An Efficient Routing Protocol Using an Improved Distance-Based Broadcasting and Fuzzy Logic System for VANET

Abstract

The high mobility and frequent changes in the network topology remain a great divergence between Vehicular Ad-hoc Network (VANET) and Mobile Ad-hoc Network (MANET). This makes existing routing protocols for MANET imperfect for a VANET environment. In order to ensure good data transmission between vehicles, it is mandatory to improve these protocols to this new environment. Ad-hoc On-demand Distance Vector (AODV) is one of the most proposed reactive protocols in the literature for MANET, but its direct application in VANET returns poor performance results. In this paper, we propose an Efficient Routing Protocol using an improved Distance-Based Broadcasting and Fuzzy Logic System (ERPFL), which is an extension of the AODV routing protocol. In the first, the proposed scheme uses an improved Distance-Based Broadcasting method in route discovery process. Each vehicle selects the most suitable neighbours to send a route request, using different mobility parameters such as distance, velocity and direction. Then, the destination vehicle selects the best route from the most reliable routes received, using the Fuzzy Logic System. Link Expiration Time and Link Reliability Model are considered input metrics for this system. The simulation results show that ERPFL saves network bandwidth resources which contributes to a higher data transmission ratio. In addition, our proposed protocol efficiently distributes data packets, decreases the hop count and reduces network routing time.

Full text

Received: May 9, 2020. Revised: June 30, 2020. 81
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
An Efficient Routing Protocol Using an Improved Distance-Based Broadcasting
and Fuzzy Logic System for VANET
Khalid Kandali1* Hamid Bennis1
1IMAGE Laboratory, SCIAM Team, Graduate School of Technology,
Moulay Ismail University of Meknes, Morocco
* Corresponding author’s Email: kandalikhalid@hotmail.com
Abstract: The high mobility and frequent changes in the network topology remain a great divergence between
Vehicular Ad-hoc Network (VANET) and Mobile Ad-hoc Network (MANET). This makes existing routing protocols
for MANET imperfect for a VANET environment. In order to ensure good data transmission between vehicles, it is
mandatory to improve these protocols to this new environment. Ad-hoc On-demand Distance Vector (AODV) is one
of the most proposed reactive protocols in the literature for MANET, but its direct application in VANET returns poor
performance results. In this paper, we propose an Efficient Routing Protocol using an improved Distance-Based
Broadcasting and Fuzzy Logic System (ERPFL), which is an extension of the AODV routing protocol. In the first, the
proposed scheme uses an improved Distance-Based Broadcasting method in route discovery process. Each vehicle
selects the most suitable neighbours to send a route request, using different mobility parameters such as distance,
velocity and direction. Then, the destination vehicle selects the best route from the most reliable routes received, using
the Fuzzy Logic System. Link Expiration Time and Link Reliability Model are considered input metrics for this system.
The simulation results show that ERPFL saves network bandwidth resources which contributes to a higher data
transmission ratio. In addition, our proposed protocol efficiently distributes data packets, decreases the hop count and
reduces network routing time.
Keywords: VANET, Routing protocol, AODV, Distance-based broadcasting, Fuzzy logic, Link expiration time, Link
reliability model.
1. Introduction
Recently, Vehicular Ad-hoc Network have
become a promising field of research to improve the
Intelligent Transportation System (ITS). This
technology has attracted great attention from ITS
designers, automakers, private research agencies, and
university research laboratories [1, 2]. Indeed,
VANET networks provide a long list of applications
that constitute Intelligent Transportation Systems [3].
These applications are classified into security
applications and comfort applications. Safety
applications aim to avoid accidents and improve road
safety, such as optimizing paths between vehicles,
monitoring traffic and preventing collisions. While
comfort applications aim to improve the level of
passenger comfort, such as parking spaces and
restaurants, weather forecasting, the nearest hotels
and petrol stations [4, 5].
In VANET network, vehicles can communicate
with each other without any infrastructure required,
or each vehicle can randomly join or leave the
network [6, 7]. These vehicles are generally equipped
with network interfaces called On Board Unit (OBU)
[8]. The VANET network is a self-configured and
self-maintained network that does not require any
centralized authority. Each node of the network can
send, receive and relay packets to the other nodes of
the network [9].
The VANET network being considered as a
subclass of the MANET networks [10-12], several
researches and studies have shown the weakness of
the routing protocols designed for MANET when
applied directly to the vehicular environment [13],
due to the very growing number of vehicles on the

Received: May 9, 2020. Revised: June 30, 2020. 82
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
roads, highly mobility and frequent disconnections
[14,15]. Indeed, the major issue of VANET remains
in the time and the reliability of routing information
through different ITS applications [16].
In the literature, several works have recently been
done to optimize the existing routing protocols and
adapt them to the VANET characteristics. There are
several types of protocols, but reactive routing
protocols are still highly recommended and popular
[17]. This type of protocol discovers the routes and
maintains them only on demand [18]. The most
studied and optimized routing protocol in the
literature is Ad-hoc On-demand Distance Vector
(AODV) [19, 20]. In fact, this protocol has several
drawbacks, notably at the level of the Route
Discovery Phase [21]. In this process, the source node
generates a large number of Route Request (RREQ)
packets when it wants to find a better routing path to
the destination node, which increases the network
overhead [22]. In addition, when the destination node
receives several routes, it searches for the best route
based on the number of nodes in each routing path
[23]. This factor is not sufficient to select the best
route, especially in a VANET networks.
Broadcasting in VANET poses more challenges
due to the mobility and density of vehicles. Due to the
mobility aspect, there is no single optimal scheme for
all scenarios. Indeed, the quality of a VANET
network lies in the ability to broadcast effectively
packets among network vehicles. Thus, efficient
packet broadcasting is a key element to obtain high
performance in VANET.
In this context, this paper proposes an improved
Distance-Based Broadcasting method in route
discovery process. Each vehicle selects the most
suitable neighbours to send a route request, using
different mobility parameters such as distance,
velocity and direction. Each vehicle sends or
forwards a certain number of RREQ packets, which
does not exceed a certain threshold. The threshold
will be defined by each vehicle according to the
characteristics of the network. In addition, the
destination vehicle selects the best route from the
most reliable routes received, by using the Fuzzy
Logic System. Link Expiration Time (LET) and Link
Reliability Model (REL) are considered input metrics
for this system.
The methodology of this paper is recapitulated as
follows:
• At first, we explain how each node calculates
and updates the threshold value, taking into
account the characteristics of the network. This
value determines the maximum number of
RREQ packets broadcasted by each node in
Route Discovery Process. We avoid considering
a fixed threshold value, so as not to have a risk
disturbing the Route Discovery Phase, when we
have a high-density network.
• Then, we present the method to select a number
of neighbouring nodes that does not exceed a
certain threshold, to broadcast RREQ packets.
The selection is made by calculating the link
weight of each neighbouring node.
• Then, to choose the most stable and reliable
route between the nodes, a Fuzzy Logic system
is introduced, when the input metrics are LET
and REL.
• Finally, an extensive simulation is carried out by
modifying the number of vehicles to verify the
effectiveness of the proposed method, and
compare it with several methods proposed in the
literature.
Briefly, the main contributions of this paper are:
• Proposing an improved Distance-Based
Broadcasting method in Route Discovery
Process.
• Selection of the reliable route to transmit data in
VANET using the Fuzzy Logic. LET and REL
are the input metrics for this system.
• A comparative evaluation of performance of
ERPFL, CALAR-DD [24] and M-GEDIR [29]
in terms of Packet Delivery Ratio, Average End-
to-End delay and hop count.
Thus, in the rest of the paper, section 2 presents
several routing protocols proposed in the literature.
Then, section 3 provides an overview of the Fuzzy
Logic system used in this paper. In section 4, we
illustrate the proposed approach. Then, section 5
details the simulation process, and displays the
results. Finally, we will conclude this paper in section
6.
2. Related works
Various researches have been proposed to correct
the aforementioned shortcomings on AODV in order
to improve this routing protocol and adapt it to
VANET.
In this section, we present the main modifications
recently made on the AODV routing protocol to
improve and adapt it to VANET.
A cache agent-based location aided routing using
distance and direction (CALAR-DD) was introduced
in [24], which combines geocasting and position
routing. In the first step, CALAR-DD selects the next
hop vehicle to forward the packet until it arrived at
the expected region. Whereas, the second step
consists in using cache agent geocasting in the
expected region, to select and forward the packets to
the destination. However, the authors have not

Received: May 9, 2020. Revised: June 30, 2020. 83
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
proposed a method for selecting the most reliable
route by the destination vehicle.
In order to improve vehicular communication in
urban environment and computational resources, in
[25], the authors presented a position-based routing
protocol named Improved Geographic Routing (IGR),
which depends on the distance to the destination and
the vehicle density of each street to select the junction.
And to transmit a data between two junctions, IGR
uses an improved greedy forwarding, which take into
account the link error rate in the path selection.
However, the proposed method did not take into
account the velocity of the vehicles. This parameter
represents a major challenge in the VANET
environment.
In [26], the authors proposed two algorithms to
design a new routing protocol called Improved
Directional LAR routing protocol. The first algorithm
is Select Next Forwarder (SNF), which considers the
nodes adjacent to the border area that has minimum
distance. While, the second algorithm is called Packet
Forwarding Algorithm (PFA), which is used to
broadcast Routing packet to the neighbouring nodes.
This approach aims to find the best next-hop
forwarder node in forward area. The main objective
of which is to reduce unnecessary transmissions. As
the selection of the best path to transmit the data is a
challenge in VANET networks, the proposed method
did not address this point. Also, the mathematical
model used during the data transmission, did not take
into account other mobility parameters such as the
velocity and direction of the vehicles.
[27] introduced a reliable path selection and
packet forwarding routing protocol (RPSPF), in order
to create a suitable route between vehicles to send
data. RPSPF considers the connectivity and the
shortest appropriate distance based on several
intersections. And in order to avoid packet loss
during packet forwarding, the proposed routing
protocol applies a novel reliable packet forwarding
technique between intersections. The selection of
next hop in the proposed method is based on the
partitioning of road networks instead of being based
on mobility parameters and the quality of connections
between vehicles.
This paper [28] introduced a fuzzy assisted
position-based routing protocol (FPBR-DTN) for
V2V communication, the main objective of which is
to ameliorate greedy routing and keep away absurd
node for routing. However, a neighbouring node that
has the highest chance value is been selected for the
greedy forwarding. This value is calculated by
applying Fuzzy Logic and different parameters, such
as the number of neighbours, the direction of
neighbouring vehicles, their speed and their distance
form a destination. Then the greedy mode, perimeter
and DTN are considered as the main modes of FPBR-
DTN. This greedy scheme enhances the efficiency of
routing protocol. However, the authors have not
proposed a method for selecting the most reliable
route by the destination vehicle. In addition, other
parameters like Link Expiration Time can be used as
input parameters for the Fuzzy Logic System.
In [29], the author introduced a Multi-metric
geographic routing technique (M-GEDIR), in order
to find vehicles for the next hop in dynamic
forwarding regions. M-GEDIR has taken into
account the future position of vehicles, critical area
vehicles at the border of transmission range, and the
strength of the received signal. In NHV selection, M-
GEDIR uses a maximum weighting factor. However,
when selecting the next hop, this work does not take
into account the quality of the link, which could
augment the possibility of link failure.
A Fuzzy Logic-Based Geographic Routing for
Urban VANET in [30], in order to improve road
safety and transportation capacity. This approach
associated several metrics, such as direction of
vehicle position, link quality and achievable
throughput, to find an appropriate next-hop node for
packet forwarding. The Fuzzy logic system
coordinated and evaluated the metrics considered.
While, the algorithm used ETX for the appreciation
of the quality of the link. However, the authors used
the Fuzzy Logic system for each node before
transmitting the packets, which causes significant
delays, in particular for the high network densities
and the low mobility of the vehicles.
In [31], a novel geographic routing protocol
called LTQGR was proposed for an urban VANET.
This protocol was based on a new routing metric
called Link Transmission Quality (LTQ), in order to
select the routing path with the lowest aggregated
weights for the transmission data. The transmission
quality of the link took into account the cost and the
reliability of transmission. However, this work does
not consider other mobility parameters such as
direction and speed in the design of the new metric
concerning the quality of transmission link.
In [17], a new version of the AODV routing
protocol was been proposed. This new version was
based on two strategies. The first was to improve
HELLO packets and control packets by sharing
information from neighbouring node in the network.
This strategy has made it possible to reduce the
network overhead. The second strategy was based on
an alternative routing option to deliver data packets
in a short time. Generally, the authors have sought
through this paper to improve the packet delivery
ratio by reducing the use of bandwidth. This paper did

Received: May 9, 2020. Revised: June 30, 2020. 84
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
not use new metrics to test the stability of the routes
between the nodes. The proposed algorithm took into
consideration Hop Number as basic metric to select
the best and the alternative paths.
[32] suggested a cross-layer optimized position-
based routing protocol for urban VANET. This
approach aimed to facilitate cooperation in the
network and to solve misbehaving nodes which
knowingly drop packets by presenting a social
aspect to its design. However, it was recommended
that nodes with a close favourable social relationship
will be net forwarder nodes, otherwise nodes with
explicit distrust connections were rated favoured in
the network.
In order to minimize the impact of the position
error, due to various factors, such as signal fading,
tunnels and line-of-sight, the paper [33] designed a
predicted Position Based Routing Protocol (PPRP)
for VANET. This approach used the Kalman Filter
(KF) to guess the vehicle location based on the
previous and current location. The selection of the
next section in the proposed method is based on the
position of the vehicles instead of being based on the
mobility parameters and the quality of the
connections between the vehicles.
A distance and direction-based location aided on
multi hop routing protocol (DD-LAR) for the city
traffic scenario was suggested in [34], where the
vehicle which had a minimum angle with maximum
distance covered from the source was selected for the
Next Hop Vehicle. But DD-LAR was interested in
relative distance to select the vehicle which are in the
source transmission range in case there is no vehicle
to choose according to the minimum angle with
maximum distance. Therefore, this approach aimed
to reduce the hop count in routing process. The
authors did not introduce the speed parameter to
examine the performance on path duration and hop
count metric.
[35] proposed a cache agent-based geocasting
(CAG). The mains additional concepts of which were
Re-caching and full range radio transmissions
assimilated with the Connectivity Assurance
Algorithm (CAA). The proposed protocol aimed to
reduce identical packets, hop count, packet loss and
hop-to-hop communications using CAA. Then the
cache packet will not be transmitted if there has been
no guaranteed packet confirmation. However, to
reduce the number of hops in route delivery, CAG
used full radio power.
The paper [18] presented a Pro-AODV routing
protocol based on information in the routing table in
order to minimize congestion in VANET networks. It
was based on the size of the routing table for each
node and did not require the execution of any
additional logic. Thanks to this, the proposed
protocol can find routes to congestion-free
destinations, which has contributed to higher delivery
and a lower drop ratio. This proposed approach aimed
to improve the packet delivery ratio, average end-to-
end, drop ratio and control overhead. This paper used
probability-based broadcasting to solve the problems
associated with the simple flooding method. Each
node has received a predetermined probability to
rebroadcast packets. However, there was a risk that
some nodes might not receive routing or data packets.
To summarize this section, these proposed works
present one or more of the following limitations: (i)
The selection of next hop and the most stable path is
based on the partitioning of road networks instead of
being based on mobility parameters, such as position,
speed and mobility; (ii) Several studies do not
provide complete approaches, which must combine
the selection of next hop and the most stable path, in
order to propose an efficient routing protocol.; (iii)
The choice of suitable input parameters for the Fuzzy
Logic System must take into consideration the quality
of the links between the vehicles. However, our
proposed scheme provides a complete routing
protocol for VANET. This approach is based on an
improved Distance-Based Broadcasting, which takes
into account several mobility parameters, and a
method of selecting the most stable path using the
Fuzzy Logic System, whose input parameters are
linked to the quality of the links between the vehicles.
3. Fuzzy logic system
Fuzzy logic is a theory formalized by professor
Lotfi Zadeh in 1965. This system is considered to be
an artificial intelligence tool, based on mathematical
logic. It extends classic Boolean logic with partial
truth values, which are real numbers between 0 and 1.
Fuzzy logic consists of a few basic concepts, such as:
fuzzy set, if-then rules, and membership function, and
it has four steps, as shown in Fig. 1: Fuzzification,
rule base, inference engine and defuzzification.
The fuzzification step converts any type of fuzzy
input into fuzzy sets. It assigns an integer number of
a system to fuzzy sets with a certain degree of
membership between 0 and 1. However, each value
corresponds to the degree of uncertainty that the
value is included in the set. In this step, the fuzzy sets
are illustrated with words that we can think of in a
linguistically natural process. The rule base includes
a limited number of rules which are used to link the
fuzzy input variables to the fuzzy output variables.
An inference engine includes a rules database, and it
is used to process the production of fuzzy decisions.

Received: May 9, 2020. Revised: June 30, 2020. 85
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
Figure. 1 Fuzzy logic system for ERPFL
When, the defuzzification step consists in obtaining
continuous variable from the fuzzy truth values [36].
4. Proposed approach
Our main intention in this paper is to improve the
Route Discovery Phase to increase the quality of
routing in VANET. This work proposes an improved
Distance-Based Broadcasting, by using the mobility
information available on the network such as the
position, the speed and the direction of the nodes.
And then to select the most stable route, using the
Fuzzy Logic System. We introduce a first factor
called Threshold Number TN. TN limits the number
of routing packets sent by each node in the network.
Each vehicle selects the most suitable neighbours to
send a route request, using different mobility
parameters such as distance, velocity and direction.
The threshold value will be updated by each node
according to the density of the network. The second
part of the proposed approach aims to select the
suitable route between the source node and the
destination node. In route discovery process, each
node that receive RREQ packet, calculates LET and
REL metrics using the position, speed and direction
parameters that are inserted into RREQ. Upon receipt
of a route request by the destination node, it
immediately sends LET and REL to the Fuzzy Logic
System, in order to calculate the fuzzy cost of each
route. Finally, the route with the higher cost function
will be selected as the appropriate route to send
RREP packet to the source node.
4.1 Selection of nodes with stable weight-based in
route discovery process
4.1.1. Threshold number
In order to update the threshold number TN to
transmit the routing packets, each node will need
several parameters like, the old stored threshold value,
the accumulated number of neighbours and hop count
Table 1. Notations and descriptions
Notation
Description
TN
Threshold number of packets sent by
each node.
TNold
Old threshold value stored in each node.
TNB
Accumulated number of neighbouring
nodes.
NB
Number of neighbouring nodes for each
node.
HC
Value stored in routing packets,
indicates the number of hops.
LW
Link Weight value.
D
Distance between two nodes.
Vi
Velocity of node i.
θ
Angle of direction.
value of the route. This information will be available
for each node upon receipt of each routing packet.
The different notations used in this section are
mentioned in Table 1.
The following formula is used to calculate the new
threshold value:

(1)
Where α + β =1.
TNB is calculated by using the following formula:
(2)
The source node adds the total number of the
neighbouring nodes TNB to the RREQ packet before
sending it. Each node that receives this packet,
calculates the number of its neighbours NB, and will
update the cumulative total number of neighbours.
Using Eq. (1), the node updates TN with the new
calculated value. The same procedure is repeated
when the destination node sends the RREP packet to
the source node.
In fact, each node updates its threshold value
according to the number of nodes connected. This
method is described in the algorithm 1.
4.1.2. Link weight
The Link Weight LW is a metric used to select
the most suitable neighbouring nodes, in order to
send route request.
Let N1 and N2 be two nodes with a speed V1, V2,
and direction θ1, θ2 respectively. LW is calculated by
Eq. (3):

(3)

Received: May 9, 2020. Revised: June 30, 2020. 86
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
Algorithm1 – Pseudo code of forwarding routing
packets
1. If Node i receive a RREQ packet Then
2. Get the value of Threshold number TN;
3. Get the number of neighbouring nodes NB;
4. Get the value of the total number of neighbouring
nodes TNB from RREQ packet;
5. Calculate the new value of TNB using Eq. (2);
6. Get Hop Count HC from RREQ packet;
7. Update the value of TN by using the Eq. (1);
8. If NB > TN Then
9. Calculate the weight link LW value for each
neighbouring node of Node i;
10. Select a number TN of the neighbouring nodes
of node i with the best weight link to forward the
RREQ packet;
11. Else
12. Broadcast RREQ packets to all neighbouring
nodes of node i;
13. End if
14. Else if Node i receive a RREP packet Then
15. Get the value of Threshold number TN;
16. Get the number of neighbouring nodes NB;
17. Get the value of the total number of
neighbouring nodes TNB from RREP packet;
18. Get Hop Count HC from RREP packet;
19. Calculate the new value of TNB using Eq. (2);
20. Update the value of TN by using the Eq. (1);
21. End if
22. End if
WD, WV, and Wθ are the weight factor of distance,
velocity and direction respectively.
Thus, the links which carry the least weight are
considered more stable links.
Upon receipt of a route request, each node
calculates the number of its neighbour nodes NB. If
this number is less than the threshold value TN, the
node broadcasts the packet RREQ to all its
neighbours; if the number of neighbours NB is
greater than TN, the node selects the most suitable
neighbours which have the least link weight to send
route request.
This process is described in the following pseudo
code.
4.2 Selection of the most stable route using fuzzy
logic system
4.2.1. Input VANET metrics
a. Link Expiration Time
In VANET networks characterized by high
mobility of vehicles, it is strongly recommended to
take into account the durability of the links when
creating routing paths. A long duration of the link
between the nodes makes the routing paths more
efficient and functional for a long time. Routing paths
Table 2. Notations and descriptions
Notation
Description
LET
Link Expiration Time.
REL
Link Reliability.
Tp
Time interval for a link between two
vehicles to be available
R
Transmission range.
L
Distance between two vehicles.
∆v
Velocity difference between two
vehicles.
based on more durable links increase data
transmission performance, and reduce network
maintenance costs. We suggest LET to test how long
two vehicles remain connected. Thus, links with a
high LET value are considered the most appropriate.
The different notations used in this section are
mentioned in Table 2.
Let N1 and N2 be two position nodes (x1, y1), (x2,
y2). These two nodes are in movement with a speed
V1 and V2, and with a direction angle θ1 and θ2
respectively. We assume that the two nodes have the
same transmission range R. As indicated in [37], the
LET between the two nodes N1 and N2 is given by Eq.
(4):

(4)
Where: a = V1 cosθ1 – V2 cosθ2
b = x1 – x2
c = V1 sinθ1 – V2 sinθ2
d = y1 – y2
b. Link Reliability Model
The reliability of the links REL between two
vehicles in VANET is an important parameter in
assessing the stability of the system. As noted in [15],
this metric shows the probability that the
transferability of data between two vehicles remain
available for a specific period of time. This
probability is calculated by Eq. (5):

(5)
Position, speed and direction information are
used to calculate REL value of the link l. This
information is obtained using the GPS installed on
each vehicle.
Let Ci and Cj two vehicles in movement. The
value of Tp shows the time interval for a link between
these two vehicles to be available, and it is calculated
according to the following possible cases:
Both vehicles have the same direction:

Received: May 9, 2020. Revised: June 30, 2020. 87
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08



(6)
The two vehicles are moving in a different
direction:
















(7)
With R is the transmission range and Lij is the
distance between the two vehicles. It is calculated by
Eq. (8):
(8)
To calculate REL, we use the following error
function:



(9)
This function is also called GAUSS error function.
It is generally used in the fields of statistics and
probability.
From the previous formula, we can calculate REL
by Eq. (10):





(10)
Where μ and  indicate the average value and the
variance of velocity,  is the velocity
difference between two vehicles.
4.2.2. Fuzzification
The membership functions are systematized into
three variables: Low, Medium and High. The
membership function of LET and REL are mentioned
in Fig. 2 and Fig. 3. Indeed, these metrics are
disposed as arranged at the Fuzzy Logic System.
Figure. 2 Membership function of LET
Figure. 3 Membership function of REL
4.2.3. IF-THEN rules
As mentioned in Table 3, fuzzy rules are
introduced based on the IF-THEN rules. These rules
are pre-defined and combined for mapping fuzzy
values, and they are used to calculate the link fuzzy
cost for each route.
Table 3. Rule base
No.
LET
REL
Fuzzy cost
1
Low
Low
Very bad
2
Low
Medium
Bad
3
Low
High
Acceptable
4
Medium
Low
Bad
5
Medium
Medium
Acceptable
6
Medium
High
Good
7
High
Low
Acceptable
8
High
Medium
Good
9
High
High
Perfect
0
0.5
1
010 20 30 40 50 60 70 80 90 100
fn
LET
Low Medium High
0
0.5
1
010 20 30 40 50 60 70 80 90 100
fn
REL
Low Medium High

Received: May 9, 2020. Revised: June 30, 2020. 88
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
Table 4. Route request packet of ERPFL
Type
Flags
Reserved
Hop count
RREQ ID
Destination IP address
Destination sequence number
Original IP address
Original sequence number
TNB
Position
Velocity
Direction
LET
REL
4.2.4. Defuzzification
The defuzzification steps take into account the
output membership function and its degree to find a
precise output value from the fuzzy space. In this
step, we use a Center Of Area (COA) method which
can be performed via the following equation:


 


 (11)
Where, μc is a discrete membership function, Xi is
defined as the fuzzy variable, and n is the number of
rules.
This step converts the Fuzzy Logic output
function to absolute numbers. This number is used to
select the most reliable route between the source node
and the destination node for transmitting data.
Thus, the route discovery process starts when a
source node has to send a data packets to a destination
node, and it does not find a route in its routing table.
In fact, it broadcasts the route request packet to
neighbouring nodes. The format of the route request
packet is described in Table 4. When a neighbouring
node receives a route request packet, it creates a
reverse route to the source node if it does not already
exist in its routing table. A reverse route is mandatory
to return response packets to the source node. If the
receiving node is the destination node, it simply
generates and sends the RREP packet.
If a node receives a RREQ packet, and it is not
the destination node, and it already has a route in the
routing table to the source node, it checks the LET
and REL values of the RREQ packet, and calculates
a new LET and REL values. If LET and REL
calculated is greater than LET and REL presented in
RREQ, the new values are updated in RREQ. Finally,
when a destination node receives RREQ, its checks
the LET and REL values, and sends them to Fuzzy
logic to calculate fuzzy cost of the route. The route
with a higher fuzzy cost will be selected to transmit
RREP to the source node.
5. Simulation and results
5.1 Simulation setup
To evaluate the performance of our proposed
routing protocol, we used the NS2 network simulator.
It is free software and is widely used in academic
research. It is implemented in C ++ with optional
Python bindings, which allows to write simulation
scripts in C ++ or python. The simulation is carried
out on a network of size 1500m x 1500m. We used
the Random Waypoint Mobility model, where each
node starts from any point. The nodes move in the
same direction or in a different direction. The
simulation lasted 1400 seconds.
The simulation process demonstrates the quality
of the proposed approach in terms of Packet Delivery
Ratio (PDR), Average End to End Delay (E2ED) and
Average number of Hop Count. In which, we varied
the number of vehicles from 40 to 200, and we varied
the speed of vehicles from 10 Km/h to 50 Km/h, in
order to assess the efficiency of ERPFL in a network
many connected nodes, and high mobility. The
performance of proposed protocol was compared
with CALAR-DD [24] and M-GEDIR [29]. All other
parameters are described in Table 5.
After several tests, we obtained the best results
when we start the simulation with the following
parameters: α= 0.3; β=0.7; WD = 0.2; WV = 0.3; Wθ
= 0.5.
5.2 Simulation results
5.2.1. Packet delivery ratio
Fig. 4 shows the impact of density on Packet
Delivery Ratio for ERPFL, CALAR-DD [24] and M-
GEDIR [29]. The proposed protocol achieves a
higher PDR than CALAR-DD and M-GEDIR. The
PDR value of CALAR-DD and M-GEDIR decreases
from 94% and 86% to 88% and 80% respectively,
when the number of nodes increases. While the PDR
of ERPFL keeps a value greater than 87%, even if the
number of nodes increases. While Fig. 5 illustrates
the impact of speed on PDR of ERPFL, CALAR-DD
[24] and M-GEDIR [29]. The PDR of ERPFL retains
a value higher than 80% for the different speeds
considered, while the PDR of CALAR-DD and M-
GEDIR decreases when the speed of the vehicles
increases. The PDR loses its value up to 60% for

Received: May 9, 2020. Revised: June 30, 2020. 89
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
Table 5. Simulation parameters
Parameter
Value
NS-2 version
NS-2.35
Simulation time
1400 s
Area size
1500 m × 1500 m
Number of nodes
40, 80, 120, 160, 200
Mobility model
Random Way Point
Minimum speed
10 Km/h
Maximum speed
50 Km/h
Transmission range
250 m
MAC protocol
IEEE 802.11p
Channel bandwidth
10 MHz
Propagation model
Two-ray ground
Traffic type
Constant bit rate (CBR)
Data rate
6 Mbps
Packet size
512 bytes
Protocols used
ERPFL, CALAR-DD, M-
GEDIR
α, β
0.3, 0.7
WD, WV, Wθ
0.2, 0.3, 0.5
CALAR-DD and 50% for M-GEDIR. This efficiency
is due to the ability of ERPFL to successfully use
speed, direction and position information in the
selection of neighbouring nodes to send routing
packets. Therefore, ERPFL selects more stable and
reliable paths to transmit data from source to
destination using Fuzzy Logic System. This saves
network bandwidth resources and contributes to a
higher data transmission ratio.
5.2.2. Average end-to-end delay
Fig. 6 illustrates the progression of the average
End-to-end delay for ERPFL, CALAR-DD [24] and
M-GEDIR [29]. The results show that our proposed
protocol requires less propagation time whether for
low or high density and mobility. However, the
Average E2ED of CALAR-DD increases from 0,5s
to 1,3s, and the Average E2ED of M-GEDIR
increases from 0,7s to 1,8s when density becomes the
main challenge. While ERPFL maintains a high
efficiency and the delay does not exceed 0,4s. Fig. 7
shows the impact of velocity on average End-to-End
delay for ERPFL, CALAR-DD [24] and M-GEDIR
[29]. The delay of ERPFL does not exceed 0,5s
despite the increase in vehicle speed, but the delay of
CALAR-DD and M-GEDIR increases respectively
by 0,5s and 0,7 s to 1,3 s and 1,7 s. The destination
node in our proposed algorithm quickly finds the
most reliable route using LET and REL as the input
metric to Fuzzy Logic System, and sends the RREP
packets along the suitable route that has the higher
fuzzy cost value. By calculating the value of the fuzzy
Figure. 4 PDR vs number of vehicles
Figure. 5 PDR vs velocity
Figure. 6 Average E2ED vs number of vehicles
70
75
80
85
90
95
100
40 80 120 160 200
PDR (%)
Number of vehicles
ERPFL CALAR-DD M-GEDIR
0
10
20
30
40
50
60
70
80
90
100
10 20 30 40 50
PDR (%)
Velocity (Km/h)
ERPFL CALAR-DD M-GEDIR
0
0.5
1
1.5
2
40 80 120 160 200
Average E2ED (seconds)
Number of vehicles
ERPFL CALAR-DD M-GEDIR

Received: May 9, 2020. Revised: June 30, 2020. 90
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
Figure. 7 Average E2ED vs velocity
cost of each route using REL and LET, ERPFL shows
its success in a high-density network and high
mobility to find a stable and shorter route, and with a
maximum lifetime.
5.2.3. Average number of hop count
Fig. 8 shows a comparison between ERPFL,
CALAR-DD [24] and M-GEDIR [29], in terms of
average number of Hop Count. Indeed, ERPFL and
CALAR-DD generally have the lowest value of
number of hops counts than M-GEDIR. The hop
count of ERPFL decreases form 0,7% to 0,2%, as the
number of vehicles increases. So that the hop count
of CALAR-DD starts better with 0,6% and decreases
to 0,2%. But the hop count of M-GEDIR is higher
than ERPFL and CALAR-DD. While Fig. 9 shows
the impact of velocity on Hops Counts for ERPFL,
CALAR-DD [24] and M-GEDIR [29]. The hop count
of ERPFL reaches a better value than CALAR-DD
and M-GEDIR. It starts with 0,7% and increases to
1,6%, while the hop count of CALAR-DD and M-
GEDIR increases from 0,6% and 0,8% to 2% and
2,7% respectively, when the vehicle speed increases.
The fact that the nodes in ERPFL know the position,
speed and direction of their neighbouring nodes. The
proposed protocol uses a minimum number of hops
counts in route discovery process. Therefore,
decreasing the Hops Counts reduces congestion,
efficiently distributes data packets and reduces
network routing time.
6. Conclusion
In this paper, we have introduced an efficient
routing protocol using Fuzzy Logic for VANET,
called ERPFL. This approach is based in the first step
Figure. 8 Average hop count vs number of vehicles
Figure. 9 Average hop count vs velocity
on a method for selecting the nodes to send routing
packets in Route Discovery Process, each node
chooses a limited number of neighbours to send the
RREQ packet, instead of broadcasting the packet to
all neighbouring nodes. In the second step, we used a
Fuzzy Logic System to select the best route to
transmit data. Indeed, ERPFL calculates a fuzzy cost
for each route, using a Fuzzy Logic System, where
the inputs of this system are LET and REL. The route
between nodes with a higher fuzzy cost is selected to
transmit data. ERPFL aims to reduce bandwidth
consumption, reduce packet routing costs and ensure
a permanent data transmission in a VANET.
We have evaluated and compared the ERPFL
proposed routing protocol with CALAR-DD [24] and
M-GEDIR [29]. The simulation results show that
ERPFL succeeded in using speed, direction and
position to select the reliable neighbouring nodes to
0
0.5
1
1.5
2
2.5
10 20 30 40 50
Average E2ED (seconds)
Velocity (km/h)
ERPFL CALAR-DD M-GEDIR
0
0.2
0.4
0.6
0.8
1
1.2
1.4
40 80 120 160 200
Average Hop Counts
Number of vehicles
ERPFL CALAR-DD M-GEDIR
0
0.5
1
1.5
2
2.5
3
10 20 30 40 50
Average Hop Counts
Velocity (Km/h)
ERPFL CALAR-DD M-GEDIR

Received: May 9, 2020. Revised: June 30, 2020. 91
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
send routing packets. Therefore, ERPFL saves
network bandwidth resources, which contributes to a
higher data transmission ratio. In addition, using
Fuzzy Logic and parameters such as LET and REL as
input metrics, ERPFL shows its efficiency in a high-
density network and high mobility to find a stable and
shorter route, and with a maximum lifetime. Finally,
our proposed protocol decreases the Hop Count and
reduces congestion. In this sense, ERPFL efficiently
distributes data packets and reduces network routing
time.
Conflicts of Interest
The authors declare no conflict of interest.
Author Contributions
Conceptualization, Kandali and Bennis;
methodology, Kandali and Bennis; software,
Kandali; validation, Kandali and Bennis; formal
analysis, Kandali and Bennis; writing—original draft
preparation, Kandali and Bennis; writing—review
and editing, Kandali and Bennis; visualization,
Bennis; supervision, Bennis; project administration,
Bennis.
References
[1] F. Mekelleche and H. Hafid, “Towards the
Development of Vehicular Ad-Hoc Networks
(VANETs): Challenges and Applications”, IoT
and Cloud Computing Advancements in
Vehicular Ad-Hoc Networks, pp. 21-47, 2020.
[2] F. Goudarzi, H. Asgari, and H. S. Al-Raweshidy,
“Traffic-Aware VANET Routing for City
Environments—A Protocol Based on Ant Colony
Optimization”, IEEE Systems Journal, Vol. 13,
No. 1, pp. 571-581, 2019.
[3] D. Lin, J. Kang, A. Squicciarini, Y. Wu, S.
Gurung, and O. Tonguz, “MoZo: A Moving Zone
Based Routing Protocol Using Pure V2V
Communication in VANETs”, IEEE
Transactions on Mobile Computing, Vol. 16, No.
5, pp. 1357-1370, 2017.
[4] C. Ghorai, and I. Banerjee, “A constrained
Delaunay Triangulation based RSUs deployment
strategy to cover a convex region with obstacles
for maximizing communications probability
between V2I”, Vehicular Communications, Vol.
13, pp. 89‑103, 2018.
[5] S. Al-Sultan, M. M. Al-Doori, A. H. Al-Bayatti,
and H. Zedan, “A comprehensive survey on
vehicular Ad Hoc network”, Journal of Network
and Computer Applications, Vol. 37, pp. 380‑392,
2014.
[6] S. Alam, S. Sulistyo, W. Mustika, and R. Adrian,
“Utility-Based Horizontal Handover Decision
Method for Vehicle-to-Vehicle Communication
in VANET”, International Journal of Intelligent
Engineering and Systems, Vol. 13, No. 2, 2020.
[7] K. C. Lee, U. Lee, and M. Gerla, “Survey of
Routing Protocols in Vehicular Ad Hoc
Networks”, Advances in Vehicular Ad-Hoc
Networks: Developments and Challenges, pp.
149-170, 2010.
[8] R. Adrian, S. Sulistyo, W. Mustika, S. Alam,
“ABNC: Adaptive Border Node Clustering
Using Genes Fusion Based on Genetic Algorithm
to Support the Stability of Cluster in VANET”,
International Journal of Intelligent Engineering
and Systems, Vol. 13, No. 1, 2020.
[9] K. Kandali, and H. Bennis, “Performance
Assessment of AODV, DSR and DSDV in an
Urban VANET Scenario”, In: Proc. of Advanced
Intelligent Systems for Sustainable Development
(AI2SD’2018), M. Ezziyyani, Éd. Cham:
Springer International Publishing, Vol. 915, pp.
98‑109, 2019.
[10] K. Kandali, H. Bennis, and M. Benyassi, “A
Novel Route Discovery Mechanism Based on
Neighborhood Broadcasting Methods in
VANET”, In: Proc. of Ezziyyani M. (eds)
Advanced Intelligent Systems for Sustainable
Development (AI2SD’2019). AI2SD 2019.
Lecture Notes in Networks and Systems,
Springer, Cham, Vol. 92, 2020.
[11] H. Fatemidokht and M. K. Rafsanjani, “F-Ant:
an effective routing protocol for ant colony
optimization based on fuzzy logic in vehicular
ad hoc networks”, Neural Computing and
Applications, Vol. 29, No. 11, pp. 1127‑1137,
2018.
[12] G. Li, L. Boukhatem, and J. Wu, “Adaptive
Quality-of-Service-Based Routing for Vehicular
Ad Hoc Networks with Ant Colony
Optimization”, IEEE Transactions on Vehicular
Technology, Vol. 66, No. 4, pp. 3249-3264,
2017.
[13] A. Naushad, G. Abbas, Z.H. Abbas, and L. Jiao,
“A novel faster failure detection strategy for link
connectivity using Hello messaging in mobile ad
hoc networks”, International Journal of Ad Hoc
and Ubiquitous Computing, Vol. 32, No. 2, pp.
75–98, 2019.
[14] S. A. Soleymani, A. H. Abdullah, M. H. Anisi,
A. Altameem, W. H. Hasan, S. Goudarzi, S.
Mandala, Z. B. Razak, and N. M. Noor,
“BRAIN-F: Beacon Rate Adaption Based on
Fuzzy Logic in Vehicular Ad Hoc Network”,

Received: May 9, 2020. Revised: June 30, 2020. 92
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
International Journal of Fuzzy Systems, Vol. 19,
No. 2, pp. 301‑315, 2017.
[15] M. H. Eiza, T. Owens, and Q. Ni, “Secure and
Robust Multi-Constrained QoS Aware Routing
Algorithm for VANETs”, IEEE Transactions on
Dependable and Secure Computing, Vol. 13, No.
1, pp. 32-45, 2016.
[16] F. Azzali, O. Ghazali, and M. H. Omar, “Fuzzy
Logic-based Intelligent Scheme for Enhancing
QoS of Vertical Handover Decision in Vehicular
Ad-hoc Networks”, In: Proc. of IOP Conf.
Series: Materials Science and Engineering, Vol.
226, pp. 012081, 2017.
[17] S. Mittal, S. Bisht, K. C. Purohit, and A. Joshi,
“Improvising-AODV routing protocol by
modifying route discovery mechanism in
VANET”, In: Proc. of 2017 3rd International
Conf. on Advances in Computing,
Communication & Automation (ICACCA) (Fall),
Dehradun, pp. 1–5, 2017.
[18] T. Kabir, N. Nurain, and M. H. Kabir, “Pro-
AODV (Proactive AODV): Simple
modifications to AODV for proactively
minimizing congestion in VANETs”, In: Proc.
of 2015 International Conf. on Networking
Systems and Security (NSysS), Dhaka, pp. 1-6,
2015.
[19] N. Javaid, A. Bibi, S. N. Mohammad, Z. A.
Khan, and N. Alrajeh, “Towards optimising
routing overhead in wireless multi-hop
networks”, International Journal of Ad Hoc and
Ubiquitous Computing, Vol. 19, No. 1, 2, pp. 4–
18, 2015.
[20] S. M. Khan, R. Nilavalan, and A. F. Sallama, “A
Novel Approach for Reliable Route Discovery
in Mobile Ad-Hoc Network”, Wireless Personal
Communications, Vol. 83, No. 2, pp. 1519–
1529, 2015.
[21] M. Wei, K. Wang, and Y. Hsieh, “A reliable
routing scheme based on vehicle moving
similarity for VANETs”, In: Proc. of TENCON
2011 - 2011 IEEE Region 10 Conf. Bali,
Indonesia, pp. 426‑430, 2011.
[22] B. Ramakrishnan, R. B. Nishanth, M. M. Joe,
and R. S. Shaji, “Comprehensive analysis of
Highway, Manhattan and Freeway mobility
models for vehicular ad hoc network”,
International Journal of Wireless and Mobile
Computing, Vol. 9, No. 1, pp. 78–89, 2015.
[23] J. H. Kwon, H. S. Chang, T. Shon, J. J. Jung, and
E. J. Kim, “Neighbor stability-based VANET
clustering for urban vehicular environments”,
The Journal of Supercomputing, Vol. 72, No. 1,
pp. 161‑176, 2016.
[24] E. Gurumoorthi, and A. Ayyasamy, “Cache
agent based location aided routing using
distance and direction for performance
enhancement in VANET”, Telecommunication
Systems, Vol. 73, No. 3, pp. 419‑432, 2020.
[25] T. Lu, S. Chang, and W. Li, “Fog computing
enabling geographic routing for urban area
vehicular network”, Peer-to-Peer Networking
and Applications, Vol. 11, No. 4, pp. 749‑755,
2018.
[26] K. K. Rana, S. Tripathi, and R. S. Raw,
“Analytical Analysis of Improved Directional
Location Added Routing Protocol for
VANETS”, Wireless Personal Communications,
Vol. 98, No. 2, pp. 2403‑2426, 2018.
[27] I. A. Abbasi, A. S. Khan, and S. Ali, “A Reliable
Path Selection and Packet Forwarding Routing
Protocol for Vehicular Ad hoc Networks”,
EURASIP Journal on Wireless Communications
and Networking, Vol. 2018, No. 1, p. 236, 2018.
[28] S. Rahimi and M. A. J. Jamali, “A hybrid
geographic-DTN routing protocol based on
fuzzy logic in vehicular ad hoc networks”, Peer-
to-Peer Networking and Applications, Vol. 12,
No. 1, pp. 88‑101, 2019.
[29] A. N. Hassan, A. H. Abdullah, O. Kaiwartya, Y.
Cao, and D. K. Sheet, “Multi-metric geographic
routing for vehicular ad hoc networks”, Wireless
Networks, Vol. 24, No. 7, pp. 2763‑2779, 2018.
[30] O. Alzamzami and I. Mahgoub, “Fuzzy Logic-
Based Geographic Routing for Urban Vehicular
Networks Using Link Quality and Achievable
Throughput Estimations”, IEEE Transactions
on Intelligent Transportation Systems, Vol. 20,
No. 6, pp. 2289-2300, 2019.
[31] L. Liu, C. Chen, Z. Ren, and C. Shi, “A link
transmission-quality based geographic routing
in Urban VANETs”, In: Proc. of IEEE 28th
Annual International Symposium on Personal,
Indoor, and Mobile Radio Communications
(PIMRC), Montreal, QC, pp. 1–6, 2017.
[32] N. Mantas, M. Louta, K. Katsaros, and S.
Kraounakis, “Social CLWPR: A socially
enhanced position-based routing protocol for
handling misbehaviour in VANETs”, In: Proc.
of 8th International Conf. on Information,
Intelligence, Systems & Applications (IISA),
Larnaca, pp. 1–6, 2017.
[33] R. K. Jaiswal and C. D. Jaidhar, “PPRP:
Predicted Position Based Routing Protocol
Using Kalman Filter for Vehicular Ad-hoc
Network”, In: Proc. of the 18th International
Conf. on Distributed Computing and
Networking - ICDCN ’17, Hyderabad, India, No.
23, pp. 1‑8, 2017.

Received: May 9, 2020. Revised: June 30, 2020. 93
International Journal of Intelligent Engineering and Systems, Vol.13, No.6, 2020 DOI: 10.22266/ijies2020.1231.08
[34] K. Pandey, S. K. Raina, and R. S. Raw,
“Distance and direction-based location aided
multi-hop routing protocol for vehicular ad-hoc
networks”, International Journal of
Communication Networks and Distributed
Systems, Vol. 16, No. 1, pp. 71, 2016.
[35] O. Kaiwartya, and S. Kumar, “Cache agent-
based geocasting in VANETs”, International
Journal of Information and Communication
Technology, Vol. 7, No. 6, p. 562, 2015.
[36] Z. Qin, M. Bai, and D. Ralescu, “A fuzzy control
system with application to production planning
problems”, Information Sciences, Vol. 181, No.
5, pp. 1018‑1027, 2011.
[37] W. Su, S. J. Lee, and M. Gerla, “Mobility
prediction and routing in ad hoc wireless
networks”, International Journal of Network
Management, Vol. 11, No. 1, pp. 3‑30, 2001.