CN103795478A - Method for detecting number of multiple primary users based on typical relational analysis - Google Patents

Abstract

The invention provides a method for detecting the number of multiple primary users based on typical relational analysis and discloses a method for a cognitive radio network to perceive the number of the multiple primary users by utilizing secondary user cooperation under a mobile environment. In a mobile cognitive radio network, secondary users send primary user signals received by the secondary users to a data fusion center in the network, and the data fusion center conducts the typical relational analysis on secondary user data and further detects the number of the primary users in work in the network. According to the method for detecting the number of multiple primary users, it is unnecessary to utilize an antenna array, only multiple secondary users are required to participate in cooperation, the number of the primary users in the network can be detected, and the method is more suitable for spectrum sensing of the mobile cognitive radio network. In addition, detection utilizing the cooperation of the secondary users can effectively solve the problems of shadow fading, hidden terminals and the like. Response to interference of space color noise further can be effectively achieved, and the method is suitable for detecting the multi-primary-user signals in the mobile cognitive radio network.

Description

A kind of many primary users number detection method based on typical association analysis

Technical field

The present invention relates to utilize in a kind of mobile cognitive radio networks the method for time many primary users of user's cooperative sensing number, particularly a kind of many primary users number detection method based on typical association analysis.

Background technology

From micro-letter, being widely used of the interconnected products of movement such as Alipay can find out, wireless mobile communications is widely used in each corner of life.But along with the continuous increase of user and business, frequency spectrum resource is also more and more in short supply.Alleviating one of important method that frequency spectrum resource is in short supply is exactly cognitive radio.It,, by utilizing the idling-resource in frequency spectrum, has improved the utilance of frequency spectrum greatly.But along with the development of technology, a large amount of appearance of multi-mobile-terminal in cognitive radio networks, whether traditional frequency spectrum perception only detects primary user and exists and can not meet the demands far away, but existing primary user's number detection method need to be set up special aerial array, if can utilize time user's cooperative detection to go out primary user's number under mobile environment, will reduce equipment expense, and improve the reliability of testing result.

In mobile cognitive radio networks, utilize time user's cooperative detection to go out primary user's number, Main Function has the following aspects:

1. facilitate dynamic spectrum resource management.Whether compare only perception primary user and exist, detecting primary user's number can, better for the dynamic spectrum resource management of network provides support, further improve the availability of frequency spectrum;

2. reduce the equipment expense of network.Traditional primary user's number detection method, need to additionally set up independent array antenna in order to detect primary user's signal.The present invention is using each user as sensing node, by analyzing the correlation of the signal that all users receive, detects primary user's signal number.Therefore no longer need to set up special array antenna, reduced the equipment expense of network;

3. successfully manage the interference of space correlation coloured noise.Under mobile environment, the noise comprising in the signal that each user is received due to the existence of multipath effect has certain correlation.The typical association analysis method adopting in the present invention, can well overcome the interference of coloured noise, reduces the requirement of network to operational environment;

4. strengthen the reliability of testing result.Under mobile wireless environment, due to shadow effect and hidden terminal problem, the reliability of single user's testing result cannot be guaranteed.Utilize user's cooperative sensing multiple times, not only can overcome shadow effect and concealed terminal, improve the reliability of testing result, can also reduce single user's equipment complexity;

At present conventional sources number detection method has the AIC(Akaiceinformation criterion based on information theory criterion) and MDL(Minimum description length) and GDE(Gerschgorin disks Estimator based on Gerschgorin radii) criterion, but these methods are all the signals based on array antenna received, need in network, set up specially array antenna to detect the number of primary user's signal, not only increase the equipment expense of network, also can face the problem such as shadow effect and concealed terminal, be not suitable for the many primary users sources number detection in mobile cognitive radio networks.

Summary of the invention

The present invention is intended to solve classical signal number detection method defect, utilizes the typical association analysis method in multiple linear analysis, analyzes the correlation between time signal of user's reception, and then calculates primary user's number.Therefore adopt mobile cognitive radio networks of the present invention not need to set up extra aerial array, only need to using in network movably time user as sensing node, still can detect the number of primary user in network.More be applicable to the frequency spectrum perception of mobile cognitive radio networks.

The object of the invention is to the detection method to primary user's number in a kind of mobile cognitive radio networks, it is characterized in that, the method comprises the following steps:

Primary user's signal in step 1, network in p time user awareness operational environment;

The primary user's signal receiving is transferred to the data fusion center in network by step 2, inferior user, and in same sense cycle, all users' sampling number is N;

P time user is divided at random two groups by step 3, data fusion center, and every group of number is respectively s ₁and s ₂, p=s ₁+ s ₂, s ₁and s ₂all be greater than the number of primary user in network; Then the data that two groups of inferior users are transferred to data fusion center are carried out typical association analysis, obtain typical incidence coefficient and ${\widehat{\mathrm{\λ}}}_1\≥{\widehat{\mathrm{\λ}}}_2\≥\·\·\·\≥{\widehat{\mathrm{\λ}}}_s,$ s=min(s ₁,s ₂)；

Step 4, the statistic that typical incidence coefficient is constructed:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

Carry out significance test, wherein k=1,2 ..., s, testing result is significant C _knumber is the primary user's number detecting.

Preferably, described primary and secondary user can be that permanent plant can be also mobile terminal, and mobile terminal can move freely at work.

Preferably, the data that described in described step 3, two groups of inferior users are transferred to data fusion center are carried out typical association analysis, obtain typical incidence coefficient

and s=min (s ₁, s ₂), comprising:

3.1) data that two groups of inferior users are transferred to data fusion center by data fusion center form respectively s ₁the matrix X of × N dimension ₁and s ₂the matrix X of × N dimension ₂;

3.2) by matrix X ₁and X ₂be configured to matrix $Z=\left[\begin{array}{c}{X}_1\\ X_2\end{array}\right],$ And try to achieve the estimate covariance matrix of Z $\widehat{\mathrm{\Σ}}=\frac{1}{N}{\mathrm{ZZ}}^H=\frac{1}{N}\left[\begin{array}{cc}{\widehat{\mathrm{\Σ}}}_{11}& {\widehat{\mathrm{\Σ}}}_{12}\\ {\widehat{\mathrm{\Σ}}}_{21}& {\widehat{\mathrm{\Σ}}}_{22}\end{array}\right];$

3.3) structural matrix $R={\widehat{\mathrm{\Σ}}}_{11}^{-1/2}{\widehat{\mathrm{\Σ}}}_{12}{\left({\widehat{\mathrm{\Σ}}}_{22}^{-1/2}\right)}^H,$ The characteristic value of matrix R ${\widehat{\mathrm{\λ}}}_i,i=\mathrm{1,2}...,s,$ S=min (s ₁, s ₂) be two groups of time users and be transferred to the typical incidence coefficient of the data at data fusion center.

5. preferably, described step 4 comprises:

4.1) level of significance α of setting significance test;

4.2) the typical incidence coefficient in step 3 is carried out to following computing, obtain k statistic C _k:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

C _k, k=1,2 ..., s obeys card side's distribution χ ²((s ₁-k+1) (s ₂-k+1));

4.3) distribute according to the card side in level of significance α and step 4.2 in step 4.1

χ ²((s ₁-k+1) (s ₂-k+1)), to C _kcarry out significance test, by significant C _knumber as detected primary user's number.

Preferably, described step 4.3 comprises the following steps:

I. according to statistic C in the level of significance α in step 4.1 and 4.2 _kdistribution, calculate the threshold T of significance test _k;

Ii. compare each C _k, k=1,2 ..., s and corresponding threshold T _ksize, if C _k>T _k, k statistic C _ksignificantly;

Iii. be significant C by testing result _knumber as the number of the primary user's signal detecting.

