CN109104257A - A kind of wireless signal detection method and device - Google Patents
A kind of wireless signal detection method and device Download PDFInfo
- Publication number
- CN109104257A CN109104257A CN201810723748.2A CN201810723748A CN109104257A CN 109104257 A CN109104257 A CN 109104257A CN 201810723748 A CN201810723748 A CN 201810723748A CN 109104257 A CN109104257 A CN 109104257A
- Authority
- CN
- China
- Prior art keywords
- spectrum signature
- wireless signal
- type
- target
- sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/309—Measuring or estimating channel quality parameters
- H04B17/318—Received signal strength
- H04B17/327—Received signal code power [RSCP]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/391—Modelling the propagation channel
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Mobile Radio Communication Systems (AREA)
- Monitoring And Testing Of Transmission In General (AREA)
Abstract
The embodiment of the invention provides a kind of wireless signal detection method and devices, are related to cognitive radio technology field, wherein the above method includes: to obtain wireless signal to be detected；Obtain the target frequency bands of spectrum signature to be extracted, and the type of the spectrum signature to be extracted in each target frequency bands is obtained, as the feature type of each target frequency bands, wherein, sample is handled using the feature selecting algorithm based on feature ordering, determines target frequency bands, the feature type of target frequency bands；For each target frequency bands, the spectrum signature that the wireless signal to be detected belongs to the feature type of the target frequency bands in the target frequency bands is extracted；Detect whether the signal to be detected belongs to target type according to extracted spectrum signature.It using the embodiment of the present invention, can be realized in the application scenarios that there is interference wireless signal, to be higher than the wireless signal of the accuracy rate identification target type of the prior art.
Description
Technical field
The present invention relates to cognitive radio technology fields, more particularly to a kind of wireless signal detection method and device.
Background technique
With the development of unmanned plane industry, using more and more extensive, unmanned plane signal detection is increasingly taken seriously.Only
The work of unmanned plane signal detection has been carried out, the follow-up works such as repatriation and the intercepting and capturing of unmanned plane could be preferably carried out.
Traditional wireless signal detection method, including wave detection and energy measuring method.It is general in the prior art to use
Following energy measuring method detects unmanned plane signal: using the window function that can at least cover 3 Frequency points, in nothing to be detected
It is translated from low to high on the power spectrum of line signal, one Frequency point of every translation calculates a window function and covered
The corresponding Signal gradient of Frequency point judge if the power spectrum of wireless signal to be detected meets following three kinds of situations
The wireless signal to be detected is doubtful unmanned plane signal, otherwise, judges that the wireless signal to be detected is not unmanned plane signal:
The first situation: when window progresses into variation along region, Signal gradient changes from small to big；
Second situation: when entire window fully enters variation along region, Signal gradient obtains maximum value, and keeps one
Fixed stationarity；
The third situation: when window leaves variation along region, Signal gradient is from large to small.
Frequency range where this method requires wireless signal to be detected is more pure, without other interference signals.And it is most of
2.4GHz frequency range where unmanned plane signal is common ISM (Industrial Scientific Medical, industrial science
Medical treatment) frequency range, there are a variety of wireless signals, such as Wi-Fi signal, Bluetooth signals in the frequency range.The presence of interference signal, affects
The accuracy rate of this method detection, keeps the accuracy rate of testing result low.
Summary of the invention
The embodiment of the present invention is designed to provide a kind of wireless signal detection method and device, thus realize exist it is dry
It disturbs in the application scenarios of wireless signal, identifies unmanned plane signal to be higher than the accuracy rate of the prior art.Specific technical solution is as follows:
The embodiment of the present invention provides a kind of wireless signal detection method, comprising:
Obtain wireless signal to be detected；
It obtains the target frequency bands of spectrum signature to be extracted, and obtains the kind of the spectrum signature to be extracted in each target frequency bands
Class, the feature type as each target frequency bands；
For each target frequency bands, extracts the wireless signal to be detected and belong to the target frequency bands in the target frequency bands
The spectrum signature of feature type；
Detect whether the signal to be detected belongs to target type according to extracted spectrum signature；
Wherein, target frequency bands, the feature type of target frequency bands are determined in the following manner:
Obtain first sample wireless signal, wherein the first sample wireless signal includes: belong to target type wireless
Signal and the wireless signal for being not belonging to target type；
The spectrum signature for extracting the first sample wireless signal, as first sample spectrum signature；
Obtain the corresponding related coefficient of every cluster spectrum signature in the first sample spectrum signature, wherein every cluster frequency
The spectrum signature that spectrum signature is included, place frequency range is identical and the type of spectrum signature is identical, and every cluster spectrum signature is corresponding
Related coefficient is for characterizing: spectrum signature is in frequency range where the cluster spectrum signature and type is the affiliated type of cluster spectrum signature
Wireless signal belong to the confidence level of target type；
The related coefficient for preferentially selecting characterized confidence level high from related coefficient obtained；
For selected each related coefficient, by frequency range where spectrum signature in the corresponding spectrum signature cluster of related coefficient
As target frequency bands, the type of spectrum signature as the feature type in target frequency bands.
In a kind of implementation of the invention, the spectrum signature for extracting the first sample wireless signal, as the
One sample spectrum signature, comprising:
On the default frequency point in default frequency range, the power spectral density of each first sample wireless signal is extracted respectively；
Based on extracted power spectral density, each first sample wireless signal is calculated separately in the every of the default frequency range
Spectrum signature on a frequency sub-band, as frequency sub-band spectrum signature；
For each frequency sub-band spectrum signature, obtain corresponding with the frequency sub-band spectrum signature by the frequency sub-band spectrum signature
The bivector that one zero flag is formed, wherein the corresponding zero flag of each frequency sub-band spectrum signature are as follows: indicated with a zero form
, spectrum signature include whether the first sample wireless signal of the frequency sub-band spectrum signature belongs to the label of target type；
Bivector obtained is polymerize, the first sample spectrum signature is obtained.
It is described preferentially to select characterized confidence level high from related coefficient obtained in a kind of implementation of the invention
Related coefficient, comprising:
Traversal selection is carried out to each related coefficient obtained in the following way, is determined in related coefficient obtained
The high related coefficient of characterized confidence level:
In related coefficient obtained, the maximum correlation coefficient of unselected mistake is selected, as currently associated coefficient；
The corresponding spectrum signature cluster of the currently associated coefficient is added to traversal feature set；
Using the corresponding markup information of spectrum signature each in the traversal feature set and the traversal feature set as first
The input of binary logistic regression model is based on ten folding cross-validation methods, and selection is so that the first binary logistic regression model
The highest model parameter of accuracy rate of testing result completes the training to the first binary logistic regression model, wherein Mei Yipin
The corresponding markup information of spectrum signature indicates: whether the first sample wireless signal with the spectrum signature is target type wireless communication
Number, the testing result of the first binary logistic regression model is for indicating: spectrum signature is each frequency in the traversal feature set
Whether the wireless signal of spectrum signature belongs to target type；
Judge whether the accuracy rate of testing result corresponding to selected model parameter is greater than the accuracy rate of history testing result；
If being not more than, the corresponding spectrum signature cluster of the currently associated coefficient is deleted from the traversal feature set；
If more than determining that the currently associated coefficient is the high phase of confidence level characterized in related coefficient obtained
Relationship number.
In a kind of implementation of the invention, the feature type of the target frequency bands, at least including following characteristics type
It is a kind of:
The mean value of power spectral density；
The variance of power spectral density；
The Differential Characteristics of power spectral density.
It is described that whether the signal to be detected is detected according to extracted spectrum signature in a kind of implementation of the invention
Belong to target type, comprising:
Extracted spectrum signature is inputted in two classifiers that training obtains in advance, whether obtains the signal to be detected
Belong to the testing result of target type；
Wherein, training obtains two classifier in the following ways:
Obtain the second sample wireless signal, wherein the second sample wireless signal includes: belong to target type wireless
Signal and the wireless signal for being not belonging to target type；
For each target frequency bands, extracts the second sample wireless signal and belong to the target frequency bands in the target frequency bands
Feature type spectrum signature, as the second sample spectra feature；
For every one second sample wireless signal, obtain by belonging to the second sample wireless communication in the second sample spectra feature
Number spectrum signature formed feature vector；
It is denoted as known quantity with feature vector obtained and the corresponding zero standard of each feature vector, with weighting parameter vector
It is unknown quantity with offset parameter, constructs the second binary logistic regression model, wherein the corresponding zero flag of each feature vector
Are as follows: whether the corresponding second sample wireless signal of feature vector indicated with a zero form, each belongs to the label of target type；
Obtain weighting parameter when the second binary logistic regression model makes maximum likelihood algorithm obtain maximum
Value and the offset parameter value；
By the value of offset parameter described in the weighting parameter vector sum be respectively obtained value when the second binomial patrol
It collects regression model and is determined as two classifier.
It is described with feature vector obtained and each feature vector corresponding 1 in a kind of implementation of the invention
Labeled as known quantity, using weighting parameter vector sum offset parameter as unknown quantity, the second binary logistic regression model is constructed, comprising:
Using following formula, the second binary logistic regression model is established:
Wherein, xiFor ith feature vector, yiFor the corresponding zero flag of ith feature vector, ω is weight ginseng
Number vector, b are the offset parameter.
In a kind of implementation of the invention, acquisition the second binary logistic regression model calculates maximum likelihood
Method obtains the value of the value of the weighting parameter and the offset parameter when maximum, comprising:
The mode that the maximum of default log-likelihood function is calculated by using gradient descent method, obtains second binomial
The value of the weighting parameter and the offset parameter takes when Logic Regression Models make maximum likelihood algorithm obtain maximum
Value；
Wherein, the default log-likelihood function are as follows:
Wherein, L (ω) is log-likelihood function, xiFor ith feature vector, yiFor ith feature vector corresponding 1
Label, N are the quantity of feature vector, and ω is the weighting parameter vector, and b is the offset parameter.
In a kind of implementation of the invention, the value by offset parameter described in the weighting parameter vector sum is distinguished
By acquisition value when the second binary logistic regression model be determined as two classifier, comprising:
Determine that by the classifier that following expression indicates be two classifier:
Wherein,For the value of the weighting parameter vector,For the value of the offset parameter, x is to input described two points
The vector of class device being made of spectrum signature, y are a zero flag of two classifier output, the one of the two classifiers output
Zero flag are as follows: whether wireless signal represented by vector indicated with a zero form, input two classifier belongs to target
The label of type.
The embodiment of the present invention also provides a kind of wireless signal detection device, comprising:
Signal acquisition module, for obtaining wireless signal to be detected；
Frequency range obtains module, for obtaining the target frequency bands of spectrum signature to be extracted, and obtains in each target frequency bands
The type of spectrum signature to be extracted, the feature type as each target frequency bands；
Characteristic extracting module extracts the wireless signal to be detected in the target frequency bands for being directed to each target frequency bands
On belong to the target frequency bands feature type spectrum signature；
Signal detection module, for detecting whether the signal to be detected belongs to target class according to extracted spectrum signature
Type；
Wherein, the frequency range obtains module, comprising:
Sample signal acquisition submodule, for obtaining first sample wireless signal, wherein the first sample wireless signal
The wireless signal for including: the wireless signal for belonging to target type and being not belonging to target type；
Sample characteristics extracting sub-module, for extracting the spectrum signature of the first sample wireless signal, as the first sample
This spectrum signature；
Related coefficient obtains submodule, corresponding for obtaining every cluster spectrum signature in the first sample spectrum signature
Related coefficient, wherein the spectrum signature that every cluster spectrum signature is included, place frequency range is identical and the type phase of spectrum signature
Together, the corresponding related coefficient of every cluster spectrum signature is for characterizing: spectrum signature is in frequency range and type where the cluster spectrum signature
Belong to the confidence level of target type for the wireless signal of the affiliated type of cluster spectrum signature；
The preferred submodule of related coefficient, the phase for preferentially selecting characterized confidence level high from related coefficient obtained
Relationship number；
Target frequency bands determine submodule, for being directed to selected each related coefficient, by the corresponding frequency spectrum of related coefficient
Frequency range where spectrum signature is as target frequency bands, the type of spectrum signature as the feature type in target frequency bands in feature cluster.
The embodiment of the present invention also provides a kind of electronic equipment, including processor, communication interface, memory and communication bus,
Wherein, processor, communication interface, memory complete mutual communication by communication bus；
Memory, for storing computer program；
Processor when for executing the program stored on memory, realizes any of the above-described wireless signal detection
The step of method.
At the another aspect that the present invention is implemented, the embodiment of the invention also provides a kind of computer readable storage medium, institutes
It states and is stored with instruction in computer readable storage medium, when run on a computer, so that computer execution is any of the above-described
The step of described wireless signal detection method.
At the another aspect that the present invention is implemented, the embodiment of the invention also provides a kind of, and the computer program comprising instruction is produced
Product, when run on a computer, so that computer executes any of the above-described wireless signal detection method.
As seen from the above, in scheme provided in an embodiment of the present invention, in training, to the wireless signal for belonging to target type
With the wireless signal for being not belonging to target type, using the feature selecting algorithm based on feature ordering, screening is obtained for detecting mesh
Mark the reliable spectrum signature of type wireless signal.When detecting, reliable spectrum signature is only extracted based on training result for examining
It surveys, the influence that can be excluded the interference signal to detection, to realize in the application scenarios that there is interference wireless signal, to be higher than
The wireless signal of the accuracy rate identification target type of the prior art.Certainly, it implements any of the products of the present invention or method and different
It is fixed to need while reaching all the above advantage.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is the flow chart of wireless signal detection method provided in an embodiment of the present invention；
Fig. 2 is the flow chart that a kind of information provided in an embodiment of the present invention determines method；
Fig. 3 is a kind of flow chart for selecting related coefficient method provided in an embodiment of the present invention；
Fig. 4 is a kind of flow chart of two classifier trainings method provided in an embodiment of the present invention；
Fig. 5 is the structural schematic diagram of wireless signal detection device provided in an embodiment of the present invention；
Fig. 6 is the structural schematic diagram provided in an embodiment of the present invention for obtaining module；
Fig. 7 is the structural schematic diagram of a kind of electronic equipment provided in an embodiment of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The embodiment of the invention provides a kind of wireless signal detection methods, below by specific embodiment, to of the invention real
The wireless signal detection method for applying example offer is described in detail.
