CN107707497A - Communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm - Google Patents
Communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm Download PDFInfo
- Publication number
- CN107707497A CN107707497A CN201710319971.6A CN201710319971A CN107707497A CN 107707497 A CN107707497 A CN 107707497A CN 201710319971 A CN201710319971 A CN 201710319971A CN 107707497 A CN107707497 A CN 107707497A
- Authority
- CN
- China
- Prior art keywords
- signal
- fuzzy clustering
- communication
- fuzzy
- clustering
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L27/00—Modulated-carrier systems
- H04L27/0012—Modulated-carrier systems arrangements for identifying the type of modulation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
Abstract
The invention discloses a kind of communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm, it includes initiation parameter；For different sub-carrier, different initial field radius values is set, the constellation point of the signal of communication of reception clustered using subtraction clustering algorithm；When the number at subtractive clustering center is less than the first presetting threshold values, the radius of neighbourhood is reduced, continues subtractive clustering；The first presetting threshold values subtractive clustering center larger using subtractive clustering center Midst density is clustered again as the initial center of fuzzy clustering algorithm using fuzzy clustering algorithm to the constellation point of signal of communication；The initial clustering number of fuzzy clustering is specified, the reasonability of joint Xie Beni indexs and the relative radius evaluation cluster of planisphere after cluster, if unreasonable, initial clustering number need to be iterated；By relative radius and standard planisphere radius ratio compared with, can draw the modulation system of signal then the classification where classical modulation signal be signal of communication classification.
Description
Technical field
The present invention relates to communication technical field, and in particular to a kind of communication letter based on subtractive clustering and fuzzy clustering algorithm
Number recognition methods.
Background technology
Modulation Signals Recognition plays the part of pivotal player in a variety of civil and military fields, such as cognitive radio, spectrum monitoring
Deng.In actual wireless communication, multipath channel can cause the distortion of signal, and signal identification becomes challenging.At present, non-
In cooperative communication, when being identified to the ofdm signal of various criterion agreement, it is important that a step be exactly effective son to OFDM
Carrier wave is identified.
In based on Modulation Identification research of the constellation clustering with neutral net using subtractive clustering and fuzzy clustering be superimposed come
Single-carrier signal is identified, the effective identification signal of energy in SNR=15dB, but the field radius value in its subtractive clustering is
Fixed value, when signal collection species to be identified is more, its discrimination can be reduced.
Believed under multipath fading environments in the research of Higher Order QAM Signals Modulation Recognition using subtractive clustering identification subcarrier
Number when, introduce signal to noise ratio in the radius of field, with the signal constellation point suitable for different dense degrees, but non-cooperating close
In communicating, signal to noise ratio is in fact unknown, it is necessary to be estimated, can so increase the amount of calculation of whole algorithm, and estimate
Also error is had after meter, error can be cumulative, finally influences discrimination, and it constructs correlation function using noise power,
And then final clusters number is controlled using this function, the complexity of whole algorithm can be increased.
The content of the invention
For above-mentioned deficiency of the prior art, the communication provided by the invention based on subtractive clustering and fuzzy clustering algorithm
Signal recognition method solves the problems, such as that existing clustering algorithm is low to communication signal recognition rate and complexity is high.
In order to reach foregoing invention purpose, the technical solution adopted by the present invention is：
A kind of communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm is provided, it includes：
Initialization field radius, Validity Function variable, the convergence threshold values and fuzzy clustering function of fuzzy clustering function
Maximum iteration；
The constellation point of the signal of communication of reception is clustered using subtraction clustering algorithm, and exports obtained multiple subtractions
Cluster centre；
When the number at subtractive clustering center is less than the first presetting threshold values, field radius is subtracted by the second presetting threshold values
It is small, and the constellation point of signal of communication is clustered using subtraction clustering algorithm again, until the number at subtractive clustering center is big
In equal to the first presetting threshold values；
Calculated as fuzzy clustering at the first presetting threshold values subtractive clustering center larger using subtractive clustering center Midst density
The initial center of method, the constellation point of signal of communication is clustered using fuzzy clustering algorithm, and export obtain it is multiple fuzzy
Cluster centre；
Distance of each fuzzy clustering center with respect to planisphere origin is calculated, and all distances are arranged in descending order, is adopted
The relative radius of signal of communication planisphere is calculated with first half distance and latter half distance；
；
Searching classical modulation signal constellation (in digital modulation) figure by fuzzy clustering center number corresponding to relative radius has identical cluster
Standard radius value corresponding to the number of center, when the difference between relative radius and standard radius value is less than the 3rd presetting threshold values
When, then the classification where classical modulation signal is the classification of signal of communication.
Beneficial effects of the present invention are：Initial field of the different communication signal in subtractive clustering half can be dynamically determined
Footpath, when the number at subtractive clustering center is less than the first presetting threshold values, field radius is reduced by the second presetting threshold values, this
Sample can obtain field radius value corresponding thereto for different modulated signals, in obtained clusters number and subtractive clustering
The heart is more accurate.
During communication signal recognition, the subtractive clustering center obtained using subtraction clustering algorithm is as fuzzy clustering
Initial center, fuzzy clustering is set to be used in Signal blind recognition, and by specified clusters number renewal process and field partly
The renewal process in footpath combines, it is possible to achieve most of signal of communication is accurately identified, and by l-G simulation test, it is obtained poly-
Class number and cluster centre are far above single clustering algorithm in the degree of accuracy and stability.
The initial cluster center of fuzzy clustering is replaced with the subtractive clustering center obtained after subtractive clustering, can be reduced fuzzy poly-
The iterations of class, and being corrected again to the center that subtractive clustering obtains using fuzzy clustering, so as to get cluster in
The heart is more reliable.
Brief description of the drawings
Fig. 1 is the flow chart of communication signal recognition method one embodiment based on subtractive clustering and fuzzy clustering algorithm.
Fig. 2 a are to obtain the imitative of subcarrier bpsk signal cluster centre using subtractive clustering, fuzzy clustering and this programme method
True figure.
Fig. 2 b are to obtain the imitative of subcarrier QPSK signal cluster centres using subtractive clustering, fuzzy clustering and this programme method
True figure.
Fig. 2 c are to obtain subcarrier 16QAM signal cluster centres using subtractive clustering, fuzzy clustering and this programme method
Analogous diagram.
Fig. 2 d are to obtain subcarrier 64QAM signal cluster centres using subtractive clustering, fuzzy clustering and this programme method
Analogous diagram.
Embodiment
The embodiment of the present invention is described below, in order to which those skilled in the art understand this hair
It is bright, it should be apparent that the invention is not restricted to the scope of embodiment, for those skilled in the art,
As long as various change in the spirit and scope of the present invention that appended claim limits and determines, these changes are aobvious and easy
See, all are using the innovation and creation of present inventive concept in the row of protection.
With reference to figure 1, Fig. 1 shows one implementation of communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm
The flow chart of example, as shown in figure 1, this method 100 includes step 101 to step 106.
In a step 101, field radius, Validity Function variable, the convergence threshold values and mould of fuzzy clustering function are initialized
Paste the maximum iteration of clustering function.When it is implemented, it is preferred that Validity Function variable OldF=10e10, fuzzy clustering letter
Several convergence threshold values ε=1e-5, the maximum iteration max_iter=100 of fuzzy clustering function.
In a step 102, the constellation point of the signal of communication of reception is clustered using subtraction clustering algorithm, and exported
The multiple subtractive clustering centers arrived.
In one embodiment of the invention, it is described that the constellation of the signal of communication of reception is clicked through using subtraction clustering algorithm
Row cluster further comprises：
Calculate the density value of each constellation point in signal of communication：
Wherein, DiFor density value；γaFor field radius；xjAnd xiFor jth, i-th of constellation point in signal of communication；N is logical
Believe the total number of constellation point in signal；Exp is e index；||xi-xj| | between i-th of constellation point and j-th of constellation point
Distance；
Using the density value of maximum as subtractive clustering center, and calculate the density value of remaining constellation point in signal of communication：
Wherein,For the maximum in last constellation point density value,For the maximum in last constellation point density value
The corresponding constellation point of value；For xiWithThe distance between；γb=1.25 γa；
The density value that all constellation points or remaining constellation point can be covered when the subtractive clustering center of acquisition is less than
During preset value, all subtractive clustering centers of acquisition are exported.
Test and find by Multi simulation running, under Gaussian channel, when SNR is 13dB, when subcarrier is bpsk signal, when
γaDuring ≈ 0.35, its Clustering Effect preferably (refers to subtractive clustering center closer to the center for standard of primary signal, clusters number
Closer to preferable clustering number mesh)；Similarly, γ is worked asaDuring ≈ 0.3, the Clustering Effect of QPSK signals is best；Work as γaDuring ≈ 0.2,
The Clustering Effect of 16QAM signals is best；Work as γaDuring ≈ 0.1, the Clustering Effect of 64QAM signals is best.
In step 103, when the number at subtractive clustering center is less than the first presetting threshold values, field radius is pressed second
Presetting threshold values reduces, and the constellation point of signal of communication is clustered using subtraction clustering algorithm again, until subtractive clustering
The number at center is more than or equal to the first presetting threshold values.
During implementation, the second presetting threshold values during preferred area radius is reduced by the second presetting threshold values is 0.01, then more
Field radius after new is new γa=γa-0.01.First presetting threshold values is preferably 2, and it is initial cluster center number really
Mesh.
At step 104, made with the first larger presetting threshold values subtractive clustering center of subtractive clustering center Midst density
For the initial center of fuzzy clustering algorithm, the constellation point of signal of communication is clustered using fuzzy clustering algorithm, and exports
The multiple fuzzy clustering centers arrived.
In step 103, it is assumed that the first presetting threshold values is 4, will when the cluster centre number after first cluster is less than 4
After the radius of neighbourhood reduces by 0.01, and the constellation point of signal of communication is clustered using subtraction clustering algorithm again, until subtracting
The number of method cluster centre is more than or equal to 4, due to being automatic cluster during subtractive clustering, often cluster once be likely to increase 1,
2,3 or more, namely increased number is indefinite every time, it is assumed that its subtractive clustering Center Number after the completion of clustering
For 6, then at step 104, during using fuzzy clustering algorithm, just subtracted with larger first 4 of subtractive clustering center Midst density
Initial center of the method cluster centre as fuzzy clustering algorithm.
In one embodiment of the invention, it is described that the constellation point of signal of communication is clustered using fuzzy clustering algorithm
Further comprise：
According to the first presetting threshold values initial center, the fuzzy clustering center of fuzzy clustering algorithm is calculated：
Wherein, viFor fuzzy clustering center；N is the total number of constellation point in signal of communication；μijFor constellation point xjIt is classified
To fuzzy clustering center viDegree of membership, 0≤μij≤1；M be fuzzy clustering algorithm fuzzy factor, m ∈ [1, ∞)；
Calculate the cost function value of fuzzy clustering algorithm：
Wherein, dij=| | vi-xj| |, it is fuzzy clustering center viWith constellation point xjThe distance between；M is fuzzy factor,
M ∈ [1, ∞)；λj, j=1 ..., n are Lagrange factor；
When convergence threshold values of the cost function value less than fuzzy clustering function, or the iterations at calculating fuzzy clustering center are small
When the maximum iteration of fuzzy clustering function, renewal constellation point is classified into the degree of membership at each fuzzy clustering center：
Wherein, dkj=| | vk-xj| |, it is fuzzy clustering center vkWith constellation point xjThe distance between；
The degree of membership that each fuzzy clustering center is classified into using the constellation point after renewal calculates fuzzy clustering algorithm
Fuzzy clustering center；
When the cost function value of calculating is more than or equal to the convergence threshold values of fuzzy clustering function, or calculate fuzzy clustering center
When iterations is equal to the maximum iteration of fuzzy clustering function, exports fuzzy clustering center and each constellation point is classified into
The degree of membership at each fuzzy clustering center.
Found by Multi simulation running, under Gaussian channel, when SNR is 13dB, to bpsk signal, QPSK signals,
During 16QAM signals and 64QAM signals carry out fuzzy clustering, it is found that as m ≈ 2, it is relative to other fuzzy factor m
Value, obtained cluster centre is all closer from the center for standard of primary signal, and obtained clusters number is also all relatively more accurate, right
Four sub-carrier signals have preferable Clustering Effect.
In step 105, distance of each fuzzy clustering center with respect to planisphere origin is calculated, and all distances are pressed
Descending is arranged, and the relative radius of signal of communication planisphere is calculated using first half distance and latter half distance；
In one embodiment of the invention, signal of communication constellation is calculated using first half distance and latter half distance
The relative radius of figure further comprises：
When fuzzy clustering center number be more than or equal to presetting number (during implementation, the preferably presetting number of this programme for 4),
Then by the average value of preceding presetting number distance in first half distance and preceding presetting number distance in latter half distance
Relative radius of the ratio of average value as signal of communication planisphere；
When fuzzy clustering center number is less than presetting number, by the average value of first half distance and latter half away from
From average value relative radius of the ratio as signal of communication planisphere.
The relative radius obtained using aforesaid way, the accuracy of relative radius can be further ensured that, and then further
It ensure that accuracy rate during communication signal recognition.
In step 106, classical modulation signal constellation (in digital modulation) figure is searched by fuzzy clustering center number corresponding to relative radius
With the standard radius value corresponding to identical cluster centre number, when the difference between relative radius and standard radius value is less than the
During three presetting threshold values, then the classification where classical modulation signal is the classification of signal of communication.
In one embodiment of the invention, the obtained multiple fuzzy clustering centers that export calculate each mould with described
Paste cluster centre further comprises between the distance of planisphere origin：
The constellation point exported according to fuzzy clustering algorithm is classified into the degree of membership at each fuzzy clustering center, calculates poly-
Class Validity Function value.
During implementation, following example can be used to calculate Cluster Validity Function value：
Wherein, xjFor j-th of constellation point in signal of communication；C is the number at fuzzy clustering center；viFor i-th of fuzzy clustering
Center；N is the total number of constellation point in signal of communication；μijFor constellation point xjIt is classified into fuzzy clustering center viDegree of membership；
Work as uij＞ ukjWhen, δij=1, otherwise, δij=0；μkjFor constellation point xjIt is classified into fuzzy clustering center xkDegree of membership, k ≠ i；
| | ... | | to ask distance to operate；||xj-vi| | it is j-th of fuzzy clustering center vjWith i-th of constellation point xiThe distance between；||
vi-vj| | it is i-th of fuzzy clustering center viWith j-th of fuzzy clustering center vjThe distance between.
By the first presetting threshold values by the first presetting threshold values is updated after setting multiple increase, it is pre- that field radius is pressed into second
More frontier radius after threshold values reduces is set, Validity Function variable is updated using Cluster Validity Function value；During implementation, setting
Multiple is preferably 2, and the second presetting threshold values and the above-mentioned second presetting threshold values herein are same parameter, and its is preferred
For 0.01.
According to the first presetting threshold values, field radius and the Validity Function variable after renewal, using subtraction clustering algorithm
The constellation point of signal of communication is clustered, until Validity Function value is more than Validity Function variable.
When evaluating the validity of fuzzy clustering, under usual situation, minimum Cluster Validity Function value F is i.e. corresponding optimal
Clusters number c.But in signal identification, following situation can be sometimes run into, although the F values obtained in c=4 compare c=8
It is small, but because the M of signal is even number, it is possible that minimum F values at c=7 or c=3.
The present invention can be increased substantially fuzzy by the relative radius after Cluster Validity Function value F and planisphere cluster
The optimal number of cluster centre, so as to ensure the accuracy of signal of communication adjustment type judgement.
During implementation, it is (excellent herein that first presetting threshold values of this programme preferably after renewal is more than or equal to the 4th presetting valve
When electing as 8), in addition to it is the first presetting valve of the first presetting threshold values of a quarter and half to calculate fuzzy clustering center
The relative radius of signal of communication planisphere during value；
The first presetting threshold values after renewal is more than or equal to the 4th presetting valve of half, and presetting less than the 4th
During valve, in addition to calculate relative half of signal of communication planisphere when fuzzy clustering center is the first presetting threshold values of half
Footpath.
It is updated during due to the first presetting threshold values by setting multiple, number less subtractive clustering center and fuzzy
Cluster centre has calculated before, in the fuzzy clustering for only needing directly to transfer the respective number obtained before herein
The heart can calculate relative radius.
When there are multiple relative radius, standard corresponding to the number of multiple fuzzy clustering centers half can be read in step 106
Footpath is worth, by disclosure satisfy that difference is less than the classification of the relative radius judge signal of communication of the 3rd presetting threshold values.If multiple phases
Pair radius all meets that difference is less than the 3rd presetting threshold values condition, can be chosen by minimal distance principle corresponding during difference minimum
Relative radius go judge signal of communication classification.
Signal of communication can select the effective sub-carrier signals of OFDM in this programme, and the effective sub-carrier signals of OFDM include
The initial field radius of mpsk signal and MQAM signals, the mpsk signal and MQAM signals is different.
The effect of the communication signal recognition method of this programme is illustrated with reference to emulation：
When being emulated, ofdm signal selects DVB-T standards, and pattern selects 2K point FFT, and subcarrier number is 2048,
The duration of symbol takes 280us, and protection interval takes 56us, sample frequency 10MHz, carrier frequency 2MHz, subcarrier-modulated side
Formula is respectively S={ BPSK, QPSK, 16QAM, 64QAM }, and QPSK phase is the integral multiple of pi/2, when subcarrier is 64QAM
OFDM symbol number is 5000, and the number of symbols of other three sub-carriers is 2000, signal to noise ratio 13dB.Gone to reception signal
After CP, FFT, one is randomly selected from numerous effectively subcarriers at random, for identification, what is extracted here is serial number
300 subcarrier.
Under Gaussian channel, for subcarrier 16QAM signals, compare subtractive clustering (algorithm one), fuzzy clustering (algorithm
Two) performance during radar recognition is carried out with this programme method (algorithm three).
