A Review on Clustering in VANET: Algorithms, Phases, and Comparisons

Abstract

A Vehicular Ad hoc Network (VANET) is a kind of mobile ad hoc network (MANET), where the nodes in VANET are vehicles. VANET is the main component for the development of the Intelligent Transportation System (ITS). VANET has a high dynamic topology due to continuous stopping and movement with different speeds of vehicles. These characteristics lead to untrustworthy information transmission in VANET. To enhance connection reliability and network scalability, clustering in VANET is introduced. It is a method of clustering network nodes into groups known as clusters. In this work, we provide a review of the most clustering algorithms presented between 1999 and 2020. Different clustering algorithms from the aspects of Cluster Head (CH) selection metrics, cluster formation according to hop distance, and cluster maintenance are explored. Also, different performance parameters used to evaluate the clustering approaches are summarized. Then, our proposed approach is summarized and compared with the most usable algorithms in literature to show its supremacy.

Full text

A Review on Clustering in VANET: Algorithms,
Phases, and Comparisons
Mays Kareem Jabbar1,2 , Hafedh Trabelsi1
1CES_Lab, ENIS , Sfax University, Tunisia
2Faculty of Engineering, University of Misan, Al Amarah City, Misan Province, 62001, Iraq
m_mays85@uomisan.edu.iq, hafedh.trabelsi@enis.tn
Abstract— A Vehicular Ad hoc Network (VANET) is a kind of
mobile ad hoc network (MANET), where the nodes in VANET
are vehicles. VANET is the main component for the
development of the Intelligent Transportation System (ITS).
VANET has a high dynamic topology due to continuous
stopping and movement with different speeds of vehicles. These
characteristics lead to untrustworthy information transmission
in VANET. To enhance connection reliability and network
scalability, clustering in VANET is introduced. It is a method
of clustering network nodes into groups known as clusters. In
this work, we provide a review of the most clustering
algorithms presented between 1999 and 2020. Different
clustering algorithms from the aspects of Cluster Head (CH)
selection metrics, cluster formation according to hop distance,
and cluster maintenance are explored. Also, different
performance parameters used to evaluate the clustering
approaches are summarized. Then, our proposed approach is
summarized and compared with the most usable algorithms in
literature to show its supremacy.
Keywords— VANET, Clustering Algorithms, Cluster Head,
Performance Parameters.
I. INTRODUCTION
Recently, transportation and wireless systems are
integrate to obtain optimum efficiency and vehicle
safety by exchanging information between vehicles on
the road. The work on this network’s type has obtained
special fame among researchers from different world
viewpoints. Studies have shown that attention to these
networks is necessary because these networks enhance
traffic efficiency, vehicle safety, and minimize
transport effects on the environment [1]. VANET is a
subset of Mobile Ad Hoc Network (MANET).
MANET is a self-configuring network of mobile nodes
connected by wireless communication. The major
advantage of VANET is that it allows vehicles to
communicate and share information, which helps to
reduce traffic congestion and improve road safety.
When the mobile nodes in MANETs are substituted for
cars and the network begins to follow fixed routes,
such as roadways, the network becomes a VANET.
VANET's average speed and mobility of nodes are
relatively high, resulting in a quick change in network
structure; these are its distinguishing characteristics [2].
VANET has two critical elements; Roadside units
(RSUs) and On-Board Units (OBUs). RSUs are placed
alongsidе the road and record all vehicle data, which is
then relayed to neighboring OBUs. The transmission of
information jobs in OBUs or cars is completely under
the control of RSUs. Furthermore, OBUs are devices
that are installed in dynamic vehiclеs to make
information to be share between automobiles and RSUs
easier. A Dedicated Short-Range Communication
(DSRC) is proposed for transmission information and
communication among vehicles with a transition range
of 100 to 1000m [3]. The DSRC system functions
similarly to Wi-Fi. The Federal Communication
Commission (FCC) of the United States has assigned a
highеr spеctrum band with a range of 75MHz. The
75MHz spectrum band is divided into sеven sub-
channels of 10 MHz each. The control channel (CCH)
broadcasts the safety message and control information
among the scattered seven channels. Service channels
(SCHs) are the remaining channels, there are
responsible for data transfer [4]. VANET
communication ensures the greatest available
communication between vehicles, vehicles, and RSUs.
