IEEE Consumer Electronics Magazine - January/February 2022 - 51

complexity of security analyses by leveraging unsupervised machine learning to learn
clusters of messages passed between ECUs that correlate with changes in the CPS state
of a vehicle as it moves throughout the world. Our approach can help to improve the
security of vehicles in a smart city, and can leverage smart city infrastructure to further
enrich and refine the quality of the machine learning output.
& THE NOTION OF smart city includes smart transportation
and the management of the infrastructure
that supports its operations. Connected
vehicles are important technologies supporting
these services. The security issues around the use
of connected vehicles in smart city ecosystems
are becoming increasingly prominent due to the
rising complexity of vehicles as a complex, integrated
cyber-physical system (CPS).1 ACPS is
described by the intercommunication between
the computational devices and their physical environment.
A modern vehicle is a CPS consisting of
hundreds of electronic control units (ECUs) to
improve the efficiency of the vehicles in terms of
safety, automation, and comfort. These ECUs also
stand between the vehicle controls and the outside
world, including other connected vehicles
and smart city infrastructure as depicted in
Figure 1. The communication among these ECUs is
facilitatedby the in-vehicle network to achieve personalized
vehicle configuration and enhanced connectivity.
However, these connectivities expose
the vehicular systems to several emergent vulnerabilities,
cyber threats, and physical compromise.
An essential design consideration of the smart city
is to account for the security of the computational
components of the CPS vehicles within their
broader ecosystemas depicted in Figure 2.
The adoption of the controller area network
(CAN) bus for communication between the ECUs
has led to the complexity of network attacks on
connected vehicles. Researchers have demonstrated
different attack surfaces that can be used
to compromise vehicle operations remotely in
modernvehicles due in large part to the integration
with CAN bus.2 A plethora of approaches have
been proposed for securing the CAN bus, which
lacks security by design.3,4 The CAN bus has a
standardized message transmission and operation,
but the implementation varies for different manufacturers
and vehicle models, make, or even trim.
A de facto standard way to describe a specific
implementation is with a CAN database file known
as DBC (after its extension, .dbc). The DBC contains
the translation for all CAN messages in a particular
vehicle, and it is typically held secret by
manufacturers and varies by implementation.
Algorithmic vulnerability analysis and attack
detection, e.g., with supervised machine learning,
requires the analysis of raw CAN messages
by manually inspecting high volumes of data to
reverse engineer message syntax and semantics
to reconstruct the vehicle operations and contextualize
the messages. Although network traffic
analysis and packet recognition exists for
enterprise networks and Internet protocols, the
prior work in this area are inapplicable to the
automotive network due to differences in the
Figure 1. In-vehicle security impacts the integrated
smart city ecosystem.
January/February 2022
Figure 2. CPS in a smart city ecosystem.
51

IEEE Consumer Electronics Magazine - January/February 2022

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - January/February 2022