CN105375998A - Multiband cooperative spectrum sensing method based on cluster optimization - Google Patents

Abstract

The present invention relates to a multiband cooperative spectrum sensing method based on cluster optimization, comprising the following steps of: establishing a cooperative sensing model composed of a spectrum sensing fusion center, N secondary users and an authorized user; the spectrum sensing fusion center clustering the secondary users according to a signal-to-noise ratio of each secondary user itself and M preset cluster signal-to-noise ratio thresholds, selecting the secondary user that has a largest signal-to-noise ratio in each cluster containing secondary users as a cluster first-time user of a corresponding cluster, and fusing spectrum sensing results of other secondary users in this cluster by using the cluster first-time user as the fusing center of the corresponding cluster so as to obtain a cooperative detection result of this cluster; at last, the spectrum sensing fusion center performing fusion detection according to a global detection probability and a global false alarm probability in the corresponding cluster sent from each cluster first-time user, and using the result of this fusion detection as the final result of multiband cooperative spectrum detection, thereby adapting to a change in signal energy sent from the users, improving secondary user detection performance, reducing computational complexity for the spectrum sensing fusion center and increasing cooperative detection efficiency.

Description

Based on the multiband cooperative frequency spectrum sensing method that sub-clustering is optimized

Technical field

The present invention relates to wireless communication field, particularly relate to a kind of multiband cooperative frequency spectrum sensing method optimized based on sub-clustering.

Background technology

Along with the emerging technology taking LTE, Wi-Fi, satellite communication and collaboration communication etc. as mark is emerged in large numbers in succession, these communication technologys propose higher demand to radio spectrum resources, being tending towards of making frequency spectrum resource become is nervous, cognitive radio technology (CognitiveRadio, CR) arises at the historic moment in this context.The basic ideas of cognitive radio are, first secondary user adopts frequency spectrum perception to continue to detect to having authorized frequency spectrum resource to carry out in surrounding environment; Then preferentially can take this mandate frequency range and transmission performance hardly under affected condition at guarantee authorized user, secondary user adjusts transceiver adaptively to the enterprising Serial Communication of idle frequency spectrum.When secondary user awareness occurs to authorization user signal, secondary user then will vacate channel fast and use for authorized user, and then avoids interference the proper communication of authorized user, thus improves frequency spectrum resource utilization rate.

In order to reduce the factors such as multipath fading in actual environment, shadow effect and incorrect noise to the adverse effect of detection perform, the frequency spectrum sensing method based on multiple user collaborations is constantly proposed.By the sensing results of each user is sent to frequency spectrum perception fusion center, make fusion by frequency spectrum perception fusion center according to certain criterion, to reach the object of accurate perceived spectral., existing cooperative frequency spectrum sensing method majority just carries out perception for one-segment.

In order to improve the availability of frequency spectrum, the cooperative frequency spectrum sensing method for multiband becomes new study hotspot.Existing in the collaborative sensing method of multiband, when secondary user adopts the multiple frequency ranges of energy detection method to multiple authorized user to carry out perception, need to set the judging threshold for signal energy accurately, to make accurate judgement when authorization user signal occurs, and frequency spectrum perception fusion center testing result is sent to carry out fusion treatment respectively by each user.

But, in the multiband collaborative spectrum sensing of reality, still there are some problems: on the one hand, the signal energy that each user receives is not changeless, therefore cause the fixing judging threshold set in existing energy detection method can not ensure that time user makes perception accurately, and then seriously have impact on the overall collaborative sensing performance of multiple users; On the other hand, frequency spectrum perception fusion center needs to carry out fusion calculation to the testing result of all users, and this increases the computation complexity of frequency spectrum perception fusion center undoubtedly, reduces cooperative detection efficiency.

Summary of the invention

Technical problem to be solved by this invention provides one can adapt to from user's received signal energy changing for above-mentioned prior art, improve the detection perform of time user, frequency spectrum perception fusion center computation complexity can be reduced again, improve the multiband cooperative frequency spectrum sensing method optimized based on sub-clustering of cooperative detection efficiency.

The present invention solves the problems of the technologies described above adopted technical scheme: the multiband cooperative frequency spectrum sensing method optimized based on sub-clustering, frequency spectrum detection is carried out for frequency spectrum perception fusion center and N number of secondary user with frequency spectrum perception function, it is characterized in that, in turn include the following steps:

(1) the collaborative sensing model be made up of frequency spectrum perception fusion center, N number of user and authorized user is set up; Wherein, frequency spectrum perception fusion center is designated as FC, and N number of time user is labeled as CR respectively _i(i=1,2 ..., N, N>=3), authorized user is designated as PU;

(2) N number of user CR _iseparately obtain self signal to noise ratio snr _i, and the signal to noise ratio snr will obtained separately respectively _ibe sent to frequency spectrum perception fusion center FC and do sub-clustering process;

(3) according to the ascending order order of snr threshold, the snr threshold SNR of M sub-clustering is preset _{wall, m}(m=1,2 ..., M and 0.5N≤M<N), frequency spectrum perception fusion center FC is by each user CR _iself signal to noise ratio snr sent _irespectively with M snr threshold SNR _{wall, m}judgement is compared, and gets M ₁the individual sub-clustering containing time user, gained sub-clustering is designated as C _l, l=1,2 ..., M ₁, 1<M ₁≤ M, SNR _{wall, 1}<SNR _{wall, 2}< ... <SNR _{wall, M}; Frequency spectrum perception fusion center FC is to each user CR _isignal to noise ratio snr _iwith each snr threshold SNR _{wall, m}judgement comparison procedure following steps (3-1) to step (3-2):

(3-1) according to the snr threshold SNR of M sub-clustering _{wall, m}, M+1 sub-clustering signal to noise ratio segment is set, is respectively [-∞, SNR _{wall, 1}), [SNR _{wall, 1}, SNR _{wall, 2}) ..., [SNR _{wall, M-1}, SNR _{wall, M}) and [SNR _{wall, M}, ∞), wherein, the signal to noise ratio being positioned at the secondary user of the first sub-clustering is in [-∞, SNR _{wall, 1}) in sub-clustering signal to noise ratio segment, the signal to noise ratio being positioned at the secondary user of the second sub-clustering is in [SNR _{wall, 1}, SNR _{wall, 2}) in sub-clustering signal to noise ratio segment, the like, the signal to noise ratio being positioned at the secondary user of M sub-clustering is in [SNR _{wall, M-1}, SNR _{wall, M}) in sub-clustering signal to noise ratio segment, the signal to noise ratio being positioned at the secondary user of M+1 sub-clustering is in [SNR _{wall, M}, ∞) and in sub-clustering signal to noise ratio segment;

(3-2) frequency spectrum perception fusion center FC is respectively by each user CR _isignal to noise ratio snr _iwith M snr threshold SNR _{wall, m}compare, to judge this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment; Wherein:

When this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment is [-∞, SNR _{wall, 1}) time, then do not grant this signal to noise ratio snr _icorresponding secondary user participates in collaborative sensing; If this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment is [SNR _{wall, M}, ∞) time, then by this signal to noise ratio snr _icorresponding secondary user is positioned in M sub-clustering;

(4) at M ₁individual containing in the sub-clustering of time user, according to secondary user's signal to noise ratio order from big to small, choosing secondary user wherein with maximum signal to noise ratio is the secondary user of cluster head in this sub-clustering, thus obtains M ₁individual cluster head time user;

(5) in the second sub-clustering containing time user, using this cluster head time user as the fusion center of this bunch, receive and the frequency spectrum perception result of other user in this bunch merged, to obtain the cooperative detection result of this bunch; Wherein, the cooperative detection process in this bunch comprises the steps that (5-1) is to step (5-3):

