CN109039503A - A kind of frequency spectrum sensing method, device, equipment and computer readable storage medium - Google Patents
A kind of frequency spectrum sensing method, device, equipment and computer readable storage medium Download PDFInfo
- Publication number
- CN109039503A CN109039503A CN201811045616.5A CN201811045616A CN109039503A CN 109039503 A CN109039503 A CN 109039503A CN 201811045616 A CN201811045616 A CN 201811045616A CN 109039503 A CN109039503 A CN 109039503A
- Authority
- CN
- China
- Prior art keywords
- sample
- frequency spectrum
- signal
- covariance matrix
- manifold
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04B—TRANSMISSION
- H04B17/00—Monitoring; Testing
- H04B17/30—Monitoring; Testing of propagation channels
- H04B17/382—Monitoring; Testing of propagation channels for resource allocation, admission control or handover
Landscapes
- Physics & Mathematics (AREA)
- Electromagnetism (AREA)
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The embodiment of the invention discloses a kind of frequency spectrum sensing method, device, equipment and computer readable storage mediums.Wherein, method includes parameterizing the covariance matrix to perceptual signal, obtains corresponding probability-distribution function, establishes the mapping relations of covariance matrix and statistical manifold；The measuring distance feature of statistical manifold is extracted, and is inputted in the disaggregated model constructed in advance, disaggregated model is obtained by the distance feature using the different other sample signals of tag class of machine learning algorithm training, and label classification is that there are primary user and primary user is not present.According to the classification results detection of disaggregated model output to whether there is primary user in perceptual signal.The application improves the detection performance under low signal-to-noise ratio based on information geometry method, the primary user in disaggregated model perceived spectral signal established based on machine learning, existing frequency spectrum perception technology is solved the problems, such as there are the low perceptual performance of low signal-to-noise ratio is poor, improves frequency spectrum perception efficiency, accuracy and stability under low signal-to-noise ratio.
Description
Technical field
The present embodiments relate to signal processing technology fields, more particularly to a kind of frequency spectrum sensing method, device, equipment
And computer readable storage medium.
Background technique
With the fast development of wireless communication technique, radio spectrum resources growing tension, however allocated frequency spectrum at present
Utilization rate it is generally relatively low, it is seen that existing frequency band will be fully utilized in traditional frequency spectrum distributing method.Therefore, how to have
The utilization rate that frequency spectrum resource is improved in the frequency spectrum resource of limit is most important to the development of wireless communication technique.Based on this, nothing is recognized
Line electricity comes into being, and frequency spectrum perception is the core of cognitive radio technology.
In cognitive radio technology, classical frequency spectrum sensing method includes energy measuring method, matched filtering detection method, circulation
Stationary nature detection method etc..Most of classical frequency spectrum sensing method is all that single node is taken to detect, and wireless communication in actual environment
There is decline in road, and cognitive user, which is difficult to detect by simple single node, carries out accurate frequency spectrum detection.Therefore, it is necessary to more
A node perceived result is merged, to improve detection reliability, i.e. collaborative sensing technology.Collaborative sensing can be largely
It is influenced caused by upper abatement decaying and masking, detection accuracy with higher.The research of collaborative spectrum sensing is recognized by multiple
Know that user executes frequency spectrum perception jointly and carries out fusion treatment to perception, the method for proposition include "AND" based on hard decision,
"or" and " K/N " collaborative sensing and the energy amalgamation judging based on soft-decision etc..In addition to this, there are also assisted using reception signal
Variance matrix characteristic value carry out frequency spectrum detection, wherein most representational characteristic value cognitive method be minimax characteristic value it
Than detection method (Maximum Minimum Eigenvalue, MME).But it above-mentioned be previously mentioned method all there is in low signal-to-noise ratio
The bad defect of lower perceptual performance.
Specifically, energy detection algorithm is easy to be influenced by noise fluctuations, uncertainty of the detection performance to noise
It is very sensitive；Cyclostationary characteristic detection algorithm complexity is higher, while can reduce the sensitivity of whole system；And matched filtering
Detection algorithm needs the prior information of authorization user signal, and versatility is poor.
Summary of the invention
The purpose of the embodiment of the present invention is that providing a kind of frequency spectrum sensing method, device, equipment and computer-readable storage medium
Matter solves the problems, such as existing frequency spectrum perception technology there are the low perceptual performance of low signal-to-noise ratio is poor, improves frequency under low signal-to-noise ratio
Spectrum perception efficiency and stability.
In order to solve the above technical problems, the embodiment of the present invention the following technical schemes are provided:
On the one hand the embodiment of the present invention provides a kind of frequency spectrum sensing method, comprising:
It treats perceptual signal and carries out statistical property processing, obtain covariance matrix；
The covariance matrix is parameterized, corresponding probability-distribution function is obtained, with establish the covariance matrix with
The mapping relations of statistical manifold；
The measuring distance feature of the statistical manifold is extracted, and is inputted in the disaggregated model constructed in advance, with basis
The classification results detection of the disaggregated model output is described to whether there is primary user in perceptual signal；
Wherein, the disaggregated model be using machine learning algorithm training multiple and different other sample signals of tag class away from
From obtained by feature, the label classification is that there are primary user and primary user is not present.
Optionally, the measuring distance feature for extracting the statistical manifold includes:
The geodesic curve distance for calculating point-to-point transmission on the statistical manifold, the measuring distance feature as the statistical manifold.
