Evaluation Methodologies of Dynamically Reconfigurable Systems in the Automotive Industry 2020-01-1363
The technology for self-driving cars and other highly-automated applications are becoming more and more advanced. At the same time, Electrical/Electronic (E/E) architectures are becoming more complex. Classical decentralized E/E architectures based on a large number of Electronic Control Units (ECU) represent an obstacle for the realization of new applications due to the computational power, energy consumption, weight, and the size of embedded components constraints in the automotive industry. Therefore the adoption of new embedded centralized E/E architectures represents a new opportunity to tackle these challenges. However, they also raise concerns and questions about their safety, hence, an appropriate evaluation must be performed to guarantee that safety requirements resulting from an Automotive Safety Integrity Level (ASIL) according to the standard ISO 26262 are met.
In this paper, an evaluation of a dynamically reconfigurable system implemented on a centralized architecture is presented. The parameters evaluated are centered in reliability, probability of failure and possible trade-offs through the implementation of redundancy into reprogrammable devices and its performance parameters.
The method used is divided into three stages. The first stage is the system decomposition and isolation into sub-systems followed by a Fault Tree Analysis on the elements of each sub-system. The second stage is the system modeling through the mathematical stochastic model of Markov Chain and obtain the reliability function and quantification of the Mean Time to Failure (MTTF). Finally, the overall system is evaluated in terms of performance and according to timing constraints such as reconfiguration latency.
Applying this method we quantify the MTTF in Failure in Time (FIT) format of an E/E architecture. Additionally, we evaluate each sub-system independently and obtain the respective ASIL decomposition of the overall system. Furthermore, we can evaluate the migration of safety-related functionalities/redundancy from traditional architectures into reprogrammable devices.
With the application of this method using the Markov Chain, we can now evaluate Dynamical Reconfigurable Systems and define new E/E automotive architectures.
Florian Oszwald, Ruben Bertelo, Manuel Gericota, Juergen Becker
BMW Group, Polytechnic of Porto, Karlsruhe Institute Of Technology