(5-1) set in the second sub-clustering and there is K time user CR _k(k=1,2 ..., K), K time user CR _kcarry out respectively based on energy frequency spectrum perception, obtain self signal to noise ratio snr independently _k, and the signal to noise ratio snr will obtained respectively _kcluster head time user CR is sent to frequency spectrum perception result ₁; Wherein, frequency spectrum perception result comprises time user CR _kdetection probability P _d,kand false alarm probability P _f,k;

(5-2) cluster head time user CR ₁receive other K-1 time user CR _kthe signal to noise ratio snr sent _kwith frequency spectrum perception result, and judge signal to noise ratio snr _kbe greater than default signal to noise ratio screening value SNR _chosetime, then the secondary user selecting this signal to noise ratio corresponding is the cognitive group membership participating in cooperative detection, and performs step (5-3); Otherwise, select the frequency spectrum perception result corresponding to secondary user with highest signal to noise ratio to be cluster head time user CR ₁final detection result;

(5-3) cluster head time user CR ₁frequency spectrum perception result according to the cognitive group membership of selected participation cooperation carries out self adaptation perception fusion; Wherein, self adaptation perception fusion process comprises the steps that (5-31) is to step (5-33):

(5-31) cluster head time user CR ₁according to the frequency spectrum perception result that K-1 user sends, to perceive authorized user PU frequency spectrum in statistics K-1 time user be the secondary number of users of seizure condition is m (1≤m≤K-1), perceive authorized user PU frequency spectrum be the secondary number of users of idle condition is K-1-m; Wherein, authorized user PU frequency spectrum is that seizure condition is designated as H ₁, authorized user PU ₁frequency spectrum is that idle condition is designated as H ₀;

(5-32) cluster head time user CR ₁according to the signal to noise ratio that K-1 time user sends, it is seizure condition H that calculating m perceives authorized user PU frequency spectrum ₁the sincere coefficient κ of secondary user _{1, j}and K-1-m to perceive authorized user PU frequency spectrum be idle condition H ₀the sincere coefficient κ of secondary user _{2, t}; Wherein, sincere coefficient κ _{1, j}and κ _{2, t}computing formula as follows:

${\mathrm{\&kappa;}}_{1,j}=\frac{{\mathrm{snr}}_j^2}{\frac{1}{m}\underset{j=1}{\overset{m}{\&Sigma;}}{\mathrm{snr}}_j^2},{\mathrm{\&kappa;}}_{2,t}=\frac{{\mathrm{snr}}_t^2}{\frac{1}{K-1-m}\underset{t=1}{\overset{K-1-m}{\&Sigma;}}{\mathrm{snr}}_t^2};$

(5-33) cluster head time user CR ₁according to respective sensing results and the sincere coefficient κ of m time user _{1, j}, calculating authorized user PU frequency spectrum is respectively seizure condition H ₁average detected probability global detection probability with this seizure condition H ₁corresponding overall false dismissal probability D _{undet, H1}, and the frequency spectrum of authorized user PU is idle condition H ₀average detected probability global detection probability this idle condition H ₀corresponding overall false dismissal probability with overall false alarm probability wherein, this process comprises the steps that (a) is to step (g):

A () sets up the global error detection probability P of m time user collaboration perception _e, obtain the energy measuring majorized function γ about decision-making thresholding ^*and the optimum gate limit value γ of energy measuring _opt, and to calculate authorized user PU frequency spectrum be seizure condition H ₁average detected probability wherein,

The global error detection probability P of m time user collaboration perception _ecomputing formula is as follows:

$P_e=P_{H_0}{P}_f+P_{H_1}{P}_m,P_{H_1}=1-P_{H_0};$

$P_f=Q\left(\frac{\mathrm{\&gamma;}-{\mathrm{\&sigma;}}_n^2}{\frac{2}{m}{\mathrm{\&sigma;}}_n^4}\right),P_d=Q\left(\frac{\mathrm{\&gamma;}-(1+\stackrel{\&OverBar;}{snr}){\mathrm{\&sigma;}}_n^2}{\frac{2}{m}(2\stackrel{\&OverBar;}{snr}+1){\mathrm{\&sigma;}}_n^4}\right),P_m=1-P_d,Q\left(z\right)={\&Integral;}_z^{\mathrm{\&infin;}}\frac{1}{\sqrt{2\mathrm{\&pi;}}}{e}^{-\frac{1}{2}{x}^2}dx;$

Wherein, for authorized user PU frequency spectrum is in idle condition H ₀probability, for authorized user PU frequency spectrum is in seizure condition H ₁probability; P _ffor overall false alarm probability, P _dfor global detection probability, P _mfor overall false dismissal probability; seizure condition H is in for correspondence is in authorized user PU frequency spectrum ₁the average signal-to-noise ratio of m user, wherein, snr _ifor secondary user CR _ithe signal to noise ratio of self, Q (z) represents the complementary integral function of normal Gaussian;

About the energy measuring majorized function γ of decision-making thresholding ^*be defined as:

${\mathrm{\&gamma;}}^{*}=\arg \underset{\mathrm{\&gamma;}}{min}{P}_e=P_{H_0}\&CenterDot;Q\left(\frac{\mathrm{\&gamma;}-{\mathrm{\&sigma;}}_n^2}{\frac{2}{m}{\mathrm{\&sigma;}}_n^4}\right)+P_{H_1}\&CenterDot;Q\left(\frac{\mathrm{\&gamma;}-(1+\stackrel{\&OverBar;}{snr}){\mathrm{\&sigma;}}_n^2}{\frac{2}{m}(2\stackrel{\&OverBar;}{snr}+1){\mathrm{\&sigma;}}_n^4}\right);$

The optimum gate limit value γ of energy measuring _optfor:

$\begin{array}{c}{\mathrm{\&gamma;}}_{opt}=\mathrm{\&gamma;}{|}_{\frac{\&part;P_e}{\&part;\mathrm{\&gamma;}}=0}\\ =\frac{{\mathrm{\&sigma;}}_n^2}{2}+{\mathrm{\&sigma;}}_n^2\sqrt{\frac{1}{4}+\frac{\stackrel{\&OverBar;}{snr}}{2}+\frac{4\stackrel{\&OverBar;}{snr}+2}{m\&CenterDot;\stackrel{\&OverBar;}{snr}}\ln \left(\frac{P_{H_0}}{P_{H_1}}\sqrt{2\stackrel{\&OverBar;}{snr}+1}\right)}\end{array};$

Authorized user PU frequency spectrum is seizure condition H ₁average detected probability computing formula is as follows:

$P_{\det ,H_1}=Q\left(\frac{{\mathrm{\&gamma;}}_{opt}-(1+\stackrel{\&OverBar;}{snr})}{\sqrt{(2/(K-1\left)\right){(1+\stackrel{\&OverBar;}{snr})}^2}}\right);$

B () is seizure condition H according to gained authorized user PU frequency spectrum ₁average detected probability and the sincere coefficient κ of m time user _{1, j}, calculating authorized user PU frequency spectrum is seizure condition H ₁global detection probability with this seizure condition H ₁corresponding overall false dismissal probability wherein, global detection probability with overall false dismissal probability computing formula is as follows:

$D_{\det ,H_1}=\sqrt[m]{\underset{j=1}{\overset{m}{\&Pi;}}{\mathrm{\&kappa;}}_{1,j}}\&CenterDot;\underset{l=m}{\overset{N}{\&Sigma;}}{\left(P_{\det ,H_1}\right)}^l{(1-P_{\det ,H_1})}^{N-l};D_{un\det ,H_1}=1-D_{\det ,H_1}$

C () is idle condition H according to gained authorized user PU frequency spectrum ₀average detected probability and the sincere coefficient κ of K-1-m time user _{2, t}, calculating authorized user PU frequency spectrum is idle condition H ₀global detection probability with this idle condition H ₀corresponding overall false dismissal probability overall situation false alarm probability wherein, average detected probability global detection probability overall situation false dismissal probability with overall false alarm probability computing formula respectively as follows:

$P_{\det ,H_0}=1-Q\left(\frac{{\mathrm{\&gamma;}}_{opt}-1}{\sqrt{(2/(K-1\left)\right)}}\right);$

$D_{\det ,H_0}=\sqrt[K-1-m]{\underset{t=1}{\overset{K-1-m}{\&Pi;}}{\mathrm{\&kappa;}}_{2,t}}\&CenterDot;\underset{l=(K-1-m)+1}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_0}\right)}^l{(1-P_{\det ,H_0})}^{K-1-l};$

$D_{Fail,H_0}=1-D_{\det ,H_0};$

(d) cluster head time user CR ₁be seizure condition H according to authorized user PU frequency spectrum ₁corresponding overall false dismissal probability and authorized user PU frequency spectrum is idle condition H ₀corresponding overall false alarm probability set up frequency spectrum perception error function Fun (m) based on secondary number of users; Wherein, frequency spectrum perception error function Fun (m) computing formula is as follows:

$\begin{array}{c}Fun\left(m\right)=P_{pu}\&CenterDot;D_{un\det ,H_1}+(1-P_{pu})\&CenterDot;D_{Fail,H_0}\\ =P_{pu}\&CenterDot;(1-D_{\det ,H_1})+(1-P_{pu})\&CenterDot;(1-D_{\det ,H_0})\\ =P_{pu}\&CenterDot;(1-\sqrt[m]{\underset{j=1}{\overset{m}{\&Pi;}}{\mathrm{\&kappa;}}_{1,j}}\&CenterDot;\underset{l=m}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_1}\right)}^l{\left(1-P_{\det ,H_1}\right)}^{K-1-l})+\\ (1-P_{pu})\&CenterDot;(1-\sqrt[K-1-m]{\underset{t=1}{\overset{K-1-m}{\&Pi;}}{\mathrm{\&kappa;}}_{2,t}}\&CenterDot;\underset{l=(K-1-m)+1}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_0}\right)}^l{\left(1-P_{\det ,H_0}\right)}^{K-1-l})\end{array};$

Wherein, P _purepresent the probability that authorized user PU signal authorizes frequency spectrum to occur at it;

E () calculates the frequency spectrum perception error minimum value Fun (m of frequency spectrum perception error function Fun (m) ₀), and with this frequency spectrum perception error function minimum value Fun (m ₀) corresponding numerical value m ₀(m ₀≤ m) as the best cooperation time number of users participating in collaborative sensing, and to the snr value snr of m time user according to its correspondence _icarry out descending, obtain the descending group of m time user;

F () chooses the front m in time user's descending group ₀individual user is as the best cooperation time user participating in collaborative sensing; Wherein, marking the best cooperation time user chosen respectively is CR' _r, wherein, r=1,2 ..., m ₀;

(g) cluster head time user CR ₁according to m in step (f) ₀the frequency spectrum perception result of individual the best cooperation time user carries out the collaborative sensing based on OR criterion, and using the testing result of collaborative sensing as K in this bunch the final detection result of a time user; Wherein, OR criterion is as follows:

$Q_{d,1}=1-\underset{r=1}{\overset{m_0}{\&Pi;}}{\mathrm{\&omega;}}_r(1-P_{d,r}),Q_{fa,1}=1-\underset{r=1}{\overset{m_0}{\&Pi;}}(1-P_{f,r});$

${\mathrm{\&omega;}}_r=\frac{{{\mathrm{SNR}}^{\&prime;\&prime;}}_r}{0.5\&CenterDot;({{\mathrm{SNR}}^{\&prime;\&prime;}}_{max}+{{\mathrm{SNR}}^{\&prime;\&prime;}}_{min})},r=1,2,...,m_0;$

Wherein, P _d,rfor cooperation time user CR best in this bunch " _rdetection probability, P _{fa, j}for cooperation time user CR best in this bunch " _rfalse alarm probability; Q _{d, 1}for the global detection probability after this bunch of cooperative detection, Q _{fa, 1}for the overall false alarm probability after this bunch of cooperative detection; ω _rrepresent signal to noise ratio snr " _rweight coefficient, SNR " _maxrepresent m in this bunch ₀the signal to noise ratio maximum of individual the best cooperation time user, SNR " _minrepresent m in this bunch ₀the signal to noise ratio minimum value of individual the best cooperation time user;

(6) according to the process of step (5), the 3rd bunch is obtained respectively to M ₁in bunch bunch in global detection probability Q _{d, 3}extremely and overall false alarm probability Q _{fa, 2}extremely

(7) frequency spectrum perception fusion center FC is according to M ₁global detection probability Q in correspondence that individual cluster head time user sends bunch _d,swith overall false alarm probability Q _{fa, s}carry out the fusion detection based on AND criterion, and using this fusion detection result as final multiband collaboration frequency spectrum testing result; Wherein, AND criterion is as follows:

$Q_d=\underset{s=2}{\overset{M_1}{\&Pi;}}{Q}_{d,s},Q_{fa}=\underset{s=2}{\overset{M_1}{\&Pi;}}{Q}_{fa,s},s=2,3,...,M_1;$

Wherein, Q _dfor the global detection probability after cooperation, Q _fafor the overall false alarm probability after cooperation.

Compared with prior art, the invention has the advantages that: frequency spectrum perception fusion center does sub-clustering according to each user self signal to noise ratio and default sub-clustering snr threshold to secondary user, and in each sub-clustering containing time user, choose the cluster head time user that the secondary user with maximum signal to noise ratio is corresponding sub-clustering, fusion center using this cluster head time user as corresponding bunch, by the adaptive adjustment of secondary user, the optimum gate limit value obtaining energy measuring, to adapt to time needs of user's received signal energy dynamics change, improve the energy measuring probability of time user; Then cluster head time user is merged the frequency spectrum perception result of other user in this bunch, to reduce the amount of calculation that traditional collaborative sensing method intermediate frequency spectrum perception fusion center need merge all user's testing results, the memory space saving frequency spectrum perception fusion center; Then in the correspondence that sends according to each cluster head time user of frequency spectrum perception fusion center bunch, global detection probability and overall false alarm probability carry out fusion detection, and are final multiband collaboration frequency spectrum testing result with this fusion detection result.Utilize the multiband cooperative frequency spectrum sensing method of this sub-clustering can adapt to from user's received signal energy changing, improve the detection perform of time user, frequency spectrum perception fusion center computation complexity can be reduced again, improve cooperative detection efficiency.

Accompanying drawing explanation

Fig. 1 is the schematic flow sheet based on the multiband cooperative frequency spectrum sensing method of sub-clustering optimization in the embodiment of the present invention.

Embodiment

Below in conjunction with accompanying drawing embodiment, the present invention is described in further detail.