Optionally, the geodesic curve distance for calculating point-to-point transmission on the statistical manifold includes:
Using Riemann's intermediate value as reference value, the geodesic curve distance of point-to-point transmission on the statistical manifold is calculated.
Optionally, the training process of the disaggregated model includes:
Each sample signal in training sample is pre-processed, corresponding sample covariance matrix is obtained；
Each sample covariance matrix is parameterized, respective probability-distribution function is obtained, to establish sample covariance
The mapping relations of matrix and sample statistics manifold；
The sample distance feature for extracting each sample statistical manifold trains each sample distance feature using k means clustering algorithm,
Obtain disaggregated model.
Optionally, described described to whether there is in perceptual signal according to the classification results detection of disaggregated model output
Primary user includes:
Judge whether following formula are true:
In formula, d is the measuring distance feature, and μ is mass center, μ={ μ1, μ2..., μk, k is integer, and ε is constant；
If so, the disaggregated model output is described to the classification results in perceptual signal there are primary user；If it is not, then institute
It is described to which primary user's classification results are not present in perceptual signal to state disaggregated model output.
On the other hand the embodiment of the present invention provides a kind of frequency spectrum sensing device, comprising:
Signal pre-processing module carries out statistical property processing for treating perceptual signal, obtains covariance matrix, assisted
Variance matrix；
Geometrical property mapping block obtains corresponding probability-distribution function for parameterizing the covariance matrix, with
Establish the mapping relations of the covariance matrix and statistical manifold；
Signal sensing module by it and is inputted and is constructed in advance for extracting the measuring distance feature of the statistical manifold
It is described primary to whether there is in perceptual signal with the classification results detection exported according to the disaggregated model in disaggregated model
Family；Wherein, the disaggregated model is special using the distance of the multiple and different other sample signals of tag class of machine learning algorithm training
Sign gained, the label classification are that there are primary user and primary user is not present.
Optionally, the signal sensing module is to calculate point-to-point transmission on the statistical manifold using Riemann's intermediate value as reference value
Geodesic curve distance module.
It optionally, further include model training module, the model training module includes:
Sample signal handles submodule, for pre-processing to each sample signal in training sample, obtains corresponding
Sample covariance matrix；
Geometrical property setting up submodule obtains respective probability point for parameterizing each sample covariance matrix
Cloth function, to establish the mapping relations of sample covariance matrix Yu sample statistics manifold；
Feature training submodule utilizes k means clustering algorithm for extracting the sample distance feature of each sample statistical manifold
Each sample distance feature of training, obtains disaggregated model.
The embodiment of the invention also provides a kind of frequency spectrum perception equipment, including processor, the processor is deposited for executing
It is realized when the computer program stored in reservoir as described in preceding any one the step of frequency spectrum sensing method.
The embodiment of the present invention finally additionally provides a kind of computer readable storage medium, the computer readable storage medium
On be stored with frequency spectrum perception program, when the frequency spectrum perception program is executed by processor realize as described in preceding any one frequency spectrum perception
The step of method.
The embodiment of the invention provides a kind of frequency spectrum sensing methods, and the covariance matrix to perceptual signal is parameterized, is obtained
To corresponding probability-distribution function, the mapping relations of covariance matrix and statistical manifold are established；Extract statistical manifold test away from
It from feature, and is inputted in the disaggregated model constructed in advance, disaggregated model is multiple and different using machine learning algorithm training
Obtained by the distance feature of the other sample signal of tag class, label classification is that there are primary user and primary user is not present.According to classification
The classification results detection of model output is to whether there is primary user in perceptual signal.
The advantages of technical solution provided by the present application, is, will utilize the geometry in statistical property and manifold to perceptual signal
Characteristic establishes mapping relations, and the method based on information geometry improves the detection performance under low signal-to-noise ratio, in addition, being based on engineering
The disaggregated model for practising algorithm training carries out feature identification, avoids the acquisition to thresholding, reduces influence of the thresholding to detection performance, and
It is identified based on the disaggregated model that machine learning is established to whether there is primary user in perceptual signal, spectrum signal sense can be effectively improved
The accuracy and efficiency known solves the problems, such as existing frequency spectrum perception technology there are the low perceptual performance of low signal-to-noise ratio is poor, improves
Frequency spectrum perception efficiency and stability under low signal-to-noise ratio.
In addition, the embodiment of the present invention provides corresponding realization device, equipment and computer also directed to frequency spectrum sensing method
Readable storage medium storing program for executing, further such that the method has more practicability, described device, equipment and computer readable storage medium
Have the advantages that corresponding.
Detailed description of the invention
It, below will be to embodiment or existing for the clearer technical solution for illustrating the embodiment of the present invention or the prior art
Attached drawing needed in technical description is briefly described, it should be apparent that, the accompanying drawings in the following description is only this hair
Bright some embodiments for those of ordinary skill in the art without creative efforts, can be with root
Other attached drawings are obtained according to these attached drawings.
Fig. 1 is a kind of flow diagram of frequency spectrum sensing method provided in an embodiment of the present invention；
Fig. 2 is a kind of statistical manifold schematic diagram provided in an embodiment of the present invention；
Fig. 3 is a kind of feature extraction schematic diagram provided in an embodiment of the present invention；
Fig. 4 is Riemann's intermediate value schematic diagram provided in an embodiment of the present invention；
Fig. 5 is many algorithms performance simulation schematic diagram provided in an embodiment of the present invention；
Fig. 6 is a kind of specific embodiment structure chart of frequency spectrum sensing device provided in an embodiment of the present invention.
Specific embodiment
In order to enable those skilled in the art to better understand the solution of the present invention, with reference to the accompanying drawings and detailed description
The present invention is described in further detail.Obviously, described embodiments are only a part of the embodiments of the present invention, rather than
Whole embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art are not making creative work premise
Under every other embodiment obtained, shall fall within the protection scope of the present invention.
The description and claims of this application and term " first ", " second ", " third " " in above-mentioned attached drawing
Four " etc. be for distinguishing different objects, rather than for describing specific sequence.Furthermore term " includes " and " having " and
Their any deformations, it is intended that cover and non-exclusive include.Such as contain a series of steps or units process, method,
System, product or equipment are not limited to listed step or unit, but may include the step of not listing or unit.
Present inventor has found that information geometry is a set of theoretical body that developed based on Differential Geometry after study
System.Many key concepts in probability theory, information theory and statistics are considered as the geometry knot in probability distribution space by information geometry
Structure studies property thereon using Differential Geometry method, so that the basic problem geometrization in probability theory and information theory be assigned
Give in it geometry essence.For the probability-distribution function race of different type or different parameters, both corresponding to one has
The statistical manifold of certain geometry.Since the form of probability-distribution function determines wherein each probability-distribution function and its
Relationship between the probability-distribution function of near its circumference, and this relationship determines the structure in the space that it is constituted.Therefore, it unites
Count manifold geometry, reflect in probability-distribution function race essential attribute.Information geometry is exactly by probability distribution
On the statistical manifold that family of functions is constituted, using Modern Differential Geometry method come the theoretical body of Research statistics and information theory problem
System.The advantage of information geometry is that it gathers probability-distribution function race as one, and by a set of principle and method, to grind
Study carefully the immanent structure information that probability distribution is contained.Since each point on statistical manifold represents a probability distribution letter
Number, and probability-distribution function is the basis of information theory and statistical research, therefore, information geometry can provide for signal detection
One new research angle.
There is geometric property and intension in view of information, an information mistake can effectively be described using geometry method
Journey, therefore, information geometry are very suitable for the research to problem of model selection.And in signal detection signal data as information,
These data can just be divided into H0(only Noise) and H1(noise+signal) two kinds of models.A usual statistical model is family
The set (being denoted as M) of probability distribution, (is denoted as S) when the set of whole probability distribution can form a statistical manifold.Therefore, these
Signal data corresponds to two points in statistical manifold, by calculating the distance in manifold between points, then according to distance
Difference judgement belongs to any model, and then reaches the purpose of signal detection namely frequency spectrum perception.
Information geometry is theoretical system emerging in recent years.Signal detecting method based on information geometry is less, in radar
There are some applications in echo detecting.And Radar Signal Processing is an important branch of signal processing, is had a wide range of applications
Background.In this field, representational implementation is that Barbaresco etc. is based on the realization pulse of positive definite matrix manifold
The research of Doppler radar Matrix C FAR detection.Sample covariance matrix of the program based on limited impulse sampling is established
Toeplitz positive definite matrix is popular, each point corresponds to the sample covariance matrix of a reference unit, these associations in manifold
Variance matrix constitutes the manifold space with negative cruvature.In the manifold, the corresponding association side of each reference unit is calculated first
The Riemann mean value (geometric mean) of poor matrix, for estimating clutter power；Then using the geometry between covariance matrix away from
From calculating unit R to be detectedDThe distance between Riemann mean value of reference unit, provides testing result compared with thresholding.
This method is equally used in frequency spectrum perception, its pulse signal is changed to the spectrum signal of perception.
But when exceptional value (outlier) occurs in sample data, Riemann mean value (geometric mean) may occur larger
Change, therefore, detection performance is not sufficiently stable.In addition to this, it is made decisions in positive definite matrix CFAR detection scheme by thresholding,
And the method for using thresholding, often there is deviation, will affect detection performance.
Frequency spectrum perception can be a kind of two classification problems (primary user exists or primary user is not present), and machine learning is calculated
Method, it is well known that be widely used in pattern classification field.Therefore using any machine learning algorithm training classification mould
Type perceives spectrum signal, to realize in perceptual signal with the presence or absence of primary user.
After describing the technical solution of the embodiment of the present invention, the various non-limiting realities of detailed description below the application
Apply mode.
Referring first to Fig. 1, Fig. 1 is a kind of flow diagram of frequency spectrum sensing method provided in an embodiment of the present invention, this hair
Bright embodiment may include the following contents:
S101: it treats perceptual signal and carries out statistical property processing, obtain covariance matrix.
S102: covariance matrix is parameterized, and obtains corresponding probability-distribution function, to establish covariance matrix and statistics
The mapping relations of manifold.
S103: the measuring distance feature of statistical manifold is extracted, and is inputted in the disaggregated model constructed in advance, with basis
The classification results detection of disaggregated model output is to whether there is primary user in perceptual signal.
