coalitional game theoretical approach for vanet clustering...

Research ArticleCoalitional Game Theoretical Approach for VANETClustering to Improve SNR

Selo Sulistyo Sahirul Alam and Ronald Adrian

Department of Electrical Engineering and Information Technology Faculty of Engineering Universitas Gadjah MadaYogyakarta Indonesia

Correspondence should be addressed to Selo Sulistyo selougmacid

Received 23 January 2019 Revised 16 April 2019 Accepted 5 May 2019 Published 17 July 2019

Academic Editor Youyun Xu

Copyright copy 2019 Selo Sulistyo et al )is is an open access article distributed under the Creative Commons AttributionLicense which permits unrestricted use distribution and reproduction in any medium provided the original work isproperly cited

Clustering is considered as the potential approach for network management in vehicular ad hoc network (VANET) )eperformance of clustering is often assessed based on the stability of the clusters Hence most of the clustering methods aim toestablish stable clusters However besides the stability of cluster good link quality must be provided especially when reliable andhigh-capacity transmission is demanded )erefore this paper proposes a clustering method based on coalitional game theorywith the purpose to improve the average of vehicle-to-vehicle (V2V) signal-to-noise ratio (SNR) and channel capacity whilemaintaining the stability of the cluster In the proposed method each vehicle attempts to form a cluster with other vehiclesaccording to coalition value To attain the purpose of clustering the value of coalition is formulated based on the V2V SNRconnection lifetime and speed difference between vehicles In fast-changing network topology the higher average of SNR can beachieved but the stability of cluster becomes hard to be maintained Based on the simulation results SNR improvement can beadjusted in order to balance with the cluster stability by setting the parameters in the proposed method accordingly Furthersimulation results show that the proposed method can obtain a higher average of V2V SNR and channel capacity than otherrelevant methods

1 Introduction

Vehicular ad hoc network (VANET) is a new form of mobilead hoc networks comprised of vehicles as the nodes Vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) com-munications are enabled by equipping vehicles with dedi-cated transceiver device VANET is provisioned tocontribute to intelligent transportation systems by providinga communication link among vehicles to support safety andtrafficmanagement purposes as well as infotainment [1])ehigher network capacity and reliable transmission link aredemanded to support these purposes However the envi-ronments of VANET bear the dynamics of the networks andraise several technical challenges )e wide range of vehiclespeed causes the fast-changing network topology and hencethe V2V connection is difficult to be maintained)e sparse-density nodes such as in rural road cause limited connectivity

[2] Meanwhile the high-density nodes such as in urban roadcause the burst of data traffic that leads to higher transmissioncollision and delay [3]

Node clustering with the hierarchical topology is con-sidered as the potential approach for network managementin VANET A cluster consists of several vehicle nodes withsome similar characteristics which form a communicationgroup A cluster commonly has one cluster head that acts asthe centre of the group and manages the cluster Other thanthe cluster head the ordinary vehicle nodes are called clustermembers Despite the effectiveness of clustering for networkmanagement in VANET clustering has some adversitiesmainly due to the dynamic environment In a survey [4] theperformance of clustering is generally assessed according tothe cluster stability It is good to have a stable clusterHowever it is insignificant unless the reliable and high-capacity communication link can be provided

HindawiJournal of Computer Networks and CommunicationsVolume 2019 Article ID 4573619 13 pageshttpsdoiorg10115520194573619

)erefore in this paper a clustering method for VANETis proposed with the aim to improve the signal-to-noise ratio(SNR) and channel capacity while maintaining the stabilityof the cluster )e proposed clustering method uses thedistributed approach based on coalitional game theory Ingeneral the coalitional game exhibits the model of in-teraction between a set of players who attempt to form agroup or coalition to reinforce their standing in the game [5])is concept matches with the investigated problem in thispaper where a vehicle interacts with the other nearby ve-hicles to establish connection and to form a cluster

As the implementation of the coalitional game inVANET each vehicle node is considered as a player whoattempts to form a coalition (cluster) with other vehiclenodes based on coalitional value )e value of coalition isdetermined by the revenue and cost Based on the purpose ofthe proposed clustering method the revenue is the linkquality or the SNR of V2V connection Meanwhile the costis the variable that decreases the revenue To obtain thehigher SNR the vehicles need to change the V2V connectionfrequently )us the stability of the cluster becomes harderto be maintained For this reason the cost in the coalitionalgame is aimed to maintain the stability of the cluster )ecost is formulated based on the connection lifetime and thespeed difference between vehicles)emain contributions ofthis paper and the advantages of the proposed clustering canbe described as follows

(1) A new clustering method for VANET based oncoalitional game theory is proposed with the aim toimprove SNR and channel capacity )e proposedclustering method also provides the flexibility onbalancing between high link quality and stability ofthe cluster by adjusting some parameters )us theimpact of parameters in the proposed method andthe guide for adjusting the parameters are alsopresented in this paper

(2) )e proposed clustering method uses a distributedapproach which is favorable due to the flexibilityespecially for VANETenvironment It is also expectedto reduce the workload of the network central ie thecluster head In this distributed clustering each ve-hicle only requires information about the link qualityand the speed of neighborhood vehicles )is in-formation is more conveniently provided than theexact position of vehicles )us some assumptionssuch as the use of GPS can be avoided)e use of GPSis still disputable due to the accuracy problem

(3) )is paper presents the simulation of clustering inVANETusing a real-world map and realistic vehiclemobility based on Simulation of Urban Mobility(SUMO) )e steps to build the simulation arepresented in detail Simulations of clustering areperformed using dedicated program built in MAT-LAB )us this work can be focused in the physicallayer of communication by specifically defining thesignal propagation model and the metrics for per-formance evaluation

2 Related Work

Some research studies related with clustering and gametheory in ad hoc networks especially VANET are presentedin this section Some technical differences between thoseresearch studies and this paper are also presented )etechnical differences can be outlined based on the criteria toform a cluster cluster control (centralized or distributed)cluster head selection procedures and goal of clustering

)e metric to form a cluster used by most of the clus-tering methods is the distance between vehicles such as in[6ndash12] )e calculation of distance is based on the vehiclersquosposition and thus it is assumed that each vehicle is equippedwith GPS However the usage of GPS is still disputable dueto the accuracy problem In this research the proposedmethod avoids the use of distance between vehicles as themetric to form a cluster Another metric used to form acluster is the vehicle speed where the adjacent vehicles withlower speed difference are grouped into the same clusterVehicle speed is used as one of the metrics to form a clusterin [6 11 13] and even the proposed method in this paperalso uses this metric )e reason is that the vehicle speed isthe dominant factor affecting the stability of the clusterAnother metric to form a cluster affecting the stability ofcluster is the connection lifetime such as that used in [6 8]and the proposed method in this paper )e other distinctmetrics to form a cluster are the node hierarchy [14] andneighborhood follow [15]

Most of the clustering methods including this researchperform clustering in a distributed manner Howeverclustering can also be performed using a centralized ap-proach with the aid of infrastructures such as in [12] wherethe clustering is performed by the cellular base station (BS))e clustering method in [13] is also centralized since thefuzzy system requires information from all vehicles to beprocessed by the centre of the networks Most of the clus-tering methods also assume the single-hop cluster structurefor the sake of simplicity )e multihop cluster structuredespite the complexity has either advantages or disadvan-tages A multihop cluster enables the flexibility and increasesthe coverage of the cluster However the delay of trans-mission may increase due to the multihops transmission)erefore the number of hops is limited to a certain numbertomaintain effectiveness)e proposedmethod in this paperassumes the multihop clustering

Cluster head selection in some clustering methods usesthe same metrics as in cluster formation while othermethods use more specific metrics for cluster head selection)e specific metric mostly used for cluster head selection isthe node connectivity such as in [13 14] Node connectivityis equal to the number of neighbor vehicles within thetransmission range )us the vehicle which has mostneighbor vehicles is considered as the best candidate of thecluster head )e other metrics to select the cluster head areas follows In [6] the cluster head is selected based on thegeographical position ie the vehicle that is located at thecentre of a cluster )e speed of vehicle closest to the averagespeed of cluster is selected as the cluster head such as in

2 Journal of Computer Networks and Communications

[12 13] In [7] the cluster head is selected based on thelowest vehicle ID while in [8] it is selected based on thenode weight which is calculated based on the history of linkquality In [10] the cluster head is the first vehicle that formsa cluster Afterwards the concept of secondary cluster head(SCH) is introduced When the cluster head leaves thecluster the SCH can switch position to be the cluster head)e SCH is selected based on the minimum speed differenceand nearest distance to the primary cluster head In [11] thecluster head is selected based on eligibility which is definedby a fuzzy logic controller In [16] the cluster head is selectedbased on the connection with minimum overhead In [12]besides based on the average cluster speed the cluster head isselected from the vehicle which has the best channel qualityto establish a connection with the base station In [15] thecluster head is selected based on neighborhood follow thatis the vehicle followed by other vehicles In this proposedclustering method cluster head selection is performed basedon the electability ie the vehicle which is selected by mostof the cluster members )e further description of elect-ability is presented in the next section

)e goal of most clustering methods is to form the stablecluster [6ndash9 11 13ndash15] )e clusters with high stability cansupport the performance of routing protocol However thestability of the cluster becomes less significant if the high linkquality cannot be provided especially when the reliable orhigh-capacity transmission is demanded )erefore someclustering methods aim different goals such as high datapacket delivery ratio (DPDR) and low delay [16] reducingsignal overhead and improving communication quality byaggregating V2I traffic transmission to the cluster head [12]Meanwhile the goals of clustering in this proposed methodare to improve the average of V2V SNR and channel ca-pacity )is goal actually can be achieved by establishingV2V connection with the higher SNR However in theenvironment of VANET which is very dynamic this cancause the vehicles change the connection frequently and thusruining the stability of the cluster )erefore a special ap-proach is needed to deal with this complex problem

)is research proposes the coalitional game theory as theapproach for clustering in VANET Game theory is knownas the potential approach for wireless communication andexpected to give contribution in VANET development In[17] game theory is proposed for ad dissemination inVANET In [18] game theory is used to observe the co-operation in VANET according to the network conditionsIn [19] game theory is proposed to jointly control the powerlevel and channel selection in VANET )e coalitional gametheory was also implemented in other wireless communi-cation systems such as cellular and WiMAX In [20] coa-litional game was utilized in multiple-input multiple-output(MIMO) cellular network to form the network and tomitigate the interference at once Each established com-munication link has interference relationship with otherlinks Hence the establishment of communication link wasmodeled as coalitional game with an aim to mitigate theinterference )e proposed algorithm converged to the Nashequilibrium to enable network formation with lower in-terference and higher data rate In [21] coalitional game was

utilized to arrange the merge process in cooperative com-munications with aim to attain energy efficiency )e pro-posed merge process consisted of three stages ietransmission request merge and cooperative transmissionWith the coalitional game the merge process was directed toattain energy efficiency of each node in the group)ereforethe total power (transmission and processing power) incooperative transmission can be lower than directtransmission

In this paper the coalitional game theory is selected dueto flexibility to formulate the rule in forming a coalitionMoreover the concepts of coalition and clustering arecompatible )e coalitional game theory has been used inclustering for mobile ad hoc networks (MANETs) [22]However besides the difference of system model betweenMANET and VANET there is a key difference betweenresearch in [22] and this research that is the formulation ofthe coalitional rule Another difference is that research in[22] does not have a cluster head in cluster structurewhereas this research has

Research in this paper utilizes the trace data from real-worldmap)e usage of trace data has become an alternativefor simulating VANETenvironment and it has an advantagedue to the proximity with the real condition Another relatedwork in VANET using the trace data was presented in [23]In this paper the trace data from Google maps is used tosimulate vehicle mobility while forming the clustersMeanwhile the trace data in [23] were used for simulation ofroute selection which considers the data rate in route se-lection Since the aim of route selection is to optimizethroughput of the network using TV white space the usageof trace data and Google spectrum data set is appropriate inthis case

3 Proposed Approach

)is paper proposes a multihop clustering method Inmultihop clustering the stability of connection betweenvehicles is the key factor which determines the stability of thecluster )erefore the establishment of V2V connectionmust be arranged appropriately to attain the longer con-nection duration and the better link quality Furthermorethe formation and maintenance of the cluster rely on theestablished V2V connection In this proposed methodcoalitional game theory is employed to improve themechanism of V2V connection establishment due to therelevance of the cluster formation model and coalitionformation model )e vehicles aim to form cluster byestablishing V2V connection with other vehicles which cangive mutually good link quality and connection lifetime)isconcept is similar to the concept of coalition formationgame where each player selects other players to form groupwhich can give mutual benefit to the group members)erefore the coalitional game is considered as the potentialapproach to be implemented in VANET clustering

)e problem of clustering investigated in this paper canbe described as an optimization problem as follows )ereare two objectives of the optimization namely to improvethe SNR of V2V connection and to establish the stable

Journal of Computer Networks and Communications 3

clusters However each of those objectives is a trade-off forthe other objective )is is due to the dynamic environmentin VANET thus the fluctuation of SNR is high and thevehicles tend to change connection frequently in order toobtain the higher value of SNR Meanwhile the frequentchange of connection disrupts the stability of cluster)erefore this is a complex optimization problem Howeveras mentioned above the coalitional game can be the ap-propriate approach to model this optimization problemHere the variables related with SNR value such as trans-mission distance and power are included in the revenue ofcoalition Meanwhile the variables influencing the stabilityof V2V connection as well as the stability of cluster such asthe speed difference and connection lifetime are included inthe cost of coalition Finally the vehicles will decide whetherto maintain current connection or change the connectionbased on the value of coalition which is the difference be-tween revenue and cost of coalition )erefore the objectiveof optimization is directed to maximize the value of co-alition Furthermore the details about the proposed clus-tering method including the implementation of coalitionalgame approach in the clustering process are described asfollows )e role of coalitional game approach in theclustering process can be seen in Figure 1

31 Vehicle-to-Vehicle Establishment Coalitional gametheory approach is used for establishment of vehicle-to-vehicle (V2V) connection )erefore the proposed clus-tering approach is called coalitional game clustering (CGC)In the coalitional game players select any other player forcooperating or forming a coalition based on the value theyexpect to get)e value of the coalition has two elements ierevenue and cost High value is implied by the high revenueand low cost )e implementation of this approach in V2Vconnection establishment is described as follows

311 Revenue )e main purpose of this clustering ap-proach is to obtain a higher transmission link quality Sincethis research is done in the physical layer of communicationthe transmission link quality observed is the signal-to-noiseratio (SNR) and the channel capacity obtained from theestablishment of a V2V connection Channel capacity isproportional to the SNR based on the ShannonndashHartleytheorem )erefore the proposed revenue function is basedon SNR as given by

uj Sij1113872 1113873 Γij (1)

where uj(Sj) is the revenue of vehicle j for using strategy Sij

and Γij is the SNR of downlink transmission from vehicle i tovehicle j )e value of SNR for the revenue is presented inwatt instead of decibel since the minimum revenue is ex-pected to be equal to zero instead of negative )e termstrategy in game theory means a set of actions that can beselected by a player In this case the strategy is the action toselect any other vehicles than itself to form a coalition(establish a V2V connection) )e list of strategies that canbe chosen by vehicle j can be denoted as

Sj S1j S

2j S

ij S

Nj (2)

where i and N are respectively the index of transmittervehicle and the total number of other vehicles within thetransmission range of vehicle j It can be noted that thenumber of strategies depends on the traffic around vehicle jIn a normal or sparse traffic the number of strategies can stillbe handled by the vehicle However in high-density trafficthe number of strategies can be numerous and difficult tohandle )erefore it is suggested to limit the number ofstrategy especially in a dense traffic condition eg by listingthe vehicles within a certain radius from vehicle j or bydefining the minimum threshold of SNR so that the vehicleswill be included in the list

312 Cost Cost is considered as the amount or price thatreduces the revenue In this proposed clustering approachthe cost is formulated to increase the lifetime of a V2Vconnection In other words it is aimed to maintain thestability of clusters In this CGC the cost consists of twoelements as follows

(i) Connection lifetime (tc) represents the duration ofV2V connection that has been established )e du-ration increases as long as the vehicle is connected withthe same transmitter vehicle Otherwise if the vehiclechanges connection to another vehicle then tc is resetIt can be understood that tc directly represents thelifetime of a V2V connection )erefore in cost for-mulation when the value of tc is lowest (eg afterreset) then the cost must be highest Hence the ve-hicle hardly changes the connection to another vehicleif it has just established a connection with a vehicleAfterwards as tc increases the cost will decrease Costelement 1 (c1) based on tc is defined as follows

V2V establishment andmaintenance using

coalitional gametheoretical approach

Cluster formation andmaintenance based on

established V2Vconnections

Cluster head selection

Beaconinterval

Figure 1 Procedure in the proposed clustering method


c1

0 tc ge δc

δc minus tc

δc tc lt δc

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(3)

where δc is the threshold for tc In other words δc isthe minimum time required by a vehicle to changeV2V connection to another vehicle with c1 equal tozero

(ii) Speed difference (Δv) between two vehicles is con-sidered as one of the factors affecting the lifetime of aV2V connection Vehicles with the higher differenceof speed will terminate the connection sooner sincethey leave their transmission range On the contraryvehicles with the lower difference of speed canmaintain their connection for the longest time Incost formulation the absolute value of Δv is pro-portional to the cost element 2 (c2) as given by

c2 max 1|Δv|

Δvmax1113888 11138891113890 1113891 (4)

where Δvmax is the defined maximum speed difference andthus c2 will be more than 1 if |Δv| is higher than ΔvmaxHowever the value of c2 is limited to 1)e purpose of usingabsolute operator is due to the vectorization of speed wherevehicles that move in the opposite direction have oppositesign of speed For example a vehicle moves from south tonorth with speed v1 thus another vehicle moving from northto south has speed minusv2 )erefore two vehicles moving indirection opposite to each other have a higher absolute valueof speed difference )is is relevant to the fact that twovehicles moving in different directions can only maintain theV2V connection in a short time )e higher cost can avoidthe establishment of a V2V connection between vehicleswhich are moving in the opposite direction as long as thereis at least another vehicle moving in the same directionwithin their transmission range

)e two cost elements in (3) and (4) are formulated sothat the values are normalized ie between 0 and 1 Af-terwards the combination of those cost elements to definethe final cost for CGC is given by

cj Sij1113872 1113873 max c1 c21113858 1113859 (5)

)e reason for using max operator is to allow each el-ement to stand out depending on the condition For ex-ample when the speed difference is above the threshold (c2 ismaximum) cj will have a maximum value regardless of thevalue of c1 In another case when some neighbor vehiclesmove with nearly same speed (thus the speed difference isvery low) then c1 can stand out to determine which vehicleto connect Certainly the connection with the same vehicleis tried to be maintained than looking for a connection withanother vehicle However the decision to use max operatoris based on intuition and the comparisons with the usage of

other operations such as mean or weighted value are notpresented in this paper

313 Value )e value obtained by vehicle j(vj) forselecting strategy Si

j is defined as the difference between therevenue obtained by forming a coalition (establishing aconnection) with vehicle i uj(Si

j) and the cost of the coalitionwith vehicle i cj(Si

j) as given by

vj Sij1113872 1113873 uj S

ij1113872 1113873 1minus cj S

ij1113872 11138731113872 1113873 (6)

Based on the above formulation vj will have the max-imum value (equal with the revenue uj) if the cost cj is zeroOn the contrary vj is minimum (zero) if the cost cj is equalto 1 regardless of the revenue In CGC vehicle j selects acooperation with vehicle i based on the value vj(Si

j) Vehiclei is selected to establish V2V connection if vehicle i gives thebest value to vehicle j However there is a constraint inselecting a vehicle to form V2V connection ie thetransmission range Hence the vehicles out of the trans-mission range cannot be selected due to the constraint )iscan be done by setting the cost of connection with vehiclesout of transmission range to the maximum Although theconnection with a vehicle which is far away has very a lowSNR or even zero setting the cost to maximum is aimed toassure that the vehicle will not be selected

32 Cluster Formation Once the vehicles select the pairedvehicle to establish V2V connections the clusters areformed Since V2V connection establishment is arranged ina distributed manner the formed clusters could be dynamic)e topology of clusters could be multihops and overlappingclusters are possible Figure 2 shows the illustration of clusterformation using CGC Two clusters are called overlap eachother when there is at least one member of a cluster withinthe coverage of cluster head from the other cluster Forexample in Figure 2 the head of the cluster I is vehicle 5Vehicles 7 and 8 belong to cluster II however they arewithin the coverage of vehicle 5 )us clusters I and IIoverlap each other )e overlapping clusters are possiblesince each vehicle is allowed to select another vehicle in-dependently to establish a V2V connection Although thevehicles aim to establish the connection with high a SNRdue to the coalitional game rule the connection with thehighest SNR is probably not selected For example vehicle 8has five neighborhood vehicles as in Figure 2 Let us assumethat V2V connection between vehicles 8 and 9 has thehighest instantaneous SNR However due to the cost factorsin the coalitional game such as connection lifetime andspeed difference vehicle 8 chooses vehicle 7 to establish V2Vconnection instead of vehicle 9 )us vehicles 7 8 and 10form their own cluster instead of joining cluster I In theworst case CGC allows a cluster to be formed although it hasonly twomembers like cluster III It seems better if vehicle 12connects to vehicle 13 However due to the coalitional valuevehicle 12 forms its own cluster with vehicle 11

One of the cost elements in CGC is the speed difference(Δv) )is cost prevents two vehicles with high speed


difference to establish a V2V connection In the calculationof Δv the speed of vehicles is vectorized )us Δv canbecome higher if the two vehicles have a different directionFrom Figure 2 it can be noticed that vehicles 11 to 16 formdifferent clusters than vehicles 1 to 10 )at is because ve-hicles 11 to 16 move in direction opposite to that of vehicles1 to 10 However cluster formation by vehicles with oppositedirection is still possible ie if the vehicles move very slowlyso that the speed difference is below the threshold (Δvmax) Itis possible in some situations such as in traffic jam junctionwith the traffic light or in the tollgate area

33 Cluster Head Selection In CGC each vehicle selectsanother nearby vehicle to establish a V2V connection In thiscase a vehicle can be selected by more than one vehicle)erefore in CGC the electability of vehicles is used as one ofthe criteria to select the head of a cluster )e vehicle with thehighest electability among the cluster members can becomethe cluster head (CH) Another criterion to select the clusterhead is the speed of the vehicle When there are two or morevehicles with the same electability then the vehicle with theslower speed becomes the cluster head Another alternativecan be used ie by selecting the vehicle with smaller speeddifference to the cluster However it requires more com-putation and adds the transmission overhead In spite of thisit is not a big problem since the main factor to select thecluster head is the electability and the additional parameter isused only when there are two or more equal cluster headcandidates Due to the dynamic environment of VANET theelectability of a vehicle can also fluctuate )erefore a pro-cedure is added in cluster head selection to increase thelifetime of a cluster head A vehicle can maintain the state ascluster head as long as there are at least two vehicles connected(has electability at least 2) even though there is anothervehicle in the cluster with higher electability Algorithm 1describes the process of cluster head selection in CGC

4 Simulation

41 System Model

411 VANET Environment In this system model the en-vironment of VANET is in toll road with the longest route

has length 9 kilometers)e road is traced from the real mapie city toll at Semarang Indonesia )e longest route startsfrom (minus6954628 110451622) at the north to (minus7026183110432788) at the south In the middle of the main routearound (minus6973131 110450006) there is a tollgate wherevehicles from either north or south direction must passthrough the gate and thus the flow of vehicles is sloweraround this point )e road has 3 lanes for each directionmeanwhile the tollgate has 4 lanes for each direction Asshown in Figure 3 other than the main road (A to B) thereare two points (C and D) where vehicles can enter or leavethe road )us there are 6 possible routes in this map A-BB-A A-C C-B B-D and D-A

)e traced coordinates of the roads from Google Maps[24] are in decimal degrees system )erefore a conversion isdone for the convenience in the distance calculation Co-ordinates in decimal degrees can be converted to Cartesian bymultiplying with 111320 )is multiplier is a particular valuefor coordinate at the equator (within 23degNS)

412 Vehicle Mobility and Distribution )e mobility ofvehicles is simulated using an open source microscopic andcontinuous road traffic simulator namely Simulation ofUrban Mobility (SUMO) [25] SUMO provides some con-veniences in simulating the mobility of the vehicle byallowing the user to plot the road manually to define thetypes and the flow of vehicles and to record the result ofsimulations )e movement of vehicles is defined by Kraussrsquocar-following model which is influenced by the surroundingvehicles )e decisions of lane-changing and overtaking are

1

3

4

78

2

5

6

109

12

13

15

16

14

11

I

II

III

IV

Figure 2 Structure of clusters using CGC

Input Ej (vehicle electability) vj (vehicle speed)CH_found⟵ 0for j 1 2 Nj (number of cluster members)if jtminus1 CHampEj ge 2 thenj becomes CHCH_found⟵ 1break

end ifend forif CH_found 0 thenfor j 1 2 Nj

CH_candidacy⟵ 1for k 1 2 Nk kne j Nk Nj minus 1

if Ej ltEk thenCH_candidacy⟵ 0

elseif Ej Ek amp vj gt vk thenCH_candidacy⟵ 0

end ifend forif CH_candidacy 1 then

j becomes CHbreak

end ifend for

end if

ALGORITHM 1 Cluster head selection


defined by Krajzewiczrsquos model [26] In this VANETsimulation the results of SUMO are saved as an XML fileconsisting of information about vehicles position andspeed along with the identity of vehicles In this work thesimulation of vehicle mobility using SUMO uses thefollowing settings )ere are three types of vehicles carcoach and trailer )e properties of the vehicle are shownin Table 1 )e distribution of vehicles based on the typeand the route are presented in Figures 4 and 5respectively

Apart from the maximum speed and speed deviation theflow of vehicle is also influenced by the lane speed limits)espeed factor as in Table 1 defines the expected multiplicatorfor the lane speed limits In the road network like theaforementioned there is a tollgate around (minus6973131110450006) Normally the flow of vehicles is slower aroundthe tollgate area )erefore a road edge with length 100mcentered at the tollgate coordinate is defined with lane speedlimit 4ms Other than this tollgate area the road edges havelane speed limit 40ms Figure 6 shows the screenshot ofSUMO taken from the graphical user interface (GUI) FromFigure 6 it can be noticed that the density of the vehicle isdifferent between the area around the tollgate and otherareas on the map )e density of the vehicle at the tollgate ishigher and the flow is slower as also shown by the red coloron the road in Figure 3

413 Signal Propagation Model )e propagation of thesignal from the transmitter vehicle to receiver vehicle isdefined by path loss and fading Path loss defines the at-tenuation of the signal proportional to the distance betweenthe transmitter and receiver Fading meanwhile defines the

variation either gain or attenuation of the signal due to thesurrounding environment Based on the path loss charac-terization for vehicular communication in [27] commu-nication link gain (G) determined by path loss and fading inthe highway environment is given by

NA

C D

B

Figure 3 Map of the road for VANET simulation

Table 1 Vehicle speed attributes

Vehicle Maximum speed (ms) Speed factor Speed deviationCar 40 075 02Coach 30 0375 01Trailer 20 05 03

1600

360

88

CarCoachTrailer

Unit in vph (vehicles per hour)

Figure 4 Distribution of vehicle based on the type of vehicle

25

25

125

125125

125

A

B

C D

Figure 5 Distribution of vehicle based on the route


G minus 3792 + 164 log10 d( 1113857 + Sσ (7)

where d represents the distance between the transmitter andreceiver in meters Fading is represented by Sσ ie randomvariable which has a normal distribution zero mean andstandard deviation of 446 )e link gain in (7) determinesthe strength of the signal received by the receiver vehicle asgiven by

S Gp (8)

where p is the transmission power level at the transmitter Inthis work all vehicles are assumed to have a fixed trans-mission power level ie 20 dBm or equal with 100mW

In wireless communication especially when the reusedspectrum is used the interference becomes a generic issueHowever this work has not dealt with the channel alloca-tion )erefore the interference is not modeled specificallyFor the sake of simplicity the interference is represented byusing additional noise with Gaussian distribution mean20 dB standard deviation of 5 dB In addition to in-terference the communication channel has noise namelywhite noise that has equal intensity at different frequencies)e noises are summed and then used to calculate the SNR(Γ) of V2V connection For the known SNR value in wattsand the bandwidth of channel (B) in hertz the capacity of thechannel (C) is calculated based on ShannonndashHartley theo-rem as given by

C B log2(1 + Γ) (9)

414 Simulations Setup )e simulation of vehicle mobilityand VANET clustering is performed separately )e vehiclemobility is simulated in SUMO and the result of the

simulation is an extensible markup language (XML) file withinformation about vehicle ID speed coordinate positionand lane position at every second Meanwhile the simula-tions of VANET clustering are performed in MATLAB Adedicated program is compiled to simulate the proposedCGC inMATLAB As the input of simulation the output filefrom SUMO is converted from XML to MATLAB data file(mat) In addition to the vehicle position file randomnumbers to represent fading (Sσ) are generated and saved asthe input of simulation )is is to ensure that the sameenvironment is used in VANET clustering simulation)erefore the simulations of clustering using different al-gorithms can be performed and the results can be fairlyevaluated Along with the clustering simulation using CGCthe simulations using lowest ID (LID) [7] density-basedclustering (DBC) [8] and mobility-based dynamic cluster-ing (MDC) [6] are performed for the purpose of comparisonand evaluation LID is the widely known clustering algo-rithm for mobile ad hoc network (MANET) as well asVANET Meanwhile DBC is the clustering method that hasthe most similarity with the proposed CGC in terms ofselection metrics for clustering Moreover DBC is the lastclustering method that uses SNR as one of the selectionmetrics according to a survey [28] )e number of clusteringmethods which use SNR as the selection metrics is verylimited thus one more comparison is done with a recentclustering method namely MDC Although MDC does notuse SNR as the metric to form the cluster it has very goodperformance in terms of cluster stability )erefore MDC isalso used for comparison in this research

)e simulation in SUMO runs for 800 seconds How-ever the output file of simulation only records the in-formation during tsim 501 to tsim 710 )e early time ofsimulation is considered as warming up time where thevehicles start to enter the road Within tsim 501minus710 somevehicles leave the road and some new vehicles enter the road)us there are around 200 vehicles in total at 1 tsim )eoutput file from SUMO is then used as the input of clusteringsimulations in MATLAB Based on the input data theduration of clustering simulation is 210 seconds Howeverthe results of the simulation are recorded for 200 secondsie from tsim 11minus210 )e reason is that the DBC algo-rithm needs some warming up times ie the first 10 secondsof clustering simulation

42 Performance Metrics )e performance of clusteringmethods are observed and evaluated using the followingmetrics

(i) Number of clusters denotes the number of clustersformed at every unit of time during simulationsUsually the fewer number of clusters is more de-sired since the fewer number of clusters is expectedto increase the transmission efficiency such as re-ducing the overhead and communication betweenclusters

(ii) Average size of the cluster represents the averagenumber of members in a cluster)e higher numberof cluster members is more desired since in the

Figure 6 Screenshot of vehicle mobility simulation in SUMO


worst case a cluster can only have one or twomembers However sometimes the number ofcluster members should be limited For example ifthe number of channels is limited then the fewernumber of members can reduce the channel con-gestion Regardless of this opinion it is assumedthat the higher number of cluster members is better

(iii) Cluster head change rate denotes the number ofcluster head changes per second Cluster headchange can be in the existing clusters or the newlyformed clusters )e fewer number of this metric isdesirable since it implies a more stable cluster

(iv) Clustering coverage defines the percentage of ve-hicles that join any clusters Normally 100 cov-erage is better However if the clustering methodallows a single vehicle to form a cluster then thecoverage can be 100

(v) Average V2V SNR denotes the average of SNR fromall V2V connections established at the time

(vi) Average channel capacity denotes the averagechannel capacity based on the V2V SNR Channelcapacity denotes the maximum bit rate that thechannel can support

43 Results and Discussion )e results of simulations areorganized into two subsections )e first subsection presentsthe results of simulations by alternating the value of pa-rameters in CGC namely connection time threshold (δc)and maximum speed difference (Δvmax) Afterwards itpresents the results of the simulation using the proposedCGC and three other clustering methods namely LIDDBC and MDC

431 Impact of CGC Parameters Setting Figures 7ndash12display the results of clustering simulation by alternatingδc (in seconds) and Δvmax (in meters per second) As shownin Figure 7 the number of clusters increases when δc isincreased It is because that if δc is increased the vehiclesbecome more difficult to change connection and thus theywill form clusters with the vehicles with lower speed dif-ference It can be said that due to the increased δc thevehicles tend to form more clusters but with the higherstability On the contrary the number of clusters decreaseswhen Δvmax is increased )e Δvmax is analogous with thetolerance of speed difference If the tolerance is low thevehicles will be separated into more clusters On the con-trary if the tolerance is high the vehicles have more chanceto unite into the same clusters Hence fewer clusters areformed )e average of cluster size as shown in Figure 8 isinversely proportional to the number of clusters Hence thehigher the number of clusters the smaller the average clustersize and vice versa As mentioned above if δc is increasedthe vehicles tend to form more clusters but with the higherstability One of the metrics to evaluate the stability of thecluster is the cluster head change rate )e more stable thecluster the lower the change rate of the cluster head InFigure 9 it can be noticed that if δc is increased the cluster

head change rate decreases)e cluster head change rate alsodecreases if Δvmax is increased )is is because Δvmax isrelevant to the cluster stability )e lower value of Δvmaxforces the vehicles to maintain a connection with the vehiclesthat have nearly similar speed )e coverage of CGC isshown in Figure 10 It can be noticed that either Δvmax or δcdo not have any significant impact toward the coverage ofclustering )e coverage of CGC is mainly influenced by thetransmission range of vehicles As long as there are at leasttwo vehicles within the transmission range of each otherthose vehicles can form a cluster However some vehicles onthe low-density area may not have another vehicle withintheir transmission range and hence they cannot form acluster Since those vehicles do not belong to any cluster thecoverage of clustering is not 100

Figure 11 shows the average of V2V SNR When δc isincreased the average of V2V SNR decreases)is is becausethat when δc is lower the chance that the vehicles change

44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters

Δvmax = 10Δvmax = 20Δvmax = 30

Figure 7 Number of clusters based on the variation of δc andΔvmax

38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize


Figure 8 Average cluster size based on the variation of δc andΔvmax


connection to obtain a higher SNR is higher since the costelement of connection lifetime is lower )e lower cost atearly connection lifetime enables the vehicles to changeconnection more frequently On the contrary the higher δcimplies that connection change at early connection lifetimeis higher and hence the vehicles choose to stick the con-nection with the current vehicle although the connectionwith other vehicles probably has the higher SNRMeanwhilethe average of V2V SNR increases if the Δvmax is increased)is is due to the flexibility of vehicles to select V2V con-nection is higher Since the vehicles have more options toestablish a V2V connection the vehicles have a higherchance to select a V2V connection with a higher SNR )echannel capacity is proportional with the SNR )us theaverage channel capacity as shown in Figure 12 and therelation with the variation of δc and Δvmax can be equallydescribed as in the explanation of average V2V SNR

432 Comparison of Clustering Performance )e results ofsimulations using LID DBC MDC and the proposed CGCare presented here For the balance between connection linkquality and cluster stability the parameters of CGC aredefined as follows δc 30 s andΔvmax 20ms Meanwhileto obtain the higher link quality DBC uses a minimum SNR40 dB and minimum group membership lifetime (GML)3 seconds to be the member of a cluster For the simulationusing the MDC method the following parameter values areused )e clustering distance (Dt) or the maximum radius ofcluster from the cluster head is 250m )e beacon interval(BI) and merge interval (MI) are 05 s and 5 s respectively)e maximum duration of temporary cluster head (TCHt)and unclustered node (TUN) are 5 s and 3 s respectively

Figures 13(a) and 13(b) respectively display the averagenumber of clusters and the average of cluster size from thesimulation using LID DBC MDC and CGC It can be seen

14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)


Figure 9 Cluster head change rate based on the variation of δc andΔvmax

0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)


Figure 10 Clustering coverage based on the variation of δc andΔvmax

Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65


Figure 11 Average V2V SNR based on the variation of δc andΔvmax

Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)


Figure 12 Average channel capacity based on the variation of δcand Δvmax


that LID has the lowest number of clusters and the biggestsize of the cluster )is is because LID allows any vehicles tojoin a cluster provided the vehicles are within the trans-mission range of the cluster head Even the vehicles can joina cluster from the opposite direction of the road Since theselection criteria are not strict more vehicles can join acluster and hence the size of the cluster becomes bigger)erefore only a few clusters are formed Meanwhile DBCMDC and CGC have more stringent criteria in forming acluster )us their average cluster size is smaller comparedto LID Moreover DBC has the lowest average of cluster sizedue to the GML minimum requirement To become amember of a cluster a vehicle has to maintain link qualityabove the threshold until the minimum duration of GML isreached Because of this a vehicle cannot directly join acluster and probably remain unclustered for some moments

Figure 14(a) shows the cluster head change rate fromsimulations using the three methods LID and MDC havelower cluster head change rate )is is because LID selectsthe cluster head based on the lowest ID of the vehicles )ecluster head will remain as long as no new vehicle with lowerID joins the cluster )is method surely has an advantage intoll road environment where only a few possible routesexist Moreover almost no vehicle stops at the edge of theroad)erefore the cluster head can remain the same for thelonger duration DBC has the highest cluster head changerate since the cluster head selection is based on weightwhich is basically the average of link quality However thismethod is vulnerable especially in a highly dynamic envi-ronment Unfortunately in toll road environment the ve-hicles have a wide range of speed unlike in urban road)erefore the communication network is highly dynamicand the link quality can change frequently CGC performscluster head selection based on the electability Eachmemberof the cluster elects the candidate of the cluster head basedon the highest value of coalition In this case the dynamicenvironment may also affect the performance of thismethod However in CGC a vehicle is allowed to maintainthe status as cluster head as long as it is elected by at least twovehicles As the result the cluster head change rate can bereduced In Figure 14(a) it can be noted that MDC has thelowest cluster head change rate as expected sinceMDC has astrong point in terms of cluster stability )e duration of

cluster head can be maintained for a longer duration inMDC and hence the cluster head change rate is very low InMDC a cluster head can retain the status as cluster head aslong as there is at least one member although there isanother event that a cluster head must relinquish the statusie during cluster merging

)e coverage of clustering is shown in Figure 14(b) LIDhas 100 coverage since the single vehicle can be counted asa cluster Meanwhile DBC and CGC do not count the singlevehicle as a cluster However in CGC the cluster can beformed by at least two vehicles located at their transmissionrange of each other)erefore CGC has great coverage sinceit only excludes the single vehicles MDC also has highcluster coverage and it is comparable with LID and CGC)is is because in MDC cluster formation with only twovehicles is also allowed )e coverage of DBC is the lowest)is is because DBC prevents a vehicle to join any clustersdirectly As results more vehicles are in unclustered statusand hence the coverage of DBC decreases

)e average of V2V SNR and channel capacity arepresented by Figures 15(a) and 15(b) respectively DBC andCGC certainly have the better average of V2V SNR than LIDsince they consider SNR as one of the criteria in forming acluster MDC also has a higher V2V SNR average than LIDalthough SNR is not included in criteria to form a cluster)is is because MDC limits the cluster size within a certainradius )erefore the higher value of SNR can still be ob-tained However CGC has the highest average of V2V SNRamong the four methods )is is because CGC selects thebest link quality to establish connection although indirectlythrough the concept of the coalitional game (revenue costand value) Even the higher V2V SNR can still be reached byselecting the best connection However the stability of theconnection cannot be maintained for a longer duration Inthis case coalitional game rule performs the work to balancebetween the higher link quality and the more stable con-nection )e capacity of the channel is proportional with theSNR according to (9) However in Figure 15(b) the averagechannel capacity using DBC is the lowest among the threemethods )is is due to the averaging where the channelcapacities from all vehicles are summed and then divided bythe number of vehicles Meanwhile in DBC there are somevehicles that are in an unclustered state Since those vehicles

LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)

Figure 13 Cluster structure based on the number of clusters and cluster size (a) Average number of clusters and (b) average size of cluster


cannot establish a V2V connection their channel capacity iscounted as zero

5 Conclusion

A distributed clustering method for VANET based oncoalitional game theory namely CGC is proposed in thispaper Each vehicle attempts to form a cluster with othervehicles according to the concept of coalition value Since thepurpose of clustering is to improve the V2V SNR whilemaintaining the stability of the cluster the coalition value isformulated based on this purpose )e value of coalition isdefined by the revenue (V2V SNR) and the cost (connectionlifetime and speed difference) In fast-changing networktopology the higher average of SNR can be obtained but thestability of the cluster becomes hard to be maintained Basedon the simulation results SNR improvement can be adjustedin order to balance with the cluster stability by setting theparameters in CGC accordingly Further simulation resultsshow that CGC can obtain a higher average of V2V SNR andchannel capacity than the other relevant methods

Data Availability

)is study is based on the datasets generated by the SUMOsoftware upon vehicle mobility model we had constructed

which are available from the corresponding author uponrequest

Disclosure

)ere are no other persons who satisfied the criteria forauthorship but are not listed )e authors further confirmthat the order of authors listed in the manuscript has beenapproved by all of them

Conflicts of Interest

)e authors declare that they have no conflicts of interest

Authorsrsquo Contributions

SS contributed mainly to the defining problem formulationmobility model and system modeling as well as discussionparts SA contributed for implementing the simulation andgraphical production while RA contributed for preparing themobility models dataset using the SUMO software )eauthors also confirm that the manuscript has been read andapproved by all named authors

Acknowledgments

)is work was supported by the Ministry of ResearchTechnology and Higher Education Indonesia under the

0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)

Figure 15 Link quality based on V2V SNR and channel capacity (a) Average V2V SNR and (b) average channel capacity

0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)

Figure 14 Cluster head change rate (a) and clustering coverage (b)


PUPT grant no 751UN1ndashPIIILTDIT-LIT2016 for theyears 2016ndash2018 and was conducted in the Department ofElectrical Engineering and Information Technology Facultyof Engineering Universitas Gadjah Mada YogyakartaIndonesia

References

[1] H Hartenstein and K P Laberteaux ldquoA tutorial survey onvehicular ad hoc networksrdquo IEEE Communications Magazinevol 46 no 6 pp 164ndash171 2008

[2] N Wisitpongphan F Bai P Mudalige V Sadekar andO Tonguz ldquoRouting in sparse vehicular ad hoc wirelessnetworksrdquo IEEE Journal on Selected Areas in Communica-tions vol 25 no 8 pp 1538ndash1556 2007

[3] N Wisitpongphan O K Tonguz J S Parikh P MudaligeF Bai and V Sadekar ldquoBroadcast storm mitigation tech-niques in vehicular ad hoc networksrdquo IEEE Wireless Com-munications vol 14 no 6 pp 84ndash94 2007

[4] R S Bali N Kumar and J J P C Rodrigues ldquoClustering invehicular ad hoc networks taxonomy challenges and solu-tionsrdquo Vehicular Communications vol 1 no 3 pp 134ndash1522014

[5] W Saad Z Han M Debbah A Hjorungnes and T BasarldquoCoalitional game theory for communication networksrdquo IEEESignal Processing Magazine vol 26 no 5 pp 77ndash97 2009

[6] M Ren L Khoukhi H Labiod J Zhang and V Veque ldquoAmobility-based scheme for dynamic clustering in vehicularad-hoc networks (VANETs)rdquo Vehicular Communicationsvol 9 pp 233ndash241 2017

[7] C R Lin and M Gerla ldquoAdaptive clustering for mobilewireless networksrdquo IEEE Journal on Selected Areas in Com-munications vol 15 no 7 pp 1265ndash1275 1997

[8] S Kuklinski and G Wolny ldquoDensity based clustering algo-rithm for VANETsrdquo in Proceedings of the 5th InternationalConference on Testbeds and Research Infrastructures for theDevelopment of Networks amp Communities and Workshopspp 1ndash6 Washington DC USA April 2009

[9] W Li A Tizghadam and A Leon-garcia ldquoRobust clustering forconnected vehicles using local network criticalityrdquo in Proceedingsof the 2012 IEEE International Conference on Communications(ICC) pp 7157ndash7161 Ottawa Canada June 2012

[10] A Ahizoune and A Hafid ldquoA new stability based clusteringalgorithm (SBCA) for VANETsrdquo in Proceedings of the 37thAnnual IEEE Conference on Local Computer Net-worksmdashWorkshops (LCN Work) pp 843ndash847 ClearwaterBeach FL USA October 2012

[11] I Tal and G Muntean ldquoUser-oriented fuzzy logic-basedclustering scheme for vehicular ad-hoc networksrdquo in Pro-ceedings of the 2013 IEEE 77th Vehicular Technology Con-ference (VTC Spring) pp 2ndash6 Dresden Germany June 2013

[12] X Duan Y Liu and X Wang ldquoSDN enabled 5G-VANETadaptive vehicle clustering and beamformed transmission foraggregated trafficrdquo IEEE Communications Magazine vol 55no 7 pp 120ndash127 2017

[13] B Jinila and Komathy ldquoRough set based fuzzy scheme forclustering and cluster head selection in VANETrdquo Elektronika IrElektrotechnika vol 21 no 1 pp 54ndash59 2015

[14] E Dror C Avin and Z Lotker ldquoFast randomized algo-rithm for hierarchical clustering in vehicular ad-hoc net-worksrdquo in Proceedings of the 2011 10th IFIP AnnualMediterranean Ad Hoc Networking Workshop pp 1ndash8Sicily Italy June 2011

[15] Y Chen M Fang S Shi W Guo and X Zheng ldquoDistributedmulti-hop clustering algorithm for VANETs based onneighborhood followrdquo EURASIP Journal on Wireless Com-munications and Networking vol 2015 no 1 pp 1ndash12 2015

[16] S Ucar S C Ergen and O Ozkasap ldquoMultihop-cluster-basedIEEE 80211p and LTE hybrid architecture for VANET safetymessage disseminationrdquo IEEE Transactions on Vehicular Tech-nology vol 65 no 4 pp 2621ndash2636 2016

[17] X Tang H Sun L Sun C Tan and G Xu ldquoGame theoreticalapproach for ad dissemination in cluster based VANETsrdquo inProceedings of the 2013 IEEE International Conference on SignalProcessing Communication and Computing (ICSPCC 2013)pp 1ndash6 Kunming China August 2013

[18] S Shivshankar and A Jamalipour ldquoAn evolutionary gametheory-based approach to cooperation in VANETs under dif-ferent network conditionsrdquo IEEE Transactions on VehicularTechnology vol 64 no 5 pp 2015ndash2022 2015

[19] S Sulistyo and S Alam ldquoDistributed channel and power levelselection in VANET based on SINR using game modelrdquo In-ternational Journal of Communication Networks and InformationSecurity (IJCNIS) vol 9 no 3 pp 432ndash438 2017

[20] T Zhou Y Chen and K J R Liu ldquoNetwork formationgames in cooperative MIMO interference systemsrdquo IEEETransactions on Wireless Communications vol 13 no 2pp 1140ndash1152 2014

[21] H-Q Lai Y Chen and K J R Liu ldquoEnergy efficient co-operative communications using coalition formation gamesrdquoComputer Networks vol 58 pp 228ndash238 2014

[22] R Massin C J Le Martret and P Ciblat ldquoA coalition for-mation game for distributed node clustering in mobile ad hocnetworksrdquo IEEE Transactions on Wireless Communicationsvol 16 no 6 pp 3940ndash3952 2017

[23] C Jiang Y Chen and K J R Liu ldquoData-driven optimalthroughput analysis for route selection in cognitive ve-hicular networksrdquo IEEE Journal on Selected Areas inCommunications vol 32 no 11 pp 2149ndash2162 2014

[24] G Maps Jl Tol Tanjungmas-Srondol httpswwwgooglecommapsplaceJl+Tol+Tanjungmas-Srondol+Sawah+Besar+Gayamsari+Kota+Semarang+Jawa+Tengah+50163-699753161104328032135zdata4m53m41s0x2e70f3355f88d1430x1c3760724e0a98908m23d-697301974d1104500701

[25] D Krajzewicz J Erdmann M Behrisch and L BiekerldquoRecent development and applications of SUMOmdashsimulationof urban mobilityrdquo International Journal on Advances inSystems and Measurements vol 5 no 3 pp 128ndash138 2012

[26] D Krajzewicz M Bonert and P Wagner ldquo)e open sourcetraffic simulation package SUMO trafficmanagement projectsrdquoin Proceedings of the Robotics and Automation 371ndash380Orlando FL USA May 2006 httpselibdlrde467401RoboCup2006_dkrajzew_SUMOpdf

[27] H Fernandez L Rubio V M Rodrigo-Penarrocha andJ Reig ldquoPath loss characterization for vehicular communi-cations at 700 MHz and 59 GHz under LOS and NLOSconditionsrdquo IEEE Antennas and Wireless Propagation Lettersvol 13 pp 931ndash934 2014

[28] C Cooper D Franklin M Ros F Safaei and M AbolhasanldquoA comparative survey of VANET clustering techniquesrdquoIEEE Communications Surveys amp Tutorials vol 19 no 1pp 657ndash681 2017


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

)erefore in this paper a clustering method for VANETis proposed with the aim to improve the signal-to-noise ratio(SNR) and channel capacity while maintaining the stabilityof the cluster )e proposed clustering method uses thedistributed approach based on coalitional game theory Ingeneral the coalitional game exhibits the model of in-teraction between a set of players who attempt to form agroup or coalition to reinforce their standing in the game [5])is concept matches with the investigated problem in thispaper where a vehicle interacts with the other nearby ve-hicles to establish connection and to form a cluster

As the implementation of the coalitional game inVANET each vehicle node is considered as a player whoattempts to form a coalition (cluster) with other vehiclenodes based on coalitional value )e value of coalition isdetermined by the revenue and cost Based on the purpose ofthe proposed clustering method the revenue is the linkquality or the SNR of V2V connection Meanwhile the costis the variable that decreases the revenue To obtain thehigher SNR the vehicles need to change the V2V connectionfrequently )us the stability of the cluster becomes harderto be maintained For this reason the cost in the coalitionalgame is aimed to maintain the stability of the cluster )ecost is formulated based on the connection lifetime and thespeed difference between vehicles)emain contributions ofthis paper and the advantages of the proposed clustering canbe described as follows

(1) A new clustering method for VANET based oncoalitional game theory is proposed with the aim toimprove SNR and channel capacity )e proposedclustering method also provides the flexibility onbalancing between high link quality and stability ofthe cluster by adjusting some parameters )us theimpact of parameters in the proposed method andthe guide for adjusting the parameters are alsopresented in this paper

(2) )e proposed clustering method uses a distributedapproach which is favorable due to the flexibilityespecially for VANETenvironment It is also expectedto reduce the workload of the network central ie thecluster head In this distributed clustering each ve-hicle only requires information about the link qualityand the speed of neighborhood vehicles )is in-formation is more conveniently provided than theexact position of vehicles )us some assumptionssuch as the use of GPS can be avoided)e use of GPSis still disputable due to the accuracy problem

(3) )is paper presents the simulation of clustering inVANETusing a real-world map and realistic vehiclemobility based on Simulation of Urban Mobility(SUMO) )e steps to build the simulation arepresented in detail Simulations of clustering areperformed using dedicated program built in MAT-LAB )us this work can be focused in the physicallayer of communication by specifically defining thesignal propagation model and the metrics for per-formance evaluation

2 Related Work

Some research studies related with clustering and gametheory in ad hoc networks especially VANET are presentedin this section Some technical differences between thoseresearch studies and this paper are also presented )etechnical differences can be outlined based on the criteria toform a cluster cluster control (centralized or distributed)cluster head selection procedures and goal of clustering

)e metric to form a cluster used by most of the clus-tering methods is the distance between vehicles such as in[6ndash12] )e calculation of distance is based on the vehiclersquosposition and thus it is assumed that each vehicle is equippedwith GPS However the usage of GPS is still disputable dueto the accuracy problem In this research the proposedmethod avoids the use of distance between vehicles as themetric to form a cluster Another metric used to form acluster is the vehicle speed where the adjacent vehicles withlower speed difference are grouped into the same clusterVehicle speed is used as one of the metrics to form a clusterin [6 11 13] and even the proposed method in this paperalso uses this metric )e reason is that the vehicle speed isthe dominant factor affecting the stability of the clusterAnother metric to form a cluster affecting the stability ofcluster is the connection lifetime such as that used in [6 8]and the proposed method in this paper )e other distinctmetrics to form a cluster are the node hierarchy [14] andneighborhood follow [15]

Most of the clustering methods including this researchperform clustering in a distributed manner Howeverclustering can also be performed using a centralized ap-proach with the aid of infrastructures such as in [12] wherethe clustering is performed by the cellular base station (BS))e clustering method in [13] is also centralized since thefuzzy system requires information from all vehicles to beprocessed by the centre of the networks Most of the clus-tering methods also assume the single-hop cluster structurefor the sake of simplicity )e multihop cluster structuredespite the complexity has either advantages or disadvan-tages A multihop cluster enables the flexibility and increasesthe coverage of the cluster However the delay of trans-mission may increase due to the multihops transmission)erefore the number of hops is limited to a certain numbertomaintain effectiveness)e proposedmethod in this paperassumes the multihop clustering

Cluster head selection in some clustering methods usesthe same metrics as in cluster formation while othermethods use more specific metrics for cluster head selection)e specific metric mostly used for cluster head selection isthe node connectivity such as in [13 14] Node connectivityis equal to the number of neighbor vehicles within thetransmission range )us the vehicle which has mostneighbor vehicles is considered as the best candidate of thecluster head )e other metrics to select the cluster head areas follows In [6] the cluster head is selected based on thegeographical position ie the vehicle that is located at thecentre of a cluster )e speed of vehicle closest to the averagespeed of cluster is selected as the cluster head such as in








3 Proposed Approach







uj Sij1113872 1113873 Γij (1)



Sj S1j S

2j S

ij S

Nj (2)









Beaconinterval



c1

0 tc ge δc

δc minus tc

δc tc lt δc

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(3)



c2 max 1|Δv|

Δvmax1113888 11138891113890 1113891 (4)



cj Sij1113872 1113873 max c1 c21113858 1113859 (5)






j) as given by

vj Sij1113872 1113873 uj S

ij1113872 1113873 1minus cj S

ij1113872 11138731113872 1113873 (6)








4 Simulation

41 System Model





1

3

4

78

2

5

6

109

12

13

15

16

14

11

I

II

III

IV








j becomes CHbreak

end ifend for

end if







NA

C D

B




1600

360

88

CarCoachTrailer



25

25

125

125125

125

A

B

C D



G minus 3792 + 164 log10 d( 1113857 + Sσ (7)


S Gp (8)



C B log2(1 + Γ) (9)


















44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia








3 Proposed Approach







uj Sij1113872 1113873 Γij (1)



Sj S1j S

2j S

ij S

Nj (2)









Beaconinterval



c1

0 tc ge δc

δc minus tc

δc tc lt δc

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(3)



c2 max 1|Δv|

Δvmax1113888 11138891113890 1113891 (4)



cj Sij1113872 1113873 max c1 c21113858 1113859 (5)






j) as given by

vj Sij1113872 1113873 uj S

ij1113872 1113873 1minus cj S

ij1113872 11138731113872 1113873 (6)








4 Simulation

41 System Model





1

3

4

78

2

5

6

109

12

13

15

16

14

11

I

II

III

IV








j becomes CHbreak

end ifend for

end if







NA

C D

B




1600

360

88

CarCoachTrailer



25

25

125

125125

125

A

B

C D



G minus 3792 + 164 log10 d( 1113857 + Sσ (7)


S Gp (8)



C B log2(1 + Γ) (9)


















44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





uj Sij1113872 1113873 Γij (1)



Sj S1j S

2j S

ij S

Nj (2)









Beaconinterval



c1

0 tc ge δc

δc minus tc

δc tc lt δc

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(3)



c2 max 1|Δv|

Δvmax1113888 11138891113890 1113891 (4)



cj Sij1113872 1113873 max c1 c21113858 1113859 (5)






j) as given by

vj Sij1113872 1113873 uj S

ij1113872 1113873 1minus cj S

ij1113872 11138731113872 1113873 (6)








4 Simulation

41 System Model





1

3

4

78

2

5

6

109

12

13

15

16

14

11

I

II

III

IV








j becomes CHbreak

end ifend for

end if







NA

C D

B




1600

360

88

CarCoachTrailer



25

25

125

125125

125

A

B

C D



G minus 3792 + 164 log10 d( 1113857 + Sσ (7)


S Gp (8)



C B log2(1 + Γ) (9)


















44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


c1

0 tc ge δc

δc minus tc

δc tc lt δc

⎧⎪⎪⎪⎪⎨

⎪⎪⎪⎪⎩

(3)



c2 max 1|Δv|

Δvmax1113888 11138891113890 1113891 (4)



cj Sij1113872 1113873 max c1 c21113858 1113859 (5)






j) as given by

vj Sij1113872 1113873 uj S

ij1113872 1113873 1minus cj S

ij1113872 11138731113872 1113873 (6)








4 Simulation

41 System Model





1

3

4

78

2

5

6

109

12

13

15

16

14

11

I

II

III

IV








j becomes CHbreak

end ifend for

end if







NA

C D

B




1600

360

88

CarCoachTrailer



25

25

125

125125

125

A

B

C D



G minus 3792 + 164 log10 d( 1113857 + Sσ (7)


S Gp (8)



C B log2(1 + Γ) (9)


















44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia




4 Simulation

41 System Model





1

3

4

78

2

5

6

109

12

13

15

16

14

11

I

II

III

IV








j becomes CHbreak

end ifend for

end if







NA

C D

B




1600

360

88

CarCoachTrailer



25

25

125

125125

125

A

B

C D



G minus 3792 + 164 log10 d( 1113857 + Sσ (7)


S Gp (8)



C B log2(1 + Γ) (9)


















44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia






NA

C D

B




1600

360

88

CarCoachTrailer



25

25

125

125125

125

A

B

C D



G minus 3792 + 164 log10 d( 1113857 + Sσ (7)


S Gp (8)



C B log2(1 + Γ) (9)


















44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia


G minus 3792 + 164 log10 d( 1113857 + Sσ (7)


S Gp (8)



C B log2(1 + Γ) (9)


















44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia











44

46

48

50

52

54

0 5 10 15 20 25 30δc (seconds)

Num

ber o

f clu

sters



38

39

4

41

42

43

44

45

0 5 10 15 20 25 30δc (seconds)

Ave

rage

clus

ter s

ize







14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia





14

16

18

20

22

24

26

0 5 10 15 20 25 30δc (seconds)

CH ch

ange

rate

(per

seco

nd)



0 5 10 15 20 25 3090

92

94

96

98

100

Clus

terin

g co

vera

ge (

)

δc (seconds)



Ave

rage

V2V

SN

R (d

B)

5 10 15 20 25 300δc (seconds)

59

60

61

62

63

64

65



Ave

rage

chan

nel c

apac

ity (M

bps)

32

34

36

38

0 10 15 20 25 305δc (seconds)









LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia







LID DBC MDC CGC

Ave

rage

num

ber o

f clu

sters

0

10

20

30

40

50

60

(a)

LID DBC MDC CGC0

2

4

6

8

10

12

Ave

rage

clus

ter s

ize

(b)




5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



5 Conclusion


Data Availability



Disclosure






Acknowledgments


0

10

20

30

40

50

60

70

LID DBC MDC CGC

Ave

rage

V2V

SN

R (d

B)

(a)

0

05

1

15

2

25

3

35

4

LID DBC MDC CGC

Ave

rage

chan

nel c

apac

ity (M

bps)

(b)


0

10

20

30

40

50

LID DBC MDC CGC

CH ch

ange

rate

(s

ec)

(a)

0

20

40

60

80

100

120

LID DBC MDC CGC

Clus

terin

g co

vera

ge (

)

(b)




References
































RoboticsJournal of




VLSI Design



Shock and Vibration







Journal of



Volume 2018



Volume 2018


Journal of




SensorsJournal of



RotatingMachinery





Propagation






Hindawi


Advances in

Multimedia



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


coalitional game theoretical approach for vanet clustering...

Documents