CN115099289A - System and method for distinguishing identity of radio radiation source based on weak fingerprint characteristics - Google Patents
System and method for distinguishing identity of radio radiation source based on weak fingerprint characteristics Download PDFInfo
- Publication number
- CN115099289A CN115099289A CN202211020314.9A CN202211020314A CN115099289A CN 115099289 A CN115099289 A CN 115099289A CN 202211020314 A CN202211020314 A CN 202211020314A CN 115099289 A CN115099289 A CN 115099289A
- Authority
- CN
- China
- Prior art keywords
- signal
- frequency
- radiation source
- radio
- frequency hopping
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
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
- G01S3/00—Direction-finders for determining the direction from which infrasonic, sonic, ultrasonic, or electromagnetic waves, or particle emission, not having a directional significance, are being received
- G01S3/02—Direction-finders for determining the direction from which infrasonic, sonic, ultrasonic, or electromagnetic waves, or particle emission, not having a directional significance, are being received using radio waves
- G01S3/14—Systems for determining direction or deviation from predetermined direction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B1/00—Details of transmission systems, not covered by a single one of groups H04B3/00 - H04B13/00; Details of transmission systems not characterised by the medium used for transmission
- H04B1/69—Spread spectrum techniques
- H04B1/713—Spread spectrum techniques using frequency hopping
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Remote Sensing (AREA)
- Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Radar, Positioning & Navigation (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Position Fixing By Use Of Radio Waves (AREA)
- Mobile Radio Communication Systems (AREA)
Abstract
The invention belongs to the field of radio positioning and radio navigation, and discloses a system and a method for distinguishing the identity of a radio radiation source based on weak fingerprint characteristics, wherein the identity of the radiation source of a fixed-frequency signal is distinguished through an identity distinguishing model; receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology; in the aspect of identification judgment of frequency hopping signals, firstly, complex frequency hopping signal data are converted into a data form similar to a fixed frequency signal, and then a method for judging the fixed frequency signal is adopted to judge radiation sources of different frequency hopping signals. The invention breaks through the traditional technical thought of artificial feature extraction and machine learning classification, overcomes the defects of excessive dependence on radiation source prior information, weak generalization capability and the like in the traditional technology by using a deep artificial neural network technology with supervision-learning capability, innovatively designs a deep learning special model for radiation source identity judgment, and can judge the radiation source equipment of fixed-frequency signals and frequency-hopping signals.
Description
Technical Field
The invention belongs to the field of radio positioning and radio navigation, and particularly relates to a system and a method for distinguishing identities of radio radiation sources based on weak fingerprint characteristics.
Background
At present, in the aspect of radiation source identity discrimination based on spectrum 'fingerprint', researchers have conducted a great deal of theoretical exploration in recent years, and have undergone the development of multiple stages of deep learning technology classification from transient signal detection, before the deep learning technology is introduced, artificial feature extraction technology is mainly used, and the technology can only identify specific equipment operating under a predefined wireless protocol. For example, the national defense science and technology university constructs a scheme for distinguishing the identity of a radiation source by using singular values and singular vectors of a time frequency spectrum in 2018, and constructs a scheme for distinguishing the identity by using the non-gaussian property of a radio frequency fingerprint in 2019; the southeast university proposes a scheme for distinguishing the identity of a low signal-to-noise ratio spread spectrum system by using information data 'estimation superposition' in 2018. The university of combined fertilizer industry in 2017 provides a scheme for judging the identity of a radiation source by integrating methods such as time sequence similarity, dynamic time warping and spectral clustering; in the earlier 2016, the university of Beijing traffic used emitter power amplifier nonlinearity to distinguish between three different radio signal sources, etc.
However, the research results have conducted meaningful exploration and research on radiation source detection, and the research is only established at a single theoretical exploration level of a small quantity of radiation source detection, so that the results have no universal and practical significance.
Through the above analysis, the problems and defects of the prior art are as follows:
the existing radio radiation source identity distinguishing mode has no universality and practicability.
The traditional technology excessively depends on radiation source prior information and has weak generalization capability.
Disclosure of Invention
Aiming at the problems in the prior art, the invention provides a system and a method for distinguishing the identity of a radio radiation source based on weak fingerprint characteristics.
