You are on page 1of 17

Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

Contents lists available at ScienceDirect

Journal of King Saud University –


Computer and Information Sciences
journal homepage: www.sciencedirect.com

Provably secure authentication for the internet of vehicles


E Haodudin Nurkifli a,b,⇑, Tzonelih Hwang a
a
Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan, Taiwan
b
Faculty of Computer Science, Universitas Singaperbangsa Karawang, West Java, Indonesia

a r t i c l e i n f o a b s t r a c t

Article history: Rapid technological development has far-reaching implications for every aspect of life, including trans-
Received 20 March 2023 portation. The emerging Intelligent Transportation System (ITS) based on the Internet of Vehicles (IoV)
Revised 10 August 2023 requires robust security measures due to the public channels used to communicate between vehicles
Accepted 19 August 2023
and trusted authorities. These channels create opportunities for adversaries to intercept and exploit
Available online 24 August 2023
the transmitted data, leading to severe attacks. Moreover, cars are often parked in public areas, making
them vulnerable to physical attacks such as theft of smart cards/devices, cloning, and impersonation. In
Keywords:
this paper, we propose a new protocol for IoV that addresses these security concerns. Our protocol offers
Internet of Vehicle
Attacks
user authentication based on biometrics, using a biometric key to prevent revealing data from theft of
PUF smart cards/devices. Additionally, we use The Physical Unclonable Function (PUF) to resist cloning
Biometric attacks. We have conducted an informal analysis, which confirms that our protocol fulfills security fea-
Analysis Formally tures. Furthermore, we have conducted a formal analysis using the RoR model and the Scyther tool to ver-
ify that our protocol achieves security properties and withstands well-known attacks. We also
demonstrate the time complexity and show that our protocol has low computation time. Therefore,
our protocol is secure and has a low computation time.
Ó 2023 The Author(s). Published by Elsevier B.V. on behalf of King Saud University. This is an open access
article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

1. Introduction ical role as the backbone of the smart city by integrating the Inter-
net of Things (IoT) and the Internet of Vehicles (IoV).
It is predicted that the number of vehicles on the road and in the In 2023, it is expected that billions of vehicles will be connected
network will increase in the next 10–20 years (Jia et al., 2016). to VANET (Vehicle Ad Hoc Network) as part of the implementation
Rapid technological development has enormous implications for of the Internet of Vehicles (IoV) (Azees et al., 2016). However, one
the Intelligent Transportation System (ITS), which provides toll of the crucial problems in IoV is ensuring privacy, data protection,
payment, insurance, traffic information, and logistics management and trusted management (Javaid et al., 2020), (Elhalawany et al.,
services. Additionally, ITS allows the use of Autonomous Vehicles 2019). Furthermore, vehicles can share acceleration, position, and
(AVs). Unfortunately, the World Health Organization (WHO) speed information with other vehicles (V2V) or with roadside units
reported that the number of road accidents every year is approxi- (V2R) to exchange important information, such as accidents in a
mately 1.25 million. Hence, the primary purpose of implementing particular region. Roadside units (RSUs) located alongside roads
a transportation system and network is to provide convenience can communicate with passing vehicles through internet access.
and safety for passengers (Nidhal et al., 2014). ITS also plays a crit- Today, AVs have shown considerable success and promise to
significantly improve driving comfort while reducing traffic colli-
sions with proper control (In et al., 2020). The IoVs provide drivers
⇑ Corresponding author. and authorities with a wide range of equipment. However,
E-mail addresses: P78067011@gs.ncku.edu.tw, dudi.nurkifli@staff.unsika.ac.id (E medium- and long-range radios and networking features make
Haodudin Nurkifli), hwangtl@csie.ncku.edu.tw, hwangtl@mail.csie.ncku.edu.tw (T.
Hwang).
cars vulnerable to new exploits and attacks. An adversary, for
Peer review under responsibility of King Saud University. Production and hosting by
example, may spread misleading information about non-existent
Elsevier. unsafe road conditions, mislead drivers, create a traffic jam, and
so on. Furthermore, modern vehicles have a new internal commu-
nication bus system that connects multiple controllers that control
vehicle components, including airbags, engines, and brake control.
Production and hosting by Elsevier Since the internal network commonly utilizes non-encrypted pro-

https://doi.org/10.1016/j.jksuci.2023.101721
1319-1578/Ó 2023 The Author(s). Published by Elsevier B.V. on behalf of King Saud University.
This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

tocols, an intruder may obtain access to it using the vehicle’s exter- e. Analyzing the proposed protocol informally and formally
nal wireless interface and inflict severe harm. In the next ten years, using the RoR model and Scyther tool to validate it. In addi-
the probability of cyberattacks will rise by 20.3 percent (Report, tion, comparing with previous related protocols in terms of
2015). Since certain VANET implementations can have life-or- performance and security feature achievements.
death implications, security designers must consider protecting
these networks from the very beginning. Security flaws must be The roadmap for this manuscript is as follows: Section 2 pre-
resolved before AVs are widely deployed. sents a literature review, and Section 3 discusses materials and
The communication model in V2R or V2V using public channels definitions. Section 4 reviews Aman et al.’s protocol, and the crypt-
opens the attacker to eavesdrop on all messages and then conduct analysis of their protocol is presented in Section 5. Section 6 outli-
severe attacks such as Sybil, DoS, replay, man-in-the-middle, tim- nes the proposed protocol, and the discussion and results are
ing, impersonation attacks, and data tampering. In addition to shown in Section 7. Finally, we conclude our paper in Section 8.
these threats, the adversary can steal the device and obtain access
to a vehicle’s storage memory. The Roadside Unit (RSU) makes IoV 2. Literature review
and especially the defense of AVs much more difficult. Vehicles
must store credential secrets, such as secret keys, to create a stable Many existing authentication protocols use public-key cryptog-
VANET. As a result, they are more likely to be the victim of physical raphy algorithms in VANET, such as those presented by (Fuentes &
threats, with the adversary revealing the secret keys. González-tablas, 2010), (Adigun et al., 2013), (Whyte et al., 2021),
The RSU is generally more secure than a vehicle because a lock, and (Salem et al., 2014). However, a public-key cryptography algo-
camera, or another physical protection method protects the RSU. rithm can increase the computational burden; hence, it is not suit-
However, the RSU remains the target of the attacker. The majority able for devices with limited memory, such as smart cards and
of researchers in VANET focus on using a public key algorithm; sensors. Additionally, (Sulaiman et al., 2013) proposed the
however, it cannot resolve the physical attack problem. Therefore, Hashed-chain-based Authentication Protocol (HAP) for VANET
we proposed a new protocol using user authentication based on authentication using symmetric-key cryptography hash-based.
biometrics and Physically Unclonable Function (PUF) to resist Their protocol is based on the assumption that every vehicle is
physical attacks on IoV. equipped with a hardware security module (HSM) that stores cre-
Since vehicles move at high speeds, the security protocol for dentials in its memory, making it vulnerable to physical attacks.
IoVs must have a communication burden slightly. Mutual authen- Other research presented by (Fuentes & González-tablas, 2010)
tication is an essential aspect of IoV. When communicating and (Committee, 2013) proposed authentication protocols using
between a vehicle and an RSU, both parties must authenticate each Elliptic Curve Cryptography (ECC) to achieve nonrepudiation based
other to ensure that data transmission comes from a legitimate on digital signature. However, their protocol has the same assump-
source. On the other hand, Authentication in IoVs must be fool- tion as (Fuentes & González-tablas, 2010), (Adigun et al., 2013),
proof and effective. Another essential criterion for IoV is the pro- (Whyte et al., 2021), and (Sulaiman et al., 2013); hence, it has sim-
tection of anonymity, which means that an adversary should not ilar issues.
track the origins of communications. Furthermore, anonymity The industrial automotive sector has attempted to solve key
means the adversary cannot obtain a real identity and cannot link management and physical attack problems. For example, PT
every message from one vehicle. Escript is an automotive cybersecurity industry that develops
A protocol design that can achieve comprehensive security Secure Hardware Extension (HSE) for companies such as Bay-
properties is needed for the system based on the Internet. In this erische Motoren Werke (BMW) and Audi (Soja, n.d.). Other pro-
research, we design security protocols specifically for the Internet jects, such as E-safety Vehicle Intrusion Protected Applications
of Vehicles (IoVs), which can generally be applied in the Internet of (EVITA) proposed by the European Union (EU), manage security
Things (IoT) environment. The industry can adopt our design, or keys to provide physical security (Seudié & Gmbh, 2009). The cur-
the government can adopt our protocol authentication to deploy rent project is Preserve Project, which develops an HSM based on
the intelligent transportation system. Additionally, our authentica- ECC and one of the public-key cryptography algorithms (Twente,
tion protocol can serve as a reference for researchers working on 2015). The National Institute of Standards and Technology (NIST)
system security and mutual authentication. Our authentication proposed the Federal Information Processing Standard (FIPS),
protocol can achieve security features and withstand attacks as which consists of four levels ranging from low to high security.
mentioned in 7.4, indicating that our protocol is very important For example, Level 1 achieves low security, and Level 4 provides
in the security system field. high security, including tamper detection and protection against
This article proposes a new authentication solution for resolv- physical attacks (Jr, 2002). Trusted Computing Group (TCG) pro-
ing the Internet of Vehicle problem, with the following posed a security key module for authentication and encryption,
contributions: called the Trusted Platform Module (TPM). TPM is an external
peripheral that stores a secret key in non-volatile memory.
a. Designing the adversary model for PUF authentication (as Additionally, most protocols that rely on HSM to achieve secu-
shown in 3.3). This adversary model is used for evaluating rity properties based on secret credentials store them in internal
previous protocols and our proposed protocol. memory, making it so that only the HSM can access the private
b. Showing a comprehensive study of Aman et al.’s protocol data. However, this solution is vulnerable to side-channel attacks.
and also denoting cryptanalysis of their protocol. On the other hand, physical security based on tamper-resistant cir-
c. Designing a secure, lightweight mutual authentication pro- cuitry is very expensive. Therefore, many protocols use external
tocol that is efficient and suitable for the IoV. memory buses for data sharing.
d. The protocol achieves the following security properties: pri- In their proposal, (X. Zhang & Chen, 2019) suggested a block-
vacy preservation, anonymity, mutual authentication, per- chain solution for securely sharing and storing data in VANET. Ini-
fect forward and backward secrecy (PFBS), session key tially, mutual authentication between the vehicle and trusted
security, and physical security. Additionally, the protocol authority is required before the vehicle can securely share the data.
withstands various attacks, including protection against However, their scheme has a high computational burden and lacks
cloning, impersonation, DoS, tracking, and replay attacks scalability. Later, (Yahiatene, 2018) proposed the Software Define