Preferably, described in, calculate the threshold T of significance test _kobtain by solving following equation

$\mathrm{\α}={\∫}_{T_k}^{\∞}\frac{1}{2^{\frac{m}{2}}\mathrm{\Γ}\left(\frac{m}{2}\right)}{y}^{\frac{m-2}{2}}{e}^{-\frac{y}{2}}\mathrm{dy}$

Wherein, Γ represents that gamma distributes, k=1, and 2 ..., s, m=(s ₁-k+1) (s ₂-k+1);

Or

The described threshold T that calculates significance test _kdraw by inquiry chi-square distribution table $T_k={\mathrm{\χ}}_{\mathrm{\α}}^2\left((s_1-k+1)(s_2-k+1)\right).$

The algorithm adopting in the present invention is the typical association analysis based in multi-variate statistical analysis, carry out significance test and determine the number of primary user's signal by the typical incidence coefficient of signal that two groups of time users are received, made up and in existing frequency spectrum perception, only detected the defect whether primary user exists.The present invention simultaneously is also no longer subject to the restriction of array antenna in classical signal number detection algorithm, utilizes the inferior user's cooperative sensing under mobile status in network, can detect the number of primary user's signal.

Accompanying drawing explanation

Fig. 1 moves cognitive radio networks model.

Fig. 2 is the method schematic diagram that detects primary user's signal number in the present invention.

Fig. 3 is typical association analysis method of the present invention and the detection effect contrast figure of traditional detection method under white noise.

Fig. 4 is typical association analysis method of the present invention and the detection effect contrast figure of traditional detection method under space correlation coloured noise.

Embodiment

Be exemplary below by the embodiment being described with reference to the drawings, only for explaining the present invention, and can not be interpreted as limitation of the present invention.

Mobile cognitive radio networks model as shown in Figure 1, supposes and in a mobile cognitive radio networks, has p time user and q primary user and a data fusion center.All users are movably in the certain limit of putting centered by fusion center.

Now provide the basic step of algorithm that the present invention adopts:

Primary user's signal in p in step 1, network time user awareness operational environment;

The primary user's signal perceiving is transferred to the data fusion center in network by step 2, a p time user, and in same sense cycle, all users' sampling number is N, and the data that the individual time user of i is transferred to data fusion center are x ⁱ(n), i=1,2 ..., p, n=1,2 ..., N;

P time user is divided at random two groups by step 3, data fusion center, and every group of number is respectively s ₁and s ₂, p=s ₁+ s ₂.S ₁and s ₂all need to be greater than the number of primary user in network.Pay special attention to be the grouping here just at fusion center in order to carry out data analysis, inferior user is entered to row stochastic grouping, the physics mobility on inferior user in network does not produce any impact.

Then typical association analysis is carried out to these the two groups primary user's signals that receive from inferior user (time user is transferred to the data at data fusion center, after referred to as inferior user data) in data fusion center, and detailed process is as follows:

3.1) two groups user data are formed respectively to s ₁the matrix X of × N dimension ₁and s ₂the matrix X of × N dimension ₂, be respectively:

$X_1=\left[\begin{array}{c}{x}_1^1\\ x_1^2\\ \·\\ \·\\ \·\\ x_1^{s_1}\end{array}\right]=\left[\begin{array}{cccc}{x}_1^1\left(1\right)& x_1^1\left(2\right)& \·\·\·& x_1^1\left(N\right)\\ x_1^2\left(1\right)& x_1^2\left(2\right)& \·\·\·& x_1^2\left(N\right)\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ x_1^{s_1}\left(1\right)& x_1^{s_1}\left(2\right)& \·\·\·& x_1^{s_1}\left(N\right)\end{array}\right]$

