CN115201768A - Active deception jamming method for generating countermeasure network based on cycle consistency - Google Patents
Active deception jamming method for generating countermeasure network based on cycle consistency Download PDFInfo
- Publication number
- CN115201768A CN115201768A CN202210655328.1A CN202210655328A CN115201768A CN 115201768 A CN115201768 A CN 115201768A CN 202210655328 A CN202210655328 A CN 202210655328A CN 115201768 A CN115201768 A CN 115201768A
- Authority
- CN
- China
- Prior art keywords
- signal sequence
- interference
- cycle
- echo signal
- signals
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
- G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
- G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
- G01S7/38—Jamming means, e.g. producing false echoes
Abstract
According to the active deception jamming method for generating the countermeasure network based on the Cycle consistency, provided by the invention, a C & I jamming signal sequence and an SMSP jamming signal sequence which have more false targets and stronger jamming strength can be generated as a judgment standard in a radar range-Doppler domain through a series of steps of radar target echo interception, interval sampling, adjacent filling and the like, and an original echo signal sequence which does not contain jamming is used as the input of a Cycle-GAN model. The Cycle-GAN model is a cyclic training model formed by a pair of GAN models, the effect of improving the stability of the GAN model on the basis of not reducing the accuracy of a generator can be realized, and the performance of the Cycle-GAN model is determined by comparing the number of the detected false targets with the total target number. Therefore, the invention can enhance the interference energy, generalize the interference pattern and obtain the interference effect which is difficult to identify in the range-Doppler domain.
Description
Technical Field
The invention belongs to the technical field of radar signals, and particularly relates to an active deception jamming method for generating an anti-network based on cycle consistency.
Background
With the continuous development of radar detection and anti-interference technologies, the insufficient interference performance of the active deception interference technology becomes an urgent problem to be solved in electronic countermeasure. In the prior art, deceptive interference modeling is performed on a pulse compression radar in a scheme, two new interference pattern slice interference (C & I) and dispersive Spectrum interference (SMSP) are provided, interference signals of the two interference patterns are intercepted, modulated and forwarded, and the interference signals generated through the two interference patterns generate false targets, so that the detection performance of the radar is limited, and an interference effect is achieved. In another scheme, an intermittent sampling direct forwarding interference pattern based on a Digital Radio Frequency Memory (DRFM) technology is provided, interference with double effects of deception and suppression is achieved, and the interference effect of the interference pattern on a linear Frequency modulation pulse compression radar is verified in an experiment. In another scheme, interference modeling is performed on two radar systems of a pulse compression radar and a pulse Doppler radar, and the problems of cooperative interference and the like are also researched.
The researchers in the above schemes in the prior art mainly focus on the aspects of "dense false target interference of chirp radar", "interference mode of intermittent sampling of DRFM", and "multiplexing of different types of interference modes". However, in these schemes, C & I interference only includes deceptive interference, and the interference strength is not high; SMSP interference has the problem that strongly scattering targets cannot be annihilated, and these schemes are prone to difficult recognition when dealing with different interference.
Disclosure of Invention
In order to solve the above problems in the prior art, the present invention provides an active spoofing interference method for generating an anti-network based on loop consistency. The technical problem to be solved by the invention is realized by the following technical scheme:
in a first aspect, the present invention provides a method for generating active spoofed interference for a countermeasure network based on loop consistency, including:
step 1: after receiving a radar transmitting signal, returning an original echo signal sequence without interference;
step 2: generating a C & I interference signal sequence and an SMSP interference signal sequence by performing interval sampling and adjacent filling on the original echo signal sequence;
and step 3: inputting the original echo signal sequence into a generator of a Cycle-GAN model, judging the interference signal generated by the generator by taking the signal sequence added with the C & I interference signal sequence and the SMSP interference signal sequence as a judgment standard of a judger, and adjusting internal parameters of the generator in a loss function reduction mode until the trained Cycle-GAN model is obtained through judgment of the judger;
and 4, step 4: setting a plurality of real targets, and transmitting signals in a real-time manner to receive real-time echo signals fed back by the real targets;
and 5: inputting the real-time echo signals into a trained Cycle-GAN model to output echo signals carrying interference;
step 6: according to the amplitude of the signal, a wave gate of an echo signal carrying interference is defined so as to detect whether a target exists or not and the number of the targets;
and 7: and comparing the number of the detected targets with the real targets to determine the performance of the Cycle-GAN model.
Optionally, generating a C & I interference signal sequence by using the original echo signal sequence in a manner of interval sampling and adjacent padding includes:
dividing the original echo signal sequence into m subsections so that the time gap number of each subsection is n;
dividing each sub-segment into smaller sub-segments;
and for each sub-section, copying a first small sub-section in the sub-section, and filling the sub-section with the copied structure until the sub-section length is reached to obtain a C & I interference signal sequence.
Optionally, the generating an SMSP interference signal sequence by using the original echo signal sequence through interval sampling and adjacent padding includes:
extracting signals from the original echo signal sequence at an interval of n times, and sequencing the extracted signals in the original echo signal sequence;
and copying the sorted and extracted signals until the length of the signals is the same as that of the original echo signal sequence to obtain an SMSP interference signal sequence.
Optionally, the step 3 includes:
inputting the original echo signal sequence into a generator of a Cycle-GAN model, taking the C & I interference signal sequence and the SMSP interference signal sequence as target sequences, generating the original echo signal sequence towards the direction of the interfered signal sequence through continuous training so that the generator continuously generates samples with interference signals, obtaining a judgment score through judging the samples with the interference signals by a judgment device, feeding back the judgment score to the generator so that the generator adjusts parameters through the judgment score, gradually approaching the training samples to the signal sequence added with the target sequences, and ending the whole training process until the judgment device is difficult to distinguish whether the generator generates the signal sequence, thereby obtaining the trained Cycle-GAN model.
Optionally, the Cycle-GAN model converts the original echo signal sequence from the source signal sequence domain X to the target signal sequence domain Y, which is denoted as mapping G: X → Y, so that the signal sequence finally generated in mapping G and the signal sequence in the target signal sequence domain cannot be identified by the determiner; the Cycle-GAN model performs the task of F (G (X)) ≈ X through the inverse map generator F: Y → X.
Optionally, step 7 includes:
comparing the number of the detected targets with the real target, and judging whether the difference value between the number of the detected targets and the real target is greater than a threshold value;
if the performance of the Cycle-GAN model is higher than the threshold value, the performance of the Cycle-GAN model is determined to be good, and if the performance of the Cycle-GAN model is not higher than the threshold value, the performance of the Cycle-GAN model is not good.
In a second aspect, the present invention provides an active spoofing interference system for generating an anti-jamming network based on cycle consistency, including:
the receiving module is used for receiving an original echo signal sequence which is returned after a radar transmitting signal and does not contain interference;
the interference generation module is used for generating a C & I interference signal sequence and an SMSP interference signal sequence in a mode of interval sampling and adjacent filling for the original echo signal sequence;
the training module is used for inputting the original echo signal sequence into a generator of a Cycle-GAN model, judging the interference signal generated by the generator by taking the signal sequence added with the C & I interference signal sequence and the SMSP interference signal sequence as a judgment standard of a judger, and adjusting internal parameters of the generator in a loss function reduction mode until the trained Cycle-GAN model is obtained through judgment of the judger;
the acquisition module is used for setting a plurality of real targets and transmitting signals in a real-time manner to acquire real-time echo signals fed back by the real targets;
the signal generation module is used for inputting the real-time echo signal to a trained Cycle-GAN model so as to output the echo signal carrying the interference signal;
the detection module is used for demarcating a wave gate of an echo signal carrying interference according to the amplitude of the signal so as to detect whether targets exist or not and the number of the targets;
and the performance determining module is used for comparing the number of the detected targets with the real targets so as to determine the performance of the Cycle-GAN model.
Optionally, the interference generating module is specifically configured to:
dividing the original echo signal sequence into m subsections so that the time gap number of each subsection is n;
dividing each sub-segment into smaller small sub-segments;
and for each sub-segment, copying a first small sub-segment in the sub-segment, and filling the sub-segment with the copied structure until the length of the sub-segment is reached to obtain the C & I interference signal sequence.
Optionally, the interference generating module is specifically configured to:
carrying out up-sampling on the original echo signal sequence in a mode of n times of original sampling to obtain an up-sampled echo signal sequence;
extracting signals from the sampled echo signal sequence at an interval of n times, and sequencing the extracted signals in the original echo signal sequence;
and copying the sorted and extracted signals until the length of the signals is the same as that of the original echo signal sequence to obtain an MSP interference signal sequence.
Optionally, the training module is specifically configured to:
inputting the original echo signal sequence into a generator of a Cycle-GAN model, taking the C & I interference signal sequence and the SMSP interference signal sequence as target sequences, generating the original echo signal sequence towards the direction of the interfered signal sequence through continuous training so that the generator continuously generates samples with interference signals, obtaining a judgment score through judging the samples with the interference signals by a judgment device, feeding back the judgment score to the generator so that the generator adjusts parameters through the judgment score, gradually approaching the training samples to the signal sequence added with the target sequences, and ending the whole training process until the judgment device is difficult to distinguish whether the generator generates the signal sequence, thereby obtaining the trained Cycle-GAN model.
The invention has the beneficial effects that:
according to the active deception jamming method and system for generating the countermeasure network based on the Cycle consistency, a C & I jamming signal sequence and an SMSP jamming signal sequence which are more in false target number and stronger in jamming strength can be generated in a radar distance-Doppler domain through a series of steps of radar target echo interception, interval sampling, adjacent filling and the like, and an original echo signal sequence which does not contain jamming is used as the input of a Cycle-GAN model. The Cycle-GAN model is a cyclic training model formed by a pair of GAN models, the effect of improving the stability of the GAN model on the basis of not reducing the accuracy of a generator can be realized, and the performance of the Cycle-GAN model is determined by comparing the number of the detected false targets with the total target number. Therefore, the invention can enhance the interference energy, generalize the interference pattern and obtain the interference effect which is difficult to identify in the range-Doppler domain.
The present invention will be described in further detail with reference to the accompanying drawings and examples.
Drawings
Fig. 1 is a flowchart of an interference generation algorithm provided by an embodiment of the present invention;
fig. 2 is a schematic diagram of C & I interference generation provided by an embodiment of the present invention;
FIG. 3 is a simulation diagram of a radar transmit waveform and a C & I interference time domain waveform provided by an embodiment of the present invention;
fig. 4 is a schematic block diagram of an SMSP interference generation provided by an embodiment of the present invention;
FIG. 5 is a simulation diagram of a radar transmitting waveform and an SMSP interference time domain waveform provided by the embodiment of the invention;
FIG. 6 is a schematic structural diagram of a Cycle-GAN model provided in an embodiment of the present invention;
FIG. 7 is a graph comparing a disturbance free echo to a disturbance containing RD provided by an embodiment of the present invention;
fig. 8 is an interference performance evaluation diagram according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to specific examples, but the embodiments of the present invention are not limited thereto.
As shown in fig. 1, the method for generating active spoofed interference for a countermeasure network based on loop consistency provided by the present invention includes:
step 1: after receiving radar emission signals, returning an original echo signal sequence which does not contain interference;
the invention obtains an original echo signal sequence without containing interference in an experimental simulation mode, and parameters of radar waveform simulation in the experiment are as follows: the working frequency is 90MHz, the bandwidth is 1.25MHz, the transmitting power is 1000W, the antenna gain is 4000, the pulse width is 200, the sampling rate is 4MHz, the number of coherent pulses is 64, and the maximum target number of each frame is 10. The simulated radar original echo signal sequence is used as input data of a network, interference signals are generated by configuring interference parameters, C & I interference and SMSP interference signal sequences are added into clean echo signals to be used as target data of the network, and the interference signals are added into the original echo signals.
Step 2: generating a C & I interference signal sequence and an SMSP interference signal sequence according to the original echo signal sequence;
the invention can divide the original echo signal sequence into m subsections, so that the time gap number of each subsection is n; dividing each sub-segment into smaller sub-segments; and for each sub-segment, copying a first small sub-segment in the sub-segment, and filling the sub-segment with the copied structure until the length of the sub-segment is reached to obtain the C & I interference signal sequence.
The C & I interference adopted by the invention is mainly realized by copying and forwarding radar clean echo signals received by an interference machine. Firstly, intercepting a radar signal by an interference machine, and simultaneously storing the radar signal in a DRFM (digital radio frequency modulation); then sampling the signal in the DRFM by adopting rectangular pulse trains with the same interval in the cropping stage; and finally, in an Interleaving stage, copying the small segment of signals intercepted and sampled in the second step to fill the adjacent intervals until the intervals are filled completely.