When calculating relative radius R values, algorithm three obtains R average closest to standard radius value, and its variance is also minimum, calculates
Method two is secondly.Algorithm three can be illustrated either in terms of the accuracy of estimation relative radius value, or in terms of stability, all
It is best, i.e., algorithm three improves the clustering performance of algorithm one really.When algorithm two and algorithm three are all in preferable clustering number
During mesh, the iterations n of algorithm three is smaller than algorithm two in terms of average and variance.It can illustrate that algorithm three shortens algorithm really
Two iterations.
The lower three kinds of algorithm sub-carriers 16QAM of the Gaussian channel of table 1 clustering performance contrast table
As shown in Figure 2 a and 2 b, for Gaussian channel, when subcarrier is BPSK and QPSK signals, after three kinds of algorithms clusters
C (cluster centre, all c being previously mentioned in the application represent cluster centre) equally, Fig. 2 a understand that algorithm one is in lineups
[- 1, -0.995], normal axis [- 5,5] * 10-3The cluster centre obtained in region is farther from a distance from center for standard.
Be 16QAM signals as Fig. 2 c can be seen that for subcarrier, algorithm two in lineups [0.3,0.4], normal axis [-
0.9, -1.05] two cluster centres have been obtained in region, this is due to that its initial Subject Matrix gives at random, and algorithm three is kept away
Exempting from the reason for such a situation occurs is, the field radius γ of subtractive clusteringaAnd γbIt is reasonable to set, the cluster centre for causing it to obtain
The distance between be separated by will not be close.
As Fig. 2 d can be seen that subcarrier be 64QAM signals when, algorithm one finally cluster out clusters number be 27,
Because the reason for the value of its field radius is not suitable for this moment, and algorithm three is avoided that be its field radius have a fine setting,
Iterative process, new γa=γa- 0.01, and Validity Function and relative radius are to the association evaluation of Cluster Validity.
By the emulation of signal in 4 as can be seen that the obtained signal modulation exponent number of algorithm three and cluster centre relatively other two
All it is best for kind algorithm.
Table 2 is the order of modulation M and relative radius R obtained after three kinds of algorithms cluster to different sub-carrier signal.It can see
Go out, under Gaussian channel, when SNR is 13dB, subcarrier BPSK, QPSK signal, the accuracy of identification of three kinds of algorithms is suitable, but
It is that now algorithm one is poor in the degree of accuracy on estimating relative radius.Subcarrier is 16QAM signals, after algorithm three clusters
Relative radius takes second place closer to center for standard, algorithm one, illustrates that the current area radius value of now algorithm one sets more
Rationally.Subcarrier is 64QAM signals, due to there is identifying mistake when algorithm one emulates, therefore the knot of the relative radius value calculated
Fruit is also very poor, and algorithm three is estimated more accurate relative to the relative radius value of algorithm two.
M, the R parameter table that the lower three kinds of algorithms of the Gaussian channel of table 2 cluster to different sub-carrier
From table 3 it is observed that under Gaussian channel, when SNR is 13dB, in the cluster process of algorithm three, sub- load
When ripple is bpsk signal, the F values at c=2 are smaller than at c=4, and its corresponding R value corresponds to clusters number closer to it
Standard radius value；When subcarrier is QPSK signals, the F values at c=8 are bigger than at c=4, illustrate the model of c corresponding to the F values of minimum
Enclosing should be between [2,8], and the R values at c=4 correspond to the standard radius value of clusters number closer to it；Similarly,
When subcarrier is 16QAM signals, after being reduced the scope using F values, 8≤c≤32, and the R values at c=16 are corresponding closer to it
The standard radius value of clusters number；When subcarrier is 64QAM signals, after being reduced the scope using F values, 64≤c≤256, and in c=
R values at 64 correspond to the standard radius value of clusters number closer to it.
Algorithm three clusters to different sub-carrier under the Gaussian channel of table 3 c, F, R parameter table
Claims (8)
1. the communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm, it is characterised in that including：
Initialization field radius, Validity Function variable, the convergence threshold values of fuzzy clustering function and the maximum of fuzzy clustering function
Iterations；
The constellation point of the signal of communication of reception is clustered using subtraction clustering algorithm, and exports obtained multiple subtractive clusterings
Center；
When the number at subtractive clustering center is less than the first presetting threshold values, field radius is reduced by the second presetting threshold values,
And the constellation point of signal of communication is clustered using subtraction clustering algorithm again, until subtractive clustering center number be more than etc.
In the first presetting threshold values；
The first presetting threshold values subtractive clustering center larger using subtractive clustering center Midst density is used as fuzzy clustering algorithm
Initial center, the constellation point of signal of communication is clustered using fuzzy clustering algorithm, and export obtained multiple fuzzy clusterings
Center；
Distance of each fuzzy clustering center with respect to planisphere origin is calculated, and all distances are arranged in descending order, before
Half part distance and latter half distance calculate the relative radius of signal of communication planisphere；
Searching classical modulation signal constellation (in digital modulation) figure by fuzzy clustering center number corresponding to relative radius has identical cluster centre
Standard radius value corresponding to number, when the difference between relative radius and standard radius value is less than three presetting threshold values,
Then the classification where classical modulation signal is the classification of signal of communication.
2. the communication signal recognition method according to claim 1 based on subtractive clustering and fuzzy clustering algorithm, its feature
It is, the relative radius that signal of communication planisphere is calculated using first half distance and latter half distance is further wrapped
Include：
When fuzzy clustering center number is more than or equal to presetting number, then by preceding presetting number distance in first half distance
The ratio of the average value of preceding presetting number distance is as the relative of signal of communication planisphere in average value and latter half distance
Radius；
When fuzzy clustering center number is less than presetting number, by the average value of first half distance and latter half distance
Average value relative radius of the ratio as signal of communication planisphere.
3. the communication signal recognition method according to claim 1 or 2 based on subtractive clustering and fuzzy clustering algorithm, it is special
Sign is that the obtained multiple fuzzy clustering centers that export calculate each fuzzy clustering center to planisphere origin with described
Further comprise between distance：
The constellation point exported according to fuzzy clustering algorithm is classified into the degree of membership at each fuzzy clustering center, and it is effective to calculate cluster
Property functional value；
By the first presetting threshold values by the first presetting threshold values is updated after setting multiple increase, it is presetting that field radius is pressed into second
More frontier radius after threshold values reduces, Validity Function variable is updated using Cluster Validity Function value；
According to the first presetting threshold values, field radius and the Validity Function variable after renewal, using subtraction clustering algorithm to logical
The constellation point of letter signal is clustered, until Validity Function value is more than Validity Function variable.
4. the communication signal recognition method according to claim 3 based on subtractive clustering and fuzzy clustering algorithm, its feature
It is, when the first presetting threshold values after renewal is more than or equal to four presetting valves, in addition to calculates fuzzy clustering center and be
The relative radius of signal of communication planisphere when the first presetting threshold values of a quarter and the first presetting threshold values of half；
The first presetting threshold values after renewal is more than or equal to the 4th presetting valve of half, and is less than the 4th presetting valve
When, in addition to calculate relative half of signal of communication planisphere when fuzzy clustering center is the first presetting threshold values of half
Footpath.
5. the communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm according to claim 3 or 4, it is special
Sign is that the specific formula of the calculating Cluster Validity Function value is：
Wherein, xjFor j-th of constellation point in signal of communication；C is the number at fuzzy clustering center；viFor in i-th fuzzy clustering
The heart；N is the total number of constellation point in signal of communication；μijFor constellation point xjIt is classified into fuzzy clustering center viDegree of membership；When
uij＞ ukjWhen, δij=1, otherwise, δij=0；μkjFor constellation point xjIt is classified into fuzzy clustering center xkDegree of membership, k ≠ i；|
| ... | | to ask distance to operate；||xj-vi| | it is j-th of fuzzy clustering center vjWith i-th of constellation point xiThe distance between；||
vi-vj| | it is i-th of fuzzy clustering center viWith j-th of fuzzy clustering center vjThe distance between.
6. the communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm according to claim 1,2,4 or 5,
Cluster is carried out to the constellation point of the signal of communication of reception using subtraction clustering algorithm further comprised characterized in that, described：
Calculate the density value of each constellation point in signal of communication：
Wherein, DiFor density value；γaFor field radius；xjAnd xiFor jth, i-th of constellation point in signal of communication；N is signal of communication
The total number of middle constellation point；Exp is e index；||xi-xj| | for the distance between i-th constellation point and j-th constellation point；
Using the density value of maximum as subtractive clustering center, and calculate the density value of remaining constellation point in signal of communication：
Wherein,For the maximum in last constellation point density value,For the maximum institute in last constellation point density value
Corresponding constellation point；For xiWithThe distance between；γb=1.25 γa；
When the subtractive clustering center of acquisition can cover the density value of all constellation points or remaining constellation point less than default
During value, all subtractive clustering centers of acquisition are exported.
7. the communication signal recognition method according to claim 6 based on subtractive clustering and fuzzy clustering algorithm, its feature
It is, it is described cluster is carried out to the constellation point of signal of communication using fuzzy clustering algorithm to further comprise：
According to the first presetting threshold values initial center, the fuzzy clustering center of fuzzy clustering algorithm is calculated：
Wherein, viFor fuzzy clustering center；N is the total number of constellation point in signal of communication；μijFor constellation point xjIt is classified into fuzzy
Cluster centre viDegree of membership, 0≤μij≤1；M be fuzzy clustering algorithm fuzzy factor, m ∈ [1, ∞)；
Calculate the cost function value of fuzzy clustering algorithm：
Wherein, dij=| | vi-xj| |, it is fuzzy clustering center viWith constellation point xjThe distance between；M is fuzzy factor, m ∈
[1, ∞)；λj, j=1 ..., n are Lagrange factor；
When the iterations of convergence threshold values of the cost function value less than fuzzy clustering function, or calculating fuzzy clustering center is less than mould
When pasting the maximum iteration of clustering function, renewal constellation point is classified into the degree of membership at each fuzzy clustering center：
Wherein, dkj=| | vk-xj| |, it is fuzzy clustering center vkWith constellation point xjThe distance between；
The degree of membership that each fuzzy clustering center is classified into using the constellation point after renewal calculates the fuzzy of fuzzy clustering algorithm
Cluster centre；
When the cost function value of calculating is more than or equal to the convergence threshold values of fuzzy clustering function, or the iteration at calculating fuzzy clustering center
When number is equal to the maximum iteration of fuzzy clustering function, exports fuzzy clustering center and be classified into each mould per constellation point
Paste the degree of membership of cluster centre.
8. known according to claim 1,2,4,5 or 7 are any described based on the signal of communication of subtractive clustering and fuzzy clustering algorithm
Other method, it is characterised in that the signal of communication is the effective sub-carrier signals of OFDM, and the effective sub-carrier signals of OFDM include
Mpsk signal and MQAM signals；The mpsk signal is different with the initial field radius of MQAM signals.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710319971.6A CN107707497B (en) | 2017-05-09 | 2017-05-09 | Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710319971.6A CN107707497B (en) | 2017-05-09 | 2017-05-09 | Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107707497A true CN107707497A (en) | 2018-02-16 |
| CN107707497B CN107707497B (en) | 2020-06-02 |
Family
ID=61169576
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710319971.6A Active CN107707497B (en) | 2017-05-09 | 2017-05-09 | Communication signal identification method based on subtraction clustering and fuzzy clustering algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107707497B (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109274626A (en) * | 2018-11-21 | 2019-01-25 | 电子科技大学 | A kind of Modulation Identification method based on planisphere orthogonal scanning feature |
| CN109978006A (en) * | 2019-02-25 | 2019-07-05 | 北京邮电大学 | Clustering method and device |
| CN112565126A (en) * | 2020-11-26 | 2021-03-26 | 杭州电子科技大学 | Modulation identification method based on receiver multi-dimensional characteristic analog-digital mixed signal |
| CN112737992A (en) * | 2020-09-23 | 2021-04-30 | 青岛科技大学 | Underwater sound signal modulation mode self-adaptive in-class identification method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008022435A1 (en) * | 2006-08-21 | 2008-02-28 | Kaveh Seyed Momen | Method, system and apparatus for real-time classification of muscle signals from self-selected intentional movements |
| CN102065056A (en) * | 2011-01-10 | 2011-05-18 | 郑州大学 | Method for realizing MQAM (Multiple Quadrature Amplitude Modulation) signal modulation mode identification of any constellation diagram on basis of clustering |
| CN103150731A (en) * | 2013-03-07 | 2013-06-12 | 南京航空航天大学 | Fuzzy clustering image segmenting method |
| CN104898562A (en) * | 2015-04-27 | 2015-09-09 | 攀枝花学院 | Modeling method of thermal error compensation of numerically-controlled machine tool |
| CN104915560A (en) * | 2015-06-11 | 2015-09-16 | 万达信息股份有限公司 | Method for disease diagnosis and treatment scheme based on generalized neural network clustering |
| CN106408569A (en) * | 2016-08-29 | 2017-02-15 | 北京航空航天大学 | Brain MRI (magnetic resonance image) segmentation method based on improved fuzzy C-means clustering algorithm |
-
2017
- 2017-05-09 CN CN201710319971.6A patent/CN107707497B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008022435A1 (en) * | 2006-08-21 | 2008-02-28 | Kaveh Seyed Momen | Method, system and apparatus for real-time classification of muscle signals from self-selected intentional movements |
| CN102065056A (en) * | 2011-01-10 | 2011-05-18 | 郑州大学 | Method for realizing MQAM (Multiple Quadrature Amplitude Modulation) signal modulation mode identification of any constellation diagram on basis of clustering |
| CN103150731A (en) * | 2013-03-07 | 2013-06-12 | 南京航空航天大学 | Fuzzy clustering image segmenting method |
| CN104898562A (en) * | 2015-04-27 | 2015-09-09 | 攀枝花学院 | Modeling method of thermal error compensation of numerically-controlled machine tool |
| CN104915560A (en) * | 2015-06-11 | 2015-09-16 | 万达信息股份有限公司 | Method for disease diagnosis and treatment scheme based on generalized neural network clustering |
| CN106408569A (en) * | 2016-08-29 | 2017-02-15 | 北京航空航天大学 | Brain MRI (magnetic resonance image) segmentation method based on improved fuzzy C-means clustering algorithm |
Non-Patent Citations (2)
| Title |
|---|
| SAN YE ，AI LING: "《Fuzzy identification based on improved clustering arithmetic and its application》", 《2011 EIGHTH INTERNATIONAL CONFERENCE ON FUZZY SYSTEMS AND KNOWLEDGE DISCOVERY (FSKD)》 * |
| 袁莹莹: "《多径衰落环境下高阶QAM信号调制识别算法研究》", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109274626A (en) * | 2018-11-21 | 2019-01-25 | 电子科技大学 | A kind of Modulation Identification method based on planisphere orthogonal scanning feature |
| CN109274626B (en) * | 2018-11-21 | 2020-11-13 | 电子科技大学 | Modulation identification method based on constellation diagram orthogonal scanning characteristics |
| CN109978006A (en) * | 2019-02-25 | 2019-07-05 | 北京邮电大学 | Clustering method and device |
| CN109978006B (en) * | 2019-02-25 | 2021-02-19 | 北京邮电大学 | Face image clustering method and device |
| CN112737992A (en) * | 2020-09-23 | 2021-04-30 | 青岛科技大学 | Underwater sound signal modulation mode self-adaptive in-class identification method |
| CN112565126A (en) * | 2020-11-26 | 2021-03-26 | 杭州电子科技大学 | Modulation identification method based on receiver multi-dimensional characteristic analog-digital mixed signal |
| CN112565126B (en) * | 2020-11-26 | 2022-06-28 | 杭州电子科技大学 | Modulation identification method based on receiver multi-dimensional characteristic analog-digital mixed signal |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107707497B (en) | 2020-06-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109802905B (en) | CNN convolutional neural network-based digital signal automatic modulation identification method | |
| CN110855591B (en) | QAM and PSK signal intra-class modulation classification method based on convolutional neural network structure | |
| CN110163282B (en) | Modulation mode identification method based on deep learning | |
| CN107707497A (en) | Communication signal recognition method based on subtractive clustering and fuzzy clustering algorithm | |
| CN110061792B (en) | Frequency spectrum sensing method based on variational modal decomposition | |
| CN112733811B (en) | Method for identifying underwater sound signal modulation modes based on improved dense neural network | |
| CN104393963B (en) | Space-Time Block Coding MC CDMA Signal blind recognition methods based on cyclostationarity | |
| CN106533590B (en) | Uplink channel quality measurement method based on receiving end EVM | |
| CN114567528B (en) | Communication signal modulation mode open set recognition method and system based on deep learning | |
| CN110113288A (en) | A kind of design and demodulation method of the ofdm demodulator based on machine learning | |
| CN105049397B (en) | A kind of power distribution method for SIM-OFDM systems | |
| Zhao et al. | Deep Learning-Based channel estimation with SRGAN in OFDM Systems | |
| CN114422311B (en) | Signal modulation recognition method and system combining deep neural network and expert priori features | |
| WO2011007334A2 (en) | Interferer region identification using image processing | |
| CN109274626B (en) | Modulation identification method based on constellation diagram orthogonal scanning characteristics | |
| Nie et al. | Automatic modulation classification based multiple cumulants and quasi-newton method for mimo system | |
| CN112636855A (en) | OFDM signal detection method | |
| CN105516044B (en) | A kind of ofdm system method for suppressing peak to average ratio based on differential evolution algorithm | |
| KR102596571B1 (en) | Deep learning-assisted index estimator for generalized LED index modulation OFDM in visible light communication | |
| CN105429925B (en) | Multi-antenna OFDMA signal decoding method based on Fast Independent Component Analysis | |
| CN101771634B (en) | MIMO-OFDM detection method for sub-carrier grouping | |
| CN108092714A (en) | Support the ofdm system of color adjustment and the detection of CSK planispheres | |
| CN107135180A (en) | Ofdm system signal method of sending and receiving | |
| CN107231323A (en) | The channel estimation methods fed back in visible light communication system based on reliable decision | |
| Hazza et al. | A novel approach for automatic classification of digitally modulated signals in HF communications |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |