresponding to the radar anti-jamming strategy, and the number of columns N is jis the jamming style that the jammer can adopt, and the number of rows in matrix b is the same as that in matrix a, that is, the number of jamming strategies corresponding to the anti-jamming strategy of the “Leader” radar. j =5.

[0135] In S2)~S4), after determining the position between the jammer and the target radar, the JSR calculation result is used as the initial value of the benefit matrix. According to the total number of anti-interference measures of the radar and the total number of interference patterns of the jammer, the strategy matrices A and B of the radar and the jammer are calculated, which overcomes the human factors existing in the traditional benefit matrix construction method that relies on the experience of domain experts.

[0136] S5) Based on the strategy matrices A and B of both parties, the radar is the "leader" and the jammer is the "follower". The probability of radar identifying false targets is P d For incomplete information, the double-matrix Stackelberg game of the composite strategy is calculated by linear programming method to calculate the benefits of each jammer's jamming measures and each radar's anti-jamming measures. The game process can be described as follows:

[0137] S5.1: Input strategy matrix A = A T ,B=B T ;T-matrix transpose;

[0138] S5.2: Take each column of A as the coefficient of function f and calculate the game result of the corresponding column of B, that is, calculate the linear programming:

[0139]

[0140] stDx<d,D eq x=d eq ,l b <x<u b

[0141] Among them, the matrix D is the coefficient matrix obtained by subtracting the interference strategy in A from other strategies in B, d,D eq and d eq is the constraint condition; b ,u b is the lower and upper bound of x;

[0142] S5.3: Record all the game results between A and B: fval = f(x j ),j=1,2,…M, select the maximum value of fval as the benefit u of the radar in the game A ;

[0143] S5.4: Benefits of radar anti-jamming u A , calculate the jammer's benefit u B;

[0144] S5.5: Calculate u d =u A -u B , if u d >0, radar outperforms jammer, conversely, jammer outperforms radar.

[0145] The above game process can also be described by the following pseudo program code:

[0146] S5.1: Input strategy matrix: A = A T ,B=B T ; T-matrix transpose; [m,n] = size(A);

[0147] S5.2: for j = 1: n

[0148] f=-A(:,j);

[0149] D = (B(:, [1:j-1, j+1:n]-repmat(B(:, j), 1, n-1)); %D is the difference between the jth interference strategy and other strategies

[0150] if D(:)>0,B(;,j)too strong,break;end

[0151] d=0; Deq=ones(n-1,1); deq=1; lb=zeros(m,1); ub=ones(m,1);

[0152] [x(j),fval]=linprog(f,D,d,Deq,deq,lb,ub);

[0153] %Solve the linear programming:

[0154]

[0155] stDx<d,D eq x=d eq ,l b <x<u b

[0156] Xpayoff(j) = -fval;

[0157] end

[0158] S5.3:[u a ,jBest]=max(Xpayoff);

[0159] S5.4: Xbest = x(jBest,:);

[0160] S5.5:E=eye(n);xBE=x*B*E;

[0161] S5.6:u b =xBE(jBest);

[0162] S5.7:Y=E(:,abs(xBE-u b ) <max(abs(B(:)))*err);%err=1e-6

[0163] S5.8:end

[0164] Among them, Xbest is the best strategy vector of the radar relative jamming strategy, and Y is the relative Xbest jamming strategy matrix.

[0165] S6) Record the number of times the radar working state changes in the previous step, and use the Bayesian method to calculate the posterior probability P of the radar being interfered. When P>0.5, the interference is effective and B remains unchanged; when P≤0.5, the interference is invalid, adjust the interference strategy allocation matrix b, and generate the interference strategy matrix B again. The simulation process ends when the maximum number of simulations is reached.

[0166] In one embodiment of the present invention, the method is used to conduct a simulation experiment through a radar-jammer "1 to 1" confrontation model. In the experiment, the recognition probability P of the radar to the false target is d As incomplete information, to prove the advanced nature of the present invention, the simulation experiment is introduced as follows.

