Adaptive Test Feedback LoopA Modelling Approach for Checking Side