$X_2=\left[\begin{array}{c}{x}_2^1\\ x_2^2\\ \·\\ \·\\ \·\\ x_2^{s_2}\end{array}\right]=\left[\begin{array}{cccc}{x}_2^1\left(1\right)& x_2^1\left(2\right)& \·\·\·& x_2^1\left(N\right)\\ x_2^2\left(1\right)& x_2^2\left(2\right)& \·\·\·& x_2^2\left(N\right)\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ x_2^{s_2}\left(1\right)& x_2^{s_2}\left(2\right)& \·\·\·& x_2^{s_2}\left(N\right)\end{array}\right]$

Wherein

be the matrix of 1 × N dimension, represent in first group user that i inferior user is transferred to the data at data fusion center, be the matrix of 1 × N dimension, represent in second group user that j inferior user is transferred to the data at data fusion center.

n=1,2 ..., N represents n the sampled point of i time user in first group user;

represent n the sampled point of j time user in second group user;

3.2) by matrix X ₁and X ₂be configured to matrix $Z=\left[\begin{array}{c}{X}_1\\ X_2\end{array}\right],$ And try to achieve the estimate covariance matrix of Z $\widehat{\mathrm{\Σ}}=\frac{1}{N}{\mathrm{ZZ}}^H=\frac{1}{N}\left[\begin{array}{cc}{\widehat{\mathrm{\Σ}}}_{11}& {\widehat{\mathrm{\Σ}}}_{12}\\ {\widehat{\mathrm{\Σ}}}_{21}& {\widehat{\mathrm{\Σ}}}_{22}\end{array}\right];$

3.3) structural matrix $R={\widehat{\mathrm{\Σ}}}_{11}^{-1/2}{\widehat{\mathrm{\Σ}}}_{12}{\left({\widehat{\mathrm{\Σ}}}_{22}^{-1/2}\right)}^H,$ The characteristic value of matrix R ${\widehat{\mathrm{\λ}}}_i,i=\mathrm{1,2}...,s,$ Be the typical incidence coefficient of two groups user data, wherein

s=min (s ₁, s ₂).

Step 4, the statistic that typical incidence coefficient is constructed:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

Carry out significance test, wherein k=1,2 ..., s.Testing result is significant C _knumber is the primary user's number detecting.

Step 4 comprises following sub-step:

4.1) set the significantly level of significance α of check according to algorithm user's requirement;

4.2) the typical incidence coefficient in step 3 is carried out to following computing:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

C _k, k=1,2 ..., s obeys card side's distribution χ ²((s ₁-k+1) (s ₂-k+1));

4.3) to C _kcarry out significance test:

I. according to statistic C in the level of significance α in step 4.1 and 4.2 _kdistribution, pass through solving equation

$\mathrm{\α}={\∫}_{T_k}^{\∞}\frac{1}{2^{\frac{m}{2}}\mathrm{\Γ}\left(\frac{m}{2}\right)}{y}^{\frac{m-2}{2}}{e}^{-\frac{y}{2}}\mathrm{dy}$

Obtain the threshold T of significance test _k.Wherein, Γ represents that gamma distributes, k=1, and 2 ..., s, m=(s ₁-k+1) (s ₂-k+1); Also can draw the thresholding of significance test by inquiry chi-square distribution table $T_k={\mathrm{\χ}}_{\mathrm{\α}}^2\left((s_1-k+1)(s_2-k+1)\right).$

Ii. compare each C _k, k=1,2 ..., s and corresponding threshold T _ksize, if C _k>T _k, k statistic C is described _ksignificantly;

Iii. testing result is significant C _knumber be the number of the primary user's signal detecting.

Come below in conjunction with accompanying drawing and concrete example that the present invention is described in further detail.

Primary user's signal in step 1, network in 16 user awareness operational environments, primary user's number of working is 3.