There are three types of communication in the VANET
known as; 1) Vehicle to Vehicle (V2V), Inside their
radio ranges, vehicles equippеd with OBUs can
communicate directly with one another as shown in
Figure.1, 3) Vehicle to Infrastructure (V2I)
communication, as well as infrastructure placement
along the roadside and numerous apps that can improve
the quality of service given by infrastructures to cars. 3)
Vehicle to X communication (V2X) includes V2V+V2I
communications between vеhicles and communications
between vehicles and other terminals, such as RSUs
[3], [5], [6].
Fig. 1. Vehicular Communications types [1].
978-1-6654-7108-4/22/$31.00 ©2022 IEEE
2022 19th International Multi-Conference on Systems, Signals & Devices (SSD'22)
444
2022 19th International Multi-Conference on Systems, Signals & Devices (SSD) | 978-1-6654-7108-4/22/$31.00 ©2022 IEEE | DOI: 10.1109/SSD54932.2022.9955850
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.

The major VANET functions are accurate vehicle
information communication, traffic management, and
vehicle safety. Because of the fast speed of vehicles,
VANET topology is dynamic and can be predicted
using GPS. A wireless communication facility is
installed in the cars to provide ad hoc access in the
VANET. The dynamic nature of the VANET causes
scalability concerns. Data distribution in VANET
architecture necessitates an effective and efficient
routing technique. The environment of VANET is
evolving in order to guarantee vehicle security and
safety [7] [8]. Cluster stability is critical for the
VANET's scalability and reliability because it ensures
minimal intra- and inter-cluster communication,
reducing the overhead associated with these concerns
[4]. The latest clustering methods are explained in
order to achieve the best communication information in
a VANET. We focus on VANET's intelligent clustering
algorithms as well. As a result, we study several
clustering algorithms. The following is the study's
structure: the clustering in VANET is introduced in
Section II, which includes algorithms, history, and
process. The performance evaluation parameters for
several clustering algorithms are described in Section
III. Then our proposed approach is summarized in
Section IV. The survey's conclusion is reported in
Section V.
II. CLUSTERING IN VANET
Clustering is a popular VANET technique that
provides a simple and effective way to simplify and
optimize network activities and services. It offers
significantly increased performance in a range of
applications when compared to the standard flat
construction. Clustering is a method of arranging
network nodes into small groups known as clusters.
Vehicles in the area are typically divided into clusters
based on a variety of key factors and data. The vehicles
in the cluster are referred to as:
1. Cluster Head (CH) – it is the node that serves as
the cluster's coordinator or leader. The CH is chosen
based on a variety of factors, and its primary function is
to allow cluster members to communicate and share
information with other members and CHs.
2. Cluster Members (CMs) – The CMs are the
cluster's remaining nodes. These nodes communicate
with one another by broadcasting messages.
3. Gateway Node (GW) – it is the node that
facilitates communication with RSU and is not required
to be present in every cluster. [9]
The VANET's cluster-based communication structure
is depicted in Figure 2. A CH is chosen by defining the
network's parameters, and the remaining vehicles are
treated as CMs. The CH is in charge of all internal
cluster communication. Internal communication within
a cluster is divided into two types: intra-cluster
communication and inter-cluster communication [4].
Fig. 2. The Cluster Structure [10]
A. Clustering Algorithms History of VANETs
Clustering algorithms for VANETs started to be
developed in the early 1990s, and they have grown in
popularity in subsequent years. In Table I, 15 clustering
algorithms are highlighted, which have been presented
between 1999 and 2020.
Coinciding with the increasing popularity of VANETs,
some MANETs clustering algorithms were introduced
to fit the specific characteristics of vehicular
communications. Most of VANET clustering
algorithms like DMAC in [11] were derived from the
previous MANETs. Also, we can see PMC and
VMaSC have the highest mean citations. Also, it is
easy to note that the clustering was highlighted by
researchers significantly after 2010.
Also, we can see although some of the clustering
algorithms have recently been presented, they have a
high mean of citations such as DHC, and CBSC.
TABLE I. CLUSTERING ALGORITHMS CITATIONS.
Ref
Year
Algorithm
Citation
Mean
[11]
1999
DMAC
482
22.8
[12]
2008
PPC
116
8.9
[13]
2009
RMAC
75
6.25
[14]
2010
ALM
127
11.5
[15]
2011
VWCA
141
14.1
[16]
2012
DBC
79
7.2
[17]
2012
TB
119
13.2
[18]
2012
AMACAD
74
8.2
[19]
2015
AWCP
54
9
[20]
2016
VMaSC
255
51
[21]
2018
PMC
180
60
[22]
2018
LRCA
29
9.6
[23]
2019
EWCA
16
8
[24]
2019
CBSC
35
17.5
[25]
2019
DHC
36
18
[26]
2020
CCA-IoV
1
0.5
These algorithms are discussed in detail in the
following parts of this work.
B. VANET Clustering Process
The most necessary part of the process of
VANET communication is cluster formation.
445
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.

The cluster generation phase includеs the cluster
formation process and the CH sеlection process. Nodes
send advertising messages to choose the principal CH
and CM, and then frequent packets of data are sent
between them during this phase. A few approaches may
be implemented between the advertising message
delivery and CH selection in order to produce a stable
cluster.
The cluster maintenance phase is responsible for some
methods like; cluster splitting, stable cluster merging,
re-clustering, and selection of secondary CH.
Only a few academics had looked into these phases
independently in the literature. This section explains
the techniques and criteria used in each clustering step,
such as CH selection, cluster creation based on hop
count, and cluster management.
1) Cluster Head Selection
CH stability has a significant impact on the
network's robustness and scalability. The stable CH
ensures that communication between clusters and
within clusters is ensured. The nodes’ interaction with
other CH and their neighbors influences the selection
of CH. A reliable vehicle can only be a CH to enhance
VANET stability. For picking the CH, the researchers
looked at a variety of metrics criteria, such as vehicle
position, relative speed, received signal strength, link
lifetime, and direction. For selecting the CH, several
clustering algorithms use a combination of various
metrics instead of a single parameter, like RMAC,
DBC, AWCP, CBSC, EWCA, and DHC. Table II lists
some of the techniques and metrics that were employed
in CH selection.
TABLE II. Metrics Used for Selecting the CH
Ref
Algorithm
CH Selection Metrics
[11]
DMAC
ID degree
[12]
PPC
relative velocity, travel time, ID
[13]
RMAC
Velocity, CH status, size, distance
[14]
ALM
Priorities relevant each vehicle
[15]
VWCA
Degree, direction, distrust level, velocity
[16]
DBC
Velocity, distance, SNR
[17]
TB
Relative velocity, distance,
[18]
AMACAD
Destination, relative velocity, distance
[19]
AWCP
Direction, highway ID, speed, position
[20]
VMaSC
Speed
[21]
PMC
Neighbors, position, speed, link lifetime
[22]
LRCA
link reliability
[23]
EWCA
Position, speed
[24]
CBSC
Relative Speed, position
[25]
DHC
Link lifetime, relative speed, signal strength
[26]
CCA-IoV
Average relative speed, link stability,
average relative acceleration, and distance
2) Cluster Formation
Clusters are formed based on several predetermined
parameters, like the maximum number of members,
cluster radius, and transmission range. The cluster
formation criteria vary in each algorithm. A hop
distance between the CH and its members can be used
to model a cluster topology in VANETs. Accordingly,
As shown in Figure 3, there are two types of
algorithms: single-hop and multi-hop algorithms.
Fig. 3. Clustering Model
The single-hop algorithms are the techniques for
forming clusters with only one hop between each
member and its CH. This ensures that each node has a
direct connection to the CH [27]. Depending on
the CH transmission range or the limiting cluster
radius, many clustering methods build direct single-
hop clusters. Some single-hop clustering techniques
are:
A single hop clustering algorithm was presented in
[13] called Robust Mobility Adaptive Clustering
(RMAC) for VANETs. RMAC utilizes a new vehicle
priority method for adaptively and easily identifying
single-hop neighbors and optimal CHs electing.
Moreover, RMAC presents a zone of interest as a new
concept that helps vehicles to maintain their neighbor
table.
