Security Breakdown in V2I Networks: A Comprehensive Analysis

This chapter discusses the components that form the basis of a functional V2I (Vehicle-to-Infrastructure) communication network. It highlights the importance of Roadside Units (RSUs), Trusted Third Parties, and vehicles in ensuring secure communication. The proposed architecture involves an aggregated vehicle-to-infrastructure approach, where each grid has a super vehicle that aggregates messages sent to the RSU infrastructure. To maintain security and privacy, a pseudo-identity technique is employed for the super vehicle user, while a lightweight key exchange algorithm is needed for super vehicle-to-RSU communication. The chapter concludes by emphasizing the need for a secured and efficient group-based encryption mechanism to guarantee all-round security, privacy, and performance requirements for V2I communication networks.

Masters Thesis
Abstract
1. Abstract:
2. Introduction: (5 pages) Including problem statement, motivation.
3. Chapter 4: in the end of the paper, including (1 page) future research and (3 pages)
conclusion.
Please, all linked and according to the paper.
2. INTRODUCTIONII. VEHICLE-TO-VEHICLE ADHOC NETWORK
The recent advances in ad hoc networks have allowed multiple implementation of architectures
for vehicular networks [1]. These architectures support different requirements, and satisfy
different constraints. When a notive is introduced to the concept of vehicular networks, all he can
think of is vehicles talking to each other. This concept is known as vehicle-to-vehicle (V2V) ad
hoc network, which will be the focus of this chapter.
V2V network allows the direct communication between vehicles without de- pending on a fixed
infrastructure support [2]. It is simply the wireless transmission of data between motor vehicles.

The principal goal of V2V is to prevent accidents by allowing vehicles share speed data and
position data with one another over an ad hoc mesh network. V2V is expected to present a 360
degree level of awareness about the surrounding to a moving vehicle [3].
V2V is becoming an important part of the intelligent transport system of the United states. Data
from vehicle-to-vehicle communication is currently being used to improve traffic management
evident in traffic lights and signs. It is expected that this technology will become a mandatory
part of the transportation system, and immensely aid in the quest for driverless-cars all across
America. The major incentive of V2V is that the technology can increase the performance of
vehivle stafety systems and help save lives. It is expected that connected vehicle technolo- gies
will provide drivers with the tools to not only survive crashes, but to avoid it altogether [4].
Without the use of a trusted authority, implementing V2V raises questions about the
communication mechanism to ensure effective connectivity, as well as address- ing the inherent
security challenges in this form of communication. Discussion of V2V in this chapter will be in
terms of the connectivity mechanism and security measures. An approach to ensuring effective
and secured communication in V2V will be discussed.
A. Proposed approach
Dedicated for wireless vehicular communications is the IEEE 802.11p standard where 6 out of
the 7 available channels are used for actual communication and the last channel for control
messages [5]. However, with the overall goal of having all communicable cars on US roads,
these channels could be easily overwhelmed. Within a mile radius, there could be several cars
especially on busy roads in highly populous geographical region all expected to provide
transportation intelligence through V2V communication. This introduces a requirement for a
virtually unlimited available communication channel in vehicular communication [6]. The
timeliness constraint of V2V networks must also be factored as time- critical emergency
messages cannot be delayed.
Subsequently in this chapter, an evaluation of the various connectivity options available for V2V
is considered, with an approach considering the communication range, speed, tavel direction etc.
Also, a trust-based security model for V2V communication is proposed.
B. Connectivity
In the context of IoV, the time to takes to fully transit emergency messages can mean the
difference between life and death. However, messages may be delayed due to a limitation in the

channel availability of the IEEE 802.11 channels, which are governed by the IEEE 802.11p
standard for wireless vehicular communications. IEEE 802.11 consists of seven channels in total.
One of these acts as the common control channel while the remaining six are employed for
vehicular communications [6]. As the popularity and prevalence of vehicles that can
communicate with one another wirelessly increases, there is a strong possibility that the IEEE
802.11 channel will become overwhelmed, and this could result in delays in messages and
scarcity in the radio spectrum [7], [8]. As such, cognitive radio technologies could represent a
viable solution by which it is possible to achieve the low latency inter-vehicle communications
require while also preventing spectrum scarcity [9]. Cognitive radio is a relatively new
technology that involves optimizing the spectrum by exploiting unutilized spectrum holes [10].
The cognitive radio network typically employs three different approaches to identify and exploit
unused channels: geo- location database, beacons, and spectrum sensing. Each of these methods
will be examined in more depth below:
Beacons utilize signals by proliferating signals across different channels with the objective of
detecting a free channel [11]. Beacons have been integrated in a range of different applications;
for example, [12], [13] , and [14]. One area in which beacons have been particularly useful
within the context of IoV is through the use of signals to detect the speed, direction, and position
of other vehicles within a network. Beacons are not without their downsides. One major issue
with the use of beacons to identify unexploited channels is that their use reduces the efficiency of
the spectrum because they rely on a high and adequate radio frequency. In addition, the signals
sent and received by beacons may suffer interference from alternative sources [15].
The geo-location database is a digital archive that obtains, processes, compiles, and proliferates
information about the spectrum band [16]. The CR system accesses the geolocation database
scheme, which contains data about the primary user systems, to identify vacant bands. When this
scheme is employed, there is no requirement for the CR users to perform spectrum sensing on
the bands; as such, in comparison to the other methods of identifying unused channels, the
overheads associated with using the geo-location database are relatively low [17]. Furthermore,
as this approach does not involve any sensing errors, it is easier to protect the primary users. To
employ this scheme effectively, there is a requirement to establish a stable connection between
the spectrum database and the CR users. As such, while it is relatively easy to employ this
scheme in a cellular CR system, it is more complex to employ it an ad-hoc situation because a
connection of this nature may not exist [8]. For this reason, it is unlikely that the geolocation
database scheme will find application in an ad-hoc CR environment.
Spectrum sensing is widely considered to represent a fundamental component of a cognitive

radio network [18]. It operates by sensing the various channels that are available. As previously
described, cognitive radio exploits the vacant spectrum as a means of serving a secondary user.
However, this can enhance the interface between those users who are licensed to use the
spectrum (the primary users) and those who are not (the secondary users) and ultimately
diminish the service that is available to the licensed user [19]. To avoid this issue, the use of
spectrum sensing can enrich spectrum utilization without reducing the parity of the service that is
available to licensed users [19]. In addition, secondary users can access the RF spectrum in an
ad-hoc manner to detect unused channels without undermining the quality of the service that is
available to the primary users. As such, the users of unlicensed vehicles need to use spectrum
sensing and evaluation to identify the idle band for opportunistic communications in IoV [20]. It
is worth employing mathematical approaches to examine the joint impacts of licensed and
unlicensed user activity on the performance of the spectrum sensing approach within the context
of opportunistic communications between vehicles that are fitted with wireless communication
and spectrum sensing functionality.
To prevent licensed users from suffering negative interference, secondary vehic- ular users can
employ spectrum sensing to verify whether licensed users are em- ploying their channels [21],
[22]. Within the context of cognitive IoV, vehicles are frequently traveling at high speeds across
various distances; as such, the network structure can rapidly change. Existing studies have been
based on the assumption that secondary users are stationary when they are in the process of
executing spectrum sensing, and PUs are assumed to be passive during ad-hoc transmissions by
secondary users of the network. Within the majority of existing literature user mobility has been
examined in non-cooperative spectrum sensing [23], cooperative spectrum sensing [24],
spectrum sensing in the presence of primary user mobility [25] and spectrum sensing using
random way point model for secondary users where primary users were motionless [26].
It is important to highlight that the contemporary sensing methods that are in use do not take into
consideration the cumulative impact of the vehicle speed, primary users mobility, and secondary
users sensing range while calculating the function of spectrum sensing in cognitive IoV where
the secondary users are traveling at a speed measured in miles per hour. The most significant
aspect of IoV is that the users travel in an identical direction or opposite directions in accordance
with the structure of the road [27]. Identifying the intersection between the protection range of
the primary user and the sensing range of the secondary user and the distance between the two
users (which is directly dependent on their relative speed) plays a critical role in determining
whether a primary user is within the sensing range. Furthermore, the dynamic spectrum access
for ad-hoc communication is also of significance from the perspective of anticipated

transmission time, realistic per-user rate, and the anticipated transmission count.
Dynamic spectrum access (DSA) represents a viable method of solving the spec- trum
inadequacy issues outlined above [6]. This novel area of study forecasts the evolution of CR
networks to enhance spectrum efficiency [28]. The licensed user indisputably has the priority on
the spectrum band and, as such, takes precedence over any unlicensed user that accesses the
spectrum band. DSA, which is also referred to as opportunistic spectrum access involves
constraining the SUs such that they can only opportunistically access the spectrum holes. In the
case of the interweave DSA model, secondary users employ the cognitive radio to detect the
proximate spectrum environment before an idle spectrum band is selected and the CR is switched
to that band to transmit a message [29]. DSA and CR have attracted significant attention from
researchers in recent years due to the potential of these technologies to significantly enhance the
efficiency of spectrum utilization.
THE APPROACH
Every conventional vehicular network consists of a collection of licensed primary spectrum users
and unlicensed (secondary) vehicular users as shown in Fig. 1. In this approach, communication
between vehicles is leveraged on the dynamic spec- trum access to radio frequency channels.
The secondary user is considered to have a sensing range s, and primary users with a protection
radii r. These constraints allows the consideration of the joint impact of secondary user mobility
and primary user activity to avoid harmful interference during the dynamic spectrum access
process.
It is assumed that the primary user protection range is shorter than the sensing range of the
spectrum sensor such that r < s. To determine the relative speed between a primary user and a
secondary user in this approach, it is assumed that the primary user is stationary while the
secondary user is mobile. Thus, this relative speed is determined by the speed of the secondary
user (vehicle). In reality, when a vehicular user is moving towards a primary user, the overlap
time is greater than when the vehicle is moving away from the primary user. Considering that a
vehicle could be moving closer or away from a primary user, the distance between them is
established as a function of the geographical distance between them, direction of travel, sensing
range of vehicular user, primay user’s protection range and the relative speed.
For a secondary user to opportunistically establish communication via a channel, the primary
user activity for a particular location and time must be checked. This activity is known as
spectrum sensing. The activities of a primary user in the channel is represented by ON and OFF
states [30]. The secpndary user detects the activity of the primary user by noise being present in
the channel. The signal received by a mobile secondary user rs(t) can be represented as a

function of two hypotheses as in (1). H0 represents the presence of only noise in the channel, and
H1 indicating the presence of a primary user.
It is important that a misdetection of primary user activity is reduced as it can cause harmful
interference. In this vein, it can be said that reducing the misdetection probability is more
important than reducing false alarm (false positive) as this does not lead to harmful interference
with primary users. An evaluation of the effect of primary user activity and secondary user
mobility defined in terms of probability of misdetection of spectrum opportunities and expected
overlapping time between static primary user and mobile secondary user is conducted. The
following propositions are made based on Definition 1 and Definition 2 to formally analyze
event A and event B.
Proposition 1: For a separation distance D between a secondary user and primary user, the
condition for the primary user being inside the sensing range of secondary user is r < D  S. The
probability for the event B Pr(B)i.e. the probability that the primary user is inside the sensing
range of secondary user and the secondary user detects that the primary user is present in a given
channel, is given as:
Proposition 2: The probability of misdetection that depends on sensing range of vehicular user,
protection range of primary user, velocity of vehicular user and energy detection threshold is
expressed as….
Proposition 3: The expected overlap time duration between static primary user
and mobile secondary user is…
Evaluation
To evaluate the approach discussed so far, simulations are conducted with numerical results
obtained used to substantiate the formal descriptions given. It is important to note that the
secondary user is assumed to be mobile, while the primary user is static in all the simulations. It
is further assumed that access points and base stations are primary users, each with its protective
range. Secondary users are not expected to use channels actively used by primary users in ithe
primary user’s protective range.
To understand the effect of the sensing range on the probability of event B, a plot of the variation
of Pr(B) vs the sensing range for a given primary user with a protective range r = 110 meter is
done. It is oberved that Pr(B) increases with increased sensing range as shown in Fig. 2.

It is important to evaluate the variation of the probability of miss-detection Pr(miss) relative to
the sped of the secondary user. For this, the primary user’s
protection range r is set to 100 meter, the initial separation distance between the primary user and
secondary user D = 200 meter, and secondary user’s sensing ranges from 0 to 1000 in multiples
of 250 meter. The probability of primary user being ON, PU(OFF!ON) is varied from 0.30 to
0.60 to 0.90 to see how primary user’s OFF ! ON activity impacts the performance of mis-
detection. In Fig. 3, it is observed that the probability Pr(miss) decreased when sensing range of
secondary user increased for a given P U(OF F !ON ) value. However, for an increased speed of
mobile secondary user with a given PU(OFF!ON) value, the probability of mis-detection
increased. The significance of this is that at a higher speed, the probability of miss-detection
becomes higher as a result of the primary user being quickly out of the sensing range of the
secondary user. it is also observed that the probability of mis-detection increased by
approximately 10% for each case of primary user being active during sensing period (and idle in
the previous) increasing from 0 to 0.9 in multiples of 0.3, coupled with the speed of the
secondary user exceeding 40km/hr as shown in Fig. 3. The reason for this being that the primary
user is expected to be idle (following being idle in the previous round), but is active resulting in a
higher mis-detection probability for a given sensing range.
To visualize the relationship between the probability of misdetection and the velocity of
secondary user with different values of PU(ON!ON), a plot is derived as shown in Fig. 4. It is
observed that the probability of misdetection increases with an increase in PU(ON!ON) when
varied from 0.3 to 0.6 to 0.9. For PU(ON!ON) greater than PU(OFF!ON), the probability of mis-
detection decreases as evident in Fig. 4. It is also observed that with P U(OF F !ON ) greater than
or equal to PU(ON!ON), mis-detection probability increases as shown in Fig. 3 and Fig. 4.
A plot of the variation of expected overlapping duration per epoch against the speed of the
secondary user as shown in Fig. 5 reveals that for higher sensing range, the primary user has a
higher possibility to fall into the secondary users sensing range for longer overlapping time
duration. This is evident in the visible decrease in the expected overlapping duration per epoch
with increasing speed of the secondary user for a given sensing range, and its increase with
increased sensing range of the mobile secondary user. In this configuration, the protection range
of the primary user is considered such that r = 110 meter. An initial separation distance between
the primary user and secondary user D = 200 meter, and secondary user’s sensing range s = 1000
meter. The secondary users sensing range was varied as s = 250, 500, 750 and 1000 meter.

For the performance evaluation of dynamic spectrum access for opportunistic communication,
the variation of per-user data rate for vehicular users vs. the number of vehicular users is plotted
as shown in Fig. 6. The per-user data rate decreases when number of user in IoV increases
because of sharing of the same channel.
Also, the variation of expected transmission time vs the achievable data rate for given number of
vehicles and probability of transmission failure is plotted as shown in Fig. 8. It is observed that
expected transmission time decreases with an increasing achievable data rate for given number
of users and probability of failure. As shown in Fig. 8, expected transmission time is highest (or
lowest) when more (or less) vehicles are present and highest (or lowest) probability of
transmission failure.
Finally, a plot of the variation of expected transmission count against the proba- bility of
transmission failure is shown in Fig. 7. It is observed that when transmis- sion failure increases,
the expected transmission count increase exponentially and when failure probability pf is 1,
transmission count tends to infinity which means there will be no successful transmission of the
packet.
C. Security in V2V
In the previous subsection, an efficient dynamic spectrum access method was described for
communication between multiple vehicles in IoV. However, with every communication is a
threat of security and privacy. Despite the recent de- velopment in VANET, the issue of security
and privacy is still under-developed. Although it has to be agreed that this is a continues process
as threat space develops with new technology and increased understanding by adversaries, it is
important to have a framework in plae to be built upon for establishment of minimum security
standards.
In an information-sensitive domain like vehicular networks where every bit of information is
defining of life or death, it is important to ensure the trustworthiness of message received from
other vehicles without violating the their privacy nor with the use of a trusted authority. This is
important if the goal of VANET in reducing road accident and fuel consumption is to be
actualized.
In V2V networks, the absence of a roadside infratructure adds an extra level of complexity to
trust establishment. Messages between a source vehicle and another destination vehicle is routed
by intermediate vehicles in V2V. The question for any of such vehicle is to verify if the
information it has received is legitimate or not. Trust is historically known to be a factor baed on

social interactions and has been introduced in information and communication technology (ICT)
during the last decade [38]. An often considered approach to V2V security is by using
cryptographic algorithms.
Each vehicle is required to receive messages sent to it in single or multiple hops with other
vehicles in the loop making this kind of communication susceptible. If a intermediate vehicle
chooses to alter a message, then the received message at the destination will be misleading.
Ensuring integrity and authenticity of such communication however must also safeguard the
private information of the vehicles and drivers alike as well as their travel path. Basic solutions
such as digital signature, symmetric and assymetric cryptography, and authentication protocols
have been used to address the security problem in V2V networks [39]. Such solutions however
do not cover the human-like psychological need of such fast- paced interactions.
In a bid to establish integrity of communication defined as a function of trust of vehicles in the
message delivery path, a probabilistic approach which observes the legitimacy of vehicles over a
period of time is a viable alternative. If a vehicle is not legitimate, it follows that message
received from such vehicle can be discarded. Such vehicle can then be warned about its
malicious activities [40].
Authentication of vehicles while preserving the privacy of vehicles has become a major
challenge in vehicular networks in general [41]. Hence, trust is an important factor in
establishing privacy and security in V2V networks. Recent research works focus on the usage of
pseudonyms and algorithms for changing them at reasonable intervals [42]. However,
establishing context-aware security and privacy mechanisms built on trust is an emerging area of
research. In the next subsection, a context-aware probabilistic trust measurement is discussed to
enforce security and privacy in V2V networks.
The approach
Considering a relay node i, the trust of i for instance can be defined as a function of the number
of packets that i successfully transmitted without modifying the message. This approach is
leveraged on the notion that a node acts the same way as it has done so far. Subsequent packets
have a probability of successful transmission equivalent to the evaluated trust value of the node.
It should be noted that due to the low contact time of vehicular nodes in a V2V network which is
a result of their high speed of mobility, trust evaluation even becomes difficult to evaluate.
A contextual approach that establishes trust of vehicular nodes using a proba- bilistic model is
proposed for V2V networks. In this approach, trust is assumed to be based on service

requirements satisfying QoS, security, privacy etc. This is enforced by having multiple copies of
the same message from a source vehicle s to a destination vehicle d over different paths.
For a given path m, the overall trust value is defined as
Xk j=1
where Tj is the trust evaluation value of node j and k is the number of nodes in path m.
Algorithm 1 describes the process of identifying malicious paths.
The evaluation of the trust value of a vehicular node is based on its current and historical trust
values using the model in [43]. Due to the temporal nature of V2V networks and how fast
components change, the historical trust values is important for trust evaluation. The trust value of
a vehicular node i at any given period is given as follows
Ti =↵⇥(Tcur)i +(1 ↵)⇥(Tμ 1)i (38)
where the value of ↵ is such that 0 < ↵ < 1 and representing the weight given to the trust value
of the current period. The weight is used to determine the importance attached to either the
current trust value or historic trust values from previous periods. A period can be represented in
terms of seconds, minutes, or even hours.
To determine the current trust value of a vehicular node Tcur, a probabilistic approach is
employed. Consider that a message Mi is sent by a vehicle i…
until message is received from other paths return max(T Tm ) modification by any relay node
will mean that the message becomes Mi ± since a portion can be added or removed from the
message. However, this is difficult to determine due to noise introduced in communication.
Hence, we consider the probability that the introduction of noise up to a certain threshold does
not alter the message. The probability of the message being altered by noise is defined as
Pmsg = Pr( i < ̄ ) (39)
where i is the signal-to-noise ratio (SNR) of the message from vehicle i and ̄ is the SNR
threshold set for which noise alters the message.
In the trust evaluation period, the beta function is used to calculate the satisfac- tion value such
that

Si = Nsuc + 1 (40) N suc+Nfail +2
where Nsuc and Nfail are the number of successful and failed interactions respec- tively by
vehicle i within the trust evaluation period. Subsequently, the trust value of the vehicle Ti is
defined as a function of its satisfaction value such that:
Ti =f(Si,Pi) (41)
Finally, as described in algorithm 1, the path with the maximum total trust value is selected such
that:
…..
where P is the set of paths for a message from a source vehicle to another destination vehicle.
To evaluate the proposed approach, the average level of trust of vehicles is estimated with
different SNR values. Blacklisted vehicles are considered to be malicious and assumed to change
the content of the message in transmission from a source to destination vehicle. The average trust
value of messages is plotted against different SNR values as shown in Fig. 9. It is observed that
the trust value increases as the SNR increases which indicates highest trust value for unchanged
message.
D. Conclusion
The importance of V2V network with respect to connectivity and security has been discussed so
far in this chapter. A study of the impact of secondary vehicular user mobility and primary user
activity has been conducted for dynamic spectrum access. Subsequently, a probabilistic approach
was formulated for trust calculation in maintaining integrity of vehicle-vehicle messaging
through intermediate vehicles. The performance of the proposed formal models are evaluated
using numerical results obtained from Monte Carlo simulations. It is found that the probability of
misdetection of primary user activity over a channel increases with increased vehicular speed
and decreasing expected overlapping time duration per epoch. Also, based on experiments
performed, the average trust value for vehicles increase with increasing SNR, with honest
vehicles showing much higher trust value than malicious vehicles.
III. CHAPTER 3: VEHICLE-TO-INFRASTRUCTURE ADHOC NETWORK

The Vehicle to Infrastructure (V2I) communication is another mode of commu- nication in
vehicular networks. Similar to V2V networks, the goal is to enable roadside intelligence through
vehicles relaying messages about their context to ensure an effective transportation system and a
step closer to autonomous vehicles plying the roads in the nearest future. However in V2I,
vehicles are expected to communicate with road-side units/infrastructures. Unlike in V2V
architecture where a multi-hop method is used allowing for rapid transmission of data, V2I
reduces the implementation concerns of V2V by using a centrlized architecture involving
stationary road side infrastructure units.
As in V2V, vehicle to infrastructure interaction is based on wireless commu- nication
technologies. The V2I communication is also commonly referred to as V2X, and is a subject of
extensive research in the United States. Primarily, to mitigate motor vehicle accidents and enable
a number of safety and environmental benefits, critical safety and operational data are exchanged
between vehicles and highway infrastructure. V2I networks are incorporated with algorithms that
utilize the messages exchanged to provide intelligence in recognizing high-risk situations
proactively and perform specific actions such as traffic light signalling, driver alerts etc.
The principal components of a V2I architecture are the vehicle on-board unit (OBU), the
roadside unit (RSU) and the communication channel. A typical ar- chitecture is shown in Fig. 10.
The OBUs are the part of the vehicles that forms the V2I network. Usually, OBUs are equipped
with a GPS system, an application processor and a radio transceiver. Also, OBUs are equipped
with (limited) storage capbility to save snapshots of data, with older data being overwritten by
the most- recent. Other vehicle data are also gathered by the OBU for transmission to the RSU.
RSUs are often dedicated equipments placed at strategic locations to provide the interface for
vehicles within their range. They could be mounted at interchangesm intersections , or any other
location. The RSU consists of a radio transceier, an application processor, a GPS unit, and an
interface through which it is connected to the V2I network. The interface of the RSU allows it to
be able to send/receive private data to/from OBUs. RSUs are equipped with prioritization
frameworks to manage the available bandwidth for V2I communication. For instance, vehicle-to-
vehicle safety messages will have the highest priority, with entertainment messages having the
lowest priority.
Fig. 10. A typical vehicle-to-infrastructure architecture []
In subsequent subsections, a description of an effective communication mecha- nism for V2I

communication will be given, as well as a framework for maintaining privacy within the V2I
communication network.
A. Communication in V2I
For V2I applications, communication mechanisms used are similar to those in V2V networks.
Advances in technology especially with the development of Cognitive Radio Technology (CRT)
and Dynamic spectrum access (DSA) has made it possible for unlicensed secondary users to use
the vehicular network systems by employing licensed bandwidth provided it does not cause
harmful inconvenience to its usage by licensed primary users. These technologies allows the
fulfillment of increasing demand for access to wireless broadband at all time in all locations,
especially in congested areas characterised by increased volume of vehicular nodes resulting in
higher need for safety applications. It is however worthy of note that unlike in other wireless
applications, networking from vehicles takes a lot of bandwidth [44]. The goal of research in this
domain is to allow seamless messaging between OBUs and RSUs in a real time format.
Dynamic spectrum access can be employed both on the OBUs and RSUs. Vehicles with
cognitive radio capabilities are capable of changing to different radio technologies depending on
the RCU they are communicating with [45].They will also be able to discover holes in the
spectrum as they travel along the highway system. Roadside units (RSUs) can employ the same
cognitive radio techniques as those employed by a vehicle’s on-board units (OBUs). As the radio
feed changes, CR permits swift adaptation to new environments.
Vehicular ad-hoc networks (VANETs) place wireless sensors in vehicles which are engaging
with the network, alongside on-board units (OBUs) that allows the in-vehicle equipment to
communicate wirelessly with other units in their location. Each vehicle must be capable of
sending, receiving and routing data to and from other vehicles or roadside units (RSUs) that are
within the range of their wireless transmissions [46].
Road Side Units (RSUs) allow vehicles to communicate with the infrastruc- ture operating in a
similar way to wireless LAN access points. It is important that RSUs have the capability of
allocating specific channels to the units on board vehicles (OBUs). RSUs must be able to provide
clear channels for the emergency services (police, fire, ambulance) when necessary [47]. Road
Side Units (RSUs) are stationary, and have the capability to communicate with each other
especially for intelligence gathering and handover. There are however two

communication requirements in such network: RSU-to-RSU usually via wired connectivity, and
vehicles-to-RSU via wireless [48]. This form of hybrid connection allows RSUs serve as routing
nodes in vehicle-to-vehicle cimmunication. Vehicle- to-Infrastructure (V21) systems allow
vehicles to communicate with roadside units. Information sent from vehicles to these units can
then be shared with wider networks as shown in Fig. 11.
RSUs are strategically placed within the domain of an intelligent transport system where they
can most effectively and reliably communicate with vehicles. They employ IEEE 802.11p
technology that allows them to communicate wirelessly over dedicated short-range wireless
communication by the support of network devices [46]. Road Side Units can send vehicles
information related to accidents, meteorological conditions, traffic jams etc., and in return, the
RSUs can harvest information regarding traffic speed, which vehicles will have to pay tolls, etc.
[49].
As mentioned earlier, applications of V2I are closely related to that of V2V systems. Generally,
V2I applications are called Intelligent Transportation System (ITS) applications. The primary
application of these systems is to provide safety services. The aim is to reduce the occurance of
road accidents by proactive pre- diction based on the information obtained through
communications between the vehicles and roadside infrastructures. Transportation advices in
form of alerts and notification is then feedback to vehicles. Typical safety applications revolve
around speed management, merging assistance, warning for hazardous situations such as
obstacles, accidents and congestions.To provide better utilization of roads and intersections,
efficiency applications are important in traffic jam notification, prior recognition of potential
traffic jam, dynamic traffic light control and interconnected navigation. Also, information ob-
tainted from vehicles could be used for parking control, congestion charge and highway toll
control.
A description of available communication mechanism for V2I systems is given below:
Short range radio Short range radio means an older technology, which is wide-spread in case of
public transport vehicles [50]. They have been installed with a short range radio transmitter
which works on a lower ISM band (such as 433 or 868 MHz) [51]. It can broadcast an identifier
which can be received by the traffic control systems roadside beacon and thus the public
transport vehicles can be prioritized in the intersections or by the stops.
DSRC

Dedicated Short Range Communications (DSRC) has emerged as the favored form of
technological solution for VANETs. DSRC has been allocated (by the FCC) the space on the
spectrum between 5850 and 5925 MHz, giving it 75 MHz bandwidth (1999). DSRC works to
IEEE 802.11 specifications (soon to be standardized under IEEE 802.11p) [52].
The Association of Radio Industries and Businesses (ARIB) created a standard for DSRC RVC
systems working on 5.8 GHz known as ARIB STD-T75 which has since been used for the
informative supply services alongside the ETC (separate from ARIB STD-T55, which is solely
for the ETC use) [53]. Safety of V2I systems for road users has been enhanced over the years
with the wireless technology of DSRC [54]. DSRC’s prime use is in enabling applications that
prevent vehicles from colliding. In order for these applications to work properly, vehicles must
be enabled to exchange data quickly and regularly between themselves and roadside units. It has
been estimated by the US Department of Transportation (DOT) that employing DSRC for V2I
communication purposes could prevent 82% of vehicle collisions (where the driver is not
impaired) [55]. This, by extension, has the potential to prevent thousands of deaths and to save
billions of dollars.
The main way in which DSRC helps to prevent collisions is hat every vehicle which is equipped
with DSRC broadcasts information about its location, acceleration and speed. It will do this
multiple times each second. A suitably equipped vehicle within several hundred meters will
receive this information, and it will, in turn, receive information from them. Each vehicle uses
the data received to plot the direction of neighboring vehicles; this is then matched against its
own direction of travel to assess the likelihood of a collision. DSRC equipped vehicles can also
receive information from RSUs pertaining to possible hazards, upcoming road layouts et cetera
[55].
Wi-Fi
Wireless local area networks (WLANs) were initially created to avoid having to install new
cabling for computer interactions within buildings. Since then the capacity and use of wireless
networks has greatly extended, and there are many different sorts of networks (WxAN), all of
which differ in their approach to the trade-off between speed and portability. These run alongside
the cellphone network, which allows users to employ their devices anywhere in the world but
works comparatively slowly. There are a trio of categories for WxAN, thus:
• WLAN: these networks provide high bandwidth access (several MB per sec- ond), allowing
personal computers and similar appliances connection with a permanent network, e.g.

Ethernet. Such networks may be found in schools, hospitals and other public buildings.
• WPAN (wireless personal area networks): these are very similar to WLANS, but are chiefly
employed in domestic situations for personal entertainment, e.g. streaming video. The
first WPANs were powered by Bluetooth technology, but as this is not fast enough to
allow for video streaming, it is anticipated that more powerful systems will be developed.
• WMANs (wireless metropolitan area networks): these networks employ broad- band radio
access across wide areas; they can be accessed by the other two types of network. With
reference to the use of wireless technology within vehicles, currently WPAN and WLAN
are being employed, but WMAN may well be used in future. The space on the spectrum
between 800 MHz and 6 GHz is likely to be the site of any such development [56]. Any
network employing IEEE 802.11 standards is known as a Wi-Fi (wireless fidelity)
network. These networks, employed as WLANs, may run up to 54 MB per second
(802.11a and 802.11g) or up to 11 MB per second (802.11b).The differences between
these 802.11 networks come under the headings of band- width, security and coverage;
these three elements determine what kind of ap- plications may be used. 802.11a is best
for multimedia applications, transmitting sound, video and large images; it is most suited
to areas where there is a large number of users. 802.11b has a wider range, and so can be
accessed over a larger area with fewer access points required. 802.11g can work in the
same way as 802.11b, but it is more secure and provides higher bandwidth, so it may
supersede 802.11b [57].The most recent version of WLAN, IEEE802.11n, has much
faster data transmission (up to 600 MB per second), is reliable and works over a greater
area, and so is expected to represent a significant improvement in WLAN standards [58].
Cellular Networks
Vehicle-to-infrastructure systems require vehicles to be connected with roadside infrastructure.
For a long time, it has been suggested that cellular networks could achieve this [2]. Cellular
networks have the great virtue of the fact that all necessary infrastructure is in situ and working.
There is a further advantage that many modern vehicles already possess cellular communication
devices. Since the spread of data will not depend on customer take-up, early adopters will
experience immediate benefits [59].
Cellular networks and short-range communication systems have recently expe- rienced
significant developments which make V2I communication real possibility, given that anything

from a single automobile to a traffic signal may now be given its own means of computing and
communicating. It may be possible to develop further through the use of the cellular network.
The increase in smartphone use [60] means that a simple application on a driver’s cellphone
could take the place of in-vehicle equipment [61]. It has been acknowledged that these systems
will work better over cellular networks than DSRC ones (offering as they do a greater range for
transmission), and in addition will be much simpler to install by simply using smartphone apps
[61].
It should be noted that both of these methods (DSRC and cellular) have inherent limitations
when used in this way. DSRC was originally intended to transmit data over short ranges without
needing large amounts of infrastructure; although cellular networks have a much greater range,
they cannot guarantee real-time data exchange in all locations. In order to reconcile these two
problems, i.e. lack of range and real- time data transmission, Heterogeneous Vehicular Networks
(HetVNETs), which combine both DSRC and cellular networks may be the best solution for
intelligent transportation systems [62].
Researchers have been investigating ways in which DSRC might be combined with Wi-Fi,
WiMAX and LTE to create the most effective network per vehicle communication [63]. It is
anticipated that DSRC RSUs will be placed at crossroads and interchanges. When used in
combination with other existing networks DSRCs relatively short range will not be a handicap
when used as part of a heterogeneous wireless network (Het-Net) [64]. By using Het-Nets a
greater range can be es- tablished, compared to simply using DSRC, and this is necessary if
applications are to provide useful information over large traffic networks. However, passing data
between different types of network will entail several seconds in the handover phase, and so Het-
Nets would not be suitable for applications where safety was dependent upon instantaneous data
transmission [64].
THE APPROACH
As highlighted earlier, one of the applications of V2I is the support for emer- gency services.
This translates to the need for real-time data traffic between vehicles and infrastructure in a
vehicular adhoc network. This is often the requirement for proactive safety applications. Having
guaranteed delay bounds is important for the reliability of drivers on such applications especially
in critical situations. This need also extends to non-critical systems where bandwidth is aimed to
be better managed to encourage user adoption and reduce development cost.
In traditional V2I systems, all vehicles report their location and speed to a RSU. Another

important feature of the V2I ystem is the ability to run queries of this data. However, the
infrastructure represents a single point of failure in this network. Also, when multiple vehicles
attempt to report their statutory data to the RSU at the same time, it queues the transmissions
however increasing the delay for data upload in some part of the network which results in a
reduction in the accuracy of the prescribed action/notification to the vehicles.
In pure V2V architecture, all the vehicles report their data to neighbour vehicles who chooses to
propagate the data further in the network without the use of a trusted central authority. However,
the proposed approach leverages on such communication as a link to the infrastructure. In the
proposed aggregated vehicle to infrastructure architecture (AV2I), an overload of the
communication channel to RSUs is aimed to be reduced. Subsequently, it is expected that this
results in a faster uplink transmission by vehicles to RSUs and downlink from RSUs to vehicles.
There is a large amount of data that must be received by the server when all vehicles are sending
their data as well as querying for other information. AV2I attempts to reduce these limitations by
building on the V2I architecture.
In this architecture as shown in Fig. 12, the transportation network is divided into pre-defined
grids which consist of one or more vehicles. Because each grid is predefined, it is expected that
every vehicle node is aware of the grid it belongs at every time based on its location coordinates.
Within each grid, a super vehicle is elected . The super vehicle is responsible for the aggregation
of the speed data of the grid to the RSU, as well as to other neighbouring super vehicles. It is
important that the grid size is such that the farthest two vehicle nodes in the grid can
communicate directly with each other. Also, the positioning of super vehicles should be such that
a super vehicle will be able to establish communication with at least one neighbor super vehicle.
This is necessary as a fail-safe mode in the case of the inavailability of the RSUs in that grid.
All vehicle nodes within a grid send their speed and location information over a wireless link to
the super vehicle, which in turn aggregates this and sends to the RSU. The frequency at which
aggregated data is sent to the RSU can however be configured based on the application need of
such data. At the RSU, the data is obtained just like with individual vehicles sending their data.
However, how a super vehicle is elected remains an important aspect of this algorithm.
Due to the mobility of vehicular users across grids, it is expected that the super vehicle will
change frequently. The super vehicle election (SVE) algorithm as depicted in algorithm 2 will be
used to create a new super vehicle and perform handover to a new super vehicle when the current
one leaves the grid. The election of a super vehicle is dependent on the estimated remaining time
to be spend in the grid at the time of election.

Algorithm 2 Super Vehicle Election algorithm G set of predefined grid 4t remaining grid time
threshold while true do timesend newelection message to all vehicles in g all vehicles calculate
the erv time…
To ensure that a known malicious vehicle node does not emerge as super vehicle giving it the
capability to hoard information and thus mislead the network, a proba- bilistic approach is
employed. To ensure that a super vehicle elected is trustworthy, the suspicion level of a vehicle v
can be defined as sv(t) ⌘ P(Tv = M|Ot) where Tv is the type of vehicle that can be malicious
(M) or Honest (H), and Ot is the observation carried out at time interval t. This suspicion level
can then be rewritten as:….
The architecture described so far provides many benefits including the reduction of bandwidth
requirement for the RSU, and offering fault tolerant mechanism in an event of hardware failure.
However, to support this claim, a series of Monte Carlo simulations were carried out with the
results from numerical analysis explained below.
Firstly, the average bandwidth utilized in a network with respect to the number of vehicular
nodes is evaluated. As shown in Fig. 13, the fraction of bandwidth utlized in V2I is much higher
than than of AV2I and it also tends towards 1 with increasing number of vehicular users. This is
because individual vehicular users in V2I report their location and speed data to RSU, whereas
only one vehicular user reports the speed and location data for all vehicles in a grid in AV2I.
Also, the delay in data upload experienced in the network is evaluated. Assuming that each data
upload to a RSU is queued, the average percentage delay experienced in data upload to RCU is
measured against increasing number of vehicular users. As shown in Fig. 14, the delay
experienced in V2I is higher than in AV2I due to the upload queue which increases waiting time
as number of users increase when they all upload their context data at once.
B. Privacy in V2I
The nature of wireless communication used in V2I communication network makes it vulnerable
to a lot of attacks. Cyberattacks on V2I communication can have devastating consequences if
V2I systems are not properly secured. What makes V2I security hard to achieve is the tight
coupling between applications, with rigid requirements, and the networking fabric, as well as the
environmental factor, and economical considerations. Solutions to this problem involve industry,
governments, and academia, and can have a broad impact.
V2I applications present a variety of vulnerabilities that create an attractive target for hackers.

For example, hackers could take control of traffic signals, create hazards, and even cause a
breakdown of the traffic system. Much research has been conducted on enabling privacy,
maintaining authentication, and providing integrity by targeting V2V communication. However,
very few studies focus on V2I-level security. V2I networks require similar security features as
V2V such as authenti- cation, confidentiality, integrity, privacy, and efficiency. In V2I networks,
there are extremely large amounts of network entities and data packets in situations of high
traffic density such as traffic jams. Thus, one of the main challenges for secure V2I
communications is to achieve a fast, efficient, and secure communication.
Since inexpensive wireless devices are available, typical wireless devices are installed at various
roadside units (RSUs), such as road signs and traffic lights [65]. In V2R based communications,
trustworthiness of the message can be easily verified since the locally centralized roadside unit
can keep track of the messages and participating vehicles [66]. By that, every node in this
communication architecture is under the control of a RSU and every message transmission can
be done only through RSU. If the RSU is alone secured form the various types of security attacks
the network communication can be carried out with less security threats.
However, the vehicles must be authenticated initially with RSU. Also, when a vehicle moves
from the current RSU to new RSU, the new RSU should authenticate the vehicle. It means that
authentication and access control functions for access network are required [67]. After the
request is accepted, the vehicle will receive the local safety messages.
Typically, VANETs consist of three components: a trusted third party, roadside infrastructures,
and vehicles.
• Trusted third party: a trusted third party (TTP), which refers to a trusted administration with
sufficient computational and storage resources where all vehicles register and get their
certificates for VANET usage, is responsible to hold the credentials and the identities of
vehicles and to reveal the real identities of nodes whose certifications have been revoked.
In addition, they are also in charge of RSUs. TTPs are fully trusted by all entities [68]. In
reality, a large number of TTPs exist and each one of them is responsible for a specific
geographical region. Each vehicle and RSU should be registered with exactly one TTP.
• Roadside units: RSUs are infrastructures fixed on the roadside, which are fully controlled by
TTPs. The RSUs are installed to help the vehicles to do massage authenticating or
communicate confidentially [69]. RSUs are quite vulnerable because they are easily
exposed to attackers, so we must put minimal trust in RSUs. For enhanced security,

RSUs could directly communicate with TTP and if TTP considers that a specific RSU has
been compromised, it could revoke the RSUs access.
• Vehicles: Vehicles are the moving nodes in the network, which are loaded with an OBU, a
tamper proof device and a reliable Global Positioning System (GPS). The OBU is used to
enable vehicles to wirelessly communicate with each other and RSUs, and the TPD is
used to store cryptographic materials, such as an Electronic License Plate (ELP) that is
installed on every new vehicle and provides a unique ID number, and process
cryptographic operations. TPD is a good second defense layer but should not be
exclusively relied upon. In V2I scenarios, vehicles send requests to the nearest RSU
when the vehicles want to get access to services provided by the RSU, such as Internet
service or, information about the nearest restaurant. Vehicles should first authenticate
them- selves with RSU to get permission to allow them to broadcast messages. When a
vehicle passes by the RSU, it should also authenticate the validity of the RSU in case it is
a fake RSU. Once the RSU is authenticated, the vehicle sends the encrypted request
messages and its certificate to the RSU. The RSU decrypts the request and then looks up
the newly updated revocation list retrieved from TTP to check whether the vehicle is
entitled to obtain the service. If the certificate is on the revocation list, the RSU rejects
the request, otherwise the vehicle is authenticated. If the vehicle is authenticated, the
RSU sends the response back to the vehicle and provide the service request. The vehicle
should also check the validity of response after receiving it.
However based on the AV2I approach discussed in the previous subsection, two pertinent
problems arise. Vehicle users all send messages in form of speed data and location data to the
super vehicle, which brings up the issue of identification. For this, it is important that
information/metadata broadcast by the uper vehicle about its identity is not harmfully-revealling.
In a bid to enforce the privacy of the super vehicle user by maintaining disclosure of its real
identity, a grid-based identity management scheme is proposed.
In this scheme, a pseudo-identity technique is employed for the super vehicle user, who is the
only vehicle whose identity is necessary to be broadcast within the grid and across. The super
vehicle is expected to a pre-shared key which will be used by neighboring vehicles to encrypt
their speed and location data. Let the message expected to be sent to a super vehicle s by a
neighboring vehicle n be expressed as Mns. This message needs to be encrypted using a session
key broadcast by the super vehicle.
The super vehicle, due to its mobility has to change the session key periodi- cally based on a
particular key change threshold. The key is only formed at the emergence of the super vehicle.

This key is the result of a hashing function of the current location of the vehicle user and a
randomly generated number such that the session key for a grid g is defined as:
keyg =hash(r,superloc) (46)
where r is the randomly generated number, and superloc is the location description of the super
vehicle defined as a function of its longitudinal and latitudinal coordi- nates.The message Mns is
then encrypted with the key to give M = E(Mns,keyg). This message is also decrypted by the
super vehicle such that Mns = D(M,keyg).
Also with regards identity, a lightweight key exchange algorithm is needed for super vehicle to
RSU communication for location and speed reporting as well as road-side services. A Diffie-
Hellman key exchange algorithm which offers perfect forward secrecy is proposed as described
below.
When a super vehicle V intends to communicate with an RSU R, V comes up with two random
prime numbers g and p. Also, a secret number a is chosen and used to compute A = gamodp
which is sent to the R. R also selects a secret number b and computes B = gbmodp which is sent
also sent to V. Henceforth, V computes Bamodp while R computes Abmodp which results in the
key kvr.
This key is then saved by the RSU together with the pseudo-identity of the super vehicle. This
enables for faster re-authentication of a vehicle with intention to communicate with an RSU as a
super vehicle. The shared key is hashed with the ID of the super vehicle H(ID||kvr) which is
stored in a cache/database associated with the RSU. Recall that ID is a pseudo ID of the super
vehicle.
C. Conclusion
In this chapter, the various components that forms the basics of a functional V2I communication
network has been discussed. The importance of RSUs and third party trusted authority as part of
the V2I architecture has also been discussed. In addition, an aggregated vehicle to infrastructure
architecture was discussed where the transportation network is divided into pre-defined grids.
Each grid then has a super vehicle which aggregates messages sent to the RSU infrastructure.
The super vehicle is elected based on the estimated remaining time it is expected to spend in the
grid. As an added security measure to avoid malicious vehicles emerging as super vehicle, a
probabilistic approach to malicious vehicle identification was also discussed.

It has become increasingly difficult to guarantee all of security, privacy while also considering
performance requirements for V2I communication networks. A secured and efficient group-
based encryption mechanism has been proposed based on the super vehicle’s hashed location.
Been proposed based on the aggregate vehicle to infrastructure architecture, the security and
privacy mechanism proposed can be scaled large implementation of Internet of Vehicles.