a double adaptive approach to tackle malicious users in...

Research ArticleA Double Adaptive Approach to Tackle Malicious Users inCognitive Radio Networks

Muhammad Sajjad Khan 1 Muhammad Jibran1 Insoo Koo 2

Su Min Kim3 and Junsu Kim 3

1Department of Electrical Engineering International Islamic University Islamabad Pakistan2School of Electrical Engineering University of Ulsan 93 Daehak-ro Nam-gu Ulsan 680-749 Republic of Korea3Department of Electronic Engineering Korea Polytechnic University 237 Sangidaehak-ro Siheung-siGyeonggi-do 429-793 Republic of Korea

Correspondence should be addressed to Junsu Kim junsukimkpuackr

Received 5 September 2018 Revised 1 February 2019 Accepted 14 March 2019 Published 27 March 2019

Guest Editor Mohamad K A Rahim

Copyright copy 2019 Muhammad Sajjad Khan et alThis is an open access article distributed under the Creative CommonsAttributionLicense which permits unrestricted use distribution and reproduction in anymedium provided the originalwork is properly cited

Cognitive radio (CR) is being considered as a vital technology to provide solution to spectrum scarcity in next generation networkby efficiently utilizing the vacant spectrum of the licensed users Cooperative spectrum sensing in cognitive radio network has apromising performance compared to the individual sensing However the existence of the malicious usersrsquo attack highly degradesthe performance of the cognitive radio networks by sending falsified data also known as spectrum sensing data falsification (SSDF)to the fusion center In this paper we propose a double adaptive thresholding technique in order to differentiate legitimate usersfrom doubtful and malicious users Prior to the double adaptive approach the maximal ratio combining (MRC) scheme is utilizedto assign weight to each user such that the legitimate users experience higher weights than the malicious users Double adaptivethreshold is applied to give a fair chance to the doubtful users to ensure their credibility A doubtful user that fails the doubleadaptive threshold test is declared as a malicious user The results of the legitimate users are combined at the fusion center byutilizing Dempster-Shafer (DS) evidence theory Effectiveness of the proposed scheme is proved through simulations by comparingwith the existing schemes

1 Introduction

Wireless network technologies are the most promising tech-nologies in the twentieth century Today we already haveover a dozen wireless technologies in use Wi Fi BluetoothZig Bee NFC LTE earlier 3G standards satellite servicesetc Due to the proliferation of these wireless networks andthe increase in the number of users the spectrum scarcityproblem is raised On the other hand various reports haveshown that the spectrum is inefficiently utilized such thatthe spectrum is underutilized at a fixed frequency and ata random geographical area [1] Federal CommunicationCommission (FCC) states that temporal and geographicalvariations in the utilization of the assigned spectrum varyfrom 15 to 85 [2] One promising solution to this problemis proposed by Joseph Mitola ie ldquoCognitive Radio (CR)rdquo[3 4]

CR is a vital technology to improve spectrum utilizationA major challenge in CR is spectrum sensing that identifiesthe presence of the licensedprimary user (LU) in the networkand whenever the LU is detected secondary user (SU) needsto vacate the channel [5]

Sensing reliability of a single SU degraded by fading andhidden terminal problems This problem is overcome by theuse of cooperative spectrum sensing (CSS) which involvesexchange of local sensing decision betweenmultiple SUs SUssend the sensing result to fusion center by utilizing either thehard decision or soft decision rules

There are two types of CSS one is centralized CSS andthe other one is distributed CSS In the centralized CSS allSUs sense the environment and send their information aboutthe presence of LU to the data fusion center (DFC) and theDFC gives the final decision about the presence of LU In thedistributed CSS in which there is no central node every SU

HindawiWireless Communications and Mobile ComputingVolume 2019 Article ID 2350694 9 pageshttpsdoiorg10115520192350694

2 Wireless Communications and Mobile Computing

senses the radio environment and different SUs share theirinformation and make their own decision with distributedmanner [1]

Various detection techniques are utilized in the literatureto detect the presence of LU Various detection techniquescan be categorized as energy detector based sensing tech-nique waveform-based sensing technique cyclostationaritybased sensing technique and matched-filtering technique[6] Among the techniques energy detector gives an effectivespectrum sensing performance with low complexity Energydetector is a noncoherent detector which detects the presenceof the LU signal by measuring its energy and comparing itwith a predetermined threshold Furthermore this techniquedoes not require any prior information about LU and it is easyto implement and to be extended for the other spectrum sensing

Meanwhile CR networks (CRNs) are highly vulnerableto security threats Security for wireless networks is animportant part which ensures secure operation of the under-lying network infrastructure [7] Various attacks are studiedin the literature which highly degrades the performanceof CRN The most common attacks in CRN are primaryuser emulation attack (PUEA) and spectrum sensing datafalsification (SSDF) attack [8] In the PUEA a malicious userbehaves like an incumbent transmitter so as to enforce SUs tovacate the spectrum band In the SSDF attack the malicioususers send false information about the presence or absenceof LU to the fusion centerThe SSDF attacks severely degradethe spectrum sensing reliability and spectrum utilization

For secure CSS from the SSDF attacks various schemeshave been proposed In [9] a schemewas proposed to preventthe SSDF attacks by calculating and updating the credit valueof the SUs malicious users are excluded to avoid the SSDFattacks in cooperative spectrum sensing In [10] a coopera-tive scheme based on adaptive threshold is proposed whichutilizes matched filter detector as a second stage detectorin a confused region between signal and noise and in theclear region between signal and noise energy efficient energydetector is used as a first stage detector Main problem of theconventional energy detector is its lowdetection performanceat low signal-to-noise-ratio (SNR) region new approach hasbeen proposed in [11] to solve the problem In [11] a bilevelthresh holding approach for energy detection is proposed In[12] an improved soft fusion-based algorithm was proposedThe authors made improvements in the traditional softfusion algorithm by establishing the reputation mechanismaccording to the SUrsquos past service qualitiesThe different SUsrsquoreputation degrees are utilized for allocation of weights tothe SUs in the fusion and using this scheme the effect ofthe malicious users can be reduced In [13ndash15] the authorsexplored Dempster-Shafer (D-S) evidence theory in CSS Itincludes four consecutive procedures which are (1) basicprobability assignment (BPA) with this approach (2) holisticcredibility calculation (3) option and amelioration for BPAand (4) evidence combination via the D-S rule respectively[15] In [16] a newmethodwas proposedwhich estimated theattack strength and applies it in the kminusoutminusN rule to obtainthe optimum value of k that minimizes the Bayes risk

On the other hand above schemes do not consider thedoubtful region where it is not clear whether a certain user

is legitimate or malicious While [10] considered the scenariowith the doubtful region matched filter detection techniqueis used It is worthy of note that the matched filter detectionhas a complexity issue where it needs prior knowledge aboutLU

In our proposed scheme we utilize D-S evidence the-ory with double adaptive threshold method to differentiateamong legitimate doubtful and malicious users In the pro-posed scheme weights are assigned to each user by utilizingmaximal ratio combining (MRC) The weights are assignedto individual SUs once weights are assigned and massesie basic probability assignment (BPA) are updated Thelegitimate users have the highest weights and the malicioususers have the lowest weights The doubtful users have theweights between those of the legitimate and the malicioususers To ensure its credibility as a legitimate user a fair chanceis given through the proposed adaptive algorithm If it provesits credibility it is declared as a legitimate user Otherwise itis declared as a malicious user and withdrawn from the finaldecision Through the simulation results we verify that ourproposed scheme is effective and efficient compared to theexisting schemes

The remaining of the paper is organized as followsystem model description is given in Section 2 Section 3gives a detailed description of the proposed double adaptivethreshold scheme and the proposed algorithm at the fusioncenter In Section 4 we evaluate the performance of theproposed scheme and compare with the existing schemesFinally the paper is concluded in Section 5

2 System Model

We consider a cognitive radio network that consists ofsecondary users (SUs) malicious users (MU) and a commonreceiver which plays a role as a fusion center as shown inFigure 1 All SUs in the network perform spectrum sensingand transmit sensing result in the fusion center determiningif a licensed user (LU) is present or absent in the networkAll the SUs including both the legitimate and malicious usersparticipate in cooperation to determine the status of thelicensed user in the network

We assume that each SU independently performs localsensing by using energy detectorThe local sensing is a binaryhypotheses testing under the absence H0 or the presence H1of the LU in the network which is given by

119910 (n) = 119911 (119899) 1198670119911 (119899) + 119904 (119899) 1198671 (1)

where z(n) denotes the additive white Gaussian noise(AWGN) and s(n) denotes the transmitted signal from theLU

Since we consider the energy detection technique for thecollection of information on the existence of the LU in thenetwork the test statistics is equivalent to an estimation ofthe received signal which is given by each SU as

119909119864119895 =119873sum119895=1

10038161003816100381610038161003816119910119895100381610038161003816100381610038162 (2)

Wireless Communications and Mobile Computing 3

SU3

SU1

MU1

SU2

PU

FUSION CENTER

Legitimate ReportMalicious Report

SUn

MU2

MUm

Figure 1 System model

where 119873 = 2119879119882 in which 119879 is the sensing duration andW is the bandwidth and 119910119895 denotes the j-th sample of thereceived signal According to the central limit theorem (CLT)when the value of N is large enough eg 119873 gt 200 thecombined signal can be well approximated as a Gaussianrandom variable under hypotheses H0 and H1 with means1205830 and 1205831 and variances 12059020 and 12059021 which are given by [17]1205830 = 119873

12059020 = 119873 (120574 + 1) 11986701205831 = 119873

12059021 = 2119873 (2120574 + 1) 1198671

(3)

where 120574 is the signal to noise ratio (SNR) the LU at the SUsIn D-S evidence theory the frame of discriminant A can

be defined as 1198670 1198671 Ω where Ω describes whether of thehypotheses is true or not Based on parameters of meansand variances the BPA mH0(i) and mH1(i) are determinedas a cumulative distribution function respectively by usinghypotheses of the absence and the presence as follows [18]

1198670 mH0 (i) larr997888 119898119894 (119909119864119894 | 1198670)= 1radic21205871205900 119890

minus(119909119864119894minus1205830)212059020

1198671 mH1 (i) larr997888 119898119894 (119909119864119894 | 1198671)= 1radic21205871205901 119890

minus(119909119864119894minus1205831)212059021

(4)

where 119898119894(119909119864119894 | 1198670) and 119898119894(119909119864119894 | 1198671) denote masses ofthe BPA values for the absence or presence of the LU in thenetwork

3 Proposed Scheme

In this section we provide detailed description of theproposed scheme In CSS SUs utilize an energy detectiontechnique to sense the existence of the LU in the networkAfter performing the spectrum sensing the SUs measuretheir mass values by using the BPA Once the mass valuesof the SUs are measured the SUs can be classified into threedifferent categories If the mass value of an SU is less thanthe lowest threshold (thr1) the SU is identified as a MU anddiscarded from the final decision If the mass value of anSU is greater than the highest threshold (thr2) the SU isidentified as a legitimate SU Finally the mass value of a SUlies between the lowest threshold (thr1) and highest threshold(thr2) the SU is categorized as a doubtful user In order toprovide a fair opportunity to doubtful users and ensure theircredibility we apply the proposed algorithm to those users Ifthe credibility is ensured the user is considered as a legitimateSU otherwise it is declared as an MU and discarded fromthe final decision at the fusion center The proposed doubleadaptive threshold is graphically described in Figure 2

In Figure 2 it is shown a single threshold fixed doublethreshold and the proposed adaptive threshold The singlethreshold does not take the doubtful users into considerationIt categorized the SU as either legitimate user or an MUwhich degrades the performance of the system In fixeddouble threshold the lowest threshold and highest threshold


Single threshold

Malicious users

Malicious users

doubtful region

doubtful region

legitimate users

legitimate users

Lowest threshold Highest threshold(ＮＢＬ1) (ＮＢＬ2)


Optimal thr

Figure 2 Proposed double adaptive threshold

are fixed In the proposed scheme we consider legitimatedoubtful andMU in a double adaptive thresholding scenarioand provide a fair chance to doubtful user to prove theircredibility

The mass value of each SU is measured in (4) After themass values of the SUs are determined the next step is tomeasure the weighting factor for each SU to update themassevalue The weightage of each user is determined by

119908 (119894) = radic 119878119873119877 (119894)119879119900119905119886119897 119878119873119877 (5)

where SNR(i) is the signal to noise ratio of the i-th SU andTotal SNR is the sum of all the SUsrsquo SNRs

In the proposed algorithm if the mass value of an SUis less than the lowest threshold (thr1) then the weightassignment to the SU is zero and considered as an MU If themass value of an SU is greater than the lowest threshold (thr1)but lower than the highest threshold (thr2) its mass value isupdated and compare with the proposed adaptive thresholdIf its mass value is still lower than the highest threshold(thr2) it is categorized as anMU and discarded from the finaldecision at the fusion center Finally the updated mass valueis sent to fusion center for final decision

Once the weights of each SU are measured using (5) thenthe mass values of SUs are updated as

m1015840H0 (i) = mH0 (i) + (mH0 (i)w (i)) m1015840H1 (i) = mH1 (i) + (mH1 (i)w (i)) (6)

where 11989810158401198670(119894) is the updated masses of hypotheses ofabsence of LU as reported by i-th secondary user and11989810158401198671(119894)

is the updated masses of hypotheses of presence of LU asreported by i-th SU

The performance of spectrum sensing in CR is enhancedby keeping the highest value of probability of detection(119875119889) and lowest value of the probability of false alarm (119875119891)According to IEEE 80222 (WRAN) to prevent any interfer-ence between the LU and SU the probability of detection (119875119889)needs to be as high as possible To prevent underutilization ofspectrum probability of false alarm (119875119891) needs to be kept aslow as possible Thus the threshold should be selected suchthat we receive optimumvalue of119875119889 and 119875119891Thus by pickingup different threshold values to obtain the lowest possiblevalues 119875119891 and highest 119875119889 we obtain an optimal threshold119875119891 and 119875119889 are calculates by (7) and (8) respectively

119875119891 = 12119890119903119891119888 ( 1radic2 (119905ℎ119903 (119896) minus 2119898 )radic4119898 ) (7)

119875119889 = 12119890119903119891119888 ( 1radic2 (119905ℎ119903 (119896) minus 2119898 (119878119873119877 (119894) + 1)radic4119898 (2119878119873119877 (119894) + 1) ) (8)

where thr(k) is the range of thresholdsThe upper and lower limit of the double adaptive thresh-

old is selected based on the requirement of the decision Theoptimal threshold formasses is also calculated using the sameformula except thr(k) is replaced by a range of masses forminimal 119875119891 and maximal 119875119889

Then the fusion center is applied in Algorithm 1Once the mass values of the SUs for both hypotheses1198981198670(119894) and 1198981198671(119894) are updated by the proposed algorithm

by utilizing (6) the updated mass values of the SUs are sentto fusion center for final decision


Radio Environment

BPA mass valuemeasurement by each

SU

Measure weight of eachSU using (5)

Yes

Yes

No

No

Assign zero weightIdentified as MU

discard it

Update mass valuesusing (6) doubtful user

Legitimate SUD-S rule of combination

FC global decisionH0H1

Mass value lt ＮＢＬ1

thr1ltMass value lt ＮＢＬ2

Figure 3 Flow chart of the proposed scheme

According to the D-S evidence theory the combinationof updated masses at the fusion center can be given as

mH0 = m1015840H0 (1) oplusm1015840H0 (2) oplus m1015840H0 (i)= sum1198601cap1198601cap119860119873=1198670prod119873119894=1mHi (Ai)1 minus 119870


(9)

where 119870 = sum1198601cap1198601cap119860119873=prod119873119894=1mHi(Ai) and the operator oplusis the sequential combination of the mass values

Thefinal decision119865119889 is determined based on the followingsimple strategy

119865119889 = H1 mH1 ≻ mH0H0 mH1 ≺ mH0 (10)

The overall flowchart of the proposed scheme is given inFigure 3


Input energy of the sample (mass value 1198981198670(119894) 1198981198671(119894)) lowest threshold (thr1) highest threshold(thr2)If energy of the sample (mass value) lt lowest threshold (thr1)

Set the corresponding weights to zeroElse if lowest threshold (thr1) lt energy of the sample (mass value) lt highest threshold (thr2)

Update masses value using (6)Else

Donrsquot update mass valuesEnd ifOutput Update mass value11989810158401198670(119894) 11989810158401198671(119894)

Algorithm 1 Proposed algorithm at fusion center

Probability of false alarm0 01 02 03 04 05 06 07 08 09 1

Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09

1ROC for Always Yes attack

Proposed Scheme without MUD-S pdf without MU[13]D-S enhanced without MU[14]Jaccard scheme with MUMalicious userk-means with MUProposed Scheme with MU

Figure 4 Performance comparison of various schemes with and without ldquoAlways Yesrdquo attack

4 Numerical EvaluationIn this section we discuss the simulation results of the pro-posed scheme and compare its performance with the existingschemes In the simulation environment we placed five SUsrandomly andmeasured their local energy by utilizing energydetector technique The probability of appearance of the LUis 05 and the bandwidth is 6 MHz and sensing period is 50120583sec The simulation environment is developed by utilizingMATLAB as an implementation tool The parameters for thesimulations are summarized in Table 1

In Figure 4 the performance comparison of the proposedscheme to other existing schemes is shown with and withoutexistence of MU in the network In this scenario we consider20 malicious users for ldquoAlways Yesrdquo It can be clearlyobserved that without malicious user in the network for119875119891 of about 015 119875119889 of the proposed scheme is 09 whenthe highest 119875119889 from all the other schemes is almost 085Similarly with the presence of MU in the network 119875119889 ofthe proposed scheme drops but it is still higher than otherexisting schemes At 119875119891 of 01 119875119889 of the proposed scheme


Table 1 Parameters of the simulation

Parameters ValueNumber of SUs in the networks 5Probability of appearance of LU 05Proportion of MUs in the network 20Number of iteration 10000


Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09

1ROC for Random attack

Proposed Scheme without MUD-S pdf without MU [13]D-S enhanced without MU[14]Malicious userJaccard distance with MUk-means with MUProposed Scheme with MU

Figure 5 Performance comparison of various schemes with and without ldquoRandomrdquo attack

is almost 068 whereas the highest 119875119889 of the other existingscheme is 064

Figure 5 shows the performance comparison of theproposed scheme with other existing schemes with andwithout having malicious user in the system The malicioususers attack considered is ldquoRandomrdquo attack scenario It can beobserved fromFigure 5 that without anymalicious user in thenetwork for119875119891 of about 01119875119889 of the proposed scheme is 086when the highest119875119889 from all the other schemes is almost 079119875119889 is increased in case of a ldquoRandomrdquo malicious user due torandom behavior of the malicious user When the malicioususer is added to the network 119875119889 of the proposed scheme isstill higher than the other existing schemes At 119875119891 of 01 119875119889offered by the proposed scheme is almost 067 which is stillhigher than 119875119889 of the other existing schemes

Figure 6 shows the comparison of the proposed schemewith other existing schemes with or without an ldquoAlways Nordquo

malicious user attack in the system It is clear from Figure 6that without any malicious user in the network for 119875119891 ofabout 02 119875119889 of the proposed scheme is almost 092 when thehighest 119875119889 from all the other schemes is almost 09 Whenthe malicious user is added to the network the performanceof the proposed is better than the other existing schemesSpecifically when the value of 119875119891 is 02 119875119889 of the proposedscheme is almost 08 and still better than the other existingscheme

5 Conclusion

The spectrum sensing data falsification attacks falsifies thesensing results which highly degrades the performance ofcooperative spectrum sensing In this paper we proposeda double adaptive approach in cognitive radio networks todeal with legitimate doubtful and malicious users in the



Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09

1ROC for Always No attack

Proposed Scheme without MUD-S pdf without MU [13]D-S enhanced without MU[14]Jaccard scheme with MUMalicious userk-means with MUProposed Scheme with MU

Figure 6 Performance comparison of various schemes with and without ldquoAlways Nordquo attack

networks Maximal ratio combining scheme is utilized forweighting of the secondary users and the proposed dou-ble adaptive thresholding approach categorized legitimatedoubtful and malicious users A fair opportunity is providedto doubtful user to ensure its credibility At the fusion centerDempster-Shafer evidence theory is utilized for combininglegitimate secondary usersrsquo and making the final decisionThe performance of the proposed scheme is tested in thepresence of various types of malicious usersrsquo attacks andcompared the results with the existing schemes The resultsshowed that the proposed scheme outperforms the existingschemes in case of ldquoAlways yesrdquo ldquoAlways Nordquo and Randomattacks

Data AvailabilityThe data used to support the findings of this study areincluded within the article

Conflicts of InterestThe authors declare no conflicts of interest

AcknowledgmentsThis work was supported in part by the MSIT (Ministryof Science and ICT) Korea under the ITRC (Information

Technology Research Center) support program (IITP-2018-0-01426) supervised by the IITP (Institute for Informationand Communication Technology Promotion) and in partby the National Research Foundation (NRF) funded by theKorea government (MISP) (no NRF-2017R1A2B1004474)

References

[1] M Sansoy A S Buttar and K Singh ldquoCognitive RadioIssues andChallengesrdquo Journal of NetworkCommunications andEmerging Technologies (JNCET vol 2 p 4 2015

[2] FCC eT Docket No 03-222 Notice of Proposed Rule-Making AndOrder 2003

[3] J Mitola ldquoCognitive Radio for Flexible Mobile MultimediaCommunicationsrdquo in Proceedings of the IEEE InternationalWorkshop on Mobile Multimedia Communications 1999

[4] Cognitive Radio An Integrated Agent Architecture for SoftwareDefined Radio [PhD thesis] KTH Royal Institute of Technol-ogy 2000

[5] A Muralidharan P Venkateswaran S G Ajay D A PrakashM Arora and S Kirthiga ldquoAn adaptive threshold method forenergy based spectrum sensing in cognitive radio networksrdquoin Proceedings of the IEEE International Conference on controlinstrumentation communication and computational technologies(ICCICCT) 2015


[6] P Geete and M Motta ldquoAnalysis of Different Spectrum Sens-ing Techniques in Cognitive Radio Networkrdquo InternationalResearch Journal of Engineering and Technology (IRJET) vol 2p 5 2015

[7] M Jenani ldquoNetwork Security A Challengerdquo International Jour-nal of Advanced Networking and Application vol 8 no 5 pp120ndash123 2017

[8] J Marinho J Granjal and E Monteiro ldquoA survey on securityattacks and countermeasures with primary user detection incognitive radio networksrdquo EURASIP Journal on InformationSecurity vol 4 p 4 2015

[9] H Du S Fu and H Chu ldquoA credibility-based defense ssdfattacks scheme for the expulsion of malicious users in cognitiveradiordquo International Journal of Hybrid Information Technologyvol 8 p 9 2015

[10] A Rauniyar and S Y Shin ldquoCooperative adaptive thresholdbased energy and matched filter detector in cognitive radionetworksrdquo Journal of Communication and Computer vol 12 pp13ndash19 2015

[11] S Mamatha and K Aparna ldquoDesign of an Adaptive EnergyDetector based on Bi-LevelThresh holding in Cognitive RadiordquoInternational Journal of Scientific Engineering and TechnologyResearch vol 4 no 2 pp 346ndash349 2015

[12] T Peng Y Chen J Xiao Y Zheng and J Yang ldquoImprovedsoft fusion-based cooperative spectrum sensing defense againstSSDF attacksrdquo in Proceedings of the 2016 International Confer-ence on Computer Information and Telecommunication SystemsCITS 2016

[13] Q Peng K Zeng W Jun and S Li ldquoA distributed spectrumsensing scheme based on credibility and evidence theoryrdquoin Proceeding of the IEEE 17th International symposium onPersonal Indoor and Mobile Radio Communication pp 1ndash5Helsinki Finland 2006

[14] N Nguyen-Thanh and I Koo ldquoAn enhanced cooperative spec-trum sensing scheme based on evidence theory and reliabilitysource evaluation in cognitive radio contextrdquo IEEE Communi-cations Letters vol 13 no 7 pp 492ndash494 2009

[15] F Ye X Zhang and Y Li ldquoCollaborative spectrum sensingalgorithm based on exponential entropy in cognitive radionetworksrdquo Symmetry vol 9 p 36 2017

[16] A A Sharifi and M J Musevi Niya ldquoDefense Against SSDFAttack in Cognitive Radio Networks Attack-Aware Collab-orative Spectrum Sensing Approachrdquo IEEE CommunicationsLetters vol 20 no 1 pp 93ndash96 2016

[17] N Nguyen-Thanh and I Koo ldquoA Robust Secure CooperativeSpectrum Sensing Scheme Based on Evidence Theory andRobust Statistics in Cognitive Radiordquo IEICE Transactions onCommunications vol 92-B no 12 pp 3644ndash3652 2009

[18] M S Khan and I Koo ldquoThe Effect of Multiple Energy Detectoron Evidence Theory Based Cooperative Spectrum Sensing forCognitive Radio Networksrdquo Journal of Information ProcessingSystems vol 12 no 2 pp 295ndash309 2016

International Journal of

AerospaceEngineeringHindawiwwwhindawicom Volume 2018

RoboticsJournal of

Hindawiwwwhindawicom Volume 2018


Active and Passive Electronic Components

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwwwhindawicom

Volume 2018

Hindawi Publishing Corporation httpwwwhindawicom Volume 2013Hindawiwwwhindawicom

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of



Journal ofEngineeringVolume 2018

SensorsJournal of



RotatingMachinery


Modelling ampSimulationin EngineeringHindawiwwwhindawicom Volume 2018


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation


Hindawi

wwwhindawicom Volume 2018

Advances in

Multimedia

Submit your manuscripts atwwwhindawicom


senses the radio environment and different SUs share theirinformation and make their own decision with distributedmanner [1]

Various detection techniques are utilized in the literatureto detect the presence of LU Various detection techniquescan be categorized as energy detector based sensing tech-nique waveform-based sensing technique cyclostationaritybased sensing technique and matched-filtering technique[6] Among the techniques energy detector gives an effectivespectrum sensing performance with low complexity Energydetector is a noncoherent detector which detects the presenceof the LU signal by measuring its energy and comparing itwith a predetermined threshold Furthermore this techniquedoes not require any prior information about LU and it is easyto implement and to be extended for the other spectrum sensing

Meanwhile CR networks (CRNs) are highly vulnerableto security threats Security for wireless networks is animportant part which ensures secure operation of the under-lying network infrastructure [7] Various attacks are studiedin the literature which highly degrades the performanceof CRN The most common attacks in CRN are primaryuser emulation attack (PUEA) and spectrum sensing datafalsification (SSDF) attack [8] In the PUEA a malicious userbehaves like an incumbent transmitter so as to enforce SUs tovacate the spectrum band In the SSDF attack the malicioususers send false information about the presence or absenceof LU to the fusion centerThe SSDF attacks severely degradethe spectrum sensing reliability and spectrum utilization

For secure CSS from the SSDF attacks various schemeshave been proposed In [9] a schemewas proposed to preventthe SSDF attacks by calculating and updating the credit valueof the SUs malicious users are excluded to avoid the SSDFattacks in cooperative spectrum sensing In [10] a coopera-tive scheme based on adaptive threshold is proposed whichutilizes matched filter detector as a second stage detectorin a confused region between signal and noise and in theclear region between signal and noise energy efficient energydetector is used as a first stage detector Main problem of theconventional energy detector is its lowdetection performanceat low signal-to-noise-ratio (SNR) region new approach hasbeen proposed in [11] to solve the problem In [11] a bilevelthresh holding approach for energy detection is proposed In[12] an improved soft fusion-based algorithm was proposedThe authors made improvements in the traditional softfusion algorithm by establishing the reputation mechanismaccording to the SUrsquos past service qualitiesThe different SUsrsquoreputation degrees are utilized for allocation of weights tothe SUs in the fusion and using this scheme the effect ofthe malicious users can be reduced In [13ndash15] the authorsexplored Dempster-Shafer (D-S) evidence theory in CSS Itincludes four consecutive procedures which are (1) basicprobability assignment (BPA) with this approach (2) holisticcredibility calculation (3) option and amelioration for BPAand (4) evidence combination via the D-S rule respectively[15] In [16] a newmethodwas proposedwhich estimated theattack strength and applies it in the kminusoutminusN rule to obtainthe optimum value of k that minimizes the Bayes risk

On the other hand above schemes do not consider thedoubtful region where it is not clear whether a certain user

is legitimate or malicious While [10] considered the scenariowith the doubtful region matched filter detection techniqueis used It is worthy of note that the matched filter detectionhas a complexity issue where it needs prior knowledge aboutLU

In our proposed scheme we utilize D-S evidence the-ory with double adaptive threshold method to differentiateamong legitimate doubtful and malicious users In the pro-posed scheme weights are assigned to each user by utilizingmaximal ratio combining (MRC) The weights are assignedto individual SUs once weights are assigned and massesie basic probability assignment (BPA) are updated Thelegitimate users have the highest weights and the malicioususers have the lowest weights The doubtful users have theweights between those of the legitimate and the malicioususers To ensure its credibility as a legitimate user a fair chanceis given through the proposed adaptive algorithm If it provesits credibility it is declared as a legitimate user Otherwise itis declared as a malicious user and withdrawn from the finaldecision Through the simulation results we verify that ourproposed scheme is effective and efficient compared to theexisting schemes

The remaining of the paper is organized as followsystem model description is given in Section 2 Section 3gives a detailed description of the proposed double adaptivethreshold scheme and the proposed algorithm at the fusioncenter In Section 4 we evaluate the performance of theproposed scheme and compare with the existing schemesFinally the paper is concluded in Section 5

2 System Model

We consider a cognitive radio network that consists ofsecondary users (SUs) malicious users (MU) and a commonreceiver which plays a role as a fusion center as shown inFigure 1 All SUs in the network perform spectrum sensingand transmit sensing result in the fusion center determiningif a licensed user (LU) is present or absent in the networkAll the SUs including both the legitimate and malicious usersparticipate in cooperation to determine the status of thelicensed user in the network

We assume that each SU independently performs localsensing by using energy detectorThe local sensing is a binaryhypotheses testing under the absence H0 or the presence H1of the LU in the network which is given by

119910 (n) = 119911 (119899) 1198670119911 (119899) + 119904 (119899) 1198671 (1)

where z(n) denotes the additive white Gaussian noise(AWGN) and s(n) denotes the transmitted signal from theLU

Since we consider the energy detection technique for thecollection of information on the existence of the LU in thenetwork the test statistics is equivalent to an estimation ofthe received signal which is given by each SU as

119909119864119895 =119873sum119895=1

10038161003816100381610038161003816119910119895100381610038161003816100381610038162 (2)


SU3

SU1

MU1

SU2

PU

FUSION CENTER


SUn

MU2

MUm



12059020 = 119873 (120574 + 1) 11986701205831 = 119873

12059021 = 2119873 (2120574 + 1) 1198671

(3)



1198670 mH0 (i) larr997888 119898119894 (119909119864119894 | 1198670)= 1radic21205871205900 119890

minus(119909119864119894minus1205830)212059020

1198671 mH1 (i) larr997888 119898119894 (119909119864119894 | 1198671)= 1radic21205871205901 119890

minus(119909119864119894minus1205831)212059021

(4)


3 Proposed Scheme




Single threshold

Malicious users

Malicious users

doubtful region

doubtful region

legitimate users

legitimate users



Optimal thr




119908 (119894) = radic 119878119873119877 (119894)119879119900119905119886119897 119878119873119877 (5)















Radio Environment


SU


Yes

Yes

No

No


discard it










(9)



119865119889 = H1 mH1 ≻ mH0H0 mH1 ≺ mH0 (10)









Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09










Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09








5 Conclusion




Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09









References






















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



SU3

SU1

MU1

SU2

PU

FUSION CENTER


SUn

MU2

MUm



12059020 = 119873 (120574 + 1) 11986701205831 = 119873

12059021 = 2119873 (2120574 + 1) 1198671

(3)



1198670 mH0 (i) larr997888 119898119894 (119909119864119894 | 1198670)= 1radic21205871205900 119890

minus(119909119864119894minus1205830)212059020

1198671 mH1 (i) larr997888 119898119894 (119909119864119894 | 1198671)= 1radic21205871205901 119890

minus(119909119864119894minus1205831)212059021

(4)


3 Proposed Scheme




Single threshold

Malicious users

Malicious users

doubtful region

doubtful region

legitimate users

legitimate users



Optimal thr




119908 (119894) = radic 119878119873119877 (119894)119879119900119905119886119897 119878119873119877 (5)















Radio Environment


SU


Yes

Yes

No

No


discard it










(9)



119865119889 = H1 mH1 ≻ mH0H0 mH1 ≺ mH0 (10)









Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09










Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09








5 Conclusion




Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09









References






















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Single threshold

Malicious users

Malicious users

doubtful region

doubtful region

legitimate users

legitimate users



Optimal thr




119908 (119894) = radic 119878119873119877 (119894)119879119900119905119886119897 119878119873119877 (5)















Radio Environment


SU


Yes

Yes

No

No


discard it










(9)



119865119889 = H1 mH1 ≻ mH0H0 mH1 ≺ mH0 (10)









Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09










Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09








5 Conclusion




Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09









References






















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



Radio Environment


SU


Yes

Yes

No

No


discard it










(9)



119865119889 = H1 mH1 ≻ mH0H0 mH1 ≺ mH0 (10)









Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09










Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09








5 Conclusion




Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09









References






















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia









Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09










Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09








5 Conclusion




Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09









References






















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09








5 Conclusion




Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09









References






















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




Prob

abili

ty o

f det

ectio

n

0

01

02

03

04

05

06

07

08

09









References






















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


















RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


a double adaptive approach to tackle malicious users in...

Documents