CN116087994B - Deception jamming detection method based on machine learning - Google Patents

Deception jamming detection method based on machine learning Download PDF

Info

Publication number
CN116087994B
CN116087994B CN202310361098.2A CN202310361098A CN116087994B CN 116087994 B CN116087994 B CN 116087994B CN 202310361098 A CN202310361098 A CN 202310361098A CN 116087994 B CN116087994 B CN 116087994B
Authority
CN
China
Prior art keywords
satellite
real
navigation
navigation signal
satellite navigation
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
CN202310361098.2A
Other languages
Chinese (zh)
Other versions
CN116087994A (en
Inventor
陈世淼
倪淑燕
雷拓峰
程凌峰
付琦玮
堵洪峰
毛文轩
张英健
王海宁
宋鑫
贾卓娅
王治宇
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.)
Peoples Liberation Army Strategic Support Force Aerospace Engineering University
Original Assignee
Peoples Liberation Army Strategic Support Force Aerospace Engineering 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 Peoples Liberation Army Strategic Support Force Aerospace Engineering University filed Critical Peoples Liberation Army Strategic Support Force Aerospace Engineering University
Priority to CN202310361098.2A priority Critical patent/CN116087994B/en
Publication of CN116087994A publication Critical patent/CN116087994A/en
Application granted granted Critical
Publication of CN116087994B publication Critical patent/CN116087994B/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
    • G01S19/015—Arrangements for jamming, spoofing or other methods of denial of service of such systems
    • 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
    • G01S19/215—Interference related issues ; Issues related to cross-correlation, spoofing or other methods of denial of service issues related to spoofing
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Abstract

The invention provides a deception jamming detection method based on machine learning, which takes the difference of other variables caused by the difference of the true satellite navigation signal and the deception satellite navigation signal as breakthrough points and utilizes the carrier phase double difference, the signal CNR double difference, the signal power double difference and the angle difference of the signal arrival direction obtained by a rotary single antenna model as characteristics to realize deception jamming detection by a machine learning training detection model; therefore, compared with the traditional rotation single-antenna deception jamming detection method, the method introduces the angle difference of the signal arrival direction to perform deception jamming detection, can realize the detection of multi-station deception jamming, utilizes a plurality of characteristics to detect, introduces machine learning, has more excellent detection performance, and greatly improves the accuracy of deception detection.

Description

Deception jamming detection method based on machine learning
Technical Field
The invention belongs to the technical field of navigation spoofing interference detection, and particularly relates to a spoofing interference detection method based on machine learning.
Background
GNSS is extremely susceptible to rogue interference due to the weak navigation signal strength, open signal modulation, predictable partial navigation data, etc. of the global satellite navigation system, which is fatal to the navigation system. The deceptive jamming is that deceptive equipment sends out false navigation signals similar to real navigation signals, and the target receiver mistakes the false navigation signals into the real navigation signals through strategies, so that the target receiver acquires wrong positioning, speed or time information.
In a navigation system, most navigation satellites are in a moving state relative to an earth coordinate system, the direction of a real navigation signal received by a receiver is also in a dynamic change, current common deception interference is generally generated by the same deception equipment, deception signals come from the same direction, complex deception interference respectively transmits deception navigation signals through a plurality of antennas, and the directionality of the real navigation signal cannot be completely simulated. Thus, detection of a spoof signal may be made with the spoof signal and the actual signal differing and differing from each other as breakthrough points for other variables caused by the differing.
Disclosure of Invention
In order to solve the problems, the invention provides a spoofing interference detection method based on machine learning, which can realize the detection of multi-station spoofing interference and has high detection accuracy.
A spoofing interference detection method based on machine learning comprises the following steps:
s1: respectively acquiring a real satellite navigation signal and a deception satellite navigation signal as sample data through a rotary single antenna, wherein the deception satellite navigation signal is generated by a satellite navigation signal simulator;
s2: extracting features for deception detection from a real satellite navigation signal and a deception satellite navigation signal respectively, wherein the features comprise carrier phase double difference observed values, CNR normalized value double differences, power normalized value double differences and angle differences of signal arrival directions; meanwhile, labeling tags of real signals or deception signals for the values of each group of characteristics, and dividing sample data into a training set and a testing set according to a set proportion;
s3: respectively inputting each group of features in the training set and labels corresponding to each group of features into a support vector machine model, a neural network model and a Bayesian classification model for training to obtain three prediction models for deception detection;
s4: and respectively predicting each group of features in the test set by adopting three prediction models, so that each group of features obtains three deception detection results, and finally obtaining a final deception detection result by a voting method.
Further, the method for calculating the carrier phase double-difference observed value between any two real navigation satellites or between any two satellite navigation signal simulators comprises the following steps:
step 1: acquiring the variation of carrier phase measurement values of M times of current time and previous time caused by rotation of a rotary single antenna corresponding to each real navigation satellite or each satellite navigation signal simulator
Figure SMS_1
：
Figure SMS_2
Where i denotes the ith real navigation satellite or the ith satellite navigation signal simulator,
Figure SMS_3
for the carrier phase of the real navigation satellite i or satellite navigation signal simulator i at the current time k,/i>
Figure SMS_4
For the carrier phase of the real navigation satellite i or satellite navigation signal simulator i at the previous instant k-1, and (2)>
Figure SMS_5
Is a real navigation guard caused by the movement between the real navigation satellites or between the satellite navigation signal simulators between two momentsThe distance change from the satellite i or the satellite navigation signal simulator i to the origin of the rotary single antenna can be obtained through calculation of ephemeris data;
step 2: calculating carrier phase double difference between any two real navigation satellites or between any two satellite navigation signal simulators according to the following formula
Figure SMS_6
：
Figure SMS_7
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_8
and the variation of the carrier phase measurement value at the kth moment caused by rotation of the rotary single antenna corresponding to the jth real navigation satellite or the jth satellite navigation signal simulator is represented.
Further, the calculation method of the CNR normalization value double difference between any two real navigation satellites or between any two satellite navigation signal simulators comprises the following steps:
step 1: acquiring carrier-to-noise ratios of M times of the current time and the previous time of each real navigation satellite or each satellite navigation signal simulator
Figure SMS_9
Wherein i represents the ith real navigation satellite or the ith satellite navigation signal simulator;
step 2: respectively calculating carrier-to-noise ratios of M moments corresponding to each real navigation satellite or each satellite navigation signal simulator
Figure SMS_10
Mean>
Figure SMS_11
And standard deviation->
Figure SMS_12
；
Step 3: guiding each real navigation satellite or each satellite respectivelyCarrier-to-noise ratio at M times corresponding to avionic signal simulator
Figure SMS_13
Normalization is carried out:
Figure SMS_14
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_15
the carrier-to-noise ratio of the normalized real navigation satellite i or the satellite navigation signal simulator i at the k moment;
step 4: calculating CNR normalized value single difference of each real navigation satellite or each satellite navigation signal simulator at k time
Figure SMS_16
：
Figure SMS_17
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_18
the carrier-to-noise ratio of the normalized real navigation satellite i or the satellite navigation signal simulator i at the time k+1;
step 5: obtaining a CNR normalized value double difference between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure SMS_19
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_20
CNR normalized value single difference at k moment for real navigation satellite i or satellite navigation signal simulator i, < ->
Figure SMS_21
For true navigation satellite j or satellite navigationCNR normalized value single difference of signal simulator j at k moment, < >>
Figure SMS_22
And (3) performing double differences on the normalized values of the CNR between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j.
Further, the calculation method of the power normalization value double difference between any two real navigation satellites or between any two satellite navigation signal simulators comprises the following steps:
step 1: acquiring the signal power of M times of the current time and the previous time of each real navigation satellite or each satellite navigation signal simulator
Figure SMS_23
Wherein i represents the ith real navigation satellite or the ith satellite navigation signal simulator;
step 2: respectively calculating the power of each real navigation satellite or M moments corresponding to each satellite navigation signal simulator
Figure SMS_24
Mean>
Figure SMS_25
And standard deviation->
Figure SMS_26
；
Step 3: respectively using the powers of the real navigation satellites or the satellite navigation signal simulators at M moments
Figure SMS_27
Normalization is carried out:
Figure SMS_28
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_29
for normalized real navigation satellite i or satellite navigation signalThe power of the simulator i at time k;
step 4: calculating signal power normalization value single difference of each real navigation satellite or each satellite navigation signal simulator at k time
Figure SMS_30
：
Figure SMS_31
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_32
the power of the normalized real navigation satellite i or satellite navigation signal simulator i at the time k+1;
step 5: acquiring signal power normalization value double differences between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure SMS_33
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_34
normalized value single difference for signal power of real navigation satellite i or satellite navigation signal simulator i at k moment, < ->
Figure SMS_35
Normalized value single difference for signal power of real navigation satellite j or satellite navigation signal simulator j at k moment, < ->
Figure SMS_36
And normalizing the double differences of the signal power between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j.
Further, the method for calculating the angle difference of the signal arrival direction comprises the following steps:
step 1: reading ephemeris information of navigation signals, and calculating current and previous timeThe position coordinates of the true navigation satellite i and the true navigation satellite j or the satellite navigation signal simulator i and the satellite navigation signal simulator j at M moments are carved, wherein the position coordinates at the kth moment are expressed as
Figure SMS_37
；
Step 2: reading the position information of M times of the current and previous times of the receiver, wherein the position information of the kth time is expressed as
Figure SMS_38
；
Step 3: acquiring the angle difference of the signal arrival direction between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure SMS_39
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_40
the angle difference of the signal arrival direction between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j at the kth time is represented.
Further, the real satellite navigation signal is obtained by receiving the real navigation signal in an open scene through a rotary single antenna and collecting and recording; the deception satellite navigation signals are generated by a plurality of satellite navigation signal simulators in a closed environment, and navigation signals of different navigation satellites are generated by different satellite navigation signal simulators and are received by a rotary single antenna.
The beneficial effects are that:
1. the invention provides a deception jamming detection method based on machine learning, which takes the difference of other variables caused by the difference of the true satellite navigation signal and the deception satellite navigation signal as breakthrough points and utilizes the carrier phase double difference, the signal CNR double difference, the signal power double difference and the angle difference of the signal arrival direction obtained by a rotary single antenna model as characteristics to realize deception jamming detection by a machine learning training detection model; therefore, compared with the traditional rotation single-antenna deception jamming detection method, the method introduces the angle difference of the signal arrival direction to perform deception jamming detection, can realize the detection of multi-station deception jamming, utilizes a plurality of characteristics to detect, introduces machine learning, has more excellent detection performance, and greatly improves the accuracy of deception detection.
2. The invention provides a deception jamming detection method based on machine learning, which can realize sample data acquisition by only a single antenna, and reduces the observation cost compared with the traditional multi-antenna deception jamming detection method.
3. The invention provides a deception jamming detection method based on machine learning, which is used for collecting deception satellite navigation signals in a closed environment and can avoid being influenced by real satellite navigation signals.
Drawings
Fig. 1 is a flowchart of a method for detecting fraud based on machine learning.
Fig. 2 is a schematic diagram of a rotary single antenna model according to the present invention.
Detailed Description
In order to enable those skilled in the art to better understand the present application, the following description will make clear and complete descriptions of the technical solutions in the embodiments of the present application with reference to the accompanying drawings in the embodiments of the present application.
The deception jamming detection is essentially a binary classification problem, and machine learning has wide application and good performance on the binary classification problem. Therefore, the traditional deception jamming detection method is combined with machine learning, and the machine learning model is used for processing navigation data to realize deception jamming detection. As shown in fig. 1, the method for detecting spoofing interference based on machine learning provided by the invention comprises the following steps:
s1: and respectively acquiring a real satellite navigation signal and a deception satellite navigation signal as sample data by rotating the single antenna, wherein the deception satellite navigation signal is generated by a satellite navigation signal simulator.
Further, the rotating single antenna model is composed as shown in FIG. 2, the base is a horizontal turntable, and the angular velocity is the same as that of the rotating single antenna model
Figure SMS_41
Rotating, wherein the antenna of the receiver is positioned above the turntable, and the included angle between the antenna and the horizontal turntable is +.>
Figure SMS_42
。
Further, the real satellite navigation signal is obtained by receiving the real navigation signal in an open scene through a rotary single antenna and collecting and recording; the deception satellite navigation signals are generated by a plurality of satellite navigation signal simulators in a closed environment, and navigation signals of different navigation satellites are generated by different satellite navigation signal simulators and are received by a rotary single antenna. The acquisition of the deception satellite navigation signals is performed in a closed environment, so that the influence of the real satellite navigation signals can be avoided.
S2: extracting features for deception detection from a real satellite navigation signal and a deception satellite navigation signal respectively, wherein the features comprise carrier phase double difference observed values, CNR normalized value double differences, power normalized value double differences and angle differences of signal arrival directions; meanwhile, labeling tags of real signals or deception signals for the values of each group of characteristics, and dividing sample data into a training set and a testing set according to a set proportion.
For example, 70% of the sample data is used as the training set and 30% of the sample data is used as the test set.
S3: and respectively inputting each group of features in the training set and labels corresponding to each group of features into a support vector machine model, a neural network model and a Bayesian classification model for training to obtain three prediction models for deception detection.
S4: and respectively predicting each group of features in the test set by adopting three prediction models, so that each group of features obtains three deception detection results, and finally obtaining a final deception detection result by a voting method.
The specific calculation methods of the carrier phase double difference observation value, the CNR normalized value double difference, the power normalized value double difference and the angle difference of the signal arrival direction are described in detail below. Meanwhile, it should be noted that the method for extracting the features for spoofing detection from the true satellite navigation signal and the spoofing satellite navigation signal is the same, so the present invention incorporates the feature extraction methods of both signals together.
Carrier phase double difference observations
The method for calculating the carrier phase double-difference observed value between any two real navigation satellites or between any two satellite navigation signal simulators comprises the following steps:
step 1: acquiring the variation of carrier phase measurement values of M times of current time and previous time caused by rotation of a rotary single antenna corresponding to each real navigation satellite or each satellite navigation signal simulator
Figure SMS_43
：
Figure SMS_44
Where i denotes the ith real navigation satellite or the ith satellite navigation signal simulator,
Figure SMS_45
for the carrier phase of the real navigation satellite i or satellite navigation signal simulator i at the current time k,/i>
Figure SMS_46
For the carrier phase of the real navigation satellite i or satellite navigation signal simulator i at the previous instant k-1, and (2)>
Figure SMS_47
The distance change between the real navigation satellite i or the satellite navigation signal simulator i and the origin of the rotary single antenna, which is caused by the movement between the real navigation satellites or between the satellite navigation signal simulators, can be obtained through ephemeris data calculation;
step 2: calculating carrier phase double difference between any two real navigation satellites or between any two satellite navigation signal simulators according to the following formula
Figure SMS_48
：
Figure SMS_49
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_50
and the variation of the carrier phase measurement value at the kth moment caused by rotation of the rotary single antenna corresponding to the jth real navigation satellite or the jth satellite navigation signal simulator is represented.
(II) double difference of CNR normalized value
The calculation method of the CNR normalization value double difference between any two real navigation satellites or between any two satellite navigation signal simulators comprises the following steps:
step 1: acquiring carrier-to-noise ratios of M times of the current time and the previous time of each real navigation satellite or each satellite navigation signal simulator
Figure SMS_51
Wherein i represents the ith real navigation satellite or the ith satellite navigation signal simulator;
step 2: respectively calculating carrier-to-noise ratios of M moments corresponding to each real navigation satellite or each satellite navigation signal simulator
Figure SMS_52
Mean>
Figure SMS_53
And standard deviation->
Figure SMS_54
；
Step 3: the carrier-to-noise ratios of M moments corresponding to each real navigation satellite or each satellite navigation signal simulator are respectively calculated
Figure SMS_55
Normalization is carried out:
Figure SMS_56
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_57
the carrier-to-noise ratio of the normalized real navigation satellite i or the satellite navigation signal simulator i at the k moment;
step 4: calculating CNR normalized value single difference of each real navigation satellite or each satellite navigation signal simulator at k time
Figure SMS_58
：
Figure SMS_59
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_60
the carrier-to-noise ratio of the normalized real navigation satellite i or the satellite navigation signal simulator i at the time k+1;
step 5: obtaining a CNR normalized value double difference between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure SMS_61
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_62
CNR normalized value single difference at k moment for real navigation satellite i or satellite navigation signal simulator i, < ->
Figure SMS_63
CNR normalized value single difference at k moment for real navigation satellite j or satellite navigation signal simulator j, < ->
Figure SMS_64
And (3) performing double differences on the normalized values of the CNR between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j.
(III) Power normalization value double difference
The calculation method of the power normalization value double difference between any two real navigation satellites or between any two satellite navigation signal simulators comprises the following steps:
step 1: acquiring the signal power of M times of the current time and the previous time of each real navigation satellite or each satellite navigation signal simulator
Figure SMS_65
Wherein i represents the ith real navigation satellite or the ith satellite navigation signal simulator;
step 2: respectively calculating the power of each real navigation satellite or M moments corresponding to each satellite navigation signal simulator
Figure SMS_66
Mean>
Figure SMS_67
And standard deviation->
Figure SMS_68
；
Step 3: respectively using the powers of the real navigation satellites or the satellite navigation signal simulators at M moments
Figure SMS_69
Normalization is carried out:
Figure SMS_70
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_71
for the normalized real navigation satellite i or satellite navigation signal simulator i at k timeIs a power of (2);
step 4: calculating signal power normalization value single difference of each real navigation satellite or each satellite navigation signal simulator at k time
Figure SMS_72
：
Figure SMS_73
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_74
the power of the normalized real navigation satellite i or satellite navigation signal simulator i at the time k+1;
step 5: acquiring signal power normalization value double differences between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure SMS_75
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_76
normalized value single difference for signal power of real navigation satellite i or satellite navigation signal simulator i at k moment, < ->
Figure SMS_77
Normalized value single difference for signal power of real navigation satellite j or satellite navigation signal simulator j at k moment, < ->
Figure SMS_78
And normalizing the double differences of the signal power between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j.
(IV) angular Difference in the arrival direction of the Signal
The method for calculating the angle difference of the signal arrival direction comprises the following steps:
step 1: reading ephemeris information of navigation signals, and calculating position coordinates of a real navigation satellite i and a real navigation satellite j or a satellite navigation signal simulator i and a satellite navigation signal simulator j at M times, wherein the position coordinates at the kth time are expressed as:
Figure SMS_79
；
step 2: reading position information of M times which are the current and previous times of the receiver, wherein the position information of the kth time is expressed as:
Figure SMS_80
；
step 3: acquiring the angle difference of the signal arrival direction between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure SMS_81
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure SMS_82
the angle difference of the signal arrival direction between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j at the kth time is represented.
The larger the value of M, the higher the detection success rate, and the slower the detection speed.
Furthermore, in step S3, the performance of the support vector machine model is mainly related to two parameters, namely RBF kernel parameter g and penalty factor C, and the selection of the parameters is generally determined according to multiple tests of a specific model.
Further, in the step 3, parameter optimization is performed on the SVM model by using an adaptive variation particle swarm optimization algorithm, and a prediction model is obtained through training, wherein the process of optimizing the support vector machine model parameters by using the adaptive variation particle swarm optimization algorithm is as follows:
step 1: randomly initializing model parameters to obtain a support vector machine model parameter combination
Figure SMS_83
The position in the solution space and the initial velocity and position of the individual particles. Each particle swarm can only optimize one model parameter, and the dimension of the particle swarm is set
Figure SMS_84
；
Step 2: the fitness of each particle is calculated.
Figure SMS_87
Root mean square error as an objective function +.>
Figure SMS_90
Setting the initial position of each particle as the current individual extremum for the fitness function>
Figure SMS_92
According to->
Figure SMS_86
Obtaining individual fitness value of each particle +.>
Figure SMS_89
If->
Figure SMS_91
Updating the individual optimum value +.>
Figure SMS_93
Selecting maximum->
Figure SMS_85
Updating to the group optimal solution->
Figure SMS_88
。
Figure SMS_94
Figure SMS_95
Step 3: and updating the particle swarm. Updating the particle position and speed according to the set rule, and calculating new fitness value
Figure SMS_96
。
Step 4: adaptive variation. When the particle position is greater than the variation threshold
Figure SMS_97
In the process, a mutation operation is performed to update the fitness value +.>
Figure SMS_98
Thereafter, the operation of step 3 is followed, comparing the individual optimum extremum of the current particle +.>
Figure SMS_99
And the current fitness value +.>
Figure SMS_100
Obtaining a population optimal solution->
Figure SMS_101
。
Step 5: the iteration is stopped. Stopping iterating to obtain the optimal SVM model parameter combination when the comparison condition is met or the iteration number is larger than the maximum iteration number
Figure SMS_102
Otherwise, continuing iteration and returning to the step 2.
In the step S3, the support vector machine model performs classification tasks by using "one-to-many" different features, and if the obtained sample features are n, then n-1 SVM classifiers are built first, in the ith classifier, the samples with the feature a are marked as +, the samples with the rest features are marked as-, then the samples with the feature a are output, then the samples with the rest features are sent to the (i+1) th classifier, the samples with the feature a+1 are continuously separated, and then the rest features are sequentially and backwardly transmitted to other SVM classifiers for training.
Further, the neural network model in step S3 is composed of four parts: an input layer, a mode layer, an add layer, and an output layer. The input layer is used to receive input samples and after normalization, the sample data is passed to the network and the number of neurons is the same as the dimension of the sample feature vector. The pattern layer first analyzes the matching relationship between the input sample feature vector and the different sample labels in the training set and then sends it in the form of euclidean distance to the node activation function to obtain the output of each neuron. The adding layer performs weighted summation on the output of the neurons in the same category in the mode layer, and then calculates probability density functions corresponding to the sample data and the labels. The output layer is composed of a threshold comparator. And judging the signal category to which the observation sample belongs according to the probability density values output by the neurons of different types in the addition layer, and selecting the label type corresponding to the neuron with the highest probability density as the signal type of the sample, thereby obtaining a cheating test result.
Further, the training process of the neural network model in the step 3 is as follows: initializing network weight, inputting the characteristics of training samples, transmitting sample characteristic vectors to a mode layer after normalization, analyzing and calculating errors of samples under different deception models, obtaining smoothing factors under different sample labels, and then establishing corresponding Gaussian kernel functions as node activation functions of the network.
Further, the core formula of the bayesian classification model is as follows:
Figure SMS_103
in the method, in the process of the invention,
Figure SMS_104
is a probability density function of the feature vector, +.>
Figure SMS_105
For the occurrence of the situation in the attack detection question studied +.>
Figure SMS_106
I.e. a priori probabilities; />
Figure SMS_107
For the mode belonging to->
Figure SMS_108
The probability density that occurs under class conditions is referred to as class conditional probability density; in the present invention->
Figure SMS_109
Classes include: the signals are real signals and deception signals, the extracted navigation data feature set is substituted into a Bayesian algorithm as initial data, and then the deception signals and the real signals in the navigation signals can be determined according to the combination of input features.
Therefore, the invention takes the difference of other variables caused by the difference of the true satellite navigation signal and the deception satellite navigation signal as breakthrough points, and uses the carrier phase double difference, the signal CNR double difference, the signal power double difference and the angle difference of the signal arrival direction obtained by the rotary single antenna model as characteristics to realize deception interference detection by the machine learning training detection model; compared with the traditional multi-antenna deception jamming detection method, the method only needs a single antenna, and reduces the observation cost; compared with the traditional rotation single-antenna deception jamming detection method, the method introduces the angle difference of the signal arrival direction to perform deception jamming detection, can realize the detection of multi-station deception jamming, utilizes a plurality of characteristics to perform detection, introduces machine learning, and has more excellent detection performance.
Of course, the present invention is capable of other various embodiments and its several details are capable of modification and variation in light of the present invention by one skilled in the art without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (6)

1. A machine learning based fraud detection method, comprising the steps of:
s1: respectively acquiring a real satellite navigation signal and a deception satellite navigation signal as sample data through a rotary single antenna, wherein the deception satellite navigation signal is generated by a satellite navigation signal simulator;
s2: extracting features for deception detection from a real satellite navigation signal and a deception satellite navigation signal respectively, wherein the features comprise carrier phase double difference observed values, CNR normalized value double differences, power normalized value double differences and angle differences of signal arrival directions; meanwhile, labeling tags of real signals or deception signals for the values of each group of characteristics, and dividing sample data into a training set and a testing set according to a set proportion;
s3: respectively inputting each group of features in the training set and labels corresponding to each group of features into a support vector machine model, a neural network model and a Bayesian classification model for training to obtain three prediction models for deception detection;
s4: and respectively predicting each group of features in the test set by adopting three prediction models, so that each group of features obtains three deception detection results, and finally obtaining a final deception detection result by a voting method.
2. The machine learning-based fraud detection method according to claim 1, wherein the method for calculating the carrier phase double difference observation value between any two real navigation satellites or between any two satellite navigation signal simulators is as follows:
step 1: acquiring the variation of carrier phase measurement values of M times of current time and previous time caused by rotation of a rotary single antenna corresponding to each real navigation satellite or each satellite navigation signal simulator
Figure QLYQS_1
：
Figure QLYQS_2
Where i denotes the ith real navigation satellite or the ith satellite navigation signal simulator,
Figure QLYQS_3
for the carrier phase of the real navigation satellite i or satellite navigation signal simulator i at the current time k,/i>
Figure QLYQS_4
For the carrier phase of the real navigation satellite i or satellite navigation signal simulator i at the previous instant k-1, and (2)>
Figure QLYQS_5
The distance change between the real navigation satellite i or the satellite navigation signal simulator i and the origin of the rotary single antenna, which is caused by the movement between the real navigation satellites or between the satellite navigation signal simulators, can be obtained through ephemeris data calculation;
step 2: calculating carrier phase double difference between any two real navigation satellites or between any two satellite navigation signal simulators according to the following formula
Figure QLYQS_6
：
Figure QLYQS_7
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_8
and the variation of the carrier phase measurement value at the kth moment caused by rotation of the rotary single antenna corresponding to the jth real navigation satellite or the jth satellite navigation signal simulator is represented.
3. The machine learning-based fraud detection method as set forth in claim 1, wherein the calculation method of the CNR normalized value double difference between any two real navigation satellites or between any two satellite navigation signal simulators is as follows:
step 1: acquiring carrier-to-noise ratios of M times of the current time and the previous time of each real navigation satellite or each satellite navigation signal simulator
Figure QLYQS_9
Wherein i represents the ith real navigation satellite or the ith satellite navigation signal simulator;
step 2: respectively calculating carrier-to-noise ratios of M moments corresponding to each real navigation satellite or each satellite navigation signal simulator
Figure QLYQS_10
Mean>
Figure QLYQS_11
And standard deviation->
Figure QLYQS_12
；
Step 3: the carrier-to-noise ratios of M moments corresponding to each real navigation satellite or each satellite navigation signal simulator are respectively calculated
Figure QLYQS_13
Normalization is carried out:
Figure QLYQS_14
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_15
the carrier-to-noise ratio of the normalized real navigation satellite i or the satellite navigation signal simulator i at the k moment;
step 4: calculating CNR normalized value single difference of each real navigation satellite or each satellite navigation signal simulator at k time
Figure QLYQS_16
：
Figure QLYQS_17
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_18
the carrier-to-noise ratio of the normalized real navigation satellite i or the satellite navigation signal simulator i at the time k+1;
step 5: obtaining a CNR normalized value double difference between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure QLYQS_19
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_20
CNR normalized value single difference at k moment for real navigation satellite i or satellite navigation signal simulator i, < ->
Figure QLYQS_21
The single difference of the normalized values of CNR at time k for the real navigation satellite j or the satellite navigation signal simulator j,
Figure QLYQS_22
and (3) performing double differences on the normalized values of the CNR between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j.
4. The machine learning-based fraud detection method of claim 1, wherein the calculation method of the power normalization value double difference between any two real navigation satellites or between any two satellite navigation signal simulators is as follows:
step 1: obtaining the current time and M times of each real navigation satellite or each satellite navigation signal simulatorSignal power
Figure QLYQS_23
Wherein i represents the ith real navigation satellite or the ith satellite navigation signal simulator;
step 2: respectively calculating the power of each real navigation satellite or M moments corresponding to each satellite navigation signal simulator
Figure QLYQS_24
Mean>
Figure QLYQS_25
And standard deviation->
Figure QLYQS_26
；
Step 3: respectively using the powers of the real navigation satellites or the satellite navigation signal simulators at M moments
Figure QLYQS_27
Normalization is carried out:
Figure QLYQS_28
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_29
the power of the normalized real navigation satellite i or satellite navigation signal simulator i at the k moment;
step 4: calculating signal power normalization value single difference of each real navigation satellite or each satellite navigation signal simulator at k time
Figure QLYQS_30
：
Figure QLYQS_31
Wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_32
the power of the normalized real navigation satellite i or satellite navigation signal simulator i at the time k+1;
step 5: acquiring signal power normalization value double differences between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure QLYQS_33
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_34
normalized value single difference for signal power of real navigation satellite i or satellite navigation signal simulator i at k moment, < ->
Figure QLYQS_35
Normalized value single difference for signal power of real navigation satellite j or satellite navigation signal simulator j at k moment, < ->
Figure QLYQS_36
And normalizing the double differences of the signal power between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j.
5. The method for detecting fraud based on machine learning according to claim 1, wherein the method for calculating the angle difference of the signal arrival direction is as follows:
step 1: reading ephemeris information of navigation signals, and calculating position coordinates of a real navigation satellite i and a real navigation satellite j or a satellite navigation signal simulator i and a satellite navigation signal simulator j at M times before and at present, wherein the position coordinates at the kth time are expressed as
Figure QLYQS_37
；
Step 2: reading the position information of M times of the current and previous times of the receiver, wherein the position information of the kth time is expressed as
Figure QLYQS_38
；
Step 3: acquiring the angle difference of the signal arrival direction between any two real navigation satellites or between any two satellite navigation signal simulators:
Figure QLYQS_39
wherein, the liquid crystal display device comprises a liquid crystal display device,
Figure QLYQS_40
the angle difference of the signal arrival direction between the real navigation satellite i and the real navigation satellite j or between the satellite navigation signal simulator i and the satellite navigation signal simulator j at the kth time is represented.
6. The machine learning-based fraud detection method of any of claims 1-5, wherein the real satellite navigation signal is obtained by receiving the real navigation signal in an open scene by rotating a single antenna and collecting and recording; the deception satellite navigation signals are generated by a plurality of satellite navigation signal simulators in a closed environment, and navigation signals of different navigation satellites are generated by different satellite navigation signal simulators and are received by a rotary single antenna.
CN202310361098.2A 2023-04-07 2023-04-07 Deception jamming detection method based on machine learning Active CN116087994B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310361098.2A CN116087994B (en) 2023-04-07 2023-04-07 Deception jamming detection method based on machine learning

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310361098.2A CN116087994B (en) 2023-04-07 2023-04-07 Deception jamming detection method based on machine learning

Publications (2)

Publication Number Publication Date
CN116087994A CN116087994A (en) 2023-05-09
CN116087994B true CN116087994B (en) 2023-07-07

Family

ID=86204769

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310361098.2A Active CN116087994B (en) 2023-04-07 2023-04-07 Deception jamming detection method based on machine learning

Country Status (1)

Country Link
CN (1) CN116087994B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116736342A (en) * 2023-05-17 2023-09-12 中山大学·深圳 GNSS spoofing detection method and system based on integrated CNN
CN116660942B (en) * 2023-06-01 2024-01-30 北京交通大学 Train satellite positioning deception jamming detection method and system based on deep learning
CN116774253B (en) * 2023-08-25 2023-10-27 中国人民解放军战略支援部队航天工程大学 Navigation deception jamming detection method based on signal arrival direction angle difference

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102007035680A1 (en) * 2007-07-30 2009-03-12 Lfk-Lenkflugkörpersysteme Gmbh Satellite navigation system e.g. global positioning system, deception system, has satellite signal simulator to transmit signals to satellite receiver, where receiving and transmitting units or simulator is arranged at mobile support system
EP2746813A1 (en) * 2012-12-21 2014-06-25 Astrium GmbH Detection of spoofing of GNSS navigation signals
CN112782728A (en) * 2021-01-26 2021-05-11 中国人民解放军92728部队 Antenna array deception jamming signal detection method based on inertia assistance
CN113376659A (en) * 2021-06-15 2021-09-10 东南大学 GNSS (Global navigation satellite System) generated deception jamming detection method based on BP (Back propagation) neural network
CN115494526A (en) * 2022-08-29 2022-12-20 中山大学 GNSS deception jamming detection method and device, electronic equipment and storage medium
CN115616619A (en) * 2022-10-19 2023-01-17 黄淮学院 GNSS deception jamming detection method based on multiple parameters

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115932900A (en) * 2023-01-10 2023-04-07 黄淮学院 GNSS deception detection method based on machine learning

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE102007035680A1 (en) * 2007-07-30 2009-03-12 Lfk-Lenkflugkörpersysteme Gmbh Satellite navigation system e.g. global positioning system, deception system, has satellite signal simulator to transmit signals to satellite receiver, where receiving and transmitting units or simulator is arranged at mobile support system
EP2746813A1 (en) * 2012-12-21 2014-06-25 Astrium GmbH Detection of spoofing of GNSS navigation signals
CN112782728A (en) * 2021-01-26 2021-05-11 中国人民解放军92728部队 Antenna array deception jamming signal detection method based on inertia assistance
CN113376659A (en) * 2021-06-15 2021-09-10 东南大学 GNSS (Global navigation satellite System) generated deception jamming detection method based on BP (Back propagation) neural network
CN115494526A (en) * 2022-08-29 2022-12-20 中山大学 GNSS deception jamming detection method and device, electronic equipment and storage medium
CN115616619A (en) * 2022-10-19 2023-01-17 黄淮学院 GNSS deception jamming detection method based on multiple parameters

Also Published As

Publication number Publication date
CN116087994A (en) 2023-05-09

Similar Documents

Publication Publication Date Title
CN116087994B (en) Deception jamming detection method based on machine learning
Pei et al. SAR automatic target recognition based on multiview deep learning framework
CN113269073B (en) Ship multi-target tracking method based on YOLO V5 algorithm
US20220094710A1 (en) Detection of cyber attacks targeting avionics systems
Beard et al. Sensor control for multi-target tracking using Cauchy-Schwarz divergence
CN102799900B (en) Target tracking method based on supporting online clustering in detection
CN109752710B (en) Rapid target angle estimation method based on sparse Bayesian learning
CN112149705A (en) Method and system for training classification model, computer equipment and storage medium
Singh et al. Machine learning-assisted anomaly detection in maritime navigation using AIS data
Moschos et al. Deep-sst-eddies: A deep learning framework to detect oceanic eddies in sea surface temperature images
CN115494526A (en) GNSS deception jamming detection method and device, electronic equipment and storage medium
Zha et al. LT-SEI: Long-tailed specific emitter identification based on decoupled representation learning in low-resource scenarios
CN111126504A (en) Multi-source incomplete information fusion image target classification method
Chen et al. A comparative study of deep-learning-based semi-supervised device-free indoor localization
Geragersian et al. Multipath detection from gnss observables using gated recurrent unit
CN114740441A (en) Low-slow small-target radar echo identification method based on small samples
CN112213697B (en) Feature fusion method for radar deception jamming recognition based on Bayesian decision theory
US20210364645A1 (en) Method for gps spoofing detection with gps receivers leveraging inaccuracies of gps spoofing devices and apparatus therefore
Ohno et al. Efficient SOM-based ATR method for SAR imagery with azimuth angular variations
Wielandner et al. Multipath-based SLAM with multiple-measurement data association
CN115616579A (en) Sea surface wind field inversion method based on AWPSO-CNN and HY-2C microwave scatterometer
Kästner et al. A bayesian approach to learning 3d representations of dynamic environments
Jayaweera A novel deep learning gps anti-spoofing system with doa time-series estimation
CN114296067A (en) Pulse Doppler radar low-slow small target identification method based on LSTM model
Hu et al. Statistical analysis of massive AIS trajectories using Gaussian mixture models

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