IEEE Consumer Electronics Magazine - January/February 2023 - 120

Security, Trust and Privacy Solutions for Intelligent Internet of Vehicular Things
Control field: Indicate two reserved and four
bits as data length code.
Data field: Carries the real data message conveyed
to other nodes.
CRC field: Assures the efficacy of the message.
All connections that accept the information
confirm the message, using this field.
ACK field: Acknowledgment bits including
ACK and delimiter part. A connection that
accepts the actual normal message returns
the ACK part. It restores the ACK part, a suspended
bit to the dominant bit (i.e., logical 1
to 0).
Figure 3. CAN-bus attack scenario, the intruders or
attackers sent malicious messages using external
communication links. These attack messages are
received at OBD port/system. The deep learningbased
intrusion detection system detects these
attacks and prevents vehicles from false
communications.
processing and communication between vehicles
other types of equipment. The internal data of the
vehicle is transferred through the CAN bus during
the communication process.
In Figure 3, as illustrated, amodern vehicle consists
of ECUs for information collection; few of
them are connected through the CAN bus. These
ECUs communicate messages between or within
vehicles by utilizing the CAN bus protocol,
which is efficient due to its low expense and
centralized system for in-vehicular network
operations. However, due to low-security features,
it usually undergoes through different
intruders attacks. The attackers can easily
inject malicious data packets (DoS and fuzzy)
into CAN message by utilizing communication
interfaces. The CAN frame message is comprised
of the following seven fields.
Start offrame: First predominant bit notifies a
start of communication of CAN message to
all nodes.
Arbitration field: Consists of two components;
the identifier (describes the ID of the frame,
utilized throughout the arbitration process)
and remote transmission request (defined
according to the type of the CAN frame).
120
End of frame: Infers the last head of the CAN
message frame.
Deep Learning-Based Intrusion Detection
System
As discussed earlier, attackers can easily access
the CAN bus due to the in-vehicle network's
vulnerability and absence of security features. In
this work, we presented a smart intrusion detection
system to investigate, classify, or detect two
kinds of attacks, i.e., DoS and Fuzzy.
In DoS attacks, too many requests are transferred
to the server of the network. As wireless
technologies are used in vehicles, thus,
it is obvious to drive a DoS attack. The
attacker and intruder can insert higher priority
information or messages in a very small
time period.
In fuzzy attacks, the intruder inserts information
of randomly spoofed identifiers, including
inconsistent data. As an outcome, all
network connections or nodes receive many
operative messages and may go to the
network's malfunction. To inject this kind of
attack, the intruder recognizes vehicle messages
and decides the targets.
To detect these attacks, a deep learning-based
architecture is used, as demonstrated in Figure 3.
The deep learning architecture detects attacks
when received on the OBD port. Our deep learning
system is based on the VGG-16 architecture, as
shown in Figure 4. We first converted the message
samples into a byteplot grayscale image,12 utilized
for the automatic classification of visual patterns.
These images are variable resolution images with a
IEEE Consumer Electronics Magazine
IEEE Consumer Electronics Magazine - January/February 2023

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - January/February 2023
