IEEE NS2 PROJECT@ DREAMWEB TECHNO SOLUTION

IEEE INTERNET OF THINGS JOURNAL, VOL. 2, NO. 2, APRIL 2015 121
A Context-Aware Trust-Based Information
Dissemination Framework for Vehicular Networks
Karim Rostamzadeh, Student Member, IEEE, Hasen Nicanfar, Student Member, IEEE, Narjes Torabi,
Student Member, IEEE, Sathish Gopalakrishnan, Member, IEEE, and Victor C. M. Leung, Fellow, IEEE
Abstract—Reliable, secure, private, and fast communication in
vehicular networks is extremely challenging due to the highly
mobile nature of these networks. Contact time between vehicles
is very limited and topology is constantly changing. Trusted com-
munication in vehicular networks is of crucial importance because
without trust, all efforts for minimizing the delay or maximizing
the reliability could be voided. In this paper, we propose a trust-
based framework for a safe and reliable information dissemination
in vehicular networks. The proposed framework consists of two
modules such that the first one applies three security checks to
make sure the message is trusted. It assigns a trust value to each
road segment and one to each neighborhood, instead of each car.
Thus, it scales up easily and is completely distributed. Once a mes-
sage is evaluated and considered to be trustworthy, our method
then in the second module looks for a safe path through which the
message is forwarded. Our frameworks are application-centric; in
particular, it is capable of preserving traffic requirements speci-
fied by each application. Experimental results demonstrate that
this framework outperforms other well-known routing protocols
since it routes the messages via trusted vehicles.
Index Terms—Information dissemination, security, trust,
vehicular networks (VNets).
I. INTRODUCTION
V EHICULAR networks (VNets) are the basis for different
applications offered in intelligent transportation systems
(ITS). These envisioned applications cover a large spectrum
ranging from delay-sensitive safety-related applications such
as accident notifications, to delay-tolerant services such as file
sharing. No matter what the application is, routing or in general,
information dissemination in a vehicular network is challenging
due to the highly dynamic and mobile nature of this net-
works [1]. Different applications in VNets can be offered using
either broadcasting the information, which is the case mainly
in safety-related applications, or unicasting/multicasting, which
happens mostly in nonsafety applications.
Over the past few years, researchers have proposed several
broadcasting and unicasting routing protocols for VNets that
minimize the delivery delay or the communication interference
[2]–[4]. However, the missing key point in those methods is
the security analysis. A node is assumed to be malicious when
Manuscript received October 21, 2014; revised December 02, 2014; accepted
December 12, 2014. Date of publication January 07, 2015; date of current
version March 13, 2015. This work was supported by the Natural Sciences and
Engineering Research Council (NSERC) of Canada.
The authors are with the Department of Electrical and Computer Engineer-
ing, University of British Columbia, Vancouver BC V6T 1Z4, Canada (e-mail:
karimr@ece.ubc.ca; hasennic@ece.ubc.ca; ntorabi@ece.ubc.ca; sathish@ece.
ubc.ca; vleung@ece.ubc.ca).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/JIOT.2015.2388581
it manipulates or intentionally drops a message. Consider this
simple example that explains the significance of the problem:
suppose one vehicle has a message and it has two neighbors
from which it chooses to forward its message. One of the neigh-
bors is a trustworthy node that relays every message without
any problem; however, routing from that node increases the
delay. The other neighbor, on the other hand, is on a fast route
to the destination but it is malicious and drops the messages
half the time. Now, based on a normal routing protocol that
its only focus is on minimizing the delay, the malicious node
will be selected to route the message faster. However, there
is a chance of 50% that the message will be dropped at this
node and it never reaches the destination. So, ignoring mali-
cious nodes can void the whole algorithm efforts to minimize
the delay. This example clearly shows why there is a need for
a safe and reliable communication framework in VNets. Note
that our focus is on having a secure packet delivery in addition
to a reliable one. Packet reliability, a.k.a. packet delivery ratio,
could be damaged by environment and network characteristics
such as channel fading or collisions. Secure packet delivery, on
the other hand, can be hurt because of attacks and intentional
manipulations by malicious nodes.
Trust is a human and psychological factor that is histori-
cally very well known in the social interactions. Using artificial
intelligence (AI) and bringing the concept of the trust to the
information and communication technology (ICT) has been
proposed and studied during the last decade [5]. For instance,
trust of a relay node i can be defined as the number of packets
that i has relayed without manipulating them, out of the total
number of packets given to the i to be relayed. Then, the prob-
ability of i relaying a new packet is the trust value of i based on
this simple rule: A node acts the same way that it has done so
far. As our first contribution, we leverage this concept to cover
a range of application requirements such as security, privacy,
delay, reliability to name a few. There is a trust value calculated
for each of these parameters. Then, we develop a function of all
these trust values to calculate the total trust for each path, which
is our reference for choosing the best route.
Trust in a VNet faces two main challenges: 1) the speed
of vehicles which changes the topology constantly and min-
imizes the contact times between the nodes and 2) the lack
of a centralized third party to evaluate and maintain the trust
values. Majority of the trust models proposed for VNets are
entity-centric in which the focus is on verifying the vehicle cre-
dentials [6]. Once the source is authenticated, the message can
be trusted. There are a few data-centric approaches that focus
on the correctness of the received message, instead [7]. Both
2327-4662 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

122 IEEE INTERNET OF THINGS JOURNAL, VOL. 2, NO. 2, APRIL 2015
of these models suffer from scalability and the entity-centric
models have the assumption that there is always a third-party
certificate issuer in the vicinity that is often not valid in the
context of VNets.
We survey the state-of-the-art in trust models/systems and
routing protocols in VNets. Then, as our next contribution, we
propose a novel two-layer framework for application-oriented
context-aware trust-based communication (FACT) in VNets,
where nodes only use their most trusted neighbors to forward
the message otherwise they carry the message by themselves.
FACT consists of two modules: Admission and Dissemination.
The key distinction of the FACT lies in its space-centric nature.
It is a combination of entity-/data-centric methods in addition
to its focus on location. Once a message is received, FACT first
applies three safety checks in the admission module to make
sure the message: 1) originated from a trusted region and tra-
versed a trusted path; 2) was not under attack on its path; and
3) has a valid content. Then, FACT admits the message and
pushes it to the dissemination module to be forwarded through
a trusted path. Each vehicle has a trust table where each road
segment in the city has its own trust value and this value is
constantly updated by the vehicle based on its experience in
that segment. The intuition behind FACT is that some areas of
every city are known to be safer with better facilities. It is fair
to assume that vehicles in those areas are more trusted, as well.
Otherwise and when there are malicious vehicles in the area,
that neighborhood’s trust is reduced and it gets a bad reputa-
tion. We divide vehicular networks’ applications into three main
categories and then identify major requirements of each cate-
gory. FACT is a general admission/communication framework
that accommodates all requirements of each application while
making sure the connection is trusted. It supports both broad-
casting and unicasting/multicasting modes. After explaining the
details of FACT, we evaluate our framework via simulation
and show that FACT outperforms other routing protocols when
some areas of the city are not trusted.
FACT gives network designers a full package, which delivers
trustworthy messages through a safe path with high reliability
and in a short amount of time. It is flexible enough meaning
network admins can tune the parameters based on the network
condition. Designers can incorporate their desired schedul-
ing and routing schemes into FACT and still disseminate the
messages safely. FACT is a framework that supports different
applications with different requirements.
II. RELATED WORK
Secure routing, privacy, and trust in VNets have gained atten-
tion from the research community over the past few years.
However, private communication is still in its early days, and
routing remains challenging. One of the first yet important rout-
ing protocols for VNets is epidemic routing [2], in which each
packet is replicated until reaches a certain destination. A carry-
and-forward strategy is used in vehicle-assisted data delivery
in vehicular ad hoc networks (VADD) [3], where messages can
be carried by vehicles in sparse networks when there are no
neighbors to which it can be forwarded. VADD always tries
to use vehicles on the fastest roads available for forwarding
the message. However, vehicles may deviate from predicted
routes and the routing path should be continuously recalcu-
lated. D-Greedy and D-MinCost are two unicast forwarding
methods proposed by Skordylis and Trigoni [4]. Unlike VADD,
they tried to minimize the number of transmissions instead of
focusing on minimizing the delivery delay.
Kamvar et al. [8] concentrate on trust in the peer-to-peer
(P2P) file-sharing networks, based on the peers’ history of
uploads, called EigenTrust. They proposed an algorithm aim-
ing at decreasing the number of downloads of inauthentic files
in the network that assigns each peer a unique global trust value.
They also described a distributed and secure method to compute
global trust values, based on power iteration. In fact, peers use
these values to choose from whom they download, whereas the
network identifies malicious peers and isolates them from the
network. This proposal mainly concentrates on P2P and directly
downloading the files. However, in a dynamic environment,
such as vehicular ad hoc network (VANET), with a high mobil-
ity, using the concepts such as score manager is not feasible.
The framework presented in [9] aimed at dealing with
potentially untrustworthy information. The framework includes
a computational trust model for estimating the amount of
received information uncertainty. In order to realize the drivers
goals, the framework also has a probabilistic beliefs-desires-
intentions agent system for reasoning about this uncertain infor-
mation. In order to secure, DTN routing toward efficient trust
establishment, iTrust scheme proposed by Zhu et al. [10], as
probabilistic misbehavior detection. They used a trusted author-
ity (TA) to judge the nodes behavior, which is periodically
available, based on the collected routing evidences and proba-
bilistically checking. For the TA to ensure the security of DTN
routing at a reduced cost, they modeled the inspection game and
use game theoretical analysis to demonstrate that by setting an
appropriate investigation probability. Also to improve the effi-
ciency of the proposed scheme, they allows a dynamic detection
probability determined by the trust of users by correlating
detection probability with the reputation of that node.
The proposed mechanism in [11] aimed at ad hoc environ-
ment by designing an ad hoc on-demand distance vector routing
(AODV)-based routing protocol that combines the hop count
and trust value. When a source studies different path opportu-
nities, it multiplies the trust value of each node to obtain the
trust value of the path. Although it may be true in the packet
delivery ratio, the trust value of the path cannot be always cal-
culated by a multiplication of the trust value of the nodes along
the path, e.g., in anonymity-based one. Moreover, the proposed
routing protocol in [12] is trust-based aimed at mobile ad hoc
networking (MANET) that concentrates on energy efficiency.
Situation-aware trust (SAT) [6], a trust architecture for
VNets, reveals the identity of the users by using a social
network. The trust system proposed in [7], designed for the
receivers and broadcast communication, shifts the concen-
tration from the sender to the data itself. In fact, the content
of the message defines the trust value even the message is
generated by different vehicles that have different trust levels.
The Markov chain-based trust model for VNets presented in
[13] is a hybrid model evaluating trust based on the data and
the entity together, in order to filter out malicious and selfish

ROSTAMZADEH et al.: CONTEXT-AWARE TRUST-BASED INFORMATION DISSEMINATION FRAMEWORK FOR VNets 123
nodes. Aside from the complexity of the proposed method,
it is not clear how this algorithm works when there is no
connectivity in the network or when the application needs a set
of delay-reliability requirements.
The aim of a trust and reputation infrastructure-based pro-
posal (TRIP) for vehicular ad hoc networks [14] is broadcast
communication in VNet. It is designed for a receiver evaluating
the trustworthiness of the received message, while the message
can be accident (safety) or a weather condition (nonsafety).
Their guideline for designing a trust system in VNet identi-
fies five characteristics such as: simple, light, and fast; accurate;
scalable; resilient to security and privacy threats; and not depen-
dent on mobility patterns. The scope of A vehicle ad hoc
network reputation system (VARS) [15] is also broadcasting
messages. A relay node in forwarding a message, generates
its own opinion based on the sender if it is known, or based
on similar received messages in the past, and sends the opin-
ion along with the message. However, considering dynamic
nature of VNets, VARS mainly relies on others opinion along
with indirect trust values for a receiver on trusting a message,
which are not enough trusted and valid sources for the receiver.
Moreover, the proposed trust model in [16] concentrates on
broadcast communication of safety related messages where is
limited to utilizing the signature of a message issued by the
originator to preserve trust of the messages.
The trust model proposed by Wang et al. [17] is designed for
senders. A sender finds similar vehicles to forward the message.
Finding similar nodes based on location, energy, and brand is
not practical and does not guarantee of being a better candidate:
the same brand does not make any difference; a node with the
same energy level cannot be a good one, if the sender is frus-
trating of low energy where requires a neighbor with a higher
energy; and energy is not a case in the VNets at all.
In a highway and reporting the traffic congestion, a mali-
cious vehicle may send bogus congestion information to
others, which is the focus of Huang et al. [18]. In their
design, they leveraged traffic fellow theory to observe the
Kinematic wave caused by congestion. So without a need to
a centralized controlled congestion detection and prediction
system, each vehicle only relied on local speed and distance
measurements to validate the congestion event that is sent
by another vehicle. In [19], an road side unit (RSU) and
beacon-based trust management system are proposed that
aimed at broadcasting message opinions quickly. Therefore,
and in a privacy-enhanced VANET environment, if an internal
attacker sends or forwards forged messages, the message will
be foiled. Eiza and Ni [20] leveraged location and velocity
information of vehicles in order to calculate link reliability
for their reliability-based routing scheme for VANETs. Their
objective was routing process in which they tried to facilitate
quality of service (QoS) support in it.
In order to receive trusted traffic information by a vehicle,
Teler and Cristea [21] designed a trust-based security system,
where regardless of source of the data, the trust is measured for
individual pieces of data referring to a specific event. In this
model, a driver evaluated the trust of received information, and
then disseminates the trust value to other vehicles in order to
improve the accuracy in the trustworthiness of an event.
In [22], a decentralized lightweight authentication scheme
for vehicle-to-vehicle communication networks is proposed,
which is called trust-extended authentication mechanism.
In fact, they aimed at improving the performance of the
authentication procedure by using the concept of transitive trust
relationships. Gazdar et al. [23] used Markov to formalize vari-
ation and its stability of the trust metric in their trust model for
VANET. They considered the dynamic nature of the trust met-
ric due to the vehicle behavior. In addition, they brought into
account constraints related to the monitoring process. In their
model, each vehicle monitors and updates the trust metric of its
neighbors based on the behavior of the neighbors.
In [24], Haddadou and Rachedi proposed DTM2
aimed at
issue of malicious and selfish nodes in the VANET. They used
the concept of cost, in which they assumed a sender transmits
a signal, which has a cost based on the message and sender
behavior, with its message to show a guarantee of the truthful-
ness of the message to the potential receivers. In this paradigm,
the cost is calculated in a way that the cost increases when the
behavior of the sender gets worse. Therefore, the cost for the
malicious and selfish nodes increases eventually, as a punish-
ment. Similarly, the cost for the benign nodes decreases as a
reward proportionally to the signals value. In [25], Liao et al.
brought into account the trustworthiness of the message orig-
inator and the forwarding nodes, if any, to calculate the trust
of the message itself, in a vehicle-to-vehicle (V2V) communi-
cation. They used the existing vehicle-to-infrastructure (V2I)
communication facilities deployed and managed by central
traffic authorities for collection of the vehicle behavior in a
crowd-sourcing fashion.
III. PROPOSED TRUST MODEL
In this section, first we describe some of the definitions for
our framework. Then, calculations regarding trust value and the
total trust are presented to be used in our design as part of
Section IV.
The level of performing the expected service by a trustee is
called satisfaction, which in fact derived the trustworthiness.
We assume the trust value to be a number between (0,1). The
four major parts of the trust system are initial trust, trust metric,
operation, and trust management. Identifying the appropriate
trust metric, which is relevant to the application, and a mecha-
nism to measure the metric, e.g., a binary or percentage unit, are
part of a trust system. In addition, designing a system that man-
ages the trust is needed, which can be centralized, distributed,
or a combination of them.
Definition 1: Trust assessment can be direct or indirect. In
direct assessment, a trustor relies on its own experience about a
trustee. For instance, in a wireless communication environment,
if node S (trustor) sends a packet to node R (trustee), which
is one hop away, to forward (service) the packet to node D,
which is two hops away, S can overhear R, and observes if R is
genuinely sending the packet. On the other hand and in indirect
assessment, a trustor relies on other nodes experience of dealing
with a trustee.
FACT is based on carry-and-forward information dissemina-
tion. Therefore, if there is no vehicle in the vicinity, the vehicle

Fig. 1. Right: map of the city divided to different neighborhoods. Left: each
neighborhood consists of different road segments.
carrying the message should continue carrying the message
until it comes across another vehicle and then decide on for-
warding the message. The rest of this section is based on the
assumption of having at least one vehicle in the transmission
range. FACT is a general framework that allows messages of
different categories of traffic to be securely delivered to their
destinations based on their requirements. As shown in Fig. 1,
all vehicles maintain a trust map, which divides the city into
several neighborhoods and each neighborhood itself is divided
to multiple segments. A segment is part of the road between
two intersections. Our work is not limited to a specific sector-
ing method; it can be symmetrical or asymmetrical sectoring.
Map division should be done in a way that meets the net-
work administrator’s resolution requirement without imposing
excessive overhead. In the mentioned trust map and for every
segment, a trust value TR is recorded for each time of the day
in each specific day of the week. For simplicity, we only con-
sider rush hour and regular hour for one day, i.e., two entries
per day and 14 entries in total per week for each segment. It
is possible to further simplify and only differentiate between
weekdays and weekends, which lowers the number of entities
for a segment to four. Every vehicle starts with an initial trust
value of 0.5 for each neighborhood. Then, it gradually updates
different entries of the trust map for that segment/neighborhood
based on its experience. For example, if this vehicle experiences
a trustworthy communication in that segment during rush hour
in weekdays, then it increases the corresponding trust value for
those times and days and makes it close to 1. Whenever a vehi-
cle is calculating the trust value of a road segment for the first
time, it uses the neighborhood’s trust instead. The details of the
trust model and the updates are discussed in Section III-A.
Note that the key component of our trust framework is that
instead of assigning a trust value to each node, which is not
scalable and hard to maintain and even inefficient, we assign a
trust value to a neighborhood and to each segment of that neigh-
borhood. In other words, our main contribution is that FACT is
space-centric, i.e., it uses location information to evaluate the
trustworthiness of messages. There is a similar scenario in a
real city where some neighborhoods are safer than the others.
This is the intuition behind FACT. Note that there is no prede-
fined trust values for different neighborhoods because all cars
start the FACT by assigning the same value of 0.5 to all neigh-
borhoods and then they gradually update their database as they
learn more about different areas. However, in different times of
the day, cars from other parts of the city may travel to these
areas and lower the actual trustworthiness of these areas. That
is why FACT assigns different trust values to different times
of the day and constantly modifies these values. The biggest
advantage of FACT is its scalability and low complexity even
for a metropolitan area. The other advantage is its efficiency. In
approaches, such as [6] and [26] where there is a trust value per
node, resources are often wasted because we never come across
that car again, but it is much more likely that we travel to the
same area more than once.
A. Trust Calculation
In this paper, trust is assumed to have multiple dimensions
based on the service requirements of the corresponding appli-
cation, e.g., delay, reliability, security/confidentiality, privacy/
anonymity to name a few. FACT guarantees security and offers
QoS based on the application requirements. Different appli-
cations on a VNet can be divided to three general categories
based on their specific set of requirements. 1) Cata: safety-
related services such as accident, high traffic, and emergency
brake notifications, and critical services when there is a dis-
astrous situation like earthquake in place. In this category,
delay, end-to-end reliability, and data integrity are important.
Some safety-related applications may be more tolerant to delay
than others like when a road condition needs to be reported
to incoming cars. As for data integrity, receivers must be
assured that the information based on which they are mak-
ing decision is genuine. 2) Catb: infotainment such as parking
availability and gas price notifications. Reliability, access con-
trol, source anonymity, and integrity are the main parameters
to be considered in this category. Access control assures that
only appropriate pairs can communicate to each other since
each piece of information is destined for a specific vehicle or a
set of vehicles. Source authentication along with access control
helps vehicles trust their communication by making sure infor-
mation is coming from a reliable source. 3) Catc: third-party
services such as when taxis working for the same company
want to share passenger availability information. Applications
in this category need source authentication to guarantee that
their information is from the right person, and reliability to
maximize number of cars receiving the information.
The overall trust value for a given path m is defined as
tTRm = FTR((α1, TR1), (α2, TR2), . . . , (αk, TRk)) (1)
where FTR is a function that combines the trust dimensions
TRi to obtain the total trust value of tTR. Note that the value
of αi represents the weight of the TRi, e.g., delay, in the
calculation of the total trust driven by the application. Given
the category of the traffic as mentioned above, FACT assigns
appropriate αi’s and calculates the overall trust for a path. One
simple example of the function is a linear combination by hav-
ing αi = 1/k for ∀i. One other example is having 0 for the
nonimportant αi’s.
The model in [27] is used for calculating trust between agents
in a multiagent environment. TRi is defined as a function of its
current and historical values as follows:
(TRτ
v)i = η × (TRcur)i + (1 − η) × (TRτ
v−1)i (2)
where τ is the period of time (e.g., an hour, one day, or
one week) and n is the number of the experiments that are

considered so far. The value of η (0 ≤ η ≤ 1) is used to give
weight to the trust value of current period (TRcur)i, compar-
ing to last v − 1 periods ((TRτ
v−1)i) of the system in final trust
calculation ((TRτ
v)i). Since we only rely on the direct trust cal-
culation in this model, the current trust value (TRcur)i is equal
to the final satisfaction level of current period as per (3)
(TRcur)i = STτ
n (3)
where STτ
n is the satisfaction value of the trustee calculated by
a trustor and is defined as
STτ
n = γ × STcur + (1 − γ) × STτ
n−1 STτ
0 = 0.5 (4)
where γ (0 ≤ γ ≤ 1) represents the weight that is given to the
current experiment (STcur) comparing to the last experiments
in this period (STτ
n−1) to calculate the final satisfaction value
(STτ
n).
Satisfaction (STτ
n or STcur) is a value between zero and one,
where one means fully satisfied and zero means completely
unsatisfied. We use the beta function to calculate the satisfaction
value via (5), as it is the most referred function in the literature
STcur =
Nsuc + 1
Nfail + Nsuc + 2
(5)
where STcur is the satisfaction level based on the number of
successful (Nsuc) and failed (Nfail) operations/services in an
experiment, which consists of one or multiple packet(s) deliv-
ery. For instance, in a safety application, there may be only one
packet while in the online social networking, the experiment
consists of multiple packets.
Finally, every vehicle uses the total trust on each path as the
reward of the path (1/cost) to choose the best path. In other
words, FACT maximizes the trust value and selects the path that
delivers the maximum trust. Let Set P denote the set of l paths
where each one has a total trust value of tTRm (1 ≤ m ≤ l).
The path with the maximum total trust value of tTRm is the
selected path as per (6)
max, tTRm
subject to, m ∈ Set P.
(6)
It is easy to be shown that (6) is a concave function with a
global maximum.
IV. FACT FRAMEWORK
We make the following assumptions for the vehicular net-
work in which FACT is implemented.
1) Vehicles know their location using global positioning
system (GPS) or other localization methods.
2) There are very few (e.g., three) access points deployed in
different parts of the city. They are used to propagate sys-
tem updates and urgent notifications. Traffic information
is uploaded into all vehicles once they join the vehicular
network. After that and every time they meet an access
point, this information is updated.
3) Similar to approaches such as [3], [4], [16], and [19],
vehicles transmit beacon periodically to let their neigh-
bors know about their existence. Each beacon has
Fig. 2. FACT’s framework and how modules function and communicate to each
other.
information about the location of the car, its velocity,
and the direction it is heading. Using these beacons,
each vehicle can determine the neighborhood in which its
neighbor is located in order to calculate its trust value.
4) Vehicles are equipped with a temper proof device.
Moreover, they can connect to a certificate authority using
an access point to receive their certificates. Therefore,
they can sign their packets, e.g., source authentication.
5) Whenever there is an incident in the network, such as
accident or road problems, two notifications are sent:
one at the beginning of that incident and another when
the incident is finished. These notifications are called the
incident and clearance notifications, respectively.
Note that the first four assumptions are typical in any vehicular
network. The last assumption increases the overhead; however,
this increase is far less than alternative approaches such as send-
ing an acknowledgment for each message. This small increase
in overhead is traded to achieve a much better security level.
When a message is received by a vehicle, it has to make
sure the message is authentic. FACT is a two-layer framework,
which consists of admission and dissemination modules. We
divide the vehicle to three types: 1) source, which originates the
message; 2) relay, which is in the middle of the path and helps
pass the message to its destination(s); and 3) destination that is
the last node on the path for which the message is intended. All
these types have the FACT framework presented in Fig. 2.
A. Admission Module
When the receiving vehicle is a relay or destination node, the
process of authenticating the message happens in the admission
module. FACT applies three checks to the message to make
sure it is authentic. First of all, the message’s origin should
be determined. To do that, each node i on the path (including
the source) is required to add its location, i.e., its latitude and

longitude (xi, yi) which can be translated to its road segment
and its neighborhood or (ri, ni) to the message header. In
addition, each node has to also add a time-stamp ti to the mes-
sage header showing the current time. Time-stamp and location
fields are both assumed to be 4B. xi and yi each has 2B allo-
cated in the location field. Since the message size is assumed
to be 512B, 8B overhead has a negligible impact on the over-
all performance of FACT. Using the mentioned fields, when a
message is received, the receiver can easily identify the loca-
tion of the source, e.g., locs = (rs, ns). Then, it can use its trust
database and based on TRrs
and TRns
, decide whether it can
trust the message’s source. FACT uses a simple rule in which
a location loci = (ri, ni) is considered trusted if the following
inequalities hold
TRri
> Φr
TRni
> Φn
(7)
where Φr and Φn are acceptable threshold for having a trusted
road segment and neighborhood, respectively. The values of
Φr and Φn are tuned by the network administrator based on
the application and the characteristics of the environment. The
same check is applied to all nodes on the path in addition to the
source in which case the message passes the first check only if
(7) holds for all previous nodes. Even the current location of
the message, i.e., (rx, nx) for node x, is evaluated using (7).
If both inequalities in (7) hold for (rx, nx), then it means that
x can trust its current neighborhood. When the message passes
the first check, FACT applies the second check.
In the second check, there is a metric that enables FACT
to detect any malicious behavior along the path: the time-
location pairs (ti, loci) attached to the message where each pair
corresponds to one vehicle on the path. If any of the nodes
manipulates its location or/and time-stamp in the message to
deceive other nodes about the trustworthiness of its (ri, ni) pair
(and consequently, the values of TRri
and TRni
), the next node
on the path x can use (ti, loci) for {∀i, i = s, 1, 2, . . . , x} and
easily spot any suspicious information in the (ti, loci) pairs.
In fact, x takes advantage of traffic information it has about
(ri, ni) pairs to come up with an estimate of the time it takes to
reach from (rs, ns) to (rx, nx) through the path traveled by the
message, denoted by tes.
To calculate tes, let ¯vi denote the average vehicle speed of the
road segment ri using the historical traffic information. When a
message is received by a car x and depending on the imple-
mented forwarding policy and the current vehicle density, it
either forwards the message instantaneously (or with a very
short-processing delay) or it decides to keep and carry the mes-
sage for a while, e.g., until it reaches another vehicle. These
two cases are called forward and carry phases, respectively. Let
tf denote the transmission time plus the processing delay and
tc(d) denote the time that it takes x to travel d meters. We
always have tc(d) tf . Assume each message goes through
k1 forward and k2 carry phases until it reaches node x. Vehicle
x calculates k1 and k2 by looking at ti − ti−1 for {∀i, i =
s, 1, 2, . . . , x} and incrementing k1 when ti − ti−1 ≤ tf ± δ1
and incrementing k2 when ti − ti−1 ≤ tc(d) ± δ2. Parameters
δ1 and δ2 are tuned based on the channel conditions and the
vehicle speed variance, respectively.
Fig. 3. Example for distance calculation in the carry phase to find tc(d).
For two successive nodes i and i + 1, tc(d) is calculated as
given in (8)
tc(d) =
⎧
⎨
⎩
d(i,i)
¯vi
, if i and i + 1 are on the same road segment
d(i,int1) + d(int1,int2)+···+ d(intj,i)
¯vi
, if otherwise
(8)
where d(i,i) is the distance that vehicle i traverses until it
decides to forward the message to vehicle i + 1. When i and
i + 1 are on different road segments, vehicle i has to travel until
it reaches i + 1 and on its way, it may cross j intersections. The
nominator in the second term in (8) shows the total distance
traveled by i before the next transmission. An example of this
scenario for j = 2 is given in Fig. 3. d(i,int1), e.g., is the dis-
tance between vehicle i and the intersection int1. Using (8), x
calculates tes as follows:
tes = k1tf +
k2
k=1
tc(d(k,k+1)). (9)
Then, x compares the calculate estimate with the actual time
tac = tx − ts and if tac − tes > Δ, x concludes that the mes-
sage is altered on its way. The value of Δ depends on the size
of a neighborhood and time it takes to traverse it because if
one vehicle decides to manipulate the message, it has to change
its location to a more trusted neighborhood different from its
current one.
If the message passes the second check as well, x applies the
third and final check. In the third check, x takes advantage of
the historic knowledge about the location of the source. In fact,
this check is a content-based check in which case x validates
the event/incident reported in the message. As an example, sup-
pose the message is about an accident at (rs, ns). x looks at
its knowledge about that location and if rs is a high-traffic
road with prior accidents, it concludes the content can be valid.
However, if rs is a road with very light daily traffic and no his-
tory of accidents, it sees an anomaly which may or may not

be true. For this purpose, we define an accident threshold ζ to
differentiate between a high accident area and a low one. Note
that the system administrator can tune this threshold in addition
to other system parameters. If x is a relay node, it sets a flag
in the message suggesting that the received message is bogus
and it cannot be trusted. It then pushes the message down to
the FACT’s dissemination module to be forwarded to the next
node. On the other hand, if x is the destination node, it waits
until it receives the same information from multiple sources
(three sources for example) then takes an appropriate action,
e.g., it uses an alternative route. If another source confirms that
incident without the flag being set, because of the contradiction
between different sources, x cannot decide whether the mes-
sage is genuine or not. Therefore, it does not take any action
except saving a copy of the message for some time Tinc, which
represents the average delay between the incident and clearance
messages and is set by the network administrator. If during Tinc,
x receives the clearance notification, meaning that the report
was correct, it recalculates the corresponding trust values in its
database. Otherwise and when Tinc expires, meaning that the
report was bogus, x just discards the message. Similarly, when
x is a relay node and it realizes that the message was actu-
ally correct (using Tinc), it updates its knowledge by adding
that incident to its database. This means that bogus messages
and the corresponding attacks do not hurt the trustworthiness of
an area.
B. Dissemination Module
When the receiving vehicle x is a relay node and if the mes-
sage passes all three checks explained in Section IV-A, x pushes
the message down to the dissemination module in order to be
forwarded to the next vehicle. Four cases can be imagined for
FACT’s dissemination module based on the type of the node
and its location. Case I: when it is a source node at intersec-
tion, it follows the pseudocode given in Algorithm 1. Note that
each vehicle knows the location of its neighbors and the corre-
sponding trust values using the periodic beacons. Case II: when
it is a relay node at an intersection that wants to decide on the
next hop, because the traffic category Cat is already in the mes-
sage, the relay node follows Algorithm 1 from step 3) to the
end. Case III: when the vehicle is a source and in the middle
of a street, it takes the first two steps of Algorithm 1 and then
forwards the packet to the closest neighbor to the destination. In
fact, all vehicles on that segment have the same segment trust
value and it does not matter, which vehicle is selected as the
next hop. Hence, it is best to send the message to the closest
node to the destination. At the end, the sender updates its trust
value for that segment. Case V: when the vehicle is a relay node
and in the middle of the road segment, it just forwards the mes-
sage to the closest neighbor to the destination and updates its
trust value for that segment.
Note that in situations where there is an adversary node in a
trusted area, STcur will be dropped temporarily and so is tTR
for that neighborhood. But, when this node moves out of the
trusted area, the total trust will go back to normal. There are
cases where all possible paths have very weak total trust val-
ues. In those cases, the vehicle carrying the message, regardless
Algorithm 1. FACT at Intersection
1: Based on the content of the message or the nature of the
event, the appropriate traffic category, Cat, is determined.
2: Based on the selected category, the corresponding αi’s are
specified.
3: For set of neighbors, j = 1 : w:
4: For set of parameters, i = 1 : k:
5: TRj
i is calculated.
6: For set of paths, m = 1 : l:
7: tTRm is calculated using the αi’s calculated in step 2.
8: tTRmax is identified.
9: The winning path, m∗
, is selected and the message includ-
ing the traffic category, Cat, is forwarded to the neighbor
on m∗
, i.e., j∗
.
10: FOR i = 1 : k:
11: Based on the forwarding experience, STm∗
i and TRm∗
i
are updated.
of being the source or a relay node, uses a threshold TRmin.
If ∃i where αi = 0 and TRi
m ≤ TRmin for ∀m ∈ Set P,
then the vehicle continues carrying the message until the next
intersection to see whether the trust condition changes on that
intersection.
There are situations when due to a temporary reason, for
instance a construction project or a festival, the trust condition
of a road segment or a neighborhood is changed and so is the
corresponding trust value in the trust map of vehicles traveling
to that area. But when that reason does not exist anymore, if
cars use (2), it will take a while for the trust value to converge
to the actual value and we say the trust value has a bias. To
address this, we consider a parameter, NUMbias as the number
of times FACT allows the current trust TRcur to be hugely dif-
ferent from the historical trust. For example, if NUMbias = 3
and this is the fourth time that the current trust is much bigger
than the historical one, then FACT uses the average of the last
three values of TRcur as the trust value for that segment. The
size of the gap between the current and the historical trust, i.e.,
the size of the bias, should be carefully selected to avoid erro-
neous decisions. One example could be 50% of the historical
trust.
One limitation of FACT is when a malicious vehicle trav-
els in a known-to-be-trusted segment. Therefore, all messages
that are relayed to this vehicle will be either manipulated or
destroyed. The other vehicles receive a very bad feedback and
they try to update their trust estimate for this segment. Using the
above-mentioned mechanism and assuming the malicious vehi-
cle corrupts the packets more than NUMbias times, the trust
value will be declined dramatically. But, as the malicious node
exits this segment, the trust value for that segment goes back to
normal after at most NUMbias times of genuine transmissions.
Based on this discussion, the impact of infiltrating a malicious
node in a trusted segment can be minimized by choosing a small
value for NUMbias, so that both discovery and recovery periods
are small.
Finally, it is obvious that the same satisfaction value caused
by a car experiment calculated via (5) will affect the trust value

of the segment as well as the neighborhood, as per (2) and (4).
However, the effect of the current value should have less influ-
ence on the neighborhood compare to the segment. Therefore,
the system administrator, which sets the parameters of the vehi-
cle, may set the values of γ and η of the neighborhood less
than the parameters γ and η of a segment in that neighborhood.
There is a similar situation when setting the values of τ and n
in (4) and τ and v in (2), where the administrator needs to be
careful in choosing the appropriate values for the segment and
for the neighborhood.
V. ANALYSIS
In this section, we analyze our proposed framework from the
security point of view and then present a qualitative comparison
between FACT and other trust-based approaches.
A. Security Analysis
We follow the well-known Dolev–Yao approach [28] and
provide an adversary analysis. Dolev–Yao assume that all pack-
ets are delivered to and received from an adversary, which is
capable of recording, deleting, replaying, rerouting, reordering,
and rescheduling the packets. To attack this system, an adver-
sary can be a sender or a relay node. Therefore, we provide three
adversary models representing three positions of the adversary.
In our first model, our adversary is a sender, and in second and
third models, our adversary is a relay node.
1) First Adversary Model: In this model, we assume our
adversary is a sender. To be more precise, he sends the bogus
messages about, e.g., traffic.
Objective: The objective of the adversary can be misleading
the receivers about the traffic in an area (e.g., street), and/or
defecting a neighborhood/segment.
Initial capability: Initially, he knows the detailed information
about our proposed framework as well as the overall topology
and map of the area.
Capability during the attack: He is capable to send many
bogus massages about an event that never happened. Moreover,
he is capable of modifying the source of the messages to be
other nodes, a benign/valid car, or a dummy car. Also, he can
keep driving and move from one area (neighborhood/segment)
to another area to harm the trust values of multiple areas.
Discussion: As per our design, explained part in Section
IV-B, having NUMbias, a receiver wait for NUMbias reports
about the same event to rely on the report. This value can be
tuned up if necessary, if the system get under attack, espe-
cially distributed attack. Note that distributed attack in this
case means our adversary should have multiple cars in differ-
ent street and areas to be able to completely destroy the trust
evaluation and misleads the receivers, which is a very expensive
attack and may not be feasible.
Finally, since the adversary is the packet generator, a receiver
may get mislead for the first time; however, after a few catching
the errors and finding out about the bogusness of the packets,
the receiver can identify that particular sender as a mali-
cious/attacker and denies entire information generated by the
sender from that point.
Note that if our adversary generates the bogus messages and
modifies the sender of each one to be a benign/valid car or a
dummy car to perform the attack, it still may affect the trust
value of an area. It is obvious that if he tries to impersonates
a benign car, the benign car will receive the message as well,
as a receiver per our design, and will distribute a correction
message to disregard the bogus one. In case of impersonating a
dummy car, as per our discussion in Section IV-A, our proposed
solution catches and ignores that attacks.
2) Second Adversary Model: In this model, we assume our
adversary is a relay node. To be more precise, he does not gen-
erate any packet, and instead, he only relies the packet sent by
pother while modifying them.
Objective: The objective of the adversary can be destroying/
modifying the senders messages about an event and about an
area.
about our proposed framework and also he is well equipped to
receive and then sends the packets as well as showing himself
as a ready node to relay.
Capability during the attack: He is capable of keep receiving
and sending the messages. He can read inside the packet and
can modify any part of the message, like the time stamp, street
name, and overall the event. For instance, he can change a mes-
sage instead of saying “there is an accident,” he can modify it to
“the accident is cleared/traffic is smooth.” He also can receive
the message and drops it, or deliver it to a wrong direction, in
case of unicast/multicast communication. In this case, he may
modify the destination as well to mislead other relay nodes
that will receive this message, which is passing the adversary
node.
Discussion: First of all, in our design, we assumed there is
an under-layer security mechanism such that the integrity of
the messages and packets are preserved. In this case, when the
affected/modified packet are received by the next relay node/ or
a receiver (if the adversary is the last node before the receiver),
the receiver and the next relay nodes are capable of catch-
ing the attack as part of their admission module of the FACT
framework (Section IV-A).
In case of the next node being a receiver, the receiver does
not rely on packet received from the adversary anymore and
expects the message from others.
In the case where the next node being a relay node, still the
next relay node is capable of finding the attack and therefore
easily can stop cooperating with the adversary for the following
communication.
Note that, since each relay node should adds a time stamp to
the packets, if the adversary attacks the system by more than
one car, meaning damages the packet by first car, and passing it
to the next attacker car to relay, still along the path, or in worth
case by a receiver, the attack can be discovered. In either cases,
the benign node(s) can stop their cooperation with the packets
passing the attackers.
As per Section IV-B by having NUMbias, a receiver waits for
NUMbias number of the reports about the same event to rely
on the report. This value can be tuned up if necessary, if the
system get under attack, especially distributed attack. Note that
distributed attack in this case means our adversary should have

TABLE I
COMPARING THE FEATURES OF THE SURVEYED PROPOSALS IN SECTION II AND OUR PROPOSAL
multiple cars in different street and areas to be able to com-
pletely destroy the trust evaluation and misleads the receivers,
which is a very expensive attack and may not be feasible.
Finally, since the adversary is the packet generator, a receiver
may get mislead for the first time; however, after a few catch-
ing the errors and finding out about the bogusness of the
packets, the receiver can identify that particular sender as a
malicious/attacker and denies entire information generated by
the sender from that point.
3) Third adversary model: In this model, we assume our
adversary is again a relay node. However, unlike the second
model, he does not change the messages; instead, he forwards
the messages along a wrong path (with low-trust value).
Objective: The objective of the adversary can be disaffect-
ing the messages about an event. The message can be about an
accident happening or about clearing the traffic caused by an
accident. With a good approximation, the former one is sim-
ilar to a good-mouthing attack, and later one is similar to a
bad-mouthing attacks.
about our proposed framework as well as the overall topology
and map of the area.
Capability during the attack: He is capable of playing the
role of a relay node. He forwards the messages, but not to the
proper direction. Therefore, he keeps his reputation to others as
a benign node although he is performing the attack. Note that
this attack can be similar to the black-hole attack in a unicast
communication scenario where the relay node sends the packets
to a malicious node to be dropped at that node.
Discussion: As per our previous discussion, our applica-
tion is data dissemination. Therefore, we do not have a single
receiver (like in the unicast communication scenario). In fact,
the messages are being forwarded and disseminated by other
cars, as well. So, if an attacker tries to send the messages
to a path with low-trust value, to perform the attack, other
benign cars will redirect the messages to the proper path (with
high-trust value) to be disseminated.
B. Comparison Analysis
Table I presents a brief summary of the comparison between
the studied models in Section II and our proposal. Rows of
the table show the characteristics of the mechanism/systems,
as follows.
1) Data-centric: If the mechanism/system concentrates on
the content of the packet.
2) Entity-centric: If the mechanism/system concentrates
on the entity, e.g., sender, receiver, and/or a relay
node/vehicle.
3) Source-centric: If the mechanism/system is designed to
be used by the senders, e.g., for choosing the next hop or
the best path.
4) Receiver-centric: If the mechanism/system is designed to
be used by the receivers, e.g., for trusting the received
message.
5) Path-centric: If the mechanism/system is designed to be
used for choosing a path or evaluating the trustworthiness
of the path.
6) Infrastructure-based: If the mechanism/system requires
having an infrastructure, including road side unit or even
certificate authority.
7) Application-oriented: If the mechanism/system considers
requirements of the application that is running.
8) Broadcast/unicast: If the mechanism/system uses broad-
casting or unicasting for communicating between
vehicles.
FACT is data-centric because it checks the content of the
message once for determining traffic category and again in the
third check of the admission module. It is entity-centric simply
because FACT assures that the message can be trusted only

TABLE II
CONFIGURATION PARAMETERS
when all the nodes (entities) on the path are trustworthy (at
least to some extent). FACT is both source-centric and receiver-
centric because each receiving car first makes sure the received
message can be trusted (in admission module), and then it
chooses a trusted path for forwarding the message (in dissemi-
nation module). Using the same reasoning, it is easy to see that
FACT is path-centric, as well. Since having road-side units are
not necessary for FACT to work, it is not infrastructure-based.
It is application-oriented because it specifies αi’s based on the
application. Finally, both broadcast and unicast are supported
by FACT.
VI. PERFORMANCE EVALUATION
To evaluate the performance of the FACT, we ran an exten-
sive set of simulations using MATLAB. We divided the whole
city into three regions. 1) Very trusted neighborhood Nt where
90% of transmissions are genuine while 10% of times packets
are destroyed. The size of Nt, on the other hand is bigger than
the other two regions and it has a lower vehicle density result-
ing in a bigger delay. 2) Untrusted neighborhood Nu where
only 10% of transmissions are genuine. However, Nu has a
smaller size and higher vehicle density compared to other two
regions, which means a smaller delivery delay. 3) Semitrusted
neighborhood Ns where 50% of transmissions are genuine. The
size and the vehicle density in Ns are between the respective
values in Nt and Nu, which results in a delay value between
other two regions’ delays. The application is assumed to be a
safety-related service with αi = 1/3 for delay, reliability, and
data integrity. We assumed a transmission range of 250 m and
Rician fading with shadowing as the propagation model. The
velocity of cars is assumed to follow a normal distribution
according to work by Wisitpongphan et al. [29] using real traffic
traces. Simulation parameters are summarized in Table II.
The results of the simulations for delay are reported in Fig. 4
for FACT versus epidemic routing and VADD. As shown in
this figure, FACT outperforms other two algorithms in terms of
delay while achieving the same level of reliability although they
use faster paths. FACT performance versus VADD is particu-
larly interesting because even though VADD chooses the fastest
path and benefits half of FACT’s communication delay per
transmission, delivery delay in FACT is significantly improved
(between 400% and 600%). The reason is the excessive num-
ber of retransmissions in VADD as a result of packet drops in
Fig. 4. Delivery delay versus different vehicle densities for 90% reliability.
Fig. 5. Number of times the source retransmits its message because of the
packet drops by malicious nodes.
Fig. 6. Packet delivery ratio versus delivery delay.
the untrusted neighborhood (see Fig. 5). Each time a packet is
dropped by a malicious vehicle, the source vehicle has to resend
the original message, which leads to higher delivery delay.
On the other hand, packet delivery ratio (or the end-to-end
reliability) is the other victim of untrusted vehicles. As shown
in Fig. 6, FACT jumps to maximum packet delivery ratio in
a much shorter time compared to epidemic routing and par-
ticularly compared to VADD. It means that FACT achieves
higher reliability in a shorter time. Note that although epidemic
routing is a close competitor to FACT in these figures, but

Fig. 7. Delivery delay versus different average vehicle speeds for 90%
reliability.
Fig. 8. FACT packet delivery ratio versus delivery delay for different average
speeds.
because of the flooding nature of it, epidemic routing imposes
a huge communication cost in terms of the interference and
congestion to the network. However, FACT has a communi-
cation cost comparable to VADD, which is much less than
epidemic.
Next, we study the effect of vehicle speed on delivery delay.
A comparison between FACT and VADD and epidemic routing
in terms of delivery delay against different speeds is presented
in Fig. 7. The average vehicle speed ranges from 50 km/h,
which is typical for an urban scenario to 110 km/h, which is
more suited for a highway environment. As shown in this figure,
vehicle speed’s impact on delivery delay is very high for VADD
and very low for FACT. As expected, increasing the speed
reduces the delivery delay for all three algorithms.
Finally, the impact of vehicle speed on the FACT packet
delivery ratio is studied. As shown in Fig. 8, there is a fairly big
gap between when the speed is 50 km/h and the rest of speeds.
It seems 50 km/h is a transitional speed in this regards. The
results for speed of 70 and 90 km/h are very close to each other.
The next transitional speed seems to be 110 km/h.
VII. CONCLUSION
In this paper, first we present a survey on routing, security
and trust systems/mechanisms, mainly in vehicular networks
and ad hoc environment. Then, we propose a two-layer trust-
based information dissemination framework called FACT and
its application in vehicular networks. FACT is an application-
oriented framework designed to support broadcast, multicast,
and unicast communication in vehicular networks. FACT main-
tains the trust values of the neighborhoods and segments of the
city. It first makes sure the message is originated from and trav-
eled through trusted nodes and the content is valid. It then uses
stored trust values to choose the best path to route the mes-
sage, or if needed, to carry the messages. Simulation results
show that FACT outperforms other well-known routing proto-
cols like VADD. They also validate the effectiveness (in terms
of communication cost) and scalability of FACT. FACT gives
network designers a full package, which delivers trustworthy
messages through a safe path with high reliability and in a
short amount of time. It is flexible enough meaning network
admins can tune the parameters based on the network condition.
Designers can incorporate their desired scheduling and routing
schemes into FACT and still disseminate the messages safely.
FACT is a framework that supports different applications with
different requirements.
REFERENCES
[1] K. Rostamzadeh and S. Gopalakrishnan, “Analysis of message dissemi-
nation in vehicular networks,” IEEE Trans. Veh. Technol., vol. 62, no. 8,
pp. 3974–3982, Oct. 2013.
[2] A. Vahdat and D. Becker, “Epidemic routing for partially connected ad
hoc networks,” Duke Univ., Durham, NC, USA, Tech. Rep. CS-200006,
2000.
[3] J. Zhao and G. Cao, “VADD: Vehicle-assisted data delivery in vehicular
ad hoc networks,” IEEE Trans. Veh. Technol., vol. 57, no. 3, pp. 1910–
1922, May 2008.
[4] A. Skordylis and N. Trigoni, “Delay-bounded routing in vehicular ad-hoc
networks,” in Proc. MobiHoc, 2008, pp. 341–350.
[5] J.-H. Cho, A. Swami, and R. Chen, “A survey on trust management for
mobile ad hoc networks,” IEEE Commun. Surveys Tuts., vol. 13, no. 4,
pp. 562–583, 2011.
[6] D. Huang, X. Hong, and M. Gerla, “Situation-aware trust architecture for
vehicular networks,” IEEE Commun. Mag., vol. 48, no. 11, pp. 128–135,
Nov. 2010.
[7] M. Raya, P. Papadimitratos, V. Gligor, and J.-P. Hubaux, “On data-centric
trust establishment in ephemeral ad hoc networks,” in Proc. IEEE 27th
Conf. Comput. Commun. (INFOCOM’08), 2008, pp. 1238–1246.
[8] S. D. Kamvar, M. T. Schlosser, and H. Garcia-Molina, “The eigentrust
algorithm for reputation management in p2p networks,” in Proc. 12th Int.
Conf. World Wide Web, 2003, pp. 640–651.
[9] A. Koster et al., “Using trust and possibilistic reasoning to deal with
untrustworthy communication in VANETs,” in Proc. 16th Int. IEEE
Annu. Conf. Intell. Transp. Syst., The Hague, The Netherlands, Oct. 6–9,
2013, pp. 2355–2360.
[10] H. Zhu, S. Du, Z. Gao, M. Dong, and Z. Cao, “A probabilistic misbehav-
ior detection scheme toward efficient trust establishment in delay-tolerant
networks,” IEEE Trans. Parallel Distrib. Syst., vol. 25, no. 1, pp. 22–32,
Jan. 2014.
[11] X. Li, Z. Jia, P. Zhang, R. Zhang, and H. Wang, “Trust-based on-demand
multipath routing in mobile ad hoc networks,” IET Inf. Secur., vol. 4,
no. 4, pp. 212–232, Dec. 2010.
[12] S. Sarkar and R. Datta, “A trust based protocol for energy-efficient
routing in self-organized manets,” in Proc. Annu. IEEE India Conf.
(INDICON), 2012, pp. 1084–1089.
[13] T. Gazdar, A. Rachedi, A. Benslimane, and A. Belghith, “A distributed
advanced analytical trust model for VANETs,” in Proc. Global Commun.
Conf. (GLOBECOM), 2012, pp. 201–206.

[14] F. Gómez Mármol and G. Martínez Pérez, “Trip, a trust and reputation
infrastructure-based proposal for vehicular ad hoc networks,” J. Netw.
Comput. Appl., vol. 35, no. 3, pp. 934–941, 2012.
[15] F. Dotzer, L. Fischer, and P. Magiera, “VARS: A vehicle ad-hoc network
reputation system,” in Proc. 6th IEEE Int. Symp. World Wireless Mobile
Multimedia Netw. (WoWMoM’05), 2005, pp. 454–456.
[16] D. Tian, Y. Wang, H. Liu, and X. Zhang, “A trusted multi-hop broadcast-
ing protocol for vehicular ad hoc networks,” in Proc. Int. Conf. Connect.
Veh. Expo. (ICCVE), 2012, pp. 18–22.
[17] J. Wang, Y. Liu, X. Liu, and J. Zhang, “A trust propagation scheme in
VANETs,” in Proc. IEEE Intell. Veh. Symp., 2009, pp. 1067–1071.
[18] D. Huang, S. A. Williams, and S. Shere, “Cheater detection in vehicular
networks,” in Proc. IEEE 11th Int. Conf. Trust Secur. Privacy Comput.
Commun. (TrustCom), 2012, pp. 193–200.
[19] Y.-C. Wei and Y.-M. Chen, “An efﬁcient trust management system for
balancing the safety and location privacy in VANETs,” in Proc. IEEE
11th Int. Conf. Trust Secur. Privacy Comput. Commun. (TrustCom), 2012,
pp. 393–400.
[20] M. H. Eiza and Q. Ni, “A reliability-based routing scheme for vehic-
ular ad hoc networks (VANETs) on highways,” in Proc. IEEE 11th
Int. Conf. Trust Secur. Privacy Comput. Commun. (TrustCom), 2012,
pp. 1578–1585.
[21] M. Teler and V. Cristea, “Securing vehicular networks using deterministic
schemes for computing trust,” in Proc. IEEE 4th Int. Conf. Intell. Netw.
Collab. Syst. (INCoS), 2012, pp. 214–221.
[22] M.-C. Chuang and J.-F. Lee, “TEAM: Trust-extended authentication
mechanism for vehicular ad hoc networks,” IEEE J. Syst. vol. 8, no. 3,
pp. 749–758, Jan. 2013.
[23] T. Gazdar, A. Rachedi, A. Benslimane, and A. Belghith, “A distributed
advanced analytical trust model for VANETs,” in Proc. IEEE Global
Commun. Conf. (GLOBECOM), 2012, pp. 201–206.
[24] N. Haddadou and A. Rachedi, “DTM2: Adapting job market signaling
for distributed trust management in vehicular ad hoc networks,” in Proc.
IEEE Int. Conf. Commun. (ICC), 2013, pp. 1827–1832.
[25] C. Liao, J. Chang, I. Lee, and K. K. Venkatasubramanian, “A trust model
for vehicular network-based incident reports,” in Proc. IEEE 5th Int.
Symp. Wireless Veh. Commun. (WiVeC), 2013, pp. 1–5.
[26] S. Buchegger and J.-Y. Le Boudec, “Performance analysis of the
CONFIDANT protocol,” in Proc. 3rd ACM Int. Symp. Mobile Ad Hoc
Netw. Comput., 2002, pp. 226–236.
[27] A. Das and M. M. Islam, “Securedtrust: A dynamic trust computation
model for secured communication in multiagent systems,” IEEE Trans.
Dependab. Secure Comput., vol. 9, no. 2, pp. 261–274, Mar./Apr. 2012.
[28] D. Dolev and A. Yao, “On the security of public key protocols,” IEEE
Trans. Inf. Theory, vol. 29, no. 2, pp. 198–208, Mar. 1983.
[29] N. Wisitpongphan, F. Bai, P. Mudalige, V. Sadekar, and O. Tonguz,
“Routing in sparse vehicular ad hoc wireless networks,” IEEE J. Sel.
Areas Commun., vol. 25, no. 8, pp. 1538–1556, Oct. 2007.
Karim Rostamzadeh (GSM’09) received the B.Sc.
and M.Sc. degrees in electrical engineering from the
Isfahan University of Technology, Khomeynishahr,
Iran, in 2006 and 2008, respectively, and is currently
working toward the Ph.D. degree at the University of
British Columbia, Vancouver, BC, Canada.
In 2009, he joined the Department of Electrical
and Computer Engineering, University of British
Columbia. His research interests include schedul-
ing and routing in wireless communication networks,
providing services to traveling vehicles under intelli-
gent transportation systems (ITS) umbrella, and modeling delay and reliability
in information dissemination in mobile networks.
Hasen Nicanfar (S’11) received the BA.Sc. degree
in electrical engineering from the Sharif University of
Technology, Tehran, Iran, in 1993, the MA.Sc. degree
in computer networks from Ryerson University,
Toronto, ON, Canada, in 2011, and is currently
working toward the Ph.D. degree in electrical and
computer engineering at the University of British
Columbia, Vancouver, BC, Canada.
From 1993 to 2010, he worked in different pro-
fessional capacities such as IT/ERP Manager, Project
Manager, and Business and System Analyst. His
research interests include trust, security and privacy in wireless communication,
computer network, and cloud computing.
Narjes Torabi (GSM’10) received the B.Sc. and
M.Sc. degrees in computer engineering from the
Isfahan University of Technology, Isfahan, Iran, in
2006 and 2008, respectively, and is currently working
toward the Ph.D. degree in electrical and computer
engineering at the University of British Columbia,
Vancouver, BC, Canada.
In 2009, she joined the Department of Electrical
and Computer Engineering, University of British
Columbia. Her research interests include the design
of coexistence management, medium access control
and scheduling schemes, and realization and implementation of public m-health
service.
Sathish Gopalakrishnan (S’02–M’06) received the
M.S. degree in applied mathematics and Ph.D. degree
in computer science from the University of Illinois
at Urbana-Champaign, Champaign, IL, USA, in 2004
and 2005, respectively.
In 2007, he joined the University of British
Columbia (UBC), Vancouver, BC, Canada, where he
is currently an Associate Professor of electrical and
computer engineering. His research interests include
center around resource allocation problems in several
contexts including real-time, embedded systems, and
wireless networks.
Dr. Gopalakrishnan was the recipient of awards for his work from the IEEE
Industrial Electronics Society (Best Paper in the IEEE TRANSACTIONS ON
INDUSTRIAL INFORMATICS in 2008) and at the IEEE Real-Time Systems
Symposium (in 2004).
Victor C. M. Leung (S’75–F’03) received the
B.A.Sc. (Hons.) and Ph.D. degrees in electrical engi-
neering from the University of British Columbia
(UBC), Vancouver, BC, Canada, in 1977 and 1981,
respectively.
He attended the Graduate School, UBC, on a
Natural Sciences and Engineering Research Council
Postgraduate Scholarship. From 1981 to 1987, he was
a Senior Member of Technical Staff and Satellite
System Specialist with MPR Teltech Ltd., Vancouver,
BC, Canada. In 1988, he was a Lecturer with the
Department of Electronics, Chinese University of Hong Kong, Shatin, Hong
Kong. He returned to UBC as a Faculty Member in 1989, and is cur-
rently a Professor and the TELUS Mobility Research Chair of Advanced
Telecommunications Engineering with the Department of Electrical and
Computer Engineering. He has coauthored more than 750 technical papers
in international journals and conference proceedings, 29 book chapters, and
has coedited 8 books. His research interests include the wireless networks and
mobile systems.
Dr. Leung is a Registered Professional Engineer in the Province of British
Columbia, Canada. He is with the Royal Society of Canada, the Engineering
Institute of Canada, and the Canadian Academy of Engineering. He was
a Distinguished Lecturer of the IEEE Communications Society. He is a
Member of the Editorial Board of the IEEE WIRELESS COMMUNICATIONS
LETTERS, Computer Communications, and several other journals, and has
previously served on the Editorial Board of the IEEE JOURNAL ON
SELECTED AREAS IN COMMUNICATIONS, the IEEE TRANSACTIONS ON
WIRELESS COMMUNICATIONS, the IEEE TRANSACTIONS ON VEHICULAR
TECHNOLOGY, the IEEE TRANSACTIONS ON COMPUTERS, and the Journal
of Communications and Networks. He has guest-edited many journal special
issues, and contributed to the Organizing Committees and Technical Program
Committees of numerous conferences and workshops. He was a recipient of the
IEEE Vancouver Section Centennial Award and the 2012 UBC Killam Research
Prize. He was the recipient of the APEBC Gold Medal as the head of the grad-
uating class in the Faculty of Applied Science, UBC. Several of his papers have
been selected for Best Paper Awards.

IEEE NS2 PROJECT@ DREAMWEB TECHNO SOLUTION

More Related Content

What's hot (19)

Viewers also liked (14)

Similar to IEEE NS2 PROJECT@ DREAMWEB TECHNO SOLUTION (20)

Recently uploaded (20)

IEEE NS2 PROJECT@ DREAMWEB TECHNO SOLUTION