Set a mobile cognitive radio networks and be operated in building, the urban district that the atural objects such as vehicle are more, the signal transmission in network is without direct-view path.And suppose that all users are movable within the scope of 100 meters take data fusion center as the center of circle.All cognitive radio mobile terminals are handheld terminal, and all users' translational speed is limited to as 0m/s～5m/s;

In step 2, network 16 users are the primary user's signal in perception present networks, and primary user's signal of reception is transferred to data fusion center, as shown in Figure 2.The time of supposing perception in all the same sense cycle of user is identical, and the sampling number that is transferred to data fusion center is N, and for guaranteeing the accuracy of testing result, N is the bigger the better, but the larger detection required time of N is longer.The N value 200 here;

16 users are divided at random two groups by step 3, data fusion center, and every group of number is respectively s ₁and s ₂, p=s ₁+ s ₂.S ₁and s ₂all need to be greater than the number of primary user in network.

Notice that the grouping is here the random grouping in data fusion center, can not have any impact in network user's mobility and operating state.In addition, in order to make full use of time user's data message, s ₁and s ₂value should equate or approach as far as possible.Here get s ₁=s ₂=8.

Then these two groups user data are carried out to typical association analysis, concrete steps are as follows:

3.1) data of two groups subscriber signals are formed respectively to 8 × 200 matrix X that tie up ₁and X ₂, be respectively:

$X_1=\left[\begin{array}{c}{x}_1^1\\ x_1^2\\ \·\\ \·\\ \·\\ x_1^8\end{array}\right]=\left[\begin{array}{cccc}{x}_1^1\left(1\right)& x_1^1\left(2\right)& \·\·\·& x_1^1\left(200\right)\\ x_1^2\left(1\right)& x_1^2\left(2\right)& \·\·\·& x_1^2\left(200\right)\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ x_1^8\left(1\right)& x_1^8\left(2\right)& \·\·\·& x_1^8\left(200\right)\end{array}\right]$

$X_2=\left[\begin{array}{c}{x}_2^1\\ x_2^2\\ \·\\ \·\\ \·\\ x_2^8\end{array}\right]=\left[\begin{array}{cccc}{x}_2^1\left(1\right)& x_2^1\left(2\right)& \·\·\·& x_2^1\left(200\right)\\ x_2^2\left(1\right)& x_2^2\left(2\right)& \·\·\·& x_2^2\left(200\right)\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ x_2^8\left(1\right)& x_2^8\left(2\right)& \·\·\·& x_2^8\left(200\right)\end{array}\right]$

Wherein

be the matrix of 1 × 200 dimension, represent 200 sampled points of i user's transmission in first group user;

be the matrix of 1 × 200 dimension, represent 200 sampled points of j user's transmission in second group user.

n=1,2 ..., 200 represent n the sampled point of i time user in first group user;

represent n the sampled point of j time user in second group user.

3.2) by matrix X ₁and X ₂be configured to matrix $Z=\left[\begin{array}{c}{X}_1\\ X_2\end{array}\right],$ And try to achieve the estimate covariance matrix of Z $\widehat{\mathrm{\Σ}}=\frac{1}{200}{\mathrm{ZZ}}^H=\frac{1}{200}\left[\begin{array}{cc}{\widehat{\mathrm{\Σ}}}_{11}& {\widehat{\mathrm{\Σ}}}_{12}\\ {\widehat{\mathrm{\Σ}}}_{21}& {\widehat{\mathrm{\Σ}}}_{22}\end{array}\right];$

3.3) structural matrix $R={\widehat{\mathrm{\Σ}}}_{11}^{-1/2}{\widehat{\mathrm{\Σ}}}_{12}{\left({\widehat{\mathrm{\Σ}}}_{22}^{-1/2}\right)}^H,$ The characteristic value of matrix R ${\widehat{\mathrm{\λ}}}_i,i=\mathrm{1,2}...,8,$ Be the typical incidence coefficient of two groups user data, wherein

Step 4, the statistic that typical incidence coefficient is constructed:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

Carry out significance test, wherein k=1,2 ..., 8.Testing result is significant C _knumber is the primary user's number detecting.

The step of carrying out significance test is as follows:

4.1) set significantly level of significance α=0.01 of check according to algorithm user's requirement;

4.2) the typical incidence coefficient in step 3 is carried out to following computing:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

Wherein, N=200, s=min (s ₁, s ₂)=8, obtain:

$C_k=-[200-k-\frac{1}{2}(16+1)]\ln \underset{i=k}{\overset{8}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

C _k, k=1,2 ..., 8 obey χ ²((8-k+1) (8-k+1)) distributes;

The detection probability Pd(Probability of Detection of the level of significance α here and testing result) there is direct relation, i.e. pd=1-α, its setting will, according to the specific requirement of network work, be got α=0.01 here.

4.3) according to statistic C in level of signifiance α=0.01 in 4.1 and 4.2 _kdistribution, calculate the threshold T of significance test _k,

k=1,2 ..., 8.Relatively 8 C _kwith corresponding 8 threshold values

size, meet

c _knumber, is the primary user's who detects number.

Fig. 3 has described the present invention and traditional AIC, MDL and the contrast of GDE algorithm under white noise environment.Abscissa in figure is signal to noise ratio, and ordinate is detection probability, and detection probability is under each signal to noise ratio, to carry out the same condition of 100 times to repeat experiment and draw.As can be seen from Figure 3 the method for typical association analysis detection probability in the time of high s/n ratio is stabilized in 0.99 left and right, and this and level of signifiance α=0.01st of setting, fit like a glove, i.e. pd=1-α, and the detection method that this explanation the present invention proposes is effective and feasible.Simultaneously comparing result also shows that method performance that the present invention adopts is better than the performance of other several detection algorithms, this is because utilize inferior to cooperative sensing primary user signal number under mobile environment, can effectively avoid the problem such as shadow effect and concealed terminal, improve the reliability of testing result.

Fig. 4 has described the present invention and traditional AIC, MDL and the contrast of GDE algorithm under space correlation coloured noise environment.The space correlation property coefficient of noise is 0.4.Abscissa in figure is signal to noise ratio, and ordinate is detection probability, and detection probability is under each signal to noise ratio, to carry out the same condition of 100 times to repeat experiment and draw.As can be seen from Figure 4, AIC and complete failure of MDL criterion under space correlation coloured noise, and the method applied in the present invention still can normally be worked, and performance is apparently higher than GDE algorithm.

In the description of this specification, the description of reference term " embodiment ", " some embodiment ", " example ", " concrete example " or " some examples " etc. means to be contained at least one embodiment of the present invention or example in conjunction with specific features, structure, material or the feature of this embodiment or example description.In this manual, the schematic statement of above-mentioned term is not necessarily referred to identical embodiment or example.And specific features, structure, material or the feature of description can be with suitable mode combination in any one or more embodiment or example.

Although illustrated and described embodiments of the invention, for the ordinary skill in the art, be appreciated that without departing from the principles and spirit of the present invention and can carry out multiple variation, modification, replacement and modification to these embodiment, scope of the present invention is by claims and be equal to and limit.

Claims (6)

1. the detection method to primary user's number in mobile cognitive radio networks, is characterized in that, the method comprises the following steps:

Primary user's signal in step 1, network in p time user awareness operational environment;

The primary user's signal receiving is transferred to the data fusion center in network by step 2, inferior user, and in same sense cycle, all users' sampling number is N;

P time user is divided at random two groups by step 3, data fusion center, and every group of number is respectively s ₁and s ₂, p=s ₁+ s ₂, s ₁and s ₂all be greater than the number of primary user in network; Then the data that two groups of inferior users are transferred to data fusion center are carried out typical association analysis, obtain typical incidence coefficient

and ${\widehat{\mathrm{\λ}}}_1\≥{\widehat{\mathrm{\λ}}}_2\≥\·\·\·\≥{\widehat{\mathrm{\λ}}}_S,$ s=min(s ₁,s ₂)；

Step 4, the statistic that typical incidence coefficient is constructed:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

Carry out significance test, wherein k=1,2 ..., s, testing result is significant C _knumber is the primary user's number detecting.

2. the detection method to primary user's number in mobile cognitive radio networks as claimed in claim 1, is characterized in that, described primary and secondary user can be that permanent plant can be also mobile terminal, and mobile terminal can move freely at work.

3. the detection method to primary user's number in mobile cognitive radio networks as claimed in claim 1, is characterized in that, the data that described in described step 3, two groups of inferior users are transferred to data fusion center are carried out typical association analysis, obtain typical incidence coefficient

and

s=min (s ₁, s ₂), comprising:

3.1) data that two groups of inferior users are transferred to data fusion center by data fusion center form respectively s ₁the matrix X of × N dimension ₁and s ₂the matrix X of × N dimension ₂;

3.2) by matrix X ₁and X ₂be configured to matrix $Z=\left[\begin{array}{c}{X}_1\\ X_2\end{array}\right],$ And try to achieve the estimate covariance matrix of Z $\widehat{\mathrm{\Σ}}=\frac{1}{N}{\mathrm{ZZ}}^H=\frac{1}{N}\left[\begin{array}{cc}{\widehat{\mathrm{\Σ}}}_{11}& {\widehat{\mathrm{\Σ}}}_{12}\\ {\widehat{\mathrm{\Σ}}}_{21}& {\widehat{\mathrm{\Σ}}}_{22}\end{array}\right];$

3.3) structural matrix $R={\widehat{\mathrm{\Σ}}}_{11}^{-1/2}{\widehat{\mathrm{\Σ}}}_{12}{\left({\widehat{\mathrm{\Σ}}}_{22}^{-1/2}\right)}^H,$ The characteristic value of matrix R ${\widehat{\mathrm{\λ}}}_i,i=\mathrm{1,2}...,s,$ S=min (s ₁, s ₂) be two groups of time users and be transferred to the typical incidence coefficient of the data at data fusion center.

4. the detection method to primary user's number in mobile cognitive radio networks as claimed in claim 1, is characterized in that, described step 4 comprises:

4.1) level of significance α of setting significance test;

4.2) the typical incidence coefficient in step 3 is carried out to following computing, obtain k statistic C _k:

$C_k=-[N-k-\frac{1}{2}(p+1)]\ln \underset{i=k}{\overset{s}{\mathrm{\Π}}}(1-{\widehat{\mathrm{\λ}}}_i^2),$

C _k, k=1,2 ..., s obeys card side's distribution χ ²((s ₁-k+1) (s ₂-k+1));

4.3) distribute according to the card side in level of significance α and step 4.2 in step 4.1

χ ²((s ₁-k+1) (s ₂-k+1)), to C _kcarry out significance test, by significant C _knumber as detected primary user's number.

5. the detection method to primary user's number in mobile cognitive radio networks as claimed in claim 4, is characterized in that, described step 4.3 comprises the following steps:

I. according to statistic C in the level of significance α in step 4.1 and 4.2 _kdistribution, calculate the threshold T of significance test _k;

Ii. compare each C _k, k=1,2 ..., s and corresponding threshold T _ksize, if C _k>T _k, k statistic C _ksignificantly;

Iii. be significant C by testing result _knumber as the number of the primary user's signal detecting.

6. the detection method to primary user's number in mobile cognitive radio networks as claimed in claim 5, described in calculate the threshold T of significance test _kobtain by solving following equation

$\mathrm{\α}={\∫}_{T_k}^{\∞}\frac{1}{2^{\frac{m}{2}}\mathrm{\Γ}\left(\frac{m}{2}\right)}{y}^{\frac{m-2}{2}}{e}^{-\frac{y}{2}}\mathrm{dy}$

Wherein, Γ represents that gamma distributes, k=1, and 2 ..., s, m=(s ₁-k+1) (s ₂-k+1);

Or

The described threshold T that calculates significance test _kdraw by inquiry chi-square distribution table $T_k={\mathrm{\χ}}_{\mathrm{\α}}^2\left((s_1-k+1)(s_2-k+1)\right).$

Images