CN103428704A - Method and device for sensing frequency spectra - Google Patents

Abstract

The invention discloses a method and a device for sensing frequency spectra, and relates to the technical field of cognitive radio. The method includes clustering all channels in the same business to obtain N channel sets and a representative channel of each channel set; determining a channel set of a channel to be sensed and a representative channel C<n> corresponding to the channel to be sensed, and computing a current occupancy state of the channel to be sensed according to a historical occupancy state of the channel to be sensed and a current occupancy state of the representative channel C<n>. The n is larger than or equal to 1 and is smaller than or equal to N. The method and the device have the advantages that only the current occupancy state of the certain representative channel needs to be detected, so that current occupancy states of all the channels in the corresponding channel sets can be computed and predicted by the method; the channel occupancy state prediction accuracy can be improved by the aid of occupancy dependency among the various channels in the same business.

Description

A kind of frequency spectrum sensing method and device

Technical field

The present invention relates to the cognitive radio technology field, particularly a kind of frequency spectrum sensing method and device.

Background technology

Along with the rapid growth of radio communication service amount, frequency spectrum problem in short supply is more and more serious.Frequency spectrum detecting result by the research countries in the world is found, authorizes the utilance of frequency range generally lower.Yet fixing spectrum allocation may policy makes the mandate frequency range of poor efficiency can't be used by unauthorized user at one's leisure, causes the great wasting of resources.The solution that appears as this problem of cognitive radio technology provides good approach and scheme.The Related Work group (as 802.22,802.16) of Institute of Electrical and Electric Engineers (being called for short IEEE) and International Telecommunications Union's (being called for short ITU) successively formulate or are actively formulating series of standards to promote the development of this technology under various application scenarioss.In cognitive radio, secondary user's can be in the situation that interfere with primary users communication insertion authority frequency range dynamically not, thereby improves the utilance of channel.In order not affect primary user's communication, secondary user's wanted the perception of advanced line frequency spectrum to determine that whether channel is occupied before the access channel, just secondary user's only can access when channel idle.So the frequency spectrum perception technology is to realize the key of cognitive radio technology.Frequency spectrum detection is the usual way of frequency spectrum perception technology.

Traditional frequency spectrum detection is taked the mode that channel is detected one by one, and the expense on time and energy is all very large.In the situation that huge, a plurality of secondary user's of channel number exists, this expense is especially obvious.The effective scheme addressed this problem is exactly spectrum prediction.Spectrum prediction is to predict next seizure condition of each channel constantly according to the account of the history of each channel occupancy.Secondary user's is pressed idle probability sequence detection from high to low to channel, thereby strengthens the probability of access, saves detection time and energy.The Chinese patent that application number is " 201010204687.2 ", disclose a kind of underwater multiplex communication means based on frequency spectrum perception and prediction.This invention utilizes the history of each channel is carried out to next acquistion probability constantly of predicted channel with statistics and the modeling of state, thereby judges its seizure condition.Its implementation procedure is: sensing node by predetermined time gap periods each channel is detected, obtain sufficient historical data; Historical occupied information by each channel is used Markov model to carry out modeling, obtains acquistion probability and the state transition probability of each channel; Then, utilize the Prediction of Markov model established at any one time the acquistion probability of each channel to be predicted.Replace the mode of direct-detection by the predicted channel seizure condition, can effectively improve message transmission rate, and save energy.

Yet, owing to carrying out channel estimating or while estimating, prior art is only considered the effect of the historic state information of channel to prediction, the influencing factor of consideration is too single, is only calculated according to the historical seizure condition of each channel, is difficult to guarantee the accuracy of prediction; In addition, only utilize the historical information of channel to be predicted the channel occupancy state, along with the time growth of prediction, can produce the phenomenon of error propagation.Therefore, concerning the spectrum prediction technology, the accuracy that how to improve as much as possible prediction is a problem of needing solution badly.

Summary of the invention

(1) technical problem to be solved

The object of the present invention is to provide a kind of frequency spectrum sensing method and device, to improve the accuracy of channel occupancy status predication, and improve frequency spectrum perception efficiency.

(2) technical scheme

In order to solve the problems of the technologies described above, the present invention proposes a kind of frequency spectrum sensing method, described method comprises:

All channels under same business are carried out to cluster, obtain the channel that represents of N channel set and each channel set, N is positive integer;

Determine the channel set treat under channel perception and the corresponding channel C that represents _n, 1≤n≤N, treat the historical seizure condition of channel perception and represent channel C according to described _nCurrent seizure condition, calculate the described current seizure condition for the treatment of channel perception.

Optionally, all channels under same business being carried out to cluster specifically comprises:

According to the historical seizure condition of all channels, calculate the coefficient correlation of every two interchannels, and according to coefficient correlation, all channels are carried out to cluster.

Optionally, the computing formula of the correlation coefficient ρ of every two interchannels is:

$\mathrm{\&rho;}=\frac{E}{E+D},$

Wherein, the identical number of corresponding bit place value in the historical seizure condition sequence that E is two channels, the different number of corresponding bit place value in the historical seizure condition sequence that D is two channels.

Optionally, according to coefficient correlation, all channels being carried out to cluster specifically comprises:

Construct correlation matrix A according to coefficient correlation;

According to default coefficient correlation thresholding Th _ρ, correlation matrix A is converted into to relevant judgement matrix B;

Convert respectively every a line of relevant judgement matrix B to the correlated channels set;

Adopt greedy algorithm, obtain the channel that represents of a described N channel set and each channel set according to described correlated channels set.

Optionally, the element of correlation matrix A is designated as ρ _Ij, ρ _IjFor channel c _iWith channel c _jBetween coefficient correlation, 1≤i≤W, 1≤j≤W, the number that W is all channels under described same business;

The element of relevant judgement matrix B is designated as b _Ij, and have:

$b_{\mathrm{ij}}=\left\{\begin{array}{cc}1& {\mathrm{\&rho;}}_{\mathrm{ij}}\&GreaterEqual;{\mathrm{Th}}_{\mathrm{\&rho;}}\\ 0& {\mathrm{\&rho;}}_{\mathrm{ij}}<{\mathrm{Th}}_{\mathrm{\&rho;}}\end{array}\right.;$

Described correlated channels set is designated as s _i, and s _i={ c _j| b _Ij=1}.

Optionally, described greedy algorithm specifically comprises:

The set of all channels under described same business is designated as to U;

Calculate and make | s _iThe correlated channels set s of ∩ U| maximum _i, and by U ∩ s _iBe designated as the first channel set, s _iCorresponding channel c _iBe designated as the channel that represents of described the first channel set;

Make U=U-s _i, repeat previous step, obtain the second channel set and represent channel;

Repeat previous step, until

Thereby, obtain the channel that represents of a described N channel set and each channel set.

Optionally, calculate and describedly treat that the current seizure condition of channel perception specifically comprises:

Obtain representing channel C by frequency spectrum detection _nCurrent seizure condition i _c

Adopt Markov model, according to described, treat the historical seizure condition sequence of channel perception and represent channel C _nHistorical seizure condition sequence, calculate the described state transition probability p (m that treats channel perception _c| m _h) and describedly treat channel perception and represent channel C _nBetween state transition probability p (i _c| m _cm _h), wherein, m _cFor the described current seizure condition for the treatment of channel perception, m _hFor the described historical seizure condition sequence for the treatment of channel perception;

Calculate joint probability p (i according to condition probability formula _cm _cm _h):

p(i _cm _cm _h)＝p(i _c|m _cm _h)p(m _c|m _h)p(m _h)，

Wherein, p (m _h) treat that for described the historical seizure condition sequence of channel perception is m _hProbability;

Adopt the MAP criterion, according to joint probability p (i _cm _cm _h) obtain m _cEstimated value

Thereby obtain the described current seizure condition for the treatment of channel perception

Optionally,

Computing formula be:

${\hat{m}}_c=\arg \underset{m_c}{\max }p\left(i_c{m}_c{m}_h\right).$

The present invention has proposed a kind of frequency spectrum sensing device simultaneously, and described device comprises cluster module and estimation module, wherein:

Described cluster module, carry out cluster for all channels by under same business, obtains the channel that represents of N channel set and each channel set, and N is positive integer;

Described estimation module, for determining the channel set treat under channel perception and the corresponding channel C that represents _n, 1≤n≤N, treat the historical seizure condition of channel perception and represent channel C according to described _nCurrent seizure condition, calculate the described current seizure condition for the treatment of channel perception.

Optionally, described device further comprises memory module and detection module, wherein:

Described memory module is for storing the historical seizure condition of all channels under described same business;

Described detection module is for obtaining representing channel C by frequency spectrum detection _nCurrent seizure condition;

Described cluster module is further used for obtaining from described memory module the historical seizure condition of all channels described same business, and carries out cluster according to the historical seizure condition of all channels;

Described estimation module is further used for obtaining the described historical seizure condition for the treatment of channel perception from described memory module, from described detection module, obtains and represents channel C _nCurrent seizure condition, and calculate the described current seizure condition for the treatment of channel perception.

(3) beneficial effect

The correlation that the frequency spectrum sensing method that the present invention proposes and device have utilized each interchannel under same business, carry out cluster by all channels under same business, and one of selection represents channel from every class channel, carry out seizure condition when prediction treating channel perception, in conjunction with the historical seizure condition information for the treatment of channel perception self and the current seizure condition that represents channel, predicted the outcome.Adopt the method, only need to detect a certain current seizure condition that represents channel, just can calculate and predict the current seizure condition of whole channels in corresponding channel set.Because the method has not only been considered the historic state information of channel, also considered the correlation that takies of interchannel, therefore, compared with prior art, the accuracy of prediction is higher, and the success rate of access is larger, and efficiency is higher.In addition, because the state information that represents channel is used in the estimation for the treatment of the channel perception seizure condition, so each all will be detected the channel that represents in each channel set constantly, can role of correcting be arranged to mistake like this, thereby error propagation is had to obvious inhibitory action.

The accompanying drawing explanation

Fig. 1 is the realization flow figure of one embodiment of the invention intermediate frequency spectrum cognitive method.

Fig. 2 is the schematic diagram of channel cluster result in one embodiment of the invention.

Embodiment

Below in conjunction with drawings and Examples, the specific embodiment of the present invention is described in further detail.

In existing scheme when carrying out the channel occupancy status predication, all only just from each channel angle independently, take situation for the history of each channel respectively and carry out model and forecast, but ignored this part important information of relevance of each interchannel objective reality under same business.In fact, by actual measurement, find, in real communication environment, have very high relevance between the seizure condition of the different channels under same business.If can be used to the high relevance between different channels in this same business, can play very large castering action to forecasting accuracy and the detection efficiency of channel.

In view of this, the embodiment of the present invention provides a kind of frequency spectrum sensing method, and as shown in Figure 1, described method comprises:

All channels under same business are carried out to cluster, obtain the channel that represents of N channel set and each channel set, N is positive integer;

Determine the channel set treat under channel perception and the corresponding channel C that represents _n, 1≤n≤N, treat the historical seizure condition of channel perception and represent channel C according to described _nCurrent seizure condition, calculate the described current seizure condition for the treatment of channel perception.

The frequency spectrum sensing method that the present invention proposes has utilized the correlation of each interchannel under same business, carry out cluster by all channels under same business, and one of selection represents channel from every class channel, carry out seizure condition when prediction treating channel perception, in conjunction with the historical seizure condition information for the treatment of channel perception self and the current seizure condition that represents channel, predicted the outcome.Adopt the method, only need to detect a certain current seizure condition that represents channel, just can calculate and predict the current seizure condition of whole channels in corresponding channel set.Because the method has not only been considered the historic state information of channel, also considered the correlation that takies of interchannel, therefore, compared with prior art, the accuracy of prediction is higher, and the success rate of access is larger, and efficiency is higher.In addition, because the state information that represents channel is used in the estimation for the treatment of the channel perception seizure condition, so each all will be detected the channel that represents in each channel set constantly, can role of correcting be arranged to mistake like this, thereby error propagation is had to obvious inhibitory action.

The precondition that the technical scheme that the present invention proposes is carried out is to have carried out long detection on each channel of target service, has enough history to detect the deposit of data, in order to future channel status predicted and estimated.That is, at first need to utilize the historical historical seizure condition information that data obtain each channel under same business that detects.Usually channel seizure condition at a time means with bit 0 or 1, and 0 means idlely, and 1 expression takies, the historical seizure condition information of channel be exactly in fact one by 0,1 sequence formed.

The frequency spectrum sensing method that the present invention proposes comprises in channel cluster and class estimates two steps.