As shown in fig. 2, a C & I interference generation simulation flow firstly divides a radar signal into N = m · N sections, where m is the number of rectangular pulse trains, and cuts the whole section of signal into m subsegment signals; n is the time gap number of the subsegment, namely, the subsegment signals are divided into smaller intervals and filled into the adjacent intervals, and the generated subsegment signals with the time width of T/N (T is the time width of the radar signals) are stored into the DRFM; then extracting sub-segment signals obtained by sampling radar signals from the DRFM through a cropping stage; and finally, duplicating the sub-segment signals and filling the sub-segment signals into the intervals of adjacent signals through an Interleaving stage, thereby generating C & I interference.
The radar parameters in the simulation are set as follows: the radar bandwidth is 1.25MHz, the signal sampling rate is 4MHz, and the pulse width is 200 mus; considering the balance relation between the interference performance and the calculated amount, the C & I interference parameter is set as follows: the number of rectangular pulse trains, the number of time gaps of the subsections, and the whole section of the signal are divided into 20 sections. Thereby obtaining the radar transmitting signal time domain waveform and the C & I interference time domain waveform shown in figure 3.
The invention can extract signals from the original echo signal sequence at an interval of n times, and sort the extracted signals according to the original sequence in the original echo signal sequence; and copying the sorted and extracted signals until the length of the signals is the same as that of the original echo signal sequence to obtain an SMSP interference signal sequence.
Another interference pattern SMSP interference generation principle employed by the present invention is shown in fig. 4. The generation process comprises the steps of firstly, after the clock frequency is increased to n times of the original clock frequency, extracting the sampling data and sequencing the sampling data in sequence according to the original sequence; in the second step, the signal generated in the first step is copied n times, so that the SMSP interference is obtained.
In the simulation process, the SMSP interference changes the clock frequency to cause the frequency of digital-to-analog conversion and the frequency of analog-to-digital conversion to be different, so that the frequency modulation slope of the generated interference signal is different from that of the original echo, and the SMSP interference signal comprising a plurality of sub-pulses with the same structure can be obtained by repeating the process.
Simulation parameters of SMSP interference are similar to C & I interference, and time width and bandwidth of adopted linear frequency modulation signals are respectively as follows: 200 mus, 1.25MHz, and divides the SMSP perturbation into four sub-pulses, the simulation results are shown in fig. 5.
And step 3: inputting the original echo signal sequence into a generator of a Cycle-GAN model, taking the signal sequence added with the C & I interference signal sequence and the SMSP interference signal sequence as a decision standard of a decision device, judging the interference signal generated by the generator, and adjusting internal parameters of the generator in a loss function reduction manner until the trained Cycle-GAN model is obtained through the decision of the decision device;
the original echo signal sequence can be input into a generator of a Cycle-GAN model, the C & I interference signal sequence and the SMSP interference signal sequence are used as target sequences, the original echo signal sequence is generated towards the direction of the interfered signal sequence through continuous training so that the generator continuously generates samples with interference signals, a judgment score is obtained through judging the samples with the interference signals by a judgment device and is fed back to the generator, the generator adjusts parameters through the judgment score fed back, training samples gradually approach the signal sequence added with the target sequence, when the judgment device is difficult to distinguish whether the signal sequence generated by the generator is true or false, the whole training process is finished, and the trained Cycle-GAN model is obtained.
As shown in FIG. 6, the Cycle-GAN model of the present invention has a structure as shown in FIG. 6. The Cycle-GAN model converts an original echo signal sequence from a source signal sequence domain X to a target signal sequence domain Y, and the original echo signal sequence is marked as mapping G: X → Y, so that a signal sequence finally generated in the mapping G and a signal sequence of the target signal sequence domain cannot be identified through a decision device; the Cycle-GAN model performs the task of F (G (X)) ≈ X through the inverse map generator F: Y → X.
In Cycle-GAN, the loss functions of a pair of GAN models are as follows:
wherein D is X And D Y For the X and Y domain deciders, respectively, G (X) and F (Y) being functions of the signal sequences generated by generators G and F, P X And P Y Respectively judging the probability that the signal sequence comes from an X domain and a Y domain for a decision device, wherein Cycle-GAN adopts Cycle consistency loss, and the formula is as follows:
calculating the loss in G and F iterative process by the loss function, wherein L cyc For continuous loss, the loss functions are then weighted and summed to obtain the loss function of the final Cycle-GAN:
L(G,F,P X ,P Y )＝L GAN2 (F,P X ,X,Y)+L GAN1 (G,P Y ,Y,X)+λL cyc (G,F) (4)
where λ is the weight of the loss of consistency, which is a real number greater than zero.
And 4, step 4: setting a plurality of real targets, and transmitting signals in a real-time manner to receive real-time echo signals fed back by the real targets;
the method can set 6000 original echo samples to be generated, and allocates the 6000 samples as target signals according to 1; then, performing network training by using the training set data and the target signal to obtain the network parameters of the Cycle-GAN; and inputting the test set sample into the network, and finally obtaining an interference signal generated by the network after training is finished.
And 5: inputting the real-time echo signals into a trained Cycle-GAN model to output echo signals carrying interference;
step 6: according to the amplitude of the signal, a wave gate of an echo signal carrying interference is defined so as to detect whether a target exists or not and the number of the targets;
referring to fig. 7, fig. 7 shows a doppler image of a raw echo signal to which interference is added in the present invention. In fig. 7, a sub-graph (a) is an original echo signal range-doppler domain image without noise, a sub-graph (b) is an image with a C & I interference range-doppler domain, a sub-graph (C) is an image with an SMSP interference range-doppler domain, a sub-graph (d) generates a C & I interference range-doppler domain image for adding Cycle-GAN, and a sub-graph (e) generates an SMSP interference range-doppler domain image for adding Cycle-GAN.
And 7: and comparing the number of the detected targets with the real targets to determine the performance of the Cycle-GAN model.
Comparing the number of the detected targets with the number of the real targets, and judging whether the difference value between the number of the detected targets and the number of the real targets is greater than a threshold value; if the performance of the Cycle-GAN model is not greater than the threshold value, the Cycle-GAN model is represented to have poor performance.
The invention transforms the echo signal without interference and the interference signal generated by the network to the RD domain, and compares the number of true and false targets to obtain the final interference performance.
As shown in FIG. 8, the invention can improve the C & I interference and SMSP interference performance through 1000 times of experimental verification through Cycle-GAN network training, and the interference performance of each point in the graph is the performance average value of 50 times of interference in the experiment. The interference performance of the C & I interference is kept between 0.51 and 0.69, the average interference performance can reach 0.64, the detection rate is 0.36, the performance of the C & I interference after network training is distributed between 0.62 and 0.89, the average interference performance can reach 0.76, the average correct detection rate is only 0.24, and the average interference performance is improved by about 12 percent after network training. In addition, the performance of the SMSP interference is distributed between 0.64 and 0.80, the interference performance can reach 0.71, and the detection rate is 0.39, however, the interference performance of the SMSP interference after network training can be kept between 0.78 and 0.93, the average interference performance can reach 0.84 through calculation, the average correct detection rate is only 0.16, and the average interference performance is improved by about 13%. The experiment shows that the interference based on network training is improved by 12-13% compared with the single C & I interference and SMSP interference. In addition, the experimental result also shows that the invention can effectively increase the energy of the interference signal and achieve the effect of deception and suppression of double interference. Therefore, the performance of the C & I interference and the SMSP interference generated based on the Cycle-GAN network is superior to that of the pure C & I interference and the SMSP interference.
The invention provides an active deception jamming system for generating a countermeasure network based on cycle consistency, which comprises:
the receiving module is used for receiving an original echo signal sequence which is returned after a radar transmitting signal and does not contain interference;
the interference generation module is used for generating a C & I interference signal sequence and an SMSP interference signal sequence in a mode of interval sampling and adjacent filling for the original echo signal sequence;
the training module is used for inputting the original echo signal sequence into a generator of a Cycle-GAN model, judging the interference signal generated by the generator by taking the signal sequence added with the C & I interference signal sequence and the SMSP interference signal sequence as a judgment standard of a judger, and adjusting internal parameters of the generator in a loss function reduction mode until the trained Cycle-GAN model is obtained through judgment of the judger;
the acquisition module is used for setting a plurality of real targets and transmitting signals in a real-time manner to acquire real-time echo signals fed back by the real targets;
the signal generation module is used for inputting the real-time echo signal to a trained Cycle-GAN model so as to output the echo signal carrying the interference signal;
the detection module is used for demarcating a wave gate of an echo signal carrying interference according to the amplitude of the signal so as to detect whether targets exist or not and the number of the targets;
and the performance determining module is used for comparing the number of the detected targets with the real targets so as to determine the performance of the Cycle-GAN model.