2
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

Vehicle Social Network (SDVSN) protocol, which uses certificates ically unclonable function (PUF). User authentication can help
and achieves data integrity through vehicle authentication. avoid stolen device/smart card attacks. Our protocol adopts the
Another proposed solution is PoolCoin by (Kaci, 2020), which is network layer proposed by Aman et al. (Aman et al., 2021), which
based on a distributed model for managing users in the network. includes a trusted authority, the RSU gateway, the RSU, and the
Their scheme assumes that one regional RSU gateway encom- vehicle (as shown in Fig. 1).
passes several RSUs for data sharing. (Yang et al., 2019) proposed
a decentralized model for managing users in VANET. (Lei et al., 3. Materials and methods
2017) proposed a blockchain-based solution that discusses key
management. However, most existing protocols in VANET are vul- This section covers Preliminaries, and the Secure Definition of
nerable to physical attacks. In summary, while several blockchain- Hash Function, PUF, and Adversary Model.
based proposals exist for securing data in VANET, most of them
suffer from scalability and computational issues and are vulnerable
3.1. Preliminaries
to physical attacks.
(Mcgrath et al., 2019), (Herder et al., 2014), (Babaei & Schiele,
This subsection briefly presents the preliminary background of
2019) introduced PUFs as an alternative to resolving the physical
the Fuzzy Extractor (FE) and Physical Unclonable Function (PUF).
attack problem, based on the uniqueness of PUFs, which are
applied in sensors or smart cards with limited resources (see defi-
3.1.1. FE
nition 2 in Subsection 3.2.2 for a more detailed discussion of secure
FE has two functions: first, FE:GenðÞ is the function for generat-
PUFs). Other researchers have applied protocol security PUF-based
ing the key and the helper data from the user’s biometric; the
for wireless sensor networks or the Internet of Things, including
equation is ðK u ; hdÞ ¼ FE:GenðBu Þ, where K u , Bu , and hd denote
(Javaid, 2019), (Aman, Basheer, & Sikdar, 2019), (Aman, Taneja,
Key, Biometric from the user, and helper data, respectively. Second,
et al., 2019), (Long et al., 2019), (Gope et al., 2019), (Chatterjee
FE:RecðÞ is the function for reconstructing the key from helper data
et al., 2019), (Aman, Basheer, & Member, 2019), (Aman et al.,
2018), (Aman et al., 2020). However, Aman et al. pointed out that and the biometric; the equation is K u ¼ FE:Recðhd; B0u Þ. In realistic
the existing protocol is unsuitable for VANETs due to the high conditions, the biometric input has noise; B0u is the noisy biometric
mobility of vehicles. Additionally, the current PUF-based protocol input. Bu denotes the biometric of a user. The approximation of the
only highlights energy efficiency and not communication burden. Hamming distance between B0u and Bu must be lower than the
Recently, Aman et al. (Aman et al., 2021) proposed a security pro- threshold (t); the equation is disðBu ; B0u Þ  t. Therefore,
tocol PUF-based on VANETs. However, their protocol has a loop- ðK u ; hdÞ ¼ FE:GenðBu Þ ! K u ¼ FE:Recðhd; B0u Þ. The similarity of the
hole and fails to achieve several security properties; we present a original and noisy biometric input is the foundation of a successful
detailed discussion in Section 5. FE. This article employs the fuzzy extractor to derive a key K u from
Kaiping Xue et al.(Xue et al., 2022) proposed a Distributed Biometric Bu , where the equation is ðK u ; hdÞ ¼ FE:GenðBu Þ. A fuzzy
Authentication protocol using a smart contract for Roaming Service extractor is also utilized to create a non-stable PUF to obtain stable
in a vehicular mobile network. However, it employs weak anonym- output; a key is produced from the response input using a fuzzy
0
ity by encrypting the identity, and it can’t provide unlinkability and extractor, where the equation is ðfkRSU ; hdRSU Þ ¼ FE:GenðRiRSu Þ. The
untraceability. Additionally, the protocol does not incorporate an keys are reconstructed from the response and help data using
unclonable device and also employs public cryptography, which Fuzzy Extractor Reconstruction, where the equation is
may increase the computational burden.  0 
fkRSU ¼ FE:Rec RiRSu ; hdRSU (Dodis et al., 2004) and (Gope & Sikdar,
Wang Jing et al. (J. Wang et al., 2022) have researched protocol
security for multiserver authentication and key agreement in IoVs. 2019).
Their research utilizes a two-factor authentication involving iden-
tity and password. However, both identity and password are short 3.1.2. PUF
words and can be easily guessed. Hence, their protocol is vulnera- The PUF is a unique characteristic in the manufacturing process
ble to online and offline password-guessing attacks. Moreover, in the fabrication of a chip to map Challenge C to a Response R (E.
their protocol may have weaknesses under the Dolev-Yao Model Suh & Devadas, 2007); The equation formally is (R) = PUF(C). The
Adversary, where the attacker can obtain the secret after guessing advantage of the PUF is that it is hard to clone (Bohm & Hofer,
the password and identity, potentially leading to various attacks, 2012). PUF has two types: Non-ideal PUF and ideal PUF. The non-
including impersonation attacks. ideal PUF, where temperature can cause PUF to give a varied
Yazid Muhammad (Yazid et al., 2023) also proposed an Authen- response output even with the same input of Challenge C, can be
tication Method for Vehicle Monitoring IoT devices. However, their stabilized using a fuzzy extractor (Tuyls & Batina, 2006), (Dodis
protocol’s authentication does not provide security features, et al., 2004).
including anonymity, untraceability, perfect forward security, While the stable PUF provides the same response when insert-
unlinkability, and perfect forward secrecy. ing the same set of challenges, researchers have recently created an
Zhang Han, Lai Yingxu, and Chen Ye (H. Zhang et al., 2023) have ideal PUF that guarantees a 0% Bit-Error Rate even if a non-stable
researched protocol security methods. However, their protocol temperature occurs (Jeon et al., 2015), (Chuang et al., 2019), (Lu
shares a similar problem with Wang Jing et al.’s authentication et al., 2018), (W. Wang et al., 2018). In this article, the smart card
protocol; that is, their protocol is vulnerable to password- is equipped with a stable PUF, and the RSU is fitted with both non-
guessing attacks, and it is fatalistic because their protocol fails to ideal PUF and a fuzzy extractor.
maintain primary security after an attacker guesses the password. Based on the types, PUFs are divided into Weak PUF and Strong
The majority of current research does not provide an unclonable PUF. Weak PUF is utilized for generating keys, while Strong PUF is
device. used for authentication methods. The architecture of Strong PUF
This article proposes a new security protocol for the Internet of includes Optical PUF and Arbiter PUF. On the other hand, the archi-
Vehicles. We consider the use of smart cards in intelligent vehicles. tecture of Weak PUF includes Ring-Oscillator PUF and SRAM PUF
Rapidly advancing technology makes it reasonable for intelligent (Herder et al., 2014). Additionally, several researchers have
vehicles to be equipped with a smart card reader and biometric extended the usage of PUFs to encryption and decryption processes
input. On the other hand, the smart card is equipped with a phys- (Choi et al., 2010), (Kleber et al., 2015), and (Guo et al., 2018).
3
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

Fig. 1. System Architecture.

3.2. Secure definition of hash function and PUF posal. In addition, we modified the DY model by adding the new
capabilities of adversaries C4, C5, and C6 to create a robust adver-
This subsection briefly presents the secure definition of Hash sary model. The proposed scheme remains secure under the attack
Function and PUF. models. The powers of the adversary based on the DY model are as
follows:
3.2.1. Definition 1(The security of hash function)
The hash function is safe according to its formal specification. a. C1: The adversary can intercept and eavesdrop on all the
Let h : f0; 1g ! f0; 1gn where the equation denotes an input bit transmitted messages via the public channel between the
of any length resulting in an output bit of a fixed length. The prob- vehicle, the RSU, the RSU gateway, and the Trusted
ability of an adversary locating a collision of a hash function at a Authority.
b. C2: The adversary can modify and delete the data.
specific time is Adv A ðtÞ ¼ Pr½ðx1 ; x2 Þ2R A : x1 –x2 ; hðx1 Þ ¼ hðx2 Þ,
Hash
c. C3: The adversary can send and reply to a message as one
where PrðXÞ is the probability of X being chosen randomly, and
participating in the parties, e.g., one vehicle. This attack is
x1 and x2 are strings that A will randomly pick. Additionally, the
called an active attack.
maximum time an adversary can obtain a collision of the hash
function is Adv A
Hash
ðtÞ  / (Gope et al., 2019). Furthermore, we added the following capabilities to the DY
adversary model to create a robust adversary model:
3.2.2. Definition 2(The security of PUF)
When PUF 1 generates output response R1 andR2 2 f0; 1gk with at d. C4: The adversary can steal Smart Card, vehicles, or the
least d1 variance for two input challenges, a secure PUF is obtained. RSU.
If an input challenge C 1 results in different output responses e. C5: The adversary can extract and compromise credentials
R1 andR2 2 f0; 1gk for any two different PUFs ðPUF 1 ; PUF 2 Þ, then stored in their memory using a side-channel attack.
f. C6: The adversary can activate PUF by programming without
C 1 andC 2 2 f0; 1gk . Consequently, Pr½HDðPUF 1 ðC 1 Þ; PUF 1 ðC 2 ÞÞ >
separating the PUF from the board unit (OBU).
d1  ¼ 1  e, and Pr½HDðPUF 1 ðC 1 Þ; PUF 2 ðC 1 ÞÞ > d2  ¼ 1  e. Here, d1
and d2 give the PUF’s error tolerance limits, and e, C 1 , and C 2 are
challenges that A randomly chooses. HD stands for Hamming Dis-
tance (Bian et al., 2020). 4. The review of Aman et al.’s protocol

3.3. Adversary model definition This section reviews the protocol of Aman et al., which includes
the system architecture, cryptography notation, registration phase,
Our adversary model uses the Dolev-Yao (DY) adversary model authentication phase, update CRP phase, and weaknesses of the
(Dolev & Yao, 1983) to evaluate Aman et al.’s protocol and our pro- protocol. A detailed discussion follows.
4
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

4.1. System architecture The third part is the registration between the RSU gateway and
the trusted authority. After the registration, the RSU gateway
This subsection denotes the system architecture. The system stores the shared key fkGS g, and the trusted authority stores the
architecture has four components: vehicle, RSU, RSU gateway, shared key and the identity of the RSU gateway fkGS ; IDG g. Every
and Trusted Authority, where the RSU gateway encompasses sev- registration phase occurs via a secure channel.‘‘
eral RSUs alongside the road. Before performing secure communi-
cation between the parties, the vehicle must initially register with
the Trusted Authorities, the RSU must register with the RSU gate- 4.4. Authentication phase
way, and the RSU gateway must register with the Trusted Author-
ity. Later, mutual authentication between the parties establishes a This subsection denotes the authentication phase as follows:
session key between the vehicle and the Trusted Authority (as Step 1: Several n Vehicles VC 1 ; VC 2 ;    ; VC n initiate communica-
shown in Fig. 1). Additionally, our authentication protocol adopts tion with the RSU, where the vehicles send
this system structure. M1 : fPIDV 1 ; a1 g; M 2 : fPIDV 2 ; a2 g;    ; M n : fPIDV n ; an g respectively
to the Roadside Unit (RSU).
Step 2: After receiving
4.2. Cryptography notations M 1 : fPIDV 1 ; a1 g; M 2 : fPIDV 2 ; a2 g;    ; M n : fPIDV n ; an g, the RSU
generates nonce N 2 , computes a response based on its stored chal-
This subsection presents the cryptography notation of Aman
lenge RiR ¼ PUFðC iR Þ, where i denotes the CRP’s index 1; 2;    ; n, that
et al.’s protocol, as shown in Table 1.    
is ðC iR ; RiR Þ 2 C 1R ; R1R ; C 2R ; R2R ;    ; ðC nR ; RnR Þ. The RSU encrypts and

4.3. Registration phase aggregates M1 ;    ; Mn in one message M RG using RiR , namely


MRG ¼ EncRi fIDR ; M1 ; M 2 ;    ; M n ; N 2 g, computes the verification
R

The registration phases are divided into three parts: the first is code V RG ¼ HðIDR jjM RG kRiR kN 2 Þ, and then the RSU sends
the registration between every vehicle and the trusted authority. PIDRSU ; M RG ; V RG to the RSU gateway.
After the registration phase, every vehicle stores the pseudo- Step 3: Upon receiving PIDRSU ; M RG ; V RG , the RSU gateway selects
identity and emergency pseudo-identity fPIDV , PIDemV g, and the CRP for the RSU’s identity IDR , verifies V RG , and decrypts
trusted authority stores the pseudo-identity of the vehicle, the MRG ¼ DecRi fIDR ; M 1 ; M 2 ;    ; M n ; N 2 g to obtain M 1 ; M 2 ;    ; M n . Sub-
R
emergency pseudo-identity, challenge-response pairs, emergency sequently, the RSU gateway generates a nonce N 3 , encrypts and
challenge-response pairs, and the identity for each vehicle aggregates M1 ;    ; M n using secret key kGS in one message M GS ,
   
PIDV ; PIDemV ; ðC i ; Ri Þ; C emi ; Remi ; IDV . where MGS ¼ EnckGS fIDG ; M 1 ; M 2 ;    ; M n ; N 3 g, computes a verifica-
The second part is the registration between the RSU and the tion code V GS ¼ HðIDG kIDS jjM GS kkGS kN 3 Þ, and sends IDG ; MGS ; V GS
RSU gateway. After the registration, the RSU stores the challenge, to the Trusted Authority.
emergency challenge, pseudo-identity, and emergency pseudo- Step 4: Upon receiving IDG ; M GS ; V GS , the Trusted Authority ver-
n o
identity C iR ; C iemR ; PIDRSU ; PIDemRSU , and the RSU gateway stores ifies V GS , decrypts M RG ¼ DeckGS fIDG ; M1 ; M 2 ;    ; M n ; N 3 g to obtain
the challenge-response, emergency challenge-response, pseudo- M1 ; M 2 ;    ; M n . Subsequently, the Trusted Authority generates a
 
identity, emergency pseudo-identity, and the identity of the RSU nonce N 4 , reads the CRPs C ij ; Rij based on the vehicle’s identity
n    o
C iR ; RiR ; C iemR ; RiemR ; PIDRSU ; PIDemRSU ; IDR . respectively, generates Token T j ¼ fIDV j ; length; SK j ; P; N 4 g, com-

Table 1
Notations of Cryptographic Functions.

Notation Definition
PIDV j Pseudo identity each vehicles
PIDRSU Pseudo identity of RSU
IDR Identity of RSU
IDG Identity of the RSU gateway
IDS Identity of the Trusted Authority
IDV Identity of the vehicle
kAS The Shared key between vehicle and trusted authority
kGS The Shared-key between RSU Gateway and trusted authority
kRG Shared key between RSU and RSU gateway
ðC iA ; RiA Þ Challenge response pairs of vehicle
ðC iR ; RiR Þ Challenge response pairs of RSU
PUF Physical Unclonable Function
K V em The Shared-key emergency between vehicle and trusted authority used if occurs desynchronization, where
 
K V em ¼ kv 1 ; kv 2 ;    ; kv n
PIDV em The Pseudo-identity emergency between vehicle and trusted authority used if occurs desynchronization,
 
where PIDV em ¼ pidv 1 ; pidv 2 ;    ; pidv n
ðC V em ; RV em Þ The Challenge-Response Pairs emergency between vehicle and trusted authority used if occurs
   
desynchronization, where C V em ¼ cv 1 ; cv 2 ;    ; cv n and RV em ¼ r v 1 ; rv 2 ;    ; r v n
K GSem The Shared-key emergency between RSU gateway and trusted authority used if occurs desynchronization,
where K GSem ¼ fkGSem1 ; kGSem2 ;    ; kGSemn g
K RGem The Shared-key emergency between RSU and RSU gateway used if occurs desynchronization, where
K RGSyn ¼ fkRGSyn1 ; kRGSyn2 ;    ; kRGSynn g
PIDRSUem The Pseudo-identity synchronization between RSU and RSU gateway used if occurs desynchronization, where
PIDRSU em ¼ fpidrem ; pidrem ;    ; pidremn g
1 2

5
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

   
putes V SAj ¼ HðIDV j jT j kRij jaj kN 4 Þ, M SAj ¼ fC j ; EncRi T j ; V SAj g, Adversary Model in Subsection 3.3). The evaluation is discussed
j
below.
where j denotes the index of each vehicle 1; 2;    ; n. Additionally,
IDj ; length; SK j ; andP denote the identity of the vehicle, the length
5.1. The protocol fails to withstand stolen device/smart card attack
of the token, a session key for each vehicle, and the expiration of
the token, respectively. Finally, the Trusted Authority encrypts
The adversary steals the vehicle or smart card and obtains a
M SG ¼ EnckGS fT 1 ; T 2 ;    ; T n ; M SA1 ; M SA2 ;    ; M SAn ; N 4 g, computes the
pseudo-identity of the vehicle, known as PIDV. The adversary can
verification code V SG ¼ HðIDG jjIDS kM SG jjN 3 kN 4 Þ, and sends
then activate the physically unclonable function (PUF) by program-
M SG ; V SG to the RSU gateway. The Trusted Authority considers
  ming it without separating it from the on-board unit (OBU). Addi-
updating the pseudo-identity where PIDnew ¼ HðIDV kRi jaj Þ. j j j tionally, based on the DY model, the adversary can eavesdrop and
Step 5: Upon receiving M SG and V SG , the RSU gateway verifies obtain all data. The steps of the attack are as follows:
V SG , decrypts M SG ¼ DeckGS fT 1 ; T 2 ;    ; T n ; M SA1 ; M SA2 ;    ; M SAn ; N 4 g Step 1: The adversary obtains all messages, including M SA1 and
to obtain T 1 ; T 2 ;    ; T n ; M SA1 ; M SA2 ;    ; M SAn , and encrypts a1 , from the public channel. Here, M SA1 ¼ fC i1 ; EncRi fT 1 g; V SA1 g,
T 1 ; T 2 ;    ; T n ; M SA1 ; M SA2 ;    ; M SAn along with N 3 using RiR . Then the
1

T 1 ¼ fIDV 1 ; length; SK 1 ; P; N 4 g, and V SA1 ¼ HðIDV 1 jjT 1 kRi1 jja1 kN 4 Þ


RSU gateway computes the verification code
(Capability C1).
V GR ¼ HðIDG jjM GR kRiR jjN 2 kN 3 Þ, and considers updating the
 Step 2: The adversary steals the Smart Card/Vehicle/Device of

pseudo-identity of the RSU as PIDnew
RSU ¼ HðIDR jRiR kN 3 Þ. Finally, the the Vehicle (Capability C4).
RSU gateway sends IDG ; M GR , and V GR to the RSU in its area. Step 3: The adversary extracts the private data from the mem-
Step 6: Upon receiving IDG ; M GR , and V GR , the RSU verifies V GR , ory of the device/smart card using a side-channel attack. (Capabil-
decrypts MGR ¼ DecRi fT 1 ; T 2 ;    ; T n ; M SA1 ; M SA2 ;    ; M SAn ; N 3 g to ity C5).
R
Step 4: The adversary activates the PUF by programming, com-
obtain T 1 ; T 2 ;    ; T n ; M SA1 ; M SA2 ;    ; M SAn , and sends
MSA1 ; M SA2 ;    ; M SAn to each vehicle VC 1 ; VC 2 ;    ; VC n , respectively. puting Ri1 ¼ PUFðC i1 Þ, decrypting DecRi fT 1 g, and obtaining T 1 (Capa-
1

Step 7: In this step, we only describe one vehicle as an example. bility C6).
Upon receiving M SA1 , the vehicle generates Ri1 ¼ PUFðC i1 Þ, decrypts Step 5: After the adversary obtains T 1 ; Ri1 , and a1 , the adversary
DecR1 fT 1 g to obtain T 1 , and verifies V SA1 . Finally, the vehicle verifies V SA1 . If the adversary can verify V SA1 , it means that the
i

updates its pseudo identity as PIDnew ¼ HðIDV 1 kRi1 jja1 jÞ. Each vehi- adversary has obtained the correct T 1 and Ri1 .
1
Based on the evaluation aforesaid, it can be concluded that
cle runs similar steps.
Aman et al.’s protocol cannot withstand a stolen device/vehicle
attack.
4.5. CRP update phase
5.2. The protocol fails to withstand impersonation attack of smart card
This Subsection presents the CRP update phase as follows:
Step 1: The Trusted Authority generates a nonce N 1 and selects
After the adversary obtains T 1 ; SK 1 ; andRi1 , the adversary can
CRP for one vehicle, for example, IDV A ; ðC iA ; RiA Þ. The Trusted Author- impersonate the vehicles in the current session using current
i
ity then generates a new challenge C new
A , encrypts it with SK 1 . Additionally, the adversary can impersonate for the next ses-
EncRi fC new
i
; N 1 g, and calculates M 1 ¼ PIDiA ; C iA ; EncRi fC new ; N 1 g.
i
sion; the steps for impersonating the vehicles are as follows:
A A
A A
Step 1: The adversary can update the new pseudo-identity of
Finally, the Trusted Authority computes the verification code
vehicle PIDnew ¼ HðIDV 1 kRi1 jja1 jÞ, the adversary can use the new
V 1 ¼ HðIDV A jjM 1 kN 1 kRiA Þ and sends M 1 and V 1 to vehicle A. 1
pseudo-identity of vehicle for next session.
Step 2: Vehicle A receives M1 and V 1 . It then generates
i i
Step 2: the adversary generates a nonce b1 and sends
RiA ¼ PUFðC iA Þ, decrypts DecRi fC new ; N 1 g to obtain C new and N 1 , M1 ¼ fPIDnew
1 , b1 g to the RSU.
A A
A

and verifies V 1 . Vehicle A subsequently generates a nonce N 2 and Step 3: the RSU generates a nonce N 2 , encrypts and aggregates it
calculates Rnew
i
¼ PUFðC new
i
Þ. It encrypts M2 ¼ EncRi fRnew ; N 1 ; N 2 g,
i
in one message M RG ¼ EncRnew
R
fIDR ; M 1 ; N 2 g,
A A A A
V RG ¼ hðIDR jjM RG kRnew new
R kN 2 Þ, and sends PIDRSU ; M RG ; V RG to the RSU
calculates the verification code V 2 ¼ HðIDV A jjM 2 kN 1 kN 2 kRiA Þ, and
gateway.
sends M 2 and V 2 to the Trusted Authority.
Step 4: Until the final step of the authentication phase, the ser-
Step 3: Upon receiving M 2 and V 2 , the Trusted Authority
i i ver recognizes the party as the legitimate user. Therefore, Aman
decrypts M 2 ¼ DecRi fRnew
A ; N 1 ; N 2 g using RiA and obtains Rnew
A ; N1 , et al.’s protocol fails to withstand an impersonation attack.
A

and N 2 . The Trusted Authority then verifies V 2 , updates the


i
pseudo-identity of vehicle PIDnew
A ¼ HðIDV A jjN 2 kRnew
A Þ, generates a 5.3. The protocol fails to achieve secret session key
nonce N3 , computes the verification code
V 3 ¼ HðIDV A jjN 2 kN 3 kRnew
i
Þ, and sends V 3 to vehicle A. Finally, the The adversary decrypts DecRi fT 1 g using Ri1 and obtains T 1 ,
A 1

Trusted Authority stores a new pseudo-identity, a new challenge, where T 1 ¼ fIDV 1 ; length; SK 1 ; P; N 4 g, hence the adversary obtains
and a new response of vehicle PIDnew new
; Rnew
i i
current session key SK 1 . Additionally, the adversary has the ability
A ; CA A .
Step 4: Vehicle A receives V 3 and verifies it. Subsequently, the to impersonate the vehicle and obtain future session key SK new . For
vehicle A updates and stores a new pseudo-identity this reason, Aman et al.’s protocol fails to achieve session key
i security.
PIDnew
A ¼ HðIDV A jjN 2 kRnew
A Þ.
5.4. The protocol fails to achieve anonymity
5. The cryptanalysis of Aman et al.’s protocol
Anonymity must achieve three security properties; hiding the
This section demonstrates security weaknesses in Aman et al.’s actual identity, unlinkable and untraceable. Unfortunately, in
protocol when evaluated under the DY model (see Definition of Aman et al.’s protocol, after the attacker impersonates a legitimate
6
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

vehicle, the adversary also can obtain new challenges (as shown in freshness to indicate that the message has not been used
5.1 and 5.2). The proof that the protocol of Aman et al. fails to before, ensuring secure mutual authentication.
achieve anonymity based on linkable and traceable messages is c. Perfect Forward and Backward Secrecy (PFBS): The protocol
as follows. must guarantee that both old and future secrets are impos-
Step 1: the adversary eavesdrops on all messages M 1 ; V 1 ; M 2 ; V 2 , sible for attackers to obtain.
V 3 in a public channel. d. Session Key Security: The protocol must provide that the
Step 2: the adversary steals the vehicle and activates the PUF by session key remains confidential. Furthermore, the session
programming without separating PUF from OBU. key must be updated for each session, and there should be
Step 3: the adversary computes RiA ¼ PUFðC iA Þ based on C iA in no relationship between past and future session keys.
i e. Physical Security: Even if an attacker carries out a dangerous
message M 1 ¼ PIDiA ; C iA ; EncRi fC new
A ; N 1 g. The adversary decrypts
A attack, such as stealing the device/smart card and conduct-
i i
DecRi fC new
A ; N 1 g and obtain new challenge C new
A . ing a side-channel attack, they should be unable to access
A

Step 4: the attacker compares the new challenge with all mes- secret data. This is due to the protocol’s protection of the
sages in the new session. If the new challenge is equal to the new secret data. Additionally, the protocol’s authentication pro-
challenge eavesdropped from the public channel, thus the adver- vides an unclonable device.
sary knows that the new challenge is sent by the same vehicle.
The proof above shows that Aman et al.’s protocol cannot 6.2. Notations of cryptography Function
achieve anonymity.
Furthermore, the protocol suffers from the insecurity of the This subsection presents the notations of the cryptography
RSU. The evidence of the RSU’s insecurity is as follows: functions throughout this paper. The cryptography functions are
shown in Table 2.
5.5. The protocol is vulnerable to stolen RSU attacks
6.3. Registration phase via secure channel
Step 1: The attacker obtains all data from the public channel,
including MRG ¼ EncRi fIDR ; M 1 ; M 2 ;    ; M n ; N 2 g. There are three registrations that take place before the protocol
R
is deployed: one between the vehicle and the trusted authority,
Step 2: The attacker steals the RSU.
another between the RSU gateway and the trusted authority, and
Step 3: The attacker extracts private data from the memory of
n o a third between the RSU and the RSU gateway. The registration
the RSU, which is C iR ; C iemR ; PIDRSU ; PIDemRSU . phases are as follows.
Step 4: The attacker activates the PUF and generates
RiR ¼ PUFðC iR Þ. 6.3.1. The registration between vehicle and Trusted Authority
Step 5: The attacker decrypts M RG , where In this subsection, we describe the registration of the vehicle
MRG ¼ DecRi fIDR ; M 1 ; M 2 ;    ; M n ; N 2 g, to obtain M 1 ; M 2 ;    ; M n . and trusted authority as follows:
R
Step 1: Vehicle A sends its identity and registration request
The attacker obtains all data, including the private data IDR and
M1 : fIDV ; RegistrationReq to the Trusted Authority.
RiR . Therefore, the protocol is vulnerable to a Stolen RSU attack.
Step 2: The Trusted Authority generates PIDV , kAS , C iA , computes
synchronization of key, pseudo-identity, and challenge, where
5.6. The protocol cannot achieve anonymity of RSU   n o
K V Syn ¼ kv 1 ; kv 2 ;    ; kv n , PIDV Syn ¼ pidv 1 ; pidv 2 ;    ; pidv n ,
 
After the attacker generates RiR ¼ PUFðC iR Þ and decrypts C V Syn ¼ cv 1 ; cv 2 ;    ; cv n . The Trusted Authority issues the smart
MRG ¼ DecRi fIDR ; M 1 ; M 2 ;    ; M n ; N 2 g, the attacker obtains the iden- card, the smart card stores PIDV ; kAS ; K V Syn ; PIDV Syn , then it sends
R

tity of the RSU. Therefore, the protocol fails to achieve anonymity M2 : fSmart card;C iA ; C V Syn g to Vehicle A.
for the RSU.
Step 3: The user inserts the smart card into the smart card
reader, and the vehicle generates the response and synchronization
 
6. Proposed authentication protocol of response RiA ¼ PUF C iA ; RV Syn ¼ PUFðC V Syn Þ. The user imprints his/
her biometricBIOuser , and the device computes:
This section presents our proposed protocol, including The ðK Bio ; hdv Þ ¼ FE:GenðBIOuser Þ, computes verification code for user
security mechanism, the cryptographic notations, assumptions, authentication V ¼ hðhdv kK Bio kIDV Þ, the vehicle encrypts every cre-
registration phase, mutual authentication, and CRP updates. dential PIDV ¼ PIDV  hðK Bio Þ, PIDV Syn ¼ PIDV Syn  hðK Bio Þ,

6.1. The security mechanism K V Syn ¼ K V Syn  hðK Bio Þ, IDV =IDV  hðK Bio Þ, kAS =kAS  hðK Bio Þ. The smart

card stores ; hdv , IDV , PIDV , kAS , K V Syn ; PIDV Syn . Finally, the vehicle
We define the security mechanism, where our protocol must sends M 3 : fRiA ; RV Syn g to the Trusted Authority.
fulfill security requirements to claim that it provides security in
Step 4: After receiving M 3 : fRiA ; RV Syn g, the Trusted Authority
the IoVs field. The security mechanism is as follows:
stores PIDV ; kAS ; C iA ; RiA ; K V Syn ; PIDV Syn ; C V Syn ; RV Syn .
a. Anonymity, Untraceability, Unlinkability: The protocol must
ensure that external entities do not possess knowledge of 6.3.2. The registration between the RSU gateway and trusted authority
the actual identities of the parties involved. It must prevent This subsection describes the registration of the RSU gateway
the tracing of the owner of transmitted messages in the and Trusted Authority as follows:
communication channel. Moreover, attackers should not be Step 1: The RSU gateway sends its identity and registration
able to link every message published in the public channel. request M1 : fIDG ; RegistrationReq g to the Trusted Authority.
b. Mutual Authentication: The protocol must ensure that the Step 2: The Trusted Authority generates the shared key, syn-
parties can ascertain that the communication is occurring chronization of the shared-key kGS ,
between the correct parties. Additionally, it must achieve K GSSyn ¼ fkGSSyn1 ; kGSSyn2 ;    ; kGSSynn g, the Trusted Authority copies
7
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

Table 2
Notations Of Cryptography Functions.

Notation Definition
PIDV j Pseudo identity each vehicles
PIDRSU Pseudo identity of RSU
IDR Identity of RSU
IDG Identity of the RSU gateway
IDS Identity of the Trusted Authority
IDV Identity of the vehicle
kAS The Shared key between vehicle and trusted authority
kGS The Shared-key between RSU Gateway and trusted authority
kRG Shared key between RSU and RSU gateway
ðC iA ; RiA Þ Challenge response pairs of device
PUF Physical Unclonable Function of device
PUF RSU Physical Unclonable Function of RSU
0
ðC iRSU ; RiRSu Þ Challenge-Response pairs of RSU
K V Syn The Shared-key synchronization between vehicle and trusted authority used if
 
occurs desynchronization, where K V Syn ¼ kv 1 ; kv 2 ;    ; kv n
PIDV Syn The Pseudo-identity synchronization between vehicle and trusted authority used
 
if occurs desynchronization, where PIDV Syn ¼ pidv 1 ; pidv 2 ;    ; pidv n
ðC V Syn ; RV Syn Þ The Challenge-Response Pairs synchronization between vehicle and trusted
 
authority used if occurs desynchronization, where C V Syn ¼ cv 1 ; cv 2 ;    ; cv n and
 
RV Syn ¼ r v 1 ; r v 2 ;    ; rv n ,
K GSSyn The Shared-key synchronization between RSU gateway and trusted authority
used if occurs desynchronization, where K GSSyn ¼ fkGSSyn1 ; kGSSyn2 ;    ; kGSSynn g
K RGSyn The Shared-key synchronization between RSU and RSU gateway used if occurs
desynchronization, where K RGSyn ¼ fkRGSyn1 ; kRGSyn2 ;    ; kRGSynn g
PIDRSUSyn The Pseudo-identity synchronization between RSU and RSU gateway used if
occurs desynchronization, where PIDRSU Syn ¼ fpidrSyn ; pidrSyn ;    ; pidrSyn g
1 2 n
0
ðC RSU Syn ; RiRSUSyn Þ The Challenge-Response Pairs synchronization between RSU and RSU gateway
 
used if occurs desynchronization, where C RSUSyn ¼ crsu1 ; crsu2 ;    ; crsun and
0
RiRSU Syn ¼ fr 0RSU1 ; r 0RSU 2 ;    ; r0RSUn g
BIOuser User’s biometric
FE Fuzzy extractor
FE:Gen FE:Gen generates the key from biometric input, where ðK Bio ; hdv Þ ¼ FE:GenðBIOuser Þ
FE:Rec FE:Rec reconstructs the key from Biometric and help data, where
K Bio ¼ FE:RecðBiouser ; hdv Þ
FERSU RSU is equipped Fuzzy extractor
FE:Gen generates the key from the response input, where
0
ðfkRSU ; hdRSU Þ ¼ FE:GenðRiRSu Þ, FE:Rec reconstructs the key from the response and
 0 
help data, where fkRSU ¼ FE:Rec RiRSu ; hdRSU .

n o
and stores: IDG , kGS , K GSSyn , then sends M 2 : kGS ; K GSSyn to the RSU C enc
RSU Syn ¼ C RSU Syn  hðK RGSyn Þ. The RSU stores kRG , K RGSyn ; PIDRSU Syn ,
i 0 0
gateway. C enc enc i i
RSU ; C RSU Syn , and sends the responses RRSu ; RRSU Syn to the RSU
n o
Step 3: The RSU gateway receives M 2 : kGS ; K GSSyn from the gateway.
Step 4: After receiving the responses from the RSU, the RSU
Trusted Authority, and stores them.
i0 0
gateway stores IDR , kRG , K RGSyn ; PIDRSUSyn ; C iRSU ; RRSu ; C RSUSyn , and RiRSUSyn .
6.3.3. The registration between the RSU and the RSU gateway All registration phases go through a secure channel. Addition-
In this subsection, we describe the registration process between ally, to avoid desynchronization problems and DoS attacks, the
the RSU and the RSU gateway as follows: proposed protocol provides the synchronization of pseudo-
Step 1: The RSU sends M1 : fIDR ; RegistrationReq g to the RSU identity, shared keys for every participant, and synchronization
gateway. of challenge-response pairs between the vehicle and the trusted
authority.
Step 2: The RSU gateway generates kRG , PIDRSU , C iRSU ,
K RGSyn ¼ fkRGSyn1 ; kRGSyn2 ;    ; kRGSynn g,
PIDRSUSyn ¼ fpidrSyn ; pidrSyn ;    ; pidrSyn g, 6.4. Mutual authentication and session key establishment
 1
2 n

C RSUSyn ¼ crsu1 ; crsu2 ;    ; crsun . Then, it stores.


This subsection describes the authentication phase; every step
IDR , kRG , K RGSyn ; PIDRSUSyn , C iRSU , C RSUSyn , and sends
n o from the authentication phase is as follows.
i
M 2 : kRG ; K RGSyn ; PIDRSUSyn ; C RSU ; C RSUSyn to the RSU, Step 1: the user inserts the smart card into the smart card
0
Step 3: The RSU generates responses RiRSu ¼ PUF RSU ðC iRSU Þ, reader, then the user imprints his/her fingerprint Biouser , and the
0 device generates K Bio ¼ FE:GenðBiouser ; hdv Þ, computes
RiRSUSyn ¼ PUF RSU ðC RSUSyn Þ. It then encrypts the challenge and chal-
IDV ¼ IDV  hðhdv kK Bio Þ, computes verification code
lenge synchronization using the shared key and shared key syn- V  ¼ hðhdv jjK Bio kIDV Þ, checks V  ¼ ?V, if the verification V  is not
i
chronization, respectively. C enc i
RSU ¼ C RSU  hðkRG Þ, equal to V, the user authentication will be denied. Otherwise, the
8
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721


vehicle decrypts PIDV ¼ PIDV  hðK Bio Þ, kVS ¼ kVS  hðK Bio Þ; gener- shared key to PIDnew
V
new
¼ hðIDV A kRiA Þ and kVS ¼ hðkVS kRiA Þ respectively.
ates a nonce N 1 . The vehicle sends M 1 : fPIDV ; N 1 g to the RSU. It also encrypts PIDVnewenc
¼ PIDnew  hðK Biou Þ and
V
Step 2: The RSU generates a nonce N 2 , decrypts newenc new enc newenc
enci
kVS ¼  hðK Biou Þ, and stores
kVS PIDnew
and kVS . Additionally,
V
C iRSU ¼ C RSU  hðkRG Þ to obtain a challenge, generates the response the trusted authority also updates its pseudo-identity and shared
0
based on a challenge RiRSu ¼ PUF RSU ðC iRSU Þ, generates fixed-key key for future sessions to PIDnew ¼ hðIDV A kRiA Þ and kVS ¼ hðkVS kRiA Þ.
new
0
V

ðfkRSU ; hdRSU Þ ¼ FE:GenðRiRSu Þ, computes ¼ hdRSU  hðkRG Þ
hdRSU In the final session, each participant stores a new credential for
encrypts and aggregates in one message that is future sessions. We consider updating the credential to achieve
MRG ¼ ðIDR ; M 1 ; N 2 Þ  hðfkRSU Þ, computes verification code perfect forward and backward secrecy.

V RG ¼ hðIDR jjM RG kkRG jjN 2 kPIDRSU khdRSU Þ. The RSU sends
  
M2 : PIDRSU ; M RG ; hdRSU ; V RG to the RSU gateway. 6.5. The renewal CRP between vehicle and trusted authority

Step 3: The RSU gateway computes hdRSU ¼ hdRSU  hðkRG Þ to
 0 
obtain help data and reconstructs fkRSU ¼ FE:Rec RiRSu ; hdRSU This subsection describes the updates of challenge-response
pairs (CRPs). The steps for updating CRPs are as follows:
using fuzzy-reconstruction to obtain a fixed key. Later, it decrypts
Step 1: The Trusted Authority generates a nonce N 1 and selects
ðIDR ; M 1 ; N 2 Þ ¼ M RG  hðfkRSU Þ and obtains IDR ; M 1 ; N 2 , verifies V RG  
i
G. If the verification of V RG fails, the communication will be discon- C A ; RiA for vehicle A with IDA . The Trusted Authority selects PIDV A
tinued. Otherwise, the RSU gateway generates a nonce N 3 , encrypts and encrypts the challenge using the shared key
and aggregates it in one message that is C enc i
A ¼ C A  hðkVS kN 1 Þ. The Trusted Authority considers updating
MGS ¼ ðIDG ; M 1 ; N 3 Þ  hðkGS Þ, computes the verification code the challenge by generating a new challenge C new A . The Trusted
V GS ¼ hðIDG jjIDS kM GS kkGS kN 3 Þ, and sends M 3 : fIDG ; M GS ; V GS g to enc
Authority encrypts C new ¼ C new  hðRiA Þ, computes the verification
the Trusted Authority. A  A
 enc

Step 4: The Trusted Authority decrypts code V 1 ¼ hðIDV A jjN 1 kRiA jC iA kPIDV A kC new
A Þ, and sends PIDV A ; C enc
A ,

ðIDG ; M 1 ; N 3 Þ ¼ M GS  hðkGS Þ and obtains IDG ; M 1 ; N 3 , verifies V GS . If C new


enc
, N 1 , and V 1 to vehicle A.
A
the verification of V GS fails, the communication will be discontin- enc
Step 2: After receiving PIDV A ; C enc new
A , CA , N 1 , and V 1 , vehicle A
ued. Otherwise, the Trusted Authority generates a nonce N 4 and
  decrypts C iA ¼ C enc i
A  hðkVS kN 1 Þ and obtains C A . Vehicle A generates
reads CRP C iA ; RiA for PIDV . The Trusted Authority considers gener-
RiA ¼ PUFðC iA Þ and verifies V 1 . If the verification of V 1 fails, the com-
ating the Token, where TokenA ¼ fIDV A ; length; SK V A ; Tmexpire ; N 4 . The munication will be terminated. Otherwise, vehicle A decrypts
length and Tmexpire denote the length of the token and the expira- C new ¼ C new
enc
 hðRiA Þ and obtains the new challenge. Later, vehicle
A A
tion time of the token, respectively. The Trusted Authority com-
A generates a new response Rnew A ¼ PUFðC new
A Þ, encrypts the new
putes V SAA ¼ hðIDV A jjTokenA kRiA jjN 1 kN 4 Þ, C encrypt ¼ C iA  hðIDV A kkAS Þ,
encrypt
A
encrypt
response using RiA and kVS : N 2 ¼ Rnew
A  hðRiA kkVS Þ. Vehicle A updates
TokenA ¼ TokenA  hðRiA Þ, M SAA ¼ fC encrypt
A ; TokenA ; V SAA g, the pseudo-identity and shared key for future sessions, encrypts
MSG ¼ ðTokenA ; M SAA ; N 4 Þ  hðkGS Þ, and updates the new shared key enc newenc new
PIDnew
V ¼ PIDnew
V  hðK Biou Þ and kVS ¼ kVS  hðK Biou Þ, and stores
between the RSU gateway and the Trusted Authority by generating newenc newenc
new newenc new PIDV ; kVS . Vehicle A sends N 2 and V 2 to the Trusted Authority.
kGS w. It encrypts kGS ¼ kGS  hðkGS kN 3 Þ. The Trusted Authority
new Step 3: The trusted authority decrypts Rnew ¼ N 2  hðRiA kkVS Þ and
computes the verification code V SG ¼ hðIDG jjIDS kMSG jjN 3 kN 4 kkGS Þ new
A

newenc
obtains the new response RA . The trusted authority then verifies
and sends M4 : fM SG ; V SG ; kGS g to the RSU gateway. V 2 and updates the pseudo-identity and shared key between vehi-
Step 5: The RSU gateway decrypts cle A and the trusted authority. The updated values of
ðTokenA ; M SAA ; N 4 Þ ¼ M SG  hðkGS Þ and obtains TokenA ; M SAA ; N 4 , PIDnew
new
¼ hðIDV A kRiA Þ and kVS ¼ hðkVS kRiA Þ will be used for future ses-
V
new newenc
kGS ¼ kGS  hðkGS kN 3 Þ, verifies V SG . If the verification of V SG fails, sions. Finally, the trusted authority stores the new values of
new
the communication will be terminated. Otherwise, the RSU gate- C new new new
A ; RA ; PIDV , and kVS .
way encrypts and aggregates TokenA ; M SAA ; N 3 in one message, The renewal of CRPs is also performed between RSU and RSU
M GR ¼ ðTokenA ; MSAA ; N 3 Þ  hðfkRSU Þ. The RSU gateway updates the gateway in a similar way as mentioned above. Finally, the RSU
shared-key and pseudo-identity between RSU and the RSU gate- gateway stores the new {challenge-response pairs}, which will be
newenc new used for the next communication session.
way: kRG ¼ kRG  hðfkRSU kN 2 Þ and PIDnew
RSU ¼ hðIDR jjfkRSU kN 2 Þ. It
then computes the verification code,
newenc new
V GR ¼ hðIDG jjM GR kkRG jjIDR kN 2 kkRG Þ, and sends 7. Discussion and result
newenc
M 5 : fMGR ; kRG ; V GR g to all RSUs in its area. The RSU gateway
new new
stores kGS , kRG , and PIDnew This section discusses an informal analysis based on solid intu-
RSU for future sessions.
Step 6: The RSU decrypts ðTokenA ; M SAA ; N 3 Þ ¼ M GR  hðfkRSU Þ and itive reasoning to prove that our proposed protocol achieves secu-
new newenc rity features and withstands various attacks. In addition, we use a
obtains TokenA ; M SAA ; N 3 . It then decrypts kRG ¼ kRG  hðfkRSU kN 2 Þ
mathematical model, the Random Oracle Model (RoR model), and a
and verifies V GR . If the verification of V GR fails, the communication
programming model (Scyther tool) for formal analysis to
will be terminated. Otherwise, the RSU updates the pseudo-
strengthen the informal study and provide evidence that our pro-
identity between RSU and the RSU gateway:
new
tocol is secure. The analyses are as follows:
PIDnew new
RSU ¼ hðIDR jjfkRSU kN 2 Þ and stores the new PIDRSU and kRG for
future sessions. The RSU sends M 6 : fMSAA g to the vehicle.
7.1. Informal analysis
Step 7: Vehicle A computes C iA ¼ C encrypt
A  hðIDV A kkAS Þ, generates
encrypt
RiA ¼ PUFðC iA Þ, computes TokenA ¼  hðRiA Þ, verifies
TokenA This subsection presents the security feature analysis and pro-
V SAA . If the verification of V SAA fails, the communication will be ter- posed protocol competencies to withstand the type of attacks.
minated. Otherwise, the vehicle A updates its pseudo-identity and The details are as follows:

9
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

a. Achievement of Mutual Authentication tication phase. In addition, RSU changes PIDRSU ; kRG by
pidrSyn ; kRGSyn1 . Afterward, the RSU gateway replaces kGS with
1
Mutual authentication between the vehicle and trusted author- kGSSyn1 , and the authentication phase is repeated. As a result, the
ity is achieved by possessing the shared key kVS and RiA ,
where the strategy can address the desynchronization problem.
trusted authority authenticates the identity of the vehicle IDV A and
response RiA , and the vehicle authenticates V SAA , while checking a g. Withstanding Impersonation Attack
nonce to ensure message freshness. Additionally, mutual authenti-
cation between the RSU and RSU gateway can be achieved, where The attacker cannot obtain the current secrets {response, ses-
the RSU authenticates the parameters kGR and fkRSU , and the RSU sion key} and new secrets {response, session key} even if the
gateway authenticates kGR and fkRSU . Later, mutual authentication attacker stole the smart card and activated PUF; hence, the
between the RSU gateway and the trusted authority can be attacker cannot impersonate the smart card as a legitimate user.
achieved by possessing the shared key kGA , where the trusted Therefore, our protocol authentication withstands an Imperson-
authority authenticates V GS and the RSU gateway authenticates ation attack.
V SG . Therefore, our protocol achieves mutual authentication.
h. Withstanding Tracking Attack
b. The Achievement of Anonymity and untraceability
In addition to achieving anonymity, the challenge C A in our
In our authentication protocol, the vehicles and RSU provide a scheme is protected by a shared key kVS . Hence, the challenge is
one-time pseudonym PIDV and PIDRSU , respectively, to interact with private, making it impossible for the attacker to know the chal-
the other parties; hence, the attacker cannot obtain the true iden- lenge owner based on the collected data in the wireless-based
tity even if the message is intercepted in a public channel. As a public channel. Therefore, the scheme can withstand a tracking
result, our system is anonymous and untraceable. Furthermore, attack.
every parameter is refreshed every session; hence, no message is
linked to another; thus, our protocol achieves unlinkability. i. Withstanding the DoS Attack

c. The Achievement of PFBS Even if the attacker spoofs communication traffic, gets all data
from the channel, and makes it seem like communication appears
In our protocol, the session key, challenge-response, and shared to stop. Our protocol is designed primarily in the condition that
key change every session, and also smart card stores non-plaintext communication may suddenly stop. Then, one participant initiated
in its memory. Even if the attacker steals and reveals current data sending the synchronization {identity, challenge-response, key} to
from the memory of the smart card, the attacker cannot obtain past ensure the protocol was still operating. Suppose one participant,
secrets, including the session key, because the smart card stores e.g., the server, does not receive the message at any specific time

non-plaintext data fV; hdv , IDV , PIDV , kAS , K V Syn ; PIDV Syn } protected from any participants. In that case, the protocol can operate by ini-
by K Bio , where K Bio is generated from user’s biometric. In addition, tiating the server sends {pseudo identity, key, challenge-response}
the attacker also cannot get the future secret data. Therefore, our synchronization. Hence, Our protocol can still work under a DoS
scheme achieves PFBS. attack. Therefore, our protocol withstands DoS attacks.

d. The Achievement of Session key security j. Withstanding replay attack

The session key in our protocol differs for each session, and Even if an attacker intercepts a message M 1 fPIDV ; N 1 g and
there is no relationship between them. Even if the attacker resends it to the server, the server quickly detects that the message
acquires one of the session keys, he or she will be unable to access has been modified or that it is still past data based on the new
prior or future session keys. As a result, our authentication process pseudo-identity PIDnew
V and new nonce.
is secure using a session key.
k. Withstanding a man-in-the-middle (MIM) attack.
e. The Achievement of Physical security
Our protocol can withstand MIM attacks. For example, if the
The smart card is equipped with PUF, based on the operation of attacker intercepts M 1 fPIDV ; N 1 g and modifies a message
PUF where R ¼ PUF ðC Þ and using challenge-response to achieve M 1 fPIDxx ; N xx g, the server can quickly identify the message as inva-
mutual authentication. Additionally, based on the uniqueness of lid based on the incorrect pseudo-identity PIDxx used by the
PUF, it cannot be cloned. Therefore, our scheme can withstand attacker.
cloning attacks. Furthermore, RSU is equipped with PUF and a
Fuzzy extractor. Rely on probability function at fuzzy extractor l. Withstanding Cloning Attack
generator; even if the adversary can fully control and obtains the
shared key, the adversary cannot obtain fixed-key simultaneously The smart card and RSU are equipped with PUF, which guaran-
in one operation. Therefore, our proposal, especially RSU, with- tees that they are unclonable devices.
stands a physical attack. In addition, to safeguard from side-
channel attacks, we can use the masking method (Kim et al., m. Withstanding Side-Channel Attack
2019; Prouff & Rivain, 2013). Hence, the adversary cannot extract
the data from memory. Our protocol protects the secret stored in memory using a bio-
metric key. Even if the attacker gets or steals the smartcard and
f. Resolve the desynchronization problem attempts to collect data from memory, they cannot retrieve the
actual secret data due to the optimized usage of biometrics in
When desynchronization occurs, the vehicle substitutes PIDV , our protocol. Therefore, our protocol can withstand side-channel
kVS with pidv 1 , kv 1 , Then conduct the identical step with the authen- attacks.

10
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

n. Withstanding Ephemeral Secret Leakage (ESL) attack hðÞ and the security of PUF ðÞ. They are both modeled under a ran-
dom oracle.
Our protocol closes the loophole to ESL attacks, where the
attacker cannot obtain every secret using biometrics and it is 7) Security Proof
impossible for the attacker to retrieve the biometric key. Addition-
ally, every secret is updated in every session, ensuring that past The security of the hash function and PUF is referred to as
and future secrets do not have a relationship. semantic security, which we use to claim that our protocol is
secure. The security of the hash function and PUF is discussed in
7.2. Analysis formally using RoR model Subsection 3.2.
By employing both of these definitions and acknowledging
Our RoR model has four participants: the vehicle (VC i ), the RSU Zipf0 s law (D. Wang et al., 2017), the proposed protocol0 s semantic
(Ri ), the RSU gateway (RGi ), and the trusted authority ðSÞ. The security can be determined by Theorem 1. Specifically, Theorem 1
model is as follows: determines the level of security provided by the protocol in terms
of semantic security.
RGi , and pS are the oracle of v c; r; rg; s
1) Participants: pvVCc i , prRi , prg s

related with vehicle (VC i ), RSU (Ri ), RSU gateway (RGi ), the Theorem 1. Suppose the adversary conducts an attack on protocol
trusted authority ðSÞ P in polynomial time t under the RoR model. In our protocol, the
2) Partnering: Partnering is only possible if, 1. Communication identity and biometric key are kept confidential. According to Zipf’s
between Vehicle pvVCc i and Trust Authority psS based on the law (D. Wang et al., 2017), l1 denotes the bit of the biometric key,
possession of the session id. 2. Updating each message sent and l2 denotes the key fkRSU . The advantages of an attacker
and received by Vehicle pvVCc i and Trust Authority psS in every breaking the semantic security for revealing the session key
session. between the vehicle and trust authority are as follows:
3) Freshness: If the attacker cannot get the session key SK
q2h q2 0 q q
between Vehicle pvVCc i and Trust Authority psS , the freshness Adv P;A ðtÞ  þ P þ 2maxfC 0 :qss ; ls ; ls g
AKE
jHashj jPUF j 21 22
can be achieved.
4) Adversary: The capabilities of Adversary A in the ROR model The definitions of the notations are as follows: qh represents
are identical to Dolev-Yao (DY) adversary model (Dolev & a query of the hash function, qp represents a query of the PUF,
Yao, 1983), and getting access to ROR model queries is as and qs represents a query of sending a message. jHashj repre-
follows: sents the size of the range space of the hash function, and
a. Executeðpv c ; ps Þ: By executing this query, the Adversary can jPUF j represents the size of the range space of the PUF. C 0 and
intercept any messages sent across the public channel s0 are Zipf’s parameters (D. Wang et al., 2017) that determine
between the vehicle and the trusted authority. This is called the distribution of the bits in the biometric key and the RSU
an eavesdropping attack. key, respectively.
b. Sendðpv c ; mÞ: By performing this query, the attacker can
send and receive messages as one of the participants in the
Proof. We follow the same proof as offered (Gope et al., 2019),
communication session, such as one vehicle. This is referred
(Roy et al., 2020), and (Roy et al., 2017), and in this work. Gj signi-
to as an active attack.
fies five-step games, j 2 ½0; 4 where j is between 0 and 4. In the
c. CorruptSmartCardðpv c Þ: The capability of the attacker is
game Gj , The adversary achieves ‘‘SUCCESS” if the adversary
where the attacker pilfers the smart card and gets the confi-
dential data stored in the memory of the smart card. This is guesses the correct c. The game’s specifics are as follows.
known as a stolen or lost smart-card attack. Game G0 : G0 indicates the adversary starts the game to break
d. CorruptRoadSideUnitðpRSU Þ: The adversary has complete con- protocol security under the RoR model. The equation is:
trol of the RSU and may reveal the credential stored in the
Adv P;A ðtÞ ¼ j2:Pr½SUCCESS0   1j
AKE
RSU’s memory. This is known as a stolen RSU attack. ð1Þ
e. Testðpv c ; ps Þ: The security is achieved if the session key Game G1 : The adversary intercepts messages by operating
between vehicle pv c and trust authority ps is secure under
Executeðpv c ; ps Þ: M 1 : fPIDV ; N 1 g,

M 2 : fPIDRSU ; MRG ; hdRSU V RG g,
RoR model (Abdalla et al., 2005). The adversary guesses, by newenc newenc
repeatedly flipping the coin for a particular time, whether M3 : fIDG ; MGS ; V GS g, M 4 : fMSG ; V SG ; kGS g, M 5 : fMGR ; kRG ; V GR g,
the guessing c = 1, the session key is fresh. Otherwise, it is and M6 : fMSAA g. Afterwards, the adversary guesses the session
a null or random key. key. In our protocol, the session key is protected by RiA . If the
5) Semantic Security Of Session Key attacker wants to obtain the session key, they must disclose the
confidential data: RiA , IDV , and kAS . Unfortunately, only a legitimate
In the RoR model, the adversary has the ability to differentiate vehicle can compute RiA , and the trusted authority knows RiA . Only a
between the correct session key and the random key. Adversary legitimate vehicle and the trusted authority can compute SK. As a
A will submit many TestðÞ queries to either pv c or ps , with the result, Adversary A’s chances of winning in G1 through eavesdrop-
results recorded in bit c. If c ¼ c0 , the opponent wins the game, ping are not increased. We have monitored the outcome
where c0 denotes a randomly chosen bit. accordingly.
Adv P;A ðt Þ ¼ j2:Pr½SUCCESS  1j denotes Adversary A’s attempt to
AKE
Pr½SUCCESS1  ¼ Pr½SUCCESS0  ð2Þ
break the semantic security of the protocol authentication in time
t, where SUCCESS represents the adversary winning the game. Game G2 : The attacker simulates Sendðpv c ; mÞ; the adversary
accepts and edits the message. All messages
6) Random Oracle {MA1 ; M A2 ; M A3 ; M A4 ; MA5 ; M A6 } are protected by a key fkRSU and hash
function. According to 3.2.1-Definition 1, even if the adversary
In this article, it is shown that all participants and adversary A operates the send queries, the adversary cannot locate a collision.
have access to the security of the collision-one-way hash function The equations based on the birthday paradox are as follows
11
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

q2h the smart card, and kRG , K RGSyn ; PIDRSUSyn ; C enc


i enc
jPr½SUCCESS2   Pr½SUCCESS1 j  ð3Þ RSU , and C RSU Syn from RSU.
2jHashj
The adversary cannot get the biometric key K Bio and the key fkRSU .
Game G3 : This game follows a similar rationale to G2 , but the According to (D. Wang et al., 2017), the probability of guessing
difference is that G3 simulates the Send and PUF requests, whereas both keys {K Bio as l1 and fkRSU as l2 ) are 21l1 and 21l2 , respectively.
G2 does not. We obtain the equation according to the definition of a Therefore, games G3 and G4 resist guessing attacks. Thus, the result
Secure PUF (as shown in 3.2.2-Definition 2): is:
q2P 0 qs qs
jPr½SUCCESS3   Pr½SUCCESS2 j  ð4Þ jPr½SUCCESS4   Pr½SUCCESS3 j  maxfC 0 :qss ; ; g ð5Þ
2jPUF j 2l1 2l2
Game G4 : In this final game, the adversary conducts The adversary must obtains c0 equal to c to win the game in
CorruptSmartCardðpv c Þ and CorruptRoadSideUnitðpRSU Þ, where the game G4 . As a result, we have

adversary A can extract V; hdv , IDV , PIDV , kAS , K V Syn ; and PIDV Syn from

Fig. 2. The result of the Scyther Tool.

12
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

1 rity properties, attack resistance, and computing complexity. The


Pr½SUCCESS4  ¼ ð6Þ
2 following table compares the two options:
Based on Eqs. (1), (2), and (6), we have Table 3 shows that P. Gope et al.’s (Gope et al., 2019), U. Chat-
  terjee et al.’s (Chatterjee et al., 2019), and M. N. Aman et al.’s pro-
1  1
Adv P;A ðt Þ ¼ Pr½SUCCESS0    tocol (Aman et al., 2021) achieves SP2, SP6, SP10 and SP12.
AKE
2 2 However, their protocols fail to achieve SP1, SP3-SP5, SP7-SP9,
  and SP11. J. Wang et al.’s protocol achieves SP3 and SP11, but it
 1 cannot achieve SP1, SP5, SP9, and SP12. Additionally, their protocol

¼ Pr½SUCCESS1   
2 does not provide SP2, SP4, SP6-SP8, and SP10. On the other hand, H.
Zhang et al.’s protocol only achieves SP3. Only our protocol can
¼ jPr½SUCCESS1   Pr½SUCCESS4 j ð7Þ achieve all security properties (SP1-SP12).
Our demonstration was performed on a virtual machine run-
We use the triangle inequality applied to formulas (3), (4), and
ning Linux Ubuntu 20.4.1, using the computing resources of an
(5).
RSU and an RSU gateway. The CPU of the RSU and the RSU gateway
jPr½SUCCESS1   Pr½SUCCESS4 j  jPr½SUCCESS1   Pr½SUCCESS3 j is an Intel Core i7 with a clock speed of 950 GHz, and the RAM is
10 GB. The Smart Card used in the demonstration has a CPU core
þ jPr½SUCCESS3   Pr½SUCCESS4 j
with a clock speed of 798 MHz and a RAM of 256 MB, which is con-
 jPr½SUCCESS1   Pr½SUCCESS2 j sistent with the specifications of an actual Smart Card (Technology,
þ jPr½SUCCESS2   Pr½SUCCESS3 j n.d.). The cryptographic operations used in our protocol include the
PUF protocol proposed by (Gope et al., 2019), (Chatterjee et al.,
þ jPr½SUCCESS3   Pr½SUCCESS4 j 2019), (Aman et al., 2021), and (Herder et al., 2014), FE.Gen and
q2h q2 FE.Rec (Dodis et al., 2004) and (Frikken et al., 2009), channel-
 þ P
jHashj jPUF j based key agreement, and the Advanced Encryption Standard
0 qs qs Cryptosystem (AES-CBC) from the Java Cryptography Architecture
þ 2maxfC 0 :qss ; ; g (JCA) library (Oracle, n.d.).
2l1 2l2
This demonstration focuses on the computational time result-
ð8Þ
ing from each algorithm used in every protocol, including our
The final result based on formulas (7) and (8): own. The purpose of this demonstration is to determine which pro-
tocol has the lowest computational burden. After obtaining the
q2h q2 0 q q
Adv P;A ðt Þ 
AKE
þ P þ 2maxfC0:qss ; ls ; ls g results, we will finally compare the computational burdens. The
jHashj jPUF j 2 22
1
results have been shown in Table 4 and Fig. 3.
The execution time for each cryptographic algorithm is as fol-
7.3. Analysis formally using Scyther tool lows: the execution time of T PUF in the vehicle/smart card is
0.304 s, and in the Roadside Unit/Gateway, it is 0.147 s. T FE:Gen in
Our protocol was also verified using the Scyther tool, a Python- the vehicle/smart card is 1.76 s, and in the Roadside Unit/Gateway,
based utility (Cremers, n.d.), (Zhu et al., 2019). The results of the it is 0.965 s. T FE:Rec in the vehicle/smart card is 3.28 s, T Bio in the
Scyther tool indicate that no attacks were found, as shown in vehicle/smart card is 4.95 s, T s in the vehicle/smart card is
Fig. 2. The code for our protocol is shown in the appendix. 0.759 s, and in the trusted authority, it is 0.248 s. T h in the vehi-
cle/smart card is 0.318 s, in the Roadside Unit/Gateway, it is
7.4. Comparison with related protocol 0.168 s, and in the Trusted Authority, it is 0.108 s. T ECC in Smart
Card/Vehicle is 7.48 s, in the Roadside Unit/Gateway, it is 5.76 s,
The subsection compares our proposed approach to previous and in the Trusted Authority, it is 4.06 s.
protocol authentication schemes presented by (Gope et al., 2019), The Table 4 shows that P. Gope et al.’s authentication protocol
(Chatterjee et al., 2019), and (Aman et al., 2021) in terms of secu- (Gope et al., 2019) gets the computational complexity of vehicle

Table 3
Comparison of Security Features.

Security Properties Authentication Protocol


(SP)
P. Gope et al. U. Chatterje at al. M. N. Aman et al. J. Wang et al. H. Zhang et al. Our scheme
SP1 No No No No No Yes
SP2 Yes Yes Yes N/A N/A Yes
SP3 No No No Yes Yes Yes
SP4 No No No N/A N/A Yes
SP5 No No No No No Yes
SP6 Yes Yes Yes N/A N/A Yes
SP7 No No No N/A N/A Yes
SP8 No No No N/A N/A Yes
SP9 No No No No N/A Yes
SP10 Yes Yes Yes N/A N/A Yes
SP11 No No No Yes N/A Yes
SP12 Yes Yes Yes No No Yes

Where SP: Security Features, SP1: Anonymity and untraceability of the vehicle, SP2: Protect against man-in-the-middle attack, SP3: Forward Secrecy, SP4: Session Key
security, SP5: Physical security of device/smart card, and SP6: resolve the loss of synchronization problem, SP7: withstanding impersonation attack, SP8: withstanding
tracking attack, SP9: withstand stolen device/smart card attack. SP10: withstand the DoS attack, SP11: withstanding reply attack, and SP12: withstanding cloning attack.
‘‘Yes” indicates that the protocol achieves the security features, ‘‘No” indicates that the protocol does not achieve the security features, and ‘‘N/A” indicates that the protocol is
not applicable for providing security features.

13
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

Table 4
Comparison of Computational Complexity.

Protocol Computation Total


(s)
Smart Card/Vehicle Roadside Unit/Gateway Trust Authority
P. Gope et al. T Bio þ 6T h þ 5T s þ 2T PUF 11.261 s 5T h 0.84 s 6T h þ 5T s 1. 888 s 13.989 s
U. Chatterje at al. 10T h þ 4T s þ T PUF +3T ECC 26.098 s 4T h þ 4T ECC 23,712 s 8T h þ 5T s þ 4T ECC 18.344 s 68.154 s
M. N. Aman et al. 2T h þ T s þ T PUF 1.426 s T PUF þ T S þ T h 0.563 s 4T h þ 3T s 1.176 s 3.165 s
J. Wang et al. 6T ECC þ 6T h 46.788 s 3T ECC þ T h 17.448 s 8T ECC þ 5T h 46.62 s 110.856 s
H. Zhang et al. 4T ECC þ T h 30.238 s 4T ECC þ T h 23.208 s 4T ECC þ 2T h 16.456 s 69.902 s
Our scheme T h þ T FE:Rec þ T PUF 3.426 s T PUF þ T FE:Gen þ T h 2,075 s 4T h 0.432 s 5.933 s

Where T FE:Gen : Time of Fuzzy Extractor Generation, T FE:Rec : Time of Fuzzy Extractor Reconstructions, T Bio : Time of Biometric operations, T h : Time of one-way hash function,
T PUF : Time of PUF process, T s : Time of symmetric encryption/decryption. T ECC : Time of the ECC operation (point multiplication/point addition).

only our proposed scheme fulfills security properties SP1-SP12


and has lower computational complexity than Gope et al.’s, U.
Chatterjee et al.’s, J. Wang et al.’s, and H. Zhang et al.’s schemes.
Additionally, our article is supported by evidence from informal
and formal analyses. The results of the RoR model analysis show
that our protocol fulfills security features and protects against var-
ious attacks. Later, the Scyther Tool also confirms that our protocol
protects against attacks. Therefore, our proposed protocol is secure
and can withstand well-known attacks. The implementation of PUF
in the real world is rarely carried out, even though security features
related to cloning need to be addressed. Similarly, devices such as
smartphones and smart cards already extensively use biometrics
for security. From this, future research can explore how to apply
biometrics and PUFs in the real world. Another important aspect
Fig. 3. Diagram of execution time comparison.
is demonstrating how real-world machine-learning and side-
channel attacks can exploit PUF vulnerabilities, as well as the
T Bio þ 6T h þ 5T s þ 2T PUF , Roadside Unit/Gateway 5T h , and the necessity for countermeasures against them.
trusted authority 6T h þ 5T s with a total of the execution time of
13.989 s. U. Chatterjee et al.’s scheme (Chatterjee et al., 2019) Declaration of Competing Interest
has a computational complexity of vehicle/smart card
10T h þ 4T s þ T PUF +3TECC , Roadside Unit/Gateway 4T h þ 4T ECC and The authors declare that they have no known competing finan-
the trusted authority 8T h þ 5T s þ 4T ECC with a total execution time cial interests or personal relationships that could have appeared
of 68.154 s. M. N. Aman et al.’s scheme (Aman et al., 2021) gets the to influence the work reported in this paper.
computational complexity of the vehicle/smart card
2T h þ T s þ T PUF , Roadside Unit/Gateway T PUF þ T S þ T h and trusted Appendix
authority 4T h þ 3T s with a total execution time of 3.165 s. J. Wang
et al.’s authentication protocol has a total computational time of The coding of our proposed protocol
110.856 s. H. Zhang et al. achieve a total computational time of
69.902 s. Our proposal has a computational complexity of vehicle /*
T h þ T FE:Rec þ T PUF , Roadside Unit/Gateway T PUF þ T FE:Gen þ T h and * Our proposal
trusted authority 4T h with a total execution time of 5.933 s. Based */
on the diagram in Fig. 3, it is shown that our protocol has a lower // The protocol description
computational complexity and execution time than P. Gope et al.’s, usertype String;
U. Chatterjee et al.’s,. J. Wang et al.’s, and H. Zhang et al.’s authen- hashfunction H;
tication protocol, and our proposal has computational complexity const XOR:Function;
higher than M. N. Aman et al.’s scheme. However, only our protocol const CON:Function;
fulfills the achievement of all security properties SP1-SP12 in const PUF:Function;
Table 3. macro TokenA = CON(CON(CON(CON(IDVA, length), SKVA),
Tmexpire), N4);
8. Conclusion macro VSAA = H(CON(CON(CON(CON(IDVA,TokenA),RA),N1),
N4));
In this article, we initially introduce a robust adversary model macro CAenc = XOR(CA,H(CON(IDVA,kVS)));
by modifying the DY structure. Then, we review and evaluate macro TokenAenc = XOR(TokenA,H(RA));
Aman et al.’s protocol and identify the weaknesses of their proto- macro MSAA=(CAenc,TokenAenc,VSAA);
col. We propose a new protocol to resolve the problems in their macro RA = PUF(CA);
protocol and provide a solution for generic IoV security issues. protocol Internet-of-Vehicle(Vehicle,TrustedAuthority){
We compare our protocol with Aman’s and several related proto- role Vehicle{
cols regarding security features, the ability to withstand attacks, secret RA, CA, IDVA, kVS, TokenA, SKVA, length, Tmexpire;
and computational complexity. The comparison table shows that fresh N1, N4, PIDVA;

14
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

var PIDVA; claim(RSUGateway,Niagree);


send_1(Vehicle,TrustedAuthority, PIDVA,N1); claim(RSUGateway,Nisynch);
recv_2(TrustedAuthority, Vehicle,MSAA); claim(RSUGateway,Alive);
claim(Vehicle,Secret,TokenA); claim(RSUGateway,Weakagree);
claim(Vehicle,Secret,SKVA); }
claim(Vehicle,Secret,CA); role TrustedAuthority{
claim(Vehicle,Secret,RA); secret RA, CA, IDVA, kVS, TokenA, SKVA, length, Tmexpire,
claim(Vehicle,Niagree); IDG, IDS, kGS,kGSnew;
claim(Vehicle,Nisynch); fresh N1, N4, N3, PIDVA;
claim(Vehicle,Alive); var PIDVA;
claim(Vehicle,Weakagree); recv_1(RSUGateway,TrustedAuthority, IDG, MSG, VGS);
} send_2(TrustedAuthority, RSUGateway,MSG,VSG,kGSne
role TrustedAuthority{ wenc);
secret RA, CA, IDVA, kVS, TokenA, SKVA, length, Tmexpire; claim(TrustedAuthority,Secret,kGS);
fresh N1, N4, PIDVA; claim(TrustedAuthority,Secret,kGSnew);
var PIDVA; claim(TrustedAuthority,Niagree);
recv_1(Vehicle,TrustedAuthority, PIDVA,N1); claim(TrustedAuthority,Nisynch);
send_2(TrustedAuthority, Vehicle,MSAA); claim(TrustedAuthority,Alive);
claim(TrustedAuthority,Secret,TokenA); claim(TrustedAuthority,Weakagree);
claim(TrustedAuthority,Secret,SKVA); }
claim(TrustedAuthority,Secret,CA); }
claim(TrustedAuthority,Secret,RA); /*
claim(TrustedAuthority,Niagree); * Our proposal
claim(TrustedAuthority,Nisynch); */
claim(TrustedAuthority,Alive); // The protocol description
claim(TrustedAuthority,Weakagree); usertype String;
} hashfunction H;
} const XOR:Function;
/* const CON:Function;
* Our proposal const PUF:Function;
*/ const FGen:Function;
// The protocol description macro M1 = CON(PIDVA,N1);
usertype String; macro fkrsu = FGen(RRSU);
hashfunction H; macro hdrsu = FGen(RRSU);
const XOR:Function; macro hdrsuenc = XOR(hdrsu,H(kRG));
const CON:Function; macro MRG = XOR(CON(CON(IDR,M1),N2),H(fkrsu));
const PUF:Function; macro VRG = H(CON(CON(CON(CON(CON(IDR,MRG),kRG),N2
macro M1 = CON(PIDVA,N1); ),PIDRSU),hdrsuenc));
macro MGS = XOR(CON(CON(IDG,M1),N3),H(kGS)); macro MGS = XOR(CON(CON(IDG,M1),N3),H(kGS));
macro VGS = H(CON(CON(CON(CON(IDG,IDS),MGS),kGS),N3) macro VGS = H(CON(CON(CON(CON(IDG,IDS),MGS),kGS),N3)
); );
macro TokenA = CON(CON(CON(CON(IDVA, length), SKVA), macro TokenA = CON(CON(CON(CON(IDVA, length), SKVA),
Tmexpire), N4); Tmexpire), N4);
macro VSAA = H(CON(CON(CON(CON(IDVA,TokenA),RA),N1), macro VSAA = H(CON(CON(CON(CON(IDVA,TokenA),RA),N1),
N4)); N4));
macro CAenc = XOR(CA,H(CON(IDVA,kVS))); macro CAenc = XOR(CA,H(CON(IDVA,kVS)));
macro TokenAenc = XOR(TokenA,H(RA)); macro TokenAenc = XOR(TokenA,H(RA));
macro MSAA=(CAenc,TokenAenc,VSAA); macro MSAA=(CAenc,TokenAenc,VSAA);
macro MSG = XOR(CON(CON(TokenA,MSAA),N4),H(kGS)); //
macro kGSnewenc = XOR(kGSnew,H(CON(kGS,N3))); macro MGR = XOR(CON(CON(TokenA,MSAA),N3),H(fkrsu));
macro VSG = H(CON(CON(CON(CON(CON(IDG,IDS),MSG),N3), macro kRGnewenc = XOR(kRGnew,H(CON(fkrsu,N2)));
N4),kGSnew)); macro VGR = H(CON(CON(CON(CON(CON(IDG,MGR),kRGne
macro RA = PUF(CA); wenc),IDR),N2),kRGnew));
protocol Internet-of-Vehicle(RSUGateway,TrustedAuthority){ macro MSG = XOR(CON(CON(TokenA,MSAA),N4),H(kGS));
role RSUGateway{ macro kGSnewenc = XOR(kGSnew,H(CON(kGS,N3)));
secret RA, CA, IDVA, kVS, TokenA, SKVA, length, Tmexpire, macro VSG = H(CON(CON(CON(CON(CON(IDG,IDS),MSG),N3),
IDG, IDS, kGS,kGSnew; N4),kGSnew));
fresh N1, N4, N3, PIDVA; macro RA = PUF(CA);
var PIDVA; protocol Internet-of-Vehicle(RSU,RSUGateway){
send_1(RSUGateway,TrustedAuthority, IDG, MSG, VGS); role RSU{
recv_2(TrustedAuthority, RSUGateway,MSG,VSG,kGSne secret fkrsu, hdrsu, RRSU, hdrsuenc, RA, CA, IDVA, kVS,
wenc); TokenA, SKVA, length, Tmexpire, IDG, IDS, kGS,kGSnew,
claim(RSUGateway,Secret,kGS); IDR,kRG,kRGnew;
claim(RSUGateway,Secret,kGSnew);
(continued on next page)

15
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

fresh N1, N4, N3, N2, PIDVA, PIDRSU,kRGnewenc; Choi, W., Kim, S., Kim, Y., Park, Y., & Ahn, K. 2010. PUF-based Encryption Processor
for the RFID Systems. 2010 10th IEEE International Conference on Computer
var PIDVA,PIDRSU; and Information Technology, Cit, 2323–2328. https://doi.org/10.1109/
send_1(RSU,RSUGateway, PIDRSU,MRG,VGR); CIT.2010.400.
recv_2(RSUGateway,RSU, MGR,kRGnewenc,VGR); Chuang, K., Member, S., Bury, E., Degraeve, R., Kaczer, B., Linten, D., Member, S.,
Verbauwhede, I., 2019. A physically unclonable function using soft oxide
claim(RSU,Secret,kRG); breakdown featuring 0 % native. IEEE Journal of Solid-State Circuits 54 (10),
claim(RSU,Secret,fkrsu); 2765–2776. https://doi.org/10.1109/JSSC.2019.2920714.
claim(RSU,Secret,kRGnew); Committee, I.T.S., 2013. IEEE Guide for wireless access in vehicular environments
(WAVE). Architecture IEEE Vehicular Technology Society.
claim(RSU,Niagree); Cremers, C. n.d. The Scyther Tool : Verification , Falsification , and Analysis of
claim(RSU,Nisynch); Security Protocols. In Proc. Int. Conf. Comput. Aided VeriFcation, 1–4.
claim(RSU,Alive); De Fuentes, J.M., González-tablas, A.I., 2010. Overview of security issues in vehicular
Ad-hoc networks. Handbook of Research on Mobility and Computing.
claim(RSU,Weakagree); Dodis, Y., Reyzin, L., Smith, A., 2004. Fuzzy extractors : How to generate strong keys
} from biometrics and other noisy data. In: Advances in Cryptology—
role RSUGateway{ EUROCRYPT’2004 (Lecture Notes in Computer Science). Springer, Heidelberg,
German, pp. 523–540.
secret fkrsu, hdrsu, RRSU, hdrsuenc, RA, CA, IDVA, kVS,
Dolev, D., Yao, A., 1983. On the security of public key protocols. IEEE Trans. Inf.
TokenA, SKVA, length, Tmexpire,IDG, IDS, kGS,kGSnew,IDR, Theory, M, 198–208.
kRG,kRGnew; E. Suh, G., & Devadas, S. 2007. Physical unclonable functions for device
fresh N1, N4, N3, N2,PIDVA, PIDRSU,kRGnewenc; authentication and secret key generation. In: 2007 44th ACM/IEEE Design
Automation Conference, San Diego, 9–14.
var PIDVA,PIDRSU; Elhalawany, B. M., El-banna, A. A. A., & Wu, K. 2019. Physical-layer security and
recv_1(RSU,RSUGateway, PIDRSU,MRG,VGR); privacy for vehicle-to-everything. IEEE Communication Magazine, October, 84–
send_2(RSUGateway, RSU, MGR,kRGnewenc,VGR); 90.
Frikken, K. B., Blanton, M., & Atallah, M. J. 2009. Robust authentication using
claim(RSUGateway,Secret,kRG); physically unclonable functions. Lecture Notes in Computer Science (Including
claim(RSUGateway,Secret,fkrsu); Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in
claim(RSUGateway,Secret,kRGnew); Bioinformatics), 5735 LNCS, 262–277. https://doi.org/10.1007/978-3-642-
04474-8_22.
claim(RSUGateway,Niagree); Gope, P., Das, A.K., Kumar, N., Cheng, Y., 2019. Lightweight and physically secure
claim(RSUGateway,Nisynch); anonymous mutual authentication protocol for real-time data access in
claim(RSUGateway,Alive); industrial wireless sensor networks. IEEE Transactions on Industrial
Informatics 15 (9), 4957–4968.
claim(RSUGateway,Weakagree); Gope, P., Sikdar, B., 2019. Lightweight and privacy-preserving two-factor
} authentication scheme for IoT devices. IEEE Internet of Things Journal 6 (1),
} 580–589. https://doi.org/10.1109/JIOT.2018.2846299.
Guo, Y., Dee, T., Tyagi, A., 2018. Barrel shifter physical unclonable function based
encryption. Cryptoraphy 2 (3), 1–20. https://doi.org/10.3390/
cryptography2030022.
Herder, C., Yu, M., Koushanfar, F., & Devadas, S. 2014. Physical Unclonable Functions
and Applications : A Tutorial. Proceedings of the IEEE, 102(8), 1126–1141.
References https://doi.org/10.1109/JPROC.2014.2320516.
In, Z. H. A. N. Q., Ieee, M., In, X. I. L., & Ieee, F. 2020. The Security of Autonomous
Abdalla, M., Fouque, P., Pointcheval, D., 2005. Password-based authenticated key Driving : Threats , Defenses , and Future Directions. Proceedings of the IEEE, 108
exchange in the three-party setting. IACR International Conference on Public- (2).
Key Cryptography 3386, 65–84. Javaid, U., Member, S., Aman, M.N., 2020. A scalable protocol for driving trust
Adigun, A., Bensaber, B.A., Biskri, I., 2013. Protocol of change pseudonyms for management in internet of vehicles with blockchain. IEEE Internet of Things
VANETs. IEEE International Workshop on Performance and Management of Journal 7 (12), 11815–11829.
Wireless and Mobile Networks, 162–167. Javaid, U. 2019. DrivMan : Dri ving Trust Man agement and Data Sharing in VANETs
Aman, M.N., Sikdar, B., Member, S., 2018. ATT-Auth : A hybrid protocol for industrial with Blockchain and Smart Contracts. IEEE 89th Vehicular Technology
IoT attestation with authentication. IEEE Internet of Things Journal 5 (6), 5119– Conference (VTC2019-Spring).
5131. Jeon, D., Baek, J. H., Kim, D. K., & Choi, B. 2015. Toward Zero Bit-Error-Rate Physical
Aman, M.N., Basheer, M.H., Member, S., 2019a. Data provenance for IoT with light Unclonable Function : Mismatch-Based vs . Physical-Based Approaches in
weight authentication and privacy preservation. IEEE Internet of Things Journal Standard CMOS Technology. 2015 Euromicro Conference on Digital System
6 (6), 10441–10457. Design, 407–414. https://doi.org/10.1109/DSD.2015.57.
Aman, M.N., Basheer, M.H., Sikdar, B., 2019b. Two-factor authentication for IoT with Jia, D., Lu, K., Member, S., Wang, J., Zhang, X., 2016. A survey on platoon-based
location information. IEEE Internet of Things Journal 6 (2), 3335–3351. vehicular cyber-physical systems. IEEE Communication Surveys and Tutorials
Aman, M.N., Taneja, S., Member, S., Sikdar, B., Member, S., Chua, K.C., Alioto, M., 18 (1), 263–284.
2019c. Token-based security for the internet of things with dynamic energy- Jr, B. L. A. 2002. Security requirements for cryptographic modules. National Institute
quality tradeoff. IEEE Internet of Things Journal 6 (2), 2843–2859. of Standards and Technology.
Aman, M.N., Basheer, M.H., Member, S., Dash, S., 2020. HAtt : Hybrid remote Kaci, A. 2020. PoolCoin: Toward a distributed trust model for miners ’ reputation
attestation for the internet of things with high availability. IEEE Internet of management in blockchain. IEEE 17th Annual Consumer Communications &
Things Journal 7 (8), 7220–7233. Networking Conference (CCNC).
Aman, M.N., Javaid, U., Member, S., 2021. A privacy-preserving and scalable Kim, H.B., Hong, S., Kim, H.S., 2019. Lightweight conversion from arithmetic to
authentication protocol for the internet of vehicles. IEEE Internet of Things Boolean masking for embedded IoT processor. Applied Sciences (Switzerland) 9
Journal 8 (2), 1123–1139. (7). https://doi.org/10.3390/app9071438.
Azees, M., Vijayakumar, P., Deborah, L.J., 2016. Comprehensive survey on security Kleber, S., Unterstein, F., Matousek, M., Kargl, F., Slomka, F., Hiller, M., 2015. Secure
services in vehicular ad-hoc networks. The Institution of Engineering and execution architecture based on puf-driven instruction level code encryption.
Technology 379–388. https://doi.org/10.1049/iet-its.2015.0072. IACR Cryptology EPrint Archive 2015, 651.
Babaei, A., Schiele, G., 2019. Physical unclonable functions in the internet of things : Lei, A., Cruickshank, H., Cao, Y., Asuquo, P., Ogah, C.P.A., Sun, Z., Member, S., 2017.
State of the art and open challenges. Sensors (Basel) 19 (14), 3208. https://doi. Blockchain-based dynamic key management for heterogeneous intelligent
org/10.3390/s19143208. transportation systems. IEEE Internet of Things Journal 4 (6), 1832–1843.
Bian, W., Gope, P., Cheng, Y., Li, Q., 2020. Bio-AKA : An efficient fingerprint based Long, J., Liang, W.E.I., Li, K., Member, S., 2019. PUF-based anonymous authentication
two factor user authentication and key agreement scheme Bio-AKA : An scheme for hardware devices and IPs in edge computing environment. IEEE
efficient fingerprint based two factor user authentication and key agreement Access 7, 124785–124796. https://doi.org/10.1109/ACCESS.2019.2925106.
scheme. Future Generation Computer Systems 109 (March), 45–55. https://doi. Lu, X., Member, S., Hong, L., Member, S., 2018. CMOS optical PUFs using noise-
org/10.1016/j.future.2020.03.034. immune process-sensitive photonic crystals incorporating passive variations
Bohm, C., Hofer, M., 2012. Physical unclonable functions in theory and practice. for robustness. IEEE Journal of Solid-State Circuits 53 (9), 2709–2721. https://
Springer, NY, USA. doi.org/10.1109/JSSC.2018.2850941.
Chatterjee, U., Govindan, V., Sadhukhan, R., Mukhopadhyay, D., 2019. Building PUF Mcgrath, T., Bagci, I.E., Wang, Z.M., Roedig, U., Young, R.J., Bagci, I.E., Wang, Z.M.,
based authentication and key exchange protocol for IoT without explicit CRPs in Roedig, U., 2019. A PUF taxonomy A PUF taxonomy. Applied Physics Reviews 6,.
verifier database. IEEE Transactions on Dependable and Secure Computing 16 https://doi.org/10.1063/1.5079407 011303.
(3), 424–437.

16
E Haodudin Nurkifli and T. Hwang Journal of King Saud University – Computer and Information Sciences 35 (2023) 101721

Nidhal, M., Ben-othman, J., Hamdi, M., 2014. Survey on VANET security challenges Wang, D., Member, S., Cheng, H., Wang, P., Member, S., 2017. Zipf ’ s Law in
and possible cryptographic solutions. Vehicular Communications 1 (2), 53–66. Passwords. IEEE Transactions on Information Forensics and Security 12 (11),
https://doi.org/10.1016/j.vehcom.2014.05.001. 2776–2791.
Oracle. n.d. Java Cryptography Architecture (JCA). Retrieved November 11, 2022, Wang, J., Wu, L., Wang, H., Choo, K.K.R., Wang, L., He, D., 2022. A secure and efficient
from https://docs.oracle.com/javase/8/docs/technotes/guides/security/crypto/ multiserver authentication and key agreement protocol for internet of vehicles.
CryptoSpec.html. IEEE Internet of Things Journal 9 (23), 24398–24416. https://doi.org/10.1109/
Prouff, E., & Rivain, M. 2013. Masking against side-channel attacks: A formal JIOT.2022.3188731.
security proof. Lecture Notes in Computer Science (Including Subseries Lecture Wang, W., Yona, Y., Diggavi, S.N., Gupta, P., 2018. Design and analysis of stability-
Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 7881 LNCS, guaranteed PUFs. IEEE Transactions on Information Forensics and Security 13
142–159. https://doi.org/10.1007/978-3-642-38348-9_9. (4), 978–992.
Report, I. 2015. Global Risks 2015 10th Edition. Whyte, W., Weimerskirch, A., Kumar, V., Hehn, T., 2021. A security credential
Roy, S., Chatterjee, S., Das, A.K., Chattopadhyay, S., 2017. On the design of provably management system for V2V communications. IEEE Vehicular Networking
secure lightweight remote user authentication scheme for mobile cloud Conference A, 1–8.
computing services. IEEE Access, October. https://doi.org/10.1109/ Xue, K., Luo, X., Ma, Y., Li, J., Liu, J., Wei, D.S.L., 2022. A distributed authentication
ACCESS.2017.2764913. scheme based on smart contract for roaming service in mobile vehicular
Roy, S., Das, A.K., Chatterjee, S., Kumar, N., Member, S., Chattopadhyay, S., Rodrigues, networks. IEEE Transactions on Vehicular Technology 71 (5), 5284–5297.
J.J.P.C., Member, S., 2020. Provably secure fine-grained data access control over https://doi.org/10.1109/TVT.2022.3148303.
multiple cloud servers in mobile cloud computing based healthcare Yahiatene, Y. 2018. Towards a Blockchain and Software-Defined Vehicular
applications. IEEE Transactions on Industrial Informatics 15 (1), 457–468. Networks approaches to secure Vehicular Social Network. IEEE Conference on
Salem, A.H., Abdel-hamid, A., El-nasr, M.A., 2014. The case for dynamic key Standards for Communications and Networking (CSCN).
dsitribution for PKI-based VANETS. International Journal of Computer Networks Yang, Z., Yang, K., Lei, L., Member, S., Zheng, K., Member, S., Leung, V.C.M., 2019.
& Communications 6 (1), 61–78. Blockchain-based decentralized trust management in vehicular networks. IEEE
Seudié, H., & Gmbh, R. B. 2009. E-Safety Vehicle Intrusion Protected Applications Internet of Things Journal 6 (2), 1495–1505.
EVITA project objectives. Escar Embedded Security in Cars Conference. Yazid, M., Fahmi, F., Sutanto, E., Setiawan, R., Aripriharta, Aziz, M., 2023. Simple
Soja, R. n.d. Automotive Security : From Standards to Implementation. Automotive authentication method for vehicle monitoring IoT device with verifiable data
Microcontrollers and Processors, NXP. integrity. IEEE Internet of Things Journal 10 (8), 7027–7037. https://doi.org/
Sulaiman, A., Raja, S.V.K., Han, S., 2013. Improving scalability in vehicular 10.1109/JIOT.2022.3228926.
communication using one-way hash chain method. Ad Hoc Networks 11 (8), Zhang, X., Chen, X., 2019. Data security sharing and storage based on a consortium
2526–2540. https://doi.org/10.1016/j.adhoc.2013.05.017. blockchain in a vehicular ad-hoc network. IEEE Access 7, 58241–58254. https://
Technology, S. A. n.d. Smart Card Standards and Specifications. Retrieved June 10, doi.org/10.1109/ACCESS.2018.2890736.
2021, from https://www.securetechalliance.org/smart-cards-intro-standards/ Zhang, H., Lai, Y., Chen, Y., 2023. Authentication methods for internet of vehicles
#::text=ISO%2FIEC 14443 compliant cards,the FIPS 201 PIV card. based on trusted connection architecture. Simulation Modelling Practice and
Tuyls, P., Batina, L., 2006. RFID-Tags for Anti-counterfeiting. In: Topics in Cryptology Theory 122, (June 2022). https://doi.org/10.1016/j.simpat.2022.102681 102681.
CT-RSA (LNCS 3860), Heidelberg. Springer, German, pp. 115–131. Zhu, F., Li, P., & Xu, H. 2019. A Lightweight RFID Mutual Authentication Protocol
Twente, U. 2015. Preparing Secure Vehicle-to-X Communication Systems. 1–3. with PUF. Sensors (Basel, Switzerland), Id, 1–22.

17

You might also like