The invention is realized in such a way, a method for distinguishing the identity of a radio radiation source based on weak fingerprint characteristics comprises the following steps:
carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model; receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology; in the aspect of identification judgment of frequency hopping signals, signals emitted by a plurality of radiation sources are simultaneously in the whole frequency hopping frequency band; judging the frequency band belonging to the signal of interest in the whole frequency band by using a radio frequency signal identity judging technology, and separating the signal of the target radiation source from other signals; and directly extracting the time domain signal corresponding to the judged frequency band through filtering, and detecting the radiation source direction and position of the signal of interest through a radio direction finding technology.
Further, the method further comprises:
collecting frequency spectrum data in the whole frequency band of a frequency hopping signal through a radio frequency spectrum supervision array and a radio receiver; dividing the frequency spectrums of all the frequency hopping signals according to the widths of all the frequency hopping signal frequency bands, the frequency bandwidth of a single frequency hopping signal and the central frequency of the single frequency hopping signal, and taking the frequency spectrums as network input; the method comprises the steps of identifying the difference between the frequency spectrum data of a signal of interest and the frequency spectrum data of other signals through a deep artificial neural network technology with supervision-learning capacity, extracting the weak signal of interest mixed in a frequency hopping frequency band, and converting a complex frequency hopping signal into a data form similar to a fixed frequency signal.
Further, the method specifically comprises:
step one, input data are classified and distinguished on different observation domains and data scales through a convolutional neural network and a cyclic neural network, and a final judgment result is given through weighted superposition neural network analysis of each domain;
secondly, by utilizing the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are judged on a finer scale by a judgment network of the next level through an attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of the radiation source is ensured to be accurately judged;
if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum monitoring array and a radio signal direction finding technology;
and step four, converting the captured ground frequency hopping signal data into a fixed frequency signal mode through sorting and data splicing at the rear end of signal processing, and judging the identity of the radiation source by utilizing the step one and the step two.
Further, in step three, the receiving devices in the radio frequency spectrum supervision array extract a smaller frequency range for monitoring within the working frequency range of the frequency hopping signal.
Further, the third step further comprises:
under the assistance of a radio signal direction finding technology, the strength of the received signal of the monitored radiation source is improved by adjusting various means of the direction of the receiving antenna, and other signals with weaker strength are filtered; when a sufficiently strong signal is measured by the receiving devices in the array, it is recognized as the signal to be captured and the captured signal data is forwarded to the signal processing back-end.
Another object of the present invention is to provide a weak fingerprint feature-based radio radiation source identification system, which includes:
the fixed-frequency signal module is used for carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model;
the frequency hopping signal acquisition module is used for receiving frequency hopping signals by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and the frequency hopping signal conversion module is used for converting the received ground frequency hopping signal data into a mode similar to a fixed frequency signal.
Further, the identity discrimination model includes:
the convolutional neural networks are used for extracting more complex local micro-features in the window signal;
the cyclic neural network is used for extracting the large-span dependence features in the long sequence signals;
the channel attention network is used for preventing the characteristics related to the radiation source fingerprint from being submerged in other characteristics extracted by the judgment network, and further reducing the misjudgment risk of the deep network;
and the attention network is used for judging the hot spot regions in the input data extracted by each level of network on a finer scale by using the characteristics extracted by each level of neural network, and simultaneously enabling the attention network to track a plurality of interested regions so as to ensure the accurate judgment of the identity of the radiation source.
Further, the radio-spectrum surveillance array is comprised of a number of radio-frequency antennas.
It is a further object of the invention to provide a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
and carrying out grading judgment on the input data on different observation domains and data scales through the convolutional neural network and the cyclic neural network respectively, and analyzing through the weighted superposition neural network of each domain to give a final judgment result.
By utilizing the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are judged by a judgment network of the next level on a finer scale through the attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of the radiation source is ensured to be accurately judged;
if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and through sorting and data splicing at the rear end of signal processing, converting captured ground frequency hopping signal data into a fixed frequency signal mode, and judging the identity of a radiation source.