Optionally, the interference generating module is specifically configured to:
dividing the original echo signal sequence into m subsegments so as to enable the time gap number of each subsegment to be n;
dividing each sub-segment into smaller small sub-segments;
and for each sub-segment, copying a first small sub-segment in the sub-segment, and filling the sub-segment with the copied structure until the length of the sub-segment is reached to obtain the C & I interference signal sequence.
Optionally, the interference generating module is specifically configured to:
the original echo signal sequence is up-sampled in an original n-time sampling mode to obtain an up-sampled echo signal sequence;
extracting signals from the sampled echo signal sequence at an interval of n times, and sequencing the extracted signals in the original echo signal sequence;
and copying the sorted and extracted signals until the length of the signals is the same as that of the original echo signal sequence to obtain an SMSP interference signal sequence.
Optionally, the training module is specifically configured to:
inputting the original echo signal sequence into a generator of a Cycle-GAN model, taking the C & I interference signal sequence and the SMSP interference signal sequence as target sequences, generating the original echo signal sequence towards the direction of the interfered signal sequence through continuous training so that the generator continuously generates samples with interference signals, obtaining a judgment score through judging the samples with the interference signals by a judgment device, feeding back the judgment score to the generator so that the generator adjusts parameters through the judgment score, gradually approaching the training samples to the signal sequence added with the target sequences, and ending the whole training process until the judgment device is difficult to distinguish whether the generator generates the signal sequence, thereby obtaining the trained Cycle-GAN model.
According to the active deception jamming method and system for generating the countermeasure network based on the Cycle consistency, provided by the invention, a C & I jamming signal sequence and an SMSP jamming signal sequence with more false targets and stronger jamming strength can be generated as a judgment standard in a radar range-Doppler domain through a series of steps of radar target echo interception, interval sampling, adjacent filling and the like, and an original echo signal sequence without jamming is used as the input of a Cycle-GAN model. The Cycle-GAN model is a circular training model formed by a pair of GAN models, the effect of improving the stability of the GAN model on the basis of not reducing the accuracy of a generator can be realized, and the performance of the Cycle-GAN model is determined by comparing the number of the detected false targets with the total number of the targets. Therefore, the invention can enhance the interference energy, generalize the interference pattern and obtain the interference effect which is difficult to identify in the range-Doppler domain.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, numerous simple deductions or substitutions may be made without departing from the spirit of the invention, which shall be deemed to belong to the scope of the invention.
Claims (10)
1. A method for generating active deceptive jamming of a countermeasure network based on cycle consistency, the method comprising:
step 1: after receiving radar emission signals, returning an original echo signal sequence which does not contain interference;
step 2: generating a C & I interference signal sequence and an SMSP interference signal sequence by performing interval sampling and adjacent filling on the original echo signal sequence;
and step 3: inputting the original echo signal sequence into a generator of a Cycle-GAN model, judging the interference signal generated by the generator by taking the signal sequence added with the C & I interference signal sequence and the SMSP interference signal sequence as a judgment standard of a judger, and adjusting internal parameters of the generator in a loss function reduction mode until the trained Cycle-GAN model is obtained through judgment of the judger;
and 4, step 4: setting a plurality of real targets, and transmitting signals in a real-time manner to receive real-time echo signals fed back by the real targets;
and 5: inputting the real-time echo signal into a trained Cycle-GAN model to output an echo signal carrying interference;
step 6: according to the amplitude of the signal, a wave gate of an echo signal carrying interference is defined so as to detect whether a target exists or not and the number of the targets;
and 7: and comparing the number of the detected targets with the real targets to determine the performance of the Cycle-GAN model.
2. The active spoofing interference method of claim 1 wherein generating a sequence of C & I interfering signals by spaced sampling and adjacent padding of the original echo signal sequence comprises:
dividing the original echo signal sequence into m subsections so that the time gap number of each subsection is n;
dividing each sub-segment into smaller small sub-segments;
and for each sub-segment, copying a first small sub-segment in the sub-segment, and filling the sub-segment with the copied structure until the length of the sub-segment is reached to obtain the C & I interference signal sequence.
