research article consensus-based distributive cooperative...

Hindawi Publishing CorporationInternational Journal of Distributed Sensor NetworksVolume 2013 Article ID 450626 10 pageshttpdxdoiorg1011552013450626

Research ArticleConsensus-Based Distributive Cooperative Spectrum Sensing forMobile Ad Hoc Cognitive Radio Networks

Anam Raza1 Waleed Ejaz2 Syed Sohail Ahmed1 and Hyung Seok Kim2

1 Department of Computer Engineering University of Engineering and Technology Taxila Pakistan2Department of Information and Communication Engineering Sejong University Seoul Republic of Korea

Correspondence should be addressed to Hyung Seok Kim hyungkimsejongedu

Received 3 September 2013 Accepted 31 October 2013

Academic Editor Khalid Al-Begain

Copyright copy 2013 Anam Raza et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Cognitive radio network (CRN) is an intelligent network that provides solution to underutilization of spectrum band by detectingspectrum holes For this purpose certain important tasks including spectrum sensing spectrum analysis and spectrum decisionare required to be performed Spectrum sensing is the crucial step as it is not only responsible for fast and reliable detection ofprimary users (PU) but also evades disturbance to their transmission Different methods of enhancements in local sensing schemehave been proposed in the literature However cooperative spectrum sensing schemes are preferred as they provide significant gainsin CRN performance by countering shading effects In this paper an efficient distributed cooperative spectrum sensing scheme isproposed for mobile ad hoc cognitive radio networks The consensus-based distributed cooperative spectrum sensing (CDCSS)scheme relies on energy detection for local sensing Random walk mobility model (RWMM) is used for the movement pattern ofnodesMobilitymodel is employed asmovement of nodes results in reduced fading effects and efficient detection Simulation resultswere compared with the existing consensus algorithm and equal gain combining (EGC) rule and the results showed improvementin sensing phenomenon

1 Introduction

Advancement in wireless communication requires efficientutilization of limited spectrum resources Recent researchshows that this limitation is because of spectrum man-agement policies [1] To overcome this limitation and forbetter utilization of spectrum one requires a new networkingstandard which is known as cognitive radio network (CRN)[2ndash4] Software defined radio (SDR) is used to build cognitiveradio (CR) CR is a smart system that senses the environmentcalled spectrum sensing and adapts to variations in operatingparameters The two prime objectives of spectrum sensingare the efficient detection of spectrum holes and interferenceavoidance with primary systemlicensed system [5ndash7] Cur-rent research divides spectrum sensing into two branchesthat are local sensing and cooperative spectrum sensingIn cooperative spectrum sensing each cognitive radio user(CRU) shares its local observation with the rest of CRUs inthe network which results in improved spectrum sensing

In cooperative spectrum sensing the CRUs either for-ward their sensing information to the central entity orthey can exchange sensing information with each otherfor cooperative decisions In infrastructure mode shown inFigure 1(a) there is central entity to fuse the sensing resultshowever there is no central entity in ad hoc mode alsoknown as cognitive radio ad hoc network (CRAHN) as shownin Figure 1(b) Distinguishing features of CRAHNrsquos are lackof central entity distributed multihop architecture dynamicnetwork topology and timelocation-varying spectrum avail-ability [8] Advancement introduced to the ad hoc networkinvolvesmobility and the network is known asmobile ad-hocnetwork (MANET) InMANETs devices canmove randomlyin all directions and links with other devices are also updatedrepeatedlyWithout utilizing a fixed infrastructure a dynamicnetwork is established by wireless nodes

To achieve a common goal MANET nodes cooperatewith each other because of the lack of centralized control inMANETs In self-organization of nodes the major respon-sibilities are topology organization neighbor discovery and

2 International Journal of Distributed Sensor Networks

Boundary ofdecodability

of primary user

Cognitive radiobase station

Cognitive radionetwork access

Cognitive radio network(with infrastructure)

(a)


of primary user

Cognitive radioad hoc access

Cognitive radio

(without infrastructure)ad hoc network

(b)

Figure 1 Types of network (a) infrastructure based CRN and (b) CRAHN

topology reorganization [9] Movement patterns of mobilenodes (MNs) are defined using mobility models It is impor-tant to keep track of node movement because nodes inMANETs are free to move This also includes informationabout their speed location velocity and acceleration changeover time

In this paper a consensus-based distributed cooperativespectrum sensing scheme (CDCSS) in CR-MANET is pro-posed which is inspired by Novel biological mechanismsThese mechanisms have become an important phenomenonin handling intricate communication networks CDCSSworks on MNs in distributive network without using acentralized center to improve the sensing performance inCR-MANETsThemain contributions of this paper are as follows

(i) A consensus-based distributed spectrum sensing forCR-MANETs is proposed

(ii) Random walk mobility model (RWMM) is used inCRN for the movement pattern of nodes

(iii) Through the simulation the proposed scheme iscompared with existing consensus algorithm andEGC-rule scheme

The rest of the paper is as follows Section 2 explains workcarried out in fields of spectrum sensing and mobilitymodels Section 3 involves the methodology adopted todevelop a scheme for cooperative spectrum sensing amongMNs Section 4 explains implementation process Section 5involves analysis and Section 6 contains conclusion of ourwork

2 Background and Related Work

The rapid advancement in wireless applications has resultedin increased usage of unlicensed band causing a problem ofnonuniform spectrum usage [10] The CR cycle is divided

into four broad fields of research to cope with spectrum uti-lization challenges (1) the spectrum sensing that determineswhich portion of the spectrum is available (2) the spectrumdecision that picks the best vacant channel (3) the spectrumsharing that allows userrsquos coordinated access to channel and(4) the spectrum mobility that allows vacating the channelwhen a PU is detected

21 Spectrum Sensing Spectrum sensing techniques are clas-sified into noncooperative and cooperative spectrum sens-ing The three most common schemes for noncooperativetransmitter detection are energy detection matched filterdetection and cyclostationary detection [11ndash14]

Due to random changing in wireless environment it isdifficult for a single CRU to detect PU signal accuratelyTo cope with factors such as noise uncertainty shadingand fading cooperative spectrum sensing is introduced byresearchers In cooperative spectrum sensing CRUs cooper-ate and share their information about PU detection Thesemethods give more accurate results as uncertainty can beminimized [15]

In decentralized spectrum sensing there is no require-ment of infrastructure and fusion center Here the CRUsexchange their information with each other to cooperateThemost well-known decentralized spectrum sensing techniqueis gossiping algorithm because it performs sensing with asignificant low overhead Other decentralized techniquesinclude clustering schemes which are already known forsensor network architectures In these schemes CRUs formclusters which coordinate among themselves [16]

A gradient based distributed cooperative spectrum sens-ing method was proposed for CRAHNs [17] The gradientfield changeswith the energy sensed byCRU and the gradientis calculated based on the components which include energysensed by CRUs and received from neighbors The proposedscheme was evaluated on the basis of reliable sensing

International Journal of Distributed Sensor Networks 3

Mobility models

Reference pointgroup model

Speed decayproblem

Gauss-Markovmodel

Randomwaypoint

model

Models withgeographicrestriction

Models with spatial

dependency

Models withtemporal

dependencyRandom models

Random walkmodel

Randomdirection

modelSmooth randommobility model

Obstacle mobility model

Set of correlatedmodels

Pathway mobility modelOther variants

Figure 2 Mobility model classification

convergence time and energy consumption This schemeconsumes less energy compared to existing consensus-basedapproach

Wu et al [18] proposed a decentralized clustering con-sensus algorithm based on two phases First phase is relatedto clustering CRU with best detection performance andsecondphase carries out distributed data fusion of the sensingoutcomes of the CRUs in the cluster Like [18] our proposedscheme is also for distributed network but this former schemedoes not take into account parameters related to mobility ofnodes which on the other hand are a major concern of ourproposed scheme

Ribeiro et al [19] presented a two-step adaptive combin-ing based technique for distributive cooperative spectrumsensing In the first step an adaptive combiner is used forthe fusion of neighborhood nodes and in the second step aconsensus decision is made by sharing local decisions withinthe neighborhoods Their results showed better performancecompared to the optimal linear fusion rule

In [20] a consensus-based spectrum sensing schemeis presented This scheme is fully distributive where localsensing information obtained by CRUs is sent to theirneighbors Information from neighbors is used by CRUsand consensus algorithm is applied for stimulating new statefor consensus variable This process is continued till theindividual states converge to a common value Spectrumsensing data falsification (SSDF) attacks are also dealt withby researchers by excluding those nodes from neighbors listthat give very much deviation from mean value

Like [20] our proposed scheme is also for distributednetwork The existing consensus algorithm required eachnode to have a prior knowledge of the upper bound of themaximum degree of the network Our proposed algorithmis not only fully distributive and robust against SSDF attacksbut also does not require prior knowledge of degree of thenetwork Further this new algorithm is applied on mobilenodes For nodes motion mobility models are investigated

22 Mobility Models The movement patterns of MN aredefined using mobility models It is important to keep

track of nodes movement as nodes in MANETs are freeto move This also includes information about their speedlocation velocity and acceleration change over time Baiand Helmy [21] provided a detailed classification of mobilitymodels according to themobility characteristics which is alsoshown in Figure 2 Some models keep track of their historymovements and are referred to as models with temporaldependency while the others are restricted by geographicbounds There is also a class of mobility models known asmodels with spatial dependency in which nodes move in acorrelatedmannerMost of thesemobilitymodels are randommodels

3 Modeling Philosophy

In this section we will first discuss the spectrum sensingmodel of CRU and then illustrate the local sensingmodel andinformation sharing with neighboring CRUs using CDCSSNetwork topology and consensus notions for CDCSS are alsopresented along with the movement pattern of CRUMNusing RWMM

Spectrum sensing process is divided into two phases thatare local spectrum sensing and cooperative spectrum sensingFor local spectrum sensing CRUs use energy detection tech-nique to detect PU and make local decision about presenceor absence of PU Reference [12] shows a block diagram ofenergy detection

The basic hypothesis model for local spectrum sensing isgiven as [1 14 16]

119909 (119905) = 119899 (119905) 119867

0

ℎ lowast 119904 (119905) + 119899 (119905) 1198671

(1)

where 119909(119905) is the received signal 119899(119905) is the additive whiteGaussian noise (AWGN) 119904(119905) is the PU signal and ℎ is theamplitude gain of PU signal119867

1indicates that a PU is present

while1198670represents absence of PU

Received signal is passed through a band-pass filter ofbandwidth 119882 and center frequency 119891

119904 Signal received from

filter is further fed to a squaring device and then to anintegrator over a time period 119879 Output from the integrator is


Start

Implement CDCSS

Compute converged

PU absentPU present

Decision

Update consensus variable states

xlowast gt 120582 xlowast lt 120582

value (xlowast)

Figure 3 Flow chart for cooperative sensing

checked against predefined threshold 120582 and makes a decisonabout presence or absence of PU and is represented as 119884Local measurement of user 119894 is denoted by 119884

119894 The output 119884

of the energy detection has the distribution [22]

119884 =

1205942

21198851198670

12059422119885(2120574) 119867

1

(2)

where 12059422119885

represents central chi-square distribution and12059422119885(2120574) represents noncentral chi-square distribution each

with 2119885 (119885 = 119879119882) degree of freedom and a non-centralityparameter of 2120574 for the latter distribution Here 120574 representsSNR So in short in local observation eachMN is representedas 119894 will detect the energy after every 119879 seconds and takemeasurement 119884

119894

31 Consensus-Based Distributive Cooperative Spectrum Sens-ing In this stageMNs communicates their local informationwith their neighbors and make a common decision aboutpresence or absence of PU They share information viacommon control channel (CCC) The CCC is responsiblefor transferring control information and CRUs coordinatewith each other using this medium The CCC not onlyfacilitates cooperation among CRUs but also provides severalother network operations that include neighbor discoverytopology change channel access negotiation and routinginformation updates In this algorithm overlay CCC wasused because CRUs temporarily allocated the spectrum notused by PU Further in this algorithm we assumed that thededicated CCC is predetermined and usually unaffected byPU activity

Cooperative sensing is an iterative process where eachMN sets 119909

119894(0) = 119884

119894as the starting value of local state

variable at time instant 119896 = 0 For further time intervalsthat is 119896 = 1 2 MNs share their local observation with

neighbors and then compute their next state 119909119894(119896 + 1) This

is the basic idea adopted from self-organizing capability ofCRAHNs The maximum degree of the network is definedas the maximum number of neighbors of a node In thepresent work the degree is represented as Δ and step size isrepresented as 1205761015840 where 0 lt 1205761015840 lt 1Δ The state of each MN isobtained by following rule

119909119899(119896 + 1) = 119909

119899(119896) + 120576

1015840sum119898120576119873

(119909119898(119896) minus 119909

119899(119896)) (3)

where119873 represents number of nodes and 119909119898is the neighbor

with which 119909119899shares information This process continues

until a converged value is achieved by all MNs This con-verged value is represented by 119909lowast This converged value isthen compared against a predefined threshold 120582 and thencooperative decision is made by all MNs about the presenceor absence of PU Figure 3 shows the flow diagram of theproposed consensus algorithm applied on MNs

Decision = 1 119909lowast gt 120582

0 119909lowast le 120582(4)

An average consensus is ensured if 0 lt 1205761015840 lt 1Δ Ifno malicious node is present in the network then the finalconverged value 119909lowast will be equal to the average of initial statevalues The convergence rate is one of the great concerns inthe field of cooperative spectrum sensing MNs continuouslydetect the presence of PU and on identifying its presenceit backs up as soon as possible For this reason reaching atconsensus that is convergence rate is a significant factorfor network topology design and also for analyzing theperformance of CDCSS

Yu et al [20] proposed (3) in which constant step size 120576is used which appeared as a limitation as it required eachnode to have a complete knowledge of network topology andmaximumdegree of the network Also it was applied on fixednetwork with predefined connections In this new proposedtechnique these limitations were overcome by the following

(1) Employing a variable 1205761015840 instead of constant step sizerather than using a predefined constant step size itsvalue is predicted by following rule [23]

Δ sim 1198731120572minus1

(5)where Δ represents the degree of network 119873 rep-resents the number of nodes in the network and 120572is a constant factor that can have a value between 2and 3 Using this rule no knowledge about completenetwork topology is required except for the number ofnodes in the network Using this rule the maximumdegree of the network (Δ) can also be predicted Thestep size is given as 0 lt 1205761015840 lt 1Δ

(2) Applying further CDCSS scheme on MNs which aretravelling considering RWMM issues faced in case offixed graph do not occur here as nodes aremobile andmotion of nodes causes small-scale fading [24] Asnodes are mobile nodes with a distance greater thana certain limit are not considered as neighbors whichchanges the degree of node (Δ) and makes step size(1205761015840) a variable factor instead of constant value


32 Network Topology Using Random Graph As stated ear-lier eachMN creates links with neighbors and communicateslocal information with them Network topology used asreference in this research can be represented as a graph 119866The graph consists of nodes 119894 = 1 2 3 119899 and a set ofedges where each edge is denoted by 119864 When two MNsare connected in an edge it means that they can send andreceive information from each other and their ordered pairwill be (119894 119895) When nodes are connected through a paththen the graph is said to be connected A path in a graphconsists of sequence of nodes such as 119894

1 1198942 1198943 119894

119898 where

119898 ge 2 In case of fixed graphs the nodes face certainproblems such as fading of signals which may cause linkfailures and packet errors and signal reception may also getaffected by moving objects between neighboring nodes Toovercome this problem we used random graph in whichthe link goes down as the nodes move away Random graphrepresentation of network topology of 10MNs which wastaken for demonstration in the present work is shown inFigure 4

33 Random Walk Mobility Model The basic idea behindthis movement emerged from unexpected movements ofparticles It was developed to follow such erratic movementpatterns and hence it became the most widely used mobilitymodel MN randomly chooses its new speed and directionand thenmoves from its current location to a new location towhich it is to travel Predefined ranges are used for selectingnew speed and direction For speed the range can be fromminimum speed to maximum speed and directions canrange from 0 to 2120587 Either a constant time interval 119905 or aconstant distance 119889 is selected for each movement in thismodel At the end of each movement a new direction andspeed are selected in a similarmanner for the nextmovementIn this model when MN reaches simulation boundary itbounces back from the simulation boundary and comesback at some angle This angle is determined by incomingdirection The MN then moves along a newly selected pathRWMM is a memoryless model as no knowledge of previousspeed and direction is retained and also these parametersare not used for future decisions The current direction andspeed of MN is independent of both past and future speedand direction

Different derivates of this model are also developed byresearchers These include 1D 2D 3D and d-D walks [25]Here we here are concerned with 2D walk which is alsoshown in Figure 5 In this example MN is allowed to choosea speed between 0 and 10ms and a direction between 0 and2120587 In this case fixed time interval is used that is MN canmove in a direction for 1 sec before changing its direction andchoosing a new destination As time is made constant thedistance to be travelled is not fixed or specifiedMN canmovein the specified simulation boundary

4 Simulation Setup

As mentioned earlier we used random graph of nodes andassumed that nodes were mobile which follow RWMM for

4

1

3

2

76

5

10

8

9

NodeBidirectional link with link failure probability

Figure 4 Network topology of 10MNs

minus600 minus400 minus200 0 200 400 600minus600

minus400

minus200

0

200

400

600

Figure 5 Mobility pattern of MN using RWMM

movement patterns Figure 6 shows the simulation result formovement of 10MNsThe time period for this simulationwas100 seconds MN took each move in constant time interval of1 second The simulation area in which nodes moved was a20m times 20m rectangle with 15m transmission range

Such transmission scenario consisting of MNs can beseen in wireless regional area network (WLAN) which hasadopted operations of mobile devices [24] As in RWMMeither a constant time interval or constant speedwas usedWeconsidered a constant time interval while speed was variablewhich made the distance travelled by MN in each move avariable factor

41 Constraints for Consensus-Based Distributive CooperativeSpectrum Sensing CDCSS algorithm is based on certainconstraints some of which were not considered in [20]Theseconstraints are as follows

(1) implementing prediction rule given in (5) to get max-imum degree of the network (Δ) which in turn gives


0 5 10 15 200

2

4

6

8

10

12

14

16

18

20Random walk mobility model

X (meters)

Y (m

eter

s)

Time (s) = 30

Figure 6 RWMM for 10MNs

range for selection of step size (1205761015840) this predictionwill omit the need of knowledge of whole networktopology

(2) sharing local information of MNs with their neigh-bors

(3) excluding those MN from neighbor list which givemaximum deviation from mean value (calculated bytaking the average of all values fromMN)

(4) considering distance restriction that is if a neighborMN exceeds a specified distance then exclude it fromneighbor list thus making step size variable (1205761015840)

Above given factor affect affect convergence rate so theaim through their execution is to fasten convergence rateImplementation of these constraints is explained below Thiswhole process of CDCSS and RWMM is also shown viaflowchart in Figure 7

42 Prediction Rule The first step for cooperation is topredict maximum degree of the network Equation (5) givesthis prediction rule Ten-node network is shown in Figure 6therefore using prediction rule the degree of network iscalculated as

Δ sim 10125minus1

(6)

where119873 = 10 and 120572which is a constant factor and can have avalue between 2 and 3 so 25 is assigned to 120572Thusmaximumdegree of network predicted for 10-node network using (5)is Δ = 5 Thus the maximum degree of network Δ is 5 for10-node network Then 0 lt 120576

1015840 lt 15 We know that thealgorithm converges faster when 1205761015840 approaches 1Δ in thiscase 02

After getting maximum degree of network all MNs willcontinue sharing their local information with neighbors andupdate their states according to (3) Given below is thesimulation results obtained by applying prediction rule and

Converged

Start

Implement mobility model

Keep track of nodes

Energy detection for local sensing

Apply consensus algorithm and share

neighborrsquos info

Predict network degree

Calculate average value and apply

Euclidian formula

Consensus achieved

PU absentPU present

Yes

No

NoYesvalue (xlowast) gt 120582

Figure 7 Flowchart of complete algorithm

achieving consensus among MNs The SNR here is a fixedfactor having value of 10 dB

Figure 8 shows estimated PU energy in a network within a network of 10 nodes From Figure 8 it is very clear thatinitially all MNs gave very different energy values which aretheir local observations These values are then shared amongneighboring MNs and after some iterations consensus isachieved This is shown in Figure 8 where after about 13iterations difference between the MNs is less than 1 dB andafter about 19 iterations it is less than 01 dBHere 30 iterationsare shown for demonstration purpose If we increase thenumber of MNs in the network then number of iterationsrequired to achieve consensus also increases accordingly Asa result the algorithm also converges slowly

43 Average Value Calculation It is stated earlier that theconsensus process takes place in iterations With every time


0 5 10 15 20 25 305

10

15

20

25

30

Number of iterations

Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10

Figure 8 Consensus algorithm with prediction rule

instant that is 119896 = 0 1 2 state of consensus variable ofeachMN119909

119894(119896) is updatedAlongwith this updating eachMN

keeps track of the energy level received from the neighborMN calculates deviation of its neighbors from mean valueThe neighbor with maximum deviation is an attacker andMN excludes this neighbor from neighborrsquos list Neighborsshowing maximum deviation can be an attacker which giveswrong energy values and eventually can lead to increasedfalse alarms and reduced correct detection of PU Due to thisconstraint a safety measure was employed for network As aresult only those energy values will be considered for stateupdating that are from verified neighbors

As there was no malicious node in this network of 10nodes the result was the same as the result for simulationwith prediction rule shown in Figure 8 The convergencevalue that is 119909lowast will be the same as the average of initialenergy levels fromMNs In this simulation the average valueachieved is 196 dB which is same as the average of initialenergy levels from all MNs

44 Distance Restriction We assumed that nodes weremobile and moving considering RWMM constraints Theywere moving in a cell of 20m times 20m area This is also shownin Figure 6 Mobility of nodes can cause an increase in signalpower and small-scale fading Considering neighboring listif the distance between nodes exceeds certain limit then theycan be excluded fromneighborrsquos list In this scenario of a 20mtimes 20m area if the distance between nodes exceeds 15m thenthey are excluded from neighbors list This in turn will causevariation in step size as maximum degree of node is varyingTo calculate distance between nodes Euclidianrsquos formula wasused which is given in (7) Simulation result for CDCSS

0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10

Figure 9 Consensus algorithm with distance restriction

which also employed this distance restriction is shown inFigure 9

119889 = radic(1199092minus 1199091)2+ (1199102minus 1199101)2 (7)

Simulation result in Figure 9 shows that now less iterationsare taken to achieve consensus compared to previous resultsThe difference between energy values is less than 1 dB after 10iterations and is less than 01 dB after 15 iterations Here alsothe converged value is equal to the average of initial energyvalues of MNs that is 196 dB

45 Complete Consensus Algorithm Finally all these con-straints were simultaneously applied to achieve consensusamong MNrsquos energy values and come to a common decisionabout presence or absence of PU Figure 10 shows a simula-tion result obtained by considering all the above rules thatis prediction rule average value calculation and distancerestriction

Figure 10 revealed that convergence rate is the fastestin this case compared to all of the rest of the cases Thedifference between energy values ofMNs is less than 1 dB after7 iterations and is less than 01 dB after 12 iterationsThe finalconverged value is 119909lowast = 196 dB

5 Simulation Results

The CDCSS for distributed network does not have a centralentity Therefore for analysis purpose it is compared withtechniques that are employed either for distributed networkor centralized networks For distributed network CDCSSwascompared with consensus algorithm given in Yu et al [20]while for centralized network it was compared with EGCrule which is a soft combining rule EGC is one of the simplest


0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10

Figure 10 Complete consensus algorithm

linear soft combining rules where estimated energy in eachnode is sent to the fusion center where they will be addedtogether The summation is compared with a predefinedthreshold and the decision ismade about presence or absenceof PU accordingly Comparisons between CDCSS and EGCrule were made on the basis of 119875

119889and 119875

119891and probability of

missed detection (119875119898)

51 Comparison with EGC Rule Firstly comparison wasmade on the basis of 119875

119898and 119875

119891 If number of false alarms

increases then 119875119891also increases which will result in low

spectrum utilization 119875119898is defined as

119875119898= 1 minus 119875

119889 (8)

As 119875119898increases this will cause high missed detection of PU

which will result in disturbance to PU transmission Thismeans that a good scheme is the one in which both of thesefactors are minimized Comparison of CDCSS with existingEGC rule is shown in Figure 11 This comparison comprises119875119891and 119875

119898

It is clear from Figure 11 that CDCSS performs bettercompared to existing EGC rule New scheme has lower 119875

119891

compared to EGC rule This will result in less interferenceto PU thus improving PUrsquos transmission and causing moreefficient and reliable spectrum sensing

Further in analysis comparison is made on the basisof varying SNR and sensitivity in detection that is 119875

119889

Figure 12 shows simulation result for SNR versus 119875119889 Here

SNR varies from minus20 dB to 20 dB Time-bandwidth factorthat is 119879119882 = 5 and 119875

119891 is fixed here having a value

of 01 From Figure 12 it is clear that in terms of averageSNR CDCSS has considerable improvements for performingdetection For 119875

119889gt 099 CDCSS required a lesser average

SNR than existing EGC-rule

0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1

Proposed algorithmExisting EGC algorithm

Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)

Probability of false alarm (Pf)

Figure 11 Comparison on the basis of119875119891versus119875

119898with the existing

EGC rule (SNR = 10 dB 119879119882 = 5)

52 Comparison with the Existing Consensus AlgorithmCDCSS was also compared with the existing consensusalgorithm which was given in Yu et al [20] This existingalgorithm was also used for distributive network Figure 13shows comparison of result for CDCSS and existing consen-sus algorithm

It is clear from Figure 13 that CDCSS has lower Pmthan existing consensus algorithm This means CDCSS hasimproved spectrum utilization and reduced PU interferenceMobility of nodes aided in better performance of CDCSSwhen compared to existing consensus algorithm As in bothschemes nodes need to communicate the degree of networkso energy consumption is approximately the same in both theschemes

Further in the analysis comparison is made on thebasis of SNR and 119875

119889 SNR varies from minus20 dB to 20 dB

Time-bandwidth factor that is 119879119882 = 5 and decisionthreshold 120582 is set so as to keep 119875

119891fixed having a value of

01 Simulation result in Figure 14 shows that CDCSS has abetter performance than existing consensus algorithm thatwas given in [20] It is also clear from Figure 14 that CDCSShas a better result for detection considering varying SNRcompared to existing consensus algorithm Even if 119875

119889is

expected to be kept above 099 then the existing algorithmrequired a higher average SNR as compared to CDCSS

6 Conclusion

Cooperative spectrum sensing is an efficient technique toimprove spectrum exploitation In this paper consensus-based distributive cooperative spectrum scheme is proposedwhich was applied to mobile nodes The local sensing inthis scheme was based on energy detection This is the mostwidely used local sensing technique as it does not require


minus20 minus15 minus10 minus5 0 5 10 15 20092

093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)

Figure 12 Comparison of the basis of SNR versus 119875119889with the

existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1

Proposed algorithmExisting algorithm

Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)


Figure 13 Comparison of the basis of 119875119891versus 119875


consensus algorithm (SNR = 10 dB 119879119882 = 5)

prior knowledge of network topology and is easy to imple-ment For cooperative sensing the proposed algorithm wasbased on a set of certain rules which includes prediction ruleand average value calculation for energy values of unlicensedusers Details of these rules are provided in Sections 3 and 4This techniquewas applied onmobile nodes becausemobilityof nodes results in reducing the fading effect thus makingsensing more efficient The goal of the proposed schemewas to provide proficient cooperative spectrum sensing tech-niqueThe results showed that our technique has the potentialto contribute effectively for efficient spectrum utilization Afuture direction of study is to consider sensing overheads


0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing consensus algorithm

and to provide optimal spectrumheterogeneousCR-MANETdesign

Acknowledgments

This work was supported by the Ministry of ScienceICT amp Future Planning (MSIP) Korea under the Con-vergence Information Technology Research Center (C-ITRC) support program (NIPA-2013-H0401-13-1003) super-vised by the National IT Industry Promotion Agency(NIPA) It was also supported by the Seoul RampBD Program(SS110012C0214831) and Special Disaster Emergency RampDProgram from National Emergency Management Agency(2012-NEMA10-002-01010001-2012)

References

[1] I F Akyildiz W Y Lee M C Vuran and S Mohanty ldquoNeXtgenerationdynamic spectrum accesscognitive radio wirelessnetworks a surveyrdquo Computer Networks vol 50 no 13 pp2127ndash2159 2006

[2] S Haykin ldquoCognitive radio brain-empowered wireless com-municationsrdquo IEEE Journal on Selected Areas in Communica-tions vol 23 no 2 pp 201ndash220 2005

[3] RWThomas DH Friend L ADaSilva andA BMacKenzieldquoCognitive networks adaptation and learning to achieve end-to-end performance objectivesrdquo IEEE Communications Maga-zine vol 44 no 12 pp 51ndash57 2006

[4] Q Zhao and B M Sadler ldquoA survey of dynamic spectrumaccessrdquo IEEE Signal Processing Magazine vol 24 no 3 pp 79ndash89 2007

[5] S S Ahmed A Raza H Asghar and G Ume ldquoImplementationof dynamic spectrum access using enhanced carrier SenseMultiple Access in Cognitive Radio Networksrdquo in Proceedingsof the 6th International Conference onWireless Communications


Networking and Mobile Computing (WiCOM rsquo10) pp 1ndash4Chengdu China September 2010

[6] D Cabric S MMishra and RW Brodersen ldquoImplementationissues in spectrum sensing for cognitive radiosrdquo in Proceedingsof the Conference Record of the 38th Asilomar Conference onSignals Systems and Computers pp 772ndash776 November 2004

[7] A Ghasemi and E S Sousa ldquoSpectrum sensing in cognitiveradio networks requirements challenges and design trade-offsrdquo IEEE Communications Magazine vol 46 no 4 pp 32ndash392008

[8] I F Akyildiz W Y Lee and K R Chowdhury ldquoCRAHNscognitive radio Ad Hoc networksrdquo Ad Hoc Networks vol 7 no5 pp 810ndash836 2009

[9] S Basagni M Conti S Giordano and I Stojmenovic MobileAd Hoc Networking John Wiley amp Sons New York NY USA2004

[10] I F AkyildizW Y Lee M C Vuran and S Mohanty ldquoA surveyon spectrum management in cognitive radio networksrdquo IEEECommunications Magazine vol 46 no 4 pp 40ndash48 2008

[11] E Axell G Leus E G Larsson andH V Poor ldquoSpectrum sens-ing for cognitive radio state-of-the-art and recent advancesrdquoIEEE Signal ProcessingMagazine vol 29 no 3 pp 101ndash116 2012

[12] D Bhargavi and C R Murthy ldquoPerformance comparison ofenergy matched-filter and cyclostationarity-based spectrumsensingrdquo in Proceedings of the IEEE 11th International Workshopon Signal Processing Advances in Wireless Communications(SPAWC rsquo10) pp 1ndash5 Marrakech Morocco June 2010

[13] D Cabric A Tkachenko andRW Brodersen ldquoSpectrum sens-ing measurements of pilot energy and collaborative detectionrdquoin Proceedings of the Military Communications Conference 2006(MILCOM rsquo06) pp 1ndash7 Washington DC USA October 2006

[14] T Yucek and H Arslan ldquoA survey of spectrum sensing algo-rithms for cognitive radio applicationsrdquo IEEE CommunicationsSurveys and Tutorials vol 11 no 1 pp 116ndash130 2009

[15] I F Akyildiz B F Lo and R Balakrishnan ldquoCooperative spec-trum sensing in cognitive radio networks a surveyrdquo PhysicalCommunication vol 4 no 1 pp 40ndash62 2011

[16] M Subhedar and G Birajdar ldquoSpectrum sensing techniquesin cognitive radio networks a surveyrdquo International Journal ofNext-Generation Networks vol 3 pp 37ndash51 2011

[17] W Ejaz N Ul Hassan and H S Kim ldquoDistributed cooperativespectrum sensing in cognitive radio for Ad Hoc networksrdquoComputer Communications vol 36 no 12 pp 1341ndash1349 2013

[18] Q H Wu G R Ding J L Wang X Q Li and Y ZHuang ldquoConsensus-based decentralized clustering for cooper-ative spectrum sensing in cognitive radio networksrdquo ChineseScience Bulletin vol 57 no 28-29 pp 3677ndash3683 2012

[19] F C Ribeiro M L de Campos and S Werner ldquoDistributedcooperative spectrum sensing with adaptive combiningrdquo inProceedings of the IEEE International Conference on AcousticsSpeech and Signal Processing (ICASSP rsquo12) pp 3557ndash3560 2012

[20] F R Yu M Huang and H Tang ldquoBiologically inspiredconsensus-based spectrum sensing in mobile Ad Hoc networkswith cognitive radiosrdquo IEEE Network vol 24 no 3 pp 26ndash302010

[21] F Bai and A Helmy ldquoA survey of mobility modelsrdquo inWirelessAd Hoc Networks vol 206 University of Southern CaliforniaLos Angeles Calif USA 2004

[22] H Urkowitz ldquoEnergy detection of unknown deterministicsignalsrdquo Proceedings of the IEEE vol 55 no 4 pp 523ndash531 1967

[23] M E J Newman ldquoThe structure and function of complexnetworksrdquo SIAM Review vol 45 no 2 pp 167ndash256 2003

[24] M Tahir H Mohamad N Ramli and Y Lee ldquoCooperativespectrum sensing using energy detection in mobile and staticenvironmentrdquo in Proceedings of the International Conference onComputer Networks and Communications Systems (CNCS rsquo12)2012

[25] T Camp J Boleng and V Davies ldquoA survey of mobility modelsfor Ad Hoc network researchrdquo Wireless Communications andMobile Computing vol 2 no 5 pp 483ndash502 2002

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics


Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks




of primary user

Cognitive radiobase station

Cognitive radionetwork access

Cognitive radio network(with infrastructure)

(a)


of primary user

Cognitive radioad hoc access

Cognitive radio

(without infrastructure)ad hoc network

(b)

Figure 1 Types of network (a) infrastructure based CRN and (b) CRAHN

topology reorganization [9] Movement patterns of mobilenodes (MNs) are defined using mobility models It is impor-tant to keep track of node movement because nodes inMANETs are free to move This also includes informationabout their speed location velocity and acceleration changeover time

In this paper a consensus-based distributed cooperativespectrum sensing scheme (CDCSS) in CR-MANET is pro-posed which is inspired by Novel biological mechanismsThese mechanisms have become an important phenomenonin handling intricate communication networks CDCSSworks on MNs in distributive network without using acentralized center to improve the sensing performance inCR-MANETsThemain contributions of this paper are as follows

(i) A consensus-based distributed spectrum sensing forCR-MANETs is proposed

(ii) Random walk mobility model (RWMM) is used inCRN for the movement pattern of nodes

(iii) Through the simulation the proposed scheme iscompared with existing consensus algorithm andEGC-rule scheme

The rest of the paper is as follows Section 2 explains workcarried out in fields of spectrum sensing and mobilitymodels Section 3 involves the methodology adopted todevelop a scheme for cooperative spectrum sensing amongMNs Section 4 explains implementation process Section 5involves analysis and Section 6 contains conclusion of ourwork

2 Background and Related Work

The rapid advancement in wireless applications has resultedin increased usage of unlicensed band causing a problem ofnonuniform spectrum usage [10] The CR cycle is divided

into four broad fields of research to cope with spectrum uti-lization challenges (1) the spectrum sensing that determineswhich portion of the spectrum is available (2) the spectrumdecision that picks the best vacant channel (3) the spectrumsharing that allows userrsquos coordinated access to channel and(4) the spectrum mobility that allows vacating the channelwhen a PU is detected

21 Spectrum Sensing Spectrum sensing techniques are clas-sified into noncooperative and cooperative spectrum sens-ing The three most common schemes for noncooperativetransmitter detection are energy detection matched filterdetection and cyclostationary detection [11ndash14]

Due to random changing in wireless environment it isdifficult for a single CRU to detect PU signal accuratelyTo cope with factors such as noise uncertainty shadingand fading cooperative spectrum sensing is introduced byresearchers In cooperative spectrum sensing CRUs cooper-ate and share their information about PU detection Thesemethods give more accurate results as uncertainty can beminimized [15]

In decentralized spectrum sensing there is no require-ment of infrastructure and fusion center Here the CRUsexchange their information with each other to cooperateThemost well-known decentralized spectrum sensing techniqueis gossiping algorithm because it performs sensing with asignificant low overhead Other decentralized techniquesinclude clustering schemes which are already known forsensor network architectures In these schemes CRUs formclusters which coordinate among themselves [16]

A gradient based distributed cooperative spectrum sens-ing method was proposed for CRAHNs [17] The gradientfield changeswith the energy sensed byCRU and the gradientis calculated based on the components which include energysensed by CRUs and received from neighbors The proposedscheme was evaluated on the basis of reliable sensing


Mobility models


Speed decayproblem

Gauss-Markovmodel

Randomwaypoint

model


Models with spatial

dependency

Models withtemporal


Random walkmodel

Randomdirection

















119909 (119905) = 119899 (119905) 119867

0

ℎ lowast 119904 (119905) + 119899 (119905) 1198671

(1)








Start

Implement CDCSS

Compute converged

PU absentPU present

Decision



value (xlowast)



119894 The output 119884


119884 =

1205942

21198851198670

12059422119885(2120574) 119867

1

(2)

where 12059422119885



119894



119894(0) = 119884





119909119899(119896 + 1) = 119909

119899(119896) + 120576

1015840sum119898120576119873

(119909119898(119896) minus 119909

119899(119896)) (3)





0 119909lowast le 120582(4)




Δ sim 1198731120572minus1





1 1198942 1198943 119894

119898 where




4 Simulation Setup


4

1

3

2

76

5

10

8

9




minus400

minus200

0

200

400

600







0 5 10 15 200

2

4

6

8

10

12

14

16

18


X (meters)

Y (m

eter

s)

Time (s) = 30








Δ sim 10125minus1

(6)




Converged

Start


Keep track of nodes



neighborrsquos info



Euclidian formula

Consensus achieved

PU absentPU present

Yes

No







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10



119889and 119875




119898and 119875




119875119898= 1 minus 119875

119889 (8)



119898


119891



119889







0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)




EGC rule (SNR = 10 dB 119879119882 = 5)








119889is


6 Conclusion




093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










Mobility models


Speed decayproblem

Gauss-Markovmodel

Randomwaypoint

model


Models with spatial

dependency

Models withtemporal


Random walkmodel

Randomdirection

















119909 (119905) = 119899 (119905) 119867

0

ℎ lowast 119904 (119905) + 119899 (119905) 1198671

(1)








Start

Implement CDCSS

Compute converged

PU absentPU present

Decision



value (xlowast)



119894 The output 119884


119884 =

1205942

21198851198670

12059422119885(2120574) 119867

1

(2)

where 12059422119885



119894



119894(0) = 119884





119909119899(119896 + 1) = 119909

119899(119896) + 120576

1015840sum119898120576119873

(119909119898(119896) minus 119909

119899(119896)) (3)





0 119909lowast le 120582(4)




Δ sim 1198731120572minus1





1 1198942 1198943 119894

119898 where




4 Simulation Setup


4

1

3

2

76

5

10

8

9




minus400

minus200

0

200

400

600







0 5 10 15 200

2

4

6

8

10

12

14

16

18


X (meters)

Y (m

eter

s)

Time (s) = 30








Δ sim 10125minus1

(6)




Converged

Start


Keep track of nodes



neighborrsquos info



Euclidian formula

Consensus achieved

PU absentPU present

Yes

No







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10



119889and 119875




119898and 119875




119875119898= 1 minus 119875

119889 (8)



119898


119891



119889







0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)




EGC rule (SNR = 10 dB 119879119882 = 5)








119889is


6 Conclusion




093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










Start

Implement CDCSS

Compute converged

PU absentPU present

Decision



value (xlowast)



119894 The output 119884


119884 =

1205942

21198851198670

12059422119885(2120574) 119867

1

(2)

where 12059422119885



119894



119894(0) = 119884





119909119899(119896 + 1) = 119909

119899(119896) + 120576

1015840sum119898120576119873

(119909119898(119896) minus 119909

119899(119896)) (3)





0 119909lowast le 120582(4)




Δ sim 1198731120572minus1





1 1198942 1198943 119894

119898 where




4 Simulation Setup


4

1

3

2

76

5

10

8

9




minus400

minus200

0

200

400

600







0 5 10 15 200

2

4

6

8

10

12

14

16

18


X (meters)

Y (m

eter

s)

Time (s) = 30








Δ sim 10125minus1

(6)




Converged

Start


Keep track of nodes



neighborrsquos info



Euclidian formula

Consensus achieved

PU absentPU present

Yes

No







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10



119889and 119875




119898and 119875




119875119898= 1 minus 119875

119889 (8)



119898


119891



119889







0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)




EGC rule (SNR = 10 dB 119879119882 = 5)








119889is


6 Conclusion




093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











1 1198942 1198943 119894

119898 where




4 Simulation Setup


4

1

3

2

76

5

10

8

9




minus400

minus200

0

200

400

600







0 5 10 15 200

2

4

6

8

10

12

14

16

18


X (meters)

Y (m

eter

s)

Time (s) = 30








Δ sim 10125minus1

(6)




Converged

Start


Keep track of nodes



neighborrsquos info



Euclidian formula

Consensus achieved

PU absentPU present

Yes

No







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10



119889and 119875




119898and 119875




119875119898= 1 minus 119875

119889 (8)



119898


119891



119889







0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)




EGC rule (SNR = 10 dB 119879119882 = 5)








119889is


6 Conclusion




093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










0 5 10 15 200

2

4

6

8

10

12

14

16

18


X (meters)

Y (m

eter

s)

Time (s) = 30








Δ sim 10125minus1

(6)




Converged

Start


Keep track of nodes



neighborrsquos info



Euclidian formula

Consensus achieved

PU absentPU present

Yes

No







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10



119889and 119875




119898and 119875




119875119898= 1 minus 119875

119889 (8)



119898


119891



119889







0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)




EGC rule (SNR = 10 dB 119879119882 = 5)








119889is


6 Conclusion




093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10







0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10



119889and 119875




119898and 119875




119875119898= 1 minus 119875

119889 (8)



119898


119891



119889







0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)




EGC rule (SNR = 10 dB 119879119882 = 5)








119889is


6 Conclusion




093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










0 5 10 15 20 25 305

10

15

20

25

30


Ener

gy le

vel o

f PU

(dB)

CR 1CR 2CR 3CR 4

CR 5CR 6CR 7CR 8

CR 9CR 10



119889and 119875




119898and 119875




119875119898= 1 minus 119875

119889 (8)



119898


119891



119889







0 10

01

02

02

03

04

04

05

06

06

07

08

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)




EGC rule (SNR = 10 dB 119879119882 = 5)








119889is


6 Conclusion




093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











093

094

095

096

097

098

099

1

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)


existing EGC rule

0 02 04 06 08 10

01

02

03

04

05

06

07

08

09

1


Prob

abili

ty o

f miss

ing

dete

ctio

n (P

m)







0985

099

0995

1

1005

Snr (dB)


Prob

abili

ty o

f det

ectio

n (P

d)




Acknowledgments


References






























RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation

































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









research article consensus-based distributive cooperative...

Documents