Only existed in system a time user (Secondary user, SU) to primary user (Primary user, PU) into
When row detection, frequency spectrum perception problem can be indicated with following dualism hypothesis:
Wherein, H0It is not present for main subscriber signal, H1For the presence of main subscriber signal, a (n) is main subscriber signal, and w (n) is only
The vertical white Gaussian noise signal for being with distribution, mean value 0, variance.
There is the collaborative spectrum sensing of M SU in cognitive radio system, constitutes a perception matrixAnd the sampling number of each SU is N, wherein xi=[xi(1),xi(2),...,xi(N)]TIt indicates
For the signal sampling value of i-th of SU.The signal matrix of an available N × M:
Therefore, the covariance matrix to perceptual signal received can be by R=E [XTX] it indicates.
According to information geometry theory, probability density function can be parameterized to obtain by corresponding covariance matrix.I.e. to any
One n-dimensional vector, when paying zero-mean gaussian distribution, distribution expression formula can are as follows:
Covariance mean value R ∈ Cn×nParametrization family of probability distribution S=p (x | R) | R ∈ Cn×n, wherein Cn×nFor n × n
Opener in dimensional vector space, X are n-dimensional vector, and R is covariance matrix.S formation one can be micro- under certain topological structure
Manifold, referred to as statistical manifold.Fig. 2 is the schematic diagram of statistical manifold, and coordinate R, sample x and statistical manifold S are learnt from the figure
Corresponding relationship between three.R is the coordinate of manifold S in figure, and x is an example of sample space Ω, probability density function
For p (x | θ).Therefore, in n-dimensional vector spaceIn, for each by the probability-distribution function p (x | θ) of θ parametrization,
It can correspond to a point S (θ) of S on statistical manifold.
It, also i.e. will be on covariance matrix and statistical manifold after establishing the mapping relations of covariance matrix and statistical manifold
Point it is corresponding after, judge to carry out it to label classification belonging to perceptual signal (having primary user and no primary user)
When classification, judged by measuring the distance between unknown signaling and known label signal size, the distance of the two is smaller, belongs to
The probability of same class label is also bigger.Therefore the distance between can put in counting statistics manifold and to be used as distance feature, for subsequent
Input the identification feature of disaggregated model.
Disaggregated model can be the distance feature for utilizing the multiple and different other sample signals of tag class of machine learning algorithm training
Gained, label classification are that there are primary user and primary user is not present.
In technical solution provided in an embodiment of the present invention, the geometry in statistical property and manifold will be utilized to perceptual signal
Characteristic establishes mapping relations, and the method based on information geometry improves the detection performance under low signal-to-noise ratio, in addition, being based on engineering
The disaggregated model for practising algorithm training carries out feature identification, avoids the acquisition to thresholding, reduces influence of the thresholding to detection performance, and
It is identified based on the disaggregated model that machine learning is established to whether there is primary user in perceptual signal, spectrum signal sense can be effectively improved
The accuracy and efficiency known solves the problems, such as existing frequency spectrum perception technology there are the low perceptual performance of low signal-to-noise ratio is poor, improves
Frequency spectrum perception efficiency and stability under low signal-to-noise ratio.
Cluster is a kind of unsupervised study, and similar object can be grouped into same cluster.Cluster identification gives cluster knot
The meaning of fruit.It clusters and is that the target of classification is previously known, and clusters different with the maximum difference of classification.Because it is generated
Result it is identical as classification, and only classification does not pre-define, and cluster is also sometimes referred to as unsupervised segmentation.
K- means clustering algorithm is the data set of given k cluster, each cluster can be described by its mass center, and mass center
For the average value of data points all in current cluster.Since frequency spectrum perception only needs to be divided into, channel is available and unavailable two class of channel,
So giving 2 clusters in K- means clustering algorithm just meets condition namely k=2.
Sample distance feature is trained based on k means clustering algorithm to obtain the process of disaggregated model can include:
Each sample signal in training sample is pre-processed, corresponding sample covariance matrix is obtained；
Each sample covariance matrix is parameterized, respective probability-distribution function is obtained, to establish sample covariance
The mapping relations of matrix and sample statistics manifold；
The sample distance feature for extracting each sample statistical manifold trains each sample distance feature using k means clustering algorithm,
Obtain disaggregated model.
Judging that measuring distance feature is input in disaggregated model, is passed through when the affiliated label classification of perceptual signal
Following formula judges affiliated classification:
In formula, d is measuring distance feature, and μ is mass center, μ={ μ1, μ2..., μk, k is integer, and ε is constant, in reality
When testing middle calculating with statistic mixed-state probability, false-alarm probability and false dismissal probability are controlled.
If measuring distance feature meets above formula, determine that PU exists, otherwise determines that PU is not present.
Pseudocode of the table 1 based on information geometry and k means clustering algorithm
From the foregoing, it will be observed that avoiding the acquisition to thresholding using k means clustering algorithm train classification models, reduce thresholding to inspection
The influence for surveying performance can effectively improve the accuracy and efficiency of spectrum signal perception.
The probability density function of parametrization can be indicated by its covariance matrix namely statistical manifold on point correspond to accordingly
Covariance matrix.Calculating the distance between probability distribution is the distance calculated between covariance matrix corresponding to probability distribution.
The mode that the distance between two probability distribution is measured in statistical manifold has very much, since geodesic curve distance considers connection two o'clock most
Short path, the i.e. structure of manifold, optionally, when extracting distance feature, can in counting statistics manifold point-to-point transmission geodesic curve away from
From measuring distance feature as statistical manifold.
For considering wherein covariance matrix point with identical mean value but the different multivariate Gaussian family of distributions of covariance matrix
It Wei not R1And R2Two members, geodesic curve distance between the two can be calculated according to the following formula:
det(R2-λjR1)=0；
In formula, λjFor matrix (R1 -1/2*R2*R1 -1/2) characteristic value, j=1,2 ..., be characterized the number of value.
It can carry out geodesic curve distance twice to calculate, obtain d1And d2Two as a result, enable d=[d1,d2] indicate to perceptual signal
Distance feature vector.
In view of being influenced the association for causing, sampling every time by uncertain factors such as multipath fading and shades in communication environments
Variance matrix has certain deviation, therefore, with T covariance matrix Rl(1,2 ..., T) covariance matrix is calculated
Average valueSuch as T=16.Through calculating covariance matrix average value can reduce environment because
Influence of the element to detection performance, in order to improve the accuracy of frequency spectrum perception signal, the geodetic of point-to-point transmission in counting statistics manifold
Linear distance, Riemann's intermediate value, which can be used, indicates covariance matrix average value as reference value, namely distance of the extraction to perceptual signal
Feature can as shown in figure 3, by covariance matrix it is corresponding with the point on statistical manifold after, calculated by geodesic curve metric form
The distance of two points in manifold, obtains measuring distance feature.
The pseudocode of the calculating Riemann's intermediate value of table 1
It is emulated according to Riemann's intermediate value, obtains Fig. 4, from analogous diagram it is recognised that needing iteration 60 times, Riemann's intermediate value is
It tends towards stability.
From the foregoing, it will be observed that being averaged using Riemann's mediant estimation covariance matrix of the embodiment of the present invention, due to Riemann's intermediate value pair
Robustness, therefore the detection performance under can further improve low signal-to-noise ratio are had more in the outlier of data.
In order to verify technical solution provided by the present application compared with prior art, there is preferably detection property under low signal-to-noise ratio
Can, the application is with collaboration user number M=5, under sampling number N=500 environment, to the detection performance of different spectral cognitive method with
The relationship of signal-to-noise ratio is emulated, and is please referred to shown in Fig. 5.
As seen from the figure, under low false-alarm probability, the detection performance of technical scheme is much than existing frequency spectrum perception side
Method performance is good；In addition, the detection performance of technical scheme also increases accordingly with the increase of signal-to-noise ratio.
The embodiment of the present invention provides corresponding realization device also directed to frequency spectrum sensing method, further such that the method
With more practicability.Frequency spectrum sensing device provided in an embodiment of the present invention is introduced below, frequency spectrum perception described below
Device can correspond to each other reference with above-described frequency spectrum sensing method.
Referring to Fig. 6, Fig. 6 is a kind of structure of the frequency spectrum sensing device provided in an embodiment of the present invention under specific embodiment
Figure, the device can include:
Signal pre-processing module 601 carries out statistical property processing for treating perceptual signal, obtains covariance matrix, obtain
To covariance matrix.
Geometrical property mapping block 602 obtains corresponding probability distribution letter for parameterizing the covariance matrix
Number, to establish the mapping relations of the covariance matrix and statistical manifold.
Signal sensing module 603 by it and inputs building in advance for extracting the measuring distance feature of the statistical manifold
Disaggregated model in, with the classification results detection exported according to the disaggregated model it is described in perceptual signal with the presence or absence of primary
Family；Wherein, the disaggregated model is special using the distance of the multiple and different other sample signals of tag class of machine learning algorithm training
Sign gained, the label classification are that there are primary user and primary user is not present.
Optionally, in some embodiments of the present embodiment, the signal sensing module 603 can be to be with Riemann's intermediate value
Reference value, the module of the geodesic curve distance of point-to-point transmission in counting statistics manifold.
In other embodiments of the present embodiment, described device for example can also include model training module, described
Model training module includes:
Sample signal handles submodule, for pre-processing to each sample signal in training sample, obtains corresponding
Sample covariance matrix；
Geometrical property setting up submodule obtains respective probability point for parameterizing each sample covariance matrix
Cloth function, to establish the mapping relations of sample covariance matrix Yu sample statistics manifold；
Feature training submodule utilizes k means clustering algorithm for extracting the sample distance feature of each sample statistical manifold
Each sample distance feature of training, obtains disaggregated model.
Specifically, the signal sensing module 603 can be to judge whether following formula are true, if so, then disaggregated model
Output is to the classification results in perceptual signal there are primary user；Otherwise, then disaggregated model output is led to be not present in perceptual signal
The module of user's classification results:
In formula, d is measuring distance feature, and μ is mass center, μ={ μ1, μ2..., μk, k is integer, and ε is constant.
The function of each functional module of frequency spectrum sensing device described in the embodiment of the present invention can be according in above method embodiment
Method specific implementation, specific implementation process is referred to the associated description of above method embodiment, and details are not described herein again.
From the foregoing, it will be observed that the embodiment of the present invention will be reflected to perceptual signal using statistical property and the geometrical property foundation in manifold
Relationship is penetrated, the method based on information geometry improves the detection performance under low signal-to-noise ratio, in addition, based on machine learning algorithm training
Disaggregated model carry out feature identification, avoid the acquisition to thresholding, reduce influence of the thresholding to detection performance, and be based on engineering
The disaggregated model identification established is practised to whether there is primary user in perceptual signal, the accuracy of spectrum signal perception can be effectively improved
And efficiency, it solves the problems, such as existing frequency spectrum perception technology there are the low perceptual performance of low signal-to-noise ratio is poor, improves low signal-to-noise ratio
Lower frequency spectrum perception efficiency and stability.
The embodiment of the invention also provides a kind of frequency spectrum perception equipment, specifically can include:
Memory, for storing computer program；
Processor realizes the step of frequency spectrum sensing method described in any one embodiment as above for executing computer program
Suddenly.
The function of each functional module of frequency spectrum perception equipment described in the embodiment of the present invention can be according in above method embodiment
Method specific implementation, specific implementation process is referred to the associated description of above method embodiment, and details are not described herein again.
From the foregoing, it will be observed that the embodiment of the present invention solves existing frequency spectrum perception technology, there are the low perceptual performance of low signal-to-noise ratio is poor
The problem of, improve frequency spectrum perception efficiency and stability under low signal-to-noise ratio.
The embodiment of the invention also provides a kind of computer readable storage mediums, are stored with frequency spectrum perception program, the frequency
Spectrum awareness program when being executed by processor as above frequency spectrum sensing method described in any one embodiment the step of.
The function of each functional module of computer readable storage medium described in the embodiment of the present invention can be according to above method reality
The method specific implementation in example is applied, specific implementation process is referred to the associated description of above method embodiment, herein no longer
It repeats.
From the foregoing, it will be observed that the embodiment of the present invention solves existing frequency spectrum perception technology, there are the low perceptual performance of low signal-to-noise ratio is poor
The problem of, improve frequency spectrum perception efficiency and stability under low signal-to-noise ratio.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with it is other
The difference of embodiment, same or similar part may refer to each other between each embodiment.For being filled disclosed in embodiment
For setting, since it is corresponded to the methods disclosed in the examples, so being described relatively simple, related place is referring to method part
Explanation.
Professional further appreciates that, unit described in conjunction with the examples disclosed in the embodiments of the present disclosure
And algorithm steps, can be realized with electronic hardware, computer software, or a combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software generally describes each exemplary composition and step according to function in the above description.These
Function is implemented in hardware or software actually, the specific application and design constraint depending on technical solution.Profession
Technical staff can use different methods to achieve the described function each specific application, but this realization is not answered
Think beyond the scope of this invention.
The step of method described in conjunction with the examples disclosed in this document or algorithm, can directly be held with hardware, processor
The combination of capable software module or the two is implemented.Software module can be placed in random access memory (RAM), memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In any other form of storage medium well known in field.
Above to a kind of frequency spectrum sensing method provided by the present invention, device, equipment and computer readable storage medium into
It has gone and has been discussed in detail.Used herein a specific example illustrates the principle and implementation of the invention, the above implementation
The explanation of example is merely used to help understand method and its core concept of the invention.It should be pointed out that for the general of the art
, without departing from the principle of the present invention, can be with several improvements and modifications are made to the present invention for logical technical staff, this
A little improvement and modification are also fallen within the protection scope of the claims of the present invention.
Claims (10)
1. a kind of frequency spectrum sensing method characterized by comprising
It treats perceptual signal and carries out statistical property processing, obtain covariance matrix；
The covariance matrix is parameterized, corresponding probability-distribution function is obtained, to establish the covariance matrix and statistics
The mapping relations of manifold；
The measuring distance feature of the statistical manifold is extracted, and is inputted in the disaggregated model constructed in advance, according to
The classification results detection of disaggregated model output is described to whether there is primary user in perceptual signal；
Wherein, the disaggregated model is special using the distance of the multiple and different other sample signals of tag class of machine learning algorithm training
Sign gained, the label classification are that there are primary user and primary user is not present.
2. frequency spectrum sensing method according to claim 1, which is characterized in that the test for extracting the statistical manifold away from
Include: from feature
The geodesic curve distance for calculating point-to-point transmission on the statistical manifold, the measuring distance feature as the statistical manifold.
3. frequency spectrum sensing method according to claim 2, which is characterized in that described to calculate point-to-point transmission on the statistical manifold
Geodesic curve distance include:
Using Riemann's intermediate value as reference value, the geodesic curve distance of point-to-point transmission on the statistical manifold is calculated.
4. according to claim 1 to frequency spectrum sensing method described in 3 any one, which is characterized in that the instruction of the disaggregated model
Practicing process includes:
Each sample signal in training sample is pre-processed, corresponding sample covariance matrix is obtained；
Each sample covariance matrix is parameterized, respective probability-distribution function is obtained, to establish sample covariance matrix
With the mapping relations of sample statistics manifold；
The sample distance feature for extracting each sample statistical manifold is obtained using each sample distance feature of k means clustering algorithm training
Disaggregated model.
5. frequency spectrum sensing method according to claim 4, which is characterized in that point according to disaggregated model output
The detection of class result is described to include: with the presence or absence of primary user in perceptual signal
Judge whether following formula are true:
In formula, d is the measuring distance feature, and μ is mass center, μ={ μ1, μ2..., μk, k is integer, and ε is constant；
If so, the disaggregated model output is described to the classification results in perceptual signal there are primary user；If it is not, then described point
Class model output is described to which primary user's classification results are not present in perceptual signal.
6. a kind of frequency spectrum sensing device characterized by comprising
Signal pre-processing module carries out statistical property processing for treating perceptual signal, obtains covariance matrix, obtain covariance
Matrix；
Geometrical property mapping block obtains corresponding probability-distribution function, for parameterizing the covariance matrix to establish
The mapping relations of the covariance matrix and statistical manifold；
Signal sensing module by it and inputs the classification constructed in advance for extracting the measuring distance feature of the statistical manifold
It is described to whether there is primary user in perceptual signal with the classification results detection exported according to the disaggregated model in model；Its
In, the disaggregated model is the distance feature institute that utilizes the multiple and different other sample signals of tag class of machine learning algorithm training
, the label classification is that there are primary user and primary user is not present.
7. frequency spectrum sensing device according to claim 6, which is characterized in that the signal sensing module is with Riemann's intermediate value
For reference value, the module of the geodesic curve distance of point-to-point transmission on the statistical manifold is calculated.
8. frequency spectrum sensing device according to claim 6 or 7, which is characterized in that it further include model training module, the mould
Type training module includes:
Sample signal handles submodule and obtains corresponding sample for pre-processing to each sample signal in training sample
This covariance matrix；
Geometrical property setting up submodule obtains respective probability distribution letter for parameterizing each sample covariance matrix
Number, to establish the mapping relations of sample covariance matrix Yu sample statistics manifold；
Feature trains submodule, for extracting the sample distance feature of each sample statistical manifold, utilizes the training of k means clustering algorithm
Each sample distance feature, obtains disaggregated model.
9. a kind of frequency spectrum perception equipment, which is characterized in that including processor, the processor is used to execute to store in memory
It is realized when computer program as described in any one of claim 1 to 5 the step of frequency spectrum sensing method.
10. a kind of computer readable storage medium, which is characterized in that be stored with frequency spectrum sense on the computer readable storage medium
Know program, the frequency spectrum sensing method as described in any one of claim 1 to 5 is realized when the frequency spectrum perception program is executed by processor
The step of.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811045616.5A CN109039503A (en) | 2018-09-07 | 2018-09-07 | A kind of frequency spectrum sensing method, device, equipment and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811045616.5A CN109039503A (en) | 2018-09-07 | 2018-09-07 | A kind of frequency spectrum sensing method, device, equipment and computer readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109039503A true CN109039503A (en) | 2018-12-18 |
Family
ID=64624185
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811045616.5A Pending CN109039503A (en) | 2018-09-07 | 2018-09-07 | A kind of frequency spectrum sensing method, device, equipment and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109039503A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110365434A (en) * | 2019-06-27 | 2019-10-22 | 广东工业大学 | Multi-antenna cooperative frequency spectrum sensing method based on information geometry and differential evolution clustering algorithm |
| CN111614421A (en) * | 2020-05-19 | 2020-09-01 | 重庆邮电大学 | Spectrum sensing method based on unsupervised machine learning classification algorithm |
| CN112350790A (en) * | 2020-09-25 | 2021-02-09 | 深圳大学 | Deep learning-based spectrum sensing detection method, device and equipment |
| CN112787736A (en) * | 2020-12-30 | 2021-05-11 | 杭州电子科技大学 | Long-short term memory cooperative spectrum sensing method based on covariance matrix |
| CN112968741A (en) * | 2021-02-01 | 2021-06-15 | 中国民航大学 | Adaptive broadband compressed spectrum sensing algorithm based on least square vector machine |
| CN114254265A (en) * | 2021-12-20 | 2022-03-29 | 军事科学院系统工程研究院网络信息研究所 | Satellite communication interference geometric analysis method based on statistical manifold distance |
| CN115577253A (en) * | 2022-11-23 | 2023-01-06 | 四川轻化工大学 | Supervision spectrum sensing method based on geometric power |
| WO2023213081A1 (en) * | 2022-05-05 | 2023-11-09 | 中兴通讯股份有限公司 | Spectrum sensing method, electronic device and computer readable storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104135327A (en) * | 2014-07-10 | 2014-11-05 | 上海大学 | Spectrum sensing method based on support vector machine |
| US20170093603A1 (en) * | 2015-04-02 | 2017-03-30 | The Board Of Trustees Of The University Of Alabama | Systems and methods for detecting unused communication spectrum |
| CN107979431A (en) * | 2017-11-28 | 2018-05-01 | 广东工业大学 | The method, apparatus and equipment of frequency spectrum perception based on Riemann's intermediate value |
| CN108462544A (en) * | 2018-03-27 | 2018-08-28 | 广东工业大学 | A kind of frequency spectrum sensing method and device |
-
2018
- 2018-09-07 CN CN201811045616.5A patent/CN109039503A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104135327A (en) * | 2014-07-10 | 2014-11-05 | 上海大学 | Spectrum sensing method based on support vector machine |
| US20170093603A1 (en) * | 2015-04-02 | 2017-03-30 | The Board Of Trustees Of The University Of Alabama | Systems and methods for detecting unused communication spectrum |
| CN107979431A (en) * | 2017-11-28 | 2018-05-01 | 广东工业大学 | The method, apparatus and equipment of frequency spectrum perception based on Riemann's intermediate value |
| CN108462544A (en) * | 2018-03-27 | 2018-08-28 | 广东工业大学 | A kind of frequency spectrum sensing method and device |
Non-Patent Citations (2)
| Title |
|---|
| VAIBHAV KUMAR等: "K-mean Clustering based Cooperative Spectrum Sensing in Generalized k-μ Fading Channels", 《2016 TWENTY SECOND NATIONAL CONFERENCE ON COMMUNICATION (NCC)》 * |
| 孙有铭等: "噪声功率不确定模型下基于CFAＲ 准则的能量检测门限优化算法", 《四川大学学报(工程科学版)》 * |
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110365434B (en) * | 2019-06-27 | 2021-08-24 | 广东工业大学 | Multi-antenna cooperative spectrum sensing method based on information geometry and differential evolution clustering algorithm |
| CN110365434A (en) * | 2019-06-27 | 2019-10-22 | 广东工业大学 | Multi-antenna cooperative frequency spectrum sensing method based on information geometry and differential evolution clustering algorithm |
| CN111614421A (en) * | 2020-05-19 | 2020-09-01 | 重庆邮电大学 | Spectrum sensing method based on unsupervised machine learning classification algorithm |
| CN112350790A (en) * | 2020-09-25 | 2021-02-09 | 深圳大学 | Deep learning-based spectrum sensing detection method, device and equipment |
| CN112350790B (en) * | 2020-09-25 | 2021-12-28 | 深圳大学 | Deep learning-based spectrum sensing detection method, device and equipment |
| CN112787736A (en) * | 2020-12-30 | 2021-05-11 | 杭州电子科技大学 | Long-short term memory cooperative spectrum sensing method based on covariance matrix |
| CN112787736B (en) * | 2020-12-30 | 2022-05-31 | 杭州电子科技大学 | Long-short term memory cooperative spectrum sensing method based on covariance matrix |
| CN112968741A (en) * | 2021-02-01 | 2021-06-15 | 中国民航大学 | Adaptive broadband compressed spectrum sensing algorithm based on least square vector machine |
| CN112968741B (en) * | 2021-02-01 | 2022-05-24 | 中国民航大学 | Adaptive broadband compressed spectrum sensing algorithm based on least square vector machine |
| CN114254265A (en) * | 2021-12-20 | 2022-03-29 | 军事科学院系统工程研究院网络信息研究所 | Satellite communication interference geometric analysis method based on statistical manifold distance |
| CN114254265B (en) * | 2021-12-20 | 2022-06-07 | 军事科学院系统工程研究院网络信息研究所 | Satellite communication interference geometric analysis method based on statistical manifold distance |
| WO2023213081A1 (en) * | 2022-05-05 | 2023-11-09 | 中兴通讯股份有限公司 | Spectrum sensing method, electronic device and computer readable storage medium |
| CN115577253A (en) * | 2022-11-23 | 2023-01-06 | 四川轻化工大学 | Supervision spectrum sensing method based on geometric power |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109039503A (en) | A kind of frequency spectrum sensing method, device, equipment and computer readable storage medium | |
| CN108600135B (en) | Method for identifying signal modulation mode | |
| CN107633227B (en) | CSI-based fine-grained gesture recognition method and system | |
| US10374902B2 (en) | Method for clustering wireless channel MPCs based on a KPD doctrine | |
| CN111163472B (en) | Signal identification attack defense method based on generative countermeasure network | |
| CN106789788B (en) | A kind of wireless digital signal Modulation Mode Recognition method and device | |
| CN104135327B (en) | frequency spectrum sensing method based on support vector machine | |
| CN111541511B (en) | Communication interference signal identification method based on target detection in complex electromagnetic environment | |
| CN106886995A (en) | Polyteny example returns the notable object segmentation methods of image of device polymerization | |
| CN106295498B (en) | Remote sensing image target area detection device and method | |
| CN109508733A (en) | A kind of method for detecting abnormality based on distribution probability measuring similarity | |
| CN108171119B (en) | SAR image change detection method based on residual error network | |
| CN104038792B (en) | For the video content analysis method and apparatus of IPTV supervision | |
| CN103310235B (en) | A kind of steganalysis method based on parameter identification and estimation | |
| CN107582061B (en) | Method for identifying human motion state and intelligent mobile device | |
| CN112364851B (en) | Automatic modulation recognition method and device, electronic equipment and storage medium | |
| CN110493221A (en) | A kind of network anomaly detection method based on the profile that clusters | |
| CN109493333A (en) | Ultrasonic Calcification in Thyroid Node point extraction algorithm based on convolutional neural networks | |
| CN110224771A (en) | Frequency spectrum sensing method and device based on BP neural network and information geometry | |
| CN109218240A (en) | A kind of signal in environment backscatter communication system is sent and detection method | |
| CN108768563A (en) | A kind of cooperative frequency spectrum sensing method and relevant apparatus | |
| CN109309538A (en) | A kind of frequency spectrum sensing method, device, equipment, system and storage medium | |
| CN104463896A (en) | Image corner point detection method and system based on kernel similar region distribution characteristics | |
| CN104680118A (en) | Method and system for generating face character detection model | |
| CN108761385A (en) | A kind of indoor location localization method carrying out fingerprint point cluster based on AP virtual coordinates |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181218 |