3. The active spoofing interference method of claim 1 wherein generating a sequence of SMSP interfering signals in the form of spaced samples and adjacent padding of the sequence of original echo signals comprises:
extracting signals from the original echo signal sequence at an interval of n times, and sequencing the extracted signals in the original echo signal sequence;
and copying the sorted and extracted signals until the length of the signals is the same as that of the original echo signal sequence to obtain an SMSP interference signal sequence.
4. The active spoof interference method of claim 1 wherein said step 3 comprises:
inputting the original echo signal sequence into a generator of a Cycle-GAN model, taking the C & I interference signal sequence and the SMSP interference signal sequence as target sequences, generating the original echo signal sequence towards the direction of the interfered signal sequence through continuous training so that the generator continuously generates samples with interference signals, obtaining a judgment score through judging the samples with the interference signals by a judgment device, feeding back the judgment score to the generator so that the generator adjusts parameters through the judgment score, gradually approaching the training samples to the signal sequence added with the target sequences, and ending the whole training process until the judgment device is difficult to distinguish whether the generator generates the signal sequence, thereby obtaining the trained Cycle-GAN model.
5. The active deception jamming method of claim 1, wherein a Cycle-GAN model converts an original echo signal sequence from a source signal sequence domain X to a target signal sequence domain Y, and is recorded as a mapping G, X → Y, so that a signal sequence finally generated in the mapping G and a signal sequence of the target signal sequence domain cannot be identified through a decision device; the Cycle-GAN model performs the task of F (G (X)) ≈ X through the inverse map generator F: Y → X.
6. The active spoof interference method of claim 1 wherein said step 7 comprises:
comparing the number of the detected targets with the real target, and judging whether the difference value between the number of the detected targets and the real target is greater than a threshold value;
if the performance of the Cycle-GAN model is not greater than the threshold value, the Cycle-GAN model is represented to have poor performance.
7. An active spoofed interference system for generating a countering network based on cycle consistency, comprising:
the receiving module is used for receiving an original echo signal sequence which is returned after a radar transmitting signal and does not contain interference;
the interference generation module is used for generating a C & I interference signal sequence and an SMSP interference signal sequence in a mode of interval sampling and adjacent filling for the original echo signal sequence;
the training module is used for inputting the original echo signal sequence into a generator of a Cycle-GAN model, judging the interference signal generated by the generator by taking the signal sequence added with the C & I interference signal sequence and the SMSP interference signal sequence as a judgment standard of a judger, and adjusting internal parameters of the generator in a loss function reduction mode until the trained Cycle-GAN model is obtained through judgment of the judger;
the acquisition module is used for setting a plurality of real targets and transmitting signals in a real-time manner to acquire real-time echo signals fed back by the real targets;
the signal generating module is used for inputting the real-time echo signal into a trained Cycle-GAN model so as to output the echo signal carrying the interference signal;
the detection module is used for delimiting a wave gate of an echo signal carrying interference according to the amplitude of the signal so as to detect whether a target exists or not and the number of the targets;
and the performance determining module is used for comparing the number of the detected targets with the real targets so as to determine the performance of the Cycle-GAN model.
8. The active spoofing interference system of claim 7, wherein the interference generating module is specifically configured to:
dividing the original echo signal sequence into m subsegments so as to enable the time gap number of each subsegment to be n;
dividing each sub-segment into smaller sub-segments;
and for each sub-section, copying a first small sub-section in the sub-section, and filling the sub-section with the copied structure until the sub-section length is reached to obtain a C & I interference signal sequence.
9. The active spoofing interference system of claim 7, wherein the interference generating module is specifically configured to:
carrying out up-sampling on the original echo signal sequence in a mode of n times of original sampling to obtain an up-sampled echo signal sequence;
extracting signals in the sampled echo signal sequence in a mode of n times of interval, and sequencing the extracted signals in the original echo signal sequence;
and copying the sorted and extracted signals until the length of the sorted and extracted signals is the same as that of the original echo signal sequence to obtain an MSP interference signal sequence.
