Today's luxury-class cars contain between 80 and 100 electronic control units (ECU) — or embedded computers — which are responsible for activating vehicle functions via software. In current automotive system architectures such as AUTOSAR, these functions are permanently allocated to specific ECUs.
Because the number of ECUs for integrating new functions cannot be increased at will, new concepts are required that will allow more flexible mapping of functions to the ECUs. In addition, a growing number of functions, such as adaptive cruise control during traffic congestion, will rely directly on driving situations. Achieving a high degree of resource efficiency in this environment calls for new dynamic concepts. That means vehicle software must also adhere to the high level of quality demanded by the automotive industry.
With the goal of creating automotive system and software architectures that can satisfy these requirements, Fraunhofer-Institute for Cognitive Systems IKS researchers are actively working on self-adaptive E/E system concepts. This new approach demonstrates several advantages over current technology:
1. Improved Resource Efficiency
Better use of the existing hardware resources can be achieved through adaptive activation and deactivation of the software-based functions in specific driving situations. Energy consumption is reduced as a result.
2. Increased Reliability without Sacrificing Efficiency
The development of reliable software systems requires extensive effort. Self-adaptive systems provide more robustness without the need for additional hardware redundancy. To cite one example, a faulty ECU could be compensated for by migrating the functions to other ECUs, thus making the E/E system more fault tolerant.
3. Maintenance and Expandability
Replacing automotive components or retrofitting/upgrading a vehicle with after-market products or consumer devices is simplified because adaptive software systems can dynamically adjust to new system conditions at runtime.