Accepted Manuscript

A robust forwarding node selection mechanism for efficient communication in urban

VANETs

Chinmoy Ghorai, Indrajit Banerjee

PII: S2214-2096(18)30060-3
DOI: https://doi.org/10.1016/j.vehcom.2018.10.003
Reference: VEHCOM 146

To appear in: Vehicular Communications

Received date: 8 March 2018

Revised date: 25 June 2018
Accepted date: 29 October 2018

Please cite this article in press as: C. Ghorai, I. Banerjee, A robust forwarding node selection mechanism for efficient communication in
urban VANETs, Veh. Commun. (2018), https://doi.org/10.1016/j.vehcom.2018.10.003

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing
this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is
published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 A robust forwarding node selection mechanism for
efficient communication in urban VANETs

Chinmoy Ghorai∗, Indrajit Banerjee

Department of Information Technology, Indian Institute of Engineering Science and
Technology, Shibpur, Howrah, West Bengal, India 711103

Abstract
In an urban VANET scenario, more number of obstacles like buildings or other
roadside constructions are present. For these obstacles, the radio propagation
may get affected which leads to poor network performance. To deal with this sit-
uation, a new approach is required which is primarily contingent on an obstacle
avoidance algorithm, capable of finding the optimized shortest path, along with
a potential forwarding vehicle selection strategy to route the packets efficiently.
As a solution, this paper proposes a robust forwarding node selection mechanism
for efficient communication between vehicles. For this, at first a path is selected
avoiding the obstacles with the help of Delaunay Triangulation. Then that path
is optimized by removing Torricelli points, priorly placed inside the Delaunay
Triangulation. After that a forwarding zone is selected in that optimized short-
est path. On that forwarding zone, a robust node is selected using fuzzy logic
for efficient data dissemination. Extensive simulation results disclose that the
proposed solution outperforms the present methods with respect to throughput,
PDR, end-to-end delay and hop-count.
Keywords: Forwarding Node, Delaunay Triangulation, Torricelli Points,
Intelligent Transportation Systems.

1. Introduction

Background. Many researchers and industrial experts envisaged in the area

of vehicular communication in order to develop a better Intelligent Transporta-
tion Systems (ITS) [1, 2, 3]. As a crucial part of ITS, VANETs are anticipated
to play a very significant part in developing the traffic safety by the support
of mutual driving among the vehicles on the road [4, 5, 6, 7, 8, 9]. VANETs
rely on Vehicle-to-anything (V2X) communication for ascertaining a competent
exchange of information among the vehicles (V2V) and roadside units (RSUs)

∗ Corresponding author.

chinmoy@it.iiests.ac.in (Chinmoy Ghorai*)

ibanerjee@it.iiests.ac.in (Indrajit Banerjee)
[10, 11, 12, 13, 14, 15]. In a Vehicular ad-hoc network, the vehicles are arranging
themselves in mobile ad-hoc communication networks without any help of fixed
infrastructural communication [16, 17, 18]. Owing to the high velocity of a large
number of vehicles and recurrent topology changes, routing packets between ve-
hicles and RSUs are becoming an exigent issue [1, 4, 19, 20, 21, 22]. Therefore,
the main confront of ITS is to route the vehicle data and RSU data to a specific
destination or a region of interest with acceptable network performances.

Motivation. Several routing protocols have been developed so far which

address impulsive velocity and movement of vehicles [23]. In an urban VANET
scenario, vehicles typically are not homogeneously dispersed around the City-
road. For these, the network may be fragmented and the competent communica-
tion between V2X get affected [24]. Furthermore, the existence of impediments
such as buildings or other constructions can be the basis of interruption to the
radio signal, leading to failure in communication, even when they are within the
communication zone of each other [25]. Most of the proposed routing protocols
designed for Urban VANETs do not take into account the possible impediments
having a negative impact on the routing, which indicates that direction of data
flow mechanism towards the destination considering the presence of obstacles
in this scenario, is an important research issue. Another important aspect is a
selection of the forwarding zone in that direction [26]. Then, the potential for-
warding node selection on that forwarding zone is another challenge for efficient
data dissemination. Based on channel states and relay location, the transmit-
ter may chosen the next relay node [27] or minimum interference path may be
choosen between source and destination to forward the packet [28].

In summary, routing limitation for VANETs in urban scenario includes:

• No measure for physical obstacles like buildings or other roadside construc-
tions that may affect the radio propagation.
• Lack of an efficient mechanism to forward the data in an appropriate direction
towards the destination which may increase the end-to-end delay.
• Lack of forwarding node selection technique that may lower the network per-
formances.

Contribution. This paper introduces a Novel Data Dissemination Protocol

(NDDP) based on forwarding node selection mechanism for multi-hop communi-
cation. The proposed NDDP has different critical applications of VANETs, like
safety-alert information in emergency situations, instructions for lane changing,
collision cautions, danger-alerts and on-traffic announcements. The proposed
scheme introduced a potential forwarding node selection strategy to forward
the sender’s data to the desired destination. The potential forwarding node is
selected by the sender node itself and aspires to progress multi-hop data dis-
semination over an urban scenario in presence of the obstacles. To lessen the
data loss at the relays, the proposed technique intends to select the potential
forwarding node. The potential forwarding node is selected not only on the dis-
tance from the source node or packet carrying node but also on the potentiality

2
of the forwarding node. Node potentiality is checked by the fuzzy rule where
the three input of the fuzzy logic system is the distance from the source node
or packet carrying node to the next forwarding node and the relative velocity
of the forwarding node and Signal-to-Interference value between them.

In summary, the main contribution of the proposed work is a robust for-

warding node selection strategy for urban VANET that:
• An obstacle avoidance shortest path finding mechanism towards the destina-
tion using Delaunay Triangulation (DT).
• The shortest path is optimized by removing redundant Torricelli point inside
the DT.
• Forwarding zone calculation along that optimized shortest path obtained from
the previous step.
• Fuzzy rule - based best forwarding node selection mechanism is introduced on
that forwarding zone to have the best potential node to forward the data, which
leads to efficient communication between source to destination with lesser delay.
For this three input fuzzy logic system is introduced here, these are - distance,
relative velocity and SIR value between packet carrying node and next forward-
ing node.

Organization. The organization of the paper is as follows: in Section 2 the

related works done so far in this area are discussed. Section 3 gives the problem
description and problem definitions. Proposed algorithm and illustration are
given in section 4. Simulation scenarios and parameter set-up are described
in Section 5. Section 6 discusses the performance evaluation of the proposed
method. Finally, concluding statement emerge in Section 7.

2. Related Study
Due to unpredictable velocity and frequent topology changes of vehicles, a
stable next node selection strategy for data dissemination is a challenging issue
for source vehicle. Conventional MANET routing protocols such as Bellman-
Ford, DSDV or DSR, etc. are no longer useful for vehicular communications due
to the sole characteristics of VANETs. The network outcomes of these routing
protocols may not be satisfactory of V2X communication always. Usually, the
vehicles have to employ a store and carry forward method to send the packets.
That means if a vehicle is carrying a packet and no neighbor node is present
in its communication zone, then the packet is carried by that vehicle until a
neighbor vehicle is found for forwarding it towards the destination. If there is
a set of neighbor vehicles inside the communication zone of the source vehicle
or packet carrying vehicle, then it is very difficult to choose the most potential
vehicle as a next forwarding vehicle among all. Beside this in an urban scenario,
for the presence of obstacles like buildings or other constructions, the efficient
communications affected because of interference and packet collision. Therefore,
the main challenge of source vehicle or packet carrying vehicle is to choose the
forwarding vehicle efficiently so that the packets can reach the destination in an

3
optimized obstacle-free path so that the end-to-end delay becomes very less.

