CN103780318A - Dynamic double-threshold cooperative spectrum sensing method - Google Patents

Abstract

The invention discloses a dynamic double-threshold cooperative spectrum sensing method. In the local sensing process, double thresholds are set dynamically, and comparison is performed between the received master user signal energy and the double thresholds. If the master user signal energy received by cognitive users are out of the range of the double thresholds, then the detection result 0 or 1 is directly transferred to a cognitive basestation; and if the received master user energy is between the double thresholds, then a trust function is calculated according to the D-S evidence theory and transferred to the cognitive basestation for comprehensive treatment, the double thresholds are set dynamically according to different channel gains, and the final judgment is made by the cognitive basestation according to the K -rank criteria. According to the method of the invention, the detection probability can be improved without improving the false-alarm probability, so that the spectrum sensing quality can be effectively improved.

Description

A kind of dynamically double-threshold cooperative spectrum sensing method

Technical field

The present invention relates to a kind of cooperative frequency spectrum sensing method, relate in particular to a kind of dynamically double-threshold cooperative spectrum sensing method, belong to cognition wireless electrical domain.

Background technology

Since 20 century 70s, along with the development of semiconductor, microelectronics and computer technology, mobile communication technology has obtained swift and violent development and application.Enter 21 century, the emphasis of radio communication needs the service of broad spectrum and higher download speed to shift to wireless Internet, multimedia communication etc.From mobile phone to wireless Internet, people expect can obtain reliable broadband network whenever and wherever possible and connect.But, due to the restriction of the conditions such as mobile terminal antenna size and power, can be very limited for the frequency range of practical application.Frequency spectrum resource has become a kind of scarce resource, becomes more and more nervous.

In order to alleviate the problem of frequency spectrum resource anxiety, improve spectrum utilization efficiency, some scholars have proposed the concept of cognitive radio (CR, Cognitive Radio).Its essence is to allow those Wireless Telecom Equipments with cognitive ability to adopt the method for chance formula to take the frequency spectrum of having authorized, and the access of unauthorized user (being these cognitive device) can not cause interference to authorized user communication.This resource of frequency range that can be utilized (time domain, spatial domain, frequency domain) is known as " frequency spectrum cavity-pocket ".Whole cognitive process comprises frequency spectrum perception, spectrum management and the large key technology of frequency spectrum share three.

Frequency spectrum perception technology is key components of cognitive radio.Cognitive radio is found frequency spectrum hole by frequency spectrum being carried out to constantly detection, and utilizes frequency spectrum hole to communicate.In the time that frequency spectrum perception technology can not effectively detect the validated user signal of authorizing in frequency range, use the wireless device of cognitive radio function to cause interference to other equipment that are operated in same frequency range, therefore accurate, reliable, continuous frequency spectrum perception technology is the basis of cognitive radio, is the prerequisite that realizes the application of the cognitive radio such as spectrum management, frequency spectrum share.

Frequency spectrum perception is divided into single user awareness and collaborative spectrum sensing.The cognition technology research of alone family relatively early, mainly contains matched filtering detection, energy measuring and cyclostationary characteristic and detects, and wherein energy measuring is because complexity is low and realize simple extensive use.The performance of single user awareness is easily subject to shadow fading impact in wireless channel.Collaborative spectrum sensing can be avoided the uncertainty of simple local frequency spectrum perception, has reduced the impact of multipath fading, concealed nodes and shadow effect.The process of cooperation perception roughly can be divided into local perception, perception information transmission and perception data and merge three phases.In existing research, major part is selected the standard of energy detection algorithm as this locality judgement.The thresholding choosing method of energy detection algorithm mainly contains two kinds of simple gate limit and double thresholds.Simple gate limit energy detection algorithm is more responsive to interference ratio such as noises, though double threshold energy detection algorithm can reduce the collision probability between cognitive user and primary user, but idle frequency spectrum fully applied, frequency spectrum service efficiency also has room for promotion.In frequency spectrum perception, conventional data fusion criterion has at present: "AND" fusion criterion, "or" fusion criterion, " K order " fusion criterion, maximum a posteriori probability fusion criterion, Bayesian Fusion criterion.

Traditional double threshold method is generally given up the cognitive user between thresholding, is unfavorable for improving detection probability.Therefore be necessary with someway the cognitive user between thresholding also being used.D-S evidence theory is one of method of processing uncertain information.The people such as Zhi Quan propose D-S evidence theory to be applied in double threshold frequency spectrum perception, have obtained good frequency spectrum perception effect.But, the double threshold of this method is fixed, and does not consider in actual wireless network, because the position of cognitive user is random distribution, the wireless channel decline of signal transmission experience is different with shade impact, and the received signal to noise ratio of each cognitive user is also different.In the time that signal to noise ratio is too low, the error probability that cognitive user is directly entered a judgement is larger, should suitably adjust double threshold.But, the number of times that thresholding is adjusted is unsuitable too many, otherwise will affect the real-time of frequency spectrum perception.

Summary of the invention

Technical problem: the object of this invention is to provide a kind of dynamically double-threshold cooperative spectrum sensing method, to solve in the situation that signal to noise ratio is too low, the large problem of error probability that cognitive user utilizes the method for traditional fixing double threshold to enter a judgement.The double threshold adjustment algorithm of this method is dynamically to adjust double threshold according to channel status, also can obtain high detection probability in low signal-to-noise ratio situation, thereby has improved the overall performance of frequency spectrum perception.

Technical scheme: dynamic double-threshold cooperative spectrum sensing method of the present invention, comprises the steps:

1), in the time of local perception, each cognitive user utilizes energy detection algorithm to draw energy detection value Yi;

2) utilize double threshold adjustment algorithm to calculate double threshold η ₀and η ₁, by the energy detection value Y calculating in described step 1) _iwith double threshold η ₀and η ₁relatively, if Y _i> η ₁, send D _i=1 to fusion center, enters step 4); If Y _i< η ₀, send D _i=0 to fusion center, enters step 4); If η ₀≤ Y _i≤ η ₁, enter step 3);

3) extract Y according to D-S evidence theory _idegree of belief information m _i(H ₀), m _i(H ₁) and m _i(Ω), send to fusion center, enter step 4);

4) fusion center synthesizes the degree of belief information transmitting in described step 3), obtains court verdict D, comprehensive described step 2) in the court verdict D that transmits _i, make final judging result according to " K order " criterion, judge whether primary user exists.

The step 2 of the inventive method) in, adjust by the following method double threshold:

1) initialization, sets and expects false alarm probability P _f ^*, expect detection probability P _d ^*, making initial value k=3/2, Δ=10logk, adjusts number of times C=0.Utilize " invariable false alerting method " to calculate threshold value η, η ₀=η/k, η ₁=k η; Calculate again actual detection probability P _d; If P _d< P _d ^*, make Δ=Δ+0.5, forward step 2 to);

2) make C=C+1, if C > 4 forwards step 4) to;

3) calculate actual detection probability P _d; If P _d< P _d ^*, make Δ=Δ+0.5, forward described step 2 to);

4) finish.

Beneficial effect: the present invention compared with prior art, has the following advantages:

1, the present invention program is in the too low situation of signal to noise ratio, and the large problem of error probability that cognitive user utilizes the method for traditional fixing double threshold to enter a judgement, has proposed a kind of dynamic double-threshold cooperative spectrum sensing method based on D-S evidence theory.In the time of local perception, dynamically adjust double threshold according to Real-time Channel state, the cognitive user of energy detection value outside double threshold directly entered a judgement, and court verdict is sent to fusion center; The cognitive user of energy detection value within double threshold extracted degree of belief information according to D-S evidence theory, then sends to fusion center.The information that fusion center comprehensively receives makes final justice to judge whether primary user exists.

2, the amount of calculation that algorithm of the present invention is introduced is substantially negligible, and has increased the detection probability in low signal-to-noise ratio situation, has improved the overall performance of frequency spectrum perception.

Accompanying drawing explanation

Fig. 1 is the perception scene schematic diagram of cognitive user of the present invention.

Fig. 2 is double threshold adjustment algorithm flow chart of the present invention.

Fig. 3 is the dynamic double-threshold cooperative spectrum sensing method flow chart based on D-S evidence theory of the present invention.

Embodiment

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

Fig. 1 is cognitive user perception scene schematic diagram of the present invention.What the present invention considered is the perception scene that has fusion center, primary user and cognitive user, and wherein each part effect is as follows:

Primary User (PU): represent primary user, claim again authorized user, the object of cognitive user is to want perception and determine that can take it authorizes frequency spectrum.

Cognitive User (CR): represent cognitive user, the signal of being responsible for primary user PU to send carries out local perception, and local sensing results is sent to cognitive base station.

Base Station (BS): represent cognitive base station, use as fusion center, the main sensing results of being responsible for receiving each cognitive user CR, and sensing results is done to integrated treatment and to whether existing primary user's signal to adjudicate.

Fig. 3 is the dynamic double-threshold cooperative spectrum sensing method flow chart based on D-S evidence theory of the present invention.When local perception, each cognitive user utilizes energy detector to obtain detection statistic Y _i:

$Y_i=\frac{1}{N}\underset{t=1}{\overset{N}{\mathrm{\&Sigma;}}}{\left|x_i\left(t\right)\right|}^2$

Wherein, N is sampling number (being average detected duration) in perceptual signal time T, x _i(t) be i the signal that cognitive user receives.

When sampling number N enough large, Y _inormal Distribution:

$\left\{\begin{array}{cc}{H}_0& Y_i~N({\mathrm{\&sigma;}}_i^2,\frac{2}{N}{\mathrm{\&sigma;}}_i^4)\\ H_1& Y_i~N({\mathrm{\&sigma;}}_i^2(1+{\mathrm{\&gamma;}}_i),\frac{2}{N}{\mathrm{\&sigma;}}_i^4(1+2{\mathrm{\&gamma;}}_i))\end{array}\right.$

Wherein N (μ, σ ²) expression average is μ, variance is σ ²normal distribution, ${\mathrm{\&gamma;}}_i=(1/N_0){\&Integral;}_0^T{\left[s\left(t\right)h_i\left(t\right)\right]}^2\mathrm{dt}$ For signal power to noise power ratio.

Traditional single threshold algorithm has provided detection probability (P _d, Detection Probability) and false alarm probability (P _f, False Alarm Probability) expression formula:

$P_d=P\{Y_i>\mathrm{\&eta;}|H_1\}=1-\mathrm{\&Phi;}\left(\frac{\mathrm{\&eta;}-{\mathrm{\&sigma;}}_i^2(1+{\mathrm{\&gamma;}}_i)}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4(1+2{\mathrm{\&gamma;}}_i)}}\right)$

$P_f=P\{Y_i>\mathrm{\&eta;}|H_0\}=1-\mathrm{\&Phi;}\left(\frac{\mathrm{\&eta;}-{\mathrm{\&sigma;}}_i^2}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4}}\right)$

Visible false alarm probability is only relevant with thresholding, if the false alarm probability of a given expectation, thresholding can be expressed as:

$\mathrm{\&eta;}=\left[{\mathrm{\&Phi;}}^{-1}(1-P_f)\right]\&CenterDot;\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4}+{\mathrm{\&sigma;}}_i^2$

This thresholding method to set up is called " invariable false alerting method ".

Due to the uncertainty of channel circumstance, in the time that reception signal energy value is near thresholding η, be difficult to do reliable judgement.For this reason, double threshold energy measuring definition ［ η/k, k η ］ is the confusion region of judgement, the scope adjustment factor that k >=1 is confusion region.

Within the energy value receiving falls within double threshold, algorithm of the present invention is obtained basic trust partition function according to D-S evidence theory, sends to BS to make conclusive judgement degree of belief information.If η ₀=η/k, η ₁=k η, in the time carrying out local double threshold detection, detection probability and false alarm probability are expressed as follows:

$P_d=P\{Y_i>{\mathrm{\&eta;}}_1|H_1\}+P\{{\mathrm{\&eta;}}_0<Y_i\&le;{\mathrm{\&eta;}}_1\left|H_1\right\}\&CenterDot;m_i\left(H_1\right)$

$=1-\mathrm{\&Phi;}\left(\frac{{\mathrm{\&eta;}}_1-\frac{1}{N}{\mathrm{\&sigma;}}_i^2(1+{\mathrm{\&gamma;}}_i)}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4(1+2{\mathrm{\&gamma;}}_i)}}\right)+[\mathrm{\&Phi;}\left(\frac{{\mathrm{\&eta;}}_1-\frac{1}{N}{\mathrm{\&sigma;}}_i^2(1+{\mathrm{\&gamma;}}_i)}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4(1+2{\mathrm{\&gamma;}}_i)}}\right)-\mathrm{\&Phi;}\left(\frac{{\mathrm{\&eta;}}_0-\frac{1}{N}{\mathrm{\&sigma;}}_i^2(1+{\mathrm{\&gamma;}}_i)}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4(1+2{\mathrm{\&gamma;}}_i)}}\right)]m_i\left(H_1\right)$

$P_f=P\{Y_i>{\mathrm{\&eta;}}_1|H_0\}+P\{{\mathrm{\&eta;}}_0<Y_i\&le;{\mathrm{\&eta;}}_1\left|H_0\right\}\&CenterDot;m_i\left(H_0\right)$

$=1-\mathrm{\&Phi;}\left(\frac{{\mathrm{\&eta;}}_1-\frac{1}{N}{\mathrm{\&sigma;}}_i^2}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4}}\right)+[\mathrm{\&Phi;}\left(\frac{{\mathrm{\&eta;}}_1-\frac{1}{N}{\mathrm{\&sigma;}}_i^2}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4}}\right)-\mathrm{\&Phi;}\left(\frac{{\mathrm{\&eta;}}_0-\frac{1}{N}{\mathrm{\&sigma;}}_i^2}{\sqrt{\frac{2}{N}{\mathrm{\&sigma;}}_i^4}}\right)]m_i\left(H_0\right)$

Wherein, m _i(H ₁) and m _i(H ₀) be the basic trust partition function of cognitive user i.

Basic trust partition function is exactly the degree of belief information of extracting according to D-S evidence theory, and specific formula for calculation is:

$m_i\left(H_0\right)={\&Integral;}_{Y_i}^{+\&infin;}\frac{1}{\sqrt{2\mathrm{\&pi;}}{\mathrm{\&sigma;}}_{0i}}{e}^{(-\frac{{(x-{\mathrm{\&mu;}}_{0i})}^2}{2{\mathrm{\&sigma;}}_{0i}^2})}\mathrm{dx}$

$m_i\left(H_1\right)={\&Integral;}_{-\&infin;}^{Y_i}\frac{1}{\sqrt{2\mathrm{\&pi;}}{\mathrm{\&sigma;}}_{1i}}{e}^{(-\frac{{(x-{\mathrm{\&mu;}}_{1i})}^2}{2{\mathrm{\&sigma;}}_{1i}^2})}\mathrm{dx}$

M _i(H ₁) degree of belief of the local sensing results of expression cognitive user i to primary user's existential proposition; m _i(H ₀) represent that local sensing results is to the not degree of belief of existential proposition of primary user.In addition, credit assignment is to H ₀and H ₁remaining part afterwards, in order to express conveniently, uses m _i(Ω) represent.M _i(H ₁), m _i(H ₀) and m _i(Ω) be exactly the degree of belief information of cognitive user i.

Suppose that each cognitive user has signal-to-noise ratio (SNR) estimation ability.If initial signal to noise ratio snr ₀=10log γ ₀(dB), in testing process, if signal to noise ratio amplitude of variation | SNR-SNR ₀| SNR ₀≤ 0.5, thresholding does not adjust; Otherwise, upgrade initial signal to noise ratio snr ₀, and adjust double threshold according to following algorithm.Fig. 2 is the flow chart of double threshold adjustment algorithm.

Double threshold adjustment algorithm:

1) initialization, sets and expects false alarm probability P _f ^*, expect detection probability P _d ^*, making initial value k=3/2, Δ=10logk, adjusts number of times C=0.Utilize " invariable false alerting method " to calculate threshold value η, η ₀=η/k, η ₁=k η; Calculate again actual detection probability P _d; If P _d< P _d ^*, make Δ=Δ+0.5, forward step 2 to);

2) make C=C+1, if C > 4 forwards step 4) to;

3) calculate actual detection probability P _d; If P _d< P _d ^*, make Δ=Δ+0.5, forward described step 2 to);

4) finish.

When cognitive user i obtains two threshold value η ₀and η ₁afterwards, if Y _i> η ₁, send D _i=1 to BS; If Y _i< η ₀, send D _i=0 to BS; If η ₀≤ Y _i≤ η ₁, by degree of belief information m _i(H ₁), m _i(H ₀) and m _i(Ω) send to BS.

In cognitive base station, there is local cognitive user number M, supposes to receive n D _i, the cognitive user number that is positioned at so confusion region is (M-n).The degree of belief information sending over for (M-n) individual cognitive user, cognitive base station adopts D-S evidence composition rule to be fused into final degree of belief function:

$m\left(H_0\right)=\frac{\underset{A_1\&cap;A_2\&cap;..\&cap;A_{M-n}=H_0}{\mathrm{\&Sigma;}}\underset{i=1}{\overset{M-n}{\mathrm{\&Pi;}}}{m}_i\left(A_i\right)}{1-\underset{A_1\&cap;A_2\&cap;..\&cap;A_{M-n}=\mathrm{\&Phi;}}{\mathrm{\&Sigma;}}\underset{i=1}{\overset{M-n}{\mathrm{\&Pi;}}}{m}_i\left(A_i\right)}$

$m\left(H_1\right)=\frac{\underset{A_1\&cap;A_2\&cap;..\&cap;A_{M-n}=H_1}{\mathrm{\&Sigma;}}\underset{i=1}{\overset{M-n}{\mathrm{\&Pi;}}}{m}_i\left(A_i\right)}{1-\underset{A_1\&cap;A_2\&cap;..\&cap;A_{M-n}=\mathrm{\&Phi;}}{\mathrm{\&Sigma;}}\underset{i=1}{\overset{M-n}{\mathrm{\&Pi;}}}{m}_i\left(A_i\right)}$

Obtain court verdict according to degree of belief function:

$D=\left\{\begin{array}{cc}1& m\left(H_1\right)/m\left(H_0\right)\&GreaterEqual;1\\ 0& m\left(H_1\right)/m\left(H_0\right)<1\end{array}\right.$

Utilize K order criterion to do final decision: to get K=M/2, obtain following judgement formula:

。

Claims (2)

1. a dynamic double-threshold cooperative spectrum sensing method, is characterized in that, the method includes the steps of:

Step 1: in the time of local perception, each cognitive user utilizes energy detection algorithm to draw energy detection value Y _i;

Step 2: utilize double threshold adjustment algorithm to calculate double threshold η ₀and η ₁, by the energy detection value Y calculating in described step 1 _iwith double threshold η ₀and η ₁relatively, if Y _i> η ₁, send D _i=1 to fusion center, enters step 4; If Y _i< η ₀, send D _i=0 to fusion center, enters step 4; If η ₀≤ Y _i≤ η ₁, enter step 3;

Step 3: extract Y according to D-S evidence theory _idegree of belief information m _i(H ₀), m _i(H ₁) and m _i(Ω), send to fusion center, enter step 4;

Step 4: fusion center synthesizes the degree of belief information transmitting in described step 3, obtains court verdict D, the court verdict D transmitting in comprehensive described step 2 _i, make final judging result according to " K order " criterion, judge whether primary user exists.

2. the dynamic double-threshold cooperative spectrum sensing method of one according to claim 1, is characterized in that, in described step 2, adjusts by the following method double threshold:

Step 1: initialization, set and expect false alarm probability P _f ^*, expect detection probability P _d ^*, making initial value k=3/2, Δ=10logk, adjusts number of times C=0.Utilize " invariable false alerting method " to calculate threshold value η, η ₀=η/k, η ₁=k η; Calculate again actual detection probability P _d; If P _d< P _d ^*, make Δ=Δ+0.5, forward step 2 to;

Step 2: make C=C+1, if C > 4 forwards step 4 to;

Step 3: calculate actual detection probability P _d; If P _d< P _d ^*, make Δ=Δ+0.5, forward described step 2 to;

Step 4: finish.

Images