Referring to Fig. 1, Fig. 1 is a kind of flow chart of wireless signal detection method provided in an embodiment of the present invention, including as follows
Step:
Step S101, wireless signal to be detected is obtained.
In the embodiment of the present invention, the wireless signal data built including equipment such as frequency spectrograph, omnidirectional antennas can be used and adopt
Collect platform, acquires wireless signal to be detected, can also be obtained by wired mode by the collected nothing to be detected of equipment
Line signal.
Step S102, the target frequency bands of spectrum signature to be extracted are obtained, and obtain the frequency to be extracted in each target frequency bands
The type of spectrum signature, the feature type as each target frequency bands.
Specifically, sample can be handled using the feature selecting algorithm based on feature ordering, target frequency bands, target frequency are determined
The feature type of section.
In the embodiment of the present invention, for the influence to exclude the interference signal, it may relate in detection target type wireless signal
Frequency range on, only select specific multiple frequency ranges and detected, using these frequency ranges as target frequency bands.In each target frequency bands
On, it only selects the diverse spectrum signature of one or more spectrum signatures and is detected.The type of spectrum signature may include
Power features, power Differential Characteristics, energy feature, waveform variation characteristic etc..
Step S103, each target frequency bands are directed to, wireless signal to be detected is extracted and belongs to the target in the target frequency bands
The spectrum signature of the feature type of frequency range.
Each target frequency bands may have the spectrum signature of one or more kinds of types, and target frequency bands may have category
In multiple spectrum signatures of same type of spectrum signature, in order to prevent spectrum signature quantity excessively cause detection time-consuming and
In one embodiment of the present of invention, it is special to extract frequency spectrum of the wireless signal to be detected in a target frequency bands for operation cost increase
When sign, can the spectrum signature to each type only extract a spectrum signature.
Step S104, detect whether signal to be detected belongs to target type according to extracted spectrum signature.
The wireless signal of target type can be unmanned plane signal, be also possible to other kinds of wireless signal.
In the embodiment of the present invention, extracted spectrum signature can be used and be compared with the threshold value being manually arranged, to sentence
Whether the signal to be detected that breaks belongs to target type, extracted spectrum signature can also be inputted two classification that training obtains in advance
In device, the testing result whether signal to be detected belongs to target type is obtained.Wherein, two classifiers are that input data is divided into two
The device of class.
In step s 102, sample is handled using the feature selecting algorithm based on feature ordering, determines target frequency bands, target
The method of the feature type of frequency range, specifically, referring to fig. 2.Fig. 2 provides a kind of information and determines method, can be with by this method
It determines target frequency bands, the feature type of target frequency bands, includes the following steps:
Step 201 obtains first sample wireless signal.
First sample wireless signal includes: the wireless signal for belonging to target type and the wireless communication that is not belonging to target type
Number.
When acquisition is not belonging to the wireless signal of target type, actually detected application scenarios can be simulated, acquisition is wireless
Environmental signal.It, can be in the channel without the wireless model of target type of script when acquisition belongs to the wireless signal of target type
In environment, communicated using target type wireless signal, then acquire wireless signal in this context.Specific acquisition method,
It is identical as wireless signal to be detected is acquired in step S101.
Step 202, the spectrum signature for extracting first sample wireless signal, as first sample spectrum signature.
It is trained for convenience, each of first sample spectrum signature spectrum signature can be marked, be slaves to
Target type is still not belonging to what extraction in the wireless signal of target type obtained.Also it is used as first sample frequency spectrum special the label
A part of sign.
Step 203 obtains the corresponding related coefficient of every cluster spectrum signature in first sample spectrum signature.
Each of same spectrum signature cluster spectrum signature, place frequency range is identical, and the type of spectrum signature is identical.Together
Each of one spectrum signature cluster spectrum signature is all that the different wireless signal from first sample wireless signal extracts.
The corresponding related coefficient of spectrum signature cluster is the spectrum signature in the spectrum signature cluster, obtains frequency spectrum spy with extraction
Whether the wireless signal of sign belongs to the related coefficient of a zero mark note of target type.The corresponding related coefficient of spectrum signature cluster is used for
Characterization: spectrum signature is in the wireless signal category that frequency range where the cluster spectrum signature and type are the affiliated type of cluster spectrum signature
In the confidence level of target type.
In one embodiment of the present of invention, can be characterized with Pearson correlation coefficients: spectrum signature is special in the cluster frequency spectrum
Frequency range where sign and type are that the wireless signal of the affiliated type of cluster spectrum signature belongs to the confidence level of target type.
One spectrum signature cluster can be indicated using following set:
Uk={ (x1, k, y1), (x2, k, y2) ..., (xI, k, yi),…,(xn,k,yn)}
Wherein, UkFor k-th of spectrum signature cluster, xi,kIt is k-th of i-th of wireless signal in first sample wireless signal
Spectrum signature, yiTo indicate whether i-th of wireless signal belongs to target type in first sample wireless signal with a zero form
Label, n are the quantity of wireless signal in first sample wireless signal.
If i-th of wireless signal belongs to target type, yi=1, if i-th of wireless signal is not belonging to target class
Type, then yi=0.
The corresponding Pearson's phase of every cluster spectrum signature in first sample spectrum signature can be calculated using following formula
Relationship number:
Wherein, ρkFor the Pearson correlation coefficients of k-th of spectrum signature cluster, xkIt is special for the frequency spectrum of k-th of spectrum signature cluster
Sign, y are to indicate whether the wireless signal in first sample wireless signal belongs to the label of target type with a zero form,For xk
Mean value,For the mean value of y,For xkStandard deviation, σyFor the standard deviation of y.
Step 204, the related coefficient for preferentially selecting characterized confidence level high from related coefficient obtained.
In the embodiment of the present invention, the related coefficient of the highest predetermined amount of related coefficient can choose, can also select
Select whole related coefficients that related coefficient numerical value is greater than preset threshold value.
If do not selected related coefficient, whole spectrum signatures are all used to train, will increase the trained time at
This and operation cost, and in a part of spectrum signature interference signal influence it is bigger, cause its related coefficient to decline, reject
Such spectrum signature helps to promote Detection accuracy.
Step 205 is directed to selected each related coefficient, by spectrum signature in the corresponding spectrum signature cluster of related coefficient
Place frequency range is as target frequency bands, the type of spectrum signature as the feature type in target frequency bands.
As seen from the above, in scheme provided in an embodiment of the present invention, in training, to the wireless signal for belonging to target type
With the wireless signal for being not belonging to target type, using the feature selecting algorithm based on feature ordering, screening is obtained for detecting mesh
Mark the reliable spectrum signature of type wireless signal.When detecting, reliable spectrum signature is only extracted based on training result for examining
It surveys, the influence that can be excluded the interference signal to detection, to realize in the application scenarios that there is interference wireless signal, to be higher than
The wireless signal of the accuracy rate identification target type of the prior art.
In a kind of implementation of the invention, the spectrum signature of first sample wireless signal is extracted in above-mentioned steps 202, is made
For first sample spectrum signature, can be realized using following steps A1- step A4:
Step A1: on the default frequency point in default frequency range, the power spectrum of each first sample wireless signal is extracted respectively
Density.
