CN105187141A - Anti-attack distributed cooperative spectrum sensing method - Google Patents

Abstract

The invention discloses an anti-attack distributed cooperative spectrum sensing method. According to the method, certain cognitive users are likely to become malicious users due to attack during the cooperative spectrum sensing so as to send unsecure sensing data and influence the cooperative sensing performance. At present, the researchers frequently research the typical malicious users in two states, namely, the cognitive users launching "Always Yes" attack and the cognitive users launching "Always No" attack; in allusion to the problem, the method is characterized in that the cognitive users with better detection performance are screened to participate the cooperation, parameters such as signal to noise ratio and sensing matching rate are utilized to define the consensus degree of the cognitive users, the consensus degree is used for representing the differences of different cognitive users in the aspects of detection performance and sensing result security and the original state values, which are iterated and updated, of the cognitive users are weighted so as to reduce the influence on the final cooperative sensing judgement from the cognitive users which are relatively bad in detection performance and unsecure in sensing result and improve the whole sensing performance of the system; and the method has the advantages of effectively resisting attack and improving the spectrum sensing performance of the distributed cooperative sensing.

Description

A kind of distributed collaborative frequency spectrum sensing method of attack resistance

Technical field

The present invention relates to a kind of distributed collaborative frequency spectrum sensing method of attack resistance, belong to cognitive radio technology field.

Background technology

The demand of modern wireless communication systems to frequency spectrum resource is increasing, the restriction condition becoming wireless communication technology development already in short supply of frequency spectrum resource.But find by carrying out investigation to the service condition of frequency spectrum resource, some authorizes the availability of frequency spectrum of frequency range extremely low, life period and idle phenomenon spatially; On the contrary, some unauthorized frequency ranges are but overused.Thus, the anxiety of frequency spectrum resource is caused by the unreasonable of spectrum management policy to a great extent, not because lack available spectrum resources physically.

Existence in view of the above problems, has scholar to propose a kind of novel intelligent wireless communication technology, i.e. cognitive radio technology.This technology in real time to needing the frequency range used to carry out perception, can adopt dynamical fashion to carry out redistributing of frequency spectrum, achieving cognitive user reusing idle frequency range, reaching the object improving the availability of frequency spectrum.Realize cognitive radio technology to need to be based upon on the basis of frequency spectrum perception, thus frequency spectrum perception is a problem being worth research.

The core technology of cognitive radio comprises frequency spectrum perception, spectrum management and frequency spectrum share.Frequency spectrum perception is the key and the prerequisite that realize cognitive radio, the study hotspot of our present stage. according to the cognitive user number participating in collaborative sensing, frequency spectrum perception can be divided into single user perception and multi-user Cooperation perception, wherein the collaborative sensing of multi-user is compared single user perception and can better be tackled the problem such as shadow effect and concealed terminal in practical radio communication environment, can effectively improve frequency spectrum detection performance.

Existing collaborative sensing method is mostly for centralized network, and in distributed network, owing to not having fusion center, centralized collaborative sensing method will be no longer applicable, and therefore, the research of distributed collaborative perception is very necessary.Existing distributed collaborative cognitive method is usually given tacit consent to all cognitive user and is all participated in cooperation, and all cognitive user carry out state updating with equal identity, such detection perform cognitive user that is poor and that be subject to malicious attack and send sense of insecurity primary data can produce worse impact to the performance of collaborative sensing.Therefore, need in distributed collaborative perception, take certain mechanism to filter out the not good cognitive user of detection perform and distinguish the detection perform of different cognitive users and the fail safe of testing result.And the present invention can solve problem above well.

Summary of the invention

The object of the invention is to provide a kind of distributed collaborative frequency spectrum sensing method of attack resistance, the method introduces the selection mechanism of cognitive user, the cognitive user choosing more excellent detection perform participates in cooperation, and with the common recognition degree of signal to noise ratio and these two parameter defined cognitive users of perception matching degree, common recognition degree is utilized to be weighted the energy value that cognitive user iteration upgrades, the detection perform of different cognitive users and the fail safe of sensing results are distinguished, reduce the lower cognitive user of common recognition degree to the impact of collaborative sensing, reach the object improving collaborative sensing performance.

The present invention solves the technical scheme that its technical problem takes: a kind of distributed collaborative frequency spectrum sensing method of attack resistance, the method comprises the steps:

Step 1: in known noise variance primary user transmitted power P, the distance d between cognitive user i and primary user _i, the perception path fading factor-beta of cognitive user i _iwhen, each cognitive user foundation estimate self signal to noise ratio, and share signal-tonoise information with neighbor user, foundation calculate decision threshold v, self signal to noise ratio is compared with v, select K cognitive user and participate in collaborative spectrum sensing.

Step 2: energy measuring is carried out to K the cognitive user filtered out, obtains energy measuring statistical value X _iand local court verdict R _i, foundation calculate common recognition degree, and by formula Z _i(0)=w _i(l) Y _i(0) the initial value Z carrying out Gradient Iteration is obtained _i(0).

Step 3: participate in the cognitive user of cooperation according to oneself state Y _it state Y that () and normal neighbor node transmit _ij(t), according to more new state, iteration renewal process is continued until that cooperative cognitive User Status value converges to common value A ^*.

Step 4: to fixed system false alarm probability, set decision threshold λ according to " invariable false alerting method ", cognitive user is by A ^*compare with decision threshold, foundation $decision=\left\{\begin{array}{ccc}{H}_0& if& A^{*}<\mathrm{\&lambda;}\\ H_1& if& A^{*}\&GreaterEqual;\mathrm{\&lambda;}\end{array}\right.$ Decision rule make the conclusive judgement whether primary user exist.

Step 5: court verdict is broadcasted by cognitive user in cognition network.

Step 6: according to court verdict, cognitive user determines whether use primary user's frequency range to communicate at data transmission slots.If cognitive user judgement primary user exist, then keep mourning in silence at data transmission slots; Otherwise, then this frequency range can be utilized to carry out the transfer of data of self.Meanwhile, when cognitive user position changes or cognitive user occurs equipment fault, re-execute initial phase, namely return 1, otherwise next round detects from 2.

The definition of the cognitive user common recognition degree in above-mentioned steps 2 of the present invention, comprising:

Common recognition degree in the present invention should reflect the detection perform of cognitive user, the sensing results whether safety (namely whether cognitive user is malicious user) of cognitive user can be differentiated again, thus its formation relates to two parameters, namely the signal to noise ratio of cognitive user and the local sensing results of cognitive user with cooperate after the matching degree (referred to as perception matching degree) of global decision, and the definition of degree of common recognition is wanted to meet and increased along with signal to noise ratio and increase, along with perception matching degree increases and the rule of increase.Suppose that the perception matching degree of starting stage cognitive user is 1, then the common recognition degree of starting stage is determined by the signal to noise ratio of cognitive user, and signal to noise ratio larger then common recognition degree is larger.

Because the signal to noise ratio of cognitive user and the change of perception matching degree all can cause the change of its common recognition degree, and these two parameters characterize the detection perform of cognitive user and the fail safe of sensing results respectively, therefore cognitive user detection perform change or under attackly can be embodied by common recognition degree, the primary iteration value of common recognition degree to the information interaction stage is then utilized to be weighted, just can different cognitive user be distinguished, make full use of the cognitive user that detection perform is good and sensing results is safe to cooperate, improve the performance of collaborative sensing.Meanwhile, when malicious user mobilizes " AlwaysYes " or " AlwaysNo " attacks, because perception matching degree is 0, then common recognition degree is 0, is also reached the object of antagonism malicious user attack by the weighting of common recognition degree.

Beneficial effect:

1, the present invention first carries out the selection of cooperative cognitive user, chooses detection perform preferably node participation cooperation, avoids the overall performance that the poor cognitive user of detection perform drags down collaborative sensing.

2, the present invention introduces the concept of common recognition degree, common recognition degree is determined by the signal to noise ratio of cognitive user and perception matching degree two parameters, signal to noise ratio can reflect the difference of cognitive user detection perform, perception matching degree effectively can differentiate whether cognitive user is malicious user, therefore signal to noise ratio is larger, perception matching degree is higher, then common recognition degree is larger.

3, the present invention utilizes the primary iteration value of common recognition degree to the cognitive user information interaction stage to be weighted, and can reduce signal to noise ratio is low, perception matching degree is not high cognitive user to the impact of collaborative sensing performance, effectively can improve the perceptual performance of system.

4, the present invention makes full use of the cognitive user that detection perform is good and sensing results is safe and cooperates, and reaches the object that antagonism malicious user is attacked, can improve the performance of collaborative sensing simultaneously.

Accompanying drawing explanation

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

Fig. 2 is method flow diagram of the present invention.

Embodiment

Below in conjunction with Figure of description, the invention is described in further detail.

As shown in Figure 1, it is considered that there is not fusion center, have the distributed frequency spectrum perception scene of 1 primary user and some cognitive user in the present invention, wherein each part effect comprises as follows:

PrimaryUser (PU): represent primary user, also known as authorized user, can the object of cognitive user want perception and determine take its frequency spectrum.

CognitiveUser (CU): represent cognitive user, is responsible for carrying out local perception to the signal that primary user PU sends, and carries out information interaction between individual cognitive user, and makes the conclusive judgement whether primary user exist.

The distributed collaborative cognitive method of a kind of attack resistance as shown in Figure 2, can be divided into four-stage, comprise:

One, initial phase: cognitive user i (i=1,2,3 ..., N) and estimate self received signal to noise ratio γ _i, then share signal-tonoise information with neighbor user, then calculate signal to noise ratio decision threshold v, finally by γ _icompare with v, judge oneself whether to participate in cooperation: work as γ _iduring>=v, this cognitive user detection perform is better, participate in collaborative sensing, otherwise detection perform is poor, does not participate in cooperation.

Two, distributed collaborative perception stage: suppose that the stage 1 filters out the preferably cognitive user participation cooperation of K detection perform, then this K cognitive user enters the distributed collaborative perception in stage 2, and this stage divides two steps to realize:

Step 1: each cognitive user participating in cooperation adopts energy detection algorithm to carry out local perception, obtains this locality judgement R whether primary user exists _ifor subsequent use, R _ithe γ obtained with the stage 1 _ithe foundation of cognitive user common recognition degree will be upgraded, according to formula Z in each frequency spectrum perception _i(0)=w _i(l) Y _i(0) iteration initial value Z is obtained _i(0), namely during t=0, the state value of cognitive user i is Z _i(0).

Step 2: cooperative cognitive user and their neighbours set up duplex communication link, suppose that the duplex link between cognitive user is not interrupted before their energy convergence, all cognitive user in network can know oneself neighbor node, and the position of cognitive user remains unchanged in the process of each frequency spectrum perception.At t=1,2 ... time, cognitive user i and its neighbor user carry out information interaction and the iteration that foundation formula (4.2) carries out state value upgrades.When t is enough large, the state value of all cognitive user will be tending towards a common value iteration stopping.Each cognitive user is by A ^*compare with the decision threshold λ preset, make according to following criterion the conclusive judgement whether primary user exist.

$decision=\left\{\begin{array}{ccc}{H}_0& if& A^{*}<\mathrm{\&lambda;}\\ H_1& if& A^{*}\&GreaterEqual;\mathrm{\&lambda;}\end{array}\right.$

Three, broadcast phase: court verdict is broadcasted by cognitive user in cognition network.

Four, data transfer phase: if cognitive user judgement primary user exists, then keep mourning in silence at data transmission slots; Otherwise then cognitive user can utilize this frequency range to carry out the transfer of data of self.Notice, initial phase estimates that the factors such as the path fading factor of the signal to noise ratio that obtains and the distance between cognitive user and primary user and cognitive user are relevant, therefore, in order to better adapt to dynamic wireless environments, once the position of cognitive user change or participate in cooperate cognitive user break down, namely the toggling init stage is needed again, again performance preferably cognitive user participation collaborative sensing is chosen, that is, when the cognitive user that cognitive user change in location or participation cooperate breaks down, new round perception will from initial phase, otherwise, new round perception is by directly from distributed collaborative perception stage.

The difference that common recognition degree distinguishes different cognitive users is defined in stage two of the present invention, the primary iteration value of common recognition degree to the cognitive user information interaction stage is utilized to be weighted, reduce the lower cognitive user of common recognition degree to the impact of collaborative sensing performance, improve the overall recognition performance of system.

Common recognition degree as the tolerance of the credibility to people, also can weigh the credibility of cognitive user sensing results.Common recognition degree is higher, and the degree that the sensing results of cognitive user is utilized is higher, otherwise then lower.Common recognition degree in this algorithm should reflect the detection perform of cognitive user, the sensing results whether safety (that is: whether cognitive user is malicious user) of cognitive user can be differentiated again, thus its formation relates to two parameters, namely the signal to noise ratio of cognitive user and the local sensing results of cognitive user with cooperate after the matching degree (referred to as perception matching degree) of global decision, and the definition of degree of common recognition is wanted to meet and is increased along with signal to noise ratio and increase, along with perception matching degree increases and the rule of increase, specifically comprise:

(1) signal to noise ratio of cognitive user

Signal to noise ratio can reflect the detection perform of cognitive user, and the larger detection perform of signal to noise ratio is better.In this algorithm, before collaborative sensing starts, in order to selectivity preferably cognitive user participation cooperation, each cognitive user is estimated self signal to noise ratio, therefore, the i-th (i=1,2 obtained by formula (4.8) can directly be utilized here,, N) and the signal to noise ratio γ of individual cognitive user _i.

(2) perception matching degree

In collaborative spectrum sensing, some cognitive user likely becomes malicious user because of under attack, sends unsafe perception data, affects the performance of collaborative sensing.At present, researcher often studies the typical malicious user of two states, the cognitive user of namely mobilizing " AlwaysYes " to attack and the cognitive user of mobilizing " AlwaysNo " to attack." AlwaysYes " attacks the perception information referring to that cognitive user always sends " 1 " or be greater than actual value, which increase the false alarm probability of cognitive user, make cognitive user can not utilize idle frequency range in time, missed communications chance, cause cognitive system throughput to reduce." AlwaysNo " attacks the perception information referring to that cognitive user always sends " 0 " or be less than actual value, that reduces the detection probability of cognitive user, well can not protect the communication of primary user.These two kinds of malicious attacks all create adverse effect to collaborative sensing, should in the algorithm by effective filtering.

The present invention can consider to utilize perception matching degree to differentiate that whether cognitive user is for malicious user, and when making perception matching degree large, common recognition degree also increases.The reflection of perception matching degree be the local court verdict of cognitive user with cooperate after the consistent degree of global decision result, as the local court verdict R of cognitive user i _iconsistent with the global decision result decision after cooperation, its perception matching degree increases, otherwise reduces.The sensing results in cognitive user i a certain moment may be one of following four kinds of states

S ₁₁: global decision result decision is H ₁time, the local court verdict R of cognitive user i _ialso be H ₁;

S ₀₀: global decision result decision is H ₀time, the local court verdict R of cognitive user i _ialso be H ₀;

S ₁₀: global decision result decision is H ₁time, the local court verdict R of cognitive user i _ifor H ₀;

S ₀₁: global decision result decision is H ₀time, the local court verdict R of cognitive user i _ifor H ₁.

Note cognitive user i l (l=1,2 ...) secondary perception time, state S ₁₁, S ₀₀, S ₁₀, S ₀₁the total degree of respective generation is then the perception matching degree of cognitive user i accumulation can be expressed as

$r_i\left(l\right)=\frac{n_{11}^i\left(l\right)}{n_{11}^i\left(l\right)+n_{10}^i\left(l\right)}\&CenterDot;\frac{n_{00}^i\left(l\right)}{n_{00}^i\left(l\right)+n_{01}^i\left(l\right)}$

Attack if cognitive user is started " AlwaysYes ", because detection probability and false alarm probability are all 1, so the Section 2 of this formula is 0, now perception matching degree is 0; In like manner, if cognitive user mobilizes " AlwaysNo " to attack, because detection probability and false alarm probability are all 0, so the Section 1 of this formula is 0, now perception matching degree is also 0.Therefore, whether it is malicious user to utilize the perception matching degree of cognitive user effectively to differentiate, when cognitive user is malicious user, its perception matching degree is 0, otherwise, be not 0.

(3) common recognition degree

Consider that signal to noise ratio is different from the evaluation criteria of these two parameters of perception matching degree, be thus normalized

${\mathrm{\&gamma;}}_i^{*}=\frac{{\mathrm{\&gamma;}}_i}{{\mathrm{\&gamma;}}_{\max }}$

$r_i^{*}\left(l\right)=\frac{r_i\left(l\right)}{\underset{i\&Element;\{1,2,...,K\}}{max}\left\{r_i\left(l\right)\right\}}$

According to above parameter, in this algorithm, the common recognition degree of defined cognitive user i when the l time perception is

$w_i\left(l\right)=\frac{w_i^{*}\left(l\right)}{\underset{i=1}{\overset{K}{\&Sigma;}}{w}_i^{*}\left(l\right)}$

Wherein, i={1,2 ..., K}, l={1,2 ..., K is collaboration user number, and

$w_i^{*}\left(l\right)={\mathrm{\&gamma;}}_i^{*}\&CenterDot;r_i^{*}(l-1)$

The cognitive user common recognition degree of the present invention's definition increases along with the increase of signal to noise ratio, increases along with the increase of perception matching degree, meets algorithm requirement.Suppose that the perception matching degree of starting stage cognitive user is 1, then the common recognition degree of starting stage is determined by the signal to noise ratio of cognitive user, and signal to noise ratio larger then common recognition degree is larger.

Because the signal to noise ratio of cognitive user and the change of perception matching degree all can cause the change of its common recognition degree, and these two parameters characterize the detection perform of cognitive user and the fail safe of sensing results respectively, therefore cognitive user detection perform change or under attackly can be embodied by common recognition degree, the primary iteration value of common recognition degree to the information interaction stage is then utilized to be weighted, just can different cognitive user be distinguished, make full use of the cognitive user that detection perform is good and sensing results is safe to cooperate, improve the performance of collaborative sensing.Meanwhile, when malicious user mobilizes " AlwaysYes " or " AlwaysNo " attacks, because perception matching degree is 0, then common recognition degree is 0, is also reached the object of antagonism malicious user attack by the weighting of common recognition degree.

Claims (3)

1. a distributed collaborative frequency spectrum sensing method for attack resistance, is characterized in that, described method step following steps:

Step 1: in known noise variance primary user transmitted power P, the distance d between cognitive user i and primary user _i, the perception path fading factor-beta of cognitive user i _iwhen, each cognitive user foundation estimate self signal to noise ratio, and share signal-tonoise information with neighbor user, foundation calculate decision threshold v, self signal to noise ratio is compared with v, select K cognitive user and participate in collaborative spectrum sensing;

Step 2: energy measuring is carried out to K the cognitive user filtered out, obtains energy measuring statistical value X _iand local court verdict R _i, foundation calculate common recognition degree, and by formula Z _i(0)=w _i(l) Y _i(0) the initial value Z carrying out Gradient Iteration is obtained _i(0);

Step 3: participate in the cognitive user of cooperation according to oneself state Y _it state Y that () and normal neighbor node transmit _ij(t), according to more new state, iteration renewal process is continued until that cooperative cognitive User Status value converges to common value A ^*;

Step 4: to fixed system false alarm probability, set decision threshold λ according to " invariable false alerting method ", cognitive user is by A ^*compare with decision threshold, foundation $decision=\left\{\begin{array}{ccc}{H}_0& if& A^{*}<\mathrm{\&lambda;}\\ H_1& if& A^{*}\&GreaterEqual;\mathrm{\&lambda;}\end{array}\right.$ Decision rule make the conclusive judgement whether primary user exist;

Step 5: court verdict is broadcasted by cognitive user in cognition network;

Step 6: according to court verdict, cognitive user determines whether use primary user's frequency range to communicate at data transmission slots; If cognitive user judgement primary user exist, then keep mourning in silence at data transmission slots; Otherwise, then this frequency range can be utilized to carry out the transfer of data of self; Meanwhile, when cognitive user position changes or cognitive user occurs equipment fault, re-execute initial phase, i.e. resumes step 1, otherwise next round detects from step 2.

2. the distributed collaborative frequency spectrum sensing method of a kind of attack resistance according to claim 1, it is characterized in that: described method introduces the selection mechanism of cognitive user, the cognitive user choosing more excellent detection perform participates in cooperation, and with the common recognition degree of signal to noise ratio and these two parameter defined cognitive users of perception matching degree, utilize common recognition degree to be weighted the energy value that cognitive user iteration upgrades, the detection perform of different cognitive users and the fail safe of sensing results are distinguished.

3. the distributed collaborative frequency spectrum sensing method of a kind of attack resistance according to claim 1, is characterized in that, the definition of the cognitive user common recognition degree in described step 2, comprising:

The common recognition degree of described method should reflect the detection perform of cognitive user, the sensing results whether safety of cognitive user can be differentiated again, namely whether cognitive user is malicious user, its formation relates to two parameters, namely the signal to noise ratio of cognitive user and the local sensing results of cognitive user with cooperate after the matching degree of global decision, referred to as perception matching degree, and the definition of degree of common recognition is wanted meet along with signal to noise ratio increase and increase, along with perception matching degree increases and the rule of increase; Suppose that the perception matching degree of starting stage cognitive user is 1, then the common recognition degree of starting stage is determined by the signal to noise ratio of cognitive user, and signal to noise ratio larger then common recognition degree is larger; The change of the cognitive user detection perform of described method or under attackly can be embodied by common recognition degree, then utilized the primary iteration value of common recognition degree to the information interaction stage to be weighted, just can be distinguished different cognitive user.