It is another object of the present invention to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
and carrying out grading judgment on the input data on different observation domains and data scales through the convolutional neural network and the cyclic neural network respectively, and analyzing through the weighted superposition neural network of each domain to give a final judgment result.
By utilizing the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are used for judging by a judging network of the next level on a finer scale through the attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of the radiation source is ensured to be accurately judged;
if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and through sorting and data splicing at the rear end of signal processing, converting captured ground frequency hopping signal data into a fixed frequency signal mode, and judging the identity of a radiation source.
Another object of the present invention is to provide a radiation source detecting terminal based on the above distinguishing system, the radiation source detecting terminal includes:
the fixed-frequency signal module is used for carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model;
the frequency hopping signal acquisition module is used for receiving the frequency hopping signal by adopting a combined radio frequency spectrum monitoring array and a radio signal direction finding technology;
and the frequency hopping signal conversion module is used for converting the received ground frequency hopping signal data into a mode similar to a fixed frequency signal.
By combining all the technical schemes, the invention has the advantages and positive effects that:
the invention breaks through the traditional technical thought of 'artificial feature extraction + machine learning classification', applies the deep artificial neural network technology with the 'supervision-learning' capability, overcomes the defects that the traditional technology excessively depends on the prior information of a radiation source, has weak generalization capability and the like, innovatively designs a 'deep learning' special model for radiation source identity judgment, and can judge the radiation source equipment of fixed-frequency signals and frequency-hopping signals.
The invention breaks through the traditional technical thought of ' artificial feature extraction + machine learning classification ', extracts a plurality of frequency hopping signals mixed in a frequency hopping frequency band by using a deep artificial neural network technology with the capability of supervision-learning ', converts the frequency hopping signals into a plurality of signals with fixed frequency, calculates the Doppler frequency shift of a radiation source by using a radio direction finding technology, and finally identifies the identity of the radiation source by using a deep learning network; the method overcomes the defects that a plurality of frequency hopping signals cannot be recognized in the traditional method, are easily interfered by different signals and rely on prior identity information, innovatively designs the technical idea of frequency hopping signal separation, signal interference removal and radiation source identity recognition and a corresponding deep learning network algorithm, and can be used for distinguishing the identities of radiation source equipment of the frequency hopping signals and the fixed frequency signals.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the embodiments of the present application will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present application, and it is obvious for those skilled in the art that other drawings can be obtained from the drawings without creative efforts.
Fig. 1 is a spectral waterfall plot of two different radiation sources provided by an embodiment of the present invention.
Fig. 2 is a schematic diagram of an identity discrimination model structure of a fixed-frequency signal radiation source according to an embodiment of the present invention.
Fig. 3 is a schematic diagram of a dedicated acquisition device for frequency hopping signals according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of receiving a frequency hopping signal by using a combination radio spectrum supervision array and radio signal direction finding technology in frequency hopping signal acquisition according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
Aiming at the problems in the prior art, the invention provides a system for distinguishing the identity of a radio radiation source based on weak fingerprint characteristics, which comprises the following steps:
the fixed-frequency signal module is used for carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model;
the frequency hopping signal acquisition module is used for receiving frequency hopping signals by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and the frequency hopping signal conversion module is used for converting the received ground frequency hopping signal data into a mode similar to a fixed frequency signal.
Further, the identity discrimination model includes:
the convolutional neural networks are used for extracting more complex local micro-features in the window signal;
the cyclic neural network is used for extracting the large-span dependence features in the long sequence signals;
the channel attention network is used for preventing the characteristics related to the radiation source fingerprint from being submerged in other massive characteristics extracted by the judgment network, and further reducing the misjudgment risk of the deep network;
and the attention network is used for judging the hot spot region in the input data extracted by each level of network on a finer scale by using the characteristics extracted by each level of neural network, and simultaneously, the attention network is enabled to track a plurality of interested regions, so that the identity of the radiation source is ensured to be accurately judged.
Further, the radio spectrum supervision array is composed of a plurality of radio signal receivers and a polyphase filter receiver.