For VANET environment the routing protocols proposed so far can be classi-
fied into the following types. In the first type on demand and multi-path routing
have been developed such as On-demand Multipath Distance Vector (AMODV)
routing protocol [29, 30, 31] that is actually an extended and adapted version
of Ad-hoc On demand Distance Vector (AODV) routing protocol [32]. Second
type is geographic position based routing protocols [33, 34]. In the [33], B.Karp
et al. developed a Greedy Perimeter Stateless Routing where the next forward-
ing node is selected by its location information. In [34], C. Lochert proposed
Greedy Perimeter Coordinate Routing where next forwarding node is selected
by right-hand rule to overcome the drawback of the Greedy Perimeter Stateless
Routing. All the protocols discussed above are applicable for dense vehicle en-
vironment. But in a sparse vehicular scenario, the topology might be detached
most of the time due to non-availability of nearby vehicles. Delay tolerant net-
work routing is applicable for this type of scenario. In the [35, 36, 37, 38, 39, 40]
[41, 42, 43, 44, 45], the authors proposed such delay tolerant routing protocol
to deal with such sparse network.

In the [1], O. Rehman et al. proposed a relay node selection technique for
multi-hop vehicular communication. They select the potential forwarding node
based on link quality and distance. Though they have evaluated their pro-
tocol in a dense area, they did not consider obstacles which are found in an
urban scenario. In [4], O. S. Oubbati et al. proposed two protocols. One is for
ground-to-air communication and another one is for air-to-air communication.
They apply their protocol only for the poorly dense network instead of real
highways and rural environment where more challenges are present. In the [19],
C. F. Wang et al. proposed a next-hop selection based routing protocol for the
heterogeneous network but they did not consider a real vanet scenario to test
their method. In the [20], T. Darwish et al. developed a lightweight-intersection
based routing protocol hinge on vehicular density, distance and road-network
connectivity. But they did not consider their protocol on real-life complicated
urban vanet scenario. In the [21], N. Benamar et al. delay tolerant routing
protocol for vanet, but they did not test their method in complicated urban
vanet scenarios.

In the [46], M. Kadadha et al. proposed a street-centric Quality-of-Service

OLSR routing protocol for urban VANETs, where bandwidth, street, and lane
information is used for selecting MultiPoint relays. The network performance of
the proposed method is satisfactory in comparison with OLSR protocol. But in
the simulation environment, they did not consider physical obstacles like build-
ings or other constructions. In the [47], Q. Ding et al. proposed a traffic light
aware routing protocol for urban VANETs. Where they calculate the street
connectivity by using the distribution and density of vehicles. The performance
of the proposed protocol performs well in respect of delivery ratio and end-to-
end delay. But they did not consider a complicated urban environment in their

4
simulation where obstacles are present. In the [48], M. Boban et al. analyzed
the impact of vehicles as obstacles on a vehicle to vehicle communications in
terms of LOS communications but they did not consider the effect of the other
obstacles like buildings or other roadside objects.

By considering all the limitations of the above related studies, the potential
next node selection in urban vehicular scenarios, where lots of physical obstacles
are present like buildings or other roadside constructions which may affect the
radio propagation for forwarding the data towards the destination seems to be
the challenging issue for efficient communications in urban VANets.

3. Problem Description & Definitions

3.1. Problem Description

In a complex urban vanet scenario, due to the presence of numerous physical
obstacles like buildings or other roadside objects, interference and collision of
packets may affect the communications. To overcome the issues of poor commu-
nication a robust forwarding node selection mechanism in an urban vehicular
scenario for multi-hop communication has been introduced.

The first step of the proposed method is to identify the shortest path and
direction of data forwarding from source node to the destination node using De-
launay Triangulation (DT). After that, the second step of the proposed method
tries to optimize the shortest path by introducing Torricelli point inside the DT.
In third step forwarding zone is calculated along the optimized shortest path.
In the proposed method it is set as 60°. Finally, in the fourth step, a fuzzy-
rule based potential forwarding node selection mechanism is introduced on that
forwarding zone. Here, three input fuzzy logic system is introduced. They are
distance, relative velocity and Signal-to-Interference value. By the above four
steps, the proposed method tries to select a robust forwarding node towards the
destination to have better network performances in urban vanet scenarios.

3.2. Definitions
Definition 1: Delaunay Triangulation (DT). In computational geometry
and mathematics, a DT is a triangulation DT(N) for a set of N-vertices in the
plane such that no vertex is inside the cir-cumcircle of any triangle in DT(N)
(Fig. 1).

5
 Figure 1: Delaunay Triangulation

Definition 2: Torricelli Point. In geometry, Torricelli point of a triangle

is a point at which the entire length from the three vertices of a triangle is
least. To place the Torricelli point of a triangle, the biggest angle of the triangle
must be less than or equal to 120°. (Fig. 2 ). To get the Torricelli point (τ )
of a triangle (ΔABC), at first construct at least two equilateral triangle of any
two chosen side of the original triangle (ΔABC), say ΔEAB and ΔDAC, then
connect a line from new vertex (E) to the reverse vertex (C) of the original
triangle (ΔABC), then draw a line from new vertex (D) to the opposite vertex
(B) of the initial triangle (ΔABC). The two line EC and DB intersect at the
Torricelli point (τ ).






Figure 2: Torricelli Point

Definition 3: Shortest Path. A shortest path is that path which be linked

the two points x and y of a graph H with least distance. Several algorithms are
present to determine shortest path between x and y, such as Dijkstra shortest

6
path algorithm.

Definition 4: Forwarding Area. The forwarding area is a fan - contour

region with a forwarding angle θ towards the destination node.

Definition 5: Signal-to-Interference Ratio (SIR). SIR is the ratio between

the average received modulated carrier power (S) and the average received co-
channel interference power (I), SIR = S/I.

Definition 6: Very-Low Priority Forwarding Node. A very low priority

forwarding node is a node whose distance from source node/packet carrier node
is least, relative velocity is slow and SIR is low or whose distance from source
node/packet carrier node is extreme, relative velocity is fast and SIR is adequate.
This types of nodes are the last choice for forwarding the packets towards the
destination. Because for the first type if the distance is minimum, relative veloc-
ity and SIR is low, the next forwarding node is nearest to the source node/packet
carrying node. If the packet is sent to that forwarding node, hop-count may in-
creases and as a result delay may increases as well as the network performances
may decreases as SIR is low. And for the second type, if the distance is maxi-
mum, relative velocity is fast and SIR is adequate of the next forwarding node,
it may goes outside the transmission zone of the source node/packet carrying
node very fast and as a result packet loss may increases.

Definition 7: Low Priority Forwarding Node. A low priority forwarding

node is a node whose distance from source node/packet carrier node is mini-
mum, relative velocity is optimum and SIR is low, or whose distance from source
node/packet carrier node is moderate, relative velocity is fast and SIR is ade-
quate. Therefore this types of nodes are the second last choice for forwarding
the packets towards the destination. Because for the first type as the distance
is minimum, relative velocity is optimum and SIR is low, the next forwarding
node is nearer to the source node/packet carrying node. If the packet is sent
to that forwarding node, hop-count may increases and as a result delay may
increases. And if the packet is transmitted with that node it may affect the
network performance as the SIR is low. And for the second type, though the
distance is moderate and SIR is adequate but the relative velocity is fast of the
next forwarding node, it may goes outside the transmission zone of the source
node/packet carrying node very fast and as a result packet loss may increases.

Definition 8: Moderate Priority Forwarding Node. A moderate priority

forwarding node is a node whose distance from source node/packet carrier node
is minimum, relative velocity is fast and SIR is adequate, or whose distance
from source node/packet carrier node is moderate, relative velocity is slow and
SIR is adequate. Therefore this type of node is the third choice for forwarding
the packets towards the destination. Because for the first type though the dis-
tance is minimum but the velocity is fast and SIR is low, the node may goes
outside the transmission zone of the source node/packet carrying node and as a

7
result packet loss may increases. And for the second type, though the distance
is moderate and SIR is adequate but the velocity is slow. If the packet is sent
to that forwarding node, hop-count may increases and as a result delay may
increases.

Definition 9: High Priority Forwarding Node. A high priority forwarding

node is a node whose distance from source node/packet carrier node is moder-
ate, relative velocity is optimum and SIR is adequate, or whose distance from
source node/packet carrier node is maximum, relative velocity is optimum and
SIR is high. Therefore this type of node is the second choice for forwarding the
packets towards the destination. Because for the both cases the next forwarding
node may reside inside the transmission zone of the source node/packet carrier
node for some time.

Definition 10: Very-High Priority Forwarding Node. A very-high priority

forwarding node is a node whose distance from source node/packet carrier node
is maximum, relative velocity is slow and SIR is high. Therefore this type
of node is the first choice for forwarding the packets towards the destination.
Because the next forwarding node reside at the furthest distance from source
node/packet carrier node’s transmission zone, for this hop-count may decreases
and as a result delay may decreases.

4. Proposed Algorithm and Illustration

The proposed algorithm is divided into four steps: i) An obstacle avoidance

shortest path finding mechanism towards the destination using Delaunay Tri-
angulation (DT) and Torricelli points, ii) Then shortest path is optimized by
removing redundant Torricelli point inside the DT, iii) Forwarding zone calcu-
lation along that optimized shortest path obtained from the previous step and
finally in iv) Fuzzy rule - based best forwarding node selection mechanism is
introduced on that forwarding zone to have the best potential node to forward
the data, which leads to efficient communication between source to destination
with lesser delay.

4.1. Obstacle avoidance Shortest path finding

This paper proposed an obstacle avoidance shortest path finding technique to
forward the sender vehicle data to the target efficiently over an urban scenario.
There may exist several obstacles in an area which are concave polygon or
convex polygon in shape. In the first step, the proposed method first identify
the presence of obstacles between source and destination from the configuration
file. This configuration file contains the corner points of the simulation area as
well as corner points of the obstacles. The intermittent beacon messages are
disseminated by each vehicle to notify its neighbors with its present position.
By analysing these message, it precisely track the position of the destination
vehicle [49]. Then it will construct a DT by the said vertices (source node,

8
destination node and corner points of the obstacles) as described in Algorithm
1 and shown in fig. 9 (c). After that, removing the edges of that DT that crosses
with the obstacles, the Torricelli points are inserted in the remaining DT that
have angles less than 120°. Then the Torricelli edges are produced by joining
the Torricelli points and their related triangles’ corner points and neighbour
Torricelli points. Finally, the proposed method constructs the connected graph
as described in Algorithm 2 and depicted in fig. 9 (d), 10 (a) & 10 (b). Then
the shortest path Ps is calculated in graph H using Ahuja-Dijkstra algorithm.
From the input set of points S, a point a is selected and search in S for one of
the nearest points b of a. Then one side of the DT is construct with ab. Then
find another point c according to the DT property to form the triangle Δabc as
depicted in figure 3.

 



 

Figure 3: A Delaunay Triangle Δ ABC

The area of triangle abc(A):

2 2
−r 2
cosη = p +q2pq & R = pqr
4A
Let τ be a Δabc where a = (x1 , y1 ), b = (x2 , y2 ) and c = (x3 , y3 ). Then
A(τ ) = 12 | D(a, b, c) |, where D(a, b, c) is the determinant.
Points a, b, c of S form a Delaunay Triangulation if the following is highest
for c amongst all the points of S \ {a, b}:
2 2
−r 2 )
R.cosη = c(p +q 8.A

c is constant for a given a & b, maximization is needed for (p2 + q 2 − r2 )/A.

Let c ∈ S \ {a, b} be the first point for which D(a, b, c) = 0. If D(a, b, c) > 0,
2
+q 2 −r 2
then the third point is identified by maximizing K(a, b, c) = pD(a,b,c) .

Theorem 1. The complexity of obstacle avoidance shortest-route-finding method

takes O(nlogn) time, where n states the number of obstacles.

Proof. The complexity of step-2 is O(nlogn), which is basically due to the

formation of Delaunay Triangulation. Edge removal is done in step-3 which

9
Algorithm 1: Delaunay Triangulation Algorithm
Input : S = {Pi , Pt , Po }
Pi = Source node
Pt = Destination node
Po = Set of Obstacles’ corner point
Output: Delaunay Triangulation: DT
1 begin
2 Choose a ∈ Pi and locate a nearest point of a, b ∈ S ;
3 Locate a third point c ∈ S \ {a, b} that maximizes K(a, b, c). Place the
 and ac,
vectors ba  in a list;
 cb
4 Since the above list is not exhausted, goto step 5, stop otherwise ;
5 Get pq off the list, set flag = 0, thirdpoint = 0, k = −∞ ;
6 For all c ∈ S \ {a, b} (goto step 6 once completed):;
7 a) if k  K(a, b, c) for increment, goto step 6.c ;
8 b) set flag = 1, K = K(a, b, c) and thirdpoint = c ;
9 c)for the next point back to the step 6 c ∈ S \ {a, b} ;
10 If flag = 0, goto step 4, otherwise c=thirdpoint ;
11 a) if ac
 is in the list, remove it; then place ac
 on the file ;
12  is in the list, remove it; then place cb
b) if cb  on the file ;
13 Goto step 4;
14 end

Algorithm 2: Obstacle Avoidance Shortest Path Algorithm

Input : Source node: Pi
Destination node: Pt
Set of Obstacles’ corner point: Po
Output: Shortest path: Ps
1 begin
2 Call the function: Delaunay Triangulation (DT) on the set of source
node Pi using algorithm 1, where destination node Pt and a set of
obstacles’ corner point Po in Euclidean plane.;
3 Remove the edges of DT that cross with the obstacles present in the
area;
4 Torricelli points are placed in that triangles which have angles lesser
120°;
5 Join Torricelli points with corner points of the corresponding triangles
and neighbour Torricelli points;
6 All obstacle edges are inserted into the graph H;
7 Find-out the shortest path from the graph H using Ahuja-Dijkstra
algorithm;
8 Call the function: path− optimization() (algorithm 3);
9 end

10
 Algorithm 3: Optimized Shortest Path Finding Algorithm:
path− optimization()
Input : Initial point: Pi
Target point: Pt
Set of Obstacles’ corner point: Po
Torricelli points: PT
Output: Optimized path: PO
1 begin
2 Delete the unnecessary Torricelli points;
3 Create the short-cut path of any two successive section;
4 Delete the short-cut path that crosses any obstacle;
5 Join the short-cuts by their corresponding length enhancement;
6 Abridge the initial-path in a descending manner if permitted;
7 end

dominated by O(nlogn). On the basis of Euler formula, the maximum number

of triangles is 2X − 2 − p and maximum number of edges is 3X − 3 − p in the
triangulation process. So, X  2 + (kx × n) = O(n) = O(n), where, X denotes
the number of vertices in a graph, p is the number of points in the convex area,
kx is the obstacles corner points. So, the formations of the Torricelli points and
the number of triangles are circumscribed by O(n). In step-5, the complexity
of edge-connections take 6 × O(n) times. Then in step-6, all the obstacle edges
are inserted in the graph. So, the complexity of step-4 to step-6 is O(n). So the
complexity of step-7 is O(e + XlogX), where X is the number of vertices and e
is the number of edges. That is, e = ((3X − 3 − p) + (6 × O(n))) = 3 × O(n) + 6 ×
O(n) = O(n). Therefore, O(e + XlogX) = O(n + nlogn) = O(nlogn). So, the
time complexity of the proposed obstacle avoidance shortest route formation
from source to destination algorithm is O(nlogn), where n states the number of
obstacles.

Theorem 2. The space-complexity of obstacle avoidance shortest-path-finding

method is O(n), where n states the number of obstacles.

Proof. After the formation of DT, the number of triangles (m) are less than
δ = 2 × (kx × n) − p − 2, where, kx is the corner-points of each obstacle, n
specifies the number of obstacles, p states the points in the convex area. So, the
space complexity is enclosed by O(n).

4.2. Optimized shortest route finding

The proposed algorithm tries to optimize the shortest route between a source
and destination. At first the unnecessary Torricelli points are deleted. Then a
short-cut path of any two successive sections is created. Then all the paths
which cross any obstacles are deleted. After that, the short-cuts are joined by
their corresponding length enhancement. And finally, abridge the original path
in a descending manner, if permissible (Algorithm 3, fig. 10 (c)).

11
4.3. Forwarding Area Calculation
The forwarding-zone of the proposed technique is shown in figure 4 as pro-
posed in [50] by K. Husain et al. The forwarding zone is a fan - contour region
towards the destination node with a forwarding angle θ. Here the proposed
method consider θ = 60°. As shown in figure 4, node S and D represents the
source node or packet carrying node and destination node or next packet for-
warding node respectively. The basis for choosing θ angle as 60° is to make
assured that all nodes in the forwarding zone are within the transmission region
of each other that holds precise due to the property of equilateral triangle. In
Figure 4, node A and node B are the furthest from each other with a distance
r.










 



Figure 4: Forwarding sector of the proposed method

The angle θ is calculated in [50] as follows:

xr − xr+1
θ = 2 × |arccos  | (1)
(xr − xr+1 )2 + (yr − yr+1 )2

where, (xr , yr ) and (xr+1 , yr+1 ) are the coordinate position of the receiving node
and previous packet carrying node respectively.

4.4. Fuzzy rule - based potential forwarding node selection

Here, forwarding node selection is carried out by fuzzy logic rule that is
subsequently illustrated in the subsequent section. Figure 5 shows the three
input fuzzy system, one is distance, second one is relative velocity and third one
is SIR value between the packet carrying vehicle and next forwarding vehicle.
Basically, the fuzzy method is divided in 4 phases: (1) Fuzzifier, (2) Fuzzy rule,
(3) Inference, and (4) Defuzzifier. The membership function change the input
into linguistic-variables in the fuzzifier phase. The fuzzifier output is send to
the inference engine. The output of the fuzzifier is revealed in Table 1. The
defuzzifier accumulates cumulative value from inference engine and makes a
non-fuzzy control which presents the types of forwarding nodes, such as - very
low priority node, low priority node, moderate priority node, high priority node
and very high priority node. This classification of the forwarding node decision

12
 If Then
Distance Relative Velocity SIR Ratio Forwarding Node Decision
0. minimum slow low very-low
1. minimum optimum low low
2. minimum fast low moderate
3. moderate slow adequate moderate
4. moderate optimum adequate high
5. moderate fast adequate low
6. maximum slow high very-high
7. maximum optimum high high
8. maximum fast high very-low

Table 1: Fuzzy rule based forwarding node selection scheme

is based on their relative velocity , distance and SIR value from the packet
carrying vehicle as described in the definition section.

  


      



 
 



  
  
    
  

Figure 5: Fuzzy rule based forwarding node selection of the proposed scheme

The input variables are:

DISTANCE, RELATIVE-VELOCITY and Signal-to-interference (SIR) ratio.
The output of the fuzzy system classifies all forwarding nodes into five, as:
V ery− Low, Low, M oderate, High and V ery− High as shown in Table 1.
The labels of input variables are as follows:
DIST AN CE = {M inimum, M oderate, M aximum}.
RELAT IV E− V ELOCIT Y = {Slow, Optimum, F ast}.
SIR = {Low, Adequate, High}.

The output labels are as follows:

F ORW ARDIN G− N ODE− DECISION =
{V ery− Low, Low, Adequate, High, V ery− High}.

The proposed method used triangulation function for reducing the compu-
tational complexity. The triangulation functions for distance, relative velocity
and SIR value between packet carrying node and forwarding node are defined

13
in equation 2 - 10.
⎧
⎪
⎪ 0, xa
⎨ x−a
a<xm
m−a ,
(μRL (x)) = (2)
⎪
⎪
b−x
b−m , m<x<b
⎩
0, xb

Here, lower limit is a, upper limit is b and a value m, where a < m < b. The
minimum distance is denoted by μRL .
⎧
⎪
⎪ 0, xc
⎨ x−c , c<xn
n−c
(μRM (x)) = (3)
⎪
⎪
d−x
, n<x<d
⎩ d−n
0, xd
Here, lower limit is c, upper limit is d and a value n, where c < n < d. The
moderate distance is denoted by μRM .
⎧
⎪ 0, xe
⎪
⎨ x−e ,
o−e e<xo
(μRU (x)) = f −x (4)
⎪
⎪ , o<x<f
⎩ −of
0, xf
Here, lower limit is e, upper limit is f and a value o, where e < o < f . The
maximum distance is denoted by μRU .

The triangulation functions for ralative velocity of forwarding node are de-
fined as follows: ⎧
⎪
⎪ 0, xa
⎨ x−a , a<xm
m−a
(μV S (x)) = (5)
⎪
⎪
b−x
, m<x<b
⎩ b−m
0, xb
Here, lower limit is a, upper limit is b and a value m, where a < m < b. The
slow velocity is denoted by μV S .
⎧
⎪
⎪ 0, xc
⎨ x−c , c<xn
n−c
(μV O (x)) = (6)
⎪
⎪
d−x
, n<x<d
⎩ d−n
0, xd
Here, lower limit is c, upper limit is d and a value n, where c < n < d. The
optimum velocity is denoted by μV O .
⎧
⎪ 0, xe
⎪
⎨ x−e ,
o−e e<xo
(μV F (x)) = f −x (7)
⎪
⎪ , o<x<f
⎩ −o
f
0, xf

14
Here, lower limit is e, upper limit is f and a value o, where e < o < f . The fast
velocity is denoted by μV F .
The triangulation functions for SIR value of forwarding node are defined as
follows: ⎧
⎪ x−a0, x  a
⎪
⎨
m−a , a < x  m
(μSL (x)) = (8)
⎪
⎪
b−x
, m<x<b
⎩ b−m
0, x  b
Here, lower limit is a, upper limit is b and a value m, where a < m < b. The
slow velocity is denoted by μSL .
⎧
⎪
⎪ 0, x  c
⎨ x−c , c < x  n
n−c
(μSA (x)) = (9)
⎪
⎪
d−x
, n<x<d
⎩ d−n
0, x  d
Here, lower limit is c, upper limit is d and a value n, where c < n < d. The
optimum velocity is denoted by μSA .
⎧
⎪ 0, x  e
⎪
⎨ x−e , e < x  o
o−e
(μSH (x)) = f −x (10)
⎪
⎪ , o<x<f
⎩ −of
0, x  f
Here, lower limit is e, upper limit is f and a value o, where e < o < f . The fast
velocity is denoted by μSH .
The inference engine output is categorised into very low priority node, low
priority node, adequate priority node, high priority node and very high priority
node according to the position of the forwarding node. The proposed scheme
computes node decision by if-then rules. The proposed fuzzy system uses Mam-
dani fuzzy controller based inference system and the output (crisp) is achieved
by the Defuzzification method center of area. The membership functions are
symbolically defined in equation 11 - 15.
⎧
⎪ 0, (b  f  x  a  e)
⎪
⎪
⎪
⎨
x−a
m−a , a<xm
b−x
(μvery−low−priority (x)) = b−m , m<x<b (11)
⎪
⎪ x−e
e<xo
⎪
⎪ o−e ,
⎩ f −x
f −o , o<x<f
⎧
⎪
⎪ 0, (b  d  f  x  a  c  e)
⎪
⎪
⎪
⎪
x−a
m−a , a<xm
⎪
⎪
⎪
⎨
b−x
b−m , m<x<b
(μlow−priority (x)) =
x−c
n−c , c<xn (12)
⎪
⎪ d−x
⎪
⎪ d−n , n<x<d
⎪
⎪
⎪
⎪
x−e
o−e , e<xo
⎪
⎩ f −x
f −o , o<x<f

15
 ⎧
⎪
⎪ 0, (b  d  f  x  a  c  e)
⎪
⎪
⎪
⎪
x−a
m−a , a<xm
⎪
⎪
⎪
⎨
b−x
b−m , m<x<b
(μmoderate−priority (x)) =
x−c
n−c , c<xn (13)
⎪
⎪ d−x
⎪
⎪ d−n , n<x<d
⎪
⎪
⎪
⎪
x−e
o−e , e<xo
⎪
⎩ f −x
f −o , o<x<f
⎧
⎪ 0, (d  f  x  c  e)
⎪
⎪
⎪
⎨
x−c
n−c , c<xn
d−x
(μhigh−priority (x)) = d−n , n<x<d (14)
⎪
⎪ x−e
e<xo
⎪
⎪ o−e ,
⎩ f −x
f −o , o<x<f
⎧
⎪ 0, (f  b  x  a  e)
⎪
⎪
⎪
⎨
x−a
m−a , a<xm
b−x
(μvery−high−priority (x)) = b−m , m<x<b (15)
⎪
⎪ x−e
e<xo
⎪
⎪ o−e ,
⎩ f −x
f −o , o<x<f
In the proposed scheme, available forwarding node set N is divided into var-
ious types according to their velocity and distance form the previous forwarding
node or last packet carrying node. The various types of nodes are very-low
priority nodes, low priority nodes, adequate priority nodes, high priority nodes
and very-high priority nodes. By these differentiation, the packet carrying node
select the most eligible or potential forwarding node to send the data towards
the destination. Figure 8 highlights the steps involved for a packet carrying ve-
hicle during data transmission. Pictorial representation of the proposed method
is given in Figure 6 and 7.

16
 (a) Initial Area







(b) Insert Obstacles boundary points







(c) Delaunay Triangulation



17





(d) Edge Deletion

Figure 6: Pictorial representation of the proposed method (step 1)








(a) Insert Torricelli Points







(b) Shortest-path finding







(c) Optimization of the Shortest-path

18

(d) Forwarding Zone

Figure 7: Pictorial representation of the proposed method (step 2)





  



 

 


  
 



 

   
  

      


 

 


 
   
 


 

 


        




 

     



 

 

  
 

  


 

 

 



 



 
  
 

  

 

  

 
 
  
 
  


 
  

  


  

 

Figure 8: The steps of a packet carrying vehicle during data diffusion

5. Simulation Scenario and Parameter Set-up

5.1. Simulation parameter setting
The simulation parameters are given in Table 2. IEEE 802.11p is considered
at the MAC layer. Mobile IPv4 is used at the network layer. For simulation
the number of vehicles are varied from 50-200 with a step length 50 and with 30
seconds of traffic period, that reflects the network density. The vehicle velocity
is varies from 30-60 km/h, that follows a normal distribution [47]. The packet
size varies from 256 bytes to 1024 bytes with a fixed number of packets 50. To
be precise the prediction omni-directional antenna model with a transmission

19
 Parameter Values
Simulator EXata 5.4
Area of Map 3000m x 3000m
Topology Urban
No of Vehicles 50, 100, 150, 200
Vehicle Velocity 30, 40, 50, 60 (km/h)
Packet Size 256, 512, 1024 (bytes)
No of Packets per Sec 50
Physical Layer Radio IEEE 802.11p
Network protocol Mobile IPv4
Routing protocol AODV, DYMO, LAR, SMPR, NDDP
Transport protocol UDP
Traffic Type CBR
Antenna Model Omni directional
Transmission Range 250m
Temperature 290 k
SNR Threshold 4 dBm
Channel Frequency 5.9 GHz
Pathloss-model Two Ray
Shadowing-Model Lognormal
Fading-Model Ricean
Simulation Time 500 s
Table 2: Simulation parameters used

range of 250 m [51], two-ray pathloss model, lognormal-shadowing model and

ricean-fading model with channel frequency 5.9 GHz are considered for the simu-
lation. The two-ray pathloss model is more appropriate here than the free space
model as it calculates both the direct path and the ground reflection path. The
temperature is fixed to 290k and signal-to-noise ratio (SNR) threshold is set to
4dBm. The Log-normal Shadowing model is used here as it is suitable for the
proposed scenarios which consider both the vehicle mobility and the obstacles
of the urban environment. The constant bit rate (CBR) data traffic and user
datagram protocol (UDP) are applied at the application layer and the transport
layer respectively. Here, the traffic light period is fixed at 30 seconds.

6. Performance Analysis
In order to assess the proposed method, AODV, DYMO, LAR and SMRP
protocols [19] are considered as the baseline algorithms. EXataCyber-5.4 Net-
work Simulator has been extensively used to evaluate the proposed technique.
An open source micro traffic simulator MOVE is used for mobility model.
The output of the MOVE is a real life mobility model which is fed into the
EXataCyber-5.4 network simulator for evaluation of the said protocol. For this
simulation a real road scenario of Salt Lake Bidhannagar area of Kolkata, In-
dia is chosen [52]. To evaluate the performance of the proposed method with

20
the baseline algorithms, four network evaluation factors are considered. Each
simulation instance is simulated 5 times and then calculate the average of the
resulting values.

6.1. Impact of vehicle number

The performance analysis for the throughput comparisons are shown in Fig.
9. The proposed method has outperformed the other conventional routing pro-
tocols: LAR, DYMO, AODV and SMRP [19] when the number of vehicles
varied under this simulation environment. That means it delivers more number
of packets compared to the above mentioned protocols. With respect to the
throughput, the simulation results of the proposed method shows an improve-
ment of about 14%, 28%, 36% and 66% as compared with the existing protocols
SMRP [19], LAR, AODV and DYMO respectively. The DYMO protocol has
the worst performance. With an increasing number of vehicles, the throughput
increases for all the five routing protocols. When the vehicle density is sparse,
the connectivity mainly affects the throughput. In this proposed method, ro-
bust neighbour vehicle selection leads to lower the packet loss which leads to
the higher throughput.

4
Throughput (mbps)

50 100 150 200

Number of vehicles
LAR
DY M O
AODV
SM RP
N DDP

Figure 9: Throughput vs number of vehicles

Fig. 10 shows the variation of PDR of the four baseline algorithm along with
the proposed algorithm with varying the vehicle size. With aspect to the packet
delivery ratio, the simulation results of the proposed method shows an improve-
ment of about 8%, 12%, 16% and 21% as compared with the existing protocols
SMRP [19], LAR, AODV and DYMO respectively. As the number of vehicles

21
increases from 50 to 200, the PDR also increases significantly. Because in the
dense network, the connectivity is high which leads to significant improvement
in PDR.

80
Packet Delivery Ratio (%)
70

60

50

40

30

20

50 100 150 200

Number of vehicles
LAR
DY M O
AODV
SM RP
N DDP

Figure 10: Packet Delivery Ratio vs number of vehicles

The variation of average end-to-end delay of the proposed protocol and the
four baseline protocols varying the number of vehicles is shown in Fig. 11. The
proposed protocol takes less time to deliver the packet to the target destination
compared to the other four protocols. With respect to the average end-to-end
delay, the simulation results of the proposed method shows an improvement of
about 23%, 45%, 63% and 50% as compared with the existing protocols SMRP
[19], LAR, AODV and DYMO respectively. AODV protocol has the worst per-
formance. As the number of vehicles increases, the end-to-end delay decreases
significantly due to efficient connectivity beteen vehicles. The proposed proto-
col performs well because in this method it tries to send the packet towards
the target destination in an optimized path. Furthermore the packet carrying
vehicle choose the best potential vehicle to send it’s data to the next forwarding
node i.e., the robust forwarding node selection in a directed way reduces the
end-to-end delay significantly.

Fig.12 shows the variation of hop-count of the four baseline algorithm and
the proposed algorithm varying the number of vehicles. With respect to the hop-
count, the simulation results of the proposed method shows an improvement of
about 8%, 16%, 24% and 27% as compared with the existing protocols SMRP
[19], LAR, AODV and DYMO respectively. The hop-count of the proposed

22
 0.8

Average end-to-end delay (s)

0.7

0.6

0.5

0.4

0.3

50 100 150 200

Number of vehicles
LAR
DY M O
AODV
SM RP
N DDP

Figure 11: Average end-to-end delay vs number of vehicles

Average number of hops

30

20

50 100 150 200

Number of vehicles
LAR
DY M O
AODV
SM RP
N DDP

Figure 12: Average number of hops vs number of vehicles

23
method is lesser than the four baseline protocol because it choose the best
potential forwarding node to route it’s packet to the target destination.

6.2. Impact of vehicle velocity

Throughput (mbps)

1
30 40 50 60
Vehicle Velocity (km/h)
LAR
DY M O
AODV
SM RP
N DDP

Figure 13: Throughput vs vehicle velocity

The performance analysis by varying the vehicle velocity for the throughput
comparisons are shown in Fig. 13. The proposed method has outperformed the
other conventional routing protocols: LAR, DYMO, AODV and SMRP [19] as
the number of vehicles varied under this simulation environment. That means
it delivers more number of packets compared to the above mentioned protocols.
Here, the vehicle velocity is varied from 30km/h to 60km/h with a step lenth
of 10. Simulation result shows that the throughput is highest when the vehicle
speed is 50km/h. The reason behind this nature of throughput may be that
in 50km/h vehicle speed, the packet carrying vehicle finds the best potential
forwarding node to forward it’s data. With respect to the throughput, the
simulation results of the proposed method shows an improvement of about 8%,
16%, 26% and 42% as compared with the existing protocols SMRP [19], LAR,
AODV and DYMO respectively.
Fig.14 shows the variation of packet delivery ratio of the four baseline algo-
rithm and the proposed algorithm varying the velocity of vehicles. With respect
to the packet delivery ratio, the simulation results of the proposed method shows
an improvement of about 9%, 14%, 24% and 28% as compared with the existing
protocols SMRP [19], LAR, AODV and DYMO respectively. As the velocity of

24
 80

Packet Delivery Ratio (%)

70

60

50

40

30

20

30 40 50 60
Vehicle Velocity (km/h)
LAR
DY M O
AODV
SM RP
N DDP

Figure 14: Packet delivery ratio vs vehicle velocity

vehicle increases, the packet delivery ratio also increases. But in 50km/h ve-
locity, the packet delivery ratio of the proposed method is maximum. Because
beyond this speed the packet carrying vehicle fails to communicate with the
efficient forwarding vehicle.
The variation of average end-to-end delay of the proposed protocol and the
four baseline protocols varying the velocity of vehicles is shown in Fig. 15. The
proposed protocol takes less time to deliver the packet to the target destination
compared to the other four protocols. With respect to the average end-to-end
delay, the simulation results of the proposed method shows an improvement of
about 22%, 40%, 63% and 45% as compared with the existing protocols SMRP
[19], LAR, AODV and DYMO respectively. AODV protocol has the worst
performance. The proposed protocol performs well because in this method it
tries to send the packet towards the target destination in an optimized path.
Furthermore the packet carrying vehicle choose the best potential vehicle to send
it’s data to the next forwarding node. The end-to-end delay of the proposed
protocol is found to be minimum when the vehicle’s speed is 50km/h. As the
velocity of the vehicles increases the end-to-end delay decreases. But beyond
50 km/h speed, the end-to-end delay slightly increases. The reason behind this
nature of end-to-end delay may be that it is hard to find out efficient forwarding
vehicle by the packet carrying vehicle.
Fig.16 shows the variation of hop-count of the four baseline algorithms and
the proposed algorithm, varying the speed of vehicles. With respect to the
hop-count, the simulation results of the proposed method takes less number of

25
 0.8

Average end-to-end delay (s)

0.7

0.6

0.5

0.4

0.3

0.2
30 40 50 60
Vehicle Velocity (km/h)
LAR
DY M O
AODV
SM RP
N DDP

Figure 15: Average end-to-end delay vs vehicle velocity

Average number of hops

30

20

10
30 40 50 60
Vehicle Velocity (km/h)
LAR
DY M O
AODV
SM RP
N DDP

Figure 16: Average number of hops vs vehicle velocity

26
hops for all the cases as compared with the existing SMRP [19] protocol and
other three conventional protocols AODV, DYMO and LAR as it chooses the
best potential forwarding node to route it’s packet to the target destination.
With respect to the hop-count, the simulation results of the proposed method
shows an improvement of about 9%, 21%, 23% and 38% as compared with the
existing protocols SMRP [19], LAR, AODV and DYMO respectively. As the
vehicle speed increases, the hop-count decreases. But, the hop-count is found to
be minimum when the vehicle is in the speed of 50km/h. Beyond that speed the
hop-count increases. The vehicle to vehicle distance increases with increasing
velocity. Beyond a distance limit, it is hard to find out stable forwarding vehicle.
For this reason, beyond 50 km/h the hop-count increases.

7. Conclusion

Conventional routing protocols are performing poor for urban vehicular envi-
ronment due to the presence of more obstacles in urban areas. Therefore, an ap-
propriate routing protocol design is an exigent area of research. This paper has
proposed a robust forwarding node selection technique for efficient data dissem-
ination in urban vehicular ad-hoc networks, aiming to improve the throughput
and packet delivery ratio. Firstly, an optimized shortest path finding technique
is introduced from source to destination using Delaunay Triangulation, Torri-
celli points and Dijkstra algorithm. Then a forwarding angle is applied in that
path to identify the best forwarding zone. Finally, fuzzy logic is applied to find
out the best potential forwarding node within that forwarding zone to forward
the packets efficiently. The proposed protocol shows an improvement of about
9% in packet delivery rate, about 11% improvement in throughput, about 9%
reduction in hop-count and about 23% reduction in end-to-end delay as com-
pared to the existing methods. Surprisingly, from simulation results, it has been
found that at 50km/h speed, the proposed protocol gives the best performance
and beyond that speed the network performances are decreasing. Because, if
the vehicles are moving faster than 50km/h, it fails to forward the packet to the
efficient neighbour vehicle due to high vehicle movement. As future continuation
of this work, it would be suggested to emphasize on investigating more efficient
forwarding zone selection mechanism with minimum delay for efficient packet
routing in urban environment. There are two limitations in this proposed work
- i. the obstacle’s corner points have assumed, and ii. the vehicles have not
been considered hare as obstacles.

Acknowledgement
The author* would like to thank University Grants Commission, Government
of India for providing financial support.

27
References

[1] Osama Rehman, Mohamed Ould-Khaoua, and Hadj Bourdoucen. An adap-

tive relay nodes selection scheme for multi-hop broadcast in vanets. Com-
puter Communications, 87:76 – 90, 2016.
[2] Sourav Kumar Bhoi and Pabitra Mohan Khilar. Sir: a secure and intelligent
routing protocol for vehicular ad hoc network. IET Networks, 4:185194,
2015.

[3] F. Outay, F. Kammoun, F. Kaisser, and M. Atiquzzaman. Towards safer

roads through cooperative hazard awareness and avoidance in connected
vehicles. In 2017 31st International Conference on Advanced Information
Networking and Applications Workshops (WAINA), pages 208–215, March
2017.

[4] Omar Sami Oubbati, Abderrahmane Lakas, Fen Zhou, Mesut Gne, Nasred-
dine Lagraa, and Mohamed Bachir Yagoubi. Intelligent uav-assisted routing
protocol for urban {VANETs}. Computer Communications, 107:93 – 111,
2017.
[5] G. Marfia, M. Roccetti, A. Amoroso, and G. Pau. Safe driving in la:
Report from the greatest intervehicular accident detection test ever. IEEE
Transactions on Vehicular Technology, 62(2):522–535, Feb 2013.
[6] C. E. Palazzi, M. Roccetti, and S. Ferretti. An intervehicular communi-
cation architecture for safety and entertainment. IEEE Transactions on
Intelligent Transportation Systems, 11(1):90–99, March 2010.

[7] W. Ben Jaballah, M. Conti, M. Mosbah, and C. E. Palazzi. Fast and se-
cure multihop broadcast solutions for intervehicular communication. IEEE
Transactions on Intelligent Transportation Systems, 15(1):433–450, Feb
2014.

[8] Chinmoy Ghorai and Indrajit Banerjee. A novel priority based exigent data
diffusion approach for urban vanets. In Proceedings of the 18th Interna-
tional Conference on Distributed Computing and Networking, ICDCN ’17,
pages 24:1–24:10, New York, NY, USA, 2017. ACM.
[9] C. Ghorai and I. Banerjee. A multi-objective data dissemination proto-
col for intelligent transportation systems. In 2017 IEEE 7th International
Advance Computing Conference (IACC), pages 144–149, Jan 2017.
[10] Fernando A. Teixeira, Vinicius F. e Silva, Jesse L. Leoni, Daniel F. Macedo,
and Jos M.S. Nogueira. Vehicular networks using the ieee 802.11p standard:
An experimental analysis. Vehicular Communications, 1(2):91 – 96, 2014.

[11] Sourav Kumar Bhoi and Pabitra Mohan Khilar. Vehicular communication:
a survey. IET Networks, 3:204–217, 2014.

28
[12] Khaled Alodadi, Ali H. Al-Bayatti, and Nasser Alalwan. Cooperative vol-
unteer protocol to detect non-line of sight nodes in vehicular ad hoc net-
works. Vehicular Communications, 9:72 – 82, 2017.

[13] Satya S. Karanki and Mohammad S. Khan. Smmv: Secure multimedia de-
livery in vehicles using roadside infrastructure. Vehicular Communications,
7:40 – 50, 2017.

[14] J. Sahoo, S. Cherkaoui, A. Hafid, and P. K. Sahu. Dynamic hierarchical

aggregation for vehicular sensing. IEEE Transactions on Intelligent Trans-
portation Systems, 18(9):2539–2556, Sept 2017.
[15] Mohammad A. Salahuddin, Ala I. Al-Fuqaha, Mohsen Guizani, and
Soumaya Cherkaoui. RSU cloud and its resource management in support
of enhanced vehicular applications. CoRR, abs/1706.06921, 2017.

[16] Alessandro Amoroso, Gustavo Marfia, and Marco Roccetti. Going realistic
and optimal: A distributed multi-hop broadcast algorithm for vehicular
safety. Computer Networks, 55(10):2504 – 2519, 2011.
[17] Cristiano M. Silva, Fabricio A. Silva, Joo F.M. Sarubbi, Thiago R. Oliveira,
Wagner Meira, and Jose Marcos S. Nogueira. Designing mobile content
delivery networks for the internet of vehicles. Vehicular Communications,
8:45 – 55, 2017. Internet of Vehicles.

[18] Maha Kadadha, Hadi Otrok, Hassan Barada, Mahmoud Al-Qutayri, and
Yousof Al-Hammadi. A stackelberg game for street-centric qos-olsr protocol
in urban vehicular ad hoc networks. Vehicular Communications, 13:64 –
77, 2018.
[19] Chu-Fu Wang, Yan-Ping Chiou, and Guan-Hsiung Liaw. Nexthop selec-
tion mechanism for nodes with heterogeneous transmission range in vanets.
Computer Communications, 55:22 – 31, 2015.

[20] Tasneem Darwish and Kamalrulnizam Abu Bakar. Lightweight

intersection-based traffic aware routing in urban vehicular networks. Com-
puter Communications, 87:60 – 75, 2016.

[21] Nabil Benamar, Kamal D. Singh, Maria Benamar, Driss El Ouadghiri, and
Jean-Marie Bonnin. Routing protocols in vehicular delay tolerant networks:
A comprehensive survey. Computer Communications, 48:141 – 158, 2014.
Opportunistic networks.

[22] Julio A. Sanguesa, Manuel Fogue, Piedad Garrido, Francisco J. Martinez,

Juan-Carlos Cano, Carlos T. Calafate, and Pietro Manzoni. Rtad: A real-
time adaptive dissemination system for vanets. Computer Communications,
60:53 – 70, 2015.

29
[23] J. Harri, F. Filali, and C. Bonnet. Mobility models for vehicular ad hoc net-
works: a survey and taxonomy. IEEE Communications Surveys Tutorials,
11(4):19–41, Fourth 2009.

[24] Kayhan Zrar Ghafoor and Marwan Aziz Mohammed. Routing protocols
in vehicular ad hoc networks: Survey and research challenges. Network
Protocols and Algorithms, 5(4):39–83, 2013.

[25] C. Sommer, D. Eckhoff, R. German, and F. Dressler. A computation-

ally inexpensive empirical model of ieee 802.11p radio shadowing in urban
environments. In 2011 Eighth International Conference on Wireless On-
Demand Network Systems and Services, pages 84–90, Jan 2011.

[26] Sang-woo Chang and Sang-sun Lee. Distance-based stable routing decision
scheme in urban vehicular ad hoc networks. Int. J. Distrib. Sen. Netw.,
2015:3:3–3:3, January 2015.

[27] Y. Richter and I. Bergel. Optimal and suboptimal routing based on partial
csi in random ad-hoc networks. IEEE Transactions on Wireless Commu-
nications, 17(4):2815–2826, April 2018.
[28] P. Fazio, F. De Rango, and C. Sottile. A predictive cross-layered interfer-
ence management in a multichannel mac with reactive routing in vanet.
IEEE Transactions on Mobile Computing, 15(8):1850–1862, Aug 2016.

[29] M. K. Marina and S. R. Das. On-demand multipath distance vector routing

in ad hoc networks. In Proceedings Ninth International Conference on
Network Protocols. ICNP 2001, pages 14–23, Nov 2001.

[30] Mohammed Tarique, Kemal E. Tepe, Sasan Adibi, and Shervin Erfani.
Survey of multipath routing protocols for mobile ad hoc networks. Journal
of Network and Computer Applications, 32(6):1125 – 1143, 2009.
[31] N. Z. Ali, R. B. Ahmad, and S. A. Aljunid. A survey on on-demand
multipath routing protocol in manets. In 2008 International Conference
on Electronic Design, pages 1–4, Dec 2008.
[32] C. E. Perkins and E. M. Royer. Ad-hoc on-demand distance vector rout-
ing. In Mobile Computing Systems and Applications, 1999. Proceedings.
WMCSA ’99. Second IEEE Workshop on, pages 90–100, Feb 1999.
[33] Brad Karp and H. T. Kung. Gpsr: Greedy perimeter stateless routing for
wireless networks. In Proceedings of the 6th Annual International Confer-
ence on Mobile Computing and Networking, MobiCom ’00, pages 243–254,
New York, NY, USA, 2000. ACM.

[34] Christian Lochert, Martin Mauve, Holger Fussler, and Hannes Hartenstein.
Geographic routing in city scenarios. SIGMOBILE Mob. Comput. Com-
mun. Rev., 9(1):69–72, January 2005.

30
[35] Danlei Yu and Young-Bae Ko. Ffrdv: Fastest-ferry routing in dtn-enabled
vehicular ad hoc networks. In 2009 11th International Conference on Ad-
vanced Communication Technology, volume 02, pages 1410–1414, Feb 2009.
[36] W. Peng, B. Zhao, W. Yu, C. Wu, and X. Yan. Ferry route design with
delay bounds in delay-tolerant networks. In 2010 10th IEEE International
Conference on Computer and Information Technology, pages 281–288, June
2010.
[37] Shuhui Yang Jie Wu and Fei Dai. Logarithmic store-carry-forward routing
in mobile ad hoc networks. pages 735–748, 2007.
[38] W. Zhao, M. Ammar, and E. Zegura. A message ferrying approach for data
delivery in sparse mobile ad hoc networks. In Proceedings of the 5th ACM
International Symposium on Mobile Ad Hoc Networking and Computing,
MobiHoc ’04, pages 187–198, New York, NY, USA, 2004. ACM.
[39] T. Wang and C. P. Low. Dynamic message ferry route (dmfr) for partitioned
manets. In 2010 International Conference on Communications and Mobile
Computing, volume 3, pages 447–451, April 2010.
[40] Chu-Fu Wang. A virtual multiple message ferry backbone routing scheme
for mobile ad-hoc networks. Ad Hoc Networks, 10(7):1399 – 1418, 2012.
[41] S. Ali, J. Qadir, and A. Baig. Routing protocols in delay tolerant networks
- a survey. In 2010 6th International Conference on Emerging Technologies
(ICET), pages 70–75, Oct 2010.
[42] Amin Vahdat and David Becker. Epidemic routing for partially-connected
ad hoc networks. Technical report, 2000.
[43] P. Costa, C. Mascolo, M. Musolesi, and G. P. Picco. Socially-aware routing
for publish-subscribe in delay-tolerant mobile ad hoc networks. IEEE J.Sel.
A. Commun., 26(5):748–760, June 2008.
[44] E. M. Daly and M. Haahr. Social network analysis for information flow in
disconnected delay-tolerant manets. IEEE Transactions on Mobile Com-
puting, 8(5):606–621, May 2009.
[45] J. Wang and W. Yan. Rbm: A role based mobility model for vanet. In 2009
WRI International Conference on Communications and Mobile Computing,
volume 2, pages 437–443, Jan 2009.
[46] M. Kadadha, H. Otrok, H. Barada, M. Al-Qutayri, and Y. Al-Hammadi.
A street-centric qos-olsr protocol for urban vehicular ad hoc networks. In
2017 13th International Wireless Communications and Mobile Computing
Conference (IWCMC), pages 1477–1482, June 2017.
[47] Q. Ding, B. Sun, and X. Zhang. A traffic-light-aware routing protocol
based on street connectivity for urban vehicular ad hoc networks. IEEE
Communications Letters, 20(8):1635–1638, Aug 2016.

31
[48] M. Boban, T. T. V. Vinhoza, M. Ferreira, J. Barros, and O. K. Tonguz.
Impact of vehicles as obstacles in vehicular ad hoc networks. IEEE Journal
on Selected Areas in Communications, 29(1):15–28, January 2011.

[49] Jizhao Liu and Quan Wang. Position prediction based frequency control of
beacons in vehicular ad hoc networks. International Journal of Distributed
Sensor Networks, 11(8):631415, 2015.

[50] Khaleel Husain and Azlan Awang. A Receiver-Based Forwarding Scheme

to Minimize Multipath Formation in VANET, pages 15–26. Springer Sin-
gapore, Singapore, 2017.
[51] Won-Il Lee, Jae-Young Pyun, Yang Sun Lee, and Sang-Woong Lee. Relative
velocity based vehicle-to-vehicle routing protocol over ad-hoc networks. Int.
J. Ad Hoc Ubiquitous Comput., 12(1):14–22, January 2013.

[52] Google Maps (2017). Salt lake bidhannagar area, kolkata, india.

32