Data-centric Rogue Node Detection in VANETs

Full text

Data Centric Rogue Node Detection in VANETs
Kamran Zaidi, Milos Milojevic, Veselin Rakocevic, Muttukrishnan Rajarajan
School of Engineering and Mathematical Sciences, City University London, London, EC1V 0HB, UK
(Kamran.Zaidi.1, Milos.Milojevic.1, Veselin.Rakocevic.1, R.Muttukrishnan)@city.ac.uk
Abstract—Vehicular ad hoc networks (VANETs) are the future
of vehicular technology and Traffic Information Systems. In
VANETs vehicles communicate by different types of beacon
messages to inform each other of their position and speed to
give them a sense of traffic around them. Vehicles can also
send emergency messages in case of accidents or other hazards.
The very fast moving nodes have to act quickly based on these
emergency messages. However, a rogue node which sends false
emergency messages can wreak havoc in the network that may
even result in fatalities. This paper develops and simulates a
technique to detect a rogue node that is sending false emergency
messages in VANETs by cooperative exchange of data without the
need of any infrastructure or revocation list. Also, the proposed
mechanism will make VANETs fault tolerant and resilient against
injection of false data.
Index Terms -Security, wireless networks, ad-hoc, cryp-
tography, fault tolerance, VANETs.
I. INTRODUCTION
VANETs are considered important due to their huge po-
tential and numerous applications. VANETs not only offer
immense safety enhancements but also many commercial
opportunities. The recent announcements by major car manu-
facturers to equip their vehicles with wireless access vehicular
environment (WAVE) devices from 2014 shows their imminent
deployment. WAVE protocols are based on IEEE 802.11p
standard and provide the basic radio standard for Dedicated
Short Range Communication (DSRC) in VANETs. Vehicles
use DSRC to communicate with each other i.e. vehicle to
vehicle (V2V) and with the infrastructure i.e. vehicle to
infrastructure (V2I) communication.
VANETs are just above the horizon and have the potential
to make road travel as safer and enjoyable as air travel. The
highways can be made much safer by integrating sensors
in vehicles and allowing the vehicles to communicate with
each other in order to have a better understanding of their
surroundings and of the road up ahead. This is exactly what
VANETs aim to do, however, there are serious challenges
in implementing VANETs. The vehicles exchange messages
with each other periodically called beacon messages and can
also send emergency messages. As nodes in VANETs are
vehicles moving at very high speeds, it is imperative that
messages received are correct and give a true picture of the
road conditions.
The existing mechanism for authenticating messages in
VANETs involves the use of cryptography and trust. Crypto-
graphic techniques involve paired keys and overhead in terms
of computing cost, storage and time. Time is of the essence
in VANETs especially in case of emergency messages when
critical decisions have to be taken quickly. Even if emergency
messages are kept unencrypted for faster processing, a false
emergency message can cause severe damage. Emergency
messages include emergency braking, accident, black ice on
road etc. These messages are to be transmitted automatically
to the vehicles behind so that effective safety measures can be
taken.
The emergency messages for cases like accidents or emer-
gency braking are time critical and require immediate action
and therefore, it is recommended to transmit these unen-
crypted. However, a false emergency message can cause severe
problems on the highways and can even result in fatalities.
The condition is exacerbated when an emergency message is
broadcast to be relayed by vehicles to others behind them in
a multi hop fashion to convey the information as far back
as possible. This raises the problem of broadcast storm in an
already bandwidth limited channel when density of vehicles is
high. Also, questions such as how far the emergency message
should travel and when should vehicles stop transmitting it has
been the focus of discussion for many years now. Furthermore,
if messages are being relayed then the messages could be
tampered with and would be impossible to detect.
A lot of research has been done in the past to secure
VANETs by encrypting messages with the help of paired
keys. The vehicles authenticate themselves with the TA and
then RSU and obtain keys or certificates that they can use
within the region of the RSU to exchange messages with other
vehicles. Other vehicles do the same and therefore, whoever
has obtained valid keys / certificates after authentication is
assumed to be a trusted user and its messages are assumed
to be correct as long as the credentials are valid. However,
if a valid user turns rogue or transmits false data due to a
faulty sensor then he can not be stopped and this can result
in serious damage. Therefore, there is a need for developing
security mechanisms for VANETs that are data centric rather
than identity centric.
A. Contributions and Outline
Our main contributions in this paper are:
•The proposed scheme enables vehicles to detect and
correct traffic parameters and highway conditions using
a traffic model - Greenshield’s model.
•The presented scheme, Co-operative Detection And Cor-
rection (C-DAC), enables vehicles to detect rogue nodes
that are falsifying emergency messages in VANETs with-
out the use of revocation list or any infrastructure.
•Make VANETs resilient against false data and enable
them to detect and correct traffic data.
•Prevent Broadcast storm in case of emergency messages.

The rest of this paper is organized as follows: related work
is discussed in Section II. In Section III, the system model
for the proposed protocol is presented. Section IV, gives the
overview of the proposed scheme. Section V analyses security
performance of the proposed scheme in detail. The results are
discussed in Section VI and the conclusion and future work
is given in Section VII.
II. RELATED WORK
Security in Vehicular ad-hoc networks has been the
focus of attention for researchers for many years now. It is
important to secure VANET communications because the user
is exchanging his location and a pseudo identity in all his
messages. Without securing the messages, an adversary would
be able to track a user by listening to their messages. In
order to prevent eavesdropping, cryptography is used by first
authenticating the user and then issuing keys / certificates. In
VANETs, authentication and non-repudiation is achieved by
digital signatures as described in [2], [3], [4]. Non-repudiation
means that a sender can’t deny sending a message and can,
therefore, be held accountable for it. Many different schemes
have been proposed including Public Key Infrastructure (PKI)
[6], [7], [8] and elliptic curve cryptographic system (ECC)
based PKI [5].
In order to preserve the location privacy of a user, the
pseudo identity or public / private key pairs are changed
frequently by each user. However, the Pseudonym (PN) or
public and private key pairs can only be used once which
means the OBU has to store them in large numbers. This
raises the question of how to replenish them in the OBU
once they have been used up. Furthermore, revocation is a
difficult issue when using PNs and public / private key pairs
e.g. if a vehicle is revoked then all the PNs or public and
private key pairs assigned to that vehicle have to be revoked
[3], [4] and added to a Revocation List (RL). Therefore,
if a single vehicle is revoked then there might be several
thousand entries added to the list [13]. This growing RL can
cause serious problems at the RSU when verifying hundreds
or thousands of messages every 300 ms (as dictated by VSC
[10]). Moreover, vehicular networks are fast moving and
highly dynamic networks in which decisions have to be taken
very quickly and RLs do not conform with these requirements
at all.
Privacy can also be achieved by using group signatures
[9]. In group signatures only one member of a group
communicates at a time on behalf of the group. A message
from a group member cannot be associated or linked to any
member of the group thereby preserving the identity of all
the users in the group. However, if a node within a group
turns rogue then it is very difficult to evict the node.
Some researchers have also proposed trust and reputation
based schemes as a solution to securing the VANETs. This
trust based on reputation can either be infrastructure based
or self organising as proposed in [17]. Self organising trust
is based on forming an opinion about a user based on
past interactions or the length of current interaction with
that user or getting feedback from other users about a new
user and assigning a trust score to them. Reputation based
schemes have been proposed in [18], [19], [20]. In [19], [20]
a decentralized infrastructure has been adopted whereas in
[18] a centralized infrastructure is used. In [18] a reputation
score is assigned to each vehicle based on its reliability in
the past. This score is collected, updated and certified by a
reputation server. The score evolves with time i.e. increases
with positive feedback and decreases with negative feedback.
However, interactions in VANETs are short lived and there
can be millions of nodes / users in the networks. Therefore,
centralized trust management is neither easy nor practical.
Similarly, self organising trust is also very difficult as a node
can be honest till a certain time and then go rogue.
Due to the highly volatile and ad-hoc nature of VANETs,
these cryptographic algorithms have to be designed to be
a trade-off between security and performance. Moreover,
malicious behaviour e.g. injection of false data is still possible
even in case of strong cryptography. Researchers in [12]
suggest using data centric techniques to make information in
VANETs more reliable by data centric trust establishment.
Some data centric misbehaviour detection techniques have
been proposed in [14], [15]. In [14] the authors propose a
model of VANETs to be used to detect and correct errors
in the data being sent out by vehicles. The messages that
conform to the model is accepted and rejected otherwise.
However, the authors do not specify the model in detail but
only the events. In the proposed scheme a VANET model is
defined and implemented against which messages are judged
for correctness. In [15] emergency messages are relayed and
false information is identified based on the kind of message
and the subsequent behaviour of the sending vehicle. Such
a technique will not be feasible for emergency messages
which need to be acted on quickly. Also, such a scheme will
increase the computation cost for the nodes.
A misbehaviour detection system and eviction mechanism
is proposed in [16] where nodes are termed misbehaving if
their info is inconsistent with the situation. Once a node is
classified as misbehaving node then the neighbouring nodes
can temporarily evict them by sharing warning messages
about them and later their credentials are passed on to the
CA which revokes them by adding them to a Revocation List
(RL). However, RLs are themselves difficult to manage which
is why data centric schemes are more suited to VANETs.
III. PRELIMINARIES
A. Authentication
In VANETs, it is imperative that vehicles can be dis-
tinguished from one another. This implies that all nodes
authenticate themselves with a Certificate Authority (CA). It
is assumed that all vehicles have authenticated themselves
with a certificate authority and obtained a valid certificate and
public/private key pairs. The vehicles use the keys to encrypt
their messages and others can authenticate and decrypt the
messages by using the relevant public keys. It is also assumed
that all vehicles have enough key pairs to last them a long time
and they keep changing these keys in a reasonable time i.e.

Fig. 1: Greenshield’s Fundamental Diagrams (a)Flow vs Density, (b)Speed vs Flow, (c)Speed vs Density
not too quickly to avoid short term linkability [1]. Moreover,
each vehicle has communicated with the other vehicle at least
once before during which the identity (certificate) has been
validated. This means that in case of an emergency message,
the vehicle recognizes the identity of the vehicle. This is a
reasonable assumption to make as vehicles are exchanging
messages every 100ms.
B. VANET Model
Greenshield’s model is considered to be a fairly reasonable
model in traffic engineering for estimating and modelling
traffic when it is uninterrupted (without traffic signals etc).
Greenshield’s model uses standard parameters such as flow
(vehicles per hour) and density (vehicles per km). The model
describes the relationship between speed (v) and density (k) of
vehicles as being negatively correlated with density increasing
with the decrease in speed as shown in Fig 1(c). In the figure
vfis the free flow speed when density is zero i.e. vehicles can
choose to move freely as there are no or very few vehicles
on the road. As the density of vehicles increases the speed
decreases till density reaches the maximum which is referred
to as jam density or kjat which point the speed becomes zero
and vehicles are stuck in a jam. In the figure kmand vmare
the optimal density and speed respectively which allows the
traffic to progress at the optimum rate of flow - qmFig 1 (a),
(b) & (c). The relationship between speed and density is given
as:
v=vf−k
kj
vf(1)
The relationship between speed, density and flow is as
follows:
q=k×v(2)
From (1) & (2) the relationship between speed and density
can be found to be:
q=vfk−k2
kj
vf(3)
C. Message Format
Each vehicle creates its own message mfor beacon and
apart from the usual values also includes the following:
m(Speedown, Densitycalc, F lowOW N )
Each beacon message mis hashed (H(m)) and signed by the
vehicle using its secret key (SK).
sig =SK (H(m))
The details of how this signature is generated and how they
are verified are beyond the scope of this paper. In case of
emergency e.g. an accident or emergency breaking, each ve-
hicle generates an emergency message which has the following
format:
EmergencyMsg(T y pe, SpeedOwn, D ensitycalc, F lowOW N )
It must be noted that the emergency messages are not en-
crypted.
D. Rogue Node Model
A node is termed as rogue if it starts to inject false data in
the network either on purpose with malicious intent or due
to faulty sensors. Moreover, the rogue node can start sending
false data at any time and can falsify values of their own
speed and their calculated values of flow and density either
Fig. 2: Estimating density of vehicles in VANETs

Fig. 3: Overview of Detection And Correction scheme (C-
DAC)
in beacon message or emergency message. However, a rogue
node can’t modify values of other nodes in the network.
In case of a false emergency message the rogue node
will start sending a low value of Flow or sudden decrease
in speed or both to indicate an accident or emergency braking.
IV. SCHEME OVERVIEW
In this paper a scheme C-DAC is proposed by which all
vehicles calculate their own values of flow. Vehicles send their
speed, flow, density and location information to other vehicles
and each vehicle can calculate their own value of flow which
gives them a very good model of the traffic in their vicinity
and up ahead as well. Each vehicle can predict the density of
vehicles on the highway by the number of messages it receives
from other vehicles by checking their IDs from messages. This
enables each vehicle to calculate the density quite accurately in
a moving window around itself as shown in Fig 2. The size of
this density window is equal to the transmission and reception
range of a vehicle (500 meters). This means that a vehicle
can receive messages from a vehicle which is up to 500m
ahead of it and 500m behind it. Therefore, each vehicle has
a communication window of 1000m around it that it can use
to calculate the density (Densitycalc ). Also, each vehicle can
calculate the average speed of vehicles (SpeedAV G) within its
communication window. In our scheme each vehicle transmits
not only its location and speed but the calculated value of
flow as well. Therefore, the vehicles calculate the traffic flow
parameter using density and average speed of other vehicles
Fig. 4: Varying Value of Flow in an Accident Scenario
through Greenshields model. The flow serves as a global
parameter which each vehicle calculates on its own and should
be very similar for vehicles that are close to each other in the
same traffic conditions.
The idea is that in case of an actual emergency situation, a
vehicle will generate a message that has a very small value of
flow that indicates that the flow of vehicles on that stretch
of road has suddenly reduced. This will be confirmed by
other vehicles as well which calculate a similar small value
of flow on their own and generate messages. However, if a
node generates a false message indicating a small value of
flow either with malicious intent or due to some fault then it
would be the only vehicle that generates such a value and
can be singled out. The vehicle’s speed has been used by
some researchers [11] to estimate density but it does not give
good results as the assumption is that given the opportunity
the vehicle will try to achieve the maximum speed possible
which is not true in real life.
Each vehicle transmits its F lowOW N which becomes
F lowRcvd for other vehicles. If a vehicle receives a value of
Flow from another vehicle that does not agree with the VANET
model then the data is rejected and vehicles’ ID is noted and
reported. If the data agrees with the model then the receiving
node checks the data with its own calculated values to confirm
if values are indeed correct (shown in Fig. 3). If the values
do not agree with the node’s own calculated parameters of
Flow, Speed and density then the values are discarded and the
sender ID is reported. The two values of flow are calculated
as follows:
F lowOW N =SpeedAV G ×Densitycalc (4)
F lowAV G =
n
XF lowRcvd
n(5)
TABLE I: SIMULATION PARAMETERS
PARAMETER VALUE
Simulation Time 500 sec
Scenario 3 Lane Highway
Highway Length 5-Kms
Max Vehicle Speed 28 m/sec or 100 Km/hr
Mobility Tool VACaMobil
Network Simulation Package OMNET++
Vehicular Traffic Generation Tool SUMO
Number of Vehicles 330
Vehicle Density 20-30 veh / Km
Wireless Protocol 802.11p
Transmission Range 500m in each direction

(a) Increasing density in case of real accident
(b) Decreasing speed in case of real accident
(c) Decreasing flow in case of real accident
Fig. 5: Real Accident Scenario
However, in case of an actual accident the low value should
be reported by all vehicles and it should propagate throughout
the highway efficiently and gradually as shown in Fig 4.
Moreover, in case of actual accident the speed of the vehicle
that is receiving the messages will come down as well as it can
detect obstacles with the help of on-board radar etc. Therefore,
the main assumption is that the vehicle will be able to trust
(a) Constant density in case of rogue nodes
(b) Constant speed in case of rogue nodes
(c) 2 rogue nodes reporting lower values of flow
Fig. 6: Rogue nodes scenario without accident
its own calculated values even if it can’t trust anyone else.
As, vehicle’s own speed comes down then from eq. (4) the
F lowOW N should come down as well which is then sent to
other vehicles. If the received data doesn’t conform to the
VANET model, own calculated values or both then the data
will be discarded and the node will be reported. The flow chart
for our scheme C-DAC is shown in Fig. 3.

Fig. 7: Decision for Data Correctness
V. PERFORMANCE EVALUATION
A. Simulation Setup
In order to check the proposed model it is simulated
using OMNET++, SUMO [22] and VACaMobil [21]. OMNET
is a modular C++ library and framework that is used for
network simulations. Simulation of Urban Mobility (SUMO)
is a software tool used to generate vehicular traffic by speci-
fying speed, types, behaviour of vehicles and road types and
conditions. VACaMobil is a car mobility manager for OMNET
that works in parallel with SUMO.
The scenario is simulated with parameters shown in Table I.
In order to validate the model an accident is simulated which
takes place at t=180 sec and the results are recorded. Nodes
0, 1, 2, 3, 4 suffer an accident and block all three lanes of
the highway. The result for DensityAV G,SpeedAV G and
F lowAV G are shown in Fig. 5 a), b) and c) respectively.
Another scenario is simulated when there are three rogue
nodes which start sending low false values of F lowOW N from
t=180 sec incorrectly indicating an accident up ahead. The
results for this scenario (where every 10th node is a rogue
node) for Densitycalc and SpeedAV G are shown in Fig. 6
a) and b) respectively whereas the , F lowAV G values for 6
vehicles (out of which 2 are rogue) are shown in Fig 6 c).
B. Simulation Results
1) Actual Accident Scenario: The results for the actual
accident scenario are shown in Fig. 5 a), b) and c). It can
be seen from Fig. 5(a) that the accident is causing the number
of vehicles (density) to build up after the accident and all
vehicles are reporting the same. Similarly, the flow value that
each vehicle is computing is decreasing immediately after the
accident (Fig 5c). Also, as the vehicles come to a stop their
speeds decrease quite abruptly(Fig. 5b). This result gives a
true - real VANET model against which received values are
compared in case of rogue nodes.
2) No Accident - Rogue Node Scenario: In Fig. 6 a),
b) and c) every 10th vehicle is a rogue node which are
travelling normally without any accident and the rogue nodes
are transmitting a low false value of Flow whereas the others
are transmitting a (true) high value. In this case, the rogue
Fig. 8: Percentage of vehicles within the distance 3000m that
received the emergency information successfully
nodes are not modifying the values of density or speed and
can easily be seen and classified as faulty or rogue values.
C. Data Centric Rogue Node Detection
The honest nodes can decide whether a value being shared is
correct or not by using a decision table shown in Fig. 7 which
is directly derived from the Greenshield’s model as shown in
Fig. 1 a). As in the case of our simulation if the value of
Flow being reported by a node is decreasing but the speed
and density reported from that node remain constant then the
value being reported is false. Similarly, another case could be
when a node reports a decreasing value of flow and increasing
value of density but the average speed remains constant in that
region then again this implies that the data being reported is
false and can be discarded.
The results show that by using our technique, messages can
be authenticated based on the relevance and freshness of data
without authenticating the identity of nodes. Such nodes can
then be reported or their messages be simply discarded. Also,
the information about an accident can be propagated down
the highway gradually and gracefully so that the traffic keeps
flowing as long as it can and comes to stop gradually.
D. Comparison
The success rate of the proposed scheme C-DAC is com-
pared to the AMBA (Adaptive and mobility based algorithm)
presented in [11]. The success rate is the percentage of vehicles
within a 3km distance that receive the emergency information
successfully and is shown in Fig. 8. In our scheme C-DAC, the
success rate reaches 100 % as there is no congestion because
the emergency messages are not being relayed as in AMBA.
Instead, in C-DAC the emergency info is being propagated
through communication of some global traffic parameters as
discussed previously and information can be relayed to all
nodes even very large distances away.
VI. DISCUSSION
In this section we discuss the direct and indirect effects
of our proposed system on the network, its reliability and
robustness.

Fig. 9: Average Flow in case of No-Accident with Rogue
Nodes
A. Fault Tolerance
Moreover, as the data is being calculated by each vehicle
and it is being compared with the readings calculated by the
vehicle itself and other vehicles, therefore, this introduces
a built-in fault tolerance in the network which is highly
useful and desirable for highly volatile and rapidly changing
VANETs. Even if a node is able to distort the values of the
reported parameters (Density, Flow and Speed) so as not to
raise a red flag with other vehicles, it results in a small error
in the overall reading as shown in Fig. 9, it shows values of
F lowown in case of no accident and two rogue nodes that
start transmitting a false value of Flow at t=180 sec and the
average value of flow shown in blue line. This value shows
that even if the false flow values are not rejected initially they
will cause little deviation if the number of rogue nodes is small
as compared to honest nodes in the neighbourhood.
B. Self Detection and Correction
In a vehicular ad-hoc network with fast moving nodes, it is
highly desirable for the nodes to be able to detect and correct
data on their own. Due to the volatile nature of VANETs it
is impractical to use any techniques that rely on reputation or
trust of users to ensure correctness of information. Moreover, a
valid identity of vehicles is important for distinguishing them
from each other but should not be used as the basis of the
acceptance of information in a protocol. This means that an
authenticated node doesn’t guarantee that the node will behave
honestly. With the latest technology being introduced in the
vehicles including radars and cameras for obstacle detection,
these technologies can be combined with a technique like ours
to ensure safety of travel. With driver-less features becoming
a reality with Google car, it is important that the vehicle
starts behaving autonomously not only in terms of driving
but also planning ahead. This means that at high speeds on
the highways, a driver-less car should be able to estimate or
predict the road and traffic conditions quite early and with
reasonable accuracy. This is only possible if the highway traffic
is modelled and used by the OBUs to detect and correct
anomalies in the information being received. The notion of
revocation quickly in a highly agile and temporary network
doesn’t look realistic till now.
C. Congestion Avoidance
In case of emergency messages in VANETs, the currently
proposed method of propagating such messages is by relaying
the message by receiving vehicles to others behind them. This
can cause a broadcast storm where every vehicle is relaying the
same message repeatedly and flooding the region in an attempt
to inform other vehicles of the emergency. This quickly,
consumes the small bandwidth available and can choke the
network. However, in our proposed scheme there is no channel
congestion as there is no need for multi-hop retransmissions
and a sudden drop in the flow or speed values can indicate an
emergency. Moreover, as only the vehicles within range behind
the vehicle experiencing the accident receive the emergency
message, they are able to identify that vehicle quickly as they
have communicated with it before. These vehicles then modify
their own values of flow and send them to others.
D. Resilience to Sybil Attacks
In case of a Sybil attack, an attacker presents multiple
identities with an intent to either vote out a user maliciously
or in our case more likely to create the illusion that there is
congestion or accidents up ahead. As all vehicles are reporting
their location along with their speed, density and flow values in
their vicinity. In case of a sybil attack, an honest vehicle which
is behind a sybil node will receive multiple (false) messages
with different identities and each message will report a low
value of flow but if the vehicle’s own speed is not decreasing
then it can start ignoring those messages. Therefore, C-DAC
provides resilience against sybil attacks.
E. Effective Dissemination of Information
In our scheme the information about an accident or similar
situation is disseminated effectively to other vehicles long way
back gradually and efficiently. This means that in case of
traffic congestion, the system will inform users to slow down
gradually and not tell them to come to a halt altogether. This
case was predicted in Fig. 4 and can be seen in Fig. 5 c) as
well. In Fig. 5 c) the value of the Flow is coming down in the
whole network gradually and gracefully validating our model.
F. Limitations of C-DAC
The presented technique, C-DAC will be unable to identify
the actual rogue node during a sybil attack when the
rogue node is presenting multiple identities / pseudonyms.
Moreover, if there are multiple attackers that collude to
launch an attack e.g. two or three cars intentionally block
the lanes of the highway and send multiple messages with
different identities reporting an accident or congestion then it
will not be detected by C-DAC. The reason is that in such
an attack the data will satisfy both the VANET model and
will also match the vehicle’s own readings. However, such

an attack is very expensive to launch as it requires multiple
rogue nodes to be present together in a region.
In case of an actual accident the density increases all
of a sudden whereas the flow value doesn’t go down as
quickly, therefore, the flow value that a vehicle will calculate
will increase for a short time before going down. During
this transition phase, it is difficult to distinguish between a
rogue node and an honest node apart from the vehicle’s own
calculations.
VII. CONCLUSION AND FUTURE WORK
In this paper we have shown that data centric rogue node
detection and data correction is possible using our proposed
scheme C-DAC without the need for any revocation list. By
using a model of the network, each node is able to predict and
estimate the current state of the network. The strength of the
proposed scheme is that an honest node can always trust its
own data regardless of what values are being reported by other
nodes. By calculating and sharing a few global parameters
each vehicle can get a reasonably accurate model of the traffic
conditions which can be used to single out the rogue node
whose data can then be discarded. Such data centric schemes
are highly suited to VANETs where trust based schemes are
impractical to implement due to very large number of nodes
and ephemeral nature of network. However, the proposed
scheme will not be able to detect false data or identify sybil
nodes if the data they are sending conforms to the model and
also to the values being calculated by the receiving vehicle
itself.
Our proposed scheme can be extended if RSUs are used to
provide confirmation of one or more global parameters. One
way in which this can be done is by calculating the flow of
vehicles along a segment of highways and then reporting this
value to all vehicles in the region to correct the data that they
are receiving from other vehicles and fine tune their estimated
traffic model. This value of flow can be calculated quite easily
by using existing traffic cameras used for traffic monitoring by
authorities and applying simple image processing techniques.
This will further improve reliability and robustness of the
system even in case of Sybil attacks.
REFERENCES
[1] A. Studer, E. Shi, F. Bai, and A. Perrig. “TACKing Together Efficient
Authentication, Revocation, and Privacy in VANETs,” Proc. 6th Annual
IEEE Commun. Society Conf. (SECON ’09)., pp. 1–9, 2009.
[2] D. Huang, S. Misra, G. Xue, and M. Verma, “PACP: An Efficient
Pseudonymous Authentication Based Conditional Privacy Protocol for
VANETS, IEEE Trans. Intell. Transp. Syst., vol. 12, no. 3, pp. 736-746,
Sep. 2011.
[3] A.-N. Shen, S. Guo, D. Zeng, and G. Mohsen. “A Lightweight Privacy-
Preserving Protocol using Chameleon Hashing for Secure Vehicular Com-
munications,” in Proc. IEEE Wireless Commun. Netw. Conf. (WCNC), pp.
2543–2548, 2012.
[4] R. Lu, X. Li, T. H. Luan, X. Liang, and X. Shen. “Pseudonym Changing
at Social Spots: An Effective Strategy for Location Privacy in VANETs,”
In IEEE Trans. Veh. Technol., vol. 61, no. 1, pp. 86–96, 2012.
[5] I. Blake, G. Seroussi, and N. Smart, “Advances in Elliptic Curve Cryptog-
raphy,” Number 317 in London Mathematical Society Lecture Note Series.
Cambridge Univ. Press, Cambridge, U.K. 2005.
[6] M. Raya, and J-P. Hubaux, “Securing Vehicular Adhoc Networks,”
Journal of Computer Security, Special Issue on Security of Ad Hoc and
Sensor Networks, vol. 15, no. 1, pp. 39–68, 2007.
[7] P. Papadimitratos, A. Kung, J. P. Hubaux, and F. Kargl, “Privacy and
Identity Management for Vehicular Communication Systems: A Position
Paper,” in Workshop on Standards for Privacy in User-Centric Identity
Management, Zurich, Switzerland, July 2006.
[8] R. Lu, X. Lin, H. Zhu, P. H. Ho, and X. Shen, “ECPP: Efficient
Conditional Privacy Preservation Protocol for Secure Vehicular Communi-
cations”, in Proc. IEEE 27th Conf. Comp. Commun. pp. 1229–1237, Apr.
2008.
[9] X. Lin, X. Sun, P.-H. Ho, and X. Shen, “GSIS: A Secure and Privacy
Preserving Protocol for Vehicular Communication,” in IEEE Trans. Veh.
Technol., vol. 56, no. 6, pp. 3442-3456, Nov. 2007.
[10] Vehicle Safety Communications Project Final Report: Identify Intelligent
Vehicle Safety Applications Enabled by DSRC. s.l. :U.S. Department of
Transportation, National Highway Traffic Safety Administration, 2005.
[11] K.A. Hafeez, L. Zhao, B. Ma, J.W. Mark, “Performance Analysis and
Enhancement of the DSRC for VANET’s Safety Applications,” IEEE
Trans. Veh. Technol., vol.62, no.7, pp.3069-3083, Sept. 2013.
[12] F. Kargl, P. Papadimitratos, L. Buttyan, M. Muter, E. Schoch, B. Wieder-
sheim, Ta-Vinh Thong, G. Calandriello, A. Held, A. Kung, J-P Hubaux,
“Secure vehicular communication systems: implementation, performance,
and research challenges,” IEEE Commun. Mag., vol.46, no.11, pp.110-118,
November 2008
[13] J. J. Haas, Y-C. Hu, and K. P. Laberteaux. “Efficient Certificate Revoca-
tion List Organization and Distribution, IEEE J. Sel. Areas Commun. vol.
29, no. 3, pp. 595-604, 2011.
[14] P. Golle, D. Greene, J. Staddon. “Detecting and correcting malicious
data in VANETs,” in Proc. 1st ACM Int. workshop Veh. ad hoc networks,
pp. 29-37. ACM, 2004.
[15] S. Ruj, M. A. Cavenaghi, Z. Huang, A. Nayak, I. Stojmenovic. “On
data-centric misbehavior detection in VANETs,” in Veh. Technol. Conf.
(VTC Fall), 2011 IEEE, pp. 1-5. IEEE, 2011.
[16] M. Raya, P. Papadimitratos, I. Aad, D. Jungels, J.-P. Hubaux, “Eviction
of misbehaving and faulty nodes in vehicular networks,” IEEE J. Sel. Areas
Commun, vol. 25, no. 8, pp. 1557-1568, Oct. 2007.
[17] P. Wex, J. Breuer, A. Held, T. Leinmuller, L. Delgrossi. “Trust Issues
for Vehicular Ad Hoc Networks,” Veh. Technol. Conf. (VTC Spring 2008),
pp.2800-2804, 11-14 May 2008.
[18] L. Qin, A. Malip, K. M. Martin, S. Ng, and J. Zhang. “A reputation-
based announcement scheme for VANETs,” IEEE Trans. Veh. Technol.,vol
61,pp 4095-4108,Nov. 2012.
[19] U. Minhas, J. Zhang, T. Tran, and R. Cohen, “Towards expanded trust
management for agents in vehicular ad hoc networks, Int. J. Comput. Intell.
Theory Pract., vol. 5, no. 1, pp. 3-15, Jun. 2010.
[20] A. Patwardhan, A. Joshi, T. Finin, and Y. Yesha, “A data intensive
reputation management scheme for vehicular ad hoc networks,” in Proc.
3rd Annual International Conf. Mobile Ubiquitous Systems, pp. 1-8. 2006.
[21] M. Baguena, S. Tornell, A. Torres, C. Calafate, J.-C. Cano, and P.
Manzoni, “Vacamobil: Vanet car mobility manager for omnet++,” in
Communications Workshops (ICC), 2013 IEEE International Conf. on,
2013, pp. 1057-1061, Jun. 2013.
[22] M. Behrisch, L. Bieker, J. Erdmann, and D. Krajzewicz. “Sumo -
Simulation of Urban Mobility: An overview,” in SIMUL 2011, The 3rd
International Conference on Advances in System Simulation, pages 63-68,
Oct. 2011.