10. The active spoofing interference system of claim 7, wherein the training module is specifically configured to:
inputting the original echo signal sequence into a generator of a Cycle-GAN model, taking the C & I interference signal sequence and the SMSP interference signal sequence as target sequences, generating the original echo signal sequence towards the direction of the interfered signal sequence through continuous training so that the generator continuously generates samples with interference signals, obtaining a judgment score through judging the samples with the interference signals by a judgment device, feeding back the judgment score to the generator so that the generator adjusts parameters through the judgment score, gradually approaching the training samples to the signal sequence added with the target sequences, and ending the whole training process until the judgment device is difficult to distinguish whether the generator generates the signal sequence, thereby obtaining the trained Cycle-GAN model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210655328.1A CN115201768A (en) | 2022-06-10 | 2022-06-10 | Active deception jamming method for generating countermeasure network based on cycle consistency |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210655328.1A CN115201768A (en) | 2022-06-10 | 2022-06-10 | Active deception jamming method for generating countermeasure network based on cycle consistency |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115201768A true CN115201768A (en) | 2022-10-18 |
Family
ID=83575965
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210655328.1A Pending CN115201768A (en) | 2022-06-10 | 2022-06-10 | Active deception jamming method for generating countermeasure network based on cycle consistency |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115201768A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116400311A (en) * | 2023-06-07 | 2023-07-07 | 清华大学 | Radar interference simulation method and device based on generation countermeasure network and electronic equipment |
| CN116930884A (en) * | 2023-09-15 | 2023-10-24 | 西安电子科技大学 | SAR deception jamming template generation and jamming method based on optical SAR image conversion |
-
2022
- 2022-06-10 CN CN202210655328.1A patent/CN115201768A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116400311A (en) * | 2023-06-07 | 2023-07-07 | 清华大学 | Radar interference simulation method and device based on generation countermeasure network and electronic equipment |
| CN116400311B (en) * | 2023-06-07 | 2023-09-19 | 清华大学 | Radar interference simulation method and device based on generation countermeasure network and electronic equipment |
| CN116930884A (en) * | 2023-09-15 | 2023-10-24 | 西安电子科技大学 | SAR deception jamming template generation and jamming method based on optical SAR image conversion |
| CN116930884B (en) * | 2023-09-15 | 2023-12-26 | 西安电子科技大学 | SAR deception jamming template generation and jamming method based on optical SAR image conversion |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115201768A (en) | Active deception jamming method for generating countermeasure network based on cycle consistency | |
| CN107272003B (en) | Active positioning method based on reliable acoustic path and target multi-path echo | |
| CN106405518A (en) | Complex system radar signal grade correlating, clustering and sorting method | |
| CN107346017B (en) | The intensive target simulation method of pulse compression radar based on frequency matching filtering | |
| CN104316914B (en) | Radar target self-adaptation detection method depending on shape parameters | |
| CN102879766A (en) | Method and apparatus for detecting and tracking faint target of high frequency ground wave radar | |
| CN112799023A (en) | Multi-false-target interference method for fast forwarding | |
| CN105137396A (en) | Detection method for SMSP interference and C&I interference | |
| CN112394353B (en) | Sea wave number spectrum reconstruction method based on steep function appraisal | |
| CN114117912A (en) | Sea clutter modeling and inhibiting method under data model dual drive | |
| CN104215944A (en) | Radar sea clutter adaptive inhibition processing method based on parameter selection | |
| CN112255607B (en) | Sea clutter suppression method | |
| Peng et al. | JLHS: A joint linear frequency modulation and hyperbolic frequency modulation approach for speed measurement | |
| CN115951315B (en) | Radar spoofing interference identification method and system based on improved wavelet packet energy spectrum | |
| Qian et al. | Recognition of radar deception jamming based on convolutional neural network | |
| CN110673118A (en) | Active sonar single-frequency pulse train waveform design and detection algorithm | |
| CN112649791B (en) | Radar echo processing method and device | |
| Tian et al. | Multiresolution jamming recognition with few-shot learning | |
| Chen et al. | A ground-penetrating radar object detection method based on deep learning | |
| Wang et al. | Study on an anti-reverberation method based on PCI-SVM | |
| CN113203991A (en) | Anti-deception jamming method of multi-base SAR (synthetic aperture radar) in multi-jammer environment | |
| Schoenecker et al. | Maximum likelihood probabilistic data association tracker applied to bistatic sonar data sets | |
| Tilley et al. | Combining Ground Penetrating Radar Scans of Differeing Frequencies Through Signal Processing. | |
| Qingxin et al. | Research on chaotic character of ship-radiated noise based on phase space reconstruction | |
| Haoquan et al. | Low-Doppler CW Target Echo Enhancement Based on Target Echo Characteristics |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |