Why software-defined car signals a paradigm shift

The quantity and complexity of electronic devices used in cars expand with each new generation. A typical vehicle built in 2000 had approximately 10 processors and featured a few thousand lines of code. Fast forward 20 years, and today’s cars have about 45 processors with hundreds of millions of lines of code. 

In the traditional vehicle architecture, with these devices connected by copper, the wiring loom becomes hugely complex and represents a high proportion of the vehicle build cost. It’s fair to say that automotive production has been evolving in this way for the last 100 years or so. For a vehicle designed using a traditional architecture, we’ve reached a ceiling of capability. 

The software-defined vehicle represents a radical, even disruptive, departure from this ethos, separating the hardware from the software – what we call abstraction — and moving away from the “flat” architecture traditionally seen in conventional vehicle design. 

The software-defined car will feature two parallel architectural changes, namely zones and domains. Three or four zones are likely, with control units merged and centralized into high-powered computers. Wiring becomes simpler in a zonal approach, and the software environment more scalable and flexible with domains connected by automotive ethernet, accessed via domain controllers. The software is easily upgradeable by centralized over the air (OTA) updates and efficiently supports the user-defined vehicle. 

Use cases can be made virtual, and the real-time requirements of safety systems with guaranteed response times (a design imperative) assured. Using machine learning and AI in such systems yields more human-like responses than the conventional software code’s “if-then-else” nature. Here, secure vehicle-to cloud connectivity becomes increasingly important — relying on the seamless integration of different ecosystems. In this instance, technological disruption is not only the emergence of specific new hardware platforms but also the convergence of several technologies and methodologies.

Features such as ADAS, further enhanced by artificial intelligence and machine learning, bring the horizon closer for Level 4 and Level 5 autonomy, the driver for much of this advancement. Hardware abstraction enables the hardware to be interchangeable, allowing new entrants to enter the automotive arena, be they tech giants or new startups. 

The product lifespan will be lengthened, too, as a vehicle’s feature set is extended and optimized through over the air updates post-point of sale (POS). They could create new revenue streams and business models from the sale of applications and driver data captured via analytics.