The car is increasingly evolving into a rolling computer on wheels. For a long time now, most innovations in the automotive sector have been centered around electronics and software. Developments such as mobility services, autonomous driving or automated driving, as well as connectivity, infotainment and e-mobility all contribute to this. With the number of software-led functions in the car, the IT architecture in the car is also changing, moving towards central, virtual high performance computers.
Volkswagen was the first OEM that appointed a software board in the management team. At the same time, the group bundled its software development department into an independent business unit. This is a significant indicator of the direction automotive development is going in. The functions in cars are becoming increasingly interconnected, with more and more sensors and electronics. The proportion of software in cars is growing significantly. According to the group CEO of Continental, the sales price of the software in a premium car in 2023 will be around 40 percent. This also shows how the professional fields are developing. At the OEMs, the proportion of engineers in the IT and software departments has doubled. The various innovation cycles of cars and electronic applications also have consequences for the development of electronics.
Over 100 control units, up to eight kilometers of wiring: cars have become rolling networks. According to a study from the Wiesbaden consultancy firm, Invensity, 90 percent of automotive innovations today are in the areas of electronics and software. However, the increase in software components in cars is posing a challenge for the existing software architecture. Car manufacturers are therefore looking for new ways to manage software efficiently, develop new functions more quickly, to guarantee scalability and reduce costs with more efficient development processes.
Electronics and software are the basis for new functions in the car. The variety of functions and proportion of car software in car electronics is growing. This increases the complexity of the system and software architectures in the car. The number of customer-specific trim levels makes the systems even more complex. Modern cars have over 100 control devices, in which computer programs control specific in-car functions. As the number and complexity of software functions will continue to rise, current standard software architectures are reaching limits. The proportionate cost of electronic components compared with the rest of the components will potentially grow from the current 16 percent to around 35 percent by 2025, according to a study by Roland Berger. For a half-autonomous driving, electric vehicle, these components equate to costs of around 7,000 dollars per car.
The introduction of the software-controlled car will lead to significant changes across the entire value chain.
The increase in electronics and software in cars poses great challenges for electric and electronic (E/E) architecture. At the very latest, automated and autonomous driving will surpass the limits of existing E/E architectures. It will be confronted with huge streams of data and will grow significantly in volume – for example, installation space, weight, number of control units, bandwidth. Until now, electronic architectures in cars were made up of decentralized control units (ECU: Electronic Control Units) that controlled individual functions, and which contained tightly integrated hardware and software. Today, the trend is moving towards centralized systems with dedicated controls that are becoming increasingly virtualized. These central control devices will then carry out multiple functions in different areas of the vehicle. Along with this, hardware and software will be increasingly disconnected from one another. Instead of trying to cover 120 separate electronic control units (ECUs), they will be less widely distributed and more centralized.
Over the next few years, the architectures will be arranged into certain domains, made up of multiple different, combined data groups. In the long term, they will be further centralized into zonal clusters of high performance computers. And ultimately, specific functions will be incorporated to ensure consistent connectivity in the cloud. In future architectures, the control units will be consolidated into just a few high performance computers (HPC). IT system architectures are coming to cars. With these architectures, hardware is being increasingly separated from software. To limit the number of additional control units, virtual control units will run on the high performance computers. Through this central architecture, many calculation steps travel directly to the sensors and actuators, thus creating intelligent sensors and actuators.
