Improving the fail-operational behavior of automotive applications by artificial DNA

Embedded systems are growing very complex because of the increasing chip integration density, larger number of chips in distributed applications, and demanding application fields, eg, in autonomous cars. Bio-inspired techniques like self-organization are a key feature to handle this complexity. In biology, the structure and organization of a system is coded in its DNA. We adapted this concept to embedded systems using an artificial DNA (ADNA). Based on the ADNA, the self-organization mechanisms can build the system autonomously at run-time providing a self-building system. This property predestines the ADNA for the use in automotive applications because modern (autonomous) cars include several highly redundant processors (electronic control units (ECUs)). The ADNA can be used to reduce the number of ECUs in a car on the one hand and to make better use of the cars' redundant ECUs on the other hand. Our contribution in this paper is to evaluate the improvements possible due to the ADNA by analyzing the fail-operational limits and failure probabilities in such scenarios. We also propose a simple graceful degradation scheme for the tasks to improve the system dependability of the cars. Finally, the usability of the concept is demonstrated by a practical evaluation.

Automated summary

Embedded systems are becoming more complex due to increasing chip integration density, a larger number of chips in distributed applications, and the demands of challenging application fields such as autonomous cars. Bio-inspired techniques like self-organization play a crucial role in handling this complexity. Utilizing artificial DNA (ADNA), self-organization mechanisms can autonomously build systems at run-time, making ADNA suitable for automotive applications with highly redundant processors.