Evolutionary Algorithms to Generate Test Cases for Safety and IT-Security in Automotive Systems

The testing process of electronic control units (ECU) is time-consuming and cost-intensive. Virtual electronic control units (vECU) can solve these problems and also offer the advantage of a variety of observation points that are not available with classical ECUs. The additional observation points can be memory access or instruction monitoring. This paper shows a novel possibility to use evolutionary algorithms, which employ the observation points, to generate new test cases for safety and IT-security in automotive systems. The novel test cases and their ability to detect error classes are described in detail. It is shown, that evolutionary test methods are able to detect different error classes. However the evolutionary methods cannot detect all given error classes within one method. Hence, a combination of different tests methods is needed.

Paper

Full text

PDF

Evolutionary Algorithms to Generate Test Cases for Safety and IT-Security in Automotive Systems

Semantic Scholar · Engineering · 2020

Abstract

The testing process of electronic control units (ECU) is time-consuming and cost-intensive. Virtual electronic control units (vECU) can solve these problems and also offer the advantage of a variety of observation points that are not available with classical ECUs. The additional observation points can be memory access or instruction monitoring. This paper shows a novel possibility to use evolutionary algorithms, which employ the observation points, to generate new test cases for safety and IT-security in automotive systems. The novel test cases and their ability to detect error classes are described in detail. It is shown, that evolutionary test methods are able to detect different error classes. However the evolutionary methods cannot detect all given error classes within one method. Hence, a combination of different tests methods is needed.

Similar papers

© 2026 NYSGPT2525 LLC