When the channel cluster, need to be according to the judgement of each inter-channel correlation under same business, utilize the channel clustering algorithm that all channels are polymerized to several classes (being channel set), and select one represent channel in each channel set, this represent channel be in class with the maximally related channel of other channels.

The channel clustering algorithm is mainly to carry out channel packet by the degree of correlation of interchannel seizure condition in same business, and what the degree of correlation of channel occupancy state was large is divided into one group.

Concrete, according to the historical seizure condition of all channels, calculate the coefficient correlation of every two interchannels, and according to coefficient correlation, all channels are carried out to cluster.

Preferably, the computing formula of the correlation coefficient ρ of every two interchannels is:

$\mathrm{\&rho;}=\frac{E}{E+D},$

Wherein, the identical number of corresponding bit place value in the historical seizure condition sequence that E is two channels, the different number of corresponding bit place value in the historical seizure condition sequence that D is two channels.For example, the historical seizure condition sequence of two channels is 00001111000 and 00010010110, E=5 here, D=6.What in fact the correlation coefficient ρ here meaned is the correlation of the seizure condition of channel, and the correlation of the interchannel of the present invention's research all refers to the correlation of seizure condition.

Correlation coefficient ρ is larger, and the correlation of two channels is larger, illustrates that the seizure condition sequence of two interchannels is more similar.When the ρ of two interchannels is enough large, just can be more exactly the seizure condition of state information estimation one other channel by a channel.Extreme case, if ρ=1, can be by detecting a channel, the seizure condition of one other channel is estimated on zero defect ground.

The purpose of channel clustering algorithm is to guarantee ρ>=Th _ρ(Th _ρExpression coefficient correlation thresholding) in situation, all channels under same business are gathered for several channel sets, and select the channel that represents of a needs detection in each channel set, make packet count minimum.That is to say, in the situation that guarantee that the predictablity rate realization detects least number of times.

Preferably, the detailed process of channel cluster is as follows:

Construct correlation matrix A according to coefficient correlation;

According to default coefficient correlation thresholding Th _ρ, correlation matrix A is converted into to relevant judgement matrix B;

Convert respectively every a line of relevant judgement matrix B to the correlated channels set;

Adopt greedy algorithm, obtain the channel that represents of a described N channel set and each channel set according to described correlated channels set.

For instance, for a certain business, suppose that it contains W channel altogether, is designated as C={c ₁, c ₂..., c _W.

The element of correlation matrix A is designated as ρ _Ij, ρ _IjFor channel c _iWith channel c _jBetween coefficient correlation, 1≤i≤W, 1≤j≤W, matrix A is expressed as follows:

The element of relevant judgement matrix B is designated as b _Ij, and have:

$b_{\mathrm{ij}}=\left\{\begin{array}{cc}1& {\mathrm{\&rho;}}_{\mathrm{ij}}\&GreaterEqual;{\mathrm{Th}}_{\mathrm{\&rho;}}\\ 0& {\mathrm{\&rho;}}_{\mathrm{ij}}<{\mathrm{Th}}_{\mathrm{\&rho;}}\end{array}\right.$

Matrix B has recorded the interchannel coefficient correlation and has been greater than default thresholding Th _ρSituation.Known according to the above, in matrix B, be that 1 two channels corresponding to element meet very high correlation.Utilize matrix B to carry out channel packet and to select the most reasonably representing channel by algorithm for design.This problem and set covering problem are similar, will guarantee to meet Th exactly _ρSituation under, cover C={c with minimum set number ₁, c ₂..., c _W.

For this problem is converted into to set covering problem, establish S set={ s ₁, s ₂... be a selection set corresponding to matrix B, each element in S is a correlated channels set, can be designated as s _i, s _iEvery a line by matrix B converts respectively, and s _i={ c _j| b _Ij=1}.

The example that S builds is as follows:

$\stackrel{\begin{array}{ccc}{c}_1& c_2& c_3\end{array}}{\left(\begin{array}{ccc}1& 1& 0\\ 1& 1& 1\\ 0& 1& 1\end{array}\right)}\stackrel{B\&RightArrow;S}{\&RightArrow;}\left\{\right\{c_1,c_2\},\{c_1,c_2,c_3\},\{c_2.c_3\left\}\right\}$

The maximum s of containing element in each S _iMean when detecting channel c _iThe time, the classification that estimable channel is maximum.

Adopt didactic greedy algorithm to be solved this problem, described greedy algorithm specifically comprises:

The set of all channels under described same business is designated as to U;

Calculate and make | s _iThe correlated channels set s of ∩ U| maximum _i, and by U ∩ s _iBe designated as the first channel set, s _iCorresponding channel c _iBe designated as the channel that represents of described the first channel set;

Make U=U-s _i, repeat previous step, obtain the second channel set and represent channel;

Repeat previous step, until

Thereby, obtain the channel that represents of a described N channel set and each channel set.

Said process also can be expressed as follows:

1 arranges U ← { c ₁, c ₂..., c _W,

2 while

3 select s _iMake | s _i∩ U| maximum

4 g←U∩ _is

5 U←U- _is

6 G←G∪{g}

7 Return G

Here g is a channel set (also can say a class), s _iCorresponding channel c _iBe the channel that represents of g.Set G has comprised all channel sets.Fig. 2 is the schematic diagram of channel cluster result in one embodiment of the invention, and in figure, each square frame represents a channel set, and the white round dot in each square frame represents the channel that represents of this channel set, and the black round dot represents the channel perception for the treatment of in this channel set.Owing to adopting greedy formula algorithm, select maximum group number to be divided into groups, so the group separated after number of channel ratio in the group first separated is large at every turn.

After the channel cluster, need to be by detecting the current seizure condition that represents channel in a channel set, accurately estimate the busy-idle condition of all channels in this channel set in conjunction with self historical seizure condition information for the treatment of channel perception.

Below algorithm for estimating in class is described in detail.

At first determine the channel set treat under channel perception and the corresponding channel C that represents _n, then to described, treat that the current seizure condition of channel perception carries out estimating in class.

The process of estimating in class is specific as follows:

Obtain representing channel C by frequency spectrum detection _nCurrent seizure condition i _c

Adopt Markov model, according to described, treat the historical seizure condition sequence of channel perception and represent channel C _nHistorical seizure condition sequence, calculate the described state transition probability p (m that treats channel perception _c| m _h) and describedly treat channel perception and represent channel C _nBetween state transition probability p (i _c| m _cm _h), wherein, m _cFor the described current seizure condition for the treatment of channel perception, m _hFor the described historical seizure condition sequence for the treatment of channel perception;

Calculate joint probability p (i according to condition probability formula _cm _cm _h):

p(i _cm _cm _h)＝p(i _c|m _cm _h)p(m _c|m _h)p(m _h)，

Wherein, p (m _h) treat that for described the historical seizure condition sequence of channel perception is m _hProbability;

Adopt the MAP criterion, according to joint probability p (i _cm _cm _h) obtain m _cEstimated value

Thereby, obtain the described current seizure condition for the treatment of channel perception.

For instance, suppose the channel that represents that the I channel is certain channel set, the M channel is that in this channel set treats channel perception.Consider to mean the impact of M channel history state information on current state with the steady Markov model in n rank, with p (m _c| m _h) mean that the state transition probability of M channel has:

p(m _c＝X _t|m _h＝X _t-1X _t-2...X _t-n...)＝p(m _c＝X _t|m _h＝X _t-1X _t-2...X _t-n)

Wherein, m _cThe state that means current time M channel, m _hMean its historic state sequence, X mean the seizure condition value be 0 or 1, t mean corresponding constantly.

Due to I channel and M channel height correlation, so the state transition probability p (i of definition interchannel _c| m _cm _h), m _cm _hThe status switch that can mean the M channel, comprise current time state and historic state sequence, i _cIt is I channel current time state.As now selected the steady Markov model in n+1 rank to describe the state transitions process of interchannel, have following formula to set up:

p(i _c＝Y _t|m _cm _h＝X _tX _t-1X _t-2...X _t-n...)＝p(i _c＝Y _t|m _cm _h＝X _tX _t-1X _t-2...X _t-n)

Because markoff process is stably, state-transition matrix time to time change not, can come out according to historical data.

In conjunction with the relevant information of historic state information and interchannel, according to condition probability formula, calculate joint probability p (i _cm _cm _h):

p(i _cm _cm _h)＝p(i _c|m _cm _h)p(m _c|m _h)p(m _h)

P(i _cm _cm _h) expression i _cm _cm _hThe joint probability of state.

In conjunction with above-mentioned several new probability formula, adopt the MAP criterion to estimate m _cValue.Because the historic state sequence of I channel current time state and M channel is known, according to MAP criterion, m _cEstimated value

The most reasonably be chosen as and make p (i _cm _cm _h) that m that value is larger _cValue, be shown below:

${\hat{m}}_c=\arg \underset{m_c}{\max p}\left(i_c{m}_c{m}_h\right)$

So far, just estimate to have obtained the current seizure condition of M channel.

The present invention has proposed a kind of frequency spectrum sensing device simultaneously, and described device comprises cluster module and estimation module, wherein:

Described cluster module, carry out cluster for all channels by under same business, obtains the channel that represents of N channel set and each channel set, and N is positive integer;

Described estimation module, for determining the channel set treat under channel perception and the corresponding channel C that represents _n, 1≤n≤N, treat the historical seizure condition of channel perception and represent channel C according to described _nCurrent seizure condition, calculate the described current seizure condition for the treatment of channel perception.

Preferably, described device further comprises memory module and detection module, wherein:

Described memory module is for storing the historical seizure condition of all channels under described same business;

Described detection module is for obtaining representing channel C by frequency spectrum detection _nCurrent seizure condition;

Described cluster module is further used for obtaining from described memory module the historical seizure condition of all channels described same business, and carries out cluster according to the historical seizure condition of all channels;

Described estimation module is further used for obtaining the described historical seizure condition for the treatment of channel perception from described memory module, from described detection module, obtains and represents channel C _nCurrent seizure condition, and calculate the described current seizure condition for the treatment of channel perception.

The correlation that the frequency spectrum sensing method that the present invention proposes and device have utilized each interchannel under same business, carry out cluster by all channels under same business, and one of selection represents channel from every class channel, carry out seizure condition when prediction treating channel perception, in conjunction with the historical seizure condition information for the treatment of channel perception self and the current seizure condition that represents channel, predicted the outcome.Adopt the method, only need to detect a certain current seizure condition that represents channel, just can calculate and predict the current seizure condition of whole channels in corresponding channel set.Because the method has not only been considered the historic state information of channel, also considered the correlation that takies of interchannel, therefore, compared with prior art, the accuracy of prediction is higher, and the success rate of access is larger, and efficiency is higher.In addition, because the state information that represents channel is used in the estimation for the treatment of the channel perception seizure condition, so each all will be detected the channel that represents in each channel set constantly, can role of correcting be arranged to mistake like this, thereby error propagation is had to obvious inhibitory action.

Simultaneously, the technical scheme that the present invention proposes also can be secondary user's and provides convenience.In existing scheme, if, in the situation that not communication between secondary user's all can may be selected access in idle channel at same group, be easy to cause the competition conflict.Under the cognitive system existed in a plurality of secondary user's, technical scheme of the present invention can arrange secondary user's to carry out dynamic access in different channel sets, the conflict of just can not competing between secondary user's like this.

The above is only the preferred embodiment of the present invention; it should be pointed out that for the person of ordinary skill of the art, under the prerequisite that does not break away from the technology of the present invention principle; can also make some improvement and replacement, these improvement and replacement also should be considered as protection scope of the present invention.

Claims (10)

1. a frequency spectrum sensing method, is characterized in that, described method comprises:

All channels under same business are carried out to cluster, obtain the channel that represents of N channel set and each channel set, N is positive integer;

Determine the channel set treat under channel perception and the corresponding channel C that represents _n, 1≤n≤N, treat the historical seizure condition of channel perception and represent channel C according to described _nCurrent seizure condition, calculate the described current seizure condition for the treatment of channel perception.

2. the method for claim 1, is characterized in that, all channels under same business carried out to cluster and specifically comprise:

According to the historical seizure condition of all channels, calculate the coefficient correlation of every two interchannels, and according to coefficient correlation, all channels are carried out to cluster.

3. method as claimed in claim 2, is characterized in that, the computing formula of the correlation coefficient ρ of every two interchannels is:

$\mathrm{\&rho;}=\frac{E}{E+D},$

Wherein, the identical number of corresponding bit place value in the historical seizure condition sequence that E is two channels, the different number of corresponding bit place value in the historical seizure condition sequence that D is two channels.

4. method as claimed in claim 2 or claim 3, is characterized in that, according to coefficient correlation, all channels carried out to cluster and specifically comprise:

Construct correlation matrix A according to coefficient correlation;

According to default coefficient correlation thresholding Th _ρ, correlation matrix A is converted into to relevant judgement matrix B;

Convert respectively every a line of relevant judgement matrix B to the correlated channels set;

Adopt greedy algorithm, obtain the channel that represents of a described N channel set and each channel set according to described correlated channels set.

5. method as claimed in claim 4, is characterized in that, the element of correlation matrix A is designated as ρ _Ij, ρ _IjFor channel c _iWith channel c _jBetween coefficient correlation, 1≤i≤W, 1≤j≤W, the number that W is all channels under described same business;

The element of relevant judgement matrix B is designated as b _Ij, and have:

$b_{\mathrm{ij}}=\left\{\begin{array}{cc}1& {\mathrm{\&rho;}}_{\mathrm{ij}}\&GreaterEqual;{\mathrm{Th}}_{\mathrm{\&rho;}}\\ 0& {\mathrm{\&rho;}}_{\mathrm{ij}}<{\mathrm{Th}}_{\mathrm{\&rho;}}\end{array}\right.;$

Described correlated channels set is designated as s _i, and s _i={ c _j| b _Ij=1}.

6. method as claimed in claim 5, is characterized in that, described greedy algorithm specifically comprises:

The set of all channels under described same business is designated as to U;

Calculate and make | s _iThe correlated channels set s of ∩ U| maximum _i, and by U ∩ s _iBe designated as the first channel set, s _iCorresponding channel c _iBe designated as the channel that represents of described the first channel set;

Make U=U-s _i, repeat previous step, obtain the second channel set and represent channel;

Repeat previous step, until

Thereby, obtain the channel that represents of a described N channel set and each channel set.

7. the method for claim 1, is characterized in that, calculates describedly to treat that the current seizure condition of channel perception specifically comprises:

Obtain representing channel C by frequency spectrum detection _nCurrent seizure condition i _c

Adopt Markov model, according to described, treat the historical seizure condition sequence of channel perception and represent channel C _nHistorical seizure condition sequence, calculate the described state transition probability p (m that treats channel perception _c| m _h) and describedly treat channel perception and represent channel C _nBetween state transition probability p (i _c| m _cm _h), wherein, m _cFor the described current seizure condition for the treatment of channel perception, m _hFor the described historical seizure condition sequence for the treatment of channel perception;

Calculate joint probability p (i according to condition probability formula _cm _cm _h):

p(i _cm _cm _h)＝p(i _c|m _cm _h)p(m _c|m _h)p(m _h)，

Wherein, p (m _h) treat that for described the historical seizure condition sequence of channel perception is m _hProbability;

Adopt the MAP criterion, according to joint probability p (i _cm _cm _h) obtain m _cEstimated value Thereby obtain the described current seizure condition for the treatment of channel perception.

8. method as claimed in claim 7, is characterized in that,

Computing formula be:

${\hat{m}}_c=\arg \underset{m_c}{\max }p\left(i_c{m}_c{m}_h\right).$

9. a frequency spectrum sensing device, is characterized in that, described device comprises cluster module and estimation module, wherein:

Described cluster module, carry out cluster for all channels by under same business, obtains the channel that represents of N channel set and each channel set, and N is positive integer;

Described estimation module, for determining the channel set treat under channel perception and the corresponding channel C that represents _n, 1≤n≤N, treat the historical seizure condition of channel perception and represent channel C according to described _nCurrent seizure condition, calculate the described current seizure condition for the treatment of channel perception.

10. device as claimed in claim 9, is characterized in that, described device further comprises memory module and detection module, wherein:

Described memory module is for storing the historical seizure condition of all channels under described same business;

Described detection module is for obtaining representing channel C by frequency spectrum detection _nCurrent seizure condition;

Described cluster module is further used for obtaining from described memory module the historical seizure condition of all channels described same business, and carries out cluster according to the historical seizure condition of all channels;

Described estimation module is further used for obtaining the described historical seizure condition for the treatment of channel perception from described memory module, from described detection module, obtains and represents channel C _nCurrent seizure condition, and calculate the described current seizure condition for the treatment of channel perception.

Images