Multi-Agent System for Secured and Reliable Routing in VANET

International Journal of Research and Scientific Innovation (IJRSI) | Volume IV, Issue VIS, June 2017 | ISSN 2321–2705
www.rsisinternational.org Page 85
Multi-Agent System for Secured and Reliable
Routing in VANET
Anil D. Devangavi1
and Dr. Rajendra Gupta2
1
Basaveshwar Engineering College, Bagalkot, Karnataka, India
2
AISECT University, Madhya Pradesh, India
Abstract:- In VANET, the emphasis is given on the exchange of
traffic information and road conditions between the vehicles and
thereby preventing the accidents. Distinctive characteristics of
VANET like restricted topology, unpredictable mobility, vehicle
density, varying channel capacity, etc. make VANET
environment exciting for developing efficient routing protocols.
Owing to the dynamic topology in VANET, the routes are
unstable and unreliable for exchange of information among the
vehicles. To enrich the performance and throughput of the
VANETs, the links between nodes must be reliable and stable. In
order to tackle the relaiblity and stability of information
communication this work proposes ‘Multi-agent system for
Secured and Reliable Routing (MSRR) in VANET. The
performance of the proposed scheme is tested in terms of packet
delivery ratio, route reliability, route discovery time and delay.
Keywords- VANET, Multi-agents, Trust value, Reliability weight,
Route Discovery Time and Delay.
I. INTRODUCTION
ehicular Ad hoc Network (VANET) is a class of Mobile
Ad Hoc Network (MANET) and one of the core
technologies in Intelligent Transportation Systems (ITS)[25].
VANET‟s have an edge over MANETs in having sufficient
computational and power resources equipped with each
vehicle. This eliminates energy related computations in
VANETs.
VANET comprises of a group of moving vehicles and some
ﬁxed infrastructure. The primary architectures for vehicular
communication in VANETs are Vehicle-to-Vehicle (V2V),
Vehicle-to-Infrastructure (V2I) and Hybrid Architecture [16].
Recently, Vehicle-to-Passenger communication (V2P) is
being discussed, which can be possible by fourth architecture.
V2V communication is primary constituent of ITS. It enables
the vehicle to communicate with other vehicles. V2I is the
communication between vehicles and roadside infrastructure,
intended primarily to enable a wide range of safety, mobility
and environmental beneﬁts. Hybrid architecture integrates
both V2V and V2I communication architectures. Vehicle-to-
Passenger communication allows straight, immediate, and
adjustable communication between moving vehicles and
roadside passengers.
The routing protocols of VANET are broadly classified as:
topology based routing protocols, position based routing
protocols, cluster based routing protocols, broadcast routing
protocols and geo cast routing protocols and [6][34]. The
topology based routing protocols makes use of existing links
information in the network to forward the packets. Position
based routing protocols use geographic positioning
information in order to select the next forwarding. In cluster
based routing protocols, the entire network is divided into
clusters. Thus an interconnected set of clusters are identified.
Broadcast routing protocols are used when message needs to
be sent far distant vehicle by using multi-hop communication.
Geo cast routing protocols are a variety of location based
multicast routing and used in VANET to deliver the packet to
all nodes within a stated geographical area.
Distinctive characteristics of VANET are restricted topology,
unpredictable mobility and vehicle density, varying channel
capacity etc. [20]. Because of the dynamic topology, routes
are frequently broken and are unreliable. Thus in VANETs,
the task of developing reliable routing protocols is very
challenging.
1.1 Related Work
A novel infrastructure based connectivity aware routing
protocol iCAR-II which enables multi-hop vehicular
applications is presented in [1]. iCAR-II consists of
algorithms that are run by vehicles to calculate local networks
connectivity. The real time information is used to
continuously update location servers. A global network
topology is thus generated. Authors in [2] have presented a
token based trust computation technique, which relies on
network connectivity duration of vehicles in VANET. The
presented technique is based on travelling time association
among vehicles or network connectivity of vehicle at the time
of driving on the road.
Authors in [3] have reviewed the most recent Traffic Aware
Routing (TAR) protocols while stressing on traffic and
network conditions awareness issues. Also authors have
explored the strengths and weaknesses of TAR protocols with
respect to measurement of routing metrics, forwarding
mechanisms and recovery techniques.
A Reliable Routing Protocol (R2P) for VANETs divides the
network into overlapping zones, which is given in [8].
Amongst the nodes in each zone, one node is designated as a
Master Node (MN). Master node maintains up-to-date routing
boards for inter/intra-zone communication. R2P employs a
special route discovery mechanism to discover available
V

routes to the destination and then elects the most reliable
route. Improved distance-based VANET routing protocol in
urban traffic environments is presented in [9]. The distance
based protocol is applied to multi-hop broadcast scheme for
reliable packet dissemination, which is based on stable routing
decisions schemes.
A Multi level-scenario-oriented Greedy Opportunity Routing
protocol (M-GOR) is depicted in [10]. A calculation method
for the connectivity probability and a Greedy Opportunity
Forwarding (GOF) algorithm to respond to the impacts of the
multilevel structure is presented.
An extension of reliable Ad-hoc On-demand Distance Vector
routing protocol (AODV), AODV-L is given in [13]. The
authors have used vehicles movement information to predict
the stability period of a route. Moving vehicles transmission
range is computed based on state of the channel. A position-
based routing approach for VANET is presented in [14].
Authors have considered the obstacles (building, tree, etc.)
blocking radio transmissions and voids, present in a city
environment.
The use of modified AOMDV protocol is given in [15].
AOMDV and AODV are compared in terms of different
traffic pattern of UDP and CBR. In the multipath routing
protocol an alternate path from source to destination is
selected when ongoing path is failed.
An intelligent multihop routing protocol that interacts with the
environment and learn the best transmission parameters is
presented in [17]. The protocol use the parameters like data
transmission rate, vehicle movement, and route length.
Authors in [19] present Long Lifetime Any paths (LLA)
routing scheme that has utilized the metric of link cost based
on the introduced link stability index. The authors focus on
anypath routing to improve the reliability of multihop
VANET communications. Authors in [20] present an
improved DSDV routing protocol MA-DSDV based on multi-
agent system approach. In this approach, each agent
periodically broadcast its routing table with a sequence
number to keep track of the recently broadcasted information.
Multi agent based system for congestion control is illustrated
in [21]. In the presented scheme, the agents in the nodes listen
to the status of neighbouring node information. Based on
neighbour node information, congestion free path is
computed. A new routing protocol Multimetric Map aware
routing protocol (MMMR) for VANETs that uses several
metrics like distance, direction of vehicles, number of vehicles
and bandwidth to calculate the multi metric value for the next
forwarding nodes is given in [22]. The node with better value
is decided as the next node.
Authors in [24] present a Secure and Intelligent Routing (SIR)
protocol to find quick and secured path. The scheme
calculates trusted vehicles at every junction, which acquires
the information of the neighbouring vehicles and compute the
secured path. Authors in [26] describe the usage of extended
evolving graph theory to develop a link reliability model
based on the speed and movements of the vehicles on a
highway. Velocity of the vehicles is used to calculate link
reliability. Reliable routes are computed pre-emptively.
The authors in [27] present an event triggered multipath
routing in WSNs by employing a set of static and mobile
agents. In this scheme, the node generating the event knows
about location information of destination node and computes
arbitrary midpoint between itself and the destination node. A
new vehicular reliability protocol AODV-R to calculate
reliable routes in VANETs is presented in [28]. The link
reliability is defined as the probability that a communication
link between two vehicles will not break for a specified time
period and is accurately calculated using the information of
vehicles along the road.
A class of routing protocols called Road-Based using
Vehicular Traffic information routing (RBVT), a reactive
protocol RBVT-R and a proactive protocol, RBVT-P are
presented in [33]. The protocols create paths consisting of
road intersections having connectivity of high probability. The
authors in [35] described the usage of vehicles movement
information to predict the chance of a link-breakage event.
Vehicles are grouped based on their velocity. A scheme that
secures geographic position-based routing is presented in [36].
The scheme safeguards the position-based routing services
likerelay communication, accurate location service and etc.
thereby increasing the robustness of the network.
Some of the drawbacks of existing routing techniques ACO-
based MCQ aware (S-AMCQ) [4], Stable Routing Protocol
(SRP) [7], Organized Topology Based Routing (OTBR) [11],
Greedy Bundle Release Scheme-Bulk Bundle Release
(GBRS-BBR) [30] and Vehicle Heading Based Routing
(VHBR) [38] in addition to the ones mentioned in related
work are as follows: (i) Lack of intelligence in path
discovery and path construction, (ii) lesser flexibility in relay
node selection mechanisms, (iii) lack of robust mechanism for
critical information transmission and (iv) reliability of
constructed routes is less and etc.
In VANET, vehicles exchange the traffic and other warning
information with each other during travelling. In these
networks, high mobility of the nodes is the major concern, due
to which routes are frequently broken. Hence the dynamics of
vehicle topology makes the routes unstable and unreliable for
exchange of information among the vehicles. To tackle these
issues, we have proposed Multi-agent System for Secured and
Reliable routing (MSRR) in VANET. The proposed work
facilitates computation of reliable and secured that enhances
the performance of the proposed system.
1.2 Methodology
The proposed MSRR protocol calculates secured and reliable
path between the source and destination using agent
technology. For faciliatating the reliable and stable
communication, the proposed protocol uses the agent

technology. For compuatation of the route the proposed
protocol uses two different agencies viz Vehicle Agency (VA)
at each vehicle and Road Side Unit Agency (RSUA).
The proposed scheme works as follows: (1) Static agent in the
source vehicle calculates ‘n’ favourable vehicle nodes
(vehicle nodes in the direction of destination based on
neighbourhood information). (2) Static agent in the source
vehicle node triggers a mobile agent and creates ‘n’ clones of
it. (3) Each clone of mobile agent traverse through one
favourable node ( one to one mapping of „n’ clones to „n’
favourable nodes) towards destination node calculating the
next intermediate nodes based on neighbourhood information.
(4) Whenever any two clones meet at any intermediate vehicle
node, a static agent is triggered at that intermediate node to
evaluate the best path amongst the paths traversed by the two
clones. (5) Clone with the best path is allowed to traverse
further and the other clone is terminated. (6) This process
continues until best path to the destination is computed. (7)
Information is encrypted and transmitted along the path. (8)
During the transit of the information, if any of the next
intermediate node is malicious (has moved out of the network
and etc.), then from the current vehicle node another
intermediate vehicle node is computed and a patch up path is
calculated.
Some of the contributions of the proposed routing technique
are as follows: (1) Usage of static node for the computation of
favourable nodes. (2) Usage of mobile agent clones for
traversing through the favourable nodes computed. (3)
Defining and usage of trust value for the selection of
intermediate nodes. (4) Defining and usage of reliability
weight factor to calculate the best reliable path. (5) The
performance of the proposed scheme is tested in terms of
performance parameters such as packet delivery ratio, route
reliability, route discovery time and delay.
The proposed work is compared with AODV-L [13]. The
advantages of proposed MSRR over AODV-L are as follows:
(1) Use of multiple agents for finding the path from source to
destination node. (2) Reliable delivery of the critical
information. (3) As it uses the trust value to select every relay
node, the selected path is more secured. (4) Path construction
mechanism is incremental and has progressive accuracy and
(5) best path is computed based on reliability weight factor.
II. STRUCTURING THE PROPOSED SYSTEM
This section presents the network environment, mathematical
models, proposed agency, scheme and algorithm.
2.1 Network Environment
Figure 1 presents the network environment in which we can
notice that number of vechiles move with different speed in
different lanes. Some of the assumptions in this work are:
vehicles travel in an urban road scenario, each vehicle is fitted
with Global Positioning System (GPS), sensors and on-board
communication devices for communication. The information
regarding location and direction of the vehicle is provided by
GPS. Each vehicle is fitted with a digital map, which provides
information regarding road connectivity. It is also assumed
that on-board communication devices are inbuilt with an agent
platform and the platform supports our proposed agency [20]
[21] [31]. Agency involves static agents, mobile agents and a
knowledge base. Agents are guarded from hosts on which
they execute and vice versa. The platform is completely
secured.
Figure 1. VANET Environment-Urban Scenario
Each vehicle communicates with other vehicle within its
communication range (VRan) via V-to-V communication. The
vehicles that lies within the transmission range of a particular
vehicle are the neighbouring vehicles of that vehicle. Each
vehicle status is being updated by itself and neighbouring
vehicles in a local database. The information is related to
vehicle ID, position of the vehicle in terms of latitude and
longitude, current speed of the vehicle, direction, trust value
and etc. Every vehicle is connected to at least one road side
unit (fixed infrastructure). Each vehicle communicates with
road side units via V-to-R communication.
Terminology
 Neighbour vehicles/nodes: Vehicle nodes which lie in the
transmission range (VRan) of vehicle „V‟.
 Favourable vehicle node (VFN): It is the vehicle node with
trust value C or P (if there are no vehicles with C value),
same speed and are in the direction of the destination
vehicle node. For any vehicle there may be several
favourable vehicle nodes towards destination vehicle.
Vehicles with trust value N are ignored.
 Intermediate vehicle node (VIN): It is the vehicle node with
trust value as C or P (if there are no vehicles with C
value), same speed and direction as the destination node. It
is the most favourable vehicle node from current vehicle
node. Vehicles with trust value N are ignored.

 Best Path / Reliable Path: Best path is the path with higher
reliability weight factor.
2.2 Mathematical models
In this subsection, mathematical models for reliability weight
and trust value are derived.
2.2.1 Mathematical Model for reliability weight
Reliability weight of any path between two vehicle nodes is
computed as follows. Let
MAV: Average mobility/speed of the vehicles on the path and
is measured in Kmts./hr.
NIP: No. of Intermediate nodes (VIN) in the path.
NPN: No. of intermediate nodes in the path with partial trust
value „P‟.
D: Distance between the two end vehicle nodes of the path
and is measured in mts.
For any 2 nodes (X1, Y1) and (X2, Y2), D=
NT: Numerical representation of trust value of the end vehicle
node.
Reliability weight of any path (WRT) is directly proportional to
the number of intermediate nodes NIP and numerical
representation of the trust value NT and is indirectly
proportional to average mobility of the vehicles in the path
MAV, number of intermediate vehicles with P as trust value
NPN and the distance between the source vehicle and
destination vehicle D (Eq. 1).
WRT NIP
1/ MAV
1/ NPN
NT
1 / D ………………………… (1)
i.e. WRT NIP * NT / MAV * NUN * D.…………………………(2)
From the Eq. 1 and Eq. 2 reliability weight of any path from
the source to the destination with n nodes in the path is
computed as:
…..…(3)
2.2.2 Mathematical Modelling of Numerical
representation of trust value:
Every vehicle is connected to atleast one RSU at any given
time. RSU allocates trust value to each vehicle. Trust values
are of three types: Completely trusted (C), Partially trusted
(P) and Not trusted (N). Trust value is subjective and time
dependent. Initially every vehicle is assigned with trust value
of „C’. Then afterwards the trust value is either retained as „C‟
or changed to „P‟ or „N‟ based on constraints like time aging
factor and positive or negative interactions.When the
intermediate vehicle not only transmits a packet to all its next
hops, but also forward devotedly (correct modification if
essential), then it is positive interaction. But in case of
negative interaction the intermediate vehicle does not forward
packet correctly by launching some attacks like black hole
attacks and etc.
 Time aging factor (TF) : It indicates that the trust fades
with time (lesser faith on vehicles with longer duration
on road) during the time period w.r.t. the current
time.
 Positive interactions (PI) : No. of Interactions of the
vehicle forwarding a packet to all its next hops devotedly
(correct modification if essential) during the time period
w.r.t. the current time.
 Negative interactions (NI) : No. of Interactions of the
vehicle that does not forward the data by launching some
attacks like black hole attacks and etc. during the time
period w.r.t. the current time [12][23].
Numerically the trust values are represented as: ,
, .
2.3 RSA algorithm
The proposed work uses RSA algorithm for secured data
transmission. The reason behind using RSA algorithm is that
(i) It is based on Public Key encryption in which the
information is encoded with someone's Public Key (everyone
knows Public Key). Nevertheless, only the intended person
can read it, by using their private key (which only they know
about). Endeavouring to utilize the Public Key to decode the
message would not work. (ii) The security of the RSA
calculation has so far been approved, since no known
endeavours to break it have yet been fruitful, generally
because of the trouble of factoring n = pq, where p and q are
huge prime numbers [18] [29] [42].
2.4 Proposed Agency
Various agencies [5] are employed to perform communication
among vehicles and road side units viz. Vehicle Agency, Road
Side Unit Agency.
2.4.1 Vehicle Agency
Vehicle agency resides in each vehicle. Vehicle agency works
on the principle of black board architecture for inter agent
communication and coordination. The components of vehicle
agency and their interaction is depicted in Figure 2. Vehicle
agency consists of a set of agents viz. Vehicle Information
Manager Agent (VIMA), Destination Path Finder Agent
(DPFA) and Path Evaluator Agent (PEA). The vehicle agency
also consists of a Vehicle KnowledgeBase (VKB).
Vehicle Information Manager Agent (VIMA): Vehicle
Information Manager Agent (VIMA) is a static agent in each
vehicle. VIMA monitors the information like transmission

range and time stamp. It gets information like Vehicle/Node
ID, location in terms of latitude and longitude, speed of the
vehicle, direction of the vehicle and trust value [2] [12] [23]
from the neighbour nodes. It also shares similar information
of the parent node with the neighbour nodes. It triggers a
mobile agent, Destination Path Finder Agent (DPFA) to find
the path to the destination vehicle and a static agent, Path
Evaluator Agent (PEA) to evaluate the best path whenever
required. VIMA of the source vehicle node computes a set of
favourable vehicle nodes.VIMA of intermediate vehicles
computes next neighbour/intermediate node depending upon
the neighbour node parameters.
Figure 2. Node/Vehicle Agency
Vehicle Knowledge Base (VKB): It works on the principle of
black board architecture. The knowledge base is updated by
the agents. It forms the base for inter agent communication
and coordination. It maintains transmission range, and time
stamp information of parent node and information like Node
ID, location in terms of latitude and longitude, speed of the
vehicle, trust value and direction of the vehicle of the parent
and neighbour vehicles/nodes.
Destination Path Finder Agent (DPFA): DPFA is a mobile
agent and is triggered by VIMA whenever a new path is to be
computed to destination vehicle. Multiple clones of DPFA are
created. The number of clones is equal to the number of
favourable nodes of the source node. Each clone migrates
from source vehicle to one favourable node. DPFA clones
traverse the network in multiple directions towards
destination. Multiple paths are thus explored. While
migrating DPFA collects the path information like distance
between the vehicles, no. of vehicles and etc. Finally upon
reaching the destination, the path is updated with the parent
vehicle node.
Path Evaluator Agent (PEA): Whenever more than one clone
meets at any intermediate vehicle node, VIMA of that
intermediate vehicle node triggers Path Evaluator Agent
(PEA).This is a static agent residing in each vehicle and is
responsible for finding the best path explored by the different
clones. It collects the path information like distance between
the vehicles, no. of intermediate vehicles, average mobility,
number of nodes with partial trust values, number of
untrusted nodes from each clone. Using this information it
computes the reliability weight of the paths explored by the
different clones. The path with better reliability weight is
selected as the best path.The same information is
communicated to the VIMA of the event node and parent
node.
2.4.2 Roadside Unit Agency: Roadside Unit Agency
resides in each Roadside Unit. It comprises of
Roadside Unit Manager Agent (RUMA), Trust Value
Allocation Agent (TVAA) and Roadside Unit
Knowledge Base (RUKB). The interaction of various
components of Roadside Unit Agency is as shown in
the Figure 3.
Figure 3. Road side unit Agency
Roadside Unit Manager Agent (RUMA): This is a static agent
and resides in the Roadside Unit. Each vehicle is connected to
atleast one RSU. It collects information like Node ID, its
direction, speed and other information from all the vehicles
directly connected to the Roadside Unit. It shares the same
information with the neighbour Roadside Units.
Roadside Unit Knowledge Base (RUKB): It works on the
principle of black board architecture. The knowledge base is
updated by the agents. It forms the base for inter agent
communication and coordination. It maintains the information
of connected vehicles such as Node ID, location in terms of
latitude and longitude, speed of the vehicle, trust value,
direction. This information can be shared with other RSU. It
maintains the trust parameters of the vehicles like time aging,
positive interactions and negative interactions.
Trust Value Allocation Agent (TVAA): This is a static agent
residing in the Roadside Unit. It is responsible for allocating
trust value to each connected vehicle. Trust allocation scheme
works as follows. All vehicles are connected to at least one
RSU and all vehicles involve in the forwarding of the data.
Three types of trust value are allocated: C: Completely
trusted, P: Partially trusted and N: Not trustworthy [2].

TVAA collects trust parameters like time aging factor,
positive and negative interactions of each vehicle connected
to the parent RSU [12][23]. Trust value is subjective and time
dependent. Initially each vehicle will be allocated with trust
value ‘C’. If either of the parameters viz. time aging factor
and positive interactions is false then the vehicle is assigned
the trust value of ‘P’ or ‘N’ , if both are false. Periodically the
trust value of each vehicle is updated by TVAA based on the
trust parameters i.e. either continued as
‘C’ or Updated to‘P’ or ‘N’.
2.5 Example execution of the algorithm to evaluate the
best path whenever more than one clone meet at any
intermediate node:
Figure 4 illustrates the scenario where three favourable nodes
(VFN) for source vehicle Vs are computed. Three clones C1,
C2 and C3 are triggered and they start exploring through the
three computed favourable nodes. Every intermediate vehicle
is selected based on neighbourhood parameters and trust
value. Two clones C1 and C2 meet at the same intermediate
vehicle node VIN. Path 1 and 2 are the two paths traversed by
the two clones till VIN.
Figure 4. Illustration of two clones meeting at the same vehicle node VIN
Static agent VIMA of VIN triggers path evaluation agent
(PEA). PEA computes the reliability weight of both paths 1
and 2 traversed by the two clones. Path 1 has the higher
reliability weight and thus it is the best path of the two. Clone
1 is allowed to continue and clone 2 is terminated and the path
traversed by it is discarded as shown in the Figure 5.
Figure 5. Illustration of retaining the clone traversing the best path
Figure 6 depicts the situation where clone 1 and clone 3 are
traversing towards the destination vehicle.
Figure 6. Illustration of retained clone 1 and clone3 moving towards the
destination
Figure 7 illustrates two clones C1 and C3 meeting at the
destination vehicle. Thus there are two paths from source
vehicle Vs to destination vehicle VD.VIMA of the destination
triggers PEA agent to evaluate the best path.
Figure7. Illustration of clone 1 and clone 3 meeting at the destination vehicle
Figure 8 illustrates PEA computing the reliability weight of
both the paths 1 and 3. Path 1 has better reliability weight
factor and is the best path. Path 3 is discarded.
Figure 8. Illustration of elimination of path with lower reliability weight
factor
Path 1 is the best path. Information is encrypted by using RSA
algorithm and communicated as shown in the Figure 9. As
each intermediate vehicle on the path is calculated based on

neighbourhood parameters and trust value, the path is secured.
Also as the path is computed based on the reliability weight
factor, it is reliable.
Figure 9. Reliable path from source vehicle to destination vehicle
Example scenario to compute patch up path in case of
malicious intermediate node
During the transit of the information the intermediate vehicle
node next to current vehicle node VIN is malicious as shown
in the figure 10. A patch up path is required.
Figure 10. Illustration of one of the malicious intermediate vehicle node
From the current intermediate node VIN another intermediate
vehicle node is calculated based on neighbour node
information and a patch up path is set to destination as
depicted in the Figure 11.
Figure 11. Illustration of computation of patch up path
2.6 Proposed Scheme
This subsection presents the proposed scheme. The steps
involved in the proposed scheme are as follows:
(1) Given a source node and destination node.
(2) VIMA of the source node computes‘n’ favourable
nodes (VFN). Favourable nodes are the ones with trust
value‘C’ or ‘P’, same speed and are in the direction
of the destination node. VIMA triggers Destination
Path Finder Agent (DPFA) and creates ‘n’ clones of
DPFA.
(3) Each clone migrates to one favourable node (‘n’
clones and ‘n’ favourable nodes with one to one
mapping).
(4) VIMA of the favourable nodes (VFN) calculate the
next intermediate vehicle node (VIN) amongst its
neighbours by considering neighbour node
information. The node with trust value as ‘C’ or ‘P’
(if there are no vehicles with ‘C’ value), same speed
and in the direction of the destination vehicle node is
selected as the next intermediate vehicle node. Node
with trust value ‘ N’ is not considered.
(5) Further next intermediate nodes (VIN) for the nodes
determined in step 5 are calculated until the
destination node is reached.‘n’ different multiple
paths towards destination are thus initiated and
explored.
(6) Whenever any two clones meet at any intermediate
vehicle node (VIN) in the process of moving towards
the destination, VIMA of that vehicle node triggers
PEA in that node. PEA calculates reliability weight
factor [Eq. 3] of the two alternative paths explored
by the two clones till this node [27]. Path with higher
reliability factor is considered as the best path and
the clone along the best path is allowed to survive
and continue towards the destination. The other path
is discarded and the clone along that path is
terminated by VIMA.
(7) Finally only one clone with the best path survives
with the destination node. This clone updates the
computed path with the VIMA of source node.
(8) Information is encrypted using RSA algorithm[18]
[29] [42].
(9) Encrypted Information is communicated along the
best path.
(10) During the transit of the information if any of the
next intermediate node (VIN) is malicious (has moved
out of the network, and etc.) then from the current
vehicle node another intermediate vehicle node is
computed and a patch up path is calculated.
2.7 Algorithms
This subsection presents the algorithms for the proposed
scheme.

2.7.1 Algorithm for compting the path between the source
vehicle and destination vehicle is as below:
Input: A set of vehicles V= {V1…..Vn}.Vs and Vd belongs
to V.
A set of RSUs R= {R1………Rn}.
Each vehicle is connected to atleast one RSU and has
vehicle agency, RSU has RSU agency.
Output: Reliable and secured Path between the source
vehicle and destination vehicle.
Algorithm 1:To compute the reliable path between the
source vehicle and destination vehicle
Begin
1. VIMA of Vs calculates „n‟ VFN s for Vs;
2. VIMA of Vs triggers mobile agent DPFA and create
„n‟ clones of it;
3. Each clone traverses through one VFN (one to one
mapping);
4. VIMA of each VFN calculates the next VIN for the
clone of that particular VFN;
5. Each clone of VFN traverses to its next VIN;
Do
6. VIMA calculates next VIN for the clone;
7. Each clone traverse to the respective next VIN;
8. If any two clones meet at the same VIN,
9. VIMA of that VIN triggers PEA to evaluate the best
path amongst the paths traversed by the two clones;
10. PEA calculates the best path.
11. Retain clone traversing the best path, terminate the
other clone;
Until next VIN is not VD;
12. Best path is communicated to VIMA of Vs.
13. Information is encrypted using RSA algorithm and
communicated. Go to End;
14. If during the transit of information any of the next
VIN is malicious (has moved out of the network, and
etc.) then from the current vehicle node another
intermediate vehicle node is computed and a patch up
path is calculated;
End
2.7.2 Algorithm 2: To assign trust value
Algorithm 2:To assign trust value:
Begin
1. For every vehicle Vi connected to RSU;
If
2. TF = New and PI where k is a constant and
time interval ;
3. Trust value= „C‟;
Else if TF= Old and PI
or TF= New and PI
4. Trust value = „P‟;
Else Trust value= „N‟;
End
III. SIMULATION MODEL
In this section simulation models, simulation procedure and
performance parameters are discussed. We have simulated
proposed model by considering Bangalore city map [37] as
shown in Figure 12. Only dense traffic roads and urban road
scenario are considered for simulation. The simulation is
carried out using NS-2.34 [32] to test the performance and
effectiveness of approach.
Figure 12. Bangalore City Map [37]
The proposed scheme has been simulated in various VANET
scenarios. Simulation environment for the proposed work
consist of following models: (1) Network model, (2) Traffic
model, (3) Mobility model and (4) Channel model. These
models are given below.
Network model: We consider V number of vehicles moving in
a ﬁxed region of length X Km. and breadth Y Km. Vehicles
are equipped with number of sensors i.e., S1,...Sn. We assume
vehicles to move in a freeway type road, with L lanes.
Communication coverage area for each vehicle is considered
as VRan meters.
Traffic model: Constant bit rate model is used to transmit
certain number of ﬁxed size packets, P pkts. Coverage area
around each vehicle has a bandwidth, BW, shared among its

neighbours. Arrival rate of information (critical/noncritical)
into the vehicle follows Poisson distribution with mean λ.
Poisson process is used as it is an efficient method for arrival
process of events to a queuing system.
Mobility model: At the beginning of a simulation, vehicles are
randomly put on the roads. They then move continuously
according to history-based speeds. Safety distance of R meters
is maintained from preceding vehicle for a certain tolerance
time and then change lane if possible. It is assumed that there
is a free ﬂow movement of vehicles and the situation of
congestions are ignored. It is also assumed that all vehicles are
furnished with a communication device and knows start
position, start time of vehicle, route that it selects and speed at
which it travels.Manhattan mobility model is used for
Bangalore city scenario[41]. This mobility model is mainly
proposed to simulate the movement in urban area, where the
streets are in an organized manner.
Channel model: Communication medium access protocol
considered for simulation is Enhanced Distributed Channel
Access (EDCA) based Distributed Coordination Function
(DCF) of IEEE 802.11 which is responsible for medium
access based on CSMA with Collision Avoidance
(CSMA/CA) [39] [40].
3.1 Simulation procedure
Simulation procedure for proposed agent model is as follows:
Begin
1. Topology generation: Generate VANET in given road
length by placing vehicles uniformly. Each vehicle
maintains a data structure to store information as speciﬁed
by scheme.
2. Apply mobility to nodes (vehicles).
3. Generate proposed agency (agents are implemented as
objects).
4. Compute the performance of system.
End
3.2 Simulation Inputs
The simulation input parameters are as below:
Table 1. Simulation input parameters
Simulation parameters Values
Network simulator ns-2.34
Simulation time Simulation time: 600 seconds
Simulation area 5000m X 5000m
Number of vehicles 50/100/150
Transmission range 150m/300m
Speed
Minimum: 20 Km/hr., Maximum: 40 and 60
Km/hr.
Data type Constant Bit Rate
MAC protocol IEEE 802.11e EDCA based DCF
Safety distance between
vehicles
4 mts
Available bandwidth 5000 Mbps
Road type Free way
3.3 Performance Parameters
1) % of packets delivered: Packet delivery ratio is the ratio of
number of packets received at the destination node to the
number of packets sent from the source node. The
performance is better when packet delivery ratio is high.
2) Route reliability:The link reliability is defined as the
stable duration of the communication link between two
vehicles. Broken link is the link which disconnects while
communication. If route has less disconnected links then
the route is more stable. Otherwise, due to high
disconnection of links there is more packet loss and more
exchange of control packets.
3) Route discovery time: Route discovery time is the time
difference between the VIMA of the source vehicle
triggering the Destination path finder agent (DPFA) and
DPFA updating the VKB of the source vehicle about the
reliable route to the destination.
4) Delay: It is the time taken to transmit the data from the
source vehicle to the destination vehicle.It is expressed in
terms of milliseconds. It represents end-to-end delay
which refers to the time taken for a packet to be
transmitted across a network from source to destination.
IV. SIMULATION RESULTS AND ANALYSIS
This section presents the results obtained during simulation.
We compare results of proposed work with an existing routing
protocol AODV-L. The below mentioned figures are
generated based on the simulation results.
% of Packet Delivery (Packet Delivery Ratio)
Figure 13 shows the Packet Delivery Ratio (PDR) evaluated
for AODV-L and MSRR protocols by increasing the number
of vehicles in the urban scenario. The PDR of MSRR is higher
than AODV-L routing protocol. The PDR of MSRR is better
when the vehicle density is more eventhough the average
mobility of vehicles is more. MSRR protocol computes the
reliable path having the highest trust value and higher
reliability weight. This path is more reliable that results in
better PDR.
Figure 13. % of Packet Delivery v/s No. of vehicles
Route reliability

Figure 14 shows the variation in route reliability for both
protocols AODV-L and MSRR. As the graph depicts the
expected disconnection degree of links in the path for MSRR
is fairly less than AODV-L which indicates more reliability of
routes for MSRR protocol. This is because in MSRR the
selected path will be of higher trust value and reliability
weight.
Figure14. Route reliability v/s No. of vehicles
Route Discovery time
Figure 15 shows variation of the route discovery time versus
the number of vehicles with varying mobility of the vehicles.
From the figure it is evident that there is considerable increase
in route discovery time as the node density and node mobility
is increased.Route discovery time taken by MSRR is lesser as
compared to AODV-L because it uses the intelligent software
agents for computation of the reliable and secured path.
Figure15. Route Discovery time v/s No. of vehicles
Delay
Figure 16(a) shows the delay versus mobility with varying
link failures. When the average speed of the vehicles increases
vehicles needs to update its information in parent node and
connected RSU, intern routes are computed using the fresh
information, which consumes more time.
The Figure 16(b) presents the delay versus the number of
vehicles with varying malicious nodes. When the number of
vehicles increases, computation of the reliable paths takes
more time. From the figure we notice that the delay
substantially increases with the increase in number of
malicious nodes. Delay introduced by MSRR as compared to
AODV-L is lower. This is because of the mobile agentswhich
in the process of generating reliable paths, migrate carrying
partially integrated results whichleads to progressive
accuracy.
Figure16(a). Delay v/s mobility
Figure16(b). Delay v/s No. of vehicles
V. CONCLUSION
The proposed scheme presented a multi-agent system based
secured and reliable routing in VANET by employing a set of
static and mobile agents. The source node activates the path
discovery mechanism by calculating the favourable nodes in
the direction of destination node dynamically.The proposed
scheme used clones of mobile agents for traversing the path
towards destination. Intermediate nodes on the path are
selected based on neighbourhood information and trust value.
Whenever two clones meet at any intermediate node best path
is calculated based on the reliability weight. Thus the
intermediate nodes on the discovered path are trustworthy and
the path is reliable. The process of computing the path is

incremental and leads to progressive accuracy. As compared
to AODV-L, the proposed MSRR performed better in terms of
packet delivery ratio, route reliability, route discovery time
and delay.
REFERENCES
[1]. Nizar Alsharif, Xuemin (Sherman) Shen, (2017) “iCAR-II:
Infrastructure-based Connectivity Aware Routing in Vehicular
Networks”, IEEE Transactions on Vehicular Technology, Vol. 66,
Issue 5, pp 4231 – 4244 .
[2]. Kapil Sharma, Brijesh Kumar Chaurasia , ShekharVerma and
Geetam Singh, (2016) “Token Based Trust Computation in
VANET ”, International Journal of Grid and Distributed
Computing Vol. 9, No. 5, pp 313-320.
[3]. Tasneem Darwish, Kamalrulnizam Abu Bakar, (2016) “Traffic
aware routing in vehicular ad hoc networks: characteristics and
challenges”, Telecommunication Systems, Vol. 61, Issue 3,
Springer, pp 489–513.
[4]. Mahmoud HashemEiza, Thomas Owens, and Qiang Ni, Senior
Member, (2016) “Secure and Robust Multi-Constrained QoS
aware Routing Algorithm for VANETs”, IEEE Transactions on
Dependable and Secure Computing , Vol. 13, Issue 1, pp 32-45.
[5]. Samira Harrabi ,Ben Jaafar ,KhaledGhedira ,(2016) “ A Novel
Clustering Algorithm Based on Agent Technology for VANET”,
Network Protocols and Algorithms, ISSN 1943-3581, Vol. 8, No.
2 (doi: 10.5296/npa.v8i2.8434)
[6]. Neha Goel, Isha Dhyani, Gaurav Sharma, “A Study of Position
Based VANET Routing Protocols”, In: Proceedings of IEEE
International Conference on Computing, Communication and
Automation, April 2016, India
(doi: 10.1109/CCAA.2016.7813803).
[7]. T. Sivakumar, R. Manoharan, (2016) “SRP: A Stable Routing
Protocol for VANETs”, International Journal of Applied
Engineering Research ISSN 0973-4562 Vol. 11, No. 5, pp 3499-
3504.
[8]. Ahmed I. Saleh a,,Samah A. Gamel a, Khaled M. Abo-Al-Ez b,
(2016) “ Reliable Routing Protocol for Vehicular Ad hoc
Networks”, Computers and Electrical Engineering, Elsevier Ltd.,
(https://doi.org/10.1016/j.compeleceng.2016.11.011).
[9]. Chang Sang-woo and Lee Sang-sun, (2016) “A Routing Protocol
for Urban Vehicular Multi-hop Data Delivery”, Chinese Journal of
Electronic Vol. 25, No.2, pp 348-356.
[10]. Lina Zhu, Changle Li, Bingbing Li, Xinbing Wang, Guoqiang
Mao, (2016) “Geographic Routing in Multilevel Scenarios of
Vehicular Ad Hoc Networks”, IEEE Transactions on vehicular
technology, Vol. 65, No. 9, pp 7740 – 7753.
[11]. JianShen , Chen Wang , Anxi Wang , Xingming Sun ,
SangmanMoh , Patrick C.K. Hung, (2016) “Organized topology
based routing protocol in incompletely predictable ad-hoc
networks” Computer Communication 99, Elsevier , Issue C, pp
107-118 (https:// doi.org/ 10.1016/ j.comcom. 2016.07.009).
[12]. Xuanxia Yao , Xinlei Zhang , HuanshengNing , Pengjian Li ,
(2016) “Using trust model to ensure reliable data acquisition in
VANETs”, Ad Hoc Networks, Elsevier, Vol. 55, pp 107–118.
[13]. He, Yang, WenjunXu, and Xuehong Lin , "A Stable Routing
Protocol for Highway mobility over Vehicular Ad-hoc
Networks." In: Proceedings of Vehicular Technology Conference
(VTC Spring), IEEE, Glassgow, UK, May 2015
(doi: 10.1109/VTCSpring.2015.7145647).
[14]. Souaad Boussoufa-Lahlaha, Fouzi Semchedinea, Louiza
Bouallouche-Medjkoune,(2015) “A position-based routing
protocol for vehicular ad hoc networks in a city environment”, In:
Proceedings of the International Conference on Advanced
Wireless, Information, and Communication Technologies,
Procedia Computer Science 73 ( 2015 ) 102 – 108 (doi: 10.1016
/.procs. 2015.12.054).
[15]. FarhanaAnjum, V.D.Bondre, Ausaf Umar Khan, (2015) “Design
of Single and Multipath Routing Protocol for Quality of Service
(QoS) in VANET” In: Proceedings of the IEEE India International
Conference on Communications and Signal Processing, April
2015 (doi: 10.1109/ ICCSP.2015.7322606).
[16]. Sergio M. Tornell, Carlos T. Calafate, Juan-Carlos Cano, and
Pietro Manzoni, (2015) “DTN Protocols for Vehicular Networks:
An Application Oriented Overview”, IEEE Communication
surveys & tutorials, Vol. 17, No. 2, pp 868 – 887.
[17]. Celimuge Wu, YushengJi, Fuqiang Liu , Satoshi Ohzahata ,
Toshihiko Kato, (2015) “Toward Practical and Intelligent Routing
in Vehicular Ad Hoc Networks”, IEEE Transactions on Vehicular
Technology, Vol. 64, No 12, pp 5503-5519.
[18]. Amrita Jain, VivekKapoor, (2015) “Secure Communication using
RSA Algorithm for Network Environment ” , International Journal
of Computer Applications (0975 – 8887) Vol. 118, No. 7, pp 6-9.
[19]. JacekRak, (2014) “LLA: A New Anypath Routing Scheme
Providing Long Path Lifetime in VANETs”, IEEE
Communications letters, Vol. 18, No. 2, pp 281-284.
[20]. Samira Harrabi ,WalidChainbi,KhaledGhedira, (2014)“ A Multi-
Agent Proactive Routing Protocol for Vehicular Ad-Hoc
Networks”, In: Proceedings of International Symposium on
Networks, Computers and Communications, Tunisia, July 2014
(doi: 10.1109 /SNCC. 2014. 6866523).
[21]. Ramesh B. Koti and Mahabaleshwar S. K., (2014) ”Multi Agent
Based Congestion Control in VANETs”, International Journal of
Future Computer and Communication, Vol. 3, No. 2, pp 102-104.
[22]. CarolinaTripp-Barba, Luis Urquiza-Aguiar,
MonicaAguilarIgartua, David Rebollo-Monedero, Luis J.dela Cruz
Llopis, Ahmad Mohamad Mezher and Jose Alfonso Aguilar-
Calder, (2014) “A Multimetric, Map-Aware Routing Protocol for
VANETs in Urban Areas ”, Sensors, Vol. 14, No. 2 , pp 2199-
2224.
[23]. Mayuri Pophali, Shraddha Mohod, T.S.Yengantiwar, (2014)
“Trust Based Opportunistic Routing Protocol for VANET
Communication”, International Journal of Engineering And
Computer Science,Vol. 3, Issue 8, pp 7408-7414.
[24]. Sourav Kumar Bhoi, Pabitra Mohan Khilar, (2014) “SIR: a secure
and intelligent routing protocol for vehicular ad hoc network”,
IET, Vol. 4, No. 3, pp 185-194.
[25]. Pavlos Sermpezis, Georgios Koltsidas, and Fotini-Niovi Pavlidou,
(2013) “Investigating a Junction-Based Multipath Source Routing
Algorithm for VANETs”, IEEE Communications letters, Vol. 17,
No. 3, pp 600-603.
[26]. Mahmoud HashemEiza and QiangNi , (2013) “An Evolving
Graph-Based Reliable Routing Scheme for VANETs”, IEEE
Transactions on Vehicular Technology, Vol. 62, No. 4, pp 1493-
1504.
[27]. Dr A.V. Sutagundar, Dr S SManvi, (2013) “Location aware event
driven multipath routing in Wireless Sensor Networks: Agent
based approach”, Egyptian Informatics Journal, Vol. 14, No. 1,
pp 55–65.
[28]. Mahmoud HashemEiza, Qiang Ni , Thomas Owens and Geyong
Min , (2013) “Investigation of routing reliability of vehicular ad
hoc networks” EURASIP Journal on Wireless Communications
and Networking (doi:10.1186/1687-1499-2013-179).
[29]. Nentawe Y. Goshwe,(2013) “Data Encryption and Decryption
Using RSA Algorithm in a Network Environment ” ,IJCSNS
International Journal of Computer Science and Network Security,
Vol.13, No.7, pp 9-13.
[30]. M. J. Khabbaz, W. F. Fawaz, and C. M. Assi,(2012) “Modeling
and delay analysis of intermittently connected roadside
communication networks,” IEEE Transactions on Vehicular
Technology, Vol. 61, No. 6, pp 2698–2706.
[31]. Oscar Urra, Sergio Ilarri, Thierry Delot and Eduardo Mena
“Mobile Agents in Vehicular Networks:Taking a First Ride”, In:
Proceedings of International Conference on Practical Applications
of Agents and Multiagent Systems , 2010 (doi: 10.1007/978-3-
642-12384-9).
[32]. Network simulator: ns-2, Available: http://www.isi.edu/nsnam/ns
[October 2010].
[33]. JosianeNzouonta, NeerajRajgure, Guiling Wang, (2009) “VANET
Routing on City Roads using Real-Time Vehicular Traffic

Information”, IEEE Transactions on Vehicular Technology,Vol.
58, No. 7, pp 3609 – 3626.
[34]. Fan Li and Yu Wang, University of North Carolina at
Charlotte,(2007) “Routing in Vehicular Ad Hoc Networks: A
Survey”, IEEE Vehicular Technology , Vol. 2, Issue 2, pp 12 – 22.
[35]. Tarik Taleb, Kazuo Hashimoto, (2007) “ A Stable Routing
Protocol to Support ITS Services in VANET Networks”, IEEE
Transactions on vehicular technology, Vol. 56, No. 6, 3337 –
3347.
[36]. Charles Harsch , Andreas Festag, PanosPapadimitratos, “Secure
Position-Based Routing for VANETs, In: Proceedings of IEEE
66th Vehicular Technology Conference, USA, October 2007
(doi: 10.1109/VETECF.2007.22).
[37]. Bangalore City Map, Available: www.mapofbangalore.com
[November 2007].
[38]. Tarik Taleb, Mitsuru Ochi, Abbas Jamalipour, Nei Kato, and
Yoshiaki Nemoto ,“An Efficient Vehicle-Heading Based Routing
Protocol for VANET Networks”, In:Proceedings of IEEE Wireless
Communications and Networking Conference, April 2006
(10.1109/ WCNC. 2006. 1696637).
[39]. W. Xiuchao and A. L. Ananda, “Link Characteristics Estimation
for IEEE 802.11 DCF based WLAN”, In: Proceedings of IEEE
International Conference on Local Computer Networks, Tampa,
USA, November 2004 (doi: 10.1109/LCN.2004.73).
[40]. S. Wiethlter and C. Hoene, “Design and Veriﬁcation of an IEEE
802.11e EDCF Simulation Model in NS-2.26”, Technical report,
Telecommunication Networks Group, Technische, Universitt
Berlin, November 2004.
[41]. F. Bai, N. Sadagopan, and A. Helmy, “Important: A Framework to
SystematicallyAnalyze The Impact of Mobility On Performance of
Routing Protocols for Ad hocNetworks”, In: Proceedings of 22th
IEEE Annual Joint Conference on Computer Communicationsand
Networking, April 2003 (doi: 10.1109/INFCOM.2003.1208920).
[42]. Rivest R, Shamir A, Aldeman L, “A method for obtaining Digital
Signatures and Public-key Cryptosystems”, Communications of
the ACM, 21(2): 120-126, February 1978 (doi:10.1145/
359340.359342).
AUTHORS
Anil D. Devangavi completed his B. E in
Computer Science and Engineering from
Karnatak University Dharwad, India and M. Tech
from Visvesvaraya Technological University
Belgaum, India. He is pursuing Ph. D in the area
of Vehicular Adhoc Networks (VANETs) from AISECT
University, Bhopal, India. Presently, he is working as
Associate Professor in Department of Information Science and
Engineering, Basaveshwar Engineering College, Bagalkot,
Karnataka, India. He has published 3 international journal
papers. His areas of interest are wireless network and
VANETs.
Dr. Rajendra Gupta is Assistant Professor in
AISECT University, Raisen. He is doctorate
in Computer Science and having 18 years of
working experiences in Government and Non-
Government Sectors. His teaching and
research areas belong to Networking,
Network Security, Data Mining, Statistical Analysis and
Computer Graphics and Multimedia. He has published 22
research papers in International and National Journals and
completed a UGC minor research project. He has designed
four SLM for MP Bhoj University, Bhopal on the topic of
„Computer Organisation and Architecture‟, „Data
Communication and Computer Networks‟, „Software
Engineering‟ and „Operating System‟. He is reviewer of two
International Journals and member of Academic Body.

Multi-Agent System for Secured and Reliable Routing in VANET

More Related Content

What's hot (19)

Similar to Multi-Agent System for Secured and Reliable Routing in VANET (20)

More from RSIS International (20)

Recently uploaded (20)

Multi-Agent System for Secured and Reliable Routing in VANET