Another object of the present invention is to provide a method for identifying a radio radiation source based on weak fingerprint features, which includes:
step one, input data are classified and distinguished on different observation domains and data scales through a convolutional neural network and a cyclic neural network, and a final judgment result is given through weighted superposition neural network analysis of each domain;
secondly, by utilizing the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are judged on a finer scale by a judgment network of the next level through an attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of the radiation source is ensured to be accurately judged;
if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum monitoring array and a radio signal direction finding technology;
and step four, converting the captured ground frequency hopping signal data into a fixed frequency signal mode through sorting and data splicing at the rear end of signal processing, and judging the identity of the radiation source by utilizing the step one and the step two.
Further, in step three, the receiving devices in the radio frequency spectrum supervision array extract a smaller frequency range for monitoring within the working frequency range of the frequency hopping signal. With the aid of radio signal direction finding technology, the strength of the received signal of the monitored radiation source is improved as much as possible by adjusting the direction of the receiving antenna and other means, and other signals with weaker strength are filtered. Under this condition, when a sufficiently strong signal is detected by the receiving devices in the array, it is recognized as the signal to be captured and the captured signal data is forwarded to the signal processing back end.
It is a further object of the invention to provide a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of:
and carrying out grading discrimination on input data on different observation domains and data scales through a convolutional neural network and a cyclic neural network respectively, and analyzing through weighted superposition neural networks of all domains to give a final judgment result.
By utilizing the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are judged by a judgment network of the next level on a finer scale through the attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of the radiation source is ensured to be accurately judged;
if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and through sorting and data splicing at the rear end of signal processing, converting captured ground frequency hopping signal data into a fixed frequency signal mode, and judging the identity of a radiation source.
It is another object of the present invention to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of:
and carrying out grading discrimination on input data on different observation domains and data scales through a convolutional neural network and a cyclic neural network respectively, and analyzing through weighted superposition neural networks of all domains to give a final judgment result.
By utilizing the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are judged by a judgment network of the next level on a finer scale through the attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of the radiation source is ensured to be accurately judged;
if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and through sorting and data splicing at the rear end of signal processing, converting captured ground frequency hopping signal data into a fixed frequency signal mode, and judging the identity of a radiation source.
The present invention will be described in detail below with reference to the accompanying drawings.
As shown in fig. 1, a method for identifying an identity of a radio radiation source based on weak fingerprint features according to an embodiment of the present invention includes:
s101, input data are judged in a grading mode on different observation domains and data scales through a convolutional neural network and a cyclic neural network, and a final judgment result is given through weighted superposition of all domains and neural network analysis.
S102, by using the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are judged on a finer scale by a judgment network of the next level through an attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of a radiation source is ensured to be accurately judged;
s103, if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum monitoring array and a radio signal direction finding technology;
s104, converting the captured ground frequency hopping signal data into a fixed frequency signal mode through sorting and data splicing at the rear end of signal processing, and judging the identity of the radiation source by utilizing S101 and S102.
The radiation source for a wireless communication device signal consists of thousands of electronic components. These devices can generally be considered to be non-ideal devices. The non-ideality of the device may cause a small distortion of the electromagnetic signal with regularity when the radiation source emits the electromagnetic signal. These minor distortions are the radio frequency "fingerprint" of the radiation source. The project group transmits a steady-state frequency spectrum waterfall diagram of a carrier wave at the same frequency point by two shackRF wireless signal radiation sources of the same manufacturer, the same specification and the same model, other spectral lines are arranged on two sides of the two waterfall diagrams, the power, the position and the number of the spectral lines are different, and the steady-state frequency spectrum waterfall diagram is shown in fig. 2.
The enormous number of radio signal radiation source devices and the diversity of device imperfections result in a radiation source whose spectral fingerprint is as unique as a human fingerprint and difficult to counterfeit. This results in a more difficult separation of the "fingerprints" of the radio signals, since the non-ideal response of the electronic components is usually weak.
Meanwhile, under the influence of channel noise and other interference factors, the radiation source fingerprint can also generate phenomena such as distortion, noise submergence and the like. In addition to the time-varying characteristics of the radio signal, it is difficult to directly indicate the existence rule and mode of the fingerprint of a specific radiation source in the radio signal a priori. Therefore, the identity of the radiation source cannot be judged by establishing an expert database for the radiation source fingerprint.
The deep learning neural network is a more effective method for solving the problem, and is commonly used for the image classification task. Similar to image recognition, the spectrum fingerprint of the radiation source is similar to the screening and extraction of the characteristics of the 'target of interest' in the image, and has regularity and certain randomness. The samples of the captured radio signal are highly complex and random, as are the "background" in the image. Therefore, the distribution characteristics of the radiation source fingerprint contained in the time-varying wireless signal are automatically learned by using a special mathematical model which is autonomously developed and verified in recent years by a subject group, and the identity of the radiation source is effectively judged.
Due to the large difference between the radio signal data and the image data. The proportion of the spectrum fingerprint data to the data volume of the radio signal is far lower than the proportion of the identification target to the background in the image identification task. A "dedicated mathematical model" for radiation source identification must be designed and developed, as shown in fig. 3.
The fixed frequency signal is: a radio signal whose center frequency does not vary with time during communication.
The model is composed of a plurality of convolutional neural networks and cyclic neural networks. The convolutional neural network is used for extracting more complex local micro-features in the window signal. The recurrent neural network is used for extracting the large-span dependent features in the long-sequence signal. The networks respectively carry out grading discrimination on input data in different observation domains and data scales, and final judgment results are given through weighted superposition neural network analysis of each domain. In the aspect of solving the problem that the proportion of the fingerprint data volume to the input data volume is too small, the model introduces a multi-stage spatial attention mechanism. By utilizing the characteristics extracted by each level of discrimination model, the attention network can judge the hot spot region in the input data extracted by each level for the discrimination network of the next level on a finer scale, and simultaneously, the attention network can track a plurality of interested regions, thereby ensuring the accurate discrimination of the identity of the radiation source. The model is also added with a channel attention network before the space attention network so as to avoid that the characteristics related to the radiation source fingerprint are submerged in other characteristics extracted by the discrimination network, thereby further reducing the misjudgment risk of the depth network.
In the aspect of identity discrimination of frequency hopping signals, the adopted thinking is as follows: in the aspect of identification judgment of frequency hopping signals, signals emitted by a plurality of radiation sources are simultaneously in the whole frequency hopping frequency band, and for the signals emitted by a single radiation source, the signals are continuous in time and continuously changed in frequency. By using a radio frequency signal identification technology, the frequency band belonging to a specific radiation source signal in the whole frequency band is judged, and the signal of the target radiation source is separated from other signals. Because the frequency hopping signals are continuous in the time domain, the time domain signals corresponding to the judged frequency bands are directly extracted through filtering, the complex frequency hopping signals are converted into data forms similar to the fixed frequency signals, and then the radiation sources of different frequency hopping signals are judged by adopting a method for judging the fixed frequency signals.
Frequency hopping refers to radio signals that use a frequency hopping technique, which is a method of spreading a frequency spectrum by using a pseudo-random code sequence to perform frequency shift keying so that a carrier frequency continuously hops within a wide operating frequency range. Compared with a fixed-frequency signal, the frequency hopping signal has stronger anti-interference capability and higher safety, but on the premise that a hopping random code sequence (a hopping sequence) is not known, the frequency hopping signal is extremely difficult to capture, so that the problem of capturing the frequency hopping signal needs to be solved before identity discrimination.
In the frequency hopping signal acquisition, the frequency hopping signal is received by adopting a combined radio frequency spectrum supervision array and radio signal direction finding technology, as shown in fig. 4.
The radio spectrum supervision array consists of a plurality of radio signal receivers and polyphase filter receivers. The receiving devices in the array extract a smaller frequency range for monitoring within the operating frequency range of the frequency hopping signal. Under the assistance of the direction finding technology, the strength of the received signal of the monitored radiation source is improved as much as possible by adjusting the direction of the receiving antenna and other means, and other signals with weak strength are filtered. Under this condition, when a sufficiently strong signal is detected by the receiving devices in the array, it is recognized as the signal to be captured and the captured signal data is forwarded to the signal processing back end.
After sorting and data splicing at the signal processing rear end, captured ground frequency hopping signal data is converted into a fixed frequency signal mode. And judging the identity of the frequency hopping signal by utilizing a neural network discrimination mode of the fixed frequency signal.
The invention provides a radio radiation source identity discrimination system based on weak fingerprint characteristics, which comprises:
the fixed-frequency signal module is used for carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model;
the frequency hopping signal acquisition module is used for receiving frequency hopping signals by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and the frequency hopping signal conversion module is used for converting the received ground frequency hopping signal data into a mode similar to a fixed frequency signal.
The identity discrimination model comprises:
the convolutional neural networks are used for extracting more complex local micro-features in the window signals;
the cyclic neural network is used for extracting large-span dependence features in the long sequence signals;
the channel attention network is used for preventing the characteristics related to the radiation source fingerprint from being submerged in other mass characteristics extracted by the judgment network, and further reducing the misjudgment risk of the deep network;
and the attention network is used for judging the hot spot region in the input data extracted by each level of network on a finer scale by using the characteristics extracted by each level of neural network, and simultaneously, the attention network is enabled to track a plurality of interested regions, so that the identity of the radiation source is ensured to be accurately judged.
The radio frequency spectrum supervision array is composed of a plurality of radio signal receivers and polyphase filtering receivers.
The invention is further described with reference to specific examples.
Example (b):
in the prior experiments of the identification discrimination of 13 civil frequency modulation broadcasts in a certain area, the discrimination accuracy rate of 97.5 percent is obtained. For the discrimination experiment of the civil hand-held interphone (the fixed frequency signal radiation source with the transmitting antenna moving in a small range), 85.7 percent of discrimination accuracy is obtained. Experiments prove that the scheme adopted by the invention can effectively judge the wireless signal radiation source.
Radiation source identity discrimination indexes:
(1) judging whether the number of radio stations is less than or equal to 20 by using a non-mobile fixed frequency signal, wherein the accuracy (top-5 requirement) is more than or equal to 90 percent;
(2) the number of the radio stations is judged to be less than or equal to 20 by non-mobile frequency hopping signals, and the judgment accuracy rate of intercepting 1 frequency hopping signal is more than or equal to 2/N (random selection, N is the number of equipment).
The above description is only for the purpose of illustrating the embodiments of the present invention, and the scope of the present invention should not be limited thereto, and any modifications, equivalents and improvements made by those skilled in the art within the technical scope of the present invention as disclosed in the present invention should be covered by the scope of the present invention.
Claims (8)
1. A weak fingerprint feature-based radio radiation source identity discrimination method is characterized by comprising the following steps:
carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model; receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology; in the aspect of identification judgment of frequency hopping signals, signals emitted by a plurality of radiation sources are simultaneously in the whole frequency hopping frequency band; judging the frequency band belonging to the signal of interest in the whole frequency band by using a radio frequency signal identity judging technology, and separating the signal of the target radiation source from other signals; and directly extracting the time domain signal corresponding to the judged frequency band through filtering, and detecting the radiation source direction and position of the signal of interest through a radio direction finding technology.
2. The weak fingerprint feature based radio radiation source identification method according to claim 1, wherein said method further comprises:
collecting frequency spectrum data in the whole frequency band of a frequency hopping signal through a radio frequency spectrum supervision array and a radio receiver; dividing the frequency spectrums of all the frequency hopping signals according to the widths of all the frequency hopping signal frequency bands, the frequency bandwidth of a single frequency hopping signal and the central frequency of the single frequency hopping signal, and taking the frequency spectrums as network input; the method comprises the steps of identifying the difference between the frequency spectrum data of a signal of interest and the frequency spectrum data of other signals through a deep artificial neural network technology with supervision-learning capability, extracting the weak signal of interest mixed in a frequency hopping frequency band, and converting a complex frequency hopping signal into a data form similar to a fixed frequency signal.
3. The weak fingerprint feature-based radio radiation source identity discrimination method according to claim 1, wherein the method specifically comprises:
step one, input data are classified and distinguished on different observation domains and data scales through a convolutional neural network and a cyclic neural network, and a final judgment result is given through weighted superposition neural network analysis of each domain;
secondly, by utilizing the characteristics extracted by each level of neural network, hot spot areas in the input data extracted by each level are judged on a finer scale by a judgment network of the next level through an attention network, and meanwhile, the attention network is enabled to track a plurality of interested areas, so that the identity of the radiation source is ensured to be accurately judged;
if the radio signal is a frequency hopping signal, receiving the frequency hopping signal by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and step four, converting the captured ground frequency hopping signal data into a fixed frequency signal mode through sorting and data splicing at the rear end of signal processing, and judging the identity of the radiation source by utilizing the step one and the step two.
4. The weak fingerprint feature based radio radiation source identification method according to claim 3, wherein in step three, the receiving devices in the radio frequency spectrum surveillance array extract a smaller frequency range for monitoring within the working frequency range of the frequency hopping signal;
the third step further comprises:
under the assistance of a radio signal direction finding technology, the strength of the received signal of the monitored radiation source is improved by adjusting various means of the direction of the receiving antenna, and other signals with weaker strength are filtered; when a sufficiently strong signal is measured by the receiving devices in the array, it is recognized as the signal to be captured and the captured signal data is forwarded to the signal processing back-end.
5. A weak fingerprint feature-based radio radiation source identity discrimination system is characterized by comprising:
the fixed-frequency signal module is used for carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model;
the frequency hopping signal acquisition module is used for receiving frequency hopping signals by adopting a combined radio frequency spectrum supervision array and a radio signal direction finding technology;
and the frequency hopping signal conversion module is used for converting the received ground frequency hopping signal data into a mode similar to a fixed frequency signal.
6. The weak fingerprint feature based radio radiation source identification system according to claim 5 wherein said identification model comprises:
the convolutional neural networks are used for extracting more complex local micro-features in the window signals;
the cyclic neural network is used for extracting the large-span dependence features in the long sequence signals;
the channel attention network is used for preventing the characteristics related to the radiation source fingerprint from being submerged in other characteristics extracted by the judgment network, and further reducing the misjudgment risk of the deep network;
and the attention network is used for judging the hot spot regions in the input data extracted by each level of network on a finer scale by using the characteristics extracted by each level of neural network, and simultaneously enabling the attention network to track a plurality of interested regions so as to ensure the accurate judgment of the identity of the radiation source.
7. The weak fingerprint based radio radiation source identification system of claim 5 wherein said radio spectrum surveillance array is comprised of a plurality of radio frequency antennas.
8. A radiation source detection terminal based on the discrimination system of any one of claims 5-6, characterized in that the radiation source detection terminal comprises:
the fixed-frequency signal module is used for carrying out identity discrimination on a radiation source of the fixed-frequency signal through an identity discrimination model;
the frequency hopping signal acquisition module is used for receiving the frequency hopping signal by adopting a combined radio frequency spectrum monitoring array and a radio signal direction finding technology;
and the frequency hopping signal conversion module is used for converting the received ground frequency hopping signal data into a mode similar to a fixed frequency signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211020314.9A CN115099289B (en) | 2022-08-24 | 2022-08-24 | System and method for distinguishing identity of radio radiation source based on weak fingerprint characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211020314.9A CN115099289B (en) | 2022-08-24 | 2022-08-24 | System and method for distinguishing identity of radio radiation source based on weak fingerprint characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115099289A true CN115099289A (en) | 2022-09-23 |
| CN115099289B CN115099289B (en) | 2022-11-01 |
Family
ID=83301243
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211020314.9A Active CN115099289B (en) | 2022-08-24 | 2022-08-24 | System and method for distinguishing identity of radio radiation source based on weak fingerprint characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115099289B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109633588A (en) * | 2018-12-29 | 2019-04-16 | 杭州电子科技大学 | Recognition Method of Radar Emitters based on depth convolutional neural networks |
| US20190342775A1 (en) * | 2018-05-07 | 2019-11-07 | Verizon Patent And Licensing Inc. | Systems and methods for identifying and locating distinct interference sources in a wireless network |
| CN111310700A (en) * | 2020-02-27 | 2020-06-19 | 电子科技大学 | Intermediate frequency sampling sequence processing method for radiation source fingerprint feature identification |
| CN112395941A (en) * | 2020-09-28 | 2021-02-23 | 中国人民解放军32802部队 | Radio signal radiation source individual identification method, computer device and storage medium |
| US20220077879A1 (en) * | 2020-09-09 | 2022-03-10 | SparkCognition, Inc. | Machine-learning based analysis, prediction, and response based on electromagnetic waveforms |
| CN114301498A (en) * | 2021-12-30 | 2022-04-08 | 中国电子科技集团公司第五十四研究所 | Behavior identification method and system for frequency hopping communication radiation source |
-
2022
- 2022-08-24 CN CN202211020314.9A patent/CN115099289B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190342775A1 (en) * | 2018-05-07 | 2019-11-07 | Verizon Patent And Licensing Inc. | Systems and methods for identifying and locating distinct interference sources in a wireless network |
| CN109633588A (en) * | 2018-12-29 | 2019-04-16 | 杭州电子科技大学 | Recognition Method of Radar Emitters based on depth convolutional neural networks |
| CN111310700A (en) * | 2020-02-27 | 2020-06-19 | 电子科技大学 | Intermediate frequency sampling sequence processing method for radiation source fingerprint feature identification |
| US20220077879A1 (en) * | 2020-09-09 | 2022-03-10 | SparkCognition, Inc. | Machine-learning based analysis, prediction, and response based on electromagnetic waveforms |
| CN112395941A (en) * | 2020-09-28 | 2021-02-23 | 中国人民解放军32802部队 | Radio signal radiation source individual identification method, computer device and storage medium |
| CN114301498A (en) * | 2021-12-30 | 2022-04-08 | 中国电子科技集团公司第五十四研究所 | Behavior identification method and system for frequency hopping communication radiation source |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115099289B (en) | 2022-11-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ureten et al. | Wireless security through RF fingerprinting | |
| CN107368732B (en) | A kind of object recognition and detection system and method based on equipment physical fingerprint feature | |
| CN107911183B (en) | Unmanned aerial vehicle identification method and system based on real-time signal characteristic analysis | |
| CN105678273A (en) | Initial point detection algorithm of transient signal in radio frequency fingerprint identification technology | |
| CN107290758A (en) | GNSS interference signals multistage identification detecting system and method | |
| CN104869630A (en) | Pseudo base station rapid positioning method and system based on offline fingerprint database | |
| CN110289918A (en) | A kind of wireless signal detection method and system | |
| CN106452716A (en) | Unmanned aerial vehicle signal recognition and detection algorithm based on Hash fingerprint | |
| CN105071830B (en) | A kind of detection recognition method of direct sequence signal | |
| CN106656303B (en) | A kind of signal detecting method based on satellite antenna acquisition and tracking pointing system | |
| CN101436879B (en) | Method for extracting interfering signal amplitude with environmental noise | |
| Ye et al. | A new frequency hopping signal detection of civil UAV based on improved k-means clustering algorithm | |
| CN111948618A (en) | Forward scattering target detection method and system based on satellite external radiation source | |
| Kaplan et al. | Measurement based FHSS–type drone controller detection at 2.4 GHz: An STFT approach | |
| Chen et al. | GNSS interference type recognition with fingerprint spectrum DNN method | |
| CN110764152B (en) | Device and method for rapid detection and identification of unmanned aerial vehicle | |
| CN106230544A (en) | The monitoring identification of a kind of automobile remote-control interference signal and localization method | |
| CN111830321B (en) | Unmanned aerial vehicle detection and identification method based on radio frequency fingerprint | |
| Sardar et al. | Vehicle detection and classification using LTE‐CommSense | |
| CN115099289B (en) | System and method for distinguishing identity of radio radiation source based on weak fingerprint characteristics | |
| CN109085613A (en) | The recognition methods of satellite cheating interference and device based on constellation track | |
| CN113300986A (en) | Unmanned aerial vehicle image transmission signal and hotspot signal identification method, medium and computer equipment | |
| CN114584227B (en) | Automatic burst signal detection method | |
| US20060212929A1 (en) | Method and technique for gathering signal intelligence of all radio communications only originating from specific selected areas | |
| KR102663240B1 (en) | Drone detection and tracking system |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |