CN111817803A - Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium - Google Patents

Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium Download PDF

Info

Publication number
CN111817803A
CN111817803A CN202010555460.6A CN202010555460A CN111817803A CN 111817803 A CN111817803 A CN 111817803A CN 202010555460 A CN202010555460 A CN 202010555460A CN 111817803 A CN111817803 A CN 111817803A
Authority
CN
China
Prior art keywords
signal
correlation coefficient
clustering algorithm
matrix
spectrum sensing
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
Application number
CN202010555460.6A
Other languages
Chinese (zh)
Inventor
卢致坚
徐维超
陈昌润
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Guangdong University of Technology
Original Assignee
Guangdong University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangdong University of Technology filed Critical Guangdong University of Technology
Priority to CN202010555460.6A priority Critical patent/CN111817803A/en
Publication of CN111817803A publication Critical patent/CN111817803A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • HELECTRICITY
    • H04ELECTRIC COMMUNICATION TECHNIQUE
    • H04BTRANSMISSION
    • H04B17/00Monitoring; Testing
    • H04B17/30Monitoring; Testing of propagation channels
    • H04B17/382Monitoring; Testing of propagation channels for resource allocation, admission control or handover
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/23Clustering techniques
    • G06F18/232Non-hierarchical techniques
    • G06F18/2321Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/08Feature extraction
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00Aspects of pattern recognition specially adapted for signal processing
    • G06F2218/12Classification; Matching

Abstract

The invention discloses a frequency spectrum sensing method, a frequency spectrum sensing system and a computer readable storage medium based on correlation coefficients and a K-means clustering algorithm, wherein the method comprises the following steps: s1: establishing a mixed Gaussian model, and taking the model as a noise distribution function; s2, establishing an nth signal sample model received by the ith secondary user; s3: receiving a sensing signal acquired by each secondary user from a known environment or an unknown environment to obtain a sensing signal matrix; s4: performing ordinal decomposition and recombination on the sensing signal matrix based on a random matrix theory to obtain two new signal matrices; s5: calculating correlation coefficients among the perception signals received by different secondary users, and deriving a two-dimensional feature vector according to the correlation coefficients; s6: and clustering and dividing the sensing signal characteristics by using a K-means clustering algorithm to determine whether a main user transmits signals or not, thereby realizing frequency spectrum sensing. The invention solves the problem of frequency spectrum sensing of weak signals under pulse noise.

Description

Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium
Technical Field
The invention relates to the technical field of cognitive radio, in particular to a frequency spectrum sensing method and system based on correlation coefficients and a K-means clustering algorithm and a computer readable storage medium.
Background
The key idea of cognitive radio is the multiplexing and sharing of frequency spectrum, that is, a cognitive user can utilize an idle frequency band of an authorized user (master user) on the premise of not causing interference to the authorized user. Therefore, the cognitive user needs to continuously detect the master user signal in the frequency band, and once the master user signal is found, the master user signal is avoided as soon as possible so as not to interfere the master user, otherwise, the frequency band can be used for communication. The above detection process is "spectrum sensing". It is easy to see that spectrum sensing is a key technology and implementation basis of cognitive radio.
At present, the conventional spectrum sensing algorithm mainly includes matched filter detection, energy detection, Likelihood Ratio detection (LRT), cyclostationary detection, and the like. Under the condition of stable Gaussian noise, if authorized user information is known, the output signal-to-noise ratio can be maximized through matched filtering detection, but the algorithm not only needs the characteristic prior information of a main user signal, but also needs accurate synchronization information of the main user and a cognitive user, and the algorithm is difficult to achieve in practical application. The energy detection is a relatively common spectrum sensing algorithm, and the cognitive user judges whether the master user signal exists or not by comparing the energy of the received signal with a preset judgment threshold value. The energy detection algorithm has low computational complexity and is easy to implement, the prior information of a main user signal is not needed, but the detection performance is greatly influenced under the condition of low signal-to-noise ratio, the threshold value is determined according to the power information of noise, and the uncertainty of the noise power seriously influences the detection probability and the false alarm probability. Likelihood ratio detection is an optimal detection method obtained by the nemann-pearson criterion, and is performed by using a difference between Probability Density Functions (PDFs) of a primary user signal and noise, but the algorithm requires the PDFs as prior information. The cyclostationary detection judges whether the main user signal exists or not by analyzing a power spectral density equation according to the inherent cyclostationary characteristic of the main user signal, the cyclostationary characteristics of different modulation signals have obvious difference, and noise and interference do not have the cyclostationary characteristic. The cyclostationary detection algorithm can overcome the defects of fading and energy detection under low signal-to-noise ratio, has good detection performance, needs larger data volume and has high calculation complexity.
Compared with the conditions required by the above traditional spectrum sensing algorithm, in a real environment, noise often appears in an unknown or incompletely known state, and the corresponding probability density function has a longer tail compared with the PDF with a Gaussian distribution. Among them, impulse noise is one of the most typical environmental noises, and may be generated by cloud discharge, ice layer crack, biological activity, etc. Due to the impulse noise mixed into the received signal, the signal-to-noise ratio drops rapidly over a corresponding period of time. In this case, the performance of the conventional spectrum sensing algorithm is also rapidly reduced and even fails. In the prior art, an invention patent with a publication number of CN103856946A discloses a dual-threshold cooperative spectrum sensing method based on differential energy detection, which solves the problems that an energy threshold in energy detection is easily affected by noise uncertainty and a sensing failure exists in a conventional dual-threshold method. Firstly, locally perceiving each cognitive user, on the basis of traditional double-threshold energy detection, adopting differential energy detection on energy values between double thresholds, and adopting traditional energy detection on energy values outside the double thresholds; and then sending the local judgment result of each cognitive user to a system fusion center, and obtaining a total system judgment result by adopting an OR fusion criterion.
Disclosure of Invention
The invention provides a spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and a computer readable storage medium, aiming at overcoming the defect that the traditional spectrum sensing method in the prior art does not solve the spectrum sensing problem of weak signals under pulse noise.
The primary objective of the present invention is to solve the above technical problems, and the technical solution of the present invention is as follows:
the invention provides a frequency spectrum sensing method based on correlation coefficient and K-mean value clustering algorithm, comprising the following steps:
s1: establishing a mixed Gaussian model, and taking the model as a noise distribution function;
s2, establishing an nth signal sample model received by the ith secondary user;
s3: receiving a sensing signal acquired by each secondary user from a known environment or an unknown environment to obtain a sensing signal matrix;
s4: performing ordinal decomposition and recombination on the sensing signal matrix based on a random matrix theory to obtain two new signal matrices;
s5: calculating correlation coefficients among the perception signals received by different secondary users, and deriving a two-dimensional feature vector according to the correlation coefficients;
s6: and clustering and dividing the sensing signal characteristics by using a K-means clustering algorithm to determine whether a main user transmits signals or not, thereby realizing frequency spectrum sensing.
In this scheme, the gaussian mixture model is represented as:
Figure BDA0002544137120000031
where w is the noise received by the secondary user,11,
Figure BDA0002544137120000032
is the mathematical expectation, mean and variance of the conventional noise,22,
Figure BDA0002544137120000033
are the mathematical expectation, mean and variance of the impulse noise, where,1+2=1, 0<2<<1,σ2→∞。
in this scheme, the establishing of the nth signal sample model received by the ith secondary user is represented as follows:
xi(n)=his(n)+wi(n)
wherein, i is 1, 2.. times.m; s (n) represents a master user transmission signal; h isi≧ 0 is a constant which is a parameter representing the signal strength of the primary user when hi0 means that the primary user transmission signal is absent, hiThe signal transmitted by the master user exists when the signal is more than 0; w is ai(n) represents samples of non-gaussian background noise,
set H0Indicating absence of primary user signal, H1Indicating the existence of a main user signal, defining the channel occupancy rate as:
Figure BDA0002544137120000034
in the scheme, the construction of the sensing signal matrix specifically comprises the following steps:
setting the number of sampling points of each secondary user as N and the number of secondary users as M, and obtaining a perception matrix X according to the nth signal sample model received by the ith secondary user in the step S2, wherein the perception matrix X is expressed as:
Figure BDA0002544137120000035
wherein the elements in the X matrix represent the perceptual signals.
In the scheme, ordinal number decomposition and recombination are carried out on the sensing signal matrix to obtain two new signal matrices, specifically:
decomposing and recombining the sensing signal matrix X to respectively obtain two signal matrixes, which are marked as Y1And Y2Expressed as:
Figure BDA0002544137120000041
Figure BDA0002544137120000042
decomposing an MxN sensing signal matrix, and recombining two Mqxk matrixes Y to obtain two Mqxk matrixes Y1And Y2The number of M secondary users is expanded to Mq, and the number of the secondary users is increased.
In the scheme,Y1Obtained by ordinal number decomposition and recombination, Y2Obtained by interval decomposition and recombination.
In the scheme, correlation coefficients between perception signals received by different users are calculated, and a two-dimensional feature vector is derived according to the correlation coefficients, specifically:
are respectively aligned with matrix Y1And Y2Extracting signal characteristics, wherein each row of the matrix represents a sample signal received by one secondary user, and acquiring a spearman rank order correlation coefficient of signal samples among different secondary users;
the process of constructing a calculation formula of the spearman rank order correlation coefficient is as follows:
respectively receiving sample signals x received by the ith secondary useriAnd the sample signal x received by the jth userjReordering to obtain a new sequence xi(1)<xi(2)<...<xi(k) And xj(1)<xj(2)<...<xj(k) Let x bei(a) (a ═ 1,2,. k) and xj(a) (a 1, 2.. k.) in the new sequence, respectively, at pi(a) A sum pj(a) Then, the spearman rank correlation coefficient is calculated as follows:
Figure 100002_1
calculating matrix Y according to the spearman rank order correlation coefficient formula1And Y2The spearman rank order correlation coefficients of different rows, and each matrix obtains MqX (Mq-1)/2 spearman rank order correlation coefficients;
the obtained spearman rank order correlation coefficient TSR(ij) the transformation is performed as follows:
Figure BDA0002544137120000051
setting L as Mq x (Mq-1)/2, and transforming L spearman rank order correlation coefficients to obtain L alphaijL is alphaijAnd averaging to obtain a decision statistical characteristic T:
Figure BDA0002544137120000052
respectively calculate to obtain matrix Y1And Y2Is marked as T1And T2The two statistical features form a two-dimensional decision statistical feature vector T ═ T1,T2]T
In the scheme, the sensing signal characteristics are clustered and divided by using a K-means clustering algorithm, whether a main user transmits a signal is determined, and spectrum sensing is realized, and the scheme specifically comprises the following steps:
a plurality of two-dimensional decision statistical feature vectors are obtained through the step S5, and the obtained two-dimensional decision statistical feature vectors are divided into a training set and a test set, wherein the training set is marked as D for training the classifier model, and the test set is marked as D
Figure BDA0002544137120000053
Where the training set and test set are represented as follows:
D={T1,T2,...,TB},
Figure BDA0002544137120000054
wherein T isb(B ═ 1, 2.., B) represents the decision statistic, Ψ, obtained in step S5kRepresenting the set of all feature vectors in class k, CkA central set representing k classes, specifically:
Ψk={Tb|Tbe.g. class K (K1, 2.., K)
Figure BDA0002544137120000055
Wherein n (Ψ)k) Denotes ΨkThe number of medium feature vectors and the cost function of the K-means are as follows:
Figure BDA0002544137120000056
the minimization objective function in the training process is:
Figure BDA0002544137120000057
after training is completed, the following formula is used for judging:
Figure BDA0002544137120000058
wherein, C1A category representing that a primary user does not exist; and xi is a measurement parameter larger than zero, when the test set meets the formula, S-1 exists in the main user signal, and otherwise, S-1 does not exist in the main user signal.
The second aspect of the present invention provides a spectrum sensing system based on correlation coefficient and K-means clustering algorithm, the system comprising: the device comprises a memory and a processor, wherein the memory comprises a spectrum sensing method program based on a correlation coefficient and a K-means clustering algorithm, and the spectrum sensing method program based on the correlation coefficient and the K-means clustering algorithm realizes the following steps when being executed by the processor:
s1: establishing a mixed Gaussian model, and taking the model as a noise distribution function;
s2, establishing an nth signal sample model received by the ith secondary user;
s3: receiving a sensing signal acquired by each secondary user from a known environment to obtain a sensing signal matrix;
s4: performing ordinal decomposition and recombination on the sensing signal matrix based on a random matrix theory to obtain two new signal matrices;
s5: calculating correlation coefficients among the perception signals received by different secondary users, and deriving a two-dimensional feature vector according to the correlation coefficients;
s6: and clustering and dividing the sensing signal characteristics by using a K-means clustering algorithm to determine whether a main user transmits signals or not, thereby realizing frequency spectrum sensing.
The third aspect of the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a program of a spectrum sensing method based on a correlation coefficient and a K-means clustering algorithm, and when the program of the spectrum sensing method based on the correlation coefficient and the K-means clustering algorithm is executed by a processor, the steps of the spectrum sensing method based on the correlation coefficient and the K-means clustering algorithm are implemented.
Compared with the prior art, the technical scheme of the invention has the beneficial effects that:
according to the method, a mixed Gaussian noise model is built, and spectrum sensing is performed by using a sampling signal matrix correlation coefficient and machine learning unsupervised clustering algorithm, so that the spectrum sensing problem of weak signals under impulse noise is solved, and the method has good detection performance under the conditions of low signal-to-noise ratio of the impulse noise and less quantity of secondary users.
Drawings
FIG. 1 is a flow chart of the method of the present invention.
Fig. 2 is a schematic diagram of a cooperative spectrum sensing system model in the present invention.
Detailed Description
In order that the above objects, features and advantages of the present invention can be more clearly understood, a more particular description of the invention will be rendered by reference to the appended drawings. It should be noted that the embodiments and features of the embodiments of the present application may be combined with each other without conflict.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, however, the present invention may be practiced in other ways than those specifically described herein, and therefore the scope of the present invention is not limited by the specific embodiments disclosed below.
Example 1
As shown in fig. 1, a first aspect of the present invention provides a spectrum sensing method based on correlation coefficients and a K-means clustering algorithm, including the following steps:
s1: establishing a mixed Gaussian model, and taking the model as a noise distribution function;
a suitable model of normal distributed thermal noise, but impulse noise is one of the most typical environmental noises in practical environments, and can be generated due to cloud layer discharge, ice layer rupture, biological activity and the like, and a spectrum sensing system is affected by non-Gaussian noise with impulse characteristics. Relative to the probability density function of the gaussian distribution, the gaussian mixture model has a longer tail, which is expressed as:
Figure BDA0002544137120000071
where w is the noise received by the secondary user,11,
Figure BDA0002544137120000072
is the mathematical expectation, mean and variance of the conventional noise,22,
Figure BDA0002544137120000073
are the mathematical expectation, mean and variance of the impulse noise, where,1+2=1, 0<2<<1,σ2→∞。
it should be noted that, in the above expression of the gaussian mixture model, the non-gaussian background is regarded as a superposition of most of the conventional noise with a smaller amplitude and a small part of the impulse noise with a larger amplitude (random pulses such as reverberation and clutter). The function is to find the 'worst distribution pair' according to a certain criterion, and then to process and find the best detection under the worst distribution according to a certain method, so that the interference is not sensitive to the change of the noise statistical characteristics (such as signal distortion), thereby ensuring the robustness of the system.
It should be noted that the present invention is developed based on the gaussian mixture model established in step 1, which is a signal characteristic in the subsequent step, such as the sample sequence w of non-gaussian background noise in step 2.
S2, establishing an nth signal sample model received by the ith secondary user;
is represented as follows:
xi(n)=his(n)+wi(n)
wherein, i is 1, 2.. times.m; s (n) transmitting signals on behalf of primary usersNumber; h isi≧ 0 is a constant which is a parameter representing the signal strength of the primary user when hi0 means that the primary user transmission signal is absent, hiThe signal transmitted by the master user exists when the signal is more than 0; w is ai(n) represents samples of non-gaussian background noise,
set H0Indicating absence of primary user signal, H1Indicating the existence of a main user signal, defining the channel occupancy rate as:
Figure BDA0002544137120000081
s3: receiving a sensing signal acquired by each secondary user from a known environment or an unknown environment to obtain a sensing signal matrix;
spectrum sensing technologies can be classified into two types, namely Single Node Sensing (SNS) and CSS, according to the number of Secondary Users (SUs). Css (cooperative Spectrum sensing) indicates that a plurality of secondary users detect a wireless channel and simultaneously transmit a received signal to a Fusion Center (FC) to uniformly process a decision. Since the geographical location of each SU is different in an actual wireless network, the communication environment may be affected by factors such as shadowing, multipath fading, and hidden terminals, resulting in different detection accuracies of different SUs. Compared with the SNS, the CSS makes full use of the sensing environment diversity of the SU, and avoids errors caused by environmental factors, such as multipath fading and shadows. The CSS improves the detection accuracy and is widely applied. Therefore, in order to guarantee the spectrum sensing performance, the spectrum sensing is performed in a cooperative sensing manner, as shown in fig. 2. Specifically, the system receives the sensing signals acquired by each antenna of each secondary user, and further obtains a sensing signal matrix corresponding to each secondary user according to the sensing signals, so that feature vector extraction and judgment of the existence of a primary user are subsequently performed based on the sensing signals acquired by each secondary user.
In the scheme, the construction of the sensing signal matrix specifically comprises the following steps:
setting the number of sampling points of each secondary user as N and the number of secondary users as M, and obtaining a perception matrix X according to the nth signal sample model received by the ith secondary user in the step S2, wherein the perception matrix X is expressed as:
Figure BDA0002544137120000082
wherein the elements in the X matrix represent the perceptual signals.
S4: performing ordinal decomposition and recombination on the sensing signal matrix based on a random matrix theory to obtain two new signal matrices;
when the number M of the secondary users is small and the dimension of the sampling matrix is low, the detection performance is not ideal. Preprocessing of the sampled signals by Decomposition And Recombination (DAR) of the introduced signal matrix logically increases the number of SUS. DAR can be divided into Ordinal-decompensation And recombination (O-DAR) And Interval-decompensation And recombination (I-DAR) depending on the division method.
In the scheme, ordinal number decomposition and recombination are carried out on the sensing signal matrix to obtain two new signal matrices, specifically:
decomposing and recombining the sensing signal matrix X to respectively obtain two signal matrixes, which are marked as Y1And Y2Expressed as:
Figure BDA0002544137120000091
Figure BDA0002544137120000092
decomposing an MxN sensing signal matrix, and recombining two Mqxk matrixes Y to obtain two Mqxk matrixes Y1And Y2The number of M secondary users is expanded to Mq, and the number of the secondary users is increased. Y is1Obtained by ordinal number decomposition and recombination, Y2Obtained by interval decomposition and recombination.
S5: calculating correlation coefficients among the perception signals received by different secondary users, and deriving a two-dimensional feature vector according to the correlation coefficients;
in the present invention, when there is impulse interference in the sample, the spearman rank correlation coefficient or the kender rank correlation coefficient is the most suitable choice because their mean and variance are only slightly affected by the magnitude of the impulse component fraction, and the spearman rank correlation coefficient or the kender rank correlation coefficient has strong robustness to impulse noise. When the signal is weak and the number of signal samples is not large, the spearman rank correlation coefficient is superior to the Kendall rank correlation coefficient in terms of relative efficiency, so the spearman rank correlation coefficient is selected in the scheme to extract the signal characteristics.
In the scheme, correlation coefficients between perception signals received by different users are calculated, and a two-dimensional feature vector is derived according to the correlation coefficients, specifically:
are respectively aligned with matrix Y1And Y2Extracting signal characteristics, wherein each row of the matrix represents a sample signal received by one secondary user, and acquiring a spearman rank order correlation coefficient of signal samples among different secondary users;
the process of constructing a calculation formula of the spearman rank order correlation coefficient is as follows:
respectively receiving sample signals x received by the ith secondary useriAnd the sample signal x received by the jth userjReordering to obtain a new sequence xi(1)<xi(2)<...<xi(k) And xj(1)<xj(2)<...<xj(k) Let x bei(a) (a ═ 1,2,. k) and xj(a) (a 1, 2.. k.) in the new sequence, respectively, at pi(a) A sum pj(a) Then, the spearman rank correlation coefficient is calculated as follows:
Figure 2
calculating matrix Y according to the spearman rank order correlation coefficient formula1And Y2The spearman rank order correlation coefficients of different rows, and each matrix obtains MqX (Mq-1)/2 spearman rank order correlation coefficients;
subjecting the obtained spearman toRank order correlation coefficient TSR(ij) the transformation is performed as follows:
Figure BDA0002544137120000102
setting L as Mq x (Mq-1)/2, and transforming L spearman rank order correlation coefficients to obtain L alphaijL is alphaijAnd averaging to obtain a decision statistical characteristic T:
Figure BDA0002544137120000103
respectively calculate to obtain matrix Y1And Y2Is marked as T1And T2The two statistical features form a two-dimensional decision statistical feature vector T ═ T1,T2]T
S6: and clustering and dividing the sensing signal characteristics by using a K-means clustering algorithm to determine whether a main user transmits signals or not, thereby realizing frequency spectrum sensing.
It should be noted that in the conventional method, the calculation of the threshold derivation is complicated. When the number of samples is limited, the threshold estimation error increases, affecting the overall performance. In the scheme, the problem of calculating the threshold error is solved by combining machine learning and frequency spectrum sensing. The invention classifies the signals by adopting a K-means algorithm.
In the scheme, the sensing signal characteristics are clustered and divided by using a K-means clustering algorithm, whether a main user transmits a signal is determined, and spectrum sensing is realized, and the scheme specifically comprises the following steps:
a plurality of two-dimensional decision statistical feature vectors are obtained through the step S5, and the obtained two-dimensional decision statistical feature vectors are divided into a training set and a test set, wherein the training set is marked as D for training the classifier model, and the test set is marked as D, and the training set and the test set are expressed as follows:
D={T1,T2,...,TB},
Figure BDA0002544137120000111
wherein T isb(B ═ 1, 2.., B) represents the decision statistic, Ψ, obtained in step S5kRepresenting the set of all feature vectors in class k, CkA central set representing k classes, specifically:
Ψk={Tb|Tbe.g. class K (K1, 2.., K)
Figure BDA0002544137120000112
Wherein n (Ψ)k) Denotes ΨkThe number of medium feature vectors and the cost function of the K-means are as follows:
Figure BDA0002544137120000113
the minimization objective function in the training process is:
Figure BDA0002544137120000114
after training is completed, the following formula is used for judging:
Figure BDA0002544137120000115
wherein, C1A category representing that a primary user does not exist; and xi is a measurement parameter larger than zero, when the test set meets the formula, the main user signal exists in the place where S is equal to 1, otherwise, the main user signal does not exist in the place where S is equal to-1.
The second aspect of the present invention provides a spectrum sensing system based on correlation coefficient and K-means clustering algorithm, the system comprising: the device comprises a memory and a processor, wherein the memory comprises a spectrum sensing method program based on a correlation coefficient and a K-means clustering algorithm, and the spectrum sensing method program based on the correlation coefficient and the K-means clustering algorithm realizes the following steps when being executed by the processor:
s1: establishing a mixed Gaussian model, and taking the model as a noise distribution function;
s2, establishing an nth signal sample model received by the ith secondary user;
s3: receiving a sensing signal acquired by each secondary user from a known environment to obtain a sensing signal matrix;
s4: performing ordinal decomposition and recombination on the sensing signal matrix based on a random matrix theory to obtain two new signal matrices;
s5: calculating correlation coefficients among the perception signals received by different secondary users, and deriving a two-dimensional feature vector according to the correlation coefficients;
s6: and clustering and dividing the sensing signal characteristics by using a K-means clustering algorithm to determine whether a main user transmits signals or not, thereby realizing frequency spectrum sensing.
The third aspect of the present invention provides a computer-readable storage medium, where the computer-readable storage medium includes a program of a spectrum sensing method based on a correlation coefficient and a K-means clustering algorithm, and when the program of the spectrum sensing method based on the correlation coefficient and the K-means clustering algorithm is executed by a processor, the steps of the spectrum sensing method based on the correlation coefficient and the K-means clustering algorithm are implemented.
It should be understood that the above-described embodiments of the present invention are merely examples for clearly illustrating the present invention, and are not intended to limit the embodiments of the present invention. Other variations and modifications will be apparent to persons skilled in the art in light of the above description. And are neither required nor exhaustive of all embodiments. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the claims of the present invention.

Claims (10)

1. A frequency spectrum sensing method based on correlation coefficient and K-means clustering algorithm is characterized by comprising the following steps:
s1: establishing a mixed Gaussian model, and taking the model as a noise distribution function;
s2, establishing an nth signal sample model received by the ith secondary user;
s3: receiving a sensing signal acquired by each secondary user from a known environment or an unknown environment to obtain a sensing signal matrix;
s4: performing ordinal decomposition and recombination on the sensing signal matrix based on a random matrix theory to obtain two new signal matrices;
s5: calculating correlation coefficients among the perception signals received by different secondary users, and deriving a two-dimensional feature vector according to the correlation coefficients;
s6: and clustering and dividing the sensing signal characteristics by using a K-means clustering algorithm to determine whether a main user transmits signals or not, thereby realizing frequency spectrum sensing.
2. The spectrum sensing method based on correlation coefficient and K-means clustering algorithm according to claim 1, wherein the Gaussian mixture model is expressed as:
Figure FDA0002544137110000011
where w is the noise received by the secondary user,11,
Figure FDA0002544137110000012
is the mathematical expectation, mean and variance of the conventional noise,22,
Figure FDA0002544137110000013
are the mathematical expectation, mean and variance of the impulse noise, where,1+2=1,0<2<<1,σ2→∞。
3. the spectrum sensing method based on correlation coefficient and K-means clustering algorithm according to claim 2, wherein the model for the nth signal sample received by the ith secondary user is established as follows:
xi(n)=his(n)+wi(n)
wherein1, 2.·, M; s (n) represents a master user transmission signal; h isi≧ 0 is a constant which is a parameter representing the signal strength of the primary user when hi0 means that the primary user transmission signal is absent, hiThe signal transmitted by the master user exists when the signal is more than 0; w is ai(n) represents samples of non-gaussian background noise,
set H0Indicating absence of primary user signal, H1Indicating the existence of a main user signal, defining the channel occupancy rate as:
Figure FDA0002544137110000021
4. the spectrum sensing method based on the correlation coefficient and the K-means clustering algorithm according to claim 3, wherein the construction of the sensing signal matrix is specifically as follows:
setting the number of sampling points of each secondary user as N and the number of secondary users as M, and obtaining a perception matrix X according to the nth signal sample model received by the ith secondary user in the step S2, wherein the perception matrix X is expressed as:
Figure FDA0002544137110000022
wherein the elements in the X matrix represent the perceptual signals.
5. The spectrum sensing method based on correlation coefficient and K-means clustering algorithm as claimed in claim 4, wherein the sensing signal matrix is subjected to ordinal decomposition and recombination to obtain two new signal matrices, specifically:
decomposing and recombining the sensing signal matrix X to respectively obtain two signal matrixes, which are marked as Y1And Y2Expressed as:
Figure FDA0002544137110000023
Figure FDA0002544137110000024
decomposing an MxN sensing signal matrix, and recombining two Mqxk matrixes Y to obtain two Mqxk matrixes Y1And Y2The number of M secondary users is expanded to Mq, and the number of the secondary users is increased.
6. The spectrum sensing method based on correlation coefficient and K-means clustering algorithm as claimed in claim 5, wherein Y is1Obtained by ordinal number decomposition and recombination, Y2Obtained by interval decomposition and recombination.
7. The spectrum sensing method based on correlation coefficient and K-means clustering algorithm according to claim 6, wherein the correlation coefficient between the sensing signals received by different secondary users is calculated, and a two-dimensional feature vector is derived therefrom, specifically:
are respectively aligned with matrix Y1And Y2Extracting signal characteristics, wherein each row of the matrix represents a sample signal received by one secondary user, and acquiring a spearman rank order correlation coefficient of signal samples among different secondary users;
the process of constructing a calculation formula of the spearman rank order correlation coefficient is as follows:
respectively receiving sample signals x received by the ith secondary useriAnd the sample signal x received by the jth userjReordering to obtain a new sequence xi(1)<xi(2)<...<xi(k) And xj(1)<xj(2)<...<xj(k) Let x bei(a) (a ═ 1,2,. k) and xj(a) (a 1, 2.. k.) in the new sequence, respectively, at pi(a) A sum pj(a) Then, the spearman rank correlation coefficient is calculated as follows:
Figure 1
according to the formula of the spearman rank order correlation coefficientMatrix Y1And Y2The spearman rank order correlation coefficients of different rows, and each matrix obtains MqX (Mq-1)/2 spearman rank order correlation coefficients;
the obtained spearman rank order correlation coefficient TSR(ij) the transformation is performed as follows:
Figure FDA0002544137110000032
setting L as Mq x (Mq-1)/2, and transforming L spearman rank order correlation coefficients to obtain L alphaijL is alphaijAnd averaging to obtain a decision statistical characteristic T:
Figure FDA0002544137110000033
respectively calculate to obtain matrix Y1And Y2Is marked as T1And T2The two statistical features form a two-dimensional decision statistical feature vector T ═ T1,T2]T
8. The spectrum sensing method based on correlation coefficients and a K-means clustering algorithm according to claim 7, wherein the sensing signal features are clustered and divided by the K-means clustering algorithm to determine whether a main user transmission signal exists or not, so as to realize spectrum sensing, and specifically:
a plurality of two-dimensional decision statistical feature vectors are obtained through the step S5, and the obtained two-dimensional decision statistical feature vectors are divided into a training set and a test set, wherein the training set is marked as D for training the classifier model, and the test set is marked as D
Figure FDA0002544137110000041
Where the training set and test set are represented as follows:
Figure FDA0002544137110000042
wherein T isb(B ═ 1, 2.., B) represents the decision statistic, Ψ, obtained in step S5kRepresenting the set of all feature vectors in class k, CkA central set representing k classes, specifically:
Ψk={Tb|Tbe.g. class K (K1, 2.., K)
Figure FDA0002544137110000043
Wherein n (Ψ)k) Denotes ΨkThe number of medium feature vectors and the cost function of the K-means are as follows:
Figure FDA0002544137110000044
the minimization objective function in the training process is:
Figure FDA0002544137110000045
after training is completed, the following formula is used for judging:
Figure FDA0002544137110000046
wherein, C1A category representing that a primary user does not exist; and xi is a measurement parameter larger than zero, when the test set meets the formula, S-1 exists in the main user signal, and otherwise, S-1 does not exist in the main user signal.
9. A spectrum sensing system based on correlation coefficient and K-means clustering algorithm is characterized by comprising: the device comprises a memory and a processor, wherein the memory comprises a spectrum sensing method program based on a correlation coefficient and a K-means clustering algorithm, and the spectrum sensing method program based on the correlation coefficient and the K-means clustering algorithm realizes the following steps when being executed by the processor:
s1: establishing a mixed Gaussian model, and taking the model as a noise distribution function;
s2, establishing an nth signal sample model received by the ith secondary user;
s3: receiving a sensing signal acquired by each secondary user from a known environment to obtain a sensing signal matrix;
s4: performing ordinal decomposition and recombination on the sensing signal matrix based on a random matrix theory to obtain two new signal matrices;
s5: calculating correlation coefficients among the perception signals received by different secondary users, and deriving a two-dimensional feature vector according to the correlation coefficients;
s6: and clustering and dividing the sensing signal characteristics by using a K-means clustering algorithm to determine whether a main user transmits signals or not, thereby realizing frequency spectrum sensing.
10. A computer-readable storage medium, wherein the computer-readable storage medium includes a program of a spectrum sensing method based on a correlation coefficient and a K-means clustering algorithm, and when the program of the spectrum sensing method based on the correlation coefficient and the K-means clustering algorithm is executed by a processor, the steps of the spectrum sensing method based on the correlation coefficient and the K-means clustering algorithm according to any one of claims 1 to 8 are implemented.
CN202010555460.6A 2020-06-17 2020-06-17 Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium Pending CN111817803A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010555460.6A CN111817803A (en) 2020-06-17 2020-06-17 Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010555460.6A CN111817803A (en) 2020-06-17 2020-06-17 Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium

Publications (1)

Publication Number Publication Date
CN111817803A true CN111817803A (en) 2020-10-23

Family

ID=72844792

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010555460.6A Pending CN111817803A (en) 2020-06-17 2020-06-17 Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium

Country Status (1)

Country Link
CN (1) CN111817803A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114330468A (en) * 2021-07-14 2022-04-12 广东工业大学 Classifier screening method and system based on dynamic programming and computer equipment
CN114422548A (en) * 2021-12-30 2022-04-29 扬州万方科技股份有限公司 Data sensing method in Internet of things
CN115021845A (en) * 2022-06-20 2022-09-06 西安航空学院 Spectrum sensing method, system, medium, device and terminal

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104821854A (en) * 2015-03-03 2015-08-05 北京科技大学 Multidimensional spectrum sensing method for multiple main users based on random sets
KR20170090104A (en) * 2016-01-28 2017-08-07 한국과학기술원 Apparatus and method for spatial channel parameter estimation in fifth generation mobile communications
US20190036629A1 (en) * 2017-07-30 2019-01-31 Nec Laboratories America, Inc. System and Method for Enabling Interference-awareness for LTE Operation in Unlicensed Spectrum
CN109981200A (en) * 2019-04-12 2019-07-05 广东工业大学 Frequency spectrum sensing method, device, equipment and storage medium based on related coefficient

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104821854A (en) * 2015-03-03 2015-08-05 北京科技大学 Multidimensional spectrum sensing method for multiple main users based on random sets
KR20170090104A (en) * 2016-01-28 2017-08-07 한국과학기술원 Apparatus and method for spatial channel parameter estimation in fifth generation mobile communications
US20190036629A1 (en) * 2017-07-30 2019-01-31 Nec Laboratories America, Inc. System and Method for Enabling Interference-awareness for LTE Operation in Unlicensed Spectrum
CN109981200A (en) * 2019-04-12 2019-07-05 广东工业大学 Frequency spectrum sensing method, device, equipment and storage medium based on related coefficient

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
卢光跃等: "基于LSTM神经网络的频谱感知算法", 《信号处理》 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114330468A (en) * 2021-07-14 2022-04-12 广东工业大学 Classifier screening method and system based on dynamic programming and computer equipment
CN114422548A (en) * 2021-12-30 2022-04-29 扬州万方科技股份有限公司 Data sensing method in Internet of things
CN115021845A (en) * 2022-06-20 2022-09-06 西安航空学院 Spectrum sensing method, system, medium, device and terminal
CN115021845B (en) * 2022-06-20 2023-04-18 西安航空学院 Spectrum sensing method, system, medium, device and terminal

Similar Documents

Publication Publication Date Title
CN108629380B (en) Cross-scene wireless signal sensing method based on transfer learning
CN111817803A (en) Frequency spectrum sensing method and system based on correlation coefficient and K-means clustering algorithm and computer readable storage medium
CN109379153B (en) Spectrum sensing method
CN109450834A (en) Signal of communication classifying identification method based on Multiple feature association and Bayesian network
CN109672485B (en) Indoor personnel real-time invasion and movement speed detection method based on channel state information
CN107820255B (en) Improved covariance absolute value cooperative spectrum sensing method
CN108038419B (en) Wi-Fi-based indoor personnel passive detection method
EP4113174A1 (en) Estimation device, estimation method, and program
CN112749633B (en) Separate and reconstructed individual radiation source identification method
CN109450573B (en) frequency spectrum sensing method based on deep neural network
CN114268388B (en) Channel estimation method based on improved GAN network in large-scale MIMO
CN113673312A (en) Radar signal intra-pulse modulation identification method based on deep learning
CN110290466B (en) Floor distinguishing method, device, equipment and computer storage medium
CN115661869A (en) Radio frequency signal fingerprint identification method and system based on multi-dimensional feature intelligent fusion
CN112350790B (en) Deep learning-based spectrum sensing detection method, device and equipment
CN108834043B (en) Priori knowledge-based compressed sensing multi-target passive positioning method
CN115017958A (en) Non-line-of-sight signal identification method based on channel characteristic weighting model
CN115002703A (en) Passive indoor people number detection method based on Wi-Fi channel state information
CN108718223A (en) A kind of blind frequency spectrum sensing method of non-co-operation signal
CN110784887B (en) Method for detecting number of abnormal signal sources in gridding radio signal monitoring system
CN112666528A (en) Multi-station radar system interference identification method based on convolutional neural network
Li et al. A spectrum sensing algorithm based on correlation coefficient and K-means
CN110912843A (en) Distributed blind estimation method and system in large-scale wireless sensor network
CN116340849B (en) Non-contact type cross-domain human activity recognition method based on metric learning
Reddy et al. Novel receiver diversity combining methods for spectrum sensing using meta-analytic approach based on P-values

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

Application publication date: 20201023

RJ01 Rejection of invention patent application after publication