In order to realize frequency spectrum perception fusion center FC and the N number of spread spectrum scenarios of secondary user to multiband with frequency spectrum perception function detects, as shown in Figure 1, based on the multiband cooperative frequency spectrum sensing method that sub-clustering is optimized in the present embodiment, in turn include the following steps:

(1) the collaborative sensing model be made up of frequency spectrum perception fusion center FC, N number of user and authorized user is set up; Wherein, N number of user is labeled as CR respectively _i(i=1,2 ..., N, N>=3), authorized user is designated as PU;

(2) N number of user CR _iseparately obtain self signal to noise ratio snr _i, and the signal to noise ratio snr will obtained separately respectively _ibe sent to frequency spectrum perception fusion center FC and do sub-clustering process; Such as, secondary user CR ₁independent self signal to noise ratio obtained is SNR ₁, secondary user CR ₂independent self signal to noise ratio obtained is SNR ₃;

(3) according to the ascending order order of snr threshold, the snr threshold SNR of M sub-clustering is preset _{wall, m}(m=1,2 ..., M and 0.5N≤M<N), namely M to preset bunch in snr threshold be respectively SNR _{wall, 1}, SNR _{wall, 2}... and SNR _{wall, M}, frequency spectrum perception fusion center FC is by each user CR _iself signal to noise ratio snr sent _irespectively with M snr threshold SNR _{wall, m}judgement is compared, and gets M ₁the individual sub-clustering containing time user, gained sub-clustering is designated as C _l, l=1,2 ..., M ₁, 1<M ₁≤ M, SNR _{wall, 1}<SNR _{wall, 2}< ... <SNR _{wall, M};

Such as, by secondary user CR ₁self signal to noise ratio be SNR ₁respectively with snr threshold SNR _{wall, 1}to SNR _{wall, M}do size to judge to compare, and then by secondary user CR ₂self signal to noise ratio be SNR ₁respectively with snr threshold SNR _{wall, 1}to SNR _{wall, M}do size to judge to compare, the like, finally by secondary user CR _nself signal to noise ratio be SNR _nrespectively with snr threshold SNR _{wall, 1}to SNR _{wall, M}do size to judge to compare;

Wherein, frequency spectrum perception fusion center FC is to each user CR _isignal to noise ratio snr _iwith each snr threshold SNR _{wall, m}judgement comparison procedure following steps (3-1) to step (3-2):

(3-1) according to the snr threshold SNR of M sub-clustering _{wall, m}, M+1 sub-clustering signal to noise ratio segment is set, is respectively [-∞, SNR _{wall, 1}), [SNR _{wall, 1}, SNR _{wall, 2}) ..., [SNR _{wall, M-1}, SNR _{wall, M}) and [SNR _{wall, M}, ∞), wherein, the signal to noise ratio being positioned at the secondary user of the first sub-clustering is in [-∞, SNR _{wall, 1}) in sub-clustering signal to noise ratio segment, the signal to noise ratio being positioned at the secondary user of the second sub-clustering is in [SNR _{wall, 1}, SNR _{wall, 2}) in sub-clustering signal to noise ratio segment, the like, the signal to noise ratio being positioned at the secondary user of M sub-clustering is in [SNR _{wall, M-1}, SNR _{wall, M}) in sub-clustering signal to noise ratio segment, the signal to noise ratio being positioned at the secondary user of M+1 sub-clustering is in [SNR _{wall, M}, ∞) and in sub-clustering signal to noise ratio segment;

Such as, setting five snr thresholds are respectively SNR now _{wall, 1}=1dB, SNR _{wall, 2}=3dB, SNR _{wall, 3}=5dB, SNR _{wall, 4}=8dB, SNR _{wall, 5}=11dB, secondary user's signal to noise ratio in first sub-clustering is in [-∞, in 1dB) sub-clustering signal to noise ratio segment, secondary user's signal to noise ratio in second sub-clustering is in [1dB, in 3dB) sub-clustering signal to noise ratio segment, secondary user's signal to noise ratio in 3rd sub-clustering is in [3dB, in 5dB) sub-clustering signal to noise ratio segment, secondary user's signal to noise ratio in 4th sub-clustering is in [5dB, in 8dB) sub-clustering signal to noise ratio segment, secondary user's signal to noise ratio in 5th sub-clustering is in [8dB, in 11dB) sub-clustering signal to noise ratio segment, secondary user's signal to noise ratio in 6th sub-clustering is in [11dB, in ∞) sub-clustering signal to noise ratio segment,

(3-2) frequency spectrum perception fusion center FC is respectively by each user CR _isignal to noise ratio snr _iwith M snr threshold SNR _{wall, m}compare, to judge this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment; Wherein:

When this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment is [-∞, SNR _{wall, 1}) time, then do not grant this signal to noise ratio snr _icorresponding secondary user participates in collaborative sensing; If this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment is [SNR _{wall, M}, ∞) time, represent that secondary user corresponding to this signal to noise ratio has extraordinary detection perform, then by this signal to noise ratio snr _icorresponding secondary user is positioned in M sub-clustering, both can reduce sub-clustering number, improves arithmetic speed, can improve again the overall collaborative sensing performance in M sub-clustering;

Such as, setting five snr thresholds are respectively SNR now _{wall, 1}=1dB, SNR _{wall, 2}=3dB, SNR _{wall, 3}=5dB, SNR _{wall, 4}=10dB, SNR _{wall, 5}=11dB, secondary user's number is six, six user CR ₁to CR ₆self corresponding signal to noise ratio is respectively SNR ₁=-1dB, SNR ₂=1.5dB, SNR ₃=2dB, SNR ₄=6dB, SNR ₅=7dB, SNR ₆=14dB; Known through multilevel iudge, SNR ₁be in [-∞, 1dB) in sub-clustering signal to noise ratio segment, then do not grant time user CR ₁participate in collaborative sensing; Due to SNR ₆be in [11dB, ∞) in sub-clustering signal to noise ratio segment, then by secondary user CR ₆be positioned over [10dB, 11dB) in sub-clustering corresponding to sub-clustering signal to noise ratio segment;

(4) at M ₁individual containing in the sub-clustering of time user, according to secondary user's signal to noise ratio order from big to small, choosing secondary user wherein with maximum signal to noise ratio is the secondary user of cluster head in this sub-clustering, thus obtains M ₁individual cluster head time user;

(5) in the second sub-clustering containing time user, using this cluster head time user as the fusion center of this bunch, receive and the frequency spectrum perception result of other user in this bunch merged, to obtain the cooperative detection result of this bunch;

Wherein, using the fusion center of time user of the cluster head in each sub-clustering as this bunch, not only can reduce the fusion calculation amount of frequency spectrum perception fusion center FC to all user's testing results, the memory space of saving frequency spectrum perception fusion center FC, but also collaborative sensing can be done independently by every bunch, effective raising completes the cooperative detection time of all users, meet the requirement in cognitive radio, secondary user being switched to spectrum efficiency, avoid time user authorized user to be taken to the interference of frequency range; Wherein, the cooperative detection process in this bunch comprises the steps that (5-1) is to step (5-3):

(5-1) set in the second sub-clustering and there is K time user CR _k(k=1,2 ..., K), K time user CR _kcarry out respectively based on energy frequency spectrum perception, obtain self signal to noise ratio snr independently _k, and the signal to noise ratio snr will obtained respectively _kcluster head time user CR is sent to frequency spectrum perception result ₁; Wherein, frequency spectrum perception result comprises time user CR _kdetection probability P _d,kand false alarm probability P _f,k;

(5-2) cluster head time user CR ₁receive other K-1 time user CR _kthe signal to noise ratio snr sent _kwith frequency spectrum perception result, and judge signal to noise ratio snr _kbe greater than default signal to noise ratio screening value SNR _chosetime, then the secondary user selecting this signal to noise ratio corresponding is the cognitive group membership participating in cooperative detection, and performs step (5-3); Otherwise, select the frequency spectrum perception result corresponding to secondary user with highest signal to noise ratio to be cluster head time user CR ₁final detection result;

Wherein, in this step (5-2), why to preset signal to noise ratio screening value SNR _chosebe because, in the secondary user participating in collaborative sensing, if when there is secondary user (being also called " rogue user ") had compared with low signal-to-noise ratio, the Detection accuracy that this " rogue user " is made is extremely low, once participate in collaborative sensing, the detection probability of the overall collaborative sensing that frequency spectrum perception fusion center FC can be caused to make is dragged down, and reduces perception efficiency.So, in collaborative sensing, must setting snr threshold be passed through, these " rogue user " to be weeded out.

(5-3) cluster head time user CR ₁frequency spectrum perception result according to the cognitive group membership of selected participation cooperation carries out self adaptation perception fusion; Wherein, self adaptation perception fusion process comprises the steps that (5-31) is to step (5-33):

(5-31) cluster head time user CR ₁according to the frequency spectrum perception result that K-1 user sends, to perceive authorized user PU frequency spectrum in statistics K-1 time user be the secondary number of users of seizure condition is m (1≤m≤K-1), perceive authorized user PU frequency spectrum be the secondary number of users of idle condition is K-1-m; Wherein, authorized user PU frequency spectrum is that seizure condition is designated as H ₁, authorized user PU ₁frequency spectrum is that idle condition is designated as H ₀;

(5-32) cluster head time user CR ₁according to the signal to noise ratio that K-1 time user sends, it is seizure condition H that calculating m perceives authorized user PU frequency spectrum ₁the sincere coefficient κ of secondary user _{1, j}and K-1-m to perceive authorized user PU frequency spectrum be idle condition H ₀the sincere coefficient κ of secondary user _{2, t}; Wherein, sincere coefficient represents the done credibility detected of corresponding secondary user, also characterizes the detectability of time user; Sincere coefficient is higher, shows that the detection probability of corresponding time user is higher; Wherein, sincere coefficient κ _{1, j}and κ _{2, t}computing formula as follows:

${\mathrm{\&kappa;}}_{1,j}=\frac{{\mathrm{snr}}_j^2}{\frac{1}{m}\underset{j=1}{\overset{m}{\&Sigma;}}{\mathrm{snr}}_j^2},{\mathrm{\&kappa;}}_{2,t}=\frac{{\mathrm{snr}}_t^2}{\frac{1}{K-1-m}\underset{t=1}{\overset{K-1-m}{\&Sigma;}}{\mathrm{snr}}_t^2};$

(5-33) cluster head time user CR ₁according to respective sensing results and the sincere coefficient κ of m time user _{1, j}, calculating authorized user PU frequency spectrum is respectively seizure condition H ₁average detected probability global detection probability with this seizure condition H ₁corresponding overall false dismissal probability and the frequency spectrum of authorized user PU is idle condition H ₀average detected probability global detection probability this idle condition H ₀corresponding overall false dismissal probability with overall false alarm probability wherein, this process comprises the steps that (a) is to step (g):

A () sets up the global error detection probability P of m time user collaboration perception _e, obtain the energy measuring majorized function γ about decision-making thresholding ^*and the optimum gate limit value γ of energy measuring _opt, and to calculate authorized user PU frequency spectrum be seizure condition H ₁average detected probability wherein,

The global error detection probability P of m time user collaboration perception _ecomputing formula is as follows:

$P_e=P_{H_0}{P}_f+P_{H_1}{P}_m,P_{H_1}=1-P_{H_0};$

$P_f=Q\left(\frac{\mathrm{\&gamma;}-{\mathrm{\&sigma;}}_n^2}{\frac{2}{m}{\mathrm{\&sigma;}}_n^4}\right),P_d=Q\left(\frac{\mathrm{\&gamma;}-(1+\stackrel{\&OverBar;}{snr}){\mathrm{\&sigma;}}_n^2}{\frac{2}{m}(2\stackrel{\&OverBar;}{snr}+1){\mathrm{\&sigma;}}_n^4}\right);P_m=1-P_d;Q\left(z\right)={\&Integral;}_z^{\mathrm{\&infin;}}\frac{1}{\sqrt{2\mathrm{\&pi;}}}{e}^{-\frac{1}{2}{x}^2}dx;$

Wherein, for authorized user PU frequency spectrum is in idle condition H ₀probability, for authorized user PU frequency spectrum is in seizure condition H ₁probability; P _ffor overall false alarm probability, P _dfor global detection probability, P _mfor overall false dismissal probability; seizure condition H is in for correspondence is in authorized user PU frequency spectrum ₁the average signal-to-noise ratio of m user, wherein, snr _ifor secondary user CR _ithe signal to noise ratio of self, Q (z) represents the complementary integral function of normal Gaussian;

About the energy measuring majorized function γ of decision-making thresholding ^*be defined as:

${\mathrm{\&gamma;}}^{*}=\arg \underset{\mathrm{\&gamma;}}{min}{P}_e=P_{H_0}\&CenterDot;Q\left(\frac{\mathrm{\&gamma;}-{\mathrm{\&sigma;}}_n^2}{\frac{2}{m}{\mathrm{\&sigma;}}_n^4}\right)+P_{H_1}\&CenterDot;Q\left(\frac{\mathrm{\&gamma;}-(1+\stackrel{\&OverBar;}{snr}){\mathrm{\&sigma;}}_n^2}{\frac{2}{m}(2\stackrel{\&OverBar;}{snr}+1){\mathrm{\&sigma;}}_n^4}\right);$

By to the energy measuring majorized function γ about decision-making thresholding ^*ask extreme value, to obtain the optimum gate limit value γ of energy measuring _optfor:

$\begin{array}{c}{\mathrm{\&gamma;}}_{opt}=\mathrm{\&gamma;}{|}_{\frac{\&part;P_e}{\&part;\mathrm{\&gamma;}}=0}\\ =\frac{{\mathrm{\&sigma;}}_n^2}{2}+{\mathrm{\&sigma;}}_n^2\sqrt{\frac{1}{4}+\frac{\stackrel{\&OverBar;}{snr}}{2}+\frac{4\stackrel{\&OverBar;}{snr}+2}{m\&CenterDot;\stackrel{\&OverBar;}{snr}}\ln \left(\frac{P_{H_0}}{P_{H_1}}\sqrt{2\stackrel{\&OverBar;}{snr}+1}\right)}\end{array};$

Namely utilize in energy measuring process, when the judging threshold for signal energy gets γ each user _opttime, secondary user can detect the existence of received signal accurately, adapts to the situation of change of time user's received signal energy, thus improves the accuracy of time user based on energy measuring;

Authorized user PU frequency spectrum is seizure condition H ₁average detected probability computing formula is as follows:

$P_{\det ,H_1}=Q\left(\frac{{\mathrm{\&gamma;}}_{opt}-(1+\stackrel{\&OverBar;}{snr})}{\sqrt{(2/(K-1\left)\right){(1+\stackrel{\&OverBar;}{snr})}^2}}\right);$

B () is seizure condition H according to gained authorized user PU frequency spectrum ₁average detected probability and the sincere coefficient κ of m time user _{1, j}, calculating authorized user PU frequency spectrum is seizure condition H ₁global detection probability with this seizure condition H ₁corresponding overall false dismissal probability wherein, global detection probability with overall false dismissal probability computing formula is as follows:

$D_{\det ,H_1}=\sqrt[m]{\underset{j=1}{\overset{m}{\&Pi;}}{\mathrm{\&kappa;}}_{1,j}}\&CenterDot;\underset{l=m}{\overset{N}{\&Sigma;}}{\left(P_{\det ,H_1}\right)}^l{(1-P_{\det ,H_1})}^{N-l};D_{un\det ,H_1}=1-D_{\det ,H_1};$

C () is idle condition H according to gained authorized user PU frequency spectrum ₀average detected probability and the sincere coefficient κ of K-1-m time user _{2, t}, calculating authorized user PU frequency spectrum is idle condition H ₀global detection probability with this idle condition H ₀corresponding overall false dismissal probability overall situation false alarm probability wherein, average detected probability global detection probability overall situation false dismissal probability with overall false alarm probability computing formula respectively as follows:

$P_{\det ,H_0}=1-Q\left(\frac{{\mathrm{\&gamma;}}_{opt}-1}{\sqrt{(2/(K-1\left)\right)}}\right);$

$D_{\det ,H_0}=\sqrt[K-1-m]{\underset{t=1}{\overset{K-1-m}{\&Pi;}}{\mathrm{\&kappa;}}_{2,t}}\&CenterDot;\underset{l=(K-1-m)+1}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_0}\right)}^l{(1-P_{\det ,H_0})}^{K-1-l};$

$D_{Fail,H_0}=1-D_{\det ,H_0};$

(d) cluster head time user CR ₁be seizure condition H according to authorized user PU frequency spectrum ₁corresponding overall false dismissal probability and authorized user PU frequency spectrum is idle condition H ₀corresponding overall false alarm probability set up frequency spectrum perception error function Fun (m) based on secondary number of users; This frequency spectrum perception error function Fun (m) characterizes works as the error condition that time number of users is m time-frequency spectrum perception; Wherein, frequency spectrum perception error function Fun (m) computing formula is as follows:

$\begin{array}{c}Fun\left(m\right)=P_{pu}\&CenterDot;D_{un\det ,H_1}+(1-P_{pu})\&CenterDot;D_{Fail,H_0}\\ =P_{pu}\&CenterDot;(1-D_{\det ,H_1})+(1-P_{pu})\&CenterDot;(1-D_{\det ,H_0})\\ =P_{pu}\&CenterDot;(1-\sqrt[m]{\underset{j=1}{\overset{m}{\&Pi;}}{\mathrm{\&kappa;}}_{1,j}}\&CenterDot;\underset{l=m}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_1}\right)}^l{\left(1-P_{\det ,H_1}\right)}^{K-1-l})+\\ (1-P_{pu})\&CenterDot;(1-\sqrt[K-1-m]{\underset{t=1}{\overset{K-1-m}{\&Pi;}}{\mathrm{\&kappa;}}_{2,t}}\&CenterDot;\underset{l=(K-1-m)+1}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_0}\right)}^l{\left(1-P_{\det ,H_0}\right)}^{K-1-l})\end{array};$

Wherein, P _purepresent the probability that authorized user PU signal authorizes frequency spectrum to occur at it;

E () calculates the frequency spectrum perception error minimum value Fun (m of frequency spectrum perception error function Fun (m) ₀), and with this frequency spectrum perception error function minimum value Fun (m ₀) corresponding numerical value m ₀(m ₀≤ m) as the best cooperation time number of users participating in collaborative sensing, and to the snr value snr of m time user according to its correspondence _icarry out descending, obtain the descending group of m time user;

Wherein, when the secondary number of users participating in collaborative sensing is m ₀time, in bunch, the collaborative sensing of time user has minimum frequency spectrum perception error, now correspond to collaborative spectrum sensing and has stronger detection perform; Signal to noise ratio due to each user remains the key affecting its frequency spectrum detection performance, therefore, does descending according to snr value size order, can conveniently make comparisons to the performance of each user after sequence, to select the secondary user with high detection performance;

F () chooses the front m in time user's descending group ₀individual user is as the best cooperation time user participating in collaborative sensing; Wherein, marking the best cooperation time user chosen respectively is CR' _r, wherein, r=1,2 ..., m ₀;

Such as, the secondary user's descending group obtained after according to signal to noise ratio descending is { CR ₁, CR ₂..., CR _m0, CR _m0+1..., CR _mtime, then m before selecting ₀individual user, i.e. { CR ₁, CR ₂..., CR _m0as the best cooperation time user participating in collaborative sensing, and difference correspondence markings CR ₁to CR _m0for the best cooperation time user CR' ₁to CR' _m0;

(g) cluster head time user CR ₁according to m in step (f) ₀the frequency spectrum perception result of individual the best cooperation time user carries out the collaborative sensing based on OR criterion, and using the testing result of collaborative sensing as K in this bunch the final detection result of a time user; Wherein, OR criterion is as follows:

$Q_{d,1}=1-\underset{r=1}{\overset{m_0}{\&Pi;}}{\mathrm{\&omega;}}_r(1-P_{d,r}),Q_{fa,1}=1-\underset{r=1}{\overset{m_0}{\&Pi;}}(1-P_{f,r});$

${\mathrm{\&omega;}}_r=\frac{{{\mathrm{SNR}}^{\&prime;\&prime;}}_r}{0.5\&CenterDot;({{\mathrm{SNR}}^{\&prime;\&prime;}}_{max}+{{\mathrm{SNR}}^{\&prime;\&prime;}}_{min})},r=1,2,...,m_0;$

Wherein, P _d,rfor cooperation time user CR best in this bunch " _rdetection probability, P _{fa, j}for cooperation time user CR best in this bunch " _rfalse alarm probability; Q _{d, 1}for the global detection probability after this bunch of cooperative detection, Q _{fa, 1}for the overall false alarm probability after this bunch of cooperative detection; ω _rrepresent signal to noise ratio snr " _rweight coefficient, ω _rlarger, represent that the detection perform of the best cooperation time user that this weight coefficient is corresponding is stronger; SNR " _maxrepresent m in this bunch ₀the signal to noise ratio maximum of individual the best cooperation time user, SNR " _minrepresent m in this bunch ₀the signal to noise ratio minimum value of individual the best cooperation time user;

(6) according to the process of step (5), the 3rd bunch is obtained respectively to M ₁in bunch bunch in global detection probability Q _{d, 3}extremely and overall false alarm probability extremely wherein, this step (6), namely according to the cooperating process in the second sub-clustering, completes the collaborative sensing in residue sub-clustering;

(7) frequency spectrum perception fusion center FC is according to M ₁global detection probability Q in correspondence that individual cluster head time user sends bunch _d,swith overall false alarm probability Q _{fa, s}carry out the fusion detection based on AND criterion, and using this fusion detection result as final multiband collaboration frequency spectrum testing result; Wherein, AND criterion is as follows:

$Q_d=\underset{s=2}{\overset{M_1}{\&Pi;}}{Q}_{d,s},Q_{fa}=\underset{s=2}{\overset{M_1}{\&Pi;}}{Q}_{fa,s},s=2,3,...,M_1;$

Wherein, Q _dfor the global detection probability after cooperation, Q _fafor the overall false alarm probability after cooperation.In step (7), frequency spectrum perception fusion center FC only needs M ₁(1<M ₁≤ M<N) the global detection probability Q that sends of individual cluster head time user _d,swith overall false alarm probability Q _{fa, s}carry out fusion calculation, and do not need to merge the testing result of N number of user again, thus reduce fusion calculation amount to a great extent, improve fusion efficiencies.

Claims (1)

1., based on the multiband cooperative frequency spectrum sensing method that sub-clustering is optimized, carry out frequency spectrum detection for frequency spectrum perception fusion center and N number of secondary user with frequency spectrum perception function, it is characterized in that, in turn include the following steps:

(1) the collaborative sensing model be made up of frequency spectrum perception fusion center, N number of user and authorized user is set up; Wherein, frequency spectrum perception fusion center is designated as FC, and N number of time user is labeled as CR respectively _i(i=1,2 ..., N, N>=3), authorized user is designated as PU;

(2) N number of user CR _iseparately obtain self signal to noise ratio snr _i, and the signal to noise ratio snr will obtained separately respectively _ibe sent to frequency spectrum perception fusion center FC and do sub-clustering process;

(3) according to the ascending order order of snr threshold, the snr threshold SNR of M sub-clustering is preset _{wall, m}(m=1,2 ..., M and 0.5N≤M < N), frequency spectrum perception fusion center FC is by each user CR _iself signal to noise ratio snr sent _irespectively with M snr threshold SNR _{wall, m}judgement is compared, and gets M ₁the individual sub-clustering containing time user, gained sub-clustering is designated as C _l, l=1,2 ..., M ₁, 1 < M ₁≤ M, SNR _{wall, 1}< SNR _{wall, 2}< ... < SNR _{wall, M}; Frequency spectrum perception fusion center FC is to each user CR _isignal to noise ratio snr _iwith each snr threshold SNR _{wall, m}judgement comparison procedure following steps (3-1) to step (3-2):

(3-1) according to the snr threshold SNR of M sub-clustering _{wall, m}, M+1 sub-clustering signal to noise ratio segment is set, is respectively [-∞, SNR _{wall, 1}), [SNR _{wall, 1}, SNR _{wall, 2}) ..., [SNR _{wall, M-1}, SNR _{wall, M}) and [SNR _{wall, M}, ∞), wherein, the signal to noise ratio being positioned at the secondary user of the first sub-clustering is in [-∞, SNR _{wall, 1}) in sub-clustering signal to noise ratio segment, the signal to noise ratio being positioned at the secondary user of the second sub-clustering is in [SNR _{wall, 1}, SNR _{wall, 2}) in sub-clustering signal to noise ratio segment, the like, the signal to noise ratio being positioned at the secondary user of M sub-clustering is in [SNR _{wall, M-1}, SNR _{wall, M}) in sub-clustering signal to noise ratio segment, the signal to noise ratio being positioned at the secondary user of M+1 sub-clustering is in [SNR _{wall, M}, ∞) and in sub-clustering signal to noise ratio segment;

(3-2) frequency spectrum perception fusion center FC is respectively by each user CR _isignal to noise ratio snr _iwith M snr threshold SNR _{wall, m}compare, to judge this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment; Wherein:

When this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment is [-∞, SNR _{wall, 1}) time, then do not grant this signal to noise ratio snr _icorresponding secondary user participates in collaborative sensing; If this signal to noise ratio snr _iresiding sub-clustering signal to noise ratio segment is [SNR _{wall, M}, ∞) time, then by this signal to noise ratio snr _icorresponding secondary user is positioned in M sub-clustering;

(4) at M ₁individual containing in the sub-clustering of time user, according to secondary user's signal to noise ratio order from big to small, choosing secondary user wherein with maximum signal to noise ratio is the secondary user of cluster head in this sub-clustering, thus obtains M ₁individual cluster head time user;

(5) in the second sub-clustering containing time user, using this cluster head time user as the fusion center of this bunch, receive and the frequency spectrum perception result of other user in this bunch merged, to obtain the cooperative detection result of this bunch; Wherein, the cooperative detection process in this bunch comprises the steps that (5-1) is to step (5-3):

(5-1) set in the second sub-clustering and there is K time user CR _k(k=1,2 ..., K), K time user CR _kcarry out respectively based on energy frequency spectrum perception, obtain self signal to noise ratio snr independently _k, and the signal to noise ratio snr will obtained respectively _kcluster head time user CR is sent to frequency spectrum perception result ₁; Wherein, frequency spectrum perception result comprises time user CR _kdetection probability P _d,kand false alarm probability P _f,k;

(5-2) cluster head time user CR ₁receive other K-1 time user CR _kthe signal to noise ratio snr sent _kwith frequency spectrum perception result, and judge signal to noise ratio snr _kbe greater than default signal to noise ratio screening value SNR _chosetime, then the secondary user selecting this signal to noise ratio corresponding is the cognitive group membership participating in cooperative detection, and performs step (5-3); Otherwise, select the frequency spectrum perception result corresponding to secondary user with highest signal to noise ratio to be cluster head time user CR ₁final detection result;

(5-3) cluster head time user CR ₁frequency spectrum perception result according to the cognitive group membership of selected participation cooperation carries out self adaptation perception fusion; Wherein, self adaptation perception fusion process comprises the steps that (5-31) is to step (5-33):

(5-31) cluster head time user CR ₁according to the frequency spectrum perception result that K-1 user sends, to perceive authorized user PU frequency spectrum in statistics K-1 time user be the secondary number of users of seizure condition is m (1≤m≤K-1), perceive authorized user PU frequency spectrum be the secondary number of users of idle condition is K-1-m; Wherein, authorized user PU frequency spectrum is that seizure condition is designated as H ₁, authorized user PU ₁frequency spectrum is that idle condition is designated as H ₀;

(5-32) cluster head time user CR ₁according to the signal to noise ratio that K-1 time user sends, it is seizure condition H that calculating m perceives authorized user PU frequency spectrum ₁the sincere coefficient κ of secondary user _{1, j}and K-1-m to perceive authorized user PU frequency spectrum be idle condition H ₀the sincere coefficient κ of secondary user _{2, t}; Wherein, sincere coefficient κ _{1, j}and κ _{2, t}computing formula as follows:

${\mathrm{\&kappa;}}_{1,j}=\frac{{\mathrm{snr}}_j^2}{\frac{1}{m}\underset{j=1}{\overset{m}{\&Sigma;}}{\mathrm{snr}}_j^2},{\mathrm{\&kappa;}}_{2,t}=\frac{{\mathrm{snr}}_t^2}{\frac{1}{K-1-m}\underset{t=1}{\overset{K-1-m}{\&Sigma;}}{\mathrm{snr}}_t^2};$

(5-33) cluster head time user CR ₁according to respective sensing results and the sincere coefficient κ of m time user _{1, j}, calculating authorized user PU frequency spectrum is respectively seizure condition H ₁average detected probability P _{det, H1}, global detection probability D _{det, H1}with this seizure condition H ₁corresponding overall false dismissal probability D _{undet, H1}, and the frequency spectrum of authorized user PU is idle condition H ₀average detected probability P _{det, H0}, global detection probability D _{det, H0}, this idle condition H ₀corresponding overall false dismissal probability D _{undet, H0}with overall false alarm probability D _{fail, H0}; Wherein, this process comprises the steps that (a) is to step (g):

A () sets up the global error detection probability P of m time user collaboration perception _e, obtain the energy measuring majorized function γ about decision-making thresholding ^*and the optimum gate limit value γ of energy measuring _opt, and to calculate authorized user PU frequency spectrum be seizure condition H ₁average detected probability P _{det, H1}; Wherein,

The global error detection probability P of m time user collaboration perception _ecomputing formula is as follows:

P _e＝P _H0P _f+P _H1P _m，P _H1＝1-P _H0；

$P_f=Q\left(\frac{\mathrm{\&gamma;}-{\mathrm{\&sigma;}}_n^2}{\frac{2}{m}{\mathrm{\&sigma;}}_n^4}\right),P_d=Q\left(\frac{\mathrm{\&gamma;}-(1+\stackrel{\&OverBar;}{snr}){\mathrm{\&sigma;}}_n^2}{\frac{2}{m}(2\stackrel{\&OverBar;}{snr}+1){\mathrm{\&sigma;}}_n^4}\right),$ P _m＝1-P _d， $Q\left(z\right)={\&Integral;}_z^{\mathrm{\&infin;}}\frac{1}{\sqrt{2\mathrm{\&pi;}}}{e}^{-\frac{1}{2}{x}^2}dx;$

Wherein, P _h0for authorized user PU frequency spectrum is in idle condition H ₀probability, P _h1for authorized user PU frequency spectrum is in seizure condition H ₁probability; P _ffor overall false alarm probability, P _dfor global detection probability, P _mfor overall false dismissal probability; seizure condition H is in for correspondence is in authorized user PU frequency spectrum ₁the average signal-to-noise ratio of m user, wherein, snr _ifor secondary user CR _ithe signal to noise ratio of self, Q (z) represents the complementary integral function of normal Gaussian;

About the energy measuring majorized function γ of decision-making thresholding ^*be defined as:

${\mathrm{\&gamma;}}^{*}=\arg \underset{\mathrm{\&gamma;}}{\min }{P}_e=P_{H0}\&CenterDot;Q\left(\frac{\mathrm{\&gamma;}-{\mathrm{\&sigma;}}_n^2}{\frac{2}{m}{\mathrm{\&sigma;}}_n^4}\right)+P_{H1}\&CenterDot;Q\left(\frac{\mathrm{\&gamma;}-(1+\stackrel{\&OverBar;}{snr}){\mathrm{\&sigma;}}_n^2}{\frac{2}{m}(2\stackrel{\&OverBar;}{snr}+1){\mathrm{\&sigma;}}_n^4}\right);$

The optimum gate limit value γ of energy measuring _optfor:

$\begin{array}{c}{\mathrm{\&gamma;}}_{opt}=\mathrm{\&gamma;}{|}_{\frac{\&part;P_e}{\&part;\mathrm{\&gamma;}}=0}\\ =\frac{{\mathrm{\&sigma;}}_n^2}{2}+{\mathrm{\&sigma;}}_n^2\sqrt{\frac{1}{4}+\frac{\stackrel{\&OverBar;}{snr}}{2}+\frac{4\stackrel{\&OverBar;}{snr}+2}{m\&CenterDot;\stackrel{\&OverBar;}{snr}}ln(\frac{P_{H0}}{P_{H1}}\sqrt{2\stackrel{\&OverBar;}{snr}+1}})\end{array};$

Authorized user PU frequency spectrum is seizure condition H ₁average detected probability P _{det, H1}computing formula is as follows:

$P_{\det ,H_1=Q}\left(\frac{{\mathrm{\&gamma;}}_{opt}-(1+\stackrel{\&OverBar;}{snr})}{\sqrt{(2/(K-1)){(1+\stackrel{\&OverBar;}{snr})}^2}}\right);$

B () is seizure condition H according to gained authorized user PU frequency spectrum ₁average detected probability P _{det, H1}and the sincere coefficient κ of m time user _{1, j}, calculating authorized user PU frequency spectrum is seizure condition H ₁global detection probability D _{det, H1}with this seizure condition H ₁corresponding overall false dismissal probability D _{undet, H1}; Wherein, global detection probability D _{det, H1}with overall false dismissal probability D _{undet, H1}computing formula is as follows:

$D_{\det ,H_1}=\sqrt[m]{\underset{j=1}{\overset{m}{\&Pi;}}{\mathrm{\&kappa;}}_{1,j}}\&CenterDot;\underset{l=m}{\overset{N}{\&Sigma;}}{\left(P_{\det ,H_1}\right)}^l{\left(1-P_{\det ,H_1}\right)}^{N-l};D_{un\det ,H_1}=1-D_{\det ,H_1}$

C () is idle condition H according to gained authorized user PU frequency spectrum ₀average detected probability P _{det, H0}and the sincere coefficient κ of K-1-m time user _{2, t}, calculating authorized user PU frequency spectrum is idle condition H ₀global detection probability D _{det, H0}with this idle condition H ₀corresponding overall false dismissal probability D _{undet, H0}, overall false alarm probability D _{fail, H0}; Wherein, average detected probability P _{det, H0}, global detection probability D _{det, H0}, overall false dismissal probability D _{undet, H0}with overall false alarm probability D _{fail, H0}computing formula respectively as follows:

$P_{\det ,H_0}=1-Q\left(\frac{{\mathrm{\&gamma;}}_{opt}-1}{\sqrt{(2/(K-1))}}\right);$

$D_{\det ,H_0}=\sqrt[K-1-m]{\underset{t=1}{\overset{K-1-m}{\&Pi;}}{\mathrm{\&kappa;}}_{2,t}}\&CenterDot;\underset{l=\left(K-1-m\right)+1}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_0}\right)}^l{\left(1-P_{\det ,H_0}\right)}^{K-1-l};$

D _Fail,H0＝1-D _det,H0；

(d) cluster head time user CR ₁be seizure condition H according to authorized user PU frequency spectrum ₁corresponding overall false dismissal probability D _{undet, H1}and authorized user PU frequency spectrum is idle condition H ₀corresponding overall false alarm probability D _{fail, H0}, set up frequency spectrum perception error function Fun (m) based on secondary number of users; Wherein, frequency spectrum perception error function Fun (m) computing formula is as follows:

$\begin{array}{c}Fun\left(m\right)=P_{pu}\&CenterDot;D_{un\det ,H_1}+\left(1-P_{pu}\right)\&CenterDot;D_{Fail,H_0}\\ =P_{pu}\&CenterDot;\left(1-D_{\det ,H_1}\right)+\left(1-P_{pu}\right)\&CenterDot;\left(1-D_{\det ,H_0}\right)\\ =P_{pu}\&CenterDot;(1-\sqrt[m]{\underset{j=1}{\overset{m}{\&Pi;}}{\mathrm{\&kappa;}}_{1,j}}\&CenterDot;\underset{l=m}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_1}\right)}^l{\left(1-P_{\det ,H_1}\right)}^{K-1-l})+\\ \left(1-P_{pu}\right)\&CenterDot;(1-\sqrt[K-1-m]{\underset{t=1}{\overset{K-1-m}{\&Pi;}}{\mathrm{\&kappa;}}_{2,t}}\&CenterDot;\underset{l=\left(K-1-m\right)+1}{\overset{K-1}{\&Sigma;}}{\left(P_{\det ,H_0}\right)}^l{\left(1-P_{\det ,H_0}\right)}^{K-1-l})\end{array};$

Wherein, P _purepresent the probability that authorized user PU signal authorizes frequency spectrum to occur at it;

E () calculates the frequency spectrum perception error minimum value Fun (m of frequency spectrum perception error function Fun (m) ₀), and with this frequency spectrum perception error function minimum value Fun (m ₀) corresponding numerical value m ₀(m ₀≤ m) as the best cooperation time number of users participating in collaborative sensing, and to the snr value snr of m time user according to its correspondence _icarry out descending, obtain the descending group of m time user;

F () chooses the front m in time user's descending group ₀individual user is as the best cooperation time user participating in collaborative sensing; Wherein, marking the best cooperation time user chosen respectively is CR' _r, wherein, r=1,2 ..., m ₀;

(g) cluster head time user CR ₁according to m in step (f) ₀the frequency spectrum perception result of the secondary user of individual final participation collaborative sensing carries out the collaborative sensing based on OR criterion, and using the testing result of collaborative sensing as K in this bunch the final detection result of time user; Wherein, OR criterion is as follows:

$Q_{d,1}=1-\underset{r=1}{\overset{m_0}{\&Pi;}}{\mathrm{\&omega;}}_r(1-P_{d,r}),Q_{fa,1}=1-\underset{r=1}{\overset{m_0}{\&Pi;}}(1-P_{f,r});$

${\mathrm{\&omega;}}_r=\frac{{{\mathrm{SNR}}^{\&prime;\&prime;}}_r}{0.5\&CenterDot;({{\mathrm{SNR}}^{\&prime;\&prime;}}_{\max }+{{\mathrm{SNR}}^{\&prime;\&prime;}}_{min})},r=1,2,...,m_0;$

Wherein, P _d,rfor cooperation time user CR best in this bunch " _rdetection probability, P _{fa, j}for cooperation time user CR best in this bunch " _rfalse alarm probability; Q _{d, 1}for the global detection probability after this bunch of cooperative detection, Q _{fa, 1}for the overall false alarm probability after this bunch of cooperative detection; ω _rrepresent signal to noise ratio snr " _rweight coefficient, SNR " _maxrepresent m in this bunch ₀the signal to noise ratio maximum of individual the best cooperation time user, SNR " _minrepresent m in this bunch ₀the signal to noise ratio minimum value of individual the best cooperation time user;

(6) according to the process of step (5), the 3rd bunch is obtained respectively to M ₁in bunch bunch in global detection probability Q _{d, 3}to Q _{d, M1}and overall false alarm probability Q _{fa, 2}to Q _{fa, M1};

(7) frequency spectrum perception fusion center FC is according to M ₁global detection probability Q in correspondence that individual cluster head time user sends bunch _d,swith overall false alarm probability Q _{fa, s}carry out the fusion detection based on AND criterion, and using this fusion detection result as final multiband collaboration frequency spectrum testing result; Wherein, AND criterion is as follows:

$Q_d=\underset{s=2}{\overset{M_1}{\&Pi;}}{Q}_{d,s},Q_{fa}=\underset{s=2}{\overset{M_1}{\&Pi;}}{Q}_{fa,s},$ s＝2,3，…,M ₁。