Over the past decade, car designers added tons of new electronic features and functions to new vehicle models. In lockstep with the number of functions increased the number of Electronic Control Units (ECUs): “Since the traditional ECUs were – and are – single-function devices, the developers had to add a new ECU for more or less every new feature,” Burcicki said. The result is known: A jungle of ECUs grew under the engine bonnets.
Now, as vehicle electronics is setting off for highspeed data communications, fueled by the trend towards compute-heavy autonomous vehicles, the situation is more complex than ever. And the upcoming electrification of the drivetrain will not exactly contribute to a relief, Burcicki noted: With today’s technology, connectivity, driving automation and electrification will more than quadruple the harness, he said. It is easy to see that without a careful, future-oriented redesign not only of the wiring systems, the road to more complexity leads to a dead end.
The rising significance of software is adding to the complexity. In today’s cars it is not unusual that the embedded software amounts to millions of lines of code (LoC). The software modules distributed across dozens of ECUs need to communicate among each other, and they trigger more communications processes between sensors and ECUs, among the various ECUs and between the car and the backend infrastructure. And all these communications processes require wires for the transport of data. What’s more, the evolution of the in-car networks is going to accelerate because future vehicle generations will increasingly add new functions even after they have left the production line – through over-the-air (OTA) updates.
The question is: What to do. In autonomous vehicles, the “thinking” part (of the “sense-think-act” chain of effects of automatic systems) is the toughest challenge for OEMs, Burcicki said. To master the challenge of this dynamic growth of complexity, it will be necessary to fundamentally re-design the harness architecture. “Wiring has to be adaptable to changes coming ten years down the road,” the Mentor mastermind said.
Guiding lines for the re-design are, among other, the functional allocation and the software configuration of future cars. Towards this end, the harness systems need to become more flexible. An important aspect is in the future, the ECUs need headroom to grow into future application worlds, which means they need to be able to accommodate even more lines of code, and, if necessary, they need to be able to provide more computing power. Therefore, a certain level of redundancy will be required – at the ECU level as well as at the cable harness level.
To meet the requirement of flexible provision of computing horsepower, automotive electronics designers are typically betting on more centralized compute platforms in the vehicles, with former physical ECUs being turned into virtual ones – the ECUs become a piece of software, running on a powerful compute platform. At first glance, it looks as if this approach reduces the cabling requirements. But Burcicki has also observed a contrary trend: “In future generations of autonomous vehicles, we will see more diversity, more different approaches, and more computing horsepower will migrate into the sensors”, he said.
Under the bottom line, it becomes obvious that adapting the wire harness design to automated driving and vehicle connectivity won’t be a trivial task. “Actually, this is one of the most challenging things in automotive design,” Burcicki said. “OEMs need to evolve their processes to integrate across domains, automated design tasks and provide robust data coherency”, he concluded.