In the embodiment of the present invention, default frequency range is the whole frequency ranges that detects target type wireless signal and may relate to, generally
Range can be bigger, including the frequency range to come in handy to detection target type wireless signal is all included.
Power spectral density is the power that per unit frequency wave carries, for indicating the power of signal how with frequency distribution.
Use the calculated result based on power spectral density as spectrum signature, function is belonged to the detection of wireless signal based on this implementation
The detection of rate feature.
Step A2: it is based on extracted power spectral density, calculates separately each first sample wireless signal in default frequency range
Each frequency sub-band on spectrum signature, as frequency sub-band spectrum signature.
In the embodiment of the present invention, default frequency range can uniformly be divided, obtain the equal frequency sub-band of multiple width.Sub- frequency
Section spectrum signature can be calculated by statistical formula, the power spectral density based on multiple default frequency points in frequency sub-band.It is more
The different descriptive statistic amounts of the power spectral density of a default frequency point, i.e., different types of frequency sub-band spectrum signature.For every
One frequency sub-band, the frequency sub-band spectrum signature of each type only one.
Step A3: being directed to each frequency sub-band spectrum signature, obtains special by the frequency sub-band spectrum signature and the frequency sub-band frequency spectrum
Levy the bivector that a corresponding zero flag is formed.
The corresponding zero flag of each frequency sub-band spectrum signature are as follows: indicated with a zero form, spectrum signature includes the son
Whether the first sample wireless signal of frequency range spectrum signature belongs to the label of target type.The value of one zero flag is a period of time,
Indicate that spectrum signature includes that the first sample wireless signal of the frequency sub-band spectrum signature belongs to target type, which takes
When value is zero, indicate that spectrum signature includes that the first sample wireless signal of the frequency sub-band spectrum signature is not belonging to target type.
In the embodiment of the present invention, which can be indicated are as follows: (xi,k,yi), wherein xi,kFor first sample wireless communication
K-th of spectrum signature of i-th of wireless signal, y in numberiTo indicate i-th of nothing in first sample wireless signal with a zero form
Whether line signal belongs to the label of target type.
Step A4: polymerizeing bivector obtained, obtains first sample spectrum signature.
In the embodiment of the present invention, first sample spectrum signature can be indicated using following bivector set: D=
{(x1,1,y1),(x1,2,y1),…,(x1,3s,y1),(x2,1,y2),…,(xi,k,yi),…,(xn,3s,yn)}
Wherein, D is first sample spectrum signature.
In this implementation, when extracting training data from sample, whether first sample wireless signal is belonged into target
Type is combined into bivector with spectral feature data, at the sample of the feature selecting algorithm based on feature ordering later
When reason, be used directly for calculate related coefficient, be conducive to training in data processing.
In a kind of implementation of the invention, the feature type of target frequency bands, is included the following three types in above-mentioned steps S102
At least one of feature type: the mean value of power spectral density；The variance of power spectral density；The Differential Characteristics of power spectral density.
So, above-mentioned steps A2 calculates descriptive statistic using the power spectral density of multiple default frequency points in frequency sub-band
It measures, may include:
Using following formula, the mean value of the power spectral density on each frequency sub-band is calculated:
Wherein,For the mean value of the power spectral density of m-th of frequency sub-band, s is the frequency sub-band that default frequency range is divided
Quantity, a is the quantity that frequency point is preset on m-th frequency sub-band, PjFor the power spectral density of j-th of frequency point on m-th of frequency sub-band.
Using following formula, the variance of the power spectral density on each frequency sub-band is calculated:
Wherein,For the variance of the power spectral density of m-th of frequency sub-band.
Using following formula, the Differential Characteristics of the power spectral density on each frequency sub-band are calculated:
Wherein, Δ PmFor the Differential Characteristics of the power spectral density of m-th of frequency sub-band, Pj+1For jth+1 on m-th of frequency sub-band
The power spectral density of a frequency point.
First sample spectrum signature in above-mentioned steps A4 can be indicated using following formula:
Wherein,For power spectral density of i-th of wireless signal in s cross-talk frequency range in first sample wireless signal
Mean value, σ2 i,sFor the side of power spectral density of i-th of wireless signal in s cross-talk frequency range in first sample wireless signal
Difference, Δ Pi,sIt is special for the difference of power spectral density of i-th of wireless signal in s cross-talk frequency range in first sample wireless signal
Sign.
Default frequency range is divided evenly into s frequency sub-band, there is mean value, the power spectrum of power spectral density on each frequency sub-band
The variance of degree and the Differential Characteristics of power spectral density, the spectrum signature of 3 kinds of types, so in first sample wireless signal
The same wireless signal shares 3s spectrum signature.So the spectrum signature cluster U in step 203k, also share 3s.
In this implementation, unmanned plane signal is detected using the spectrum power feature and power Differential Characteristics of signal
With identification.Compared with the energy feature of signal, waveform variation characteristic, the feature that this method uses realizes that simple, complexity is low, can
To reflect target type signal well in the Energy distribution of frequency domain and jump situation, Detection accuracy is higher.In addition, target frequency
The feature type of section is the statistical result that multiple power spectral densities are calculated by statistical formula in a frequency range, reduces list
The measurement data of a frequency point reduces a possibility that causing detection to malfunction due to measurement error to the influence degree of testing result.
Fig. 3 show a kind of flow chart for selecting related coefficient method provided in an embodiment of the present invention, specific aforementioned step
Rapid 204 when preferentially selection characterizes execution degree high related coefficient from related coefficient obtained, can be in accordance with the following steps
301- step 308 carries out traversal selection to each single item related coefficient obtained, and then determines institute in related coefficient obtained
Characterize the high related coefficient of execution degree.
Specifically, step 301- step 308 is as follows:
Step 301, in related coefficient obtained, the maximum correlation coefficient of unselected mistake is selected, as current phase
Relationship number.
Particularly, if the numerical value of multiple related coefficients is equal and maximum, one of work can be chosen at random
For currently associated coefficient.If only a surplus related coefficient is not selected, then the phase in related coefficient obtained
Relationship number can be regarded as the maximum correlation coefficient of unselected mistake, and be taken as currently associated coefficient.
The corresponding spectrum signature cluster of currently associated coefficient is added to traversal feature set by step 302.
In the embodiment of the present invention, traversal feature set is the union of spectrum signature cluster, can be indicated using following set:
V=Uα∪Uβ∪…∪Uk
Wherein, V is traversal feature set, UαFor the maximum spectrum signature cluster of Pearson correlation coefficients, UβFor by traversal selection
The spectrum signature cluster of traversal feature set, U are stayed in afterwardskThe corresponding frequency of currently associated coefficient of traversal feature set is added to for this step
Spectrum signature cluster.
Before the maximum spectrum signature cluster of Pearson correlation coefficients to be added to traversal feature set, traversal feature set is sky
Collection.
Step 303 will traverse each spectrum signature zero flag corresponding with the spectrum signature in feature set as the
The input of one binary logistic regression model, is based on ten folding cross-validation methods, and selection makes the detection of the first binary logistic regression model
As a result the highest model parameter of accuracy rate.