[0167] 1. “1 to 1” confrontation simulation

[0168] The cognitive confrontation based on the double-matrix Stackelberg game can be expressed by formula (17):

[0169] G={(R,J),(A,B),(u A ,u B )} (17)

[0170] In the formula, R-radar, J-jammer, A is the anti-jamming strategy matrix of the radar, and B is the jammer's jamming strategy matrix; u A The payoff for the radar, u B Benefits obtained for the jammer.

[0171] The set of all anti-interference measures taken by the radar is denoted as S R = {R1, R2…R m}, the set of all jamming measures that the jammer can take is denoted as S J ={J1,J2…J n}. For convenience of representation, R m With J n They are respectively recorded as the pure strategies of radar and jammer. The anti-jamming measures of radar and the jammer’s jamming measures are both limited, that is, S R and S J They are all finite sets.

[0172] In this game, the radar is regarded as the leader and the jammer as the follower. The anti-jamming strategy of the radar is not completely known to the jammer. Only reasonable assumptions can be made about the anti-jamming measures in the matrix A based on domain knowledge or public information. For example,

[0173] S r ={'PulseCompress','WaveAgile','F_Agile','RF_decept','FalseT_ID'}

[0174] In the formula, 'PulseCompress'-pulse compression, 'WaveAgile'-waveform agility, 'F_Agile'-frequency agility, 'RF_decept'-radio frequency deception, 'FalseT_ID'-false target identification.

[0175] In practice, when the radar is not interfered with, it will send out detection pulses according to its established mission, x = [A m ,f c ,B w ,M d ,prf] is the radar state variable, which remains unchanged for a certain period of time. The jammer intercepts x through its perception system and analyzes the radar's current anti-jamming strategy. Usually, 'PulseCompress' is an anti-jamming measure implemented by radar hardware, which is definitely a must in A, while other options can be estimated by x.

[0176] S r By making different permutations and combinations, we can get the anti-interference strategy matrix A. For example, when a certain anti-interference measure is adopted, S r The corresponding position is set to "1", otherwise it is set to "0". r The formula can be used to set the radar's countermeasure strategy allocation matrix a:

[0177]

[0178] Among them, the first column of a is all "1" according to the working principle of the radar. The fifth column of a is set to "1" to reflect that the radar's signal processing process runs through the entire confrontation process and is an unpredictable item, that is, the radar anti-interference measure μ; and the second, third, and fourth columns of a are state parameters extracted from the radar state x that can be measured by the interference party through the perception system. For example, the first row of a represents the working state of the radar when it is not interfered with, and the other rows represent the response strategies taken after being interfered with.

[0179] Similarly, for radar, the jammer's jamming strategy can also be set based on prior knowledge, such as:

[0180] S j ={'WBJ','NBJ','C&I','DenseTargets','R&VDecept'}

[0181] Where, 'WBJ'-wideband noise suppression, 'NBJ'-narrowband smart noise suppression, 'C&I'-slice interference, 'DenseTargets'-dense false targets, 'R&VDecept'-range-velocity deception; S j By making different permutations and combinations, we can get the interference strategy matrix B. For example, suppose the interference strategy allocation matrix b is:

[0182]

[0183] After determining the strategy allocation matrix, the radar's anti-interference improvement factor is calculated to generate the strategy matrix A. Similarly, the strategy matrix B is generated by calculating the jammer's interference benefit factor.

[0184] In order to verify the effectiveness of the proposed method, a confrontation simulation scenario is established with AN / SPY-1 radar as the target: the protected target is assumed to have its own self-defense jammer, and its radar cross section (RCS) σ = 0.03m 2 , the slant range from the radar is the maximum range of the radar relative to the RCS R t =R max =310Km, the vertical height of the target relative to the ground is H=20Km.

[0185] According to public information, the main technical parameters of the radar are as follows:

[0186] Center frequency f c =3.1~3.5GHz(λ=8.6~9.7cm)

[0187] Instantaneous bandwidth Wr=40MHz, up to 400MHz

[0188] With ISAR capability, distance resolution Δr = 0.5 ~ 1m

[0189] Antenna aperture: 12m 2

[0190] Antenna gain G = 42dB

[0191] Beam width (MLBr): 1.7°×1.7°

[0192] Instantaneous transmission power P t =4~6MW; average power P av =58KW

[0193] Transmit pulse width (SPY-1B): 6.4, 12.7, 25, 51 μs

[0194] Pulse pressure ratio k = 128

[0195] Receiver noise figure: 4.25dB

[0196] Maximum detection distance: RCS = 0.0025m 2 ,R max =165Km@3.3GHz; RCS=1m 2 ,R max =740Km, RCS=0.03m 2 ,R max =310Km.

