Virtual Test Method for Complex and Variant-Rich Automotive Systems

Abstract

The fast development of embedded automotive systems in form of connected Electronic Control Units (ECUs) has led to complex development processes. Especially for safetycritical functions, the testing activities are essential to check if the designed system complies with the requirements. Nowadays, the continuous development of mobile electronic devices through software updates is performed almost on a daily basis. This trend is now starting to be observed in cyber-physical systems with higher safety priorities. In the automotive field, the rising software portion in the vehicles and the shortening technology life-cycles are accentuating the need for Software Over The Air (SOTA) updates. Despite the opportunities offered by SOTA updates, the current test processes and methods must be adapted to manage the resulting complexity throughout the life-cycle of the vehicles. Especially the typical variants abundance in automotive product lines is considered as an important challenge, which cannot be solved only by ”classical” testing methods such as Hardware-In-the-Loop. In this paper, we present a testing method for variantrich systems, which can be applied for automotive software updates. It uses virtual platforms for automated delta testing to handle the abundance of system configurations. Virtual testing is introduced as a powerful tool to reduce the amount of real tests and allow efficient variants verification. As a proof of concept, an Adaptive Cruise Control (ACC) composed of two ECUs has been implemented both in real hardware and using a virtual platform. With this approach, virtual delta tests, i. e. specific test-benches targeting the differences to a basic variant, can be rapidly executed for various system configurations. To prove the feasibility of the presented test method in more complex systems, a scalability study has been conducted.