Content uploaded by Shah Khalid Khan
Author content
All content in this area was uploaded by Shah Khalid Khan on Nov 05, 2020
Content may be subject to copyright.
Content uploaded by Nirajan Shiwakoti
Author content
All content in this area was uploaded by Nirajan Shiwakoti on Oct 29, 2020
Content may be subject to copyright.
Cyber-Attacks in the Next-Generation Cars, Mitigation Techniques,
Anticipated Readiness and Future Directions
Shah Khalid Khan*, Nirajan Shiwakoti, Peter Stasinopoulos, Yilun Chen
School of Engineering, RMIT University, Carlton, Victoria 3053, Australia
*Corresponding author, Email: s3680269@student.rmit.edu.au
Abstract- Modern-day Connected and Autonomous Vehicles (CAVs) with more than 100
million code lines, running up-to a hundred Electronic Control Units (ECUs) will create and
exchange digital information with other vehicles and intelligent transport networks.
Consequently, ubiquitous internal and external communication (controls, commands, and data)
within all CAV-related nodes is inevitably the gatekeeper for the smooth operation. Therefore,
it is a primary vulnerable area for cyber-attacks that entails stringent and efficient measures in
the form of "cybersecurity". There is a lack of systematic and comprehensive review of the
literature on cyber-attacks on the CAVs, respective mitigation strategies, anticipated readiness,
and research directions for the future.
This study aims to analyse, synthesise, and interpret critical areas for the roll-out and
progression of CAVs in combating cyber-attacks. Specifically, we described in a structured
way a holistic view of potentially critical avenues, which lies at the heart of CAV cybersecurity
research. We synthesise their scope with a particular focus on ensuring effective CAVs
deployment and reducing the probability of cyber-attack failures. We present the CAVs
communication framework in an integrated form, i.e., from In-Vehicle (IV) communication to
Vehicle-to-Vehicle (V2X) communication with a visual flowchart to provide a transparent
picture of all the interfaces for potential cyber-attacks. The vulnerability of CAVs by proximity
(or physical) access to cyber-attacks is outlined with future recommendations. There is a
detailed description of why the orthodox cybersecurity approaches in Cyber-Physical System
(CPS) are not adequate to counter cyber-attacks on the CAVs. Further, we synthesised a table
with consolidated details of the cyber-attacks on the CAVs, the respective CAV
communication system, its impact, and the corresponding mitigation strategies. It is believed
that the literature discussed, and the findings reached in this paper are of great value to CAV
researchers, technology developers, and decision-makers in shaping and developing a robust
CAV-cybersecurity framework.
Keywords: Connected and Autonomous Vehicles (CAVs), Electronic Control Units (ECUs),
Cybersecurity, CAVs communication framework, Driverless cars.
1. Introduction
In the last few decades, the transition of automotive systems from electromechanical to
electronic and software-driven systems has changed the dynamics of the automotive vehicle
industry. There is a massive increase in the software-contents due to built-in applications in the
vehicle's Electronic Control Units (ECUs) (Möller and Haas, 2019). This is attributed to the
enhanced sophistication of the fundamental control systems but is also representative of more
technological growth than hardware advancement, and these innovations are primarily
occurring in the automobile, electronics, software, and telecommunications sectors, as shown
in Fig.1. This boom aims to adopt the emerging technologies developed for fully automated
cars, predictive intelligence, Advanced Driver Assistance System (ADAS), V2X
communication, state-of-the-art infotainment/telematics systems, and shift to connected and
autonomous driving. Connected and Autonomous Vehicles (CAVs) utilise wireless networks
and sensors to collect traffic and other critical information while their driving is governed by
one of the six automation stages (SAE-International, 2018). Levels range from 0 to 5; SAE
Level 0 means human driver performing all tasks related to the vehicle operation, and SAE
Level 5 is the full autonomous navigation without human involvement. Based on these policy
concepts, an automated vehicle at levels 4 and 5 is a self-driving car, but a self-driving vehicle
at level 3 is not automated because it is restricted in the operating area and needs a human
driver who can take over as appropriate.
The digitalisation of the automotive industry would generate $3.1 trillion in socio-economic
benefits and $67 billion in business revenue (West, 2016). Various firms, from mainstream
automotive makers like Volkswagen to internet-based businesses such as Google and Apple
are spending immense human resources and capital in the production of CAVs (Behere and
Torngren, 2015). These vehicles will be inter-connected, having access to the internet and
interact with the surrounding environment through various sensors without the driver's
involvement (Gehrig and Stein, 1999). The consequence of this transition is that the CAV
should be a fully-fledged cyber-physical system (CPS), a collaborative network of electronic
components regulating mechanical parts (Scalas and Giacinto, 2019).
Connected and autonomous driving technology accelerates the need for continuous
communication between the CAV-ECU and a range of cloud services. It will develop
sophisticated computing and efficient vehicle manoeuvring techniques, along with the prospect
of delivering new software upgrades and other relevant content. As a result, the key to the
successful integration of CAVs into the Intelligent Transport System (ITS) is the seamless
internal and external communication within all CAV-related nodes. CAVs communication is
the primary area vulnerable to cyber-attacks, requiring a rigorous and efficient approach in the
form of cybersecurity. Cybersecurity is a set of technologies, procedures and activities
designed to secure devices, files, networks and systems from interference, disruption or
unwanted exposure to cyber-attacks.
Cyber-attacks in the CAVs communication contribute to two security issues, i) limited
access: showing that CAV has an established wireless connection, but still does not have access
to the internet, and ii) limited computational capacity: halting or stopping any
calculation/judgment by the CAV that requires both arithmetical and non-arithmetic measures
to follow an algorithm or a well-defined model. These vulnerabilities lead to unforeseen assault
scenarios and may pose vital hazards to CAV-users and infrastructure (El-Rewini et al., 2019).
In July 2015, Fiat Chrysler recalled 1.4 million cars due to doubts about car software and
alleged remote control. Software coding errors caused the Nissan Leaf to be hacked using the
Nissan Connect EV application. An error has made it possible for hackers to remotely access
in-car infotainment system and display driver identification details (Morris and Madzudzo,
2018).
Worldwide research shows many vulnerabilities and uncertainty in CAVs in terms of cyber-
attacks. In this uncertainty are several possibilities for autonomous vehicle technology to
underdeliver on its promise of safety and security; and, instead, become a net burden on society
(Stasinopoulos et al.). There is a range of continuing initiatives to maintain the protection and
security of CAVs, which are primarily focused on the extension of strategies currently utilised
to secure different cyber and physical elements. However, there is no structured paradigm for
CAVs that tackle protection in a unified framework that addresses; i) hardware challenges, (ii)
network challenges, (iii) human threats, as well as iv) software risks. The current literature
lacks; i) the transparent picture of all the possible interfaces for cyber-attacks in a consolidated
form, and ii) the analysis of all possible avenues for the cessation of cyber-attacks on the CAVs.
Potential cybersecurity mitigation strategies in CAVs are mostly demonstrated for specific
scenarios without considering the overall picture of CAV operation in an integrated ITS.
Adversaries access to CAVs through proximity (or physical) access is a doorway for hackers,
which needs an in-depth understanding and analysis. Similarly, synthesis from a review of
existing literature shows that among the CAVs security performance evaluation metrics
availability, authenticity and confidentiality are discussed in detail by various researchers. The
scope of integrity is comparatively less discussed; however, reliability, robustness, and
trustworthiness performance metrics are still subject to in-depth research and review, as
depicted in Fig.2 (Lopez et al., 2019; Möller and Haas, 2019). The purpose of this study is to
address these limitations and to assess critical areas for the advancement of CAVs in combating
cyber-attacks.
1.1. Contributions of the study
Currently, there is a lack of systematic and comprehensive review of the literature on cyber-
attacks on the Connected and Autonomous Vehicles (CAVs), respective mitigation strategies,
anticipated readiness, and research directions for the future. Our study aims to fulfil these
knowledge gaps in the existing literature. The main contributions of this study are as below:
1) We described in a structured way a holistic view of potentially critical avenues, which
lies at the heart of CAV cybersecurity research. We analyse their scope with a particular
focus on ensuring effective mass deployment of CAVs and reducing the likelihood of
cyber-attack failures.
2) For each potential avenue, we present a taxonomy of different threats and illustrate the
generalization of attack surfaces for autonomous and connected vehicle applications.
3) Moving forward, we present the CAVs communication framework in an interconnected
manner, i.e., from In-Vehicle (IV) communication to Vehicle to Everything (V2X)
Fig.1. Digital transformations
in the automotive industry (Source: Authors’
synthesis).
Fig.2. Performance evaluation metrics for
CAVs security (Source: Authors’ synthesis).
communication with visual flowcharts and pictorial representations to provide a clear
description of all future cyber-attack interfaces.
4) We highlight and emphasized that why conventional CPS cybersecurity approaches are
not adequate to counter cyber-attacks on CAVs. Subsequently, we review and synthesis
a consolidated summary table and description of CAVs cyber-attacks, the respective
CAV-communication system, and the corresponding counterstrategies with future
directions.
5) Finally, the vulnerability of CAVs to adversaries by proximity (or physical) access to
cyber-attacks is synthesized. There is a segregation of cyber-attacks in the relevant
performance evaluation metrics. We illustrate research gaps with future directions for
each potential avenue, which needs further investigation.
The paper is organized as follows. The next section briefly explains the methodology used for
the study. We then present the state-of-the-art to enlighten the scope, challenges, defensive
strategies, anticipated readiness, and research gaps of cyber-attacks in CAVs. Finally, we
conclude by summarizing the major contributions of this study. A list of the abbreviations used
in this manuscript is provided in Table 1.
Table 1. A list of abbreviations used in this study.
Abbreviation
Explanation
ADAS
Advanced Driver Assistance System
AI
Artificial intelligence
BCM
Body Control Module
CAN
Controller Area Network
CAV
Connected and Autonomous Vehicles
CPS
Cyber-Physical System
DSRC
Dedicated Short-Range Communication
DoS
Denial of Service
ECM
Engine Control Module
ECU
Electronic Control Units
ITS
Intelligent Transport System
IV
In-Vehicle
JTAG
Joint Test Action Group
LIN
Local Interconnect Network
ML
Machine Learning
MOST
Media oriented systems transport
NCM
Navigation Control module
OBD
On-Board Diagnostics
OEM
Original Equipment Manufacturer
PEPS
Passive Entry Passive Start
RAT
Routine Activity Theory
RFID
Radio Frequency Identification
TCM
Transmission Control Module
UWB
Ultra-wide band
V2C
Vehicle to Cloud
V2I
V2R
Vehicle to Infrastructure
Vehicle to Road
V2IoT
Vehicle-to-IoT
V2V
Vehicle-to-Vehicle
V2X
Vehicle to Everything
VANET
Vehicular ad-hoc Network
VCM
Vehicle Control Module
VCS
Voice Controllable Systems
VVS
Vehicle Vision systems
2. Methodology
Literature was retrieved from online archives, including Web of Science, Google Scholar,
TRID, and SCOPUS using Boolean operators. Books, news articles, and relevant industry
reports were also used to supplement background research. Specific keywords that were used
include; “connected and autonomous vehicle/cars (s)”, “driverless”, “driverless vehicle/car
(s)”, and “automated vehicle/car (s)” in combination with i) cybersecurity-related terms-
“cybersecurity”, “hacking”, “attack(s)”, “hacker(s)”, “cyber-attack(s)”, ii) safety-related terms-
“safety”, “accident(s)”, “crash(es)”, “concerns”, “risk(s)”, and iii) privacy-related terms-
“privacy”, “data protection”, “surveillance”, “location”, and “tracking”. Cybersecurity for AV
vehicles got attention in the past decade, and therefore post-2010 literature is considered. The
literature search was further intensified by forward and backward snowballing in the related
papers with the intention of having a thorough analysis of the topic, putting together the content
and drawing some relevant conclusions for a robust CAV-cybersecurity framework.
3. State-of-the-Art
This section is structured into seven subsections, identifying six avenues for identifying,
tracking, and preventing cyber-attacks on the CAVs as shown in Fig.3. Literature synthesis
shows that stakeholders ought to work on these factors to ensure; i) seamless roll-out of CAVs,
ii) efficient mass implementation of the CAVs, and ii) a decrease in the risk of cyber-attack
vulnerabilities.
Fig.3. A holistic view of potential avenues essential in combating CAVs cyberattacks
(Source: Authors’ Synthesis)
The first potential avenue outlines the CAVs communication framework in an integrated
form, i.e., from IV communication to V2X communication with a visual flowchart to provide
a transparent picture of all the interfaces for potential cyber-attacks including CAV wireless
communication technologies. The second avenue demonstrates the susceptibility of CAVs to
adversaries by synthesising proximity (or physical) exposure to cyber-attacks. Then, it
describes the scope and importance of CAVs supply chain for the detection and eradication of
cyber-attacks. It is followed by avenue on how psychologists and human-factor researchers can
augment cybersecurity in CAVs by exploring behavioural models and learning from
criminology theory to reduce the risk of a successful attack. Moreover, there is a concise
overview of the regulatory regulations and policy framework avenue for fostering smooth
deployment and operation of the CAVs on highways. Furthermore, the significance of an
integrating mechanism avenue for ensuring acceptable levels of cybersecurity risks in the
CAVs is highlighted and emphasised. Finally, we illustrate research gaps for each potential
avenue, which needs further investigation.
3.1 CAVs Communication Framework
The key benefit of CAVs is the potential to interact with other vehicles, the smart road
technology network, and other internet-connected applications, thereby improving driving
behaviour and road safety (Ali et al., 2020). Several facets of autonomous transport rely on
these interactions, such as platooning, shared route management, and so on. As a result, there
has been a growing interest in the creation of reliable vehicle communication. At the same
time, the ubiquitous nature of communication components is making it a sweet point for white-
hat hackers. Based on the literature review, the leading automotive company reports, and the
study of relevant govt research bodies, CAVs communication framework is illustrated in Fig.4.
This figure presents the CAVs communication framework for all possible interfaces in the form
of a flow-chart. The rationale for presenting this is three-fold:
1) It is imperative to have a systematic understanding of the CAVs communication
framework.
2) It is beneficial for monitoring, assessing, tracking, and combating potential cyber-
attacks on various communication interfaces.
3) It will facilitate the development of a robust CAVs cybersecurity-by-design paradigm
by application developers.
CAVs communication framework can be classified into two main categories; i) In-Vehicle (IV)
communication, and ii) Vehicle to Everything (V2X) communication. Both these facets of
CAVs (Fig. 4) are explained and synthesised in detail in the following sub-sections.
3.1.1. In-Vehicle (IV) Communication
The IV communication as presented in Fig. 4 involves; a) data flow of sensors, b) intra-vehicle
communication, and c) coordination of ECUs. In Fig. 5 below, we present the relationship of
attack vectors in IV communication in an automated vehicle. As shown in Fig. 5, different
types of sensors (e.g., LIDAR, camera, Laser, etc.), Electronic control units (ECUs) (e.g.,
engine control module, transmission control module, etc.) and intra-vehicle communication
(e.g., Local Interconnect Network (LIN), Controller Area Network (CAN), Ethernet, etc.) need
to work together to guide and navigate the automated vehicle in the transport network. In the
next sub-sections, the components of IV communication and the cyber-attack vectors are
described in more detail.
Fig.4. CAVs communication framework (Source: Authors' synthesis)
Fig.5. Cyber-attack vectors in CAVs In-Vehicle (IV) Communication (Source: Authors'
synthesis).
A. Sensors’ Data Flow
The data flow of the sensors is the correspondence between the external physical environment
and the CAV-computer unit. Autonomous vehicles are mounted with a variety of sensors that
gather an understanding of the environment. Types of autonomous vehicle-mounted sensors
include LASER, RADAR, LiDAR, GPS, TPMS, Camera, Ultrasonic and Gyroscopic sensors,
all contributing to better vision and recognition algorithms (Levinson et al., 2011; Raiyn, 2018).
The input obtained from these sensors enhances the capacity of vehicles to navigate.
B. Intra-vehicle communication
Intra-vehicle communication consists of automotive bus systems, interaction through the On-
Board Diagnostics (OBD) device or dongle, and the Automotive Ethernet for Control
Automation Technology-EtherCAT. Premium segment automobiles would have 5 bus systems
which are; i) LIN (Local Interconnect Network): is a bus protocol used for low-cost
multiplexed connectivity in automobile networks designed for high latency and specialised
error management with a data rate of up to 10 Kbps, primarily used for electric seats, mirrors
and tailgates, ii) Controller Area Network (CAN): is the most widely used bus interface
allowing various components to connect with each with other having a speed of 1 Mbps.
Typical uses include ABS, power rail, and engine management (Krishnapriya et al., 2012), iii)
TTCAN and Flex-Ray: is CAN equivalent with a data rate of up to 10 Mbps, mostly used in
the areas of the engine, braking, suspension, acceleration, steering, and diagnostics, iv) MOST
(Media oriented systems transport): is designed for digital data transfer utilising optical fibre
cables or coaxial cables. It has faster transmission rates than LIN, CAN and Flex Ray with a
data rate of up to 23 Mbps. It is primarily used for video plays and car infotainment systems,
and v) Automotive Ethernet: its use is still minimal but will play a crucial position in the next
phase of automotive networks. The broad bandwidth is a valuable attribute for new automobiles
with trade between cost and weight.
C. ECUs coordination
The coordination of ECUs is the exchange of information for the control of a series of
actuators to ensure optimum CAV performance. ECUs are the embedded controller that
controls one or more of the automobile’s subsystems. Examples of different ECU units
according to their importance are Transmission Control Module (TCM), Engine Control
Module (ECM), Navigation Control module (NCM), Body Control Module (BCM), Vehicle
Control Module (VCM), and Vehicle Vision systems (VVS) (Al Zaabi et al., 2019).
3.1.2. Vehicle to Everything (V2X) Communication
The key aspect of the ITS is that CAVs have the communication capability to interact with
each other or with the other sections of the transport network. V2X communication can be
classified into Vehicle to Cloud (V2C) and Vehicle to Vehicle (V2V) communication. V2C
communication ranges from Vehicle to Infrastructure/Road (V2I/V2R) to Vehicle-to-IoT
(V2IoT) interactions. Moreover, the primary wireless communication technologies that will
enable the operation of CAVs are shown in Fig.4, and the CAVs V2X communication attack
vectors are depicted in Fig.6.
Fig.6. Cyber-attack vectors in CAVs Vehicle to Everything (V2X) Communication.
Various wireless access technologies used for V2X communications are:
i) Dedicated Short-Range Communication (DSRC)/VANETs: is a protocol stack that
allows low latency, secure and high-speed communication (Dey et al., 2016; Liu et
al., 2020).
ii) Cellular Network: The DSRC suffers from low bandwidth, so cellular networks are
deemed to be the successful candidates. 3GPP has developed LTE (4G) for V2X
communication and is currently focused on V2X communication integration into
5G-New Radio (Muhammad and Safdar, 2018; Khan et al., 2020). Innovation in the
telecommunications industry and the use of unmanned aerial vehicles have further
eased on-the-go connectivity in mountainous or challenging terrain areas for self-
driving cars, especially in the mmWave band (Khan, 2019; Khan et al., 2020).
iii) Zigbee: is a short-range networking protocol designed for low-speed
communication. In El-Rewini et al. (2019), the authors suggested the usage of
Zigbee within V2C communication, ADAS, and Forward Collision Alert System.
iv) Ultrawideband (UWB): is an another form of wireless networking and can transfer
large data speeds at low transmitting strength. UWB has been introduced in VANET
for accident avoidance and vehicle positioning systems, and defence to counter
relay attacks in Passive Entry Passive Start (PEPS) system (El-Rewini et al., 2019).
v) Wi-Fi, WiMAX: is also a potential contender for V2X communication referred to
as IEEE 802.16, low-latency standard, secure, and all-IP core network support
(Tanuja et al., 2015).
vi) Radio Frequency Identification (RFID): permits the identification of radio signals
using VANET applications including public transport passes and traffic systems
(Pawade et al., 2013; Vijaykumar and Elango, 2014; Hennessy, 2016; Al Zaabi et
al., 2019; El-Rewini et al., 2019).
vii) Blue-tooth: is mostly used to connect smartphones with automotive infotainment
and telematics systems for access to calls, music streaming, calendars, and car
diagnostics units (Onishi et al., 2017).
3.1.3. Orthodox CPS Cybersecurity Solutions are Insufficient in CAVs
Existing research suggests a number of cybersecurity approaches for the CAVs, most of
which are based on CPSs security frameworks. However, these orthodox cybersecurity
approaches cannot be used directly to counter cyber-attacks in the automotive industry; there
is a need for robust hardware and easy-to-use applications. The reasons why these
cybersecurity solutions are inadequate in CAVs are highlighted in Fig.7. First, CAVs would
remain in the field over a prolonged period relative to current IT networks, allowing white-
hackers ample opportunities to identify flaws in the deployed vehicles. Second, current
protection approaches are typically cost-effective and challenging to introduce in CAVs due to
resource and technological constraints in the ECUs, as well as their sensitivity to stressful
environments, such as low or high temperatures, vibrations, and electromagnetic interference
(Scalas and Giacinto, 2019). Third, several ECUs are needed to carry out real-time activities
that are often safety-critical. Furthermore, the presence of strict authentication constraints can
lead to the extraction and misuse of various private details, i.e. identity surveillance, driving
records, behavioural inferences, subscribed services, and mobility trends. Similarly, to protect
intellectual property, vendors often supply (software) components without source code; it may
be more challenging to change them to enhance security.
Fig.7. Reasons why orthodox CPS cybersecurity solutions are inadequate in the CAVs.
(Source: Authors’ synthesis)
3.1.4. Overview of Cyber-Attacks, their Impact on the CAVs Operation and Mitigation
Techniques.
CAVs communication is vulnerable to two types of attacks; i) active: when communication
is disrupted or replaced by fake messages, and ii) passive: when the intruder collects
information for a potential malicious intent in a long-term timeframe. In contrast to passive
attacks, active attacks are more challenging to protect but easier to detect. Table2 presents a
summary of cyber-attacks, their impact on the CAVs in ITS, relevant mitigation techniques,
communication layer details and recommendations for the future.
Voice Controllable Systems (VCS) are susceptible to hidden voice instructions that are
ignored or unintelligible to humans. Depending on the target level, these threats may be
categorised as inaudible voice commands (attacking the voice capture point) and audio-
adversarial cases (attacking the speech recognition stage) (Zhou et al., 2019). For example,
malicious voice commands encoded in the sound of online videos can sneakily control the
vehicle as people watch these videos in the car (Zhou et al., 2019). A spoofing attack is when
the intruder uses a false identity or sends out fake data (Linkov et al., 2019). This would impact
on shared route management (platooning) of CAVs in two ways; i) pretend to be a (fake)
neighbour CAV, or ii) send false information about a neighbour’s CAV location. Man-in-the-
middle attacks/passive attacks/relay attacks are where the adversary sends the original message
to the CAV, updates it and sends a new message to the vehicle that causes an incorrect message
switch between the CAVs communication (He et al., 2017). Eavesdropping, disclosure of
information, and traffic-data analysis are the consequence of such assaults. The jamming attack
is where radio signal noise interrupts the frequency of communication, intended to obstruct or
hinder V2C communication (Parkinson et al., 2017). In Czerwinski et al. (2019), the author
emphasises the impact of the radio power level of the ZigBee network module on the use of
energy and performance in the CAVs communication under jamming conditions. A
masquerade attack is an intrusion that uses a false identity, such as a network identity to obtain
unauthorised access to the CAVs without valid access authentication. The authors in Choi et
al. (2018) and Liu et al. (2017) described two CAN vulnerabilities that allow masquerade
attacks. An eavesdropping attack is known as a sniffing or snooping attack, is an incursion
where white-hat hackers attempt to intercept information that CAVs send over a network. The
black-hole attack is where the message is blocked without the CAV being aware of the missing
message (Carsten et al., 2015). The impact of this attack is blocking communication to the
target CAV, blocking or disabling the response of the receiver-CAV.
Similarly, in a replay attack, the perpetrators continually forward a valid frame to prevent the
CAV from working in real-time Liu et al. (2017). In bus-off attacks, intruder constantly sends
bits both in the ID field and in other fields, which allows the ECU transmission error counter
to be raised, if the value of the TEC is higher than 255, the resulting CAV-ECU would be shut
down and will result in halting the CAV operation (Choi et al., 2018). Denial of Service (DoS)
attacks happen when adversaries consistently send high priority messages that circumvent
genuine low priority messages. DoS attacks may be used as a way of monitoring override
attacks that enable attackers to gain control of the CAV (Carsten et al., 2015). In message
spoofing attacks, intruders send unauthorised messages containing false information to
interrupt vehicle communication (El-Rewini et al., 2019). Injection attacks happen when
attackers are inserting unauthorised and harmful messages inside the in-vehicle network. In
timing attacks, a deceptive CAV gets a notification, introduces a time delay, and then transfers
the notification to other CAVs causing incorrect scheduling of instructions and fake road
congestion. This intrusion may be catastrophic to transportation networks that rely on real-time
applications (Sumra et al., 2011). Impersonation attacks are conducted by creating a CAV with
a fake identity (Amirtahmasebi and Jalalinia, 2010). Impersonation is harmful to the integrity
of the overall transportation network infrastructure and is especially destructive in the case of
a crash, as the CAV under review is untraceable and misleading to other CAVs. The
alteration/replay attack happens when an intruder utilises previously created frames to submit
and interact with other nodes (CAVs), with or without modification causing incorrect location
and speed value or wrong route adaption (Parno and Perrig, 2005).
Cyber-attacks don't often end in accidents, but they do cause broader oscillations (Cui et al.,
2018). The slight attack is the case where the reported CAV data randomly deviate from the
real data, and the deviations do not exceed the threshold-making the difference less noticeable.
The authors in Li et al. (2018) indicated three findings; a) when one CAV is under slight cyber-
attacks, it becomes riskier if the locations transmitted are targeted than the speeds, b) if multi-
CAVs are under attack, the scenario of more CAVs under low severity attack is expected to be
more hazardous than those of fewer CAVs under high-intensity attack, and c) the effect of
slight cyber-attacks during the deceleration phase is more severe than the acceleration phase.
In a spamming attack, the transmission latency in VANET is increased, causing a severe delay
in V2X communication. During a tunnelling attack, two parts of the network are linked by an
attacker using an external communication channel that may result from eavesdrop on V2C
communication to halting of CAVs real-time operation. CAVs cameras are also vantage points
for white-hat aggressors. For example, attackers can hide or remove traffic signal photos at key
locations or attach lines to the path to avoid track identification. Similarly, GPS signals are
weak and susceptible to remote malicious intrusion. GPS jamming attacks are usually
considered to be the deadliest attack, there is a high probability of an accident when the GPS
of the CAV is under attack (Cui et al., 2018). Furthermore, the attack on LiDAR is an expansion
of the replay attack, and the intention is to transmit the original signal from the LiDAR device
of the target vehicle from another location to produce false echoes impacting autonomous
navigation (Nayegandhi, 2007).
Table2. Summary of cyber-attacks, its impact on the CAVs and mitigation techniques.
Attack Type
Communication
System
Attack
Vehicles
Attack Impact
Counterstrategies and
Potential Mitigation
Techniques
Communi
cation
Layer
Future Work
Hidden voice command/
dolphin attack/
commander song
(Greenberg, 2015; Zhang
et al., 2017; BMW-Group,
VCS, IV
communication
BMW,
GM,
Chevy
Impala
Infotainment
malfunction,
Vehicle driving
control.
Pop-noise-based general
defence strategy,
audio turbulence and
audio squeezing.
Network
Induction of Machine
Learning (ML) and Deep
Learning (DL) algorithms to
combat such attacks.
2018; Roy et al., 2018;
Yuan et al., 2018; Zhou et
al., 2019)
Sibel/
Spoofing/
Botnet attacks
(Wesson et al., 2011; Yu
et al., 2013; Anouar et al.,
2016; Solon, 2016; Van
der Heijden et al., 2016;
Muciaccia and Passaro,
2017; Nie et al., 2017; He
and Chow, 2019)
GPS,
IV
communication
Tesla
Model S
P85,
Model
75D,
Jeep
Cherokee
Incorrect
location and
speed value,
wrong route
adaption.
Speed-deviation such as
acceptance range
thresholds, Intrusion
Detection System (IDS),
correlating messages
from neighbours.
Using a reputation-
based mechanism.
Network
Transport
Blockchain is an evolving
technology and has the
potential to circumvent the
security challenges of
existing VANETs and can
help to combat these cyber-
attacks. Its scope and
utilization in CAV
communication are still in the
early stages.
RF Jamming attacks
(Azogu et al., 2013; Puñal
et al., 2014; Parkinson et
al., 2017; Lyamin et al.,
2019)
Over-the-air
communication,
LiDAR,
autonomous
navigation
Performance of
platooning,
in-accurate
communication,
object detection
failure.
ML-based jamming
detection,
anti-jamming technique
for VANET metrics-
directed security
defence.
Jamming DoS detection
using data mining,
intrusion, and network
coding.
Physical
Network
Need for fast-detection and
fast-reacting anti-jamming
mechanisms in CAV (non-
static) networks.
The implementation of
various techniques such as
artificial intelligence, mobile
agent, game theory,
consistency check, cross-,
spatial retreat, and frequency
hopping needs to be assessed
in this context.
Slight/
Greyhole attacks
(Miller and Valasek,
2015; Verma et al., 2015;
Liu et al., 2017; BMW-
Group, 2018; Li et al.,
2018)
V2V, V2C
2014 Jeep
Cherokee,
BMW
Cars
Longitudinal
safety,
traffic
congestion.
Cyber-attacks should be
considered for the entire
CAV fleet rather than
for individual’s CAV.
Physical
Network
Designing of dynamic
defence mechanisms for
CAVs platooning-as an entity
of integrated ITS.
Passive Attacks/ Man-in-
the-middle attacks/
Relay attacks
(Francillon et al., 2011;
Weiß, 2011; Gagandeep et
al., 2012; Nissan, 2016;
He et al., 2017; BMW-
Group, 2018;
TencentKeen-
SecurityLab, 2019).
V2V, V2C
Tesla
model S
75,
Nissan
leaf
Eavesdropping,
disclosure of
information,
traffic-data
analysis.
Encryption techniques
using public keys in
V2X communications.
Network
Transport
Developing and configuration
of CAVs communication
networks to encrypt not only
data but also information and
electromagnetic shielding.
Replay /
Playback attacks
(He et al., 2017;
TencentKeen-
SecurityLab, 2019)
V2V, V2C
Tesla
Model S
75
Impact on real-
time CAV-
operation.
Tagging each encrypted
component with a
session ID and a
component number.
Network
Designing and induction of
bio-metric identification
systems, i.e. fingerprints, iris,
speech, face, retinal
recognition and hand
morphology, for swift and
effective identification
purposes.
DoS/
Message
Delaying/
Blackhole attacks
(Hasbullah, 2010;
Hortelano et al., 2010;
Raiyn, 2013; Bergin,
2015; He et al., 2017)
Automotive Bus
System,
Over-the-air
communication
blocking
communication
to the target
CAV, blocking
or disabling the
response of the
receiver.
Cryptographic
techniques such as MAC
and digital signatures to
secure information.
Physical
Network
The CAV programme needs
to be reviewed and tested
regularly,
protocols need to be
constantly reinforced,
induction of firewall set up,
watchdogs and data-based
malicious behaviour
detection.
Modification/
Injection attacks
(Koscher et al., 2010;
Weiß, 2011; Foster et al.,
2015; He et al., 2017;
Raiyn, 2018)
V2V, V2C
2009
Chevy
Malibu
Modify
messages e.g.
GPS details on
the
communication
channel.
Use of public keys as
encryption: which uses
biometric data for
message authentication
and secured
communication-based
on iris recognition.
Network
Transport
Machine learning method to
filter the data, analyse it, and
cross-check it with other
input parameters.
Timing-faking attacks
(Sumra et al., 2011; Taylor
et al., 2015; Cho and Shin,
2016)
Over-the-air
communication
Incorrect
judgements by
the Road-Side
Units (RSUs)
directing the
CAVs to follow
sub-optimal
paths, i.e. high-
traffic or crash
paths.
Timing-based CAN-Bus
anomaly detectors.
Physical
layer
Clock-based ECU
fingerprinting,
fixed processing time
algorithms.
Tunnelling attacks
(Rawat et al., 2012; Zhang
et al., 2014; Xu et al.,
2018; Shrestha et al.,
2019; Zhang et al., 2019).
V2V, V2C
Modification of
packets,
delaying
communication.
Split Horizon DNS
concept,
cloud-based on-board
malware defence
manager.
Network
Transport
Developing CAVs
communication networks to
encrypt not only data but also
information and
electromagnetic shielding.
Impersonation attacks
(Amirtahmasebi and
Jalalinia, 2010)
V2V, V2C
Vehicle under
investigation
becomes un-
trackable.
Digital signatures
without certificates.
Network
Designing a two-layer (cross-
verification) approach for
decision-making by the CAV.
For example, one input from
CAV-generated data and
another input from
Infrastructure-generated data.
Cyber-attacks on sensor
networks
(Rouf et al., 2010;
TencentKeen-
SecurityLab, 2019)
Autonomous
navigation,
IV
communication
Tesla
Model S
75
Providing
inaccurate
inputs to the
CAV-ECUs,
low network
performance.
Used of encryption and
cryptography
techniques,
updated firmware, and
use of a VPNs.
Physical
Using a non-standard network
and diagnostic ports.
Developing a sensor
tampering detector so that the
functionality and sensitivity
of the CAV-sensors can be
checked and notified
immediately.
Attacks on GPS
(Spoofing and Jamming)
(Rouf et al., 2010; Wesson
et al., 2011; Lopez et al.,
Autonomous
navigation,
IV
communication
Incorrect
location and
speed value,
i) Cryptography: use of
authentication and
encryption procedures,
and
Physical
Improvements to the
traditional firmware and
applications running on the
underlying ECUs.
2019; TencentKeen-
SecurityLab, 2019)
wrong route
adaption.
ii) Signal check:
Utilizing distortions of
correlation function in
the receiver.
Attacks on LiDAR
(Petit et al., 2014; Petit et
al., 2015; Shin et al., 2017)
Autonomous
navigation,
IV
communication
Deception with
toxic signs,
failed pedestrian
detection.
Reduced range of
LiDAR connections,
shortening pulsing time
interval, and
pulsing laser multiple
times.
Physical
The impact of attacks may be
mitigated by reducing the
receiving angle.
Using multiple LiDAR input
as cross-verification.
Attacks on camera
(Petit et al., 2014; Petit et
al., 2015; Raiyn, 2018;
Decisions, 2019; Lopez et
al., 2019; TencentKeen-
SecurityLab, 2019)
Traffic sign
recognition,
Lane obstacle
detection,
Hide or remove
traffic signal
photos at key
locations,
avoid track
identification.
Filtering options, using
photochromic lenses.
Physical
layer
Use of machine-readable road
signs (e.g. UV QR overlay or
PCM-modulated light
signals).
Attacks on TPMS
(Rouf et al., 2010)
IV
communication
Proving
inaccurate
inputs to the
ECUs,
gaining entrance
via exploitable
inputs.
IDS with updated
firmware,
prevention of spoofed
activation.
Physical
Development of the sensor-
tampering detectors.
Attacks on RADAR
(Yan et al., 2016)
IV
communication
Tesla
Model S
Failed
adaptative cruise
control,
failed pedestrian
detection.
ML-based jamming
detection,
anti-jamming strategy
for VANET metrics-
directed security
defence.
Physical
Using multiple RADARs for
cross-checking and validation
of input data with robust and
efficient models.
Attacks on Ultrasonic
sensors
(Zhang et al., 2017; Yuan
et al., 2018; Zhou et al.,
2019)
IV
communication
Autonomous
navigation,
parking
assistance
Pop-noise-based general
defence strategy, audio
turbulence and audio
squeezing.
Physical
Development of the sensor-
tampering detector,
using multiple sensors, cross-
validation of input data.
Existing research suggests different classification requirements for categorising CAVs cyber-
attacks. For example, in Mejri et al. (2014), the authors classified CAVs cyber-attacks into five
categories depending on the related cryptographic classification, i.e. attacks on authenticity,
availability, confidentiality, non-repudiation, and data trust. The authors in Cui et al. (2019)
divided cyber-attack on the vehicles into four types; availability, data integrity, authenticity,
and privacy based on security requirements concerning the CAVs. In Amoozadeh et al. (2015),
such attacks are graded into the network layer, application layer, privacy leakage attacks, and
system-level. Likewise, in Wang et al. (2020) CAVs cyber-attacks are fragmented into the
group of three; bogus messages, collusion attacks, and replay/delay based on; i) cyber-attacks
characteristics, and ii) attacking principles.
The severity of cyber-attacks on the CAVs cannot be explicitly categorised in prior as critical
or minor because it depends on; i) the degree of interruption that white-hat hackers want, ii)
the privileges they have access to, and iii) the CAV-driving scenario, i.e., vehicle platooning
etc. The classification of such attacks under different security performance evaluation metrics
is shown in Fig.8. Visualising the role of CAVs in ITS and its operational framework, cyber-
attacks that impact the availability and authentication of CAVs are usually of extreme severity
(Lopez et al., 2019; Möller and Haas, 2019). Attacks on the reliability, robustness and integrity
are typical of mild magnitude. Similarly, attacks impacting confidentiality and trustworthiness
can be graded as low severity.
Fig.8. CAVs cyber-attacks classification under security performance evaluation metrics
(Source: Authors' synthesis).
3.1.5. Counterstrategies and Potential Mitigation Techniques
CAVs will rely on both sensor data and mapping data for real-time operation. Sensor data is
the one collected from sensors, i.e., LiDAR, RADAR, camera etc. Mapping data can be of two
types; i) static-terrain details, and ii) dynamic-real-time updates from communication
infrastructure (Liu et al., 2017). For secure CAVs communication, stringent low-latency
authentication techniques are required. CAVs and relevant service platforms should be
authenticated by one another or by a reputable third-party. Biometrics authentication can be
introduced, which is safe and reliable, i.e. passwords cannot be lost or forgotten (Raiyn, 2018).
Commonly used biometric authentication are fingerprints, iris, speech, face, retina
recognition, and hand geometry to secure data CAVs communication. Similarly, encryption
techniques using a public key may be used in V2X communications (Weiß, 2011).
Cryptographic techniques such as MAC and digital signatures may secure information against
spoofing and forgeries. Blockchain is an evolving technology and has the potential to
circumvent the security challenges of existing VANETs and can facilitate to combat cyber-
attacks. Its scope and utilization of in-vehicle communication are still in the early stages
(Shrestha et al., 2019; Zhang et al., 2019). The computing resources required for the CAVs are
significantly higher to support the processing of large quantities of sensors data. In effect, this
includes improvements to the traditional firmware and applications running on the underlying
ECUs (Lopez et al., 2019).
3.2 Physical Access Attacks
Physical access to the CAVs can provide hackers with a multitude of entry points, but
physical attacks are generally not very flexible, hard to be untraceable, and difficult to treat
frivolously. Adversaries may have access to autonomous vehicles by either physical access or
proximity access. Similarly, the operation of CAVs can be halted by vehicle diagnostic ports
or by illusion attacks, depicted in Fig.9.
Potential unauthorised access attacks are categorised in i) obvious interference, and ii) sensor
manipulation. Typically, sensors are the direct security risks to the CAVs, adversaries can; i)
produce incorrect messages; ii) obstruct sensor data; or iii) intervene with autonomous driving
by hacking into the CAV computing device (Liu et al., 2019). In Becher et al. (2006), the
authors analysed various physical attacks against sensor node hardware, i.e. node captures,
Joint Test Action Group (JTAG ) attacks, bootstrap loader attacks, and external flash attacks,
though, such kinds of assaults require specialised skills and costly hardware. Proximity attacks
can be classified as inaudible voice commands and PKES. Inaudible voice commands required
the physical presence of an attack device to transmit synthetic ultrasound signals. PKES is used
to unlock the door or to start the engine simply by holding the key in its pocket vulnerable to
relay attacks, with hardware available just under $100 (Choi et al., 2018). Similarly, an attacker
will tamper with on-board sensors and other hardware to alter the perceived position, direction,
and distance (Petit et al., 2015; Wang et al., 2020).
Physical inputs and outputs contained within a CAV include ports such as diagnostic port
(OBD-II), USB port, infotainment interactions, EV charging, and other open ports. Exploiting
these ports is normally more difficult for an attacker because they will generally require
physical access to the vehicle but due to the availability of additional devices that connect to
these ports, there are several ways in which attacks can be carried out through a remote link.
The diagnostic port directly connects via CAN to multiple on-board computers; an intruder
with physical access to the car can trigger attacks and compromise sensitive vehicle systems
(Lopez et al., 2019). There is an opportunity for an attacker to reach the vehicle's internal
network in the presence of the USB port. CAVs can also be targeted by placing an infected CD
Fig.9. Physical access attacks (Source: Authors’ synthesis)
on a CD player that may execute malware automatically (Linkov et al., 2019). Entertainment
systems and CAN bus networking to upgrade ECU software in CAVs could be the victim of
hackers as well. CAVs can also be targeted when charging using an electrical adapter that
attaches and records data. Furthermore, some cars have open ports where the intruder can
quietly connect to the D-Bus without further authentication (Lopez et al., 2019). Moreover,
illusion attacks could be in the forms of drones that projected pictures of fake machine-readable
road signs (e.g. UV QR overlay or PCM-modulated light signals) for an instant, i.e., 100ms-
too short for human sight, but long enough for autopilot sensors to be deceived (Gurion, 2019).
3.2.1. Defence Strategies
Physical access to a CAV communication bus may be like root access to the system, so both
physical and cyber risks must be addressed. However, as the priority for vehicle safety has
increased in recent years, efforts are being made to prevent physical attacks. An external sensor
tampering detection system is needed to track the sensitivity of the sensors on the CAVs. It's
also effective to use vehicle alerts to detect and curtail proximity access attacks. A sound-based
proximity-detection method, pop-noise-based general defence strategy, audio turbulence and
audio squeezing can prevent relay attacks on PKES systems and combat inaudible voice
commands (Greenberg, 2015; Zhang et al., 2017; BMW-Group, 2018; Roy et al., 2018; Tang
et al., 2018; Yuan et al., 2018; Zhou et al., 2019). In Markham and Chernoguzov (2017), the
authors proposed a role-oriented access control scheme to tackle vulnerabilities caused through
diagnostic ports, i.e., any commercial OBD-II device would be certified by the OEM and would
supply the vehicle with X.509 certificate and a public key to show its identity, and access will
be granted accordingly with the relevant privileges. Uncertified gadgets would only be allowed
to read the bus, while a licenced mechanic scanning tool would have both the read and write
permissions. Similarly, cyber protection techniques will be used to counter information
security risks. In addition, simple physical access points to a CAV, such as the OBD-II port,
can be moved to a non-standard position. Furthermore, mitigation techniques for the threats to
EV charging unit are; i) incorporation of secure encryption in the controller boards of EV
charging unit, i.e., for both board-to-board communication and flash memory, ii) tampering
alarm induction and activation for any breach, and iii) using secure coding standards.
3.3 CAVs Supply Chain
CAVs consist of complex digital systems and subsystems developed by several
geographically distributed vendors. The increasing complexity of the software system and the
dynamically interconnected IT subsystems have paved the way for the advent of new vendors
within the CAVs supply chain. New entrants offer services, especially in the field of
engineering and design, rather than physical products (Morris and Madzudzo, 2018). Two main
issues concerning the supply chain of CAVs are as follows:
1- Cybersecurity is deemed a secondary task in most automotive business models. It offers
minimal incentives for monetization and value generation within a highly profit-driven
operating framework. Potential suppliers in this supply chain can incorporate
confidential information and obtain access to assets produced by the other players. At
the same time, the manufacturer would want to secure the product from other players
in the supply chain. Such security leaks could result in the malfunctioning of CAVs or
the espionage of users and even access to OEM confidential information as well.
2- Device manufacturers build systems and components based on the technical
requirements and performance criteria given by the automotive OEMs, but often
product manufacturers make design choices without OEM input. The relationship
between the limitations of technology information and the inefficiency of many
cybersecurity techniques results in an overuse of intuition, a dependence on static and
generic knowledge, and a lack of governance of cyber presence (Julisch, 2013).
The prerequisite commitment of CAVs OEMs/vendors/suppliers to impartial security testing
will be very beneficial. Similarly, log files must be preserved and secured for a given period of
time for all CAVs operations including physical access and anti-malware updates etc. In
addition, OEMs/vendors/suppliers must fulfil audit requests and initiatives such as incident
response, penetration testing and vulnerability scanning etc.
3.4 Human Factor in the Safety of CAVs and Use of the Criminological Theory
CAVs cybersecurity is an evolving field of traffic safety. Since human weakness is the most
likely explanation for a successful cyber-attack, psychologists and human-factor researchers
can augment cybersecurity in CAVs by exploring how to reduce the risk of a successful attack.
A major consultancy firm expressed it as “Cybersecurity is not only about technology but
rather about psychology and sociology. It's convenient for developers to think that the most
appropriate approach is the one with the most blinking lights, but in the field of cybersecurity,
it's mostly people's actions that decide the result (PWC, 2014)”.
Similarly, mixed traffic flow would be very normal in the near future, and the driver's reaction
to the lead vehicle relies on their conditional perception of CAV technology rather than on the
real driving behaviour. This suggests that in the immediate future, classic car-following
behaviour in pure human vehicle traffic will need to be revised to model mixed traffic, based
on the CAVs characteristics and layout (Zhao et al., 2020). Only the penetration of 100 per
cent CAVs will contribute to the maximum protection (Sinha et al., 2020). The main human
factors that need to be investigated in the effective mass adoption of CAVs and to minimise
the risk of human-related failures are discussed below (Linkov et al., 2019):
Characteristics of people prone to cybersecurity failures–people vary in their ability to
accurately identify cybersecurity risks. The research conducted in Katerina and Nicolaos
(2018) found that 23% of people correctly manage fewer than half of the cybersecurity
scenarios, and only 4% are able to handle more than 90% of situations. Internet users are likely
to be more aggressive towards cybersecurity if they are more extraverted, less attentive, and
addicted to the internet (Hadlington, 2017). Those who use computers more frequently for non-
work purposes have little understanding of internet security. Anxious people are less effective
in identifying cyber-threats (Welk et al., 2015). Generally, men have more knowledge about
cybersecurity than women (Anwar et al., 2017). Risky cybersecurity attitude is related to the
over-trust of automated technology (Noy et al., 2018). Similarly, older drivers and female
drivers benefit maximally from the CAV communication facilities compared with the young
drivers (Ali et al., 2020; Sharma et al., 2020). Nevertheless, the features of individuals with
riskier conduct towards cyber-CAVs are still unclear.
Human behaviour–during cyber-attacks is an integral part of research to inform people about
cyber threats in the CAVs. The level of multitasking that the driver indulges in can have an
impact on the effectiveness of the response to cybersecurity breaches. Based on a study in this
area, when CAV drivers have to respond to unusual incidents, they have a wide range of
response times and their ability to react varies (Brandts et al., 2016; Dixit et al., 2016; Dogan
et al., 2017). Similarly, due to ubiquitous CAVs communication and prior information
availability about anticipated incidents, for example, informing in advance about pedestrians
approaching from the pavement would make it safer for drivers to be more willing to give way
and retain a higher margin of safety (Ali et al., 2020).
Motivations and characteristics of attackers–Knowledge of attackers' intentions and
behaviours can help to deter potential CAVs attacks. In King et al. (2018), the authors found
that attackers may be characterised by low socioeconomic status, socialisation towards law-
breaking behaviour, hyperactivity, and dark triad personality traits. Such characteristics can
help to establish methods for the identification of attackers (Holt et al., 2010; King et al., 2018).
Criminological theory for mitigating and preventing cybersecurity threats–the authors in
Kennedy et al. (2019) proposed criminological theory, i.e., Routine Activity Theory (RAT)
for mitigating and preventing cybersecurity threats. Based on the discussion and in-depth
literature review, the authors evaluated the use of RAT as a suitable framework for preventing
and mitigating cybersecurity risks.
3.5 Regulatory Laws and Policy Framework
Cities need to start planning for this modern mode of travel, which is evolving rationally.
Regulatory bodies need to develop, review plans and policies for the anticipated development
of CAVs and implement the measures necessary to ensure an efficient and sustainable transport
network. Various factors of regulatory laws and policy framework need to be considered, such
as data confidentiality and privacy, short-term economic considerations, operational safety,
public acceptance, legal and ethical issues (Seuwou et al., 2020). In Taeihagh and Lim (2019),
the authors studied and analysed government approaches taken globally in response to CAVs
advancement. Based on their analysis, in most situations, state policymakers have resisted strict
legislation to encourage innovation in the CAVs, relying only on the formation of committees
or focused groups for further exploration. The implications are lack of clarification as to
whether CAV passengers, CAV OEMs or other third parties are to be held liable in an accident.
Another main concern is the obligations of service providing operators, which requires the
mandate for operators to store CAVs related personal data inside a country and its availability
to the overseas business partners (He, 2018). Although, China, Singapore and the US have
adopted cybersecurity regulations, that are not distinct to the CAVs (Taeihagh and Lim, 2019).
Australia and Germany are also becoming conscious of the threats involved with the CAV's
cybersecurity, while the UK has plans to utilise cybersecurity hazards as an incentive to boost
the competitive ability of the country (Cabinet Office, 2016; Lim and Taeihagh, 2018). CAVs
and connected infrastructure require a continuous surveillance system to alert the relevant
operation centre immediately about any data or vehicle breaches, including unauthorised
access, control or operation (Hodge et al., 2019).
3.6 Integrated Management Framework for CAVs Cybersecurity
This section underlines the importance of addressing safety and security standards for the roll-
out and operation of CAVs in an integrated manner. CAVs telematics network would be the
most appropriate example. The key components of the telematics are; i) subscriber
identification module (SIM) cards, ii) input/output controllers, iii) engine interfaces, iv)
accelerometers, and v) GPS receivers. Telematics is typically connected to external fleet
operating systems of third parties, such as communications infrastructure and service database
(Hodge et al., 2019). This establishes an external link to one or more CAV in a fleet. Since
telematics connects with vehicles to gather data from the CAN and other communication
systems; vulnerabilities in the CAV sensing capability may arise if the external network or
physical telematics interface is compromised (Li et al., 2019; Wang et al., 2019). Although, a
potential approach will be to give a unique cryptographical key to each telematics unit;
however, to make it pragmatic, an integrated management co-operation framework is required.
Developments involved in the implementation of the CAVs are occurring primarily in the
automotive, electronics and telecommunications industries. Value-creation of networks and
closing information gaps between stakeholders are essential as well as the development of a
shared vision for the entire value chain. Another concern is that most of the established
processes are quality-oriented, with only a few safety-oriented, and no security-oriented.
Moreover, there is no complete mapping of these standards. Therefore, it is of utmost
importance to have an integration mechanism to ensure acceptable levels of cybersecurity risks.
4. Research Gaps and Future Directions
Due to the high degree of non-linearity and variation of CAV cyber systems, risks cannot be
asserted or calculated in conventional risk theory terms. Most publicised work is reactive, and
vulnerabilities are typically undiscovered by white-hat hackers, hobbyists, and researchers.
Cybersecurity challenges are on the rise, and cybercriminals are penetrating quickly. In this
section, we spotlight a few promising areas which need further investigation.
4.1. Robust Defensive Techniques to Identify, Deter, and Prevent CAVs Cyber-
attacks
V2C communication will be crucial for the successful operation of CAVs, but at the same
time, it will be vulnerable to cyber-attacks on various interfaces and will require a robust ML
and DL algorithms to identify, deter, and prevent cyber threats (Liang et al., 2018; Ye et al.,
2018). While the essential elements are not modelling or algorithms, they are the abundance of
data. A recent study reported that a CAV would have to travel up to hundreds of billions of
miles across several scenarios to ensure statistical reliability relative to human drivers (Kalra
and Paddock, 2016). Two main areas for further improvement of the attack surface analysis
are; i) the integration of the system analysis, and ii) the understanding of rapidly evolving
threats in various environments and scenarios.
The vulnerability of vehicle infotainment systems using the internet is an open field for in-
depth research and review. The safety of V2C communication has not been investigated in
detail and, unfortunately, security is not currently viewed by the automotive industry as a
priority for V2C development. Because V2C technology is in its infancy, it would be
appropriate for researchers to concentrate on strategies for testing the security of vehicle
communication systems. CAVs safety is endangered if safety is compromised at any level. The
task, though, is to ensure protection at all stack stages (Wang et al., 2019). Therefore, the
drafting of industrial safety authorization standards will be very beneficial in this regard (Liu
et al., 2019).
Furthermore, we visualised the aggregate degree of keyword frequency in VOSviewer
application for around 40 research articles (of the last 5 years), specifically related to the CAVs
cyber safety, depicted in Fig.10. Using VOSviewer app, users can visualise the main topics or
thematic clusters, the interrelationships between the topics and the occurrence of those fields
of study through infographics (Shaharudin et al., 2019). Visualising for the seven metrics of
security performance assessment in Fig.10, it is apparent that some of the popular thematic
clusters have been “security”, “attack”, “communication”, “data”, “information” and
“networks”. Small cluster on availability, reliability, robustness, and confidentiality in CAVs
cyber-attacks assessment are observed suggesting relatively lesser work in this area.
Nevertheless, cyber-attacks which fall under the categories of integrity, reliability and
trustworthiness needs further rigorous analysis, assessment, and development of counter-
mitigation strategies. A formal, systemized, dynamic, and multi-layered (cross-verification)
approach to address these metrics based cyber-attacks will boost the CAVs safety.
Fig. 10. Network visualization of keywords theme occurrences
4.2. Combating Physical/Proximity Access Attacks
Security and safety are processes, system designers need to stay up to date with the advances
in attacks on the CAV-embedded system. First, an illusion attack or deceiving with a false
traffic signal is an open field for study so that CAVs can operate smoothly without slowing
down or limiting overall efficiency, especially at the intersections and multi-lane roads (Ali et
al., 2019). Efficient detection mechanisms and algorithms are required to differentiate between
genuine and fake or counterfeit signs. Machine-readable road signs (e.g. UV QR overlay or
PCM-light signals) with electronic signatures would make it secure in this context. Second,
sensor tampering detectors required in-depth study and analysis, i.e., functionality and
sensitivity of CAV-sensors need to be verified and notified instantaneously prior to CAV start-
up. Development of anti-tampering devices in such a way that the physical access to the device
does not opt for partial or full CAV network access. Besides, the safety assessment of the EV
during charging is another area for further study, considering the aspects of grid protection, the
safety of charging facilities, and security of communication.
4.3. Combating Adversarial ML Attacks on CAVs
ML and AI strategies are central ingredients in the success of several primary CAV functions.
Nevertheless, ML/AI systems are susceptible to subtle adversarial perturbations that are
deliberately designed. These perturbations could occur in voice (audio), vision (image), text
and networking. The current defence mechanisms are ineffective in tracking and countering
these perturbations. It has been observed that these defence mechanisms are only successful
for a particular form of attack and fails for more vigorous or unknown assaults. It is therefore
vital to establish new ML/AI protection strategies that are successful and productive, especially
the auto-navigation framework of CAVs and preserving privacy as well.
4.4. CAVs Infotainment System: Ubiquitous Connectivity
Infotainment systems in vehicles present the greatest potential of attacks on vehicle networks.
OEM telematics service providers allow multiple infotainment systems to reach and
communicate with the Internet, leading to vulnerabilities in CAV interconnections (Li et al.,
2019). CAV infotainment systems compile and show details on the current state of various
functionalities; however, infotainment device communication setup varies from supplier to
vendor, model to model, and year to year (Hodge et al., 2019). Therefore, designing and
countering specific attacks and generalising them is more challenging. This necessitates for the
design of low latency two-fold mechanism; i) ensuring infotainment systems operate on a
segregated communication network (other than the operational and safety network), i.e.,
ceasing IV bridged communication, and ii) configuring stringent user’s access and augmenting
digital signatures for mobile apps connected to the CAVs.
4.5. CAVs Supply Chain Trust Provisioning Schemes
There is no existing mechanism for a formal engagement between road operators and
suppliers, i.e. OEMs, data aggregators and suppliers to discuss the possibilities for addressing
supply chain vulnerabilities; all present encounters are on an ad-hoc basis. Tech companies
such as Google and Apple have made their claims on the driving markets of CAVs.
Nonetheless, no one can win if security issues weaken consumer trust in CAVs. To alleviate
these concerns, manufacturers need to integrate security into every part of their designs.
Trust provisioning schemes-The automotive industry is developing confidence-provisioning
systems to tackle this issue at the architectural level; the notion is that assets are received from
different stakeholders through a common, unified trust model. The trust model is usually
established by the component manufacturer, which enables different stakeholders to inject
assets at different stages. Blockchain security techniques and powerful AI approaches will
make it easier to confront these difficulties. Each CAV and any outside communicator need to
have a specific digital signature. Nevertheless, it is always the OEMs who are accountable for
the reliability, quality, and safety of cars brandishing their logo.
4.6. Analysis of Human Behaviour Concerning the Safety of CAVs
Human factor researchers aim to recognise the most susceptible individuals in the CAVs
cybersecurity, identify the types of scenarios they are struggling with, and develop tailored
educational materials. Similarly, the drop in driving skills due to the use of CAVs and its
relation to cybersecurity skills is also worth researching (Gkartzonikas and Gkritza, 2019;
Linkov et al., 2019). Human-factor experts are conscious that improvements to CAVs
cybersecurity may not necessarily maximise traffic safety. In Macher (2017), the authors
describe a scenario in which the steering wheel is blocked in a risky condition. This is safe
from a cybersecurity point of view because the intruder cannot alter the direction of the steering
wheel. However, it is not safe from a traffic perspective because the driver cannot react
appropriately when the steering wheel is blocked. Investigators should consider these types of
situations, i.e. where cybersecurity threats might be overestimated. Research on human
behaviour concerning CAVs cybersecurity will be more appropriate in the future when people
use CAVs regularly. Dedicated CAVs lanes are yet another future approach, but their impact
on traffic safety and efficiency is still to be explored while taking into account two key factors;
i) driver behaviour adaptation, and ii) manual and automated transitions in operational control
(Rad et al., 2020).
4.7. Legal Readiness in terms of Policy, Regulation, and Liability of the CAVs
It has been witnessed that focusing solely on the market players for the production and
distribution of CAVs is an ineffective solution in resolving the increasing death and accident
tolls on our highways (Geels and Penna, 2015; Taeihagh and Lim, 2019). Strong legislation is
therefore required that sets minimum vehicle safety performance before being exposed to
cyber-attacks, i.e., hijacks and user privacy concerns. Similarly, where an incident happens that
involves damage, such as the most recent one in Tesla, the liability needs to be clearly described
that governs the legislation. (Boudette, 2017). Legal readiness in terms of policy and regulation,
liability, privacy, and cybersecurity context will facilitate the untroubled roll-out of CAVs
(Rosique et al., 2019). At the same time, state initiatives for the user acceptance, i.e. consumer
acceptance, marketing and advertising, and affordable cost of CAVs will boost up these efforts.
Finally, the legislation of relevant acts and rules will ensure smooth traffic and safety in terms
of travel behaviour, transport supply, land use, economy, governance, and environment.
4.8. Integrated Management Framework for CAVs Cybersecurity
Current state-of-the-art policy solutions do not provide a realistic mechanism to achieve
the adaptability necessary for the complex, competitive, and diverse conditions under which
CAVs operate. The journey to this destination is not easy since 84% of OEM employees and
their vendors are concerned that cybersecurity measures are not keeping up with emerging
technologies (Scalas and Giacinto, 2019). A recent study conducted by Austroads highlights
that automotive manufacturers, data aggregators, vendors, and road operators are all interested
in exploring techniques to realise value from crowdsourced and fleet-sourced CAVs-data for
asset management purposes (Infrastructure-Australia, 2017). Providing a coordinated approach
to CAV cybersecurity development would avoid duplication and sharing of the lessons learned
could accelerate the learning process. This co-creation would adopt a shared problem-solving
approach involving both the road operators (as customers) and suppliers such as automotive
manufacturers, equipment manufacturers, data aggregators and data processors.
5. Conclusion
Connected and autonomous vehicles can see their environment, evaluate the optimum route,
and travel without human involvement for the entire journey. Nonetheless, there are challenges
to the complete adoption of the CAV technology, and cybersecurity is one of them. In this
review paper, we presented the CAV communication framework in an immersive way with a
graphical flow chart to provide a concise picture of all the interfaces of CAV cyber-attacks in
the ITS. We addressed the ineffectiveness of the orthodox CPS cybersecurity approaches for
the CAVs. The summary of cyber-attacks on the CAVs, the respective CAV communication
system, its impact and the corresponding mitigation strategies are presented. Besides, we
identified the potential avenues that need to be addressed for the successful and timely battle
against cyber-attacks. Moreover, each focused area is supported with research gaps and future
direction which would facilitate CAV academics, application creators and decision-makers in
defining and creating a comprehensive CAV-cybersecurity framework.
Acknowledgements
The authors would also like to acknowledge the Australian Government, Department of
Industry, Innovation and Science for the financial support received for this study through the
Automotive Engineering Graduate Program (GRANT NO: AEGP000050). The views
expressed by the authors do not necessarily reflect those of the funding agency. The authors
would also like to acknowledge three anonymous reviewers whose feedback have helped us to
improve the paper.
References
Al Zaabi, A.O., Yeun, C.Y., Damiani, E., 2019. Autonomous Vehicle Security: Conceptual Model,
2019 IEEE Transportation Electrification Conference and Expo, Asia-Pacific (ITEC Asia-Pacific).
IEEE, pp. 1-5.
Ali, Y., Haque, M.M., Zheng, Z., Washington, S., Yildirimoglu, M., 2019. A hazard-based duration
model to quantify the impact of connected driving environment on safety during mandatory lane-
changing. Transportation Research Part C: Emerging Technologies 106, 113-131.
Ali, Y., Sharma, A., Haque, M.M., Zheng, Z., Saifuzzaman, M., 2020. The impact of the connected
environment on driving behavior and safety: A driving simulator study. Accident Analysis & Prevention
144, 105643.
Ali, Y., Zheng, Z., Haque, M.M., Yildirimoglu, M., Washington, S., 2020. Understanding the
discretionary lane-changing behaviour in the connected environment. Accident Analysis & Prevention
137, 105463.
Amirtahmasebi, K., Jalalinia, S.R., 2010. Vehicular networks-Security, Vulnerabilities and
Countermeasures. http://publications.lib.chalmers.se/records/fulltext/123778.pdf.
Amoozadeh, M., Raghuramu, A., Chuah, C.-N., Ghosal, D., Zhang, H.M., Rowe, J., Levitt, K., 2015.
Security vulnerabilities of connected vehicle streams and their impact on cooperative driving. IEEE
Communications Magazine 53(6), 126-132.
Anouar, B., Mohammed, B., Abderrahim, G., Mohammed, B., 2016. Vehicular navigation spoofing
detection based on V2I calibration, 2016 4th IEEE International Colloquium on Information Science
and Technology (CiSt). IEEE, pp. 847-849.
Anwar, M., He, W., Ash, I., Yuan, X., Li, L., Xu, L., 2017. Gender difference and employees'
cybersecurity behaviors. Computers in Human Behavior 69, 437-443.
Azogu, I.K., Ferreira, M.T., Larcom, J.A., Liu, H., 2013. A new anti-jamming strategy for VANET
metrics-directed security defense, 2013 IEEE Globecom Workshops (GC Wkshps). IEEE, pp. 1344-
1349.
Becher, A., Benenson, Z., Dornseif, M., 2006. Tampering with Motes: Real-World Physical Attacks on
Wireless Sensor Networks. Springer Berlin Heidelberg, Berlin, Heidelberg, pp. 104-118.
Behere, S., Torngren, M., 2015. A functional architecture for autonomous driving, 2015 First
International Workshop on Automotive Software Architecture (WASA). IEEE, pp. 3-10.
Bergin, D.L., 2015. Cyber-attack and defense simulation framework. The Journal of Defense Modeling
and Simulation 12(4), 383-392.
BMW-Group, 2018. Experimental Security Assessment of BMW Cars: A Summary Report.
https://keenlab.tencent.com/en/whitepapers/Experimental_Security_Assessment_of_BMW_Cars_b
y_KeenLab.pdf.
Boudette, N.E., 2017. Tesla’s self-driving system cleared in deadly crash. The New York Times 19,
https://www.nytimes.com/2017/2001/2019/business/tesla-model-s-autopilot-fatal-crash.html.
Brandts, J., Rott, C., Solà, C., 2016. Not just like starting over-Leadership and revivification of
cooperation in groups. Experimental economics, 19(4), 792-818.
Cabinet Office, 2016. National security and intelligence, HM Treasury, and The Rt Hon Philip
Hammond MP. National cyber security strategy 2016–2021.
https://www.gov.uk/government/publications/national-cyber-security-strategy-2016-to-2021.
Carsten, P., Andel, T.R., Yampolskiy, M., McDonald, J.T., 2015. In-vehicle networks: Attacks,
vulnerabilities, and proposed solutions, Proceedings of the 10th Annual Cyber and Information Security
Research Conference, pp. 1-8.
Cho, K.-T., Shin, K.G., 2016. Fingerprinting electronic control units for vehicle intrusion detection,
25th {USENIX} Security Symposium ({USENIX} Security 16), pp. 911-927.
Choi, W., Joo, K., Jo, H.J., Park, M.C., Lee, D.H., Security, 2018. Voltageids: Low-level
communication characteristics for automotive intrusion detection system. IEEE Transactions on
Information Forensics 13(8), 2114-2129.
Choi, W., Seo, M., Lee, D.H., 2018. Sound-proximity: 2-factor authentication against relay attack on
passive keyless entry and start system. Journal of Advanced Transportation 2018.
Cui, J., Liew, L.S., Sabaliauskaite, G., Zhou, F., 2019. A review on safety failures, security attacks, and
available countermeasures for autonomous vehicles. Ad Hoc Networks 90, 101823.
Cui, L., Hu, J., Park, B.B., Bujanovic, P., 2018. Development of a simulation platform for safety impact
analysis considering vehicle dynamics, sensor errors, and communication latencies: Assessing
cooperative adaptive cruise control under cyber attack. Transportation Research Part C: Emerging
Technologies 97, 1-22.
Czerwinski, D., Nowak, J., Przylucki, S., 2019. An Impact of Jamming Signal on the Energy Efficiency
of ZigBee Network Elements, International Conference on Computer Networks. Springer, pp. 62-75.
Decisions, T., 2019. Outdated firmware could be putting IP camera security at risk.
https://www.technologydecisions.com.au/content/security/news/outdated-firmware-could-be-
putting-ip-camera-security-at-risk-994422321.
Dey, K.C., Rayamajhi, A., Chowdhury, M., Bhavsar, P., Martin, J., 2016. Vehicle-to-vehicle (V2V)
and vehicle-to-infrastructure (V2I) communication in a heterogeneous wireless network–Performance
evaluation. Transportation Research Part C: Emerging Technologies 68, 168-184.
Dixit, V.V., Chand, S., Nair, D., 2016. Autonomous vehicles: disengagements, accidents and reaction
times. PLoS one 11(12), e0168054.
Dogan, E., Rahal, M.-C., Deborne, R., Delhomme, P., Kemeny, A., Perrin, J., 2017. Transition of
control in a partially automated vehicle: Effects of anticipation and non-driving-related task
involvement. Transportation Research Part F: Traffic Psychology Behaviour 46, 205-215.
El-Rewini, Z., Sadatsharan, K., Selvaraj, D.F., Plathottam, S.J., Ranganathan, P., 2019. Cybersecurity
challenges in vehicular communications. Vehicular Communications, 100214.
Foster, I., Prudhomme, A., Koscher, K., Savage, S., 2015. Fast and vulnerable: A story of telematic
failures, 9th {USENIX} Workshop on Offensive Technologies ({WOOT} 15).
Francillon, A., Danev, B., Capkun, S., 2011. Relay attacks on passive keyless entry and start systems
in modern cars, Proceedings of the Network and Distributed System Security Symposium
(NDSS),Eidgenössische Technische Hochschule Zürich, Department of Computer Science.
Gagandeep, A., Kumar, P., Technology, A., 2012. Analysis of different security attacks in MANETs
on protocol stack A-review. International Journal of Engineering 1(5), 269-275.
Geels, F.W., Penna, C.C., 2015. Societal problems and industry reorientation: Elaborating the Dialectic
Issue LifeCycle (DILC) model and a case study of car safety in the USA (1900–1995). Research Policy
44(1), 67-82.
Gehrig, S.K., Stein, F.J., 1999. Dead reckoning and cartography using stereo vision for an autonomous
car, Proceedings 1999 IEEE/RSJ International Conference on Intelligent Robots and Systems. Human
and Environment Friendly Robots with High Intelligence and Emotional Quotients (Cat. No.
99CH36289). IEEE, pp. 1507-1512.
Gkartzonikas, C., Gkritza, K., 2019. What have we learned? A review of stated preference and choice
studies on autonomous vehicles. Transportation Research Part C: Emerging Technologies 98, 323-337.
Greenberg, A., 2015. GM took 5 Years to fix a full-takeover hack in millions of Onstar cars. Wired,
September 10, 1059-1028.
Gurion, B., 2019. Autonomous vehicles fooled by drones that project too-quick-for-humans road-signs.
https://boingboing.net/2019/2007/2006/flickering-car-ghosts.html.
Hadlington, L., 2017. Human factors in cybersecurity; examining the link between Internet addiction,
impulsivity, attitudes towards cybersecurity, and risky cybersecurity behaviours. Heliyon 3(7), e00346.
Hasbullah, H.S., Irshad Ahmed 2010. Denial of service (dos) attack and its possible solutions in
VANET. International Journal of Electronics Communication Engineering 4(5), 813-817.
He, B.Y., Chow, J.Y., 2019. Optimal privacy control for transport network data sharing. Transportation
Research Part C: Emerging Technologies.
He, H., 2018. Cybersecurity law causing “mass concerns” among foreign firms in China. South China
Morning Post https://www.scmp.com/news/china/economy/article/2135338/cybersecurity-law-
causing-mass-concerns-among-foreign-firms-china.
He, Q., Meng, X., Qu, R., 2017. Survey on cyber security of CAV, 2017 Forum on Cooperative
Positioning and Service (CPGPS). IEEE, pp. 351-354.
Hennessy, A.P., 2016. Implementation of physical layer security of an ultra-wideband transceiver
(Doctoral dissertation, Engineering).
Hodge, C., Hauck, K., Gupta, S., Bennett, J.C., 2019. Vehicle Cybersecurity Threats and Mitigation
Approaches, in: National Renewable Energy Lab.(NREL), G., CO (United States) (Ed.).
Holt, T.J., Burruss, G.W., Bossler, A.M., 2010. Social learning and cyber-deviance: Examining the
importance of a full social learning model in the virtual world. Journal of Crime Justice 33(2), 31-61.
Hortelano, J., Ruiz, J.C., Manzoni, P., 2010. Evaluating the usefulness of watchdogs for intrusion
detection in VANETs, 2010 IEEE International Conference on Communications Workshops. IEEE, pp.
1-5.
Infrastructure-Australia, 2017. AUTOMATED VEHICLES DO WE KNOW WHICH ROAD TO
TAKE? , https://infrastructure.org.au/wp-content/uploads/2017/2009/AV-paper-FINAL.pdf.
Julisch, K., 2013. Understanding and overcoming cyber security anti-patterns. Computer Networks
57(10), 2206-2211.
Kalra, N., Paddock, S.M., 2016. Driving to safety: How many miles of driving would it take to
demonstrate autonomous vehicle reliability? Transportation Research Part A: Policy
Practice 94, 182-193.
Katerina, T., Nicolaos, P., 2018. Mouse behavioral patterns and keystroke dynamics in End-User
Development: What can they tell us about users’ behavioral attributes? Computers in Human Behavior
83, 288-305.
Kennedy, J., Holt, T., Cheng, B., 2019. Automotive cybersecurity: assessing a new platform for
cybercrime and malicious hacking. Journal of Crime Justice, 1-14.
Khan, S.K., 2019. Performance evaluation of next generation wireless UAV relay with millimeter-wave
in access and backhaul. Master Thesis, School of Engineering, RMIT University, Melbourne, Australia.
Khan, S.K., Al-Hourani, A., Chavez, K.G., 2020. Performance Evaluation of Amplify-and-Forward
UAV Relay in Millimeter-Wave, 2020 27th International Conference on Telecommunications (ICT).
IEEE, pp. 1-5.
Khan, S.K., Farasat, M., Naseem, U., Ali, F., 2020. Performance Evaluation of Next-Generation
Wireless (5G) UAV Relay. Wireless Personal Communications, 1-16.
King, Z.M., Henshel, D.S., Flora, L., Cains, M.G., Hoffman, B., Sample, C., 2018. Characterizing and
measuring maliciousness for cybersecurity risk assessment. Frontiers in Psychology 9, 39.
Koscher, K., Czeskis, A., Roesner, F., Patel, S., Kohno, T., Checkoway, S., McCoy, D., Kantor, B.,
Anderson, D., Shacham, H., 2010. Experimental security analysis of a modern automobile, 2010 IEEE
Symposium on Security and Privacy. IEEE, pp. 447-462.
Krishnapriya, V., Sikha, M., Nandakumar, R., Kidav, J.U., 2012. Hardware efficiency comparison of
IP Cores for CAN & LIN protocols, 2012 Third International Conference on Computing,
Communication and Networking Technologies (ICCCNT'12). IEEE, pp. 1-3.
Levinson, J., Askeland, J., Becker, J., Dolson, J., Held, D., Kammel, S., Kolter, J.Z., Langer, D., Pink,
O., Pratt, V., 2011. Towards fully autonomous driving: Systems and algorithms, 2011 IEEE Intelligent
Vehicles Symposium (IV). IEEE, pp. 163-168.
Li, Y., Luo, Q., Liu, J., Guo, H., Kato, N., 2019. TSP security in intelligent and connected vehicles:
Challenges and solutions. IEEE Wireless Communications 26(3), 125-131.
Li, Y., Tu, Y., Fan, Q., Dong, C., Wang, W., 2018. Influence of cyber-attacks on longitudinal safety of
connected and automated vehicles. Accident Analysis Prevention 121, 148-156.
Liang, L., Ye, H., Li, G.Y., 2018. Toward intelligent vehicular networks: A machine learning
framework. IEEE Internet of Things journal 6(1), 124-135.
Lim, H.S.M., Taeihagh, A., 2018. Autonomous vehicles for smart and sustainable cities: An in-depth
exploration of privacy and cybersecurity implications. Energies 11(5), 1062.
Linkov, V., Zámečník, P., Havlíčková, D., Pai, C.-W., 2019. Human Factors in the Cybersecurity of
Autonomous Vehicles: Trends in Current Research. Frontiers in Psychology 10, 995.
Liu, J., Ma, D., Weimerskirch, A., Zhu, H., 2017. A functional co-design towards safe and secure
vehicle platooning, Proceedings of the 3rd ACM Workshop on Cyber-Physical System Security, pp.
81-90.
Liu, J., Zhang, S., Sun, W., Shi, Y., 2017. In-vehicle network attacks and countermeasures: Challenges
and future directions. IEEE Network 31(5), 50-58.
Liu, L., Wu, B., Shi, W., 2020. A Comparison of Communication Mechanisms in Vehicular Edge
Computing, 3rd {USENIX} Workshop on Hot Topics in Edge Computing (HotEdge 20).
Liu, S., Liu, L., Tang, J., Yu, B., Wang, Y., Shi, W., 2019. Edge computing for autonomous driving:
Opportunities and challenges. Proceedings of the IEEE 107(8), 1697-1716.
Liu, S., Tang, J., Wang, C., Wang, Q., Gaudiot, J.-L., 2017. A unified cloud platform for autonomous
driving. Computer 50(12), 42-49.
Lopez, A., Malawade, A.V., Al Faruque, M.A., Boddupalli, S., Ray, S., 2019. Security of Emergent
Automotive Systems: A Tutorial Introduction and Perspectives on Practice. IEEE Design Test 36(6),
10-38.
Lyamin, N., Kleyko, D., Delooz, Q., Vinel, A., 2019. Real-time jamming DoS detection in safety-
critical V2V C-ITS using data mining. IEEE Communications Letters 23(3), 442-445.
Macher, G., Eric Armengaud, Richard Messnarz, Andreas Riel, Eugen Brenner, and Christian Kreiner,
2017. Integrated Safety and Security Development in the Automotive Domain. SAE Technical Papers
Vol.2017-(March).
Markham, T.R., Chernoguzov, A., 2017. A balanced approach for securing the OBD-II port. SAE
International Journal of Passenger Cars-Electronic Electrical Systems 10(2017-01-1662), 390-399.
Mejri, M.N., Ben-Othman, J., Hamdi, M., 2014. Survey on VANET security challenges and possible
cryptographic solutions. Vehicular Communications 1(2), 53-66.
Miller, C., Valasek, C., 2015. Remote exploitation of an unaltered passenger vehicle. Black Hat USA
2015, 91.
Möller, D.P., Haas, R.E., 2019. Guide to Automotive Connectivity and Cybersecurity. Springer.
Morris, D., Madzudzo, G., 2018. Cybersecurity and the auto industry: the growing challenges presented
by connected cars. International Journal of Automotive Technology Management 18(2), 105-118.
Muciaccia, T., Passaro, V.M.N., 2017. Future Scenarios for Software-Defined Metro and Access
Networks and Software-Defined Photonics. Photonics 4(1), 1.
Muhammad, M., Safdar, G.A., 2018. Survey on existing authentication issues for cellular-assisted V2X
communication. Vehicular Communications 12, 50-65.
Nayegandhi, A., 2007. LIDAR technology overview. ETI – US Geological Survey, Florida Integrated
Science Center, St. Petersburg, FL 33701. 17,
http://www.ce.siue.edu/faculty/hzhou/Highway%20Inventory%20Refereces/Nayegandhi_Lidar_Tec
hnology_Overview.pdf.
Nie, S., Liu, L., Du, Y., 2017. Free-fall: Hacking tesla from wireless to can bus. Briefing, Black Hat
USA, 1-16.
Nissan, 2016. Hacker takes control of Nissan electric vehicle from other side of the world through Leaf
app. https://www.ibtimes.co.uk/hacker-takes-control-nissan-electric-vehicle-other-side-world-
through-leaf-app-1545808.
Noy, I.Y., Shinar, D., Horrey, W.J., 2018. Automated driving: Safety blind spots. Safety science 102,
68-78.
Onishi, H., Wu, K., Yoshida, K., Kato, T., 2017. Approaches for vehicle cyber-security in the US.
International Journal of Automotive Engineering 8(1), 1-6.
Parkinson, S., Ward, P., Wilson, K., Miller, J., 2017. Cyber threats facing autonomous and connected
vehicles: Future challenges. IEEE Transactions on Intelligent Transportation Systems 18(11), 2898-
2915.
Parno, B., Perrig, A., 2005. Challenges in securing vehicular networks, Workshop on hot topics in
networks (HotNets-IV). Maryland, USA, pp. 1-6.
Pawade, S., Shah, S., Tijare, D., 2013. Zigbee Based Intelligent Driver Assistance System. International
Journal of Engineering Research Applications 3, 1463-1468.
Petit, J., Feiri, M., Kargl, F., 2014. Revisiting attacker model for smart vehicles, 2014 IEEE 6th
International Symposium on Wireless Vehicular Communications (WiVeC 2014). IEEE, pp. 1-5.
Petit, J., Stottelaar, B., Feiri, M., 2015. Remote attacks on automated vehicles sensors: Experiments on
camera and lidar. Black Hat Europe 11, 2015.
Puñal, O., Aktaş, I., Schnelke, C.-J., Abidin, G., Wehrle, K., Gross, J., 2014. Machine learning-based
jamming detection for IEEE 802.11: Design and experimental evaluation, Proceeding of IEEE
International Symposium on a World of Wireless, Mobile and Multimedia Networks 2014. IEEE, pp.
1-10.
Rad, S.R., Farah, H., Taale, H., van Arem, B., Hoogendoorn, S.P., 2020. Design and operation of
dedicated lanes for connected and automated vehicles on motorways: A conceptual framework and
research agenda. Transportation Research Part C: Emerging Technologies 117, 102664.
Raiyn, J., 2013. Handoff self-management based on SNR in mobile communication networks.
International Journal of Wireless Mobile Computing 6(1), 39-48.
Raiyn, J., 2018. Data and cyber security in autonomous vehicle networks. Transport and
Telecommunication 19(4), 325-334.
Rawat, A., Sharma, S., Sushil, R., 2012. VANET: Security attacks and its possible solutions. Journal of
Information Operations Management 3(1), 301.
Rosique, F., Navarro, P.J., Fernández, C., Padilla, A., 2019. A systematic review of perception system
and simulators for autonomous vehicles research. Sensors 19(3), 648.
Rouf, I., Miller, R.D., Mustafa, H.A., Taylor, T., Oh, S., Xu, W., Gruteser, M., Trappe, W., Seskar, I.,
2010. Security and Privacy Vulnerabilities of In-Car Wireless Networks: A Tire Pressure Monitoring
System Case Study, USENIX Security Symposium.
Roy, N., Shen, S., Hassanieh, H., Choudhury, R.R., 2018. Inaudible voice commands: The long-range
attack and defense, 15th {USENIX} Symposium on Networked Systems Design and Implementation
({NSDI} 18), pp. 547-560.
SAE-International, 2018. Taxonomy and definitions for terms related to driving automation systems for
on-road motor vehicles.
Scalas, M., Giacinto, G., 2019. Automotive Cybersecurity: Foundations for Next-Generation Vehicles,
2019 2nd International Conference on new Trends in Computing Sciences (ICTCS). IEEE, pp. 1-6.
Seuwou, P., Banissi, E., Ubakanma, G., 2020. The Future of Mobility with Connected and Autonomous
Vehicles in Smart Cities, Digital Twin Technologies and Smart Cities. Springer, pp. 37-52.
Shaharudin, M.S., Fernando, Y., Jabbour, C.J.C., Sroufe, R., Jasmi, M.F.A., 2019. Past, present, and
future low carbon supply chain management: A content review using social network analysis. Journal
of cleaner production 218, 629-643.
Sharma, A., Zheng, Z., Kim, J., Bhaskar, A., Haque, M., 2020. Is an informed driver a better decision
maker? A grouped random parameters with heterogeneity-in-means approach to investigate the impact
of the connected environment on driving behaviour in safety-critical situations. Analytic Methods in
Accident Research, 100127.
Shin, H., Kim, D., Kwon, Y., Kim, Y., 2017. Illusion and dazzle: Adversarial optical channel exploits
against lidars for automotive applications, International Conference on Cryptographic Hardware and
Embedded Systems. Springer, pp. 445-467.
Shrestha, R., Bajracharya, R., Shrestha, A.P., Nam, S.Y., 2019. A new type of blockchain for secure
message exchange in VANET. Digital Communications Networks.
Sinha, A., Chand, S., Wijayaratna, K.P., Virdi, N., Dixit, V., 2020. Comprehensive safety assessment
in mixed fleets with connected and automated vehicles: A crash severity and rate evaluation of
conventional vehicles. Accident Analysis & Prevention 142, 105567.
Solon, O., 2016. Team of hackers take remote control of Tesla Model S from 12 miles away. The
Guardian 20.
Stasinopoulos, P., Shiwakoti, N., Beining, M., Use-stage life cycle greenhouse gas emissions of the
transition to an autonomous vehicle fleet: A System Dynamics approach. J Journal of Cleaner
Production 278, 123447.
Sumra, I.A., Ab Manan, J.-L., Hasbullah, H., 2011. Timing attack in vehicular network, Proceedings of
the 15th WSEAS International Conference on Computers, World Scientific and Engineering Academy
and Society (WSEAS), Corfu Island, Greece, pp. 151-155.
Taeihagh, A., Lim, H.S.M., 2019. Governing autonomous vehicles: emerging responses for safety,
liability, privacy, cybersecurity, and industry risks. Transport Reviews 39(1), 103-128.
Tang, J., Liu, S., Yu, B., Shi, W., 2018. Pi-edge: A low-power edge computing system for real-time
autonomous driving services. arXiv preprint arXiv:.04978.
Tanuja, K., Sushma, T., Bharathi, M., Arun, K., 2015. A survey on VANET technologies.
Taylor, A., Japkowicz, N., Leblanc, S., 2015. Frequency-based anomaly detection for the automotive
CAN bus, 2015 World Congress on Industrial Control Systems Security (WCICSS). IEEE, pp. 45-49.
TencentKeen-SecurityLab, 2019. Experimental Security Research of Tesla Autopilot, in:
https://keenlab.tencent.com/en/whitepapers/Experimental_Security_Research_of_Tesla_Autopilot.
pdf (Ed.).
Van der Heijden, R.W., Kargl, F., Abu-Sharkh, O.M., 2016. Enhanced position verification for
VANETs using subjective logic, 2016 IEEE 84th Vehicular Technology Conference (VTC-Fall). IEEE,
pp. 1-7.
Verma, S., Mallick, B., Verma, P., 2015. Impact of gray hole attack in VANET, 2015 1st International
Conference on Next Generation Computing Technologies (NGCT). IEEE, pp. 127-130.
Vijaykumar, V., Elango, S., 2014. Hardware implementation of tag-reader mutual authentication
protocol for RFID systems. Integration 47(1), 123-129.
Wang, J., Shao, Y., Ge, Y., Yu, R., 2019. A survey of vehicle to everything (V2X) testing. Sensors
19(2), 334.
Wang, P., Wu, X., He, X., 2020. Modeling and analyzing cyberattack effects on connected automated
vehicular platoons. Transportation Research Part C: Emerging Technologies 115, 102625.
Wang, Y., Liu, L., Zhang, X., Shi, W., 2019. HydraOne: An indoor experimental research and education
platform for CAVs, 2nd {USENIX} Workshop on Hot Topics in Edge Computing (HotEdge 19).
Weiß, C., 2011. V2X communication in Europe–From research projects towards standardization and
field testing of vehicle communication technology. Computer Networks 55(14), 3103-3119.
Welk, A.K., Hong, K.W., Zielinska, O.A., Tembe, R., Murphy-Hill, E., Mayhorn, C.B., 2015. Will the
“Phisher-Men” Reel You In?: Assessing individual differences in a phishing detection task.
International Journal of Cyber Behavior, Psychology Learning 5(4), 1-17.
Wesson, K.D., Shepard, D.P., Bhatti, J.A., Humphreys, T.E., 2011. An evaluation of the vestigial signal
defense for civil GPS anti-spoofing, Radionavigation Laboratory Conference Proceedings.
West, D.M., 2016. Moving forward: self-driving vehicles in China, Europe, Japan, Korea, and the
United States. Center for Technology Innovation at Brookings: Washington, DC, USA.
Xu, C., Liu, H., Li, P., Wang, P., 2018. A remote attestation security model based on privacy-preserving
blockchain for V2X. IEEE Access 6, 67809-67818.
Yan, C., Xu, W., Liu, J., 2016. Can you trust autonomous vehicles: Contactless attacks against sensors
of self-driving vehicle. DEF CON 24.
Ye, H., Liang, L., Li, G.Y., Kim, J., Lu, L., Wu, M., 2018. Machine learning for vehicular networks:
Recent advances and application examples. IEEE Vehicular Technology Magazine 13(2), 94-101.
Yu, B., Xu, C.-Z., Xiao, B., 2013. Detecting sybil attacks in VANETs. Journal of Parallel Distributed
Computing 73(6), 746-756.
Yuan, X., Chen, Y., Zhao, Y., Long, Y., Liu, X., Chen, K., Zhang, S., Huang, H., Wang, X., Gunter,
C.A., 2018. Commandersong: A systematic approach for practical adversarial voice recognition, 27th
{USENIX} Security Symposium ({USENIX} Security 18), pp. 49-64.
Zhang, G., Yan, C., Ji, X., Zhang, T., Zhang, T., Xu, W., 2017. Dolphinattack: Inaudible voice
commands, Proceedings of the 2017 ACM SIGSAC Conference on Computer and Communications
Security, pp. 103-117.
Zhang, R., Xue, R., Liu, L., 2019. Security and privacy on blockchain. ACM Computing Surveys 52(3),
1-34.
Zhang, T., Antunes, H., Aggarwal, S., 2014. Defending connected vehicles against malware: Challenges
and a solution framework. IEEE Internet of Things journal 1(1), 10-21.
Zhao, X., Wang, Z., Xu, Z., Wang, Y., Li, X., Qu, X., 2020. Field experiments on longitudinal
characteristics of human driver behavior following an autonomous vehicle. Transportation Research
Part C: Emerging Technologies 114, 205-224.
Zhou, M., Qin, Z., Lin, X., Hu, S., Wang, Q., Ren, K., 2019. Hidden voice commands: Attacks and
defenses on the vcs of autonomous driving cars. IEEE Wireless Communications 26(5), 128-133.
