Data management: the challenge of connected cars

autonomous driving

A recent event hosted by the Cambridge Wireless Automotive & Transport Special Interest Group and focused on autonomous driving was called “Are we nearly there yet?”.

Judging from the topics the event looked at, which included autonomous vehicles from the perspectives of explainable AI, emerging standards and regulations, ethics and accountability, legal aspects and data generation and ownership, we’re still a long way from our destination.

It’s true that technology is making connected vehicles a reality – and will eventually enable autonomous vehicles. But the technological advances that enable this will take time to perfect. They will also bring new challenges, such as the ones that informed my own presentation at the event titled ‘How much data is enough, who decides and who owns it?’.

A vehicle isn’t just about wheels, an engine and a chassis anymore. It’s about hardware, software and the cloud. Onboard sensors, V2X, edge, cloudification, onboard computing, storage and support for AI and ML will eventually be part of almost every vehicle, for the benefit of safety, comfort and convenience.

Real Wireless is more aware of this than most. We have strong and relevant experience of techno-economic modelling for autonomous driving testbeds – in particular AutoAir, a UK Department for Digital, Culture, Media, and Sport funded 5G Testbed looking at future wireless vehicle connectivity.

There seems to be strong support for the role of wireless in the development of connected and autonomous vehicles. The evolution of 4G/5G vehicle connectivity standards is continuing and in particular the incorporation of V2X services within 5G standards. Our own work on the AutoAir project aimed to help mature and prove the technology that will make fully connected and autonomous vehicles a commercial reality.

The AutoAir consortium demonstrated, through the use of both mmWave and sub 6GHz wireless networks, the ability to connect vehicles moving at high speed and to stream data – both between vehicles and to fixed infrastructure.

This was useful to prove that the connected vehicle future can and will happen. AutoAir also showed how the different connected vehicle use cases could be enabled.

But, even at this early stage, it has also become clear that connected vehicles and autonomous driving will generate vast amounts of data. Thus, while we are still looking at the enabling technologies for connected vehicles, this work has led us to look a little further forward – to a time when vehicles generate ever-increasing amounts of data. My presentation looked at why vehicles will generate such large amounts of data, the scale of this data generation and who will use or monetise this data? And who owns it?

A vehicle that can communicate is about more than just not bumping to things. For example, a vehicle-based software ecosystem can be programmed to manage everything from insurance, payments and financial services to energy use, maintenance and repair.

But not bumping into things is still important – and it generates large amounts of useful data from a growing array of on-board sensors. Advanced driver assistance systems (ADAS), 3D environment scans using LiDAR, forward, sideways and rear facing cameras spotting everything and everyone on or near the road, sonar, GPS, digital mapping, vehicle operational sensors, interior occupant presence and alertness sensing, driver understanding systems and more will be built into cars. Can the information generated by these sensors be used for AI training to improve safety and the driver experience – not to mention advertising of targeted products and services?

In theory, yes. This information could benefit predictive maintenance, component analytics, collision avoidance, real-time route management, road operator traffic optimisation, mobility as a service (MaaS) operations, concierge services and more, for the benefit not just of drivers but police, roadside assistance, vehicle component providers, insurance providers, other road users and many more.

But who will control access to this and other forms of vehicle data? No one wants to be accused of collecting or using sensitive information without permission. Thus the likelihood is that GDPR-style consent will be involved, backed up by penalties and consequences. We need to start thinking about this now.

But there’s another issue. Forecasts suggest that there be 400 million connected vehicles on the road by 2025 and that globally, connected vehicles could produce up to 10 exabytes of data per month by 2025. 10 exabytes of data per month could, potentially, be a drop in the ocean by, say, 2030. It is certainly true to say that opinions on the volume of data generated by tomorrow’s vehicles vary widely – a range of 4 – 100 GB per vehicle per day being typical.

Then there’s the question of how much of this data will remain in the vehicle and how much will be passed into the cloud, for use by others. But what uses and by whom? This inevitably raises many questions – Who manages all of this data? Which data can be ignored? Which data is valuable, to whom and for what purposes? How complex is the ecosystem in accessing and utilising this data? How is it monetised or valorised? Who regulates its use? And – yet again – who owns it?

As I did some research for my presentation, it became apparent that this ecosystem is in its infancy and many of the questions I asked have no real answers today. Indeed, with vehicle ownership patterns changing and Mobility as a Service (MaaS) growing the ecosystem is getting more complex with vehicle OEMs starting to own vehicle fleets, recognising the shift to MaaS.

Real Wireless may be better positioned than most to answer the many questions around connected and autonomous vehicles, as we have a foot in the technology and data supervision camps, both through projects like AutoAir and through our understanding and experience of the sort of regulatory processes necessary to guide the management of CAV-supplied data.

So, while we wouldn’t answer the question “Are we nearly there yet?” with a resounding “Yes”, we feel that, where data management in the connected vehicle future is concerned, we’re definitely on the right road.