[0197] The main technical parameters of the jammer are as follows:

[0198] Interference power P j =1KW

[0199] Antenna gain G j =20dB

[0200] Interference bandwidth W j =2Wr

[0201] Interference beamwidth MLB j =2MLBr

[0202] Polarization mismatch coefficient γ j =0.5 (assuming the jammer is circularly polarized).

[0203] The radar anti-interference improvement factor calculated based on the above parameters is as follows:

[0204] Interference-to-signal ratio at the radar antenna port: JSR = 69.5397

[0205] Pulse compression cumulative benefit: S r1 =41.4257

[0206] Waveform Agility Benefit:S r2 =20(pulse accumulation number n p=100)

[0207] Frequency diversity benefit: S r3 =13.0103(frequency diversity number N f =20)

[0208] RF shielding effect: S r4 =7.7815(Number of protected radio frequencies d =5)

[0209] False target identification benefit: S r5 =34.7698 (let false target recognition probability P d =0.5)

[0210] Benefits of Jammers:

[0211] Broadband suppression benefits: S j1 =72.55

[0212] Narrowband aiming noise suppression benefit:S j2 =101.5567

[0213] C&I Slice Interference Benefit:S j3 =23.6866 (η=0.24; slice width 1 / 4 pulse width)

[0214] Dense false target jamming effect: S j4 =10(Number of false targets N ft =100)

[0215] Distance-speed drag interference benefit: S j5 =34.7698.

[0216] The radar strategy matrix A is generated according to the anti-interference strategy allocation matrix a in the previous section; the jammer strategy matrix B is generated according to the anti-interference strategy allocation matrix b. It should be noted that the radar, as a "leader", has a relatively fixed strategy matrix, while the jammer, as a "follower", can formulate different strategy matrices. The jammer strategy allocation matrix here is a response strategy formulated for A, that is, at least one power suppression and deception jamming style is adopted at the same time.

[0217] The probability of radar identifying a false target is P d As unknown information, let P d =0.1:0.1:1 (i.e. step length 0.1) increasing, the game results are as shown in the attached Figure 2 shown.

[0218] By the attached Figure 2 It can be seen that when the radar detects a false target, the probability P d <0.6, the benefit of the jammer ub Greater than the benefit obtained by radar a On the contrary, u a Follow P d Linear increase, u b Jump to less than u a This is the result of using linear programming in the double matrix Stackelberg game.

[0219] As attached Figure 3 As shown, it is the detection probability P of the false target d =0.5, let the number of false targets N f =100:10:1000 (i.e., the step size is 10) in the confrontation simulation results. The jammer outperforms the radar. Figure 4 P d =0.6, the number of false targets N f =100:10:1000 confrontation simulation results, the radar is better than the jammer. It can be seen that the radar's false target recognition ability is the key factor that determines the success or failure of the confrontation. In practice, the jammer can only observe the radar state variable x = [A m ,f c ,B w ,M d ,prf] estimates whether the interference is effective. Here, "0" and "1" are used to represent the changes of each element of the radar state variable x after a confrontation. There are x=2 5 states; let the prior probability density of x be g m Following the Gaussian mixture distribution model (GMM), after a confrontation, the posterior probability of the radar state is calculated using the Bayesian method: P = posterior (g m ,x), where g m =fitgmdist(x,5) is to fit the GMM distribution model; if P<0.5, the current interference strategy B is considered invalid and needs to be changed; otherwise, the current interference strategy is considered effective and B remains unchanged. To truly realize this countermeasure simulation, it is necessary to use a semi-physical experimental system in a microwave darkroom and collect a large amount of experimental data to fit g m .

[0220] Due to the incompleteness of the actual detection information of each other by the two opposing parties in the electromagnetic spectrum warfare, the traditional evaluation of the confrontation effect has become subjective. With the support of domain knowledge and public information, the present invention proposes a radar cognitive anti-interference modeling and simulation evaluation method under the condition of incomplete perception information. The proposed radar anti-interference improvement factor and the interference benefit calculation method of the jammer meet the actual engineering needs and have strong universality, providing a new way to solve this problem. The proposed method is based on the fact that the jammer can accurately obtain the working status of the radar through the electronic reconnaissance system, and does not consider the impact of measurement errors on the game results. This premise is based on the fact that the existing electronic reconnaissance system has a high signal analysis accuracy, and its measurement error will not cause subversive results.

[0221] From a methodological point of view, the proposed method can be extended to the modeling and evaluation of the system confrontation between radar networks and swarm jammers. As long as the elements of system confrontation and the types of incomplete information are sorted out, a reasonable evaluation of the confrontation effect under different conditions can be made.

Claims

1. A cognitive adversarial modeling method under incomplete information conditions, characterized in that: The steps include: Step 1: Construct radar anti-interference improvement factor D i Calculation method, i=1,2…N R , N R is the total number of anti-jamming measures of the radar; Step 2: After the simulation confrontation scenario is determined, the interference-to-signal ratio JSR of the jammer power on the radar antenna aperture is calculated according to the location information of the jammer and the radar; Step 3: Take the total number of radar anti-interference measures N R Based on this, the radar anti-interference strategy allocation matrix a is constructed, where "0" indicates the anti-interference measures not adopted, and "1" indicates the anti-interference measures adopted; the radar benefit matrix A is generated; Step 4: Take the total number of jammer measures N j Based on this, the interference strategy allocation matrix b is constructed, where "0" represents the interference measure not adopted, and "1" represents the interference measure adopted; Generate jammer benefit matrix B; Step 5: With the radar as the leader and the jammer as the follower, the probability of radar identifying false targets is P d For incomplete information, a double-matrix Stackelberg game with a compound strategy is conducted to calculate the benefits of each jamming measure of the jammer and each anti-jamming measure of the radar. Step 6, record the number of times the radar working state changes in step 5, and use the Bayesian method to calculate the posterior probability P of the radar being interfered. When P>0.5, the interference is effective and B remains unchanged; When P≤0.5, the interference is invalid, the interference strategy allocation matrix b is adjusted, and step 4 is repeated. The simulation process ends when the maximum number of simulations is reached.

2. According to the cognitive adversarial modeling method under incomplete information conditions according to claim 1, it is characterized in that: In step 1, radar anti-interference improvement factor D i The calculation methods include: 1), matched filtering and pulse compression technology Pulse pressure signals include: linear / nonlinear frequency modulation signals, pulse coding signals, and the corresponding improvement factors are: D LFM =BτL D pm =NL Where D LFM is the anti-interference improvement factor of the linear frequency modulation (LFM) signal, Bτ is the bandwidth-time product of the radar pulse, and L is the pulse compression loss; D pm is the anti-interference improvement factor of the pulse coding signal, N is the phase coding length, when the Barker code is used, the value of N is 7 to 13, when the M sequence is used, the value of N is 16 to 128, and when the 4-phase code is used, the value of N is 16; After taking the decibel number for the improvement factor, the anti-interference improvement factor obtained is: D0 = pulse pressure ratio or code length - 1.3dB 2), Pulse accumulation improvement factor Radar pulse accumulation is divided into coherent accumulation and incoherent accumulation; the number of accumulated pulses is calculated as follows: In the formula, θ 0.5 is the radar main lobe half-power point width, T is the radar beam dwell time, f PRF is the pulse repetition frequency; For coherent accumulation, D1 = n 0.8 , for incoherent accumulation, D1 = n V , V∈[0.7,0.9]; 3), Waveform Agility Factor D2=10lgN w Where N w The number of waveform agility changes. Radar increases the difficulty of analyzing the jammer through waveform changes, forcing the jammer to increase the jamming bandwidth, thereby reducing the jamming effectiveness. 4), frequency diversity factor D3=10lgN F Where N F is the number of radar frequency diversity; 5), Deceptive emission factor: D4=10lg(N D +1) Where N D The number of frequencies of radar deception signals. Deception emission means that the radar transmits low-power deception signals from the auxiliary antenna at multiple frequencies, making it impossible for the jammer to accurately aim at the radar's main operating frequency. 6) False target identification technology Where N r is the number of false target echoes identified, N totle is the total number of echoes; Order r1 =D0+D1,s r2 =D2,s r3 =D3,s r4 =D4,s r5 =D5, then A = a.*[s r1 ,s r2 ,s r3 ,s r4 ,s r5 ], .* is the dot multiplication operator, which means that each element of a and s r1 ~s r5 Multiply, a is M×N R The number of rows M is the total number of anti-interference strategies that the radar can adopt.

3. According to the cognitive adversarial modeling method under incomplete information conditions as described in claim 1, it is characterized in that: In step 2, the calculation formula of the interference signal ratio JSR of the interference power at the radar antenna aperture is as follows: Where P j is the interference power, and the distance R from the jammer to the radar antenna j Inversely proportional to the square of P r is the target echo power and the distance R from the target to the radar antenna t Inversely proportional to the fourth power, in the self-defense jamming scenario, there is R j =R t ; G j is the gain of the jammer antenna relative to the radar direction, γ j is the polarization mismatch coefficient of the jammer antenna relative to the radar antenna, P t is the radar transmitting pulse power, G r is the gain of the radar receiving antenna toward the jammer, and σ is the RCS of the protected target.

4. According to the cognitive adversarial modeling method under incomplete information conditions as claimed in claim 3, it is characterized in that: In step 4, the interference benefit factor of the jammer includes: (1)Broadband noise suppression benefits: Where W j is the interference bandwidth, W r is the radar receiver bandwidth; (2) Narrowband smart noise suppression benefits: Where, T r =τ is the radar transmission pulse width, k = W r / T r is the radar pulse bandwidth-time width ratio; (3) 'C&I' Slice Interference Benefit: s j3 =101g(h 2 W r T r )+12,η∈[0,1] Where η is the slice coefficient; (4) Benefits of dense false targets: s j4 =10lg(N f )(1-P d ) Where N f is the number of false targets; (5) Distance-speed deception benefits: s j5 =JSR(1-P d ) Then the interference strategy matrix B = b.*[s j1 ,s j2 ,s j3 ,s j4 ,s j5 ], after A and B are determined, a double matrix Stackelberg game with a composite strategy is performed, and b is M×N j The matrix whose row number M is the jammer's jamming strategy corresponding to the radar's anti-jamming strategy.

5. According to the cognitive adversarial modeling method under incomplete information conditions as claimed in claim 4, it is characterized in that: The double matrix Stackelberg game process of the composite strategy is as follows: S1: Input strategy matrix A = A T ,B=B T ;T-matrix transpose; S2: Take each column of A as the coefficient of function f and calculate the game result of the corresponding column of B, that is, calculate the linear programming: Among them, the matrix D is the coefficient matrix obtained by subtracting the interference strategy in A from other strategies in B, d,D eq and d eq is the constraint condition; b ,u b is the lower and upper bound of x; S3: Record all the game results between A and B: fval = f(x j ),j=1,2,…M, select the maximum value of fval as the benefit u of the radar in the game A ; S4: Benefits of radar anti-jamming u A , calculate the jammer's benefit u B ; S5: Calculate u d =u A -u B , if u d >0, radar outperforms jammer, conversely, jammer outperforms radar.

6. The cognitive adversarial modeling method under incomplete information conditions according to claim 5 is characterized in that: In step 6, the jammer observes the radar state variable x=[A m ,f c ,B w ,M d ,prf] estimates whether the jamming is effective, and uses "0" and "1" to indicate whether the radar state variable x has changed significantly after a confrontation. There are x = 2 5 states; let the prior probability density of x be g m Following the Gaussian mixture distribution model, after a confrontation, the posterior probability of the radar state is calculated using the Bayesian method: P = posterior (g m ,x), where g m =fitgmdist(x,5) is to fit the Gaussian mixture distribution model. If P<0.5, it is considered that the current interference strategy B is invalid and B needs to be changed. Otherwise, it is considered that the current interference strategy is effective and B remains unchanged.