Page 29 - EngineerIT August 2022 Digital

ARTIFICIAL INTELLIGENCE

AI and defending the future of autonomy

By Prof Ben Azvine, global head of security research and distinguished engineer, BT

onnected and Autonomous
Vehicles (CAVs) are changing the
Cway we think about mobility -
and transforming the digital network
infrastructure that supports these vehicles.
Soon, every car will be connected and
able to talk to everything around it using
Vehicle-to-Everything (V2X) technology,
including other cars and smart city
infrastructure.
But as with any connected computing
device, the advent of CAVs brings with
it security threats, as the number of
computing and network components both
inside and outside a car will dramatically
increase - raising the attack surface for
criminals to wreak havoc.
The infamous Jeep hack in 2015 is a Automakers and OEMs not only have to consider the security of the firmware and
prime example. Two white-hat hackers software against the typical threat of cyberattacks, but it’s also made more complicated by
were successfully able to hack a Jeep on a connected IoT system where one vulnerability could open up the system to even more
the motorway - stalling the engine to bring threats. With a lack of standards and potentially millions of OEM devices connecting with
it to a complete stop. Fiat Chrysler recalled each other, it’s nearly impossible to know exactly where the threat will come from.
1.4m cars in response, and it resulted in
the first legal dispute of its kind. Building our defences
While in some industries security breaches only disrupt business operations, for CAVs,
New threats on the horizon breaches can be a matter of life or death. For autonomous vehicles to become a reality, they
While the hackers in the Jeep incident need to be safe, reliable and resilient enough to earn the public’s trust. Fortunately, the
didn’t harbour any sinister intentions, solution lies with the technology that will make autonomous driving a reality in the first place
others won’t be so kind. As connectivity - AI and machine learning.
grows to become the nervous system of One project showing AI’s budding potential is the BT Mobius project. Trialled in 2020,
our new mobility infrastructure, connecting the project showed that AI powered epidemiological methods - typically used to model
our cars to base stations and our future the spread of pathogens - could be deployed to model the spread of malware over V2X
smart cities, we will start to see malicious communications in CAVs. This is potentially game changing in turning the tide against
actors spread code with increased virality malicious actors. If deployed across our entire mobility infrastructure, it could function as the
- much like how biological pathogens immune system for the network, responding to an attack in the same way our white blood
spread. For automakers and Original cells would respond to infection - using SDN-based security applications to quickly block
Equipment Manufacturer (OEM) device malicious intruders at the entry point where possible, otherwise change security controls to
producers, tracking all the vulnerabilities slow down and limit the spread of the attack or divert them to firewalls before they do more
across the systems running their CAVs is a damage to the network.
challenging task.
Firstly, because there are no On the road to autonomous driving
established standards for vehicle Connectivity is the key force bringing us closer to a future where automation plays a central
security. For instance, the EU only offers role in our daily lives, particularly where mobility is concerned. However, as we advance
recommendations on how auto and OEMs towards that goal, we must keep our eye on the hazards that threaten to throw us off course.
should cope with cybersecurity challenges Connectivity providers can help automakers and OEMs monitor and protect their connected
for autonomous driving. The absence of vehicles from future threats, creating a secure operational environment that will enable
clear, defined technical requirements for them to maintain control over their cars and safeguard the passengers that ride them. AI and
autonomous driving security means getting machine learning are the indispensable tools that will protect against cyber threats to the
security right in autonomous cars is tricky. mobility networks of today, and tomorrow. n

EngineerIT | August 2022 | 27