Vehicular Communications 37 (2022) 100502

Contents lists available at ScienceDirect

Vehicular Communications
www.elsevier.com/locate/vehcom

Blockchain based optimized data storage with secure communication

for Internet of Vehicles considering active, passive, and double
spending attacks
Adamu Sani Yahaya a , Nadeem Javaid a,b,∗ , Sherali Zeadally c , Hassan Farooq a
a
Department of Computer Science, COMSATS University Islamabad, Islamabad 44000, Pakistan
b
School of Computer Science, University of Technology Sydney, Ultimo, NSW, 2007, Australia
c
College of Communication and Information, University of Kentucky, Lexington, KY 40506-0224, USA

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

Article history: In the study, we propose a double blockchain based vehicular system for optimizing data storage and
Received 13 August 2021 securing communication in Internet of Vehicles (IoV) networks. We propose an incentive mechanism and
Received in revised form 28 May 2022 a reputation system to reduce the selfish and distrustful behavior of IoV nodes. Moreover, we propose
Accepted 24 June 2022
an encryption technique to protect the proposed system from passive and active attacks. The proposed
Available online 4 July 2022
technique is a hybrid of Rivest Shamir Adleman and Advanced Encryption Standard-256 (AES-256+RSA).
Keywords: We use this hybrid approach to efficiently and secretly share the users’ secret keys. To address the issues
Blockchain of high resource consumption and communication delay, we propose a cache memory technique for
Double spending attack miner and controller nodes. In addition, we propose an authentication mechanism to remove malicious
IoT nodes from the system. The simulation results depict that our system provides security, scalability,
IoVs optimized data storage, node’s trustworthiness, and transparent data sharing.
WSN © 2022 Elsevier Inc. All rights reserved.

1. Introduction Units (OBUs), which have different devices, such as cameras, auto-
matic driving and braking systems, different sensors and so on [6].
Today, we are witnessing a rapid growth in the amount of ve- These nodes generate diverse types of data and share it through
hicular data as the number of vehicles increases in Vehicular Ad- wireless communication devices. The nodes share data with each
hoc Networks (VANETs) [1]. The impact of the increasing amount other as well as the RSUs to improve the service quality of In-
of vehicular data on VANETs is twofold. First, a large amount of telligent Transportation System (ITS) and ensure safe driving [7,8].
vehicular data improves traffic and transport management system ITS is an advanced system that provides services related to var-
in terms of timely decision making. Second, the vehicular data re- ious modes of traffic and transport management. It also enables
quires additional computing resources and large storage capacity, users to be well informed about traffic conditions and make safe
which are the major concerns for resource constrained vehicles. coordinated use of the transportation network [9]. Different types
This mentioned impact has become a hot research topic, which of communication technologies and protocols are used in VANETs,
has attracted the attention of both industry and academia. It is such as Wireless Access Vehicular Environment (WAVE), Dedicated
estimated that the number of vehicles over the next decade will Short Range Communication (DSRC), Long Term Evolution (LTE),
reach 140 million around the world [2,3]. LTE-Vehicle to X (LTE-V2X), IEEE 802.11p, and Wi-Fi [10]. However,
In VANETs, vehicles can interact with one another using Vehi- there are various challenging issues in VANETs that hinder the se-
cle to Vehicle (V2V) communication and they can also commu- curity and integrity of data that is transmitted in the network.
nicate with Road Side Units (RSUs) using Vehicle to Infrastruc-
VANETs provide several services such as traffic information,
ture (V2I) communication [4]. When these vehicles connect and
sensory information and road condition information. The informa-
share information in a distributed manner through the Internet,
tion is gathered by IoV nodes and is stored at a central server
they become part of an Internet of Vehicles (IoV) network and are
using a centralized system approach. In general, IoV nodes are
referred as IoV nodes [5]. The nodes are equipped with On-Board
resource constrained because of their limited battery size, less
computational resources, and low data storage. Hence, they face
several challenges. The challenges include data insecurity, storage
* Corresponding author.
E-mail address: nadeemjavaidqau@gmail.com (N. Javaid). overhead, high delay in services’ delivery, single point of failure,
URL: http://www.njavaid.com (N. Javaid). and inefficient data authentication. To address these challenges,

https://doi.org/10.1016/j.vehcom.2022.100502
2214-2096/© 2022 Elsevier Inc. All rights reserved.
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

distributed and decentralized approaches have been proposed for mechanism cannot be applied to resource constrained devices due
VANETs [11,12,18,20]. However, there are still some issues that re- to its high consumption of resources. The authors of [27] proposed
main unresolved in distributed systems, such as malicious attacks, a secure blockchain based decentralized and distributed storage
and data leakage. Blockchain system is an emerging technology, management system for a vehicular network. The proposed sys-
which can address some of the aforementioned challenges. tem is used to secure data in a storage device, which increases
the efficiency and performance of the vehicular network. However,
1.1. Preliminary the network’s communication channels are not reliable in terms of
privacy and efficiency when dealing with constrained devices. The
1) Blockchain has several unique characteristics, which include authors of [28] proposed a trust model using blockchain. The au-
decentralization, security, anonymity, immutability, and trans- thors also proposed a scheme for validating traffic events based
parency [13,14]. It is a Peer to Peer (P2P) distributed public on a proof of event consensus mechanism. However, the method
ledger that is maintained by all peer nodes using cryptographic that is used to update the reputation scores relies on pieces of
hash functions and Merkle tree [15]. The records of peer nodes evidence collected from vehicles, which can cause an increase in
are stored in the public ledger. When the records are grouped the transmission delay. The authors of [29] developed a reliable
together, a block is generated, which is validated by the peer and secure smart parking system using linkable group signature
nodes using consensus mechanisms. Each block contains pre- and blockchain. The proposed system mitigates multi-reservation
vious block’s hash, current block’s hash, timestamp, a list of attack, malicious occupation of parking spots and dishonest park-
transactions, and a nonce value as Fig. 1 shows. Moreover, Proof ing lots. However, the proposed system does not address issues
of Work (PoW) consensus protocol is commonly used in public of double spending attack and storage overheads. To address the
blockchain [16,17]. In PoW, all nodes take part in the mining aforementioned issues, we propose a consortium blockchain based
process to verify and authenticate the new blocks. vehicular network to secure the communication channels. This
2) A hashing algorithm is a mathematical mapping operation be- paper is an extension of [30] where encryption techniques, au-
tween an integer set with a fixed size output and a set of thentication mechanism, security analysis, incentive mechanism,
dynamic size input [19]. Hashing algorithms are used for vari- and reputation system were discussed. The double spending attack
ous purposes, e.g., comparison between output data and input occurs when an attacker acts as an honest user and broadcasts
data, data recognition, and so on. They are extensively used to two payment transactions to the blockchain: paying himself/her-
enforce security and privacy of data. self and at the same time paying the receiver. This means that the
3) A MTP-Argon2 technique is a method used to filter unnecessary attacker can mine a block comprised of the false transaction in
and duplicate data from the raw data in a cloud system. parallel with other non-malicious miners. The leakage of informa-
tion occurs when confidential information is exposed by attackers
1.2. Related work to unauthorized users or other parties in the system. It is only
possible when the system is not monitored and is not secured
Recently, blockchain has been used in various domains such as against threats. This type of threat is known as a passive attack.
vehicular networks, Internet of Things (IoT) networks, underwa- In contrast, when the content of the information is changed by an
ter sensor networks, Wireless Sensor Networks (WSNs), and many attacker, it is called an active attack.
others [18,20–22].
Due to the characteristics of blockchain, in [23], a blockchain 1.3. Motivations and problem statement
based model for big data sharing using smart contract is proposed
to secure data storage. However, the storage overhead is still an is- The main challenging issues in vehicular networks include stor-
sue because the amount of sensed data is rapidly increasing. The age overhead [31], communication delays [32], lack of trust and
increase is because some IoT nodes gather irrelevant and incorrect selfishness of nodes [33], insecure communication channels, and
data from different sources, such as selfish nodes, defective sensors the presence of malicious nodes [34]. The issues of low data stor-
or malicious nodes. Moreover, as the amount of vehicular data is age and high congestion in traffic have been addressed by [35]
growing rapidly, trust concerns and insecure communication chan- wherein the authors propose a distributed transportation manage-
nels have become critical issues that cannot be overlooked [24]. ment system using a blockchain based vehicular network. How-
The latency and waiting time of vehicular messages increase when ever, the absence of data filtering and verification mechanisms to
the amount of data increases, which affect the communication in remove duplicate and unnecessary data causes storage overhead
the vehicular network. To ensure sharing high quality data, a rep- and delay during communication. The authors of [32] solve the
utation system is required in which the data provided or shared issue of communication delays using on-chain and off-chain ser-
by IoV nodes is rated by other IoV nodes based on quality of the vice mechanisms where a secure service provisioning mechanism
shared data (i.e., valid or correct data). based on the blockchain technology for lightweight IoT clients
A large amount of data is produced by the vehicular network is proposed. However, the IoT devices do not trust edge servers
[25]. Moreover, it is difficult to maintain and store such a large (base stations), which reduces the rate of participation for data
amount of data. In the network, selfish node does not forward sharing. To address the selfishness and trust issues of IoT devices
the data packet to the remaining nodes. To resolve the issue, the and encourage them to participate and share data in the network,
authors of [25] proposed a blockchain based incentive system for an incentive mechanism is required. Moreover, a reputation sys-
storing data using the Provable Data Possession (PDP) mechanism. tem can be used to solve the trust issue. Furthermore, the selfish
However, the data storage capacity is not optimized. Also, the dam- behavior of nodes in the vehicular network has been addressed
aged nodes have a high rate of failure, which is a problem when by [33] wherein the authors propose a two-stage security mech-
requesting data. In [26], the authors proposed a Device-to-Device anism based on blockchain enabled IoV network using contract
(D2D) system to authenticate every user using blockchain technol- theory and reputation system. However, the proposed mechanism
ogy and Channel State Information (CSI). In the proposed system, requires much time for secure data sharing, which causes delay in
the authors applied the Practical Byzantine Fault Tolerance (PBFT) the network communication. To minimize the data sharing time,
consensus algorithm during consensus for blocks’ validation. The the authors of [36] propose a cache technique for an IoT network
D2D mechanism overcomes the spectral inefficiency of mobile de- using blockchain based edge computing system. Using the cache
vices and increases the power capacity. However, the proposed technique in the proposed system, the authors minimize the data

2
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

Fig. 1. Structure of blockchain.

Fig. 2. The proposed system model.

sharing time, resource consumption for service response, and com- 2. Overview of the proposed system model
munication delay. However, in the network, the communication
channels are not protected from passive, active, and double spend- Motivated by the study done in [35], Fig. 2 shows a blockchain
ing attacks. To secure the communication channels of the vehicu- based vehicular network, which consists of IoV nodes, miner
lar network, an encryption and decryption mechanism is required. nodes, controller nodes and a cloud server. Also, the proposed net-
Furthermore, the authors of [34] present a blockchain based trust work has two blockchains as shown in Figs. 3 and 4: local and
system using Software Defined Networking (SDN). However, a ro- main. The miner nodes store their data temporarily in the local
bust mechanism for detecting malicious nodes is still needed to blockchain storage. Whereas, the other blockchain, i.e., the main
reduce the number of malicious nodes and fake messages that are blockchain, is used by the controller nodes for metadata stor-
produced by them. Moreover, to prevent malicious nodes from par- age and later forward data to cloud server. The local and main
ticipating in the proposed network, an authentication mechanism blockchains are thoroughly discussed in Section 2.1 and Section 2.2
is required, which can protect the network from spoofing attacks. respectively. IoV nodes are Smart Vehicles (SVs), which have low
Table 1 presents the summary of the related work. computational resources, small battery size, low data storage, and
limited capabilities for data sharing [35]. These nodes are con-
nected in a P2P manner to communicate with miner nodes to
1.4. Our research contributions
share information and get services such as traffic information, road
condition information, charging services information, and so on.
The major contributions of the research study are as follows. Miner nodes are IoV nodes (shown in red circles) with high com-
putational resources, large battery size, large data storage, and
• We proposed a blockchain based secure vehicular network for more capabilities for data sharing as compared to ordinary IoV
IoV’s data storage to provide an efficient communication chan- nodes. These nodes are responsible for storing vehicles’ data and
nel in smart cities. providing information to other IoV nodes. Besides, the controller
• We implemented a smart contract in the network through nodes act as base stations, which provide the facilities of ser-
which incentives and ratings are provided to IoV nodes to in- vice request and response. These nodes have high computational
crease the trust level. Furthermore, to minimize the service resources, high data storage, large battery size, and high capabili-
delay, we proposed a caching technique. ties for data sharing and processing as compared to miner nodes.
• We proposed an authentication mechanism to prevent mali- The IoV nodes’ data from different locations is stored in the lo-
cious nodes from entering the network. Furthermore, we con- cal blockchain for a limited amount of time. In contrast, all of the
ducted a security analysis to assess the robustness of the net- controller nodes’ data and local blockchains’ data are permanently
work against passive, active, and double spending attacks. stored in the main blockchain. The cloud server aggregates the pro-

3
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

Table 1
Related work summary.

Problems addressed Techniques Limitations

The central network stores huge amount of PDP and consortium blockchain. Data storage capacity is not optimized. The damaged
data, which is difficult to maintain [25]. nodes are recovered with high failure rate.

User authentication [26]. PBFT and blockchain. High computational complexity for resource constrained
devices and data transmission delay.

To investigate on how blockchain network will PoW and public blockchain. Trust management is not considered, communication
be extended and reduce the delay in service re- channels are not reliable, and system has high compu-
sponse time [27]. tational complexity.

Trust management and reduction in dissemina- Proof of event and blockchain. Scalability and delay in data transmission problems are
tion of fake messages [28]. not resolved.

Low throughput, centralized system, and high Consortium blockchain and PoA. Reliable trusted providers and incentive mechanism are
security risk for the system [32]. not included. Issues of computational overhead and high
storage cost are not solved.

VANET does not provide user comfort and PoW and public blockchain. Lack of storage due to constrained storage devices and
driver’s security. Also scalability and availabil- high communication latency for vehicles.
ity of network’s nodes are achieved [35].
Network and blocks scalability issues are PoC and private blockchain. Security analysis is neglected, optimal size of the sen-
solved in WSN [37]. sor’s memory is not considered when optimizing the
system’s performance.

Lack of network coverage, trust of private net- LoRaWAN and blockchain. Network scalability and data communication delay are
work, and high cost of spectrum resources for not considered. Also, no security analysis is done.
IoT edge devices [38].
Scalability and security are handled [39]. PoC, smart contract, and private Computational overhead and trust management prob-
blockchain. lems are not handled. Also, security analysis is not per-
formed.

The third party involvement and lack of trust FTF, PoCo, and blockchain. The security of design system is compromised. The per-
in data sharing causes high latency [40]. formance is also reduced in a real-world environment.

A single point of failure, network scalabil- PoC, smart contract, and blockchain. High latency and trust management problems are not
ity, and access management issues are han- solved.
dled [41].
Security and trust development between vehi- A branching and un-branching algo- Issues related to high latency and storage cost are not
cles [42]. rithm, intelligent vehicle trust point, solved.
blockchains, and PoW.
Lack of trust and security [43]. Joint PoW and PoS, and public Computational overhead.
blockchain.
Delay in data communication [44]. Three-weight subjective logic, Duplication of data and high storage cost problems are
blockchain, and edge computing. not resolved.

Data sharing efficiency is improved [45]. Batch verification, PoW, and Data credibility and duplication are not properly han-
blockchain. dled.

Prevent the distribution of fake messages [46]. Lexicographic Merkle tree technique, Transmission delay and high storage cost issues are not
blockchain, and PoW. solved.

Anonymous communication and trust manage- Logistic regression, blockchain, PoW, Communication delay problem is not solved.
ment [47]. and PBFT.

cessed data from controller nodes and stores it in a distributed request to its nearest miner node. If the required information is
manner. stored in cache storage of the miner node, then it quickly responds
to the IoV node. If the miner node does not get the required in-
2.1. The miner node model formation, then it searches the local blockchain. If the miner node
finds the required information, then it responds to the IoV node. If
The miner nodes have high data storage, sensing, and comput- the search of miner node fails, then it forwards the request to the
ing capabilities. In the blockchain, each miner node has a cache controller node. Next, the controller node checks its cache storage,
storage and a public ledger that contains all of the IoV nodes’ main blockchain, and cloud server to search for the information.
transactions, as shown in Fig. 3. Each miner node stores recently If the information is not found, then it returns “information not
requested services of the local blockchain in the cache memory found” message to the miner nodes. Otherwise, it responds to the
during data sharing with other IoV nodes. All of the miner nodes miner node with the required information. After receiving the re-
are connected in a P2P manner with each other. All of the IoV sponse from the controller node, the miner node responds to the
nodes’ information is stored in the local blockchain after valida- IoV node. After getting the response from the miner node, the IoV
tion using a PoW consensus mechanism. node forwards the response to controller node for validation. If
Before an IoV node’s sensed data is stored in the system’s stor- the received information is valid or is truly not found, then the
age, it must be broadcasted and validated by the miners. Each IoV node adds a positive rating to the miner node’s profile, else
miner node validates the data. If at least 51% of the miner nodes it adds a negative rating. Based on this rating, the trust or repu-
agree on the validity of the data, then a block is created in the tation of a miner node is increased or decreased. Positive rating
local blockchain. When an IoV node needs any information such means the reputation value will increase when valid information
as traffic congestion, and road and weather condition, it sends a is received. While negative rating means the reputation value will

4
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

2.2. The controller node model

Fig. 4 shows that the controller nodes are linked with each
other using the main blockchain and the cloud server. In the figure,
the controller nodes also contain cache storage that provides the
necessary information related to traffic, weather, routes, and the
nearest charging station to miner nodes. The controller nodes have
high computational resources, large battery size, and high capabil-
ity to process large amounts of vehicular data. Each controller node
has a public ledger, which contains the shared transactional data.
In the proposed system, we assume that the controller nodes are
trusted entities. The cloud server gathers the filtered data of con-
troller nodes in a distributed manner. Whenever IoV nodes require
Fig. 3. Operation of the miner node.
data, they send a request to the closest miner nodes. If none of the
miner nodes is near to the IoV node, then it directly sends the re-
Algorithm 1 Data storage process in main blockchain and cloud quest to the controller node. To store data on the main blockchain,
server. we used a PoW consensus mechanism.
1: Send data to miner node from IoV node; Algorithm 2 presents the process for data storage in the main
2: if New data then
blockchain and cloud server. Initially, an IoV node senses its sur-
3: Processes the data;
rounding environment using different sensors. Then, it aggregates
4: Send the processed data to controller nodes;
and forwards the sensed data to a miner node. After receiving the
5: if Data not received then
data of IoV nodes, the miner node checks if it already has sim-
6: Send again;
7: else
ilar data in storage using the Keccak-256 hashing algorithm. The
8: Filter the data; Keccak-256 hashing algorithm [19] is used because it is a vari-
9: Validate the data; ant of Secure Hash Algorithms (SHA) 3, i.e., SHA-3 family, which
10: Create a block; is developed to overcome the weakness of SHA-1 family and to
11: Update the main blockchain and provide services; enhance the strength of SHA-2 family [19]. Hashing algorithm is
12: Upload data to cloud server; a mathematical mapping operation between an integer set with a
13: end if fixed size output and a set of dynamic size input [19]. Hashing al-
14: end if gorithms are used for various purposes, e.g., comparison between
output data and input data, data recognition, and so on. They are
extensively used to enforce security and privacy of data. If the re-
reduce when fake or incorrect information is received. The values ceived data is the same as the previously stored data, then the
increase and decrease by 0.01 for positive rating and negative rat- miner node responds with “Data is the Same” message. If the re-
ing respectively. ceived data is new, then it forwards the data to the controller
IoV nodes constantly share sensors’ data with the miner nodes, node. After getting the raw data, the controller node again veri-
which introduces storage overhead issues in the local blockchain. fies it. If the data is not received, then it responds with a “Not
To address this issue, we propose an optimized storage mechanism, Received” message to the miner node. If the raw data is received,
motivated by the study [37]. In this mechanism, we consider the
then the controller node starts the processing phase using the
number of blocks and the time limit of the local blockchain. The
MTP-Argon2 technique. By applying this method, the raw data is
data of the local blockchain is sent to the main blockchain after
filtered and unnecessary or duplicate data is removed. Before the
a certain period. However, there is a possibility that data genera-
raw data is recorded in the main blockchain, its unique hash is
tion during a specific interval may increase, resulting in an increase
generated and is compared with the existing hash values. If the
in the number of blocks generated and the storage overhead. To
existing hash values do not match with that of the raw data, then
mitigate this situation, we apply a limit on number of blocks, i.e.,
data is considered valid and useful. Otherwise, the raw data is
the data is transferred from the local chain to the main chain, if
considered invalid, unnecessary or duplicate and it is discarded.
the number of blocks reaches a certain limit. Otherwise, the data
In the validation phase, the filtered raw data is shared with all
is transferred when the transmission time reaches the maximum
other controller nodes. Using the PoW consensus mechanism, we
limit. After the data transfer, the data of all blocks is deleted from
the local chain except the data that is stored in the last block. In ensure the validity and integrity of data. Next, a new block is gen-
this way, the procedure of storing block data continues and the erated and updated in the main blockchain by the controller nodes.
local blockchain data storage is optimized. Algorithm 1 presents a Moreover, the data of main blockchain is uploaded to the cloud
data storing process in local blockchain. Initially, we assume the to- server after a specific period. Using the main blockchain, the con-
tal number of blocks to be n, time limit to be tm, arbitrary to be k, troller node provides information to IoV nodes and miner nodes. In
which varies from 1 to n, and the first block of local blockchain to this way, the main blockchain’s storage is efficiently optimized and
be B0, which is also known as the genesis block. The miner nodes the service delay is reduced. In the proposed study, we used PoW
store sensors’ data of IoV nodes in the local blockchain. When the for both local and main blockchain as a consensus mechanism to
number of blocks exceeds n, the miner nodes save all the local ensure the validity and integrity of data of the IoV system. The
blockchain’s data in the main blockchain. After storing the local mechanism secures the system using validators for verifying the
blockchain’s data, all the blocks are deleted except the last one, submitted transactions, which are to be added in the blockchain. It
which is then considered as the genesis block. On the other hand, also does not allow one validator to modify the data in the system.
when the time exceeds tm, then the process of storing the data This mechanism keeps the whole network secure from intruders’
in the blocks is stopped. By applying this method, we resolve the activities during which intruders invest a lot of money and time in
data storage issue. order to take control of the system [48].

5
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

Algorithm 3 Smart contract for incentive and reputation mecha-

nism.
Input g j = Rating j , h j = Re ward j ;
1: if {Transaction honest} then
2: g j = g j + 0.01;
3:
4: h j = h j + 0.01;
5: else
6: g j = g j − 0.01;
7: end if
Fig. 4. Operation of the controller node. 8: if { g j ≥ κ } then
9: The node j is added to the list of miners;
10: end if
11: if { g j < 0} then
12: The node j will be added to the blacklisted list;
13: end if

threshold point is set for IoV nodes by the controller nodes to

meet the requirements of becoming miner nodes (Static threshold
is selected based on the condition at that given time). When the
number of points of an IoV node is equal to or greater than the
threshold value (it means having enough points to upgrade their
functionalities of both hardware and software), it is converted into
a miner node. On the other hand, if the incentive points are lower
than or equal to zero, then the node is blacklisted and it is sub-
Fig. 5. Incentive and reputation mechanisms.
sequently removed from the network. It is worth noting that the
incentive mechanism provides non-monetary incentives. Therefore,
Algorithm 2 Local blockchain process.
it is used for providing more priority to nodes that have more in-
Input n, tm, B0;
centive points than the one with low incentive points. Moreover,
1: i = 1
in the proposed model, it assumes that all the nodes with high
2: while i ≤ n do
incentive have high computational resources.
3: j=1
Since there are different miner nodes collecting requests and
4: while j ≤ tm do
5: F (B) = Bi ;
providing information from or to several IoV nodes without a
proper validation mechanism, some IoV nodes may not trust the
6: end while
miner nodes because of invalid information response. To overcome
7: Save local blockchain’s data on controller node’s
storage;
the distrust issue of the nodes, we propose a reputation system,
which allows IoV nodes to rate the miner nodes based on the valid
8: end while
information response. When IoV nodes send a request to the miner
9: Delete: B k−n to B k−1
node for getting the required information, such as traffic conges-
tion, weather condition and road condition, the miners check the
2.3. Incentive and reputation mechanisms IoV nodes’ requests in the local blockchain and process it to find
the required information response. If the required information is
In our proposed system, we implement the incentive and repu- not available, they forward the request to the controller nodes,
tation mechanism using a smart contract given in Algorithm 3. In which follows the same steps as those of the miner nodes and
the smart contract, the IoV nodes share information with other IoV respond to the miner nodes. Next, the miners respond to the IoV
nodes or miner nodes as Fig. 5 shows. When an IoV node needs nodes. If both the miner and the controller nodes fail to find the
any type of information, it sends a request to miner nodes. If the information, then the “information not found” message is sent to
information needed is not stored in the miner node’s cache stor- the IoV node. If the response message is valid or truly not found,
age as well as the local blockchain, then the request is forwarded then the IoV node assigns a positive rating to the miner node. Oth-
to the controller node. The controller node processes the request erwise, the IoV node assigns a negative rating. To confirm whether
in the main blockchain and responds to the IoV node through the the message is truly not found, the hash value at the receiver side
miner node. Some IoV nodes do not participate in data sharing will be matched with the hash value of the requested information
process because their resources are limited and they may become from the source. If they are the same, then the message is truly
selfish. This means that the nodes may refuse to participate in not found, otherwise, the message is found and a negative point
the network to save their computational and storage resources. To will be given to miner nodes which refuse to forward the infor-
overcome selfishness of the nodes, we propose an incentive mech- mation requested. Based on the rating, the trust of miner nodes
anism. IoV nodes send the sensory information to the miner nodes, is increased or decreased. If the miner node’s rating is lower than
which store it in the local blockchain. Whenever new information the pre-defined threshold value (assumed to be 0.5), it is consid-
is shared with the miner nodes, it is validated by other miner and ered to be a malicious node. As the minimum value is 0 and the
controller nodes. If the new information is valid, then the miner maximum value is 1 of the reputation, and 0.5 is at the middle of
node gives incentives to the IoV node through the controller node both the maximum and minimum values, therefore, it is selected
using the incentive mechanism. If the new information is invalid, as the initial value for all the nodes. When a miner node is iden-
then the miner node penalizes the IoV node. The IoV nodes that tified as malicious, its status is broadcasted to the network. It is
send correct information are rewarded with non-monetary incen- then removed from the vehicular network by the controller node,
tives as Fig. 6 shows. Furthermore, if a node provides more correct which acts as a trusted authority. The controller node maintains
information, the node receives higher incentive points. A static the reliability of the network by preventing malicious activities.

6
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

nodes request data from the controller nodes. This is because the
controller nodes process and retrieve required information for the
requester node from the main blockchain. To reduce both the com-
putational time and response delay, we propose a caching tech-
nique in which recently processed information is stored. Using this
technique, the controller node stores recently processed informa-
tion into the cache memory. In the proposed system, we consider
the size and the amount of data to be stored as a criterion for
storing data in the cache. If the size of data is more than 4 GB or
Fig. 6. Non-monetary incentive mechanism. the number of transactions exceeds 100, the data will be moved
to blockchain storage. When an IoV node or a miner node requests
When an IoV node shares information with miner nodes, some any information, the controller node responds with the required
malicious nodes can interrupt or attack the sensors’ information of information from the cache memory instead of retrieving it from
IoV nodes and add some fake messages. For example, an IoV node the main blockchain. Processing the information directly from the
shares road accident conditions with the miner node. If encryp- main blockchain takes a lot of computational resources and time,
tion is not used, malicious nodes can alter the shared information which cannot be ignored.
of the IoV node and change the accident condition to normal traf-
fic condition on road. To mitigate these types of attacks, we use 2.5. Attacker model
hybrid of two cryptographic techniques: Rivest Shamir Adleman
(RSA) and Advanced Encryption Standard-256 (AES-256) encryp- We make the following assumptions for an adversary: (i) The
tion (AES-256+RSA). AES-256 is a symmetric encryption algorithm data of the nodes is stored in the blockchain, which is tamper-
that requires both the decryptor and encryptor to use the same
proof and cannot be compromised. (ii) Double spending attacks
key. Key management is one of the drawbacks of symmetric en-
can occur when adversaries take control of the communication
cryption. To address this issue, the strengths of RSA and AES-256
channels in the proposed system.
encryption are combined. In the proposed system, the data to be
exchanged is encrypted using the speedy AES-256 encryption tech- The idea of double spending attacks in this study is motivated
nique. To solve the key management problem, the secret key of from [49], which is justified using the following three parame-
AES-256 is encrypted using public key of the receiver node and ters: a catch-up function (C ), a potential progress function ( P ), and
the RSA algorithm. Both the encrypted key and data are forwarded an adversary’s probability for performing a double spending at-
to the receiver. The receiver uses the private key to decrypt the tack (D S). A double spending attack’s probability that is performed
encrypted key to get the secret key, which is then used to decrypt on the basis of the attacker’s expected branch size is denoted as
the data. Later, the data is stored in the local blockchain after val- P . C and P are independent functions that generate D S where
idation and the miner node sends a response message to the IoV the system’s vulnerability to double spending attack is measured
node. as a probability. The following are the parameters used to an-
alyze the behavior of D S for achieving the primary goal of the
2.4. Nodes’ authentication and caching techniques
attack model: (i) C (q, z) is an attacker’s branch likelihood that can
become larger than the branch of a non-malicious node on the
Once the V2V communication is established between IoV and
basis of z blocks’ initial disadvantage (initial disadvantage means
miner nodes, there is a possibility of different attacks that can
the number of blocks the attacker produces ahead of the honest
compromise the stability of vehicular communication channels. To
users). (ii) The attackers’ probability to mine n blocks before the
prevent these attacks, we propose an authentication mechanism.
When a new vehicle is added to the vehicular network, all its in- non-malicious nodes mine mth block is P (q, m, n, t ), assuming the
formation is gathered by the controller nodes and a unique crypto- attacker is secretly mining during τ sec. τ ∈ R≥0 is an additional
ID is assigned to the vehicle before it is allowed to participate in parameter that is used in the adversary model. It is an average
the network. The crypto-ID serves as a vehicle’s identification tag duration for the attacker as well as the honest nodes to mine a
and can be easily verified by the controller node. Using the crypto- block. The percentage of computing power used by the attacker
ID, adversary nodes are easily identified whenever they want to is q ∈ [0, 1] and t is the time advantage allocated to an attacker
participate in the proposed system. The reason is that the mali- whenever nth block is mined. The time difference taken for an
cious nodes have no crypto-IDs. When the vehicles share any infor- attacker to produce and mine fraudulent blocks is called time ad-
mation with other vehicles, their trust values increase or decrease vantage. (iii) The probability for an attacker to perform the attack
based on the valid or invalid information being shared. Moreover, is D S (q, K , n, t ). Assuming that the largest share of the network
the trust value for each vehicle is computed via the controller node computing power is controlled by the attacker and the mth block
and it is calculated using Equation (1). The trust value is computed is mined by the non-malicious nodes. The least amount of confir-
as the ratio of successful transactions that are performed (i.e., valid mations needed for accepting blocks and transactions as legitimate
or correct) to the total number of transactions that are executed is defined as K ∈ N .
(i.e., valid and invalid transactions). Based on the trust value, the This attacker model uses the generalized attack model of [49].
reliability of both the IoV and miner nodes, and the system is de- We need to have knowledge for possible blocks’ number that are
termined. mined at τ sec to define P . Assume a(q, t , n) represents the likeli-
Number of successful transactions performed hood to mine n number of blocks if a node consistently mines for
Trust Value = . (1) τ sec with a proportion of q. P is determined using Equation (2)-
Total number of transactions executed
Equation (4) [49].
When IoV nodes exchange data with miner nodes, the miner
nodes respond by providing the appropriate information from the

n
controller nodes’ storage or from their storage. Both the compu- P (q, m, n, t ) = a(q, t , n) P R (q, m, n − z), (2)
tational time and response delay increase whenever the miner
z =0

7
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

⎧
⎪
⎨ 1, if t = 0 and n = 0, Theorem 2. An adversary cannot reveal information or intercept data
a(q, t , n) = 0, if t ≤ 0, (3) within the sending node and the receiving node whether in the block-
⎪
⎩ (qt )n chains or out of them in order to modify the message’s content.
n!
exp (−qt ), otherwise.
⎧
⎪
⎨1 if m = n = 0,
,  Proof. To protect data from passive and active attacks, a decryp-
P R (q, m, n) = m+n−1 (4) tion and encryption mechanism is used, i.e., AES-256+RSA algo-
⎪
⎩ qn (1 − q)m , otherwise. rithm. This algorithm allows IoV nodes to encrypt their requests
n
with miners’ unique public keys and send the requests to them.
P R (q, m, n) is the attacker’s probability to mine n number of Next the miner nodes decrypt the requests using their private keys
blocks prior to the non-malicious node mines the mth block. and store them in the local blockchain after validation. Afterwards,
The probability is derived from M. Rosenfeld double spending at- the miner nodes send response messages to the IoV nodes. This
tack model [49]. The probability is modeled as given in Equa- algorithm helps to protect the system from active and passive at-
tions (5)-(6) [49]. It represents the probability that an adversary tacks. 
performs the attack having an initial benefit of τ sec and n num-
ber of blocks ahead of the non-malicious users. 4. Discussion of simulation results


K −n
In this section, we discuss the proposed system’s performance.
D S (q, K , n, t ) = 1 − P (q, K , z, t )(1 − C (q, K − n − z)), (5)
We developed the system on the Ethereum platform using dif-
z =0
ferent tools. Microsoft’s Visual Studio (VS) code is an environ-
q ( z +1 ) ment used for debugging, editing, and testing code. Solidity is
, if q < 0.5 and z > 0,
C (q, z) = p (6) an object-oriented scripting language for smart contract develop-
1, otherwise.
ment. Ganache allows to create numerous virtual accounts, each
Here p + q = 1, i.e., p is the probability that the attacker with the with 100 ethers. MetaMask is an application that connects the ac-
highest computational resources cannot takeover the network. counts to a web server. While the Remix Integrated Development
Environment (IDE) is a web-based platform that allows users to
3. Security analysis compile, test, and debug smart contract code. The specifications
of the system we used in our tests are 8.00 GB RAM, an Intel(R)
A security analysis of the proposed model is presented in this Core(TM) i7-7500 CPU @ 2.70 GHz, and Windows 10 operating sys-
section. Two theorems define the model’s resistance against double tem. Moreover, in this section, we evaluate the proposed system’s
spending, active, and passive attacks. Theorem 1 defines incentive performance using performance metrics such as execution cost,
based double spending. The analysis demonstrates the efficiency of transaction cost, computational time to store data, computational
our proposed approach in terms of average computational power, time for encryption and decryption, and the system’s robustness
and mining time and block advantage (the number of blocks pro- based on double spending attacks. The performance metrics are
duced by attacker ahead of an honest user is called block advan- explained as follows:
tage). Theorem 2 shows that passive and active attacks are not
possible in the proposed system. • Execution cost: is the cost of operation for each line of code
in a smart contract’s function. It is measured in gas unit.
Theorem 1. A malicious IoV node cannot get any non-monetary incen- • Transaction cost: is the total amount of gas used by the smart
tive without sending a valid or correct sensory information to miner contract functions to send data to the blockchain. It is mea-
nodes and the miner node cannot spend the same incentive twice in the sured in gas unit.
proposed system. • Computational time to store data: is the amount of time re-
quired to store data in storage system.
Proof. Here, an attacker acts as a non-malicious node and broad- • Computational time for encryption and decryption: is the
casts two payment transactions to the blockchain: paying the re- amount of time taken to encrypt and decrypt data in the pro-
ceiver and paying himself/herself. This means that the attacker posed system.
can mine a block comprised of the false transaction in parallel • The system’s robustness against the double spending attacks:
with other non-malicious miners. Once the receiver receives the is the probability that the proposed system can resist double
payment, the attacker releases the secretly mined block for prop- spending attacks.
agating a fork in the network. As a result, the transaction of the
receiver becomes invalid. Once the attacker succeeds in paying 4.1. Results for the proposed blockchain and smart contracts
himself/herself, a new block is added, which creates a false branch
in the network. If the attacker is lucky, his/her false branch be- In Fig. 7a, gas consumption for various functions is evaluated.
comes the longest in the network, which means that there is a There are four functions, which are performed in the proposed ve-
high likelihood that other non-malicious nodes will accept the hicular network: registerVehicle(), requestService(), response(), and
branch of the attacker. validate(). The registerVehicle() function allows the IoV nodes to
In the proposed model, if the attacker node intends to breach register in the network and the requestService() function allows the
the agreement by not giving incentive to the sender node, then IoV nodes to send request for services to both miner and controller
the proposed smart contract withdraws the stipulated token from nodes. The response() function is used by either the miner or the
the attacker’s account after verifying the token to avoid double controller nodes to respond the service requests that are made
spending attack. Upon finding out that no transaction is made from by the IoV nodes, and the validate() function is used to verify if
attacker’s account to the honest IoV node, the smart contract au- the IoV, miner, and controller nodes are allowed to participate in
tomatically forwards the actual incentive to the sender node. Also, the network or not. Transaction cost is more than execution cost
the smart contract waits until a confirmation message is received because it involves both the costs of executing the functions and
from the sender and keeps the record of the incentive spent in the contract deployment cost. We evaluate the execution cost and
order to mitigate the double spending attack.  transaction cost in terms of gas consumption.

8
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

Fig. 7. (a) Gas consumption of service request and response. (b) Gas consumption of incentive and reputation functions.

In the Ethereum platform, each action is considered as a IoV nodes request information, which is not available from miners’
transaction. For every transaction, a gas unit is charged. The storage, the request is sent to the controller nodes, which respond
gas consumption for each opcode is given in the study [50]. by searching for the relevant data. As a result of this process, the
1 gas unit is equivalent to 4 gwei, i.e., 1 Ether is equivalent to computing time increases the delay of the services. In our system,
1, 000, 000, 000 gwei. It is also observed that the execution cost the computing time is shorter than the existing systems since the
and transaction cost of registerVehicle() function are 148925 gas miner nodes are interconnected with the local blockchain and min-
units and 174101 gas units respectively, which are more than the ing nodes have their own cache memory and data storage. Thus,
execution costs and transaction costs of other functions. The reason the searching process takes less time. As a result, the proposed
is that whenever a vehicle is registered in the system, its informa- system is effective in terms of data storage and service delay.
tion is recorded in both the main blockchain and local blockchain. In Fig. 8b, gas consumptions for authentication, cache memory,
Hence, the amount of gas consumed by a function depends on and malicious node detection are evaluated. The functions exe-
the complexity of its opcode. Following the vehicles’ registration, cuted in the proposed system are CacheMemory(), Authentication(),
they interact and share information with one another. The vehi- and MaliciousNodeDetection(). It is observed from the figure that
cles request for information either from miners or controllers. The the execution cost of every function is lower than the transac-
execution cost and transaction cost of the requestService() func- tion cost. The reason is that transaction cost consists of both the
tion are 62635 gas units and 86275 gas units respectively, which costs of executing the functions and the contract deployment cost.
are more than the gas consumption of the response() function The CacheMemory() function is executed when new information is
with 67630 gas units and 42838 gas units of transaction and exe- recorded in the main blockchain or the local blockchain. While the
cution costs respectively. The transaction and execution costs of
recent data is recorded in miners’ cache memory and the con-
the validate() function are 34255 gas units and 9463 gas units
trollers’ cache memory. The Authentication() function is performed
respectively, which are less than the other functions because it
whenever information is communicated with the IoV nodes or the
performs a few numbers of operations. In Fig. 7b, we evaluate
miner nodes. The controllers execute the function when sensory
the gas consumption of the incentive and reputation functions. In
information is to be authenticated. While the MaliciousNodeDetec-
the proposed system, four functions are used to provide incen-
tion() function is executed whenever miners obtain fake informa-
tive and reputation to IoV nodes and miner nodes based on the
tion from IoV nodes. The collected information is validated and
valid response. The functions include: giveIncentive(), giveReputa-
verified by forwarding it to the nearest IoV nodes or miner nodes.
tion(), and viewIncentive(). The execution cost and transaction cost
If the information is confirmed to be fake, then the information
of the giveIncentive() function are 41931 gas units and 65571 gas
is forwarded to the controller nodes. After verifying the histor-
units respectively. The execution cost and transaction cost of the
ical records of the IoV nodes, the controller nodes broadcast a
giveReputation() function are 41921 gas units and 65497 gas units
respectively. Both function costs are approximately the same be- message to the network that malicious nodes are detected. The
cause the functions are triggered whenever IoV nodes give ratings MaliciousNodeDetection() function has the highest transaction and
to miners that give incentives to IoV nodes. For all the functions, execution costs as compared to the other functions because it has
the information of the miners is shared with other IoV nodes and more operational code to execute. Moreover, the function is exe-
miner nodes. The execution and transaction costs of the viewIncen- cuted by the controller nodes and it is done off the chain. The re-
tive() function are 4859 gas units and 27539 gas units respectively. sults obtained from the offchain are stored in the main blockchain.
The costs of viewIncentive() function are lower than compared to The transaction and execution costs of the function are low be-
that of viewReputation() function, i.e., 5277 gas units for execution cause it is executed by the controller node. Moreover, it is obvious
cost and 27957 gas units for transaction cost. that transaction cost and execution cost vary based on implemen-
tation scenarios and different features. In Fig. 9, it is shown that
4.2. Results for the proposed system based on data storage, the overall transaction cost of the proposed system is better than
computational time, and encryption technique compared to the existing system [51]. The reason is that in our sys-
tem, the deployment cost for all the nodes is reduced, and a very
Fig. 8a shows the comparison of the existing [35] and the pro- low cost for the miner and IoV nodes is incurred. This reduction
posed systems based on computational time and data storage. The helps in minimizing the deployment cost of the proposed system.
computing time increases as the size of the data storage grows. Moreover, for the existing model, the transaction cost is only col-
We see that the data generated by vehicles is saved in the con- lected from the study for one city VN and data hash table based
troller node’s storage in the existing systems. In the system, when on PoW consensus mechanism for comparison purpose.

9
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

Fig. 8. (a) Computational time with respect to stored data. (b) Gas consumption of malicious node detection, authentication and cache memory.

less than compared to the existing system because of the usage of

cache memory.
Fig. 10b presents the analysis of service request and response
time with respect to transactions. The request time for information
is large at first while the response time for returning the informa-
tion is small. From transaction t7, the request service time starts
increasing gradually, however, an abrupt change is observed in the
response service time at the same transaction. The reason of this
abrupt change is that miner nodes have to search for the requested
information in the local blockchain. The request time grows at
transaction t9 while the response and request time of services be-
comes almost the same at transaction t11. At that time, the miners
are gathering information of the requested services from the con-
troller nodes. The memory of miner nodes becomes overburdened
with response and information requests when the number of re-
quests grows.

4.3. Results for the double spending attack

Through simulations, we evaluate the impact of double spend-

Fig. 9. Our system and existing system comparison based on overall transaction cost ing attack on the proposed system. The number of generated
deployment. blocks and their generation times during implementation are uti-
lized as experimental inputs. In the test subject, random blocks
In our proposed scenario, an attacker can be either the IoV advantage [49] from -5 to 5 is used. In this paper, if the likelihood
node or the miner node. Thus, the attacker can execute neither the of double spending attacks approaching 1, it means the attack is
MaliciousNodeDetection() nor the authentication() function for the successful. On the other hand, if the likelihood of double spend-
purpose of circumventing the detection and authentication mech- ing attacks approaching 0, it means the attack is unsuccessful. As
anisms. The findings reveal that our system is secured against seen in Fig. 11a, the probability of a double spending attack grows
adversaries and legitimate data is obtained. Moreover, the encryp- as the value of block advantages rises. Furthermore, as shown
tion algorithms convert the actual texts obtained from the source in Fig. 11, the likelihood of double spending attacks rises as the
nodes into cypher texts. As a result, the security of the proposed amount of both pre-mined block as well as computational power
system is improved. In Fig. 10a, AES-128, Affine cipher, and Triple rises. To understand the behavior of the double spending attack,
Data Encryption Standard (3-DES) encryption techniques are com- the results show four individual scenarios for different q values,
pared with the proposed encryption algorithm based on average i.e., 10% to 40% with an interval of 10 for each step.
executional time. The average executional time relies on the en- Fig. 11b depicts the likelihood of double spending attacks
crypted data, size of public key, and encryption technique. It is against the attackers’ computing power, i.e., the green line. From
observed that the 3-DES scheme has the highest executional time the figure, it is worth noting that as the computational power of
for both decryption and encryption than the rest of algorithms. In the attacker increases, the likelihood of double spending attacks
our system, an AES-256+RSA technique is used because it provides increases as well. This occurs as a result of enough time and high
high security; however, it uses a large amount of computational re- speed computing support given to the nodes to create a block
sources and incurs more time than AES-128. Using the encryption where it is more likely that the malicious nodes build an ille-
algorithm in our system, malicious activities are reduced; how- gitimate branch that is long enough to be considered as a valid
ever, the system’s computational complexity is increased when it blockchain. However, in our system, the response time and request
is compared to AES-128, which is the trade-off. However, the pro- time are low, which reduce the likelihood of double spending at-
posed encryption is still better than the rest of the techniques. So, tacks even if the computing power of the attackers is high. Fig. 11b
the proposed system is secured against malicious activities at the shows the probability of double spending attacks against the num-
cost of increased service delays over the network. However, we ob- ber of accepted valid transactions, i.e., the red line. From the figure,
serve from Fig. 8a that the computing time for our system is still we observe that as the number of accepted valid transactions in-

10
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

Fig. 10. (a) Encryption techniques comparison in terms of average execution time. (b) Transactions in terms of response and request of services.

Fig. 11. (a) Double spending attack’s probability vs attackers’ block pre-mined. (b) Double spending attack’s probability vs the acceptance of valid transaction confirmation
and the computing power.

creases, the probability that a double spending attack will occur mechanism and an encryption technique to mitigate malicious at-
reduces. It is also worth noting that if the attacker has more com- tacks on the communication path and to secure sensory informa-
puting power than the non-malicious nodes, the probability of the tion of miner or IoV nodes. We have presented a security analysis
attack is approximately one. This is because the attacker creates of the proposed model, which demonstrated that the model is
blocks before the non-malicious nodes. However, in the proposed robust against double spending, active and passive attacks. From
model, a reputation mechanism is incorporated to discourage ma- the simulation results obtained, the proposed system provides se-
licious activities. Also, we use a token tracking mechanism in the curity, transparency, scalability, and decentralization by utilizing
smart contract to track each token that is used in the system dur- blockchain. Our system is 10% more efficient than the benchmark
ing and after a transaction has taken place. Therefore, the mech- system in terms of computational time.
anisms help to reduce the risk of double spending attack in the
proposed system. Declaration of competing interest

5. Conclusion
The authors declare that they have no known competing finan-
cial interests or personal relationships that could have appeared to
In this study, we proposed a double blockchain based vehicu-
influence the work reported in this paper.
lar system to optimize data storage and secure the communication
channel. In the literature, when the number of IoV nodes increases
in the vehicular network to communicate and share information, a Acknowledgements
secure communication channel and an efficient storage system are
required. Due to the low storage capacities of IoV nodes, a vehic- Sherali Zeadally was supported by a Fulbright U.S. scholar grant
ular network faces high communication delays. In addition, some award administered by the U.S. Department of State Bureau of Ed-
IoV nodes do not trust service providers due to their selfishness. To ucational and Cultural Affairs, and through its cooperating agency
overcome the storage overhead and minimize communication de- the Institute of International Education (“IIE”).
lays, we improved the existing communication and storage mech-
anisms using local and main blockchains with a cloud server. In References
the cloud server, we used the MTP-Argon2 filtration technique to
filter the unnecessary and duplicate data. We also introduced the [1] Sherali Zeadally, Ray Hunt, Yuh-Shyan Chen, Angela Irwin, Aamir Hassan, Ve-
hicular ad hoc networks (VANETS): status, results, and challenges, Telecom-
cache memory technique to minimize the delay in providing in-
mun. Syst. 50 (4) (2012) 217–241.
formation. To encourage the IoV nodes to share data and to build [2] Tianyang Zhang, Himanshu Pota, Chi-Cheng Chu, Rajit Gadh, Real-time renew-
a trusted relationship with each other, we proposed an incentive able energy incentive system for electric vehicles using prioritization and cryp-
mechanism and a reputation system. We used an authentication tocurrency, Appl. Energy 226 (2018) 582–594.

11
A.S. Yahaya, N. Javaid, S. Zeadally et al. Vehicular Communications 37 (2022) 100502

[3] Global EV Outlook 2020, https://www.iea.org/reports/global-ev-outlook-2020, [28] Yao-Tsung Yang, Li-Der Chou, Chia-Wei Tseng, Fan-Hsun Tseng, Chien-Chang
2021. (Accessed 31 March 2021). Liu, Blockchain-based traffic event validation and trust verification for VANETs,
[4] Muhammad Arif, Guojun Wang, Md Zakirul Alam Bhuiyan, Tian Wang, Jianer IEEE Access 7 (2019) 30868–30877.
Chen, A survey on security attacks in VANETs: communication, applications [29] Yan Zhang, Lei Zhang, Burong Kang, Yue Ma, Tong Chen, Secure and reliable
and challenges, Veh. Commun. 19 (2019) 100179. parking protocol based on blockchain for VANETs, in: 2021 IEEE Wireless Com-
[5] Juan Contreras-Castillo, Sherali Zeadally, Juan Antonio Guerrero-Ibanez, Internet munications and Networking Conference (WCNC), IEEE, 2021, pp. 1–6.
of vehicles: architecture, protocols, and security, IEEE Int. Things J. 5 (5) (2018) [30] Hassan Farooq, Muhammad Usman Arshad, Muhammad Faraz Akhtar, Shahid
3701–3709. Abbas, Bilal Zahid, Nadeem Javaid, Block-VN: a distributed blockchain-based
[6] Juan Guerrero-Ibáñez, Sherali Zeadally, Juan Contreras-Castillo, Sensor tech- efficient communication and storage system, in: International Conference
nologies for intelligent transportation systems, Sensors 18 (4) (2018) 1212. on Broadband and Wireless Computing, Communication and Applications,
[7] Wenchao Xu, Haibo Zhou, Nan Cheng, Feng Lyu, Weisen Shi, Jiayin Chen, Springer, Cham, 2019, pp. 56–66.
Xuemin Shen, Internet of vehicles in big data era, IEEE/CAA J. Autom. Sin. 5 (1) [31] Muhammad UmarJaved, Mubariz Rehman, Nadeem Javaid, Abdulaziz
(2017) 19–35. Aldegheishem, Nabil Alrajeh, Muhammad Tahir, Blockchain-based secure
[8] Adia Khalid, Muhammad Sohaib Iftikhar, Ahmad Almogren, Rabiya Khalid, data storage for distributed vehicular networks, Appl. Sci. 2020 10 (6) (2011).
Muhammad Khalil Afzal, Nadeem Javaid, A blockchain based incentive provi- [32] Yang Xu, Guojun Wang, Jidian Yang, Ju Ren, Yaoxue Zhang, Cheng Zhang,
sioning scheme for traffic event validation and information storage in VANETs, Towards secure network computing services for lightweight clients using
Inf. Process. Manag. 58 (2) (2021) 102464. blockchain, Wirel. Commun. Mob. Comput. (2018) 1–13.
[9] Salabat Khan, Liehuang Zhu, Xiaoyan Yu, Zijian Zhang, Mussadiq Abdul Rahim, [33] Jiawen Kang, Zehui Xiong, Dusit Niyato, Dongdong Ye, Dong In Kim, Jun Zhao,
Maqbool Khan, Xiaojiang Du, Mohsen Guizani, Accountable credential manage- Toward secure blockchain-enabled Internet of Vehicles: optimizing consensus
ment system for vehicular communication, Veh. Commun. 25 (2020) 100279. management using reputation and contract theory, IEEE Trans. Veh. Technol.
[10] Sherali Zeadally, Muhammad Awais Javed, Elyes Ben Hamida, Vehicular com- 68 (3) (2019) 2906–2920.
munications for ITS: standardization and challenges, IEEE Commun. Stand. Mag. [34] Lixia Xie, Ying Ding, Hongyu Yang, Xinmu Wang, Blockchain-based secure
4 (1) (2020) 11–17. and trustworthy Internet of things in SDN-enabled 5G-VANETs, IEEE Access 7
[11] Ikram Ali, Alzubair Hassan, Fagen Li, Authentication and privacy schemes for (2019) 56656–56666.
vehicular ad hoc networks (VANETs): a survey, Veh. Commun. 16 (2019) 45–61. [35] Pradip Kumar Sharma, Seo Yeon Moon, Jong Hyuk Park, Block-VN: a distributed
[12] Lei Zhang, Mingxing Luo, Jiangtao Li, Man Ho Au, Kim-Kwang Raymond Choo, blockchain based vehicular network architecture in smart city, J. Inf. Process.
Tong Chen, Shengwei Tian, Blockchain based secure data sharing system for Syst. 13 (1) (2017) 184–195.
Internet of vehicles: a position paper, Veh. Commun. 16 (2019) 85–93. [36] Fangmin Xu, Fan Yang, Chenglin Zhao, Chao Fang, Edge computing and caching
[13] Zehui Xiong, Yang Zhang, Nguyen Cong Luong, Dusit Niyato, Ping Wang, Nadra based blockchain IoT network, in: 2018 1st IEEE International Conference on
Guizani, The best of both worlds: a general architecture for data management Hot Information-Centric Networking (HotICN), 2018, pp. 238–239.
in blockchain-enabled Internet-of-Things, IEEE Netw. 34 (1) (2020) 166–173. [37] Sergii Kushch, Francisco Prieto-Castrillo, Blockchain for dynamic nodes in a
[14] David Berdik, Safa Otoum, Nikolas Schmidt, Dylan Porter, Yaser Jararweh, A smart city, in: 2019 IEEE 5th World Forum on Internet of Things (WF-IoT),
survey on blockchain for information systems management and security, Inf. 2017, pp. 29–34.
Process. Manag. 58 (1) (2021) 102397. [38] Jun Lin, Zhiqi Shen, Chunyan Miao, Siyuan Liu, Using blockchain to build
[15] Satoshi Nakamoto, Bitcoin: a peer-to-peer electronic cash system, http:// trusted LoRaWAN sharing server, Int. J. Crowd Sci. 1 (3) (2017) 270–280.
bitcoin.org/bitcoin.pdf, 2008. [39] Raphael Rosa, Christian Esteve Rothenberg, Blockchain-based decentralized
[16] Adamu Sani Yahaya, Nadeem Javaid, Ahmad Almogren, Abrar Ahmed, Sardar applications for multiple administrative domain networking, IEEE Commun.
Muhammad Gulfam, Ayman Radwan, A two-stage privacy preservation and Stand. Mag. 2 (3) (2018) 29–37.
secure peer-to-peer energy trading model using blockchain and cloud-based [40] Chenhan Xu, Kun Wang, Peng Li, Song Guo, Jiangtao Luo, Baoliu Ye, Minyi Guo,
aggregator, IEEE Access 9 (2021) 143121–143137. Making big data open in edges: a resource-efficient blockchain-based approach,
[17] Adamu Sani Yahaya, Nadeem Javaid, Muhammad Umar Javed, Ahmad Almo- IEEE Trans. Parallel Distrib. Syst. 30 (4) (2018) 870–882.
gren, Ayman Radwan, Blockchain based secure energy trading with mutual [41] Oscar Novo, Scalable access management in IoT using blockchain: a perfor-
verifiable fairness in a smart community, IEEE Trans. Ind. Inform. (2022). mance evaluation, IEEE Int. Things J. 6 (3) (2018) 4694–4701.
[18] Andreas Kamilaris, Ian R. Cole, Francesc X. Prenafeta-Boldu, Blockchain in agri- [42] Madhusudan Singh, Shiho Kim, Branch based blockchain technology in intelli-
culture, in: Food Technology Disruptions, Academic Press, 2021, pp. 247–284. gent vehicle, Comput. Netw. 145 (2018) 219–231.
[19] Fuqin Wang, Yijiang Chen, Ruochen Wang, Akindipe Olusegun Francis, Bug- [43] Zhe Yang, Kan Yang, Lei Lei, Kan Zheng, Victor CM Leung, Blockchain-based
ingo Emmanuel, Wei Zheng, Jinjun Chen, An experimental investigation into decentralized trust management in vehicular networks, IEEE Int. Things J. 6 (2)
the hash functions used in blockchains, IEEE Trans. Eng. Manag. 67 (4) (2019) (2018) 1495–1505.
1404–1424. [44] Jiawen Kang, Rong Yu, Xumin Huang, Maoqiang Wu, Sabita Maharjan, Shengli
[20] Anichur Rahman, Md Jahidul Islam, Ziaur Rahman, Md Mahfuz Reza, Adnan An- Xie, Yan Zhang, Blockchain for secure and efficient data sharing in vehicular
war, MA Parvez Mahmud, Mostofa Kamal Nasir, Rafidah Md Noor, DistB-Condo: edge computing and networks, IEEE Int. Things J. 6 (3) (2018) 4660–4670.
distributed blockchain-based Iot-SDN model for smart condominium, IEEE Ac- [45] Xiaohong Zhang, Xiaofeng Chen, Data security sharing and storage based on
cess 8 (2020) 209594–209609. a consortium blockchain in a vehicular ad-hoc network, IEEE Access 7 (2019)
[21] Partha Pratim Ray, Dinesh Dash, Khaled Salah, Neeraj Kumar, Blockchain for 58241–58254.
IoT-based healthcare: background, consensus, platforms, and use cases, IEEE [46] Zhaojun Lu, Wenchao Liu, Qian Wang, Gang Qu, Zhenglin Liu, A privacy-
Syst. J. 15 (1) (2020) 85–94. preserving trust model based on blockchain for VANETs, IEEE Access 6 (2018)
[22] Weiwei Li, Zhou Su, Ruidong Li, Kuan Zhang, Yuntao Wang, Blockchain-based 45655–45664.
data security for artificial intelligence applications in 6G networks, IEEE Netw. [47] Xingchen Liu, Haiping Huang, Fu Xiao, Ziyang Ma, A blockchain-based trust
34 (6) (2020) 31–37. management with conditional privacy-preserving announcement scheme for
[23] Li Yue, Huang Junqin, Qin Shengzhi, Wang Ruijin, Big data model of security VANETs, IEEE Int. Things J. 7 (5) (2019) 4101–4112.
sharing based on blockchain, in: 2017 3rd International Conference on Big Data [48] Ammar Riadh Kairaldeen, Nor Fadzilah Abdullah, Asma Abu-Samah, Rosdiadee
Computing and Communications (BIGCOM), 2017, pp. 117–121. Nordin, Data integrity time optimization of a blockchain IoT smart home net-
[24] Rasheed Hussain, Jooyoung Lee, Sherali Zeadally, Trust in VANET: a survey of work using different consensus and hash algorithms, Wirel. Commun. Mob.
current solutions and future research opportunities, IEEE Trans. Intell. Transp. Comput. 2021 (2021).
Syst. 22 (5) (2020) 2553–2571. [49] Carlos Pinzón, Camilo Rocha, Double-spend attack models with time advantage
[25] Yongjun Ren, Yepeng Liu, Sai Ji, Arun Kumar Sangaiah, Jin Wang, Incentive for bitcoin, Electron. Notes Theor. Comput. Sci. 329 (2016) 79–103.
mechanism of data storage based on blockchain for wireless sensor networks, [50] Gavin Wood, Ethereum: a secure decentralised generalised transaction ledger,
Mob. Inf. Syst. 2018 (2018) 1–10. Ethereum Project Yellow Paper 151(2014) (2014) 1–32..
[26] Di Lin, Yu Tang, Blockchain consensus based user access strategies in D2D net- [51] Sushil Kumar Singh, Pradip Kumar Sharma, Yi Pan, Jong Hyuk Park, BIIoVT:
works for data-intensive applications, IEEE Access 6 (2018) 72683–72690. blockchain-based secure storage architecture for intelligent Internet of vehicu-
[27] Tigang Jiang, Hua Fang, Honggang Wang, Blockchain-based Internet of vehicles: lar things, IEEE Trans. Consum. Electron. (2021).
distributed network architecture and performance analysis, IEEE Int. Things J.
6 (3) (2018) 4640–4649.

12