Aktuelle Ansätze für den Steuergerätetest

Current Approaches for ECU Testing
What really matters

Download PDFFurther development of methods and tools for comprehensive and structured testing of ECUs is necessary not only for economic reasons. New highly complex technologies such as ADAS and autonomous driving also call for suitable testing strategies. This article presents current development activities and approaches as well as appropriate strategies by Vector.

A large portion of the value created in a modern vehicle is associated with the electronic and software systems. Malfunctions of electronics and software are accordingly highrisk and are not tolerable especially when used in safetycritical systems. That is why significant investments have been made in automated testing of these systems for many years. With each new technology, the question arises about the correct test methodology. Important factors here are the effectiveness, i.e., whether and how the test objectives are achieved, and the efficiency, i.e., optimization of costs and effort.

Model-based test design

In model-based testing, test sequences and data are generated from models. These models can either be a model of the ECU functionality or the test cases themselves. The first approach, thus the derivation of test cases from function models, should be viewed somewhat critically. If both the productive software and the corresponding test cases are generated from the same model, that promises large efficiency gains – the test practically drops out during development free of charge. But one should be aware of what is actually tested in such a constellation – in the extreme case, the code generator is merely being tested against the test case generator. Moreover, a function model largely contains the “good case” and thus describes how a device is supposed to function. The corresponding operating situations for the test of these functions can still be effectively derived from this model. However, it is difficult if not impossible to derive the many different error scenarios from the function model. But it is exactly these error scenarios that make the test so valuable since most problems don’t arise in the "straightforward case".

vTESTstudio provides various graphical test notations for
test modelingFor this reason, the Vector test solution works with a test model. This means that the test designer systematically models the situations in which the System Under Test must exhibit correct functioning. A graphical test notation enables a clear and transparent representation of the abstract test sequence (Figure 1). The test cases are generated from this high-level description. Variants are visible in the diagram and are taken into account during test case generation. The test models connect the specification and implementation of the test in an abstract and easy-to-understand way. This will also facilitate the frequently neglected quality assurance measures for the tests themselves, such as reviews by developers. Abstraction enables use of the same test specifications on different test systems and the reuse of previously specified tests – both leads to significant savings.

Page 1/3
Did you find this page helpful?

Please comment: