CN111399002B - GNSS receiver combined interference classification and identification method based on two-stage neural network - Google Patents

GNSS receiver combined interference classification and identification method based on two-stage neural network Download PDF

Info

Publication number
CN111399002B
CN111399002B CN202010276113.XA CN202010276113A CN111399002B CN 111399002 B CN111399002 B CN 111399002B CN 202010276113 A CN202010276113 A CN 202010276113A CN 111399002 B CN111399002 B CN 111399002B
Authority
CN
China
Prior art keywords
interference
stage
signal
peak
identification module
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.)
Active
Application number
CN202010276113.XA
Other languages
Chinese (zh)
Other versions
CN111399002A (en
Inventor
张国梅
张欣
李国兵
贾小林
马小辉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian Jiaotong University
Original Assignee
Xian Jiaotong University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Xian Jiaotong University filed Critical Xian Jiaotong University
Priority to CN202010276113.XA priority Critical patent/CN111399002B/en
Publication of CN111399002A publication Critical patent/CN111399002A/en
Application granted granted Critical
Publication of CN111399002B publication Critical patent/CN111399002B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

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
    • G01S19/00—Satellite radio beacon positioning systems; Determining position, velocity or attitude using signals transmitted by such systems
    • G01S19/01—Satellite radio beacon positioning systems transmitting time-stamped messages, e.g. GPS [Global Positioning System], GLONASS [Global Orbiting Navigation Satellite System] or GALILEO
    • G01S19/13—Receivers
    • G01S19/21—Interference related issues ; Issues related to cross-correlation, spoofing or other methods of denial of service
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • 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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12—Classification; Matching

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Biomedical Technology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Position Fixing By Use Of Radio Waves (AREA)

Abstract

The invention discloses a GNSS receiver combined interference classification identification method based on a two-stage neural network, wherein a receiver receives navigation signals sent by N visible satellites, adopts a two-stage identification scheme based on a BP neural network according to a received GNSS signal model and an interference source, extracts time domain and frequency domain characteristics from digital intermediate frequency signals after A/D conversion through a first-stage identification module, and sends the time domain and frequency domain characteristics to the BP neural network for suppression type interference detection and classification; if the identification result of the first-stage identification module is non-interference or deception interference, capturing the digital intermediate frequency signal, extracting the relevant peak characteristics by using the captured two-dimensional search matrix, and sending the relevant peak characteristics to the second-stage identification module for deception interference detection; and when the final identification result of the two-stage identification module is non-interference, judging that the received signal is a real satellite signal, and after identifying the interference type, adopting a corresponding interference processing means. The invention can quickly and accurately identify the interference.

Description

GNSS receiver combined interference classification and identification method based on two-stage neural network
Technical Field
The invention belongs to the technical field of signal processing, and particularly relates to a GNSS receiver combined interference classification and identification method based on a two-stage neural network.
Background
The Global Navigation Satellite System (GNSS) is a Navigation System with wide coverage, all weather, real time and high precision. With the continuous development of satellite navigation technology, GNSS has been widely used in various military and civil fields. The GNSS mainly includes a Global Positioning System (GPS) in the united states, GALILEO (GALILEO) in the european union, GLONASS (GLONASS) in russia, and a beidou navigation satellite system (BDS) in China. The safety of the global satellite navigation system is especially important to guarantee due to wide application and large influence range. Because the navigation satellite is generally far away from the earth surface, when the navigation signal transmitted by the satellite is transmitted to a user terminal on the ground, the satellite signal is very weak, the signal format of the civil satellite is open, and the anti-interference capability of the satellite navigation system is limited, so that the GNSS receiver is easily attacked by interference. The jamming and the deceptive jamming are two most typical jamming modes in the GNSS jamming, and are also two jamming modes which are common in the part special for the user end of the navigation system. The compressive interference is a high-power strong interference signal, and the receiver cannot normally receive and lock a satellite navigation signal so that the satellite navigation signal cannot be positioned. The deceptive jamming is that a jammer retransmits or generates a signal which is the same as or similar to a navigation satellite signal, so that a receiver of a target user of a satellite navigation system mistakenly recognizes the deceptive signal as a real navigation satellite signal, thereby capturing and tracking the deceptive signal and calculating the wrong position, speed or time. The research on the anti-interference technology of the GNSS receiver has important significance for improving the working performance of the satellite navigation system in a complex electromagnetic environment and enhancing the reliability of the satellite navigation system in various environments.
Interference identification is an important link of anti-interference, and therefore, the method becomes a research hotspot in the field of GNSS anti-interference. Most of the current technologies for identifying relevant interference are based on specific systems and specific interference types, and the universality is poor. And the used related algorithms such as decision trees, clustering algorithms, neural networks and the like are only used for judging whether the interference exists or not, and classifying and detecting the pure compression type interference or the pure deception type interference. In an actual combat environment, the interference source generally performs suppression type interference for a certain time, so that the target GNSS receiver is switched into a search state, and then sends deception interference to lock the interfered GNSS receiver on a deception signal. Therefore, the medium and deceptive interferences alternate in the same scene and may switch randomly. In order to enhance the anti-interference capability of the receiver, a uniform scheme capable of automatically classifying and identifying the compression type interference and the deceptive interference is necessary to be designed.
Disclosure of Invention
The technical problem to be solved by the present invention is to provide a GNSS receiver combined interference classification and identification method based on two-stage neural networks, which all use a three-layer fully-connected neural network to implement classification decision-making.
The invention adopts the following technical scheme:
a GNSS receiver combined interference classification identification method based on a two-stage neural network is characterized in that a receiver receives navigation signals sent by N visible satellites, a two-stage identification scheme based on a BP neural network is adopted according to a received GNSS signal model and an interference source, time domain and frequency domain characteristics are extracted from digital intermediate frequency signals after A/D conversion through a first-stage identification module, and the digital intermediate frequency signals are sent to the BP neural network for suppression type interference detection and classification; if the identification result of the first-stage identification module is non-interference or deception interference, capturing the digital intermediate frequency signal, extracting the relevant peak characteristics by using the captured two-dimensional search matrix, and sending the relevant peak characteristics to the second-stage identification module for deception interference detection; and when the final identification result of the two-stage identification module is non-interference, judging that the received signal is a real satellite signal, and after identifying the interference type, adopting a corresponding interference processing means.
Specifically, the BP neural network of the first-stage identification module is trained by utilizing the ratio of the maximum value and the second maximum value of the signal spectrum amplitude, the single-frequency energy aggregation degree, the average spectrum flatness coefficient, the time domain kurtosis, the power spectrum skewness, the ratio of the power spectrum kurtosis spectrum variance to the mean square, the normalized spectrum peak-to-average ratio and the 3dB bandwidth of the normalized spectrum, and the output labels are divided into 8 types.
Further, the ratio of the maximum value to the next maximum value of the signal spectrum amplitude is:
x1＝|X(k)|1stmax/|X(k)|2ndmax
the single frequency energy concentration is:
Figure BDA0002444848140000031
the average spectral flatness coefficient is:
Figure BDA0002444848140000032
the time domain kurtosis is:
x4＝E(|x(n)-μt|4)/σt 4
the power spectrum skewness is:
x5＝E[X(ω)-μP]3/σP 3
the power spectrum kurtosis is:
x6＝E[X(ω)-μP]4/σP 4
the ratio of the spectral variance to the mean squared is:
x7＝σf 2/μf 2
the normalized spectral peak-to-average ratio is:
x8＝max{Xu(k)}/E[Xu(k)]
the 3dB bandwidth of the normalized spectrum is:
x9＝card{k|X′u(k)＞V3dB}/card{k|X′u(k)}
wherein, Pp(k) Denotes the result of partial extraction of the impulse in P (k), μtIs the mean of x (n), σ is the standard deviation of x (n); mu.sPIs the mean value of X (ω), σPIs the standard deviation of X (ω); card { } denotes the number of collection elements, Xu' (k) is X (k) the result of normalization with the mean value, V3dB＝0.707max[X′u(k)]。
Specifically, the global accumulation amount of the correlation value, the local accumulation amount of the correlation value, the peak-to-peak value of the correlation value, and A are usedfNumber of correlation peaks in, AcNumber of correlation peaks in, AfCorrelation peak width of, AcCorrelation peak width of, AfThe cumulative amount of the symmetrical difference of the correlation peaks, AcThe cumulative amount of the symmetric difference of the correlation peaks, AfDifference in slope and AcTraining BP of second-stage recognition module by 11 features of slope differenceNeural networks, output labels fall into class 2.
Further, the correlation value global accumulation amount is:
Figure BDA0002444848140000041
the local cumulative amount of correlation values is:
Figure BDA0002444848140000042
the correlation peak-to-peak value is:
x13＝max{ai,j|ai,j∈A}
AfThe number of correlation peaks in (a) is:
x14＝card{i|pAf(i)＞VT}
Acthe number of correlation peaks in (a) is:
x15＝card{j|pAc(j)＞VT}
Afthe correlation peak width of (a) is:
x16＝card{Af|Af＞VT}
Acthe correlation peak width of (a) is:
x17＝card{Ac|Ac＞VT}
Afthe cumulative amount of the symmetrical difference of the correlation peaks is as follows:
Figure BDA0002444848140000051
Acthe cumulative amount of the symmetrical difference of the correlation peaks is as follows:
Figure BDA0002444848140000052
Afthe slope difference is:
x20＝Af(ip+0.5/Δfd)-Af(ip-0.5/Δfd)
Acthe slope difference is:
x21＝Ac(jp+0.5fs/Rc)-Ac(jp-0.5fs/Rc)
wherein, ai,jIs the i, j th element, V, of the matrix ATIs the acquisition threshold of the receiver and,
Figure BDA0002444848140000053
is represented by AfA set of all peak-to-peak values;
Figure BDA0002444848140000054
is represented by AcA set of all peak-to-peak values; b isfIs AfThe result after the translation and the amplitude limiting; i'PIs the maximum correlation peak at BfCoordinates of (5), BcIs AcThe result after the translation and the amplitude limiting; j'PIs the maximum correlation peak at BcCoordinate of (1), ipIs the coordinate of the maximum peak on the Doppler shift axis, Δ fdSearch step size for Doppler shift, ip±0.5/ΔfdDenotes the abscissa, j, of the correlation peak at 0.5kHz left and right on the Doppler frequency shift axispIs the coordinate of the maximum correlation peak on the pseudo-code phase axis, fsFor the receiver sampling frequency, RcIs the code rate, j, of the spreading codep±0.5fs/RcRepresenting the abscissa of the correlation peak at 0.5 symbols left and right on the pseudo-code phase axis.
Specifically, the first-stage identification module and the second-stage identification module both adopt three-layer fully-connected BP neural networks, the number of input nodes of the first-stage identification module is 9, 9 characteristic parameters are used, the number of input nodes of the second-stage identification module is 11, and 11 characteristic parameters are used; the number of hidden layer nodes of the first-stage identification module is 12, and the number of hidden layer nodes of the second-stage identification module is 10; the number of output nodes of the first-stage identification module is 8, the number of output nodes of the second-stage identification module is 2, and the number corresponds to the number of classification labels at each stage.
In particular, the model of the received GNSS signals of the receiver may be represented as
Figure BDA0002444848140000055
Wherein the subscript i denotes the satellite number, AiRepresenting the amplitude of the signal, Ci(t) denotes a spreading code, D (t) denotes a navigation message, τiRepresenting the pseudo-code phase offset of the signal, fi-CRepresenting the carrier frequency, fi-DWhich is indicative of the doppler shift frequency and,
Figure BDA0002444848140000061
indicating the initial phase of the carrier.
Specifically, the interference source comprises single-tone interference STI, multi-tone interference MTI, linear frequency modulation interference LFMI, pulse interference PI, BPSK narrowband interference BPSKNBI, BPSK broadband interference BPSKWBI and deception interference SI, and the state of a received signal corresponding to a certain moment is divided into H0 without interference; h1, presence of SI; h2: the presence of MTI; h3, presence of LFMI; h4, presence of PI; h5, BPSK narrowband interference; h6, BPSK wideband interference; h7, there is spoofing interference.
Further, single-tone interference STI is modeled as:
Figure BDA0002444848140000062
the multi-tone interference MTI is modeled as:
Figure BDA0002444848140000063
the chirp interference LFMI is modeled as:
Figure BDA0002444848140000064
the impulse interference PI is modeled as:
Figure BDA0002444848140000065
BPSK narrow-band interference BPSKNBI is modeled as:
Figure BDA0002444848140000066
BPSK broadband interference BPSKWBI is modeled as:
Figure BDA0002444848140000067
the deceptive jamming SI is modeled as:
Figure BDA0002444848140000071
wherein, P represents the power of various types of pressing interference signals, f is the frequency of the interference signals,
Figure BDA0002444848140000072
subject to a uniformly distributed random phase on [0,2 π), f 0Represents the center frequency of the sweep frequency, K represents the linear sweep frequency, tau is the pulse duty cycle, TPIIs the pulse period, N is the number of pulses, aiRepresenting a random binary non-return-to-zero bit stream, g (T) representing a rectangular window, TbRepresenting the symbol width of a binary bit, BBPSKRepresenting the bandwidth of BPSK modulated signals, BGNSSIndicating GNSS signal bandwidth, the subscript "-S" indicates spoofed signals.
Further, the interference-to-real signal power ratio is expressed as JSR-10 lgPJ/PS，PJPower of interference, PSIs the power of the real satellite signal.
Compared with the prior art, the invention has at least the following beneficial effects:
the invention relates to a GNSS receiver combined interference classification identification method based on a two-stage neural network, which comprehensively considers scenes of suppressed interference and deceptive interference combined interference, utilizes two-stage identification modules aiming at the uncertainty of two major types of interference to finally achieve higher identification rate, and utilizes GPS signals and Beidou signals for testing.
Furthermore, the characteristic parameters of the first-stage identification module are mainly used for identifying the suppressed interference, and the set characteristic parameters refer to the characteristic parameter design of partial radar active interference identification and modulation mode identification. For example, the ratio of the maximum value to the second maximum value of the signal spectrum amplitude can effectively distinguish single-tone interference from other suppressed interference, the single-frequency energy concentration can effectively distinguish single-tone interference, multi-tone interference, impulse interference from other suppressed interference, and other characteristic parameters are similar.
Further, the characteristics used by the second-stage identification module are designed based on the difference of the navigation signal without the deception signal and the navigation signal capture result with the deception signal. When deception jamming exists, the correlation peak value is enlarged, 2 peaks are displayed when the code phase difference is large in the number of correlation peaks, the slope difference is generated when the code phase difference is small, the width of the correlation peak is widened, and the like.
Furthermore, the BP neural network has extremely strong nonlinear mapping capability, has the capability of associative memory of external stimulation and input information, has strong self-learning and self-adaptive capabilities and good generalization capability and fault-tolerant capability, and is commonly used in the aspects of classification, prediction and the like. Therefore, the algorithm adopted by the invention is a related algorithm of the BP neural network.
Furthermore, the model description of the received GNSS signal can be split into three parts, namely, a navigation message part, a spreading code part and a carrier part.
Further, various models of jamming and spoofing are described, the most common types of jamming that are typical.
Furthermore, the interference-signal ratio is an index for measuring the interference intensity, and the larger the interference-signal ratio is, the larger the interference power is, the larger the damage degree to the navigation system is.
In summary, the invention designs an interference identification scheme based on a two-stage neural network aiming at a pressing type and deception combined interference scene, wherein the two-stage module adopts a BP neural network, and can quickly and accurately identify a certain pressing type interference or deception interference which randomly appears by extracting different characteristic parameters, and has good effect on both GPS and Beidou signals.
The technical solution of the present invention is further described in detail by the accompanying drawings and embodiments.
Drawings
FIG. 1 is a schematic diagram of a scenario in which both squashed and spoofed interference are present;
FIG. 2 is a block diagram of a GNSS receiver and identification module framework;
FIG. 3 is a schematic diagram of the training of a neural network;
FIG. 4 is a schematic diagram of a decision tree-based interference identification process;
FIG. 5 shows the results of the individual tests for 7 types of interference under the GPS system, and the average recognition rate for 7 types of interference under the GPS system compared to the average recognition rate of the comparison scheme;
fig. 6 shows the results of the individual tests for 7 kinds of interferences under the BD system, and the average recognition rate for 7 kinds of interferences under the BD system is compared with the average recognition rate of the comparison scheme.
Detailed Description
The invention provides a GNSS receiver combined interference classification identification method based on a two-stage neural network.A first-stage identification module utilizes 9 characteristic parameters of a time domain, a frequency domain and a power domain extracted from a digital intermediate frequency signal to be sent to a BP neural network for identifying six typical suppressed interferences of single tone, multi-tone, linear frequency modulation, pulse, BPSK narrow band and BPSK broadband. Since the deceptive jamming has the same structure as the real satellite signal, effective discrimination cannot be realized by using the characteristic parameters adopted by the first-stage network. For this purpose, a second-stage identification module is introduced, and the real satellite signals and the deception signals are distinguished by using 11 characteristic parameters extracted from a two-dimensional array obtained by the signal acquisition of the receiver.
Referring to fig. 1, in consideration of a scenario in which a suppressing type interference and a deceptive interference exist in combination, a receiver may receive navigation signals sent by N visible satellites, a suppressing type interference source and a deceptive type interference source randomly launch an attack to the receiver, data information and directivity in the navigation signals are ignored, and a model of a received GNSS signal of the receiver is represented as
Figure BDA0002444848140000091
Wherein the subscript i denotes the satellite number, AiRepresenting the amplitude of the signal, Ci(t) denotes a spreading code, D (t) denotes a navigation message, τiRepresenting the pseudo-code phase offset of the signal, fi-CRepresenting the carrier frequency, fi-DWhich is indicative of the doppler shift frequency and,
Figure BDA0002444848140000092
indicating the initial phase of the carrier.
The signal received by the receiver, in addition to the useful satellite signal, also has background noise and possibly interference. The invention considers six types of pressing interference and forwarding deception interference, and the signal models of the pressing interference and the forwarding deception interference are shown in table 1, wherein STI is single tone interference, MTI is multi-tone interference, LFMI is linear frequency modulation interference, PI is pulse interference, BPSKNBI is BPSK narrow-band interference, BPSKWBI is BPSK wide-band interference, and SI is deception interference.
TABLE 1 interference types and modeling
Figure BDA0002444848140000101
In table 1, P represents the power of each type of the compressed interference signal, f is the frequency of the interference signal,
Figure BDA0002444848140000102
Obeying a random phase uniformly distributed over 0,2 pi).
In the LFMI signal model, f0Representing the center frequency of the sweep, K represents the lineThe frequency of the sexual sweep. In the PI model, τ is the pulse duty cycle, TPIIs the pulse period, and N is the number of pulses.
In the BPSKNBI and BPSKWBI models, aiRepresenting a random binary non-return-to-zero bitstream, g (T) representing a rectangular window, TbRepresenting the symbol width of a binary bit, BBPSKRepresenting the bandwidth of BPSK modulated signals, BGNSSRepresenting the GNSS signal bandwidth.
In the Spoofing Interference (SI) model, the lower subscript "-S" indicates the Spoofing signal, and the other parameters have the same meaning as in equation (1).
It is assumed that at any one time, if there is an attack, there is only one type of interference in table 1. Therefore, the received signal state at a certain time can be divided into 8 cases:
h0, no interference;
h1, presence of SI;
h2: the presence of MTI;
h3, presence of LFMI;
h4, presence of PI;
h5, BPSK narrowband interference;
h6, BPSK wideband interference;
h7, there is spoofing interference.
The Ratio of interference to real Signal Power (JSR) is expressed as JSR-10 lgPJ/PSIn which P isJPower of interference, PSIs the power of the real satellite signal.
Referring to fig. 2, a conventional GNSS software receiver may be divided into three modules, an antenna and rf front end, a baseband processing module and an application processing module, as shown in the upper half of fig. 2.
The invention relates to a GNSS receiver combined interference classification identification method based on a two-stage neural network, which is based on the structural design of a GNSS software receiver and adopts a two-stage identification scheme based on a BP neural network, wherein a first-stage identification module extracts time domain and frequency domain characteristics from digital intermediate-frequency signals after A/D conversion and sends the digital intermediate-frequency signals to the BP neural network for compression type interference detection and classification. Because the deception jamming code structure is the same as the real satellite signal, the deception jamming and the real satellite signal cannot be distinguished only by the basic time and frequency domain characteristics of the intermediate frequency signal. If the first-stage module identification result is that the digital intermediate frequency signal is not interfered or the deception jamming exists, the digital intermediate frequency signal is further captured, the relevant peak characteristics are extracted by using the captured two-dimensional search matrix, and then the two-dimensional search matrix is sent to the second-stage module for deception jamming detection. And when the final identification result of the two-stage module is non-interference, the received signal can be considered as a real satellite signal, otherwise, a corresponding interference processing means can be adopted according to the identified interference type.
The two-stage modules all adopt three layers of fully-connected BP neural networks: the number of input nodes is 9 and 11 respectively, the first-stage module uses 9 characteristic parameters, and the second-stage module uses 11 characteristic parameters; the number of hidden layer nodes is 12 and 10 respectively; the number of output nodes is 8 and 2 respectively, corresponding to the number of classification labels at each level. A schematic diagram of a two-stage network training network is shown in fig. 3.
First-level identification module
Firstly, the digital intermediate frequency signal is subjected to power normalization:
Figure BDA0002444848140000121
then, the normalized signal is corrected
Figure BDA0002444848140000122
Fourier transform is carried out to obtain a frequency spectrum X (k), and the following formula is not particularly explained, wherein the value range of k is 1-N. Then, the frequency domain normalization is performed to obtain the normalized frequency spectrum
Xu(k)＝X(k)/max[X(k)] (3)
Further, a power spectrum P (k) ([ X (k))]2. Normalizing the power spectrum to obtain a normalized power spectrum
Figure BDA0002444848140000123
Wherein
Figure BDA0002444848140000124
Represents the mean value of P (k).
In order to accurately identify various types of pressing type interference, the characteristic parameters adopted by the first-stage identification module are shown in table 2:
table 2 characteristic parameters used by the first level network
Figure BDA0002444848140000125
Figure BDA0002444848140000131
At x3In the calculation of (1), Pp(k) Representing the result of the impulse portion extraction in P (k), i.e. with normalized power Pu(k) The result of subtracting the sliding average applied to itself is expressed as
Figure BDA0002444848140000132
Where L is the length of the moving average window, and in the subsequent simulation, L is 1. At x4In the calculation of (2), mutIs the mean of x (n) and σ is the standard deviation of x (n). At x5And x6In the calculation of (2), muPIs the mean value of X (ω), σPIs the standard deviation of X (ω). At x9In the calculation of (2), card { } represents the number of collection elements, X'u(k) Is the result of X (k) normalization with mean, V 3dB＝0.707max[X′u(k)]. And training the BP neural network of the first-stage recognition module by using the 9 characteristics, wherein output labels are divided into 8 types corresponding to H0-H7.
Level 2 identification module
The data used by the second-stage identification module comes from a two-dimensional array generated after the digital intermediate frequency signal is subjected to capturing operation. If there is a satellite signal or a spoofed interference signal, there will be a correlation peak in the acquisition output. And calculating the plane projection of the correlation peaks on a code phase axis and a Doppler frequency shift axis, extracting corresponding characteristic parameters, and inputting the characteristic parameters into a neural network of a second-stage identification module for training. And recording the projection of the two-dimensional matrix A on the pseudo code phase axis and the Doppler frequency shift axis as Ac and Af respectively. The feature parameter set used by the second level identification module is shown in table 3:
TABLE 3 characteristic parameters used by the second level network
Figure BDA0002444848140000141
Wherein, ai,jIs the i, j th element, V, of the matrix ATIs the acquisition threshold of the receiver and,
Figure BDA0002444848140000142
is represented by AfThe set of all peak-to-peak values in the spectrum.
Figure BDA0002444848140000143
Is represented by AcThe set of all peak-to-peak values in the spectrum. At x18In the calculation of (A), BfIs AfI.e. the maximum correlation peak is shifted to the middle position and will be less than the capture threshold VTThe value of (a) is set to 0 while the remaining values are unchanged; i' PIs the maximum correlation peak at BfCoordinates of (2). At x19In the calculation of (A), BcIs AcThe result after the translation and the amplitude limiting; j'PIs the maximum correlation peak at BcOf (2) is calculated. x is the number of20In the calculation formula, ipIs the coordinate of the maximum peak on the Doppler shift axis, Δ fdSearch step size for Doppler shift, ip±0.5/ΔfdThe abscissa of the correlation peak at 0.5kHz left and right on the doppler shift axis is shown. At x21In the calculation of (1), jpIs the coordinate of the maximum correlation peak on the pseudo-code phase axis, fsFor the receiver sampling frequency, RcIs the code rate, j, of the spreading codep±0.5fs/RcThe abscissa representing the correlation peak at 0.5 symbol left and right on the pseudo-code phase axis. Training the second-stage recognition module by using the 11 characteristicsThe output labels of the BP neural network of (1) are classified into 2 types, i.e., H0 and H7.
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations. Thus, the following detailed description of the embodiments of the present invention, presented in the figures, is not intended to limit the scope of the invention, as claimed, but is merely representative of selected embodiments of the invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
Consider a scenario in which both squashing and spoofing are present, as shown in fig. 1. In a simulation experiment, by using simulated intermediate frequency data of a GPS L1 frequency point and a BDS B1 frequency point, a receiver can receive signals of 6-8 visible satellites, the receiving signal-to-noise ratio is-20 dB, the sampling frequency is 10.23MHz, the number of satellites forwarded by a deception interference source is 2-4, the pseudo-code phase difference between a multipath signal and a direct signal is 0.1-1 chip, and the Doppler frequency shift difference between the multipath signal and the direct signal is +/-100 Hz. Other relevant simulation parameters are shown in table 4.
The detailed simulation parameters are shown in table 1.
TABLE 4 simulation parameters
Figure BDA0002444848140000151
Figure BDA0002444848140000161
Comparison scheme:
because the press type interference and the deceptive interference are not considered simultaneously in the prior workIn order to illustrate the effectiveness of the proposed scheme, a classical Decision Tree (DT) scheme based on a threshold method is introduced, and according to the combined interference scenario considered by the present invention, the characteristic parameters and the threshold are properly designed and adjusted, so as to compare the performance with the scheme in this document. The basic idea of the decision tree is to compare each characteristic parameter value with a threshold value and classify the two classes step by step until only one interference type exists in each class set finally. The identification process is shown in fig. 4. Characteristic value f used therein 1～f6The calculation formula and the threshold value are shown in table 5.
TABLE 5 characteristic parameters and threshold values of decision trees
Figure BDA0002444848140000171
Figure BDA0002444848140000181
Fig. 5 shows the identification accuracy rate of the proposed scheme for each interference under the GPS system, and also shows the average identification rate of the proposed scheme and the threshold-based decision tree scheme for all interference types. It should be noted that the attack objectives of the jamming and the spoofing interference are different, and therefore, the power ranges adopted by the jamming and the spoofing interference are different. The former JSR is typically larger, typically greater than 10dB, while when spoofing attacks are made JSR is typically lower, typically less than 20 dB. In order to conveniently show the average identification effect of the scheme on all interferences, the result graph only shows the JSR range that two interference types can possibly obtain: 10-20 dB.
Fig. 6 shows the test results for BD data. It can be seen that the average recognition rate of the proposed scheme is above 96% in the JSR range under consideration, regardless of the GPS system or the BD system. When JSR is more than 17dB, the identification accuracy can reach 100%. And the identification precision of the decision tree scheme based on the threshold is only between 80 and 95 percent. The main reason is that when the interference detection and identification are carried out, the specific JSR cannot be accurately measured, so that an accurate threshold cannot be set according to the JSR, and the threshold can only be set according to the JSR interval. In addition, the decision tree is classified step by step, with the risk of cumulative errors. And thus the classification effect is poor. The invention extracts effective characteristic parameters according to the interference characteristics and utilizes the excellent classification capability of the neural network, thereby obtaining better identification effect.
In summary, the invention designs an interference identification scheme based on a two-stage neural network for a pressing type and deception combined interference scene, wherein the two-stage modules both adopt a BP neural network, and identification of pressing type interference and deception interference is respectively realized by extracting different characteristic parameters. The test results in the attached drawings show that the method can accurately classify and detect the randomly occurring pressing type and deception type interferences, and has high average recognition rate; and the data utilized by the scheme is positioned in the early stage of GNSS receiver processing and is not later than the signal capturing stage, so that the interference detection and identification can be realized as early as possible, and the timeliness of the navigation system for resisting interference is enhanced.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above-mentioned contents are only for illustrating the technical idea of the present invention, and the protection scope of the present invention is not limited thereby, and any modification made on the basis of the technical idea of the present invention falls within the protection scope of the claims of the present invention.

Claims (5)

1. A GNSS receiver combined interference classification identification method based on a two-stage neural network is characterized in that a receiver receives navigation signals sent by N visible satellites, a two-stage identification scheme based on a BP neural network is adopted according to a received GNSS signal model and an interference source, time domain and frequency domain characteristics are extracted from digital intermediate frequency signals after A/D conversion through a first-stage identification module, and the digital intermediate frequency signals are sent to the BP neural network for suppression type interference detection and classification; if the identification result of the first-stage identification module is non-interference or deception interference, capturing the digital intermediate frequency signal, extracting the relevant peak characteristics by using the captured two-dimensional search matrix, and sending the relevant peak characteristics to the second-stage identification module for deception interference detection; when the final identification result of the two-stage identification module is non-interference, judging that the received signal is a real satellite signal, and after identifying the interference type, adopting a corresponding interference processing means;
The projection of the two-dimensional capture matrix A generated in the receiver capture stage on the pseudo code phase axis and the Doppler frequency shift axis is respectively marked as AcAnd AfUsing the global accumulation of correlation values, the local accumulation of correlation values, the peak-to-peak value of correlation, AfNumber of correlation peaks in, AcNumber of correlation peaks in, AfCorrelation peak width of, AcCorrelation peak width of, AfThe cumulative amount of the symmetrical difference of the correlation peaks, AcThe cumulative amount of the symmetrical difference of the correlation peaks, AfDifference in slope and AcThe slope difference is 11 characteristics, a BP neural network of a second-stage identification module is trained, and output labels are divided into 2 types;
the correlation value global accumulation is:
Figure FDA0003598943170000011
the local cumulative amount of correlation values is:
Figure FDA0003598943170000012
the correlation peak-to-peak value is:
x13＝max{ai,j|ai,j∈A}
Afthe number of correlation peaks in (a) is:
Figure FDA0003598943170000013
Acthe number of correlation peaks in (a) is:
Figure FDA0003598943170000014
Afthe correlation peak width of (a) is:
x16＝card{Af|Af＞VT}
Acthe correlation peak width of (a) is:
x17＝card{Ac|Ac＞VT}
Afthe cumulative amount of the symmetrical difference of the correlation peaks is as follows:
Figure FDA0003598943170000021
Acthe cumulative amount of the symmetrical difference of the correlation peaks is as follows:
Figure FDA0003598943170000022
Afthe slope difference is:
x20＝Af(ip+0.5/Δfd)-Af(ip-0.5/Δfd)
Acthe slope difference is:
x21＝Ac(jp+0.5fs/Rc)-Ac(jp-0.5fs/Rc)
wherein, ai,jIs the i, j th element, V, of the matrix ATIs the acquisition threshold of the receiver and is,
Figure FDA0003598943170000023
is shown as AfA set of all peak-to-peak values;
Figure FDA0003598943170000024
is represented by AcThe set of all peak values in the wave; b isfIs AfThe result after the translation and the amplitude limiting; i'PIs the maximum correlation peak at BfCoordinate of (1), B cIs AcThe result after the translation and the amplitude limiting; j'PIs the maximum correlation peak at BcCoordinate of (1), ipIs the coordinate of the maximum peak on the Doppler shift axis, Δ fdSearch step size for Doppler shift, ip±0.5/ΔfdDenotes the abscissa, j, of the correlation peak at 0.5kHz left and right on the Doppler frequency shift axispIs the coordinate of the maximum correlation peak on the pseudo-code phase axis, fsFor the receiver sampling frequency, RcIs the code rate, j, of the spreading codep±0.5fs/RcThe abscissa representing the correlation peak at 0.5 symbol left and right on the pseudo code phase axis;
the first-stage identification module and the second-stage identification module both adopt three-layer fully-connected BP neural networks, the number of input nodes of the first-stage identification module is 9, 9 characteristic parameters are used, the number of input nodes of the second-stage identification module is 11, and 11 characteristic parameters are used; the number of hidden layer nodes of the first-stage identification module is 12, and the number of hidden layer nodes of the second-stage identification module is 10; the number of output nodes of the first-stage identification module is 8, the number of output nodes of the second-stage identification module is 2, and the output nodes correspond to the number of classification labels at each stage;
the received GNSS signal model of the receiver is represented as
Figure FDA0003598943170000031
Wherein the subscript i denotes the satellite number, AiRepresenting the amplitude of the signal, Ci(t) denotes a spreading code, D (t) denotes a navigation message, τ iRepresenting the pseudo-code phase offset of the signal, fi-CRepresenting the carrier frequency, fi-DWhich is indicative of the doppler shift frequency of the signal,
Figure FDA0003598943170000032
representing the initial phase of the carrier wave;
interference sources comprise single tone interference STI, multi-tone interference MTI, linear frequency modulation interference LFMI, pulse interference PI, BPSK narrow-band interference BPSKNBI, BPSK broadband interference BPSKWBI and deception interference SI, and the state of a received signal corresponding to a certain moment is divided into H0 without interference; h1, presence of SI; h2: the presence of MTI; h3, presence of LFMI; h4, presence of PI; h5, BPSK narrowband interference; h6, BPSK wideband interference; h7, there is spoofing interference.
2. The GNSS receiver combined interference classification recognition method based on the two-stage neural network as claimed in claim 1, wherein the BP neural network of the first stage recognition module is trained by using the ratio of the maximum value to the second maximum value of the signal spectrum amplitude, the single frequency energy concentration, the average spectrum flatness coefficient, the time domain kurtosis, the power spectrum skewness, the ratio of the power spectrum kurtosis spectrum variance to the mean square, the normalized spectrum peak-to-average ratio and the 3dB bandwidth of the normalized spectrum, and the output labels are classified into 8 types.
3. The GNSS receiver combined interference classification recognition method based on two-stage neural network as claimed in claim 2, wherein the ratio of the maximum value to the second maximum value of the signal spectrum amplitude is:
x1＝|X(k)|1stmax/|X(k)|2ndmax
The single frequency energy concentration is:
Figure FDA0003598943170000033
the average spectral flatness coefficient is:
Figure FDA0003598943170000034
the time domain kurtosis is:
x4＝E(|x(n)-μt|4)/σt 4
the power spectrum skewness is:
x5＝E[X(ω)-μP]3/σP 3
the power spectrum kurtosis is:
x6＝E[X(ω)-μP]4/σP 4
the ratio of the spectral variance to the mean squared is:
x7＝σf 2/μf 2
the normalized spectral peak-to-average ratio is:
x8＝max{Xu(k)}/E[Xu(k)]
the 3dB bandwidth of the normalized spectrum is:
x9＝card{k|X′u(k)＞V3dB}/card{k|X′u(k)}
wherein, Pp(k) Denotes the result of the impulse portion extraction in P (k), μtIs the mean value of x (n), σtIs the standard deviation of x (n); mu.sPIs the mean value of X (ω), σPIs the standard deviation of X (ω); card { } denotes the number of collection elements, X'u(k) Is the result of X (k) normalization with mean, V3dB＝0.707max[X′u(k)]。
4. The GNSS receiver combined interference classification recognition method based on two-stage neural network as claimed in claim 1, wherein the single tone interference STI is modeled as:
Figure FDA0003598943170000041
the multi-tone interference MTI is modeled as:
Figure FDA0003598943170000042
the chirp disturbance LFMI is modeled as:
Figure FDA0003598943170000043
the impulse interference PI is modeled as:
Figure FDA0003598943170000051
BPSK narrow-band interference BPSKNBI is modeled as:
Figure FDA0003598943170000052
BPSK broadband interference BPSKWBI is modeled as:
Figure FDA0003598943170000053
the deceptive jamming SI is modeled as:
Figure FDA0003598943170000054
wherein, P represents the power of various types of pressing interference signals, f is the frequency of the interference signals,
Figure FDA0003598943170000055
subject to a uniformly distributed random phase on [0,2 π), f0Representing the center frequency of the sweep frequency, K representing the linear sweep frequency, τ being the pulse duty cycle, T PIIs the pulse period, N is the number of pulses, aiRepresenting a random binary non-return-to-zero bitstream, g (T) representing a rectangular window, TbRepresenting the symbol width of a binary bit, BBPSKRepresenting the bandwidth of a BPSK modulated signal, BGNSSIndicating GNSS signal bandwidth, the subscript "-S" indicates spoofed signals.
5. The GNSS receiver combined interference classification recognition method based on two-stage neural network as claimed in claim 1, wherein the interference to true signal power ratio is expressed as JSR-10 lgPJ/PS，PJPower of interference, PSIs the power of the real satellite signal.
CN202010276113.XA 2020-04-09 2020-04-09 GNSS receiver combined interference classification and identification method based on two-stage neural network Active CN111399002B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010276113.XA CN111399002B (en) 2020-04-09 2020-04-09 GNSS receiver combined interference classification and identification method based on two-stage neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010276113.XA CN111399002B (en) 2020-04-09 2020-04-09 GNSS receiver combined interference classification and identification method based on two-stage neural network

Publications (2)

Publication Number Publication Date
CN111399002A CN111399002A (en) 2020-07-10
CN111399002B true CN111399002B (en) 2022-06-07

Family

ID=71435061

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010276113.XA Active CN111399002B (en) 2020-04-09 2020-04-09 GNSS receiver combined interference classification and identification method based on two-stage neural network

Country Status (1)

Country Link
CN (1) CN111399002B (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20210334626A1 (en) * 2020-04-28 2021-10-28 Novatel Inc. Gnss-receiver interference detection using deep learning
CN112083448B (en) * 2020-09-04 2023-08-15 哈尔滨工程大学 Satellite navigation system-oriented interference signal classification recognition feature extraction method and system
CN112859116B (en) * 2020-12-28 2022-11-15 杭州电子科技大学 Single-antenna GNSS deception jamming detection method based on main feature blind source separation
CN113221781B (en) * 2021-05-20 2022-12-06 成都海擎科技有限公司 Carrier signal detection method based on multitask deep convolutional neural network
CN113376659B (en) * 2021-06-15 2022-05-13 东南大学 GNSS (global navigation satellite system) generated deception jamming detection method based on BP (back propagation) neural network
CN113970762B (en) * 2021-09-26 2024-05-24 西安交通大学 Multistage interference source positioning method and system
CN116125499B (en) * 2021-11-12 2024-04-09 北京六分科技有限公司 Method, device and system for detecting intermediate frequency data
CN114325600B (en) * 2021-12-22 2023-09-26 广东邦盛北斗科技股份公司 Anti-interference method, system and device for Beidou navigation system and cloud platform
CN114114227B (en) * 2022-01-27 2022-05-31 安徽京淮健锐电子科技有限公司 Radar signal modulation type identification method based on dual-channel heterogeneous fusion network
GB2621176A (en) * 2022-08-05 2024-02-07 Ordnance Survey Ltd Detecting jamming in OSNET using machine learning
CN116009034B (en) * 2022-12-21 2023-11-07 湖南工商大学 Satellite signal capturing method, baseband signal processing unit, receiver and medium
CN116482720B (en) * 2023-06-26 2023-09-05 山东科技大学 GNSS deception jamming detection method based on machine learning theory

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107238847A (en) * 2017-04-21 2017-10-10 中国科学院光电研究院 A kind of satellite navigation interferer signal detection method and system
CN109061577A (en) * 2018-07-05 2018-12-21 西安电子科技大学 A kind of recognition methods of different types of covering type interference and Deceiving interference
CN109782304A (en) * 2019-01-29 2019-05-21 西安交通大学 A kind of the GNSS cheating interference detection method and system of signal acquisition phase
CN110826630A (en) * 2019-11-08 2020-02-21 哈尔滨工业大学 Radar interference signal feature level fusion identification method based on deep convolutional neural network

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9651671B2 (en) * 2014-08-14 2017-05-16 Honeywell International Inc. Interference detection using multiple automatic gain controllers
US20180024248A1 (en) * 2016-07-20 2018-01-25 Honeywell International Inc. Systems and methods for nequick modeling using neural networks

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN107238847A (en) * 2017-04-21 2017-10-10 中国科学院光电研究院 A kind of satellite navigation interferer signal detection method and system
CN109061577A (en) * 2018-07-05 2018-12-21 西安电子科技大学 A kind of recognition methods of different types of covering type interference and Deceiving interference
CN109782304A (en) * 2019-01-29 2019-05-21 西安交通大学 A kind of the GNSS cheating interference detection method and system of signal acquisition phase
CN110826630A (en) * 2019-11-08 2020-02-21 哈尔滨工业大学 Radar interference signal feature level fusion identification method based on deep convolutional neural network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于BAS-BP神经网络的遮盖干扰信号识别;杨洁等;《西安邮电大学学报》;20190930;第24卷(第5期);正文第1-4小节，附图1-4 *

Also Published As

Publication number Publication date
CN111399002A (en) 2020-07-10

Similar Documents

Publication Publication Date Title
CN111399002B (en) GNSS receiver combined interference classification and identification method based on two-stage neural network
CN109782304B (en) GNSS deception jamming detection method and system in signal capture stage
CN110231633B (en) GNSS deception jamming identification and inhibition method and system based on LSTM in signal capture stage
CN109188470B (en) GNSS deception jamming detection method based on convolutional neural network
CN106125053B (en) Pulse Doppler radar polarization anti jamming method
Yang et al. Dense people counting using IR-UWB radar with a hybrid feature extraction method
CN104267379B (en) A kind of active radar and passive radar based on Waveform Design works in coordination with anti-interference method
CN102323601B (en) Method for capturing GNSS-BOS (Global Navigation Satellite System-Binary Offset Carrier) modulation signal
CN105137396B (en) The detection method that a kind of SMSP interference and C＆I are disturbed
CN113406671B (en) Based on C/N0GNSS forwarding type deception jamming detection method of-MV
CN110231634A (en) A kind of GNSS cheating interference detection method and system of the signal acquisition phase based on LSTM
CN103954943A (en) Networked radar system deceptive jamming resisting method
CN116577808B (en) Navigation deception jamming detection method based on output of receiver correlator
CN108983165B (en) Substation selection-based anti-deception jamming method for multi-station radar system
Lu et al. An improved sliding matched filter method for interrupted sampling repeater jamming suppression based on jamming reconstruction
CN107272030B (en) Target detection method based on information fusion of multiple GPS satellites
CN113534067B (en) Multi-station radar deception jamming resisting method and system under multiple interference sources
FR2821163A1 (en) METHOD OF DETECTING MOBILE OBJECTS USING PASSIVE SUBSEA BUOYS
CN107064903A (en) Object detection method based on GLRT under a kind of multiple isomery satellites
Chen et al. GNSS interference type recognition with fingerprint spectrum DNN method
Cao et al. DNN-based peak sequence classification CFAR detection algorithm for high-resolution FMCW radar
Brieger et al. Multimodal learning for reliable interference classification in GNSS signals
Wu et al. Jammer classification with federated learning
Colone et al. Reference-free amplitude-based WiFi passive sensing
Guven et al. Classifying LPI radar waveforms with time-frequency transformations using multi-stage CNN 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