In [15], vehicular clustering based on weighted
clustering (VWCA) was proposed. It is a one-hop
clustering approach that enhances security, connection,
and stability. The adaptive transmission range
algorithm (AART), which is based on recognized
short-range communication standards, can improve
connectivity. Based on the density of the cars, the
AART assists in dynamically extending the
transmission range from 100 to 1000 meters.
In [18], an Adaptable mobility-aware clustering
algorithm based on destination positions
(AMACAD) was proposed as another one-hop
clustering approach for VANETs.
The authors in [22] proposed a Link Reliability-based
Clustering Algorithm (LRCA) to ensure that data
transmission in the urban VANET is both secure and
efficient. To choose a set of vehicles with steady
neighboring, LLT-based neighbor sampling strategy is
applied. The routing technique in this proposed is
designed to serve VANET infotainment services that
do not have tight delay requirements.
In [23], an Enhanced Weight-based Clustering
Algorithm (EWCA) was proposed. This strategy
decreases the development of unstable clusters and
improves clustering stability in an emergency message
delivery scenario.
446
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.

A Center-Based Clustering algorithm (CBSC) was
proposed in [24] to assist self-organized VANETs in
forming stable clusters and reducing the frequency of
vehicle status changes on highways. The results
showed that the proposed obtains a low packet loss
rate and high stability.
In [25], Under various settings and scenarios, a
resilient Double-Head Clustering (DHC) algorithm for
VANET was presented with an emphasis on increasing
cluster stability and minimizing the number of clusters
in the network. In terms of efficiency and cluster
stability, the suggested approach outperformed
previous algorithms.
Single-hop clustering methods allow CHs to
communicate more reliably within their cluster and
coordinate more effectively. This type of cluster has a
small coverage area, which results in a large number of
clusters and a high maintenance overhead.
In conclusion, single-hop methods provide minimal
latency and strong cluster stability, but cluster
coverage needs to be improved further.
Clusters are created using multi-hop distance, which
means that each node is at least two hops away from
its CH. In this part, we'll look at some multi-hop
clustering algorithms.
In [12], the authors proposed a multi-hop clustering
technique known as priority-based clustering (PPC) in
which cluster structure is determined by the
geographic location information and priorities
assigned to vehicles. In this approach, the CH election
process is similar to the computation of minimum
dominating sets used in graph theory. A node priority
is calculated on the basis of its ID, current time, and
eligibility function.
In [14] a hybrid backbone-based clustering algorithm
was presented. Nodes with a substantially higher
degree of connectivity create a backbone that is
selected as leader during cluster creation. The leader
then takes part in CH election and effective cluster re-
organization based on the combined relative velocity
of the leader's vehicles.
Also, another multi-hop clustering model using the
YATES algorithm to achieve stability in clusters was
introduced in [16].
Threshold Based (TB) is a multi-hop clustering
approach that was presented in [17]. This algorithm
aims to increase the network topology stability and
minimize network dynamic. During the cluster
creation process, the speed differential between cars,
along with their direction and position, were taken into
account.
The Vehicular Multi-hop algorithm for Stable
Clustering (VMaSC) was proposed in [20]. This
algorithm selects the vehicle which has minimum
speed as a CH. The VMaSC network had a less
overhead delay and greater PDR than other algorithms.
An Adaptive Weighted Clustering Protocol (AWCP)
was created by selecting the most stable vehicles
among present vehicles to function as CHs based on
speed, highway ID, position, and direction [19].
Highway ID information is utilized in order to
maximize the cluster stability structure.
The authors in [21] proposed a Passive Multi-hop
Clustering (PMC) algorithm in VANET to address the
lack of performance in clustering methods in terms of
reliability and stability. Clustering is presented in this
method based on the priority neighbor following
strategy, and this technique is used to elect the best
CH.
The number of clusters can decrease using the multi-
hop clustering techniques while also increasing cluster
coverage and stability. To summary, multi-hop
algorithms give good cluster stability and great
coverage, particularly in terms of CM re-affiliation,
CH changes, and cluster longevity. Multi-hop cluster
construction, on the other hand, is more complicated,
and it will take a long time to form a cluster, which
may cause a delay in delivering data. In addition, the
cluster overhead needs to be improved.
Table III presents a comparison between the single and
multi-hop clustering algorithms in terms of vehicle
density, transmission range, hop count, and traffic
scenario.
TABLE III. SINGLE AND MULTI-HOP CLUSTERING ALGORITHMS
COMPARISON
Algorithm
Vehicle
Dеnsity
Transmission
Rangе
Hop
Count
Traffic
Scеnario
EWCA
50-150
300m
Single
Highway
VWCA
10-350
100-1000m
Single
Highway
RMAC
25-75
250
Single
Highway
CBSC
-
-*
Single
Highway
DHC
50-200
300m
Single
Highway
LRCA
1500
200,500m
Single
Urban
AMACAD
50
100-200m
Single
Urban
DMAC
30-200
-
Multi
Random
ALM
50-
1000
-
Multi
Highway
VMaSC
100
200m
Multi
Highway
PPC
100
250m
Multi
Highway
DBC
100-
500
250m
Multi
Urban
TB
400
150-300m
,800-1000m
Multi
Highway
AWCP
25-200
1000
Multi
Highway
PMC
100
100-300m
Multi
Random
CCA-IoV
100
-
Single
Highway
* (-) Means Not mentioned.
3) Cluster Maintenance
Due to the VANET's dynamic topology, frequent
vehicle re-connection and disconnection cause
substantial packet loss. By minimizing frequent
vehicle re-clustering, the cluster maintenance
procedure provides strong connectivity and achieves
steady link longevity through CH. Cluster merging,
and Vehicle joining, vehicle leaving, are and some of
cluster maintenance methods.
447
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.

The CH transmits frequently signals during the vehicle
joining and vehicle leaving process, and if it detects
any signal from a vehicle, that new vehicle is allocated
to that cluster and becomes the CM of that cluster. The
CH's local data base will then be updated. When the
CH loses contact with a member vehicle, the
information for that member is deleted from the CH's
local database. AWCP in [19] uses this maintenance
method.
The cluster merging process is the second method;
which is more difficult than the first. When two or
more clusters may be represented by a single merged
cluster, this method is used to reduce the number of
clusters and enhance clustering efficiency. For each
approach, the cluster merging requirements are
different. For example, the cluster merging occurs in
ALM algorithm [14] when two CHs are in
communication range of each other. Whereas, some
algorithms like PPC in [12] used a distance threshold
to control cluster merging, where the cluster merging
occurs between two CHs if their distance is less than
the dismiss threshold. In [20], the VMaSC is
compared to the AVGREL-SPEED of two neighboring
CHs. the higher CH relative speed gives up its CH
position and becomes a CM for the lower CH speed.
The cluster merging method is also used by the PMC
algorithm in [21], to request cluster merging; the CH
transmits merge request packets to other surrounding
CH nodes. The merging process is carried out if one of
these CHs has a high relative speed and lesser
following cars.
Some algorithms like TB [17], and LRCA [22]
handled the two previous methods in the maintenance
phase.
Another technique utilized in the cluster maintenance
phase is a selected secondary CH. This approach was
employed in some algorithms like EWCA proposed in
[23]. This method increases the clustering stability
because it overcomes the CH’s unavailability.
C. Clustering Algorithms Comparison.
When comparing the clustering algorithms, a variety
of parameters are commonly used. Any clustering
algorithm is generated and characterized using these
parameters [4]. Latency, cluster stability, and
convergence are some of these parameters. Table IV
shows a comparison of the benchmark algorithms on
the basis of these parameters.
Stability: A good clustering algorithm provides the
stability to the network.
Latency: refers to the time needed to send a message
from a source to a destination.
Convergence: It is the time taken for the cluster
formation process.
A good cluster algorithm should provide high stability,
low latency during sending packets, and low
convergence time to form the cluster.
TABLE IV. CLUSTERING ALGORITHMS COMPARISON
Algorithm
Clustеr stability
Latеncy
Convеrgence
EWCA
High
Low
Medium
VWCA
High
-*
Low
AWCP
Low
Low
Medium
VMaSC
High
Low
Low
LRCA
Improves
Low
-
ALM
Low
Low
Low
AMACAD
Medium
High
Low
PPC
High
-
Low
DBC
High
-
Medium
TB
High
-
High
RMAC
Medium
High
High
CBSC
High
-
-
PMC
High
Low
Low
DHC
High
-
Low
* (-) Means Not mentioned.
III. PERFORMANCE EVALUATION METRICS
Different parameters can be used to measure and
evaluate the performance of any clustering technique;
the most frequent parameters for measuring the
performance of clustering techniques are cluster
performance and network performance: [4], [27].
A. Cluster Performance Parameters: These
parameters reflect the stability of the network's
backbone nodes and indicate how well
clustering strategies perform. These
parameters describe the overall cluster
performance and stability. Cluster stability,
Cluster numbers, CH Lifetime, CM Lifetime,
CH change Rate, and Cluster size are some of
the cluster performance parameters.
Stable clustering methods should have a high CH and
CM lifetime, large cluster size, a low CH change rate,
and few cluster numbers. However, cluster
performance parameters only cannot explain
connection links' between vehicles in the network.
B. Network Performance Parameters: These
parameters define the overall network
performance, which include: Throughput,
Packet loss ratio, Packet Delivery Ratio
(PDR), overhead, and End to End Delay (E2E
Delay) or Latency.
Context-based clustering techniques are estimated
using these parameters. Efficient clustering algorithms
lead to low packet loss rate, short E2E delay, large
throughput, small overhead, and high PDR. The
performance parameters and the simulator tools used
for different algorithms are presented in Table V.
448
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.

IV. SUMMARY OF OUR PROPOSED ALGORITHM
A. Literature problems
Our literature survey is for two decades, where the
primary focus has been the stability of CH. Some of
the most important issues in this literature are: Firstly,
the mobility and nеighborhood arе the most mеtrics
takеn. Secondly, the work on the rеal scеnario is
mostly limitеd to the highway; analysis on the urban
scenario is missing. Designing a clustering algorithm
for an urban environment is more complicated than
designing it for a highway, due to the large number of
intersections and the varied speed of vehicles as a
result of congestion. Thirdly, the methods designed
have overlooked the effect of dynamic change of these
networks. Finally, in the urban scenario, the mеtrics
like speеd and mobility are lost whеre the vehicle
speed is low and there is huge congеstion in pеak
hours. Thus, there is a need to search for new metrics
to effectively analyze the networks.
B. Our Proposed Solution
We create a comprehensive solution for VANET
issues, focusing on the urban environment. A newly
suggested vehicular-hypergraph-based spectral
clustering model is used for cluster generation. Our
approach has cluster formation, CH selection, and
maintenance. The concept of hypergraphs is well-
suited to capturing the dynamic nature of VANETs,
and clusters are constructed using the designed
hypergraph algorithm.
For CH selection, a cumulative multimetric is designed
to consider four factors: relative speed, trust score,
neighboring degree, and eccentricity.
In the cluster maintenance phase, we use the vehicle
joining and vehicle leaving method.
Simulation is implemented using MATLAB,
Simulation of Urban Mobility (SUMO), and Traffic
Control Interface (TraCI). The area considered for the
study is a crowded market area of Baghdad’s real map;
it is extracted from Open Street Map (OSM), Figure. 4.
Fig. 4. The Taken Area from Baghdad City Map
Figure.5 shows a comparison between our proposed
and two others algorithms in the literature, PMC, and
one hop VMaSC. The comparison occurs for low
density with maximum speed of 25m/s at different
transmission ranges (100m, 200m, and 300m
respectively). We can see, when the vehicle
transmission range increases, the percentage of CH
lifetime increases, so the cluster stability will increase.
Our proposed achieves high stability with
approximately %18 and %30 at low and high
transmission ranges respectively. This is due to the
effectiveness of the cluster formation phase using
hypergraph theory and the novel set of CH selection
parameters.
Fig. 5. Comparision Our Proposed at Low Density.
V. CONCLUSION
VANET is a new technology that can be used in
future ITSs. Many researchers are concentrating on the
issue of routing scalability and reliability in VANETs.
The clustering technique is one of the existing
mechanisms designed to adapt to the VANET
environment. An extensive survey of the most
common clustering approaches was presented in this
paper to handle various VANETs issues. Each of the
clustering steps' algorithms have been described;
including CH selection, cluster formation according to
the hop distance, and cluster maintenance. In addition,
we have compared these algorithms based on some
key parameters to see the performance of these
algorithms. The evaluation parameters used for some
clustering algorithms have also been presented. Then,
we have summarized our proposed algorithm and
showed its performance in terms of stability. Our
proposed achieved between 18% and 30% of stability
at low and high transmission ranges respectively
compared with the most usable algorithms which have
high mean citations in the literature.
449
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.

TABLE V. CLUSTERING ALGORITHMS EVALUATION PARAMETERS.
REFERENCES
[1]
J. Barrachina, J. A. Sanguesa, M. Fogue and P.
Garrido, "V2X-d: a Vehicular Density Estimation
System that combines V2V and V2I
Communications," Nov 2013.
[2]
B. Marzak and K. e. Guemmat, "A Survey on
Routing Protocols for Vehicular Ad-Hoc
Networks," Indian Journal of Science and
Technology, Dec 2016.
[3]
R. Naja, Wireless Vehicular Networks for Car
Collision Avoidance, New York, 2013.
[4]
M. . K. J. Alsabah, H. Trabelsi and W. Jerbi,
"Survey on Clustering in VANET Networks," in
18th International Multi-Conference on Systems,
Signals & Devices (SSD), Monastir, Tunisia,
2021.
[5]
P. Panse, T. Shrimali and M. Dave, "An
Approach for Preventing Accidents and Traffic
Load Detection on Highways," January 2016.
[6]
Y. SHI , "AN EFFICIENT CLUSTER-BASED
SERVICE MODEL FOR VEHICULAR AD-
HOC NETWORKS ON MOTORWAYS," Aston
University, 2017.
[7]
Venkatesh, A. Indra and R. Murali, "Routing
Protocols for Vehicular Adhoc Networks
(VANETs): A Review," CIS Journal, January
2014.
[8]
K. N. Qureshi, H. Abdullah, F. Ullah and R. W.
Anwar, "VEHICULAR AD HOC NETWORKS
ROUTING PROTOCOLS: SURVEY,"
Sci.Int.(Lahore), November 2015.
[9]
P. Thakur and A. Ganpati, "A Comparative Study
of Cluster-Head Selection Algorithms in
VANET," 2019.
[10]
W. Ahsan, M. F. Khan, F. Aadil, M. Maqsood, S.
Ashraf, Y. Nam and S. Rho, "Optimized Node
Clustering in VANETs by Using Meta-Heuristic
Algorithms," 27 Febuary 2020.
[11]
S. Basagni, "Distributed clustering for ad hoc
networks," 1999.
[12]
Z. Wang, L. Liu, M. Zhou and N. Ansari, "A
position-based clustering technique for ad hoc
intervehicle communication," March 2008.
[13]
Goonewardene RT, Ali FH and Stipidis E,
"Robust mobility adaptive clustering scheme with
support for geographic routing for vehicular ad
hoc networks," IET Intell Transp Syst, pp. 148-
158, 2009.
[14]
E. Souza, I. Nikolaidis and P. Gburzynski, "A
new aggregate local mobility (ALM) clustering
algorithm for VANETs," in Communications
(ICC), 2010.
[15]
A. Daeinabi , A. G. Pour Rahbar and A.
Khademzadeh, "VWCA: An efficient clustering
algorithm in vehicular ad hoc networks," Journal
of Network and Computer Applications, vol. 34,
pp. 207-222, 2011.
[16]
W. Li, A. Tizghadam and A. Tizghadam,
"Density based clustering algorithm for vehicular
ad-hoc networks," in Communications (ICC),
2012 IEEE International Conference, 2012 June.
[17]
Z. Y. Rawashdeh and S. Mahmud, "A novel
algorithm to form stable clusters in vehicular ad
hoc networks on highways," 2012.
[18]
M. Morales, E. J. Cho, C. s. Hong and S. Lee,
"An adaptable mobility-aware clustering
algorithm in vehicular networks," Journal of
Ref
Algorithm
Simulator Tool
Evaluation Paramеters
[11]
DMAC
NS2
Cluster dеnsity, cluster rеsidence time, cluster lifеtime,
[12]
PPC
NS2
Clusterreconfiguration rate, Mean cluster diamеter, Throughput.
[13]
RMAC
NS2
Cluster residеnce times, error estimated
Locations, node re-clustering time
[14]
ALM
SIDE/ SMURPH, SUMO
Status changes per node, CH lifеtime, CH density
[15]
VWCA
Matlab
PDR, CH and CM lifetime
[16]
DBC
Java, VANET MobiSim
Avеrage clustеr size, average number of clusters, average percentage of
clustered nodes, average number of CH changes per node, average time
which node spend being clustered.
[17]
TB
C++
Average clustеr lifetime, average clustеr lifetime
[18]
AMACAD
Java
Membеrship lifetime , CH lifetime, Re-affiliation rate
[19]
AWCP
SUMO, JOSM, MOVE, NS2,
PDR, Average Cluster Lifetime, Packet overhead
[20]
VMaSC
SUMO, NS3
CH Changе Rate, CH/CM Duration, Number of Vehicles in SE state,
Ovеrhead,
[21]
PMC
VanetMobiSim, NS2
Number of Average Cluster Head Changes, Average CH/CM Duration
Time, overhead
[22]
LRCA
SUMO, NS2
CH change rate, CH/CM duration, PDR, routing overhead, E2E dеlay
[23]
EWCA
SUMO, NS2
Cluster stability, number of clusters, and E2E
[24]
CBSC
SUMO, OMNeT++
Avеrage Number of Re-affiliation times, Avеrage CH/CM Lifеtime,
Packеt Loss Rate
[25]
DHC
SUMO
Cluster formation rate, numbеr of changеd states, CH/CM lifеtime, packet
overhеad, CH Aliеnation
[26]
CCA-IoV
SUMO,NS2
CH Lifetime, CM lifetime, Cluster size
450
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.

Computing Science and Engineering, Sep 2012.
[19]
M. Hadded, R. Zagrouba, A. Laouiti, P.
Muhlethaler and A. S. Leila, "A Multi-Objectif
Genetic Algorithm-Based Adaptive Weighted
Clustering Protocol in VANET," 2015.
[20]
S. Ucar, S. C. Ergen and O. Ozkasap, "VMaSC:
Vehicular Multi-hop algorithm for Stable
Clustering in Vehicular Ad Hoc Networks," in
2013 IEEE Wireless Communications and
Networking Conference (WCNC): NETWORKS.
[21]
Degan Zhang and Yu-Ya Cui, "New Multi-Hop
Clustering Algorithm for Vehicular Ad Hoc
Networks," 2018.
[22]
Xiang Ji, Huiqun Yu, Guisheng Fan, Huaiying
Sun and Liqiong Chen, "Efficient and Reliable
Cluster-Based Data Transmission for Vehicular
Ad Hoc Networks," Hindawi, 2018.
[23]
A. Bello Tambawal, R. Md Noor, R. Salleh, C.
Chembe and M. Oche, "Enhanced weight-based
clustering algorithm to provide reliable delivery
for VANET safety applications," 4 April 2019.
[24]
Xiaolu Cheng and Baohua Huang, "A Center-
Based Secure and Stable Clustering Algorithm for
VANETs on Highways," Hindawi Wireless
Communications and Mobile Computing, 2019.
[25]
Ghada H. Alsuhli, Ahmed Khattab and Yasmine
A. Fahmy, "Double-Head Clustering for Resilient
VANETs," Hindawi Wireless Communications
and Mobile Computing, 2019.
[26]
M. Bersali, A. Rachedi and H. Bouarfa, "A New
Collaborative Clustering Approach for the
Internet of Vehicles (CCA-IoV)," 2020.
[27]
Oussama Senouci, Saad Harous and Zibouda
Aliouat, "Survey on vehicular ad hoc networks
clustering algorithms: Overview, taxonomy,
challenges, and open research issues," WILEY,
March 2020.
451
Authorized licensed use limited to: UNIVERSITE DE SFAX. Downloaded on December 04,2022 at 09:09:12 UTC from IEEE Xplore. Restrictions apply.