CN101860878A - Method for defending frequency spectrum detection data falsification attack and in cognitive wireless network - Google Patents

Abstract

The invention discloses a method for defending frequency spectrum detection data falsification attack in a cognitive wireless network, which comprises the following steps: establishing a frequency spectrum detection model, acquiring local energy estimation level-values by detecting a target frequency band by a secondary user, establishing a wireless communication link and exchanging updating local energy estimation level-values by the secondary user and neighbor users to generate a neighbor user set, calculating local update by the secondary user and sending energy estimation level-values to the neighbor users for consistency processing/repeating the above steps so as to converge all energy estimation level-values in preset sharing values within an error range. The method for defending frequency spectrum detection data falsification attack in the cognitive wireless network ensures detection-loss rate and false-alarm rate are lower than the regular level, improves safety in cognitive wireless field and maximizes the use of wireless resources. The method has the advantages of stable and reliable working performance and wider application range.

Description

In the cognitive radio networks to the method for defending frequency spectrum detection data falsification attack

Technical field

The present invention relates to cognitive radio (CR) field, particularly cognitive radio networks safe practice field specifically is meant the method that realization is on the defensive to frequency spectrum detection data falsification attack in a kind of cognitive radio networks.

Background technology

The latest developments of radio communication constantly increase the demand of frequency spectrum resource.Management organization as Federal Communications Committee (FCC), is considering under glitch-free principle, opens (master) frequency range from permission to the operation of objectionable (inferior) frequency range to licensed-in user.It is that a kind of can making do not have licensed-in user to carry out method of operating at a certain spectral band that dynamic spectrum obtains (DSA), is lower than a certain thresholding as long as disturb.

A kind of method that realizes DSA is to adopt the thought of cognitive radio (CR), and its can perception surrounding environment and adjusts inherent state by the corresponding change to operating parameter.Because CRs is time user, so be the interference that to avoid near potential main user to the basic demand of CRs.In addition, can not require main user to change network foundation structure and CRs common spectrum.Therefore, the inferior user existence that is merely able to detect main user by continuous frequency spectrum detection whether.

The frequency spectrum detection of CRs can be individually or cooperation ground.Recently, as a more efficient methods, cooperation frequency spectrum detects and has obtained great concern.Compared to non-cooperative detection, cooperative detection has many advantages, specifically please refer to following document:

●Ghasemi?and?E.Sousa，“Collaborative?spectrum?sensing?for?opportunistic?access?in?fading?environments，”in?Proc.IEEE?DySPAN?2005，(Baltimore，Maryland)，Nov.2005.

●G.Ganesan?and?Y.G.Li，“Agility?improvement?through?cooperation?diversity?in?cognitive?radio，”in?Proc.IEEE?GLOBECOM’05，(St.Louis，MO)，Nov?2005.

●S.Mishra，A.Sahai，and?R.Brodersen，“Cooperative?sensing?among?cognitive?radios，”in?Proc.IEEE?ICC’06，(Istanbul，Turkey)，June?2006.

●E.Peh?and?Y.-C.Liang，“Optimization?for?cooperative?sensing?in?cognitive?radio?networks，”inProc.IEEE?WCNC’07，(Hong?Kong，P.R.China)，Mar.2007.

●J.Unnikrishnan?and?V.V.Veeravalli，“Cooperative?sensing?for?primary?detection?in?cognitive?radio，”IEEE?J.Sel.Topics?Signal?Proc.，vol.2，pp.18-27，Feb.2008.

●Z.Quan，S.Cui，and?A.H.Sayed，“Optimal?linear?cooperation?for?spectrum?sensing?incognitive?radio?networks，”IEEE?J.Sel.Topics?Signal?Proc.，vol.2，pp.28-40，Feb.2008.

For example, if inferior user is among the deep fading, it is very difficult distinguishing one " white space " from a deep fading's effect.But cooperation method can address this problem by sharing frequency spectrum detection information with inferior user.And, being in the main subscriber signal that seriously covers in order to listen to, CR must possess the sensitivity of height.But this can cause CR terminal cost to increase, thus any wide-area deployment of restriction CR network.By cooperation, the CR terminal of muting sensitivity also can obtain reliable frequency spectrum detection cheaply.

Although done a few thing in the CR network, to compare with other field, the security fields of cognitive radio (CR) network are far from attracting much attention.Certainly, on the whole, remain the point of interest of CR example for the threat of non-cognitive radio networks.Yet some distinguishing features that dynamic spectrum obtains (DSA) have been incorporated into the risk on the safety in the CR network.For example, the local frequency spectrum detection information of collecting and exchanging is used to set up the perception environment that a meeting influences the CR user behavior.This provides chance for malicious attack.In below with reference to document:

R.Chen，J.-M.Park，Y.Hou，and?J.Reed，“Toward?secure?distributed?spectrum?sensing?incognitive?radio?networks，”IEEE?Comm.Mag.，vol.46，pp.50-55，Apr.2008.

The author has described two kinds of threats to the CR network: incumbent emulation (IE) and frequency spectrum detection data falsification (SSDF).In an IE attacked, the invador imitated current main user's characteristics of signals, to cheat other times user.IE attacks and understands serious interference spectrum testing process and reduce time retrievable channel resource of user significantly.Being used for discerning IE below with reference to the method that has proposed a transmitter checking in the document attacks:

R.Chen，J.-M.Park，and?J.Reed，“Defense?against?primary?user?emulation?attacks?in?cognitive?radio?networks，”IEEE?J.Sel.Areas?Commun.，vol.26，pp.25-37，Jan.2008.

In SSDF attacked, the invador sent wrong local testing result, caused the wrong frequency spectrum detection judgement of CR with this.List of references (R.Chen, J.-M.Park, Y.Hou, and J.Reed, " Toward secure distributed spectrum sensing in cognitive radio networks, " IEEE Comm.Mag., vol.46, pp.50-55, author Apr.2008.) has done good trial by the method for advising several calculating SSDF number of times of attack.Yet, do not have the further report of development.

Summary of the invention

The objective of the invention is to have overcome above-mentioned shortcoming of the prior art, provide a kind of can effectively defend in the cognitive radio malicious attack, improve the cognition wireless electrical domain safely, make that cognitive user can be in interfere with primary users not obtain more frequency spectrum, maximized Radio Resource, stable and reliable working performance, the scope of application method that realization is on the defensive to frequency spectrum detection data falsification attack in the cognitive radio networks comparatively widely of utilizing in advance down.

In order to realize above-mentioned purpose, realize in the cognitive radio networks of the present invention that the method that frequency spectrum detection data falsification attack is on the defensive is as follows:

The method that realization is on the defensive to frequency spectrum detection data falsification attack in this cognitive radio networks, its main feature is that described method may further comprise the steps:

(1) sets up the frequency spectrum detection model of cognitive radio networks;

(2) all inferior users in the cognitive radio networks detect target band based on this frequency spectrum detection model, and obtain local Energy Estimation level value;

(3) all inferior user sets up radio communication with its adjacent user and is connected, the local Energy Estimation level value that upgrades according to default time interval exchange, and produce adjacent user's collection;

(4) all inferior user Energy Estimation level value after carrying out will upgrading after local update calculates is issued each adjacent user that adjacent user concentrates and is carried out the consistency processing;

(5) repeat above-mentioned steps (3) and (4), all converge on common value in the default error range up to all Energy Estimation level values.

The frequency spectrum detection model of setting up cognitive radio networks in the method that realizes in this cognitive radio networks frequency spectrum detection data falsification attack is on the defensive is specially:

Set up the frequency spectrum detection model of cognitive radio networks according to following formula:

$x\left(t\right)=\left\{\begin{array}{cc}n\left(t\right),& H_0\\ h\·s\left(t\right)+n\left(t\right),& H_1\end{array}\right.;$

$Y=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{2\mathrm{TW}}^2\left(2\mathrm{\γ}\right),& H_1\end{array}\right.;$

Wherein, x (t) is time received signal of user, and s (t) is main user's a transmission signal, and n (t) is the white Gaussian noise that adds, and h is the amplitude gain of channel; T is the time, and W is a bandwidth,

Be center χ ²Distribute, the degree of freedom is 2TW,

Be acentric χ ²Distribute, the degree of freedom is 2TW, and 2 γ are this acentric χ ²Center of distribution.

Obtain local Energy Estimation level value in the method that realizes in this cognitive radio networks frequency spectrum detection data falsification attack is on the defensive, be specially:

Obtain the local Energy Estimation level value Y of each time user i according to following formula _i:

$Y_i=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{(2\mathrm{TW}-2)}^2+e_{(2\stackrel{\‾}{\mathrm{\γ}}+2)},& H_1\end{array}\right.;$

Wherein, γ is an exponential distribution,

Be average signal-to-noise ratio,

Be the stochastic variable of exponential distribution, parameter is

Be acentric χ ²Distribute, the degree of freedom is 2TW-2.

The local Energy Estimation level value that the time interval exchange according to default in the method that realizes in this cognitive radio networks frequency spectrum detection data falsification attack is on the defensive is upgraded also produces adjacent user's collection, may further comprise the steps:

(11) described user i obtains local reception signal average μ according to following formula when time interval k-1 _i(k-1):

${\mathrm{\μ}}_i(k-1)=\frac{x_i(k-1)+\underset{j\∈N_i}{\mathrm{\Σ}}{x}_j(k-1)}{1+\left|N_i\right|};$

Wherein, N _iBe adjacent user's set of inferior user i, j is the adjacent user of i, x _j(k-1) received signal that is exchanged for adjacent user j,

N is a time total number of users,

| N _i| be N _iThe user element number;

(12) described user i obtains the maximum of the received signal that exchanged in adjacent user's set according to following formula, thereby identifies the adjacent user who carries out frequency spectrum detection data falsification attack

$\hat{j}=\arg \underset{j\∈N_i}{\max }|x_j\left(k\right)-{\mathrm{\μ}}_i(k-1)|;$

(13) described user i according to following formula with adjacent user

Gather N from adjacent user _iMiddle deletion, thus formation is familiar with and the adjacent user of process checking collects

${\hat{N}}_i\left(k\right)=N_i\backslash \left\{\hat{j}\right\}.$

Energy Estimation level value after will upgrade in the method that realizes in this cognitive radio networks frequency spectrum detection data falsification attack is on the defensive is issued each adjacent user that adjacent user concentrates and is carried out the consistency processing, may further comprise the steps:

(21) the Energy Estimation level value Y after described user i will upgrade _iWith received signal x _i(k) issue adjacent user's collection

In each adjacent user;

(22) described user i is according to following formula x to received signal _i(k) carry out interative computation:

$x_i(k+1)=x_i\left(k\right)+\mathrm{\ϵ}\underset{j\∈{\hat{N}}_i\left(k\right)}{\mathrm{\Σ}}(x_j\left(k\right)-x_i\left(k\right));$

Wherein,

Δ is the maximal degree of network, k=0, and 1,2 ..., x _j(k) be the received signal that adjacent user j sends.

Adopted and realized method that frequency spectrum detection data falsification attack is on the defensive in the cognitive radio networks of this invention, carry out data fusion for conclusive judgement and come SSDF is attacked counting owing to wherein need not public receiver, cognitive user wherein only need be set up local interaction, and the information exchange that need not concentrate, in cognitive process, adopted simultaneously consistency algorithm, by adopting consistency algorithm to come the degree of writing that the local frequency spectrum detection that each sense terminals is received is reported is carried out difference, thereby can effectively get rid of the malicious attack object, do not need centralized exchange message mode, implementation procedure is simple and convenient, and can improve the identification that SSDF is attacked significantly and resist, and guaranteed that rate of lapsing and false alarm rate are lower than conventional levels, improved the safety of cognition wireless electrical domain, what make that cognitive user can be in interfere with primary users not obtains more frequency spectrum in advance down, the maximized Radio Resource that utilized, stable and reliable working performance, moreover, method of the present invention can also be calculated SSDF number of times of attack in the CR network, the algorithm with biological characteristics that is wherein adopted provides reference for the CR Network Design in future, and the scope of application is comparatively extensive.

Description of drawings

Fig. 1 is energy testing apparatus high-level schematic functional block diagram in the method that realizes in the cognitive radio networks of the present invention frequency spectrum detection data falsification attack is on the defensive.

Embodiment

In order more to be expressly understood technology contents of the present invention, describe in detail especially exemplified by following examples.

See also shown in Figure 1ly, realize method that frequency spectrum detection data falsification attack is on the defensive in this cognitive radio networks, comprising following steps:

(1) set up the frequency spectrum detection model of cognitive radio networks, be specially:

Set up the frequency spectrum detection model of cognitive radio networks according to following formula:

$x\left(t\right)=\left\{\begin{array}{cc}n\left(t\right),& H_0\\ h\·s\left(t\right)+n\left(t\right),& H_1\end{array};\right.$

$Y=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{2\mathrm{TW}}^2\left(2\mathrm{\γ}\right),& H_1\end{array}\right.;$

Wherein, x (t) is time received signal of user, and s (t) is main user's a transmission signal, and n (t) is the white Gaussian noise that adds, and h is the amplitude gain of channel; T is the time, and W is a bandwidth,

Be center χ ²Distribute, the degree of freedom is 2TW,

Be acentric χ ²Distribute, the degree of freedom is 2TW, and 2 γ are this acentric χ ²Center of distribution;

(2) all inferior users in the cognitive radio networks detect target band based on this frequency spectrum detection model, and obtain local Energy Estimation level value, are specially:

Obtain the local Energy Estimation level value Y of each time user i according to following formula _i:

$Y_i=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{(2\mathrm{TW}-2)}^2+e_{(2\stackrel{\‾}{\mathrm{\γ}}+2)},& H_1\end{array}\right.;$

Wherein, γ is an exponential distribution,

Be average signal-to-noise ratio,

Be the stochastic variable of exponential distribution, parameter is

Be acentric χ ²Distribute, the degree of freedom is 2TW-2;

(3) all inferior user sets up radio communication with its adjacent user and is connected, the local Energy Estimation level value that upgrades according to default time interval exchange, and produce adjacent user's collection, may further comprise the steps:

(a) described user i obtains local reception signal average μ according to following formula when time interval k-1 _i(k-1):

${\mathrm{\μ}}_i(k-1)=\frac{x_i(k-1)+\underset{j\∈N_i}{\mathrm{\Σ}}{x}_j(k-1)}{1+\left|N_i\right|};$

Wherein, N _iBe adjacent user's set of inferior user i, j is the adjacent user of i, x _j(k-1) received signal that is exchanged for adjacent user j,

N is a time total number of users,

| N _i| be N _iThe user element number;

(b) described user i obtains the maximum of the received signal that exchanged in adjacent user's set according to following formula, thereby identifies the adjacent user who carries out frequency spectrum detection data falsification attack

$\hat{j}=\arg \underset{j\∈N_i}{\max }|x_j\left(k\right)-{\mathrm{\μ}}_i(k-1)|;$

(c) described user i according to following formula with adjacent user

Gather N from adjacent user _iMiddle deletion, thus formation is familiar with and the adjacent user of process checking collects

${\hat{N}}_i\left(k\right)=N_i\backslash \left\{\hat{j}\right\};$

(4) all inferior user Energy Estimation level value after carrying out will upgrading after local update calculates is issued each adjacent user that adjacent user concentrates and is carried out the consistency processing, may further comprise the steps:

(a) the Energy Estimation level value Y after described user i will upgrade _iWith received signal x _i(k) issue adjacent user's collection

In each adjacent user;

(b) described user i is according to following formula x to received signal _i(k) carry out interative computation:

$x_i(k+1)=x_i\left(k\right)+\mathrm{\ϵ}\underset{j\∈{\hat{N}}_i\left(k\right)}{\mathrm{\Σ}}(x_j\left(k\right)-x_i\left(k\right));$

Wherein,

Δ is the maximal degree of network, k=0, and 1,2 ..., x _j(k) be the received signal that adjacent user j sends;

(5) repeat above-mentioned steps (3) and (4), all converge on common value in the default error range up to all Energy Estimation level values.

In the middle of reality is used, method of the present invention be based on consistency algorithm latest developments below with reference to document:

W.Ren，R.Beard，and?E.Atkins，“A?survey?of?consensus?problems?in?multi-agentcoordination，”in?Proc.American?Control?Conference’05，(Portland，OR)，June?2005.

An important trigger point of method of the present invention is the research bevy, the natural phenomena that a group fish and a army of bees etc. are complicated.Recently, consistency problem is also in distributed controlling models, and radio sensing network and random noise are played an important role in measuring.Consider safe frequency spectrum detection model, the basic demand of cognitive user is that filtering SSDF attacks the data falsification that adds and judges correctly whether main user is current exists that this situation can be regarded as typical many co-operative environments.Its advantage is as follows:

(1) this is a complete distributed upgradeable method.Different with existing scheme, method of the present invention does not need a public receiver to carry out data fusion for conclusive judgement to come SSDF is attacked counting.Because be difficult to a centralized node in the number of C R network, local interaction only need be set up in the cognitive family in the method for the present invention, and the information exchange that need not concentrate.

(2) the most of decision rule that adopted unlike existing method, as OR rule and n/N rule, method of the present invention has adopted conforming method in cognition.By adopting conforming method to come the degree of writing that the local frequency spectrum detection that each sense terminals is received is reported is carried out difference, this method can be resisted SSDF well and attack.

(3) because the CR example has unforgettable human characteristics (as study, adapting to and cooperation) in wireless network, the algorithm with biological characteristics that adopts in the method for the present invention provides reference for the CR Network Design in future.

Frequency spectrum detection and SSDF attack model be will at first introduce below, employed network model of method of the present invention and conforming specific implementation then introduced.

(1) frequency spectrum detection model

In the CR network, inferior user checks by measuring whether main user exists.The frequency spectrum detecting method that has 3 kinds of methods to be widely used sees also following document:

D.Cabric，S.Mishra，and?R.Brodersen，“Implementation?issues?in?spectrum?sensing?for?cognitive?radios，”in?Proc.Thirty-Eighth?Asilomar?Conference?on?Signals，Systems?and?Computers，vol.1，pp.772-776，2004.

Matched filtering method is best selection in theory, but it needs the priori of main system, this means that the adaptive detection circuit of exploitation different radio main system becomes more complicated, and it is higher that cost also becomes.Energy measuring is not a best choice, but it is easy to install and main user's position is not had too many requirement.Cyclostationary characteristic detects can detect the very low signal of SNR, but it also needs some prioris of main user, sees also following document:

C.Sun，W.Zhang，and?K.B.Letaief，“Cluster-based?cooperative?spectrum?sensing?incognitive?radio?systems，”in?Proc.IEEE?ICC’07，(Glasgow，UK)，pp.2511-2515，June?2007.

The model of method supposition of the present invention is the priori of not knowing main user.For the ease of installing, wherein used the frequency spectrum detecting method of a measuring energy, specifically see also following document:

A.Ghasemi?and?E.Sousa，“Collaborative?spectrum?sensing?for?opportunistic?access?in?fading?environments，”in?Proc.IEEE?DySPAN?2005，(Baltimore，Maryland)，Nov.2005.

Wherein Fig. 1 has shown the block diagram of energy testing apparatus.The centre frequency of input tape bandpass filter is f _s, bandwidth is W.What be contained in its back is a squarer, and it is used for measuring the energy that receives.Another one is that integrator decides observation to be spaced apart T.At last, the output Y of integrator compares with thresholding λ, judges with this whether main subscriber signal exists.The purpose of frequency spectrum detection is to judge two following supposition,

$x\left(t\right)=\left\{\begin{array}{cc}n\left(t\right),& H_0\\ h\·s\left(t\right)+n\left(t\right),& H_1\end{array}\right.---\left(1\right)$

X (t) is the signal that time user receives, and s (t) is main user's a transmission signal, and n (t) is the white Gaussian noise (AWGN) that adds, and h is the amplitude gain of channel.We also represent signal to noise ratio (snr) with γ.The output of integrator is Y among Fig. 1, and it is the decision statistic amount.According to list of references (H.Urkowitz, " Energy detection of unknown eterministic signals, " Proc.IEEE, vol.55, pp.523-531, the Apr.1967.) method in, Y has following distribution,

$Y=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{2\mathrm{TW}}^2\left(2\mathrm{\γ}\right),& H_1\end{array}\right.---\left(2\right)$

With

Represent central distribution and acentric χ respectively ²Distribute, their degree of freedom all is 2TW, acentric χ ²Center of distribution is 2 γ.For simply, the present invention supposes that the product TW of time and bandwidth is an integer, represents with m.

Under the condition of Rayleigh fading, γ is an exponential distribution, so the present invention has used average signal-to-noise ratio in this case

In addition, h is uncertain.Therefore, according to below with reference to document:

V.Kostylev，“Energy?detection?of?a?signal?with?random?amplitude，”in?IEEE?Proc.ICC’02，(New?York，NY)，Apr.2002.

$Y=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{(2\mathrm{TW}-2)}^2+e_{(2\stackrel{\‾}{\mathrm{\γ}}+2)},& H_1\end{array}\right.---\left(3\right)$

Be the stochastic variable of an exponential distribution, its parameter is

Be an acentric χ ²Distribute, its degree of freedom is (2TW-2).

In a word, after the T, each time user i detects energy and obtains the Energy Estimation level

(2) the SSDF attack model during cooperation frequency spectrum detects

In cooperation frequency spectrum detected, one group time the user did frequency spectrum detection by the local information of obtaining of mutual exchange.An assailant sends wrong local frequency spectrum detecting result causes the frequency spectrum detection of a mistake to other time user judgement.As first line of defence,,, can be used for protecting the cooperation frequency spectrum of CR network to detect as checking based on the method for prevention.Yet, always have weakness in the complete experience display system of conventional wireless network case and be difficult to prediction.Particularly the inferior user of malice may utilize cooperation frequency spectrum to detect to start SSDF to attack.Be 3 kinds of attack model below.

● the inferior user of first kind of attack model a---malice sends high relatively main user's energy, to show existing of main user.Although do not exist main user and detected energy very low.In this case, other time user can do the judgement that makes mistake, and thinks main user to exist and do not remove to use frequency spectrum.Malice time user's purpose is to monopolize target spectrum.We are called selfish SSDF to this attack and attack.

● the inferior user of second kind of attack model---malice sends low relatively main user's energy, does not have main user to show.Although it is very high having main user and its energy.In this case, other time user makes false judgment and thinks and do not have main user and use frequency spectrum.The inferior user's of malice purpose is an interfere with primary users.We are called interference SSDF to this attack and attack.

● the inferior user of the third attack model---malice launches main user's energy at random in the cooperation frequency spectrum testing process.That is to say that it sometimes sends correct main user's energy, send wrong main user's energy sometimes.Its objective is other time of fascination user, make them can't reach consistent.We are called fascination SSDF to this attack and attack.

● attack in order to alleviate these SSDF, we adopt state-of-the-art consistency algorithm in the present cooperation frequency spectrum detection.

(3) network model and consistency thought

In based on conforming frequency spectrum detection, inferior user and adjacent inferior user establish a communications link, to carry out the local information exchange.The network that inferior users set up can describe with the icon model of a standard.Briefly, it can with comprise a group node i=1,2 ..., n} and one group of line Icon G=(N, ε) (please refer to document: [20] M.Huang and J.H.Manton, " Stochastic lyapunov analysis for consensus algorithms with noisy measurements; " in Proc.American Control Conference ' 07, (New York, NY), July 2007.) represent.Each line be a unordered pair (i, j).Therefore, if two users connect in order to line, mean their exchange messages mutually.The paths of G figure comprises a string node i ₁, i ₂..., i _l, to all 1≤m≤l-1, there is (i l 〉=2 _m, i _M+1) ∈ ε.If two nodes among any G figure are interconnective, then G figure is in connection.For convenience of explanation, the node i that refers to of the present invention is exactly time user i.If (then time user j is the neighboring user of time user i for j, i) ∈ ε, and j ≠ i.The adjacent user of node i uses

Represent.N _iUser element use | N _i| expression.

Laplacian among the G figure is defined as: L=(l _Ij) _{N * n},

Wherein L is a positive semidefinite matrix.Furtherly, if G figure is a non-direct connection figure, then the order of G is n-1.Specifically see also below with reference to document:

R.Olfati-Saber，J.Fax，and?R.Murray，“Consensus?and?cooperation?in?networked?multi-agent?systems，”Proc.IEEE，vol.95，pp.215-233，Jan.2007.

Because cooperation frequency spectrum detection problem is regarded as a consistency problem before the user reaches an agreement about this locality exchange of their testing results, the present invention has done this as giving a definition:

N user distributes shown in illustraton of model G, and the present invention has distributed state scalar x for them _i, (primary power is estimated horizontal Y to i ∈ N _iAs previously mentioned).Each x _iBe considered to a consistency variable, it is used in cooperation frequency spectrum detects to the estimation of node i energy measuring.When reaching unanimity, we think each x _iState reach a common value x gradually ^*, for example:

x _i(k) → x ^*As k → ∞ ... (5)

To each i ∈ N, k is the discrete time interval, k=0, and 1,2 ... x _i(k) be based on that node i and adjacent User Status thereof upgrade.

Special case

With

Be called average homogeneity respectively, maximum consistency and minimum consistency.

(4) based on conforming frequency spectrum detection algorithm

The inventive method has proposed to come SSDF attacked based on conforming frequency spectrum detection scheme to count.Wherein hypothesis time user has set up wireless connections with its neighboring user.Exist up to reaching consistent, connecting all the time.The present invention is called definite figure to this topology.On the basis of this assumption, algorithm comprises two steps:

A. the first step---all inferior users detect target band on the frequency spectrum detection model based, and obtain local Energy Estimation level, use Y _iExpression.

B. second going on foot---all inferior users set up radio communication with its adjacent user and are connected, from the time interval

Begin to exchange the local Energy Estimation level of renewal.This process is carried out repeatedly.Here the user i of time interval k=0 and measurement Y _iBe expressed as Each time user's consistency variable be updated in one-period sampling interval k=0,1,2 ... take place.At each time interval k, in case receive the horizontal x of Energy Estimation from neighboring user _j(k), each time user similarly is assailant's adjacent user according to selected standard exclusion at first.This process will produce adjacent user's collection, and it is used to upgrade the state of user i.

C. after local update calculated, each user issued the Energy Estimation level value x after adjacent user upgrades _i(k+1).Adjacent user then selects and the state renewal process will repeat, up to the horizontal x of all Energy Estimation _i(k) converge on the interior common value x of regulation error range ^*

At last, and compare with the thresholding λ of definition among Fig. 1, inferior user i obtains final data and merges:

Next, at length explain screening rule.The present invention supposes k 〉=1 He | N _i|＞2.

(1) at first, inferior user i obtains local average when time interval k-1:

${\mathrm{\μ}}_i(k-1)=\frac{x_i(k-1)+\underset{j\∈N_i}{\mathrm{\Σ}}{x}_j(k-1)}{1+\left|N_i\right|}---\left(7\right)$

(2) then, user i identifies adjacent user by the maximum in the equation:

$\hat{j}=\arg \underset{j\∈N_i}{\max }|x_j\left(k\right)-{\mathrm{\μ}}_i(k-1)|---\left(8\right)$

(3) last, user i forms a quilt and is familiar with the adjacent user's collection through verifying

${\hat{N}}_i\left(k\right)=N_i\backslash \left\{\hat{j}\right\}---\left(9\right)$

When k=0 or | N _i|≤2, screening equation ψ directly specifies ψ _i(k)=N _iReason as for k=0 is that desired value is clear and definite.Negative value appears in the k-1 in equation ψ, definition k 〉=1.

Next, the inventive method will be illustrated in the consistency algorithm in the adjacent user's selection of cooperation.From k=0,1,2 ... the iteration form of consistency algorithm is as follows:

$x_i(k+1)=x_i\left(k\right)+\mathrm{\ϵ}\underset{j\∈{\hat{N}}_i\left(k\right)}{\mathrm{\Σ}}(x_j\left(k\right)-x_i\left(k\right))---\left(10\right)$

Here:

$0<\mathrm{\&epsiv;}<{\underset{i}{\left(\max \right|N_i\left|\right)}}^{-1}\stackrel{\mathrm{\&Delta;}}{=}{\mathrm{\&Delta;}}^{-1}---\left(11\right)$

Δ is called the maximal degree of network.

If use fundamental equation (N in the above-mentioned equation (10) _i(k) always represent N _iSpecifically see also document: R.Olfati-Saber, J.Fax, and R.Murray, " Consensus and cooperation in networked multi-agent systems, " Proc.IEEE.vol.95, pp.215-233, Jan.2007.), average homogeneity one is obtained surely so, and final common value

Will be the mean value of initial vector x (0) or after pre-energy measuring screening finishes, can obtain Y ^T={ Y ₁, Y ₂..., Y _nAverage.Furtherly, this will carry out in compromise in performance and secure context.Proposed by the invention has only used a common random matrix based on conforming algorithm, the adding and be 1 of its row, rather than for the average input value of obtaining k=0 doubly stochastic matrix is averaged.The consistency of this application no longer is an average, is not two-way because connect in the adjacent user's communications of each time interval k.This moment communicate to connect the time dynamic.

In the CR network, safety is unusual the important point.In cooperation frequency spectrum detected, malice CRs can send wrong local frequency spectrum detecting result.In this method, we have showed in the CR network, and scheme that detects based on the consistency cooperation frequency spectrum is come SSDF attacked and calculated.By inferior user's consistency, the scheme of proposition can be distinguished credible frequency spectrum detection terminal, and this makes that it can be very reliable before SSDF, and no longer needs a public receiver in final judgement.Simulation result has shown the validity that proposes a plan.Studying simultaneously and carrying out, such as the aspects such as checking in the security framework of the CR network that proposes based on the future network of the scheme of prevention.

About help to understand and implement technical scheme of the present invention can reference technical literature also comprise following:

●J.Mitola，Cognitive?radio：An?integrated?agent?architecture?for?software?defined?radio.Doctor?of?Technology，Royal?Inst.Technol.(KTH)，Stockholm，Sweden，2000.

●S.Haykin，“Cognitive?radio：Brain-empowered?wireless?communications，”IEEE?J.Sel.Areas?Commun.，vol.23，pp.201-220，Feb.2005.

●R.Olfati-Saber，J.Fax，and?R.Murray，“Consensus?and?cooperation?in?networked?multi-agentsystems，”Proc.IEEE，vol.95，pp.215-233，Jan.2007.

●L.Xiao，S.Boyd，and?S.Lall，“A?scheme?for?robust?distributed?sensor?fusion?based?onaverage?consensus，”in?Proc.Fourth?International?Symposium?on?Information?Processing?inSensor?Networks?IPSN?2005，pp.63-70，2005.

●M.Huang?and?J.Manton，“Stochastic?consensus?seeking?with?measurement?noise：convergence?and?asymptotic?normality，”in?Proc.American?Control?Conference’08，(Seattle，WA)，June?2008.

●J.L.Burbank，“Security?in?cognitive?radio?networks：the?required?evolution?in?approaches?towireless?network?security，”in?Proc.IEEE?CrownCom’08，(Singapore)，pp.1-7，May?2008.

Adopted the method that realization is on the defensive to frequency spectrum detection data falsification attack in the above-mentioned cognitive radio networks, carry out data fusion for conclusive judgement and come SSDF is attacked counting owing to wherein need not public receiver, cognitive user wherein only need be set up local interaction, and the information exchange that need not concentrate, in cognitive process, adopted simultaneously consistency algorithm, by adopting consistency algorithm to come the degree of writing that the local frequency spectrum detection that each sense terminals is received is reported is carried out difference, thereby can effectively get rid of the malicious attack object, do not need centralized exchange message mode, implementation procedure is simple and convenient, and can improve the identification that SSDF is attacked significantly and resist, and guaranteed that rate of lapsing and false alarm rate are lower than conventional levels, improved the safety of cognition wireless electrical domain, what make that cognitive user can be in interfere with primary users not obtains more frequency spectrum in advance down, the maximized Radio Resource that utilized, stable and reliable working performance, moreover, method of the present invention can also be calculated SSDF number of times of attack in the CR network, the algorithm with biological characteristics that is wherein adopted provides reference for the CR Network Design in future, and the scope of application is comparatively extensive.

In this specification, the present invention is described with reference to its certain embodiments.But, still can make various modifications and conversion obviously and not deviate from the spirit and scope of the present invention.Therefore, specification and accompanying drawing are regarded in an illustrative, rather than a restrictive.

Claims (5)

1. realize in the cognitive radio networks method that frequency spectrum detection data falsification attack is on the defensive it is characterized in that described method may further comprise the steps:

(1) sets up the frequency spectrum detection model of cognitive radio networks;

(2) all inferior users in the cognitive radio networks detect target band based on this frequency spectrum detection model, and obtain local Energy Estimation level value;

(3) all inferior user sets up radio communication with its adjacent user and is connected, the local Energy Estimation level value that upgrades according to default time interval exchange, and produce adjacent user's collection;

(4) all inferior user Energy Estimation level value after carrying out will upgrading after local update calculates is issued each adjacent user that adjacent user concentrates and is carried out the consistency processing;

(5) repeat above-mentioned steps (3) and (4), all converge on common value in the default error range up to all Energy Estimation level values.

2. realization is characterized in that to the method that frequency spectrum detection data falsification attack is on the defensive the described frequency spectrum detection model of setting up cognitive radio networks is specially in the cognitive radio networks according to claim 1:

Set up the frequency spectrum detection model of cognitive radio networks according to following formula:

$x\left(t\right)=\left\{\begin{array}{cc}n\left(t\right),& H_0\\ h\&CenterDot;s\left(t\right)+n\left(t\right),& H_1\end{array}\right.;$

$Y=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{2\mathrm{TW}}^2\left(2\mathrm{\&gamma;}\right),& H_1\end{array}\right.;$

Wherein, x (t) is time received signal of user, and s (t) is main user's a transmission signal, and n (t) is the white Gaussian noise that adds, and h is the amplitude gain of channel; T is the time, and W is a bandwidth,

Be center χ ²Distribute, the degree of freedom is 2TW, Be acentric χ ²Distribute, the degree of freedom is 2TW, and 2 γ are this acentric χ ²Center of distribution.

3. realize in the cognitive radio networks according to claim 2 it is characterized in that method that frequency spectrum detection data falsification attack is on the defensive, describedly obtain local Energy Estimation level value, be specially:

Obtain the local Energy Estimation level value Y of each time user i according to following formula _i:

$Y_i=\left\{\begin{array}{cc}{X}_{2\mathrm{TW}}^2,& H_0\\ X_{(2\mathrm{TW}-2)}^2+e_{(2\stackrel{\&OverBar;}{\mathrm{\&gamma;}}+2)},& H_1\end{array}\right.;$

Wherein, γ is an exponential distribution,

Be average signal-to-noise ratio,

Be the stochastic variable of exponential distribution, parameter is

Be acentric χ ²Distribute, the degree of freedom is 2TW-2.

4. the method that realization is on the defensive to frequency spectrum detection data falsification attack in the cognitive radio networks according to claim 3, it is characterized in that, the described local Energy Estimation level value that upgrades according to default time interval exchange also produces adjacent user's collection, may further comprise the steps:

(11) described user i obtains local reception signal average μ according to following formula when time interval k-1 _i(k-1):

${\mathrm{\&mu;}}_i(k-1)=\frac{x_i(k-1)+\underset{j\&Element;N_i}{\mathrm{\&Sigma;}}{x}_j(k-1)}{1+\left|N_i\right|};$

Wherein, N _iBe adjacent user's set of inferior user i, j is the adjacent user of i, x _j(k-1) received signal that is exchanged for adjacent user j,

N is a time total number of users, | N _i| be N _iThe user element number;

(12) described user i obtains the maximum of the received signal that exchanged in adjacent user's set according to following formula, thereby identifies the adjacent user who carries out frequency spectrum detection data falsification attack

$\hat{j}=\arg \underset{j\&Element;N_i}{\max }|x_j\left(k\right)-{\mathrm{\&mu;}}_i(k-1)|;$

(13) described user i according to following formula with adjacent user Gather N from adjacent user _iMiddle deletion, thus formation is familiar with and the adjacent user of process checking collects

${\hat{N}}_i\left(k\right)=N_i\backslash \left\{\hat{j}\right\}.$

5. the method that realization is on the defensive to frequency spectrum detection data falsification attack in the cognitive radio networks according to claim 4, it is characterized in that, Energy Estimation level value after described will the renewal is issued each adjacent user that adjacent user concentrates and is carried out consistency and handle, and may further comprise the steps:

(21) the Energy Estimation level value Y after described user i will upgrade _iWith received signal x _i(k) issue adjacent user's collection

In each adjacent user;

(22) described user i is according to following formula x to received signal _i(k) carry out interative computation:

$x_i(k+1)=x_i\left(k\right)+\mathrm{\&epsiv;}\underset{j\&Element;{\hat{N}}_i\left(k\right)}{\mathrm{\&Sigma;}}(x_j\left(k\right)-x_i\left(k\right));$

Wherein,

Δ is the maximal degree of network, k=0, and 1,2 ..., x _j(k) be the received signal that adjacent user j sends.

Images