Binary logistic regression model is the binomial Discrete Choice Model of logic-based distribution, is most widely used binomial mould
Type, model parameter generally use Maximum Likelihood Estimation to calculate value.
Ten folding cross-validation methods are a kind of method of testing algorithm accuracy, data set are divided into ten parts, in turn by it
In nine parts be used as training data, portion be used as test data, tested.
In the embodiment of the present invention, the input data of the first binary logistic regression model can be divided into ten parts, it will be therein
Nine parts are used as training data, and the first binary logistic regression model of training obtains the value of model parameter, by remaining a conduct
Test data tests the value of obtained model parameter.Specific test method are as follows: the spectrum signature part of test data is defeated
Enter into the first binary logistic regression model, the model parameter of the first binary logistic regression model is joined using obtained model
Several values, one or zero result exported.Test number by the one of output or zero result and corresponding thereto
One or zero label compares in, calculates the frequency that the two is consistent, the accuracy rate as testing result.
It regard nine parts in ten parts of data as training data in turn, portion is used as test data, calculates the standard of testing result
True rate.The accuracy rate of highest testing result is selected in the accuracy rate for ten testing results being calculated, and selects to make to examine
Survey the highest model parameter of accuracy rate of result.
Step 304 judges whether the accuracy rate of testing result corresponding to selected model parameter is greater than history testing result
Accuracy rate, if it is not greater, 305 are entered step, if it does, entering step 306.
The accuracy rate of history testing result, during traversing currently associated coefficient, to generate since traversal
All accuracys rate.If the accuracy rate of testing result corresponding to selected model parameter is greater than the accurate of history testing result
Rate, then, which is since traversal, to traversing highest accuracy rate during currently associated coefficient.
Particularly, when executing this step for the first time, it is special to traverse the only maximum frequency spectrum of Pearson correlation coefficients in feature set
Levy cluster, not before history testing result accuracy rate.At this moment the accuracy rate that history testing result can be regarded as is zero, in this way
When executing this step for the first time, it is judged as greater than the accuracy rate of history testing result, enters step 306.
Step 305 deletes the corresponding spectrum signature cluster of currently associated coefficient from traversal feature set, and enters step 307.
Step 306 determines that frequency range where spectrum signature is a target in the corresponding spectrum signature cluster of currently associated coefficient
Frequency range, the feature type that the type of spectrum signature is the target frequency bands.
In the embodiment of the present invention, can be realized by recording the number of currently associated coefficient determine a target frequency bands,
The type of spectrum signature is a feature type of the target frequency bands, if currently associated coefficient is ρk, then record positive integer
k。
Step 307 judges whether that all related coefficients are all selected, and if not being all selected, enters step
301, if be all selected, enter step 308.
Step 308 determines target frequency bands, the feature type of target frequency bands.
In the embodiment of the present invention, each currently associated coefficient number can be determined by recording step 306, so that it is determined that
Target frequency bands, the feature type of target frequency bands.The value of currently associated coefficient number k comes from:
For the numerical value of each number being recorded, a corresponding spectrum signature, the spectrum signature pair can be found
The frequency range answered is a target frequency bands, and the type of the spectrum signature is a feature type of the target frequency bands.
In this implementation, from high to low according to related coefficient, spectrum signature cluster is added sequentially to traversal training set, is tested
Detection accuracy after card addition, retains the spectrum signature cluster for improving accuracy after being added, accuracy cannot be made by removing
The spectrum signature cluster of raising, the corresponding related coefficient of spectrum signature cluster stayed are exactly that preferential the characterized confidence level of selection is high
Related coefficient.The high spectrum signature cluster of related coefficient is grouped together, Detection accuracy may not be high.The optimization of this implementation
The selection method of related coefficient, the related coefficient of not choosing not instead of simply related coefficient high pass through spectrum signature and combine
To the effect after together, find can by being added to the spectrum signature cluster that Detection accuracy be made to be promoted in traversal training set, thus
It is preferred that the related coefficient that characterized confidence level is high.
Specifically, abovementioned steps S104 detects whether signal to be detected belongs to target type according to extracted spectrum signature
When, extracted spectrum signature can be inputted in two classifiers that training obtains in advance, obtain whether signal to be detected belongs to
The testing result of target type.
Wherein, two classifiers can be trained by the way of being illustrated in fig. 4 shown below and be obtained.
Fig. 4 show a kind of flow chart of two classifier trainings method provided in an embodiment of the present invention, can specifically include
Following steps:
Step 401 obtains the second sample wireless signal.
Second sample wireless signal includes: the wireless signal for belonging to target type and the wireless communication that is not belonging to target type
Number.
In the embodiment of the present invention, the second sample wireless signal can be identical with first sample signal, can also be with the first sample
This signal is different.
Step 402 is directed to each target frequency bands, extracts the second sample wireless signal and belongs to the target in the target frequency bands
The spectrum signature of the feature type of frequency range, as the second sample spectra feature.
Step 403 is directed to every one second sample wireless signal, obtains by belonging to second sample in the second sample spectra feature
The feature vector that the spectrum signature of this wireless signal is formed.
In the embodiment of the present invention, feature vector obtained can indicate in the following way:
Wherein, xiFor i-th of feature vector obtained.
If N be the second sample wireless signal in wireless signal quantity, this step can obtain altogether N number of feature to
Amount.
Step 404, the second binary logistic regression model of building.
In the embodiment of the present invention, it can be denoted as with feature vector obtained and the corresponding zero standard of each feature vector
The amount of knowing, using following formula, establishes the second binary logistic regression model using weighting parameter vector sum offset parameter as unknown quantity:
Wherein, xiFor ith feature vector, yiFor the corresponding zero flag of ith feature vector, ω be weighting parameter to
Amount, b is offset parameter.
Step 405 obtains weighting parameter when the second binary logistic regression model makes maximum likelihood algorithm obtain maximum
Value and offset parameter value.
In the embodiment of the present invention, the side of the maximum of default log-likelihood function can be calculated by using gradient descent method
Formula obtains the value of weighting parameter and biasing ginseng when the second binary logistic regression model makes maximum likelihood algorithm obtain maximum
Several values.
Wherein, log-likelihood function is preset are as follows:
Wherein, L (ω) is log-likelihood function, xiFor ith feature vector, yiFor ith feature vector corresponding 1
Label, N are the quantity of feature vector, and ω is weighting parameter vector, and b is offset parameter.
Step 406, by the value of weighting parameter vector sum offset parameter be respectively obtained value when the second binomial patrol
It collects regression model and is determined as two classifiers.
In the embodiment of the present invention, it can determine that by the classifier that following expression indicates be two classifiers:
Wherein,For the value of weighting parameter vector,For the value of offset parameter, x is two classifiers of input by frequency spectrum
The vector that feature is constituted, y are a zero flag of two classifiers output, a zero flag of two classifiers output are as follows: with a zero form
Whether wireless signal represented by vectors indicate, two classifiers of input belongs to the label of target type.
X is the vector being made of spectrum signature for inputting two classifiers, each element of the vector is, for each mesh
Frequency range is marked, the spectrum signature that wireless signal belongs to the feature type of the target frequency bands in the target frequency bands is extracted.
In order to detect whether signal to be detected belongs to target type according to extracted spectrum signature, the embodiment of the present invention is mentioned
A kind of training method of two classifiers has been supplied, extracted spectrum signature can have been inputted into two classifiers that training obtains in advance
In, obtain the testing result whether signal to be detected belongs to target type.Using two classifier provided in an embodiment of the present invention
When training method carrys out detected wireless signals, the wireless signal of target type and the wireless signal of target type is not belonging to using belonging to
Training binary logistic regression model, obtains two classifiers.Spectrum signature is inputted into two detection of classifier, than directly using artificial
The threshold value of setting judges there is higher Detection accuracy.
Based on the same inventive concept, the wireless signal detection method provided according to that above embodiment of the present invention, correspondingly, this
Inventive embodiments additionally provide a kind of wireless signal detection device, and structural schematic diagram is as shown in figure 5, specifically include:
Signal acquisition module 501, for obtaining wireless signal to be detected；
Frequency range obtains module 502, for obtaining the target frequency bands of spectrum signature to be extracted, and obtains in each target frequency bands
The type of upper spectrum signature to be extracted, the feature type as each target frequency bands；
Characteristic extracting module 503 extracts the wireless signal to be detected in target frequency for being directed to each target frequency bands
Belong to the spectrum signature of the feature type of the target frequency bands in section；
Signal detection module 504, for detecting whether the signal to be detected belongs to mesh according to extracted spectrum signature
Mark type；
Wherein, the frequency range obtains module 502, as shown in Figure 6, comprising:
Sample signal acquisition submodule 601, for obtaining first sample wireless signal, wherein the first sample is wireless
Signal includes: the wireless signal for belonging to target type and the wireless signal that is not belonging to target type；
Sample characteristics extracting sub-module 602, for extracting the spectrum signature of the first sample wireless signal, as first
Sample spectra feature；
Related coefficient obtains submodule 603, for obtaining every cluster spectrum signature pair in the first sample spectrum signature
The related coefficient answered, wherein the spectrum signature that every cluster spectrum signature is included, place frequency range is identical and the kind of spectrum signature
Class is identical, and the corresponding related coefficient of every cluster spectrum signature is for characterizing: spectrum signature in frequency range where the cluster spectrum signature and
Type is that the wireless signal of the affiliated type of cluster spectrum signature belongs to the confidence level of target type；
The preferred submodule 604 of related coefficient, for preferentially selecting characterized confidence level high from related coefficient obtained
Related coefficient；
Target frequency bands determine submodule 605, for being directed to selected each related coefficient, by the corresponding frequency of related coefficient
Frequency range where spectrum signature is as target frequency bands, the type of spectrum signature as the characteristic species in target frequency bands in spectrum signature cluster
Class.
Using wireless signal detection device detected wireless signals provided in an embodiment of the present invention, in training, to belonging to mesh
It marks the wireless signal of type and is not belonging to the wireless signal of target type, using the feature selecting algorithm based on feature ordering, sieve
Choosing is obtained for detecting the reliable spectrum signature of target type wireless signal.When detecting, it is only extracted reliably based on training result
Spectrum signature for detecting, there is interference wireless signal to realize in the influence that can be excluded the interference signal to detection
In application scenarios, to be higher than the wireless signal of the accuracy rate identification target type of the prior art.
Based on the same inventive concept, the wireless signal detection method provided according to that above embodiment of the present invention, correspondingly, this
Inventive embodiments additionally provide a kind of electronic equipment, as shown in fig. 7, comprises processor 701, communication interface 702, memory 703
With communication bus 704, wherein processor 701, communication interface 702, memory 703 are completed mutual by communication bus 704
Communication,
Memory 703, for storing computer program；
Processor 701 when for executing the program stored on memory 703, is realized any wireless in above-described embodiment
The step of signal detecting method.
The communication bus that above-mentioned electronic equipment is mentioned can be Peripheral Component Interconnect standard (Peripheral Component
Interconnect, PCI) bus or expanding the industrial standard structure (Extended Industry Standard
Architecture, EISA) bus etc..The communication bus can be divided into address bus, data/address bus, control bus etc..For just
It is only indicated with a thick line in expression, figure, it is not intended that an only bus or a type of bus.
Communication interface is for the communication between above-mentioned electronic equipment and other equipment.
Memory may include random access memory (Random Access Memory, RAM), also may include non-easy
The property lost memory (Non-Volatile Memory, NVM), for example, at least a magnetic disk storage.Optionally, memory may be used also
To be storage device that at least one is located remotely from aforementioned processor.
Above-mentioned processor can be general processor, including central processing unit (Central Processing Unit,
CPU), network processing unit (Network Processor, NP) etc.；It can also be digital signal processor (Digital Signal
Processing, DSP), it is specific integrated circuit (Application Specific Integrated Circuit, ASIC), existing
It is field programmable gate array (Field-Programmable Gate Array, FPGA) or other programmable logic device, discrete
Door or transistor logic, discrete hardware components.
Electronic equipment detected wireless signals are detected using wireless signal provided in an embodiment of the present invention, in training, to category
Wireless signal in target type and the wireless signal for being not belonging to target type, are calculated using the feature selecting based on feature ordering
Method, screening are obtained for detecting the reliable spectrum signature of target type wireless signal.When detecting, it is only extracted based on training result
Reliable spectrum signature is for detecting, the influence that can be excluded the interference signal to detection, there is interference wireless communication to realize
Number application scenarios in, be higher than the prior art accuracy rate identification target type wireless signal.
In another embodiment provided by the invention, a kind of computer readable storage medium is additionally provided, which can
It reads to be stored with instruction in storage medium, when run on a computer, so that computer executes any nothing in above-described embodiment
The step of line signal detecting method.
In another embodiment provided by the invention, a kind of computer program product comprising instruction is additionally provided, when it
When running on computers, so that computer executes any wireless signal detection method in above-described embodiment.
It should be noted that, in this document, relational terms such as first and second and the like are used merely to a reality
Body or operation are distinguished with another entity or operation, are deposited without necessarily requiring or implying between these entities or operation
In any actual relationship or order or sequence.Moreover, the terms "include", "comprise" or its any other variant are intended to
Non-exclusive inclusion, so that the process, method, article or equipment including a series of elements is not only wanted including those
Element, but also including other elements that are not explicitly listed, or further include for this process, method, article or equipment
Intrinsic element.In the absence of more restrictions, the element limited by sentence "including a ...", it is not excluded that
There is also other identical elements in process, method, article or equipment including the element.
Each embodiment in this specification is all made of relevant mode and describes, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device,
For electronic equipment, computer readable storage medium and computer program product embodiments, since it is substantially similar to method reality
Example is applied, so being described relatively simple, the relevent part can refer to the partial explaination of embodiments of method.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the scope of the present invention.It is all
Any modification, equivalent replacement, improvement and so within the spirit and principles in the present invention, are all contained in protection scope of the present invention
It is interior.
Claims (10)
1. a kind of wireless signal detection method characterized by comprising
Obtain wireless signal to be detected；
It obtains the target frequency bands of spectrum signature to be extracted, and obtains the type of the spectrum signature to be extracted in each target frequency bands,
Feature type as each target frequency bands；
For each target frequency bands, the feature that the wireless signal to be detected belongs to the target frequency bands in the target frequency bands is extracted
The spectrum signature of type；
Detect whether the signal to be detected belongs to target type according to extracted spectrum signature；
Wherein, target frequency bands, the feature type of target frequency bands are determined in the following manner:
Obtain first sample wireless signal, wherein the first sample wireless signal includes: the wireless signal for belonging to target type
With the wireless signal for being not belonging to target type；
The spectrum signature for extracting the first sample wireless signal, as first sample spectrum signature；
Obtain the corresponding related coefficient of every cluster spectrum signature in the first sample spectrum signature, wherein every cluster frequency spectrum is special
Levy included spectrum signature, place frequency range is identical and the type of spectrum signature is identical, the corresponding correlation of every cluster spectrum signature
Coefficient is used to characterize: spectrum signature is in the nothing that frequency range where the cluster spectrum signature and type are the affiliated type of cluster spectrum signature
Line signal belongs to the confidence level of target type；
The related coefficient for preferentially selecting characterized confidence level high from related coefficient obtained；
For selected each related coefficient, using frequency range where spectrum signature in the corresponding spectrum signature cluster of related coefficient as
Target frequency bands, the type of spectrum signature are as the feature type in target frequency bands.
2. the method according to claim 1, wherein the frequency spectrum for extracting the first sample wireless signal is special
Sign, as first sample spectrum signature, comprising:
On the default frequency point in default frequency range, the power spectral density of each first sample wireless signal is extracted respectively；
Based on extracted power spectral density, each first sample wireless signal is calculated separately in every height of the default frequency range
Spectrum signature in frequency range, as frequency sub-band spectrum signature；
For each frequency sub-band spectrum signature, obtain by the frequency sub-band spectrum signature and the frequency sub-band spectrum signature corresponding 1
Mark the bivector formed, wherein the corresponding zero flag of each frequency sub-band spectrum signature are as follows: being indicated with a zero form,
Spectrum signature includes whether the first sample wireless signal of the frequency sub-band spectrum signature belongs to the label of target type；
Bivector obtained is polymerize, the first sample spectrum signature is obtained.
3. the method according to claim 1, wherein described preferentially select institute's table from related coefficient obtained
Levy the high related coefficient of confidence level, comprising:
Traversal selection is carried out to each related coefficient obtained in the following way, determines institute's table in related coefficient obtained
Levy the high related coefficient of confidence level:
In related coefficient obtained, the maximum correlation coefficient of unselected mistake is selected, as currently associated coefficient；
The corresponding spectrum signature cluster of the currently associated coefficient is added to traversal feature set；
Using the corresponding markup information of spectrum signature each in the traversal feature set and the traversal feature set as the first binomial
The input of Logic Regression Models is based on ten folding cross-validation methods, selects the detection so that the first binary logistic regression model
As a result the highest model parameter of accuracy rate completes the training to the first binary logistic regression model, wherein each frequency spectrum is special
Levying corresponding markup information indicates: whether the first sample wireless signal with the spectrum signature is target type wireless signal,
The testing result of the first binary logistic regression model is for indicating: spectrum signature is that each frequency spectrum is special in the traversal feature set
Whether the wireless signal of sign belongs to target type；
Judge whether the accuracy rate of testing result corresponding to selected model parameter is greater than the accuracy rate of history testing result；
If being not more than, the corresponding spectrum signature cluster of the currently associated coefficient is deleted from the traversal feature set；
If more than determining that the currently associated coefficient is the high phase relation of confidence level characterized in related coefficient obtained
Number.
4. method according to any one of claim 1-3, which is characterized in that the feature type of the target frequency bands, packet
Include at least one of following characteristics type:
The mean value of power spectral density；
The variance of power spectral density；
The Differential Characteristics of power spectral density.
5. method according to any one of claim 1-3, which is characterized in that described to be examined according to extracted spectrum signature
Survey whether the signal to be detected belongs to target type, comprising:
Extracted spectrum signature is inputted in two classifiers that training obtains in advance, obtains whether the signal to be detected belongs to
The testing result of target type；
Wherein, training obtains two classifier in the following ways:
Obtain the second sample wireless signal, wherein the second sample wireless signal includes: the wireless signal for belonging to target type
With the wireless signal for being not belonging to target type；
For each target frequency bands, the spy that the second sample wireless signal belongs to the target frequency bands in the target frequency bands is extracted
The spectrum signature for levying type, as the second sample spectra feature；
For every one second sample wireless signal, obtain by belonging to the second sample wireless signal in the second sample spectra feature
The feature vector that spectrum signature is formed；
It is denoted as known quantity with feature vector obtained and the corresponding zero standard of each feature vector, it is inclined with weighting parameter vector sum
Setting parameter is unknown quantity, constructs the second binary logistic regression model, wherein the corresponding zero flag of each feature vector are as follows: with
Whether one zero form indicates, the corresponding second sample wireless signal of each feature vector belongs to the label of target type；
Obtain taking for weighting parameter when the second binary logistic regression model makes maximum likelihood algorithm obtain maximum
The value of value and the offset parameter；
The second binomial logic time when by the value of offset parameter described in the weighting parameter vector sum being respectively obtained value
Model is returned to be determined as two classifier.
6. according to the method described in claim 5, it is characterized in that, described with feature vector obtained and each feature vector
A corresponding zero standard is denoted as known quantity, using weighting parameter vector sum offset parameter as unknown quantity, constructs the second binary logistic regression
Model, comprising:
Using following formula, the second binary logistic regression model is established:
Wherein, xiFor ith feature vector, yiFor the corresponding zero flag of ith feature vector, ω be the weighting parameter to
Amount, b are the offset parameter.
7. according to the method described in claim 5, it is characterized in that, acquisition the second binary logistic regression model makes
Maximum likelihood algorithm obtains the value of the value of the weighting parameter and the offset parameter when maximum, comprising:
The mode that the maximum of default log-likelihood function is calculated by using gradient descent method, obtains the second binomial logic
The value of the value of the weighting parameter and offset parameter when regression model makes maximum likelihood algorithm obtain maximum；
Wherein, the default log-likelihood function are as follows:
Wherein, L (ω) is log-likelihood function, xiFor ith feature vector, yiFor the corresponding zero standard of ith feature vector
Note, N are the quantity of feature vector, and ω is the weighting parameter vector, and b is the offset parameter.
8. according to the method described in claim 5, it is characterized in that, described by offset parameter described in the weighting parameter vector sum
Value be respectively obtained value when the second binary logistic regression model be determined as two classifier, comprising:
Determine that by the classifier that following expression indicates be two classifier:
Wherein,For the value of the weighting parameter vector,For the value of the offset parameter, x is to input two classifier
The vector being made of spectrum signature, y is a zero flag of two classifier output, a zero standard of two classifier output
Be denoted as: whether wireless signal represented by vector indicated with a zero form, input two classifier belongs to target type
Label.
9. a kind of wireless signal detection device characterized by comprising
Signal acquisition module, for obtaining wireless signal to be detected；
Frequency range obtains module, for obtaining the target frequency bands of spectrum signature to be extracted, and obtains in each target frequency bands wait mention
The type for taking spectrum signature, the feature type as each target frequency bands；
Characteristic extracting module is extracted the wireless signal to be detected and is belonged in the target frequency bands for being directed to each target frequency bands
In the spectrum signature of the feature type of the target frequency bands；
Signal detection module, for detecting whether the signal to be detected belongs to target type according to extracted spectrum signature；
Wherein, the frequency range obtains module, comprising:
Sample signal acquisition submodule, for obtaining first sample wireless signal, wherein the first sample wireless signal packet
It includes: belonging to the wireless signal of target type and be not belonging to the wireless signal of target type；
Sample characteristics extracting sub-module, for extracting the spectrum signature of the first sample wireless signal, as first sample frequency
Spectrum signature；
Related coefficient obtains submodule, for obtaining the corresponding correlation of every cluster spectrum signature in the first sample spectrum signature
Coefficient, wherein the spectrum signature that every cluster spectrum signature is included, place frequency range is identical and the type of spectrum signature is identical, often
The corresponding related coefficient of cluster spectrum signature is for characterizing: spectrum signature in frequency range where the cluster spectrum signature and type be should
The wireless signal of the affiliated type of cluster spectrum signature belongs to the confidence level of target type；
The preferred submodule of related coefficient, the phase relation for preferentially selecting characterized confidence level high from related coefficient obtained
Number；
Target frequency bands determine submodule, for being directed to selected each related coefficient, by the corresponding spectrum signature of related coefficient
Frequency range where spectrum signature is as target frequency bands, the type of spectrum signature as the feature type in target frequency bands in cluster.
10. a kind of electronic equipment, which is characterized in that including processor, communication interface, memory and communication bus, wherein processing
Device, communication interface, memory complete mutual communication by communication bus；
Memory, for storing computer program；
Processor when for executing the program stored on memory, realizes any method and step of claim 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810723748.2A CN109104257B (en) | 2018-07-04 | 2018-07-04 | Wireless signal detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810723748.2A CN109104257B (en) | 2018-07-04 | 2018-07-04 | Wireless signal detection method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109104257A true CN109104257A (en) | 2018-12-28 |
| CN109104257B CN109104257B (en) | 2020-08-21 |
Family
ID=64845643
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810723748.2A Active CN109104257B (en) | 2018-07-04 | 2018-07-04 | Wireless signal detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109104257B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110518924A (en) * | 2019-09-12 | 2019-11-29 | 北京维普无限智能技术有限公司 | A method of improving wireless signal transceiver compatibility |
| US20210150286A1 (en) * | 2019-11-20 | 2021-05-20 | Rohde & Schwarz Gmbh & Co. Kg | Method and system for detecting and/or classifying a wanted signal |
| CN113300986A (en) * | 2021-04-17 | 2021-08-24 | 湖南红船科技有限公司 | Unmanned aerial vehicle image transmission signal and hotspot signal identification method, medium and computer equipment |
| CN114978410A (en) * | 2022-07-28 | 2022-08-30 | 杭州恒生数字设备科技有限公司 | Method and system for shielding radio signal |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102364885A (en) * | 2011-10-11 | 2012-02-29 | 宁波大学 | Frequency spectrum sensing method based on signal frequency spectrum envelope |
| CN104468001A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Signal identification method and system based on radio signal frequency spectrum feature template |
| US9094080B1 (en) * | 2013-01-04 | 2015-07-28 | Marvell International Ltd. | Method and apparatus for estimating statistics in wireless systems |
| CN107222875A (en) * | 2017-07-10 | 2017-09-29 | 东南大学 | The small base station switch control method analyzed based on big data |
-
2018
- 2018-07-04 CN CN201810723748.2A patent/CN109104257B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102364885A (en) * | 2011-10-11 | 2012-02-29 | 宁波大学 | Frequency spectrum sensing method based on signal frequency spectrum envelope |
| US9094080B1 (en) * | 2013-01-04 | 2015-07-28 | Marvell International Ltd. | Method and apparatus for estimating statistics in wireless systems |
| CN104468001A (en) * | 2014-11-26 | 2015-03-25 | 北京邮电大学 | Signal identification method and system based on radio signal frequency spectrum feature template |
| CN107222875A (en) * | 2017-07-10 | 2017-09-29 | 东南大学 | The small base station switch control method analyzed based on big data |
Non-Patent Citations (1)
| Title |
|---|
| 常婧云: "无线射频信号的发射源及信息处理方式的识别", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110518924A (en) * | 2019-09-12 | 2019-11-29 | 北京维普无限智能技术有限公司 | A method of improving wireless signal transceiver compatibility |
| US20210150286A1 (en) * | 2019-11-20 | 2021-05-20 | Rohde & Schwarz Gmbh & Co. Kg | Method and system for detecting and/or classifying a wanted signal |
| CN113300986A (en) * | 2021-04-17 | 2021-08-24 | 湖南红船科技有限公司 | Unmanned aerial vehicle image transmission signal and hotspot signal identification method, medium and computer equipment |
| CN114978410A (en) * | 2022-07-28 | 2022-08-30 | 杭州恒生数字设备科技有限公司 | Method and system for shielding radio signal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109104257B (en) | 2020-08-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110222170B (en) | Method, device, storage medium and computer equipment for identifying sensitive data | |
| CN109104257A (en) | A kind of wireless signal detection method and device | |
| JP6942741B2 (en) | Identification for spectroscopic classification with reduced false positives | |
| CN109842513A (en) | Network exception event analytical equipment, method and its computer storage medium | |
| CN109983348A (en) | Realize the technology of portable frequency spectrum analyzer | |
| JP6476861B2 (en) | Electromagnetic field feature classification presentation device | |
| CN102955902B (en) | Method and system for evaluating reliability of radar simulation equipment | |
| CN109934269A (en) | A kind of opener recognition methods of electromagnetic signal and device | |
| US11640525B2 (en) | Synthesis of sequential, spectral, and time-series data | |
| CN115271003B (en) | Abnormal data analysis method and system for automatic environment monitoring equipment | |
| CN108780048A (en) | A kind of method, detection device and the readable storage medium storing program for executing of determining detection device | |
| CN107241696A (en) | Multipath effect discriminating conduct and method for estimating distance based on channel condition information | |
| CN103310235B (en) | A kind of steganalysis method based on parameter identification and estimation | |
| CN104680185B (en) | Hyperspectral image classification method based on boundary point reclassification | |
| CN111582387A (en) | Rock spectral feature fusion classification method and system | |
| CN110348494A (en) | A kind of human motion recognition method based on binary channels residual error neural network | |
| CN109389176A (en) | Grain measurement of moisture content method and system based on WIFI channel state information | |
| CN106250913B (en) | A kind of combining classifiers licence plate recognition method based on local canonical correlation analysis | |
| CN109165665A (en) | A kind of category analysis method and system | |
| Supekar et al. | Multi-parameter based mango grading using image processing and machine learning techniques | |
| CN108280429A (en) | The nuclide identification method, apparatus and computer-readable medium of feature based peak match | |
| CN110009012A (en) | A kind of risk specimen discerning method, apparatus and electronic equipment | |
| CN108663334A (en) | The method for finding soil nutrient spectral signature wavelength based on multiple Classifiers Combination | |
| US10789477B2 (en) | Method and apparatus for real-time detection of a scene | |
| CN110428012A (en) | Brain method for establishing network model, brain image classification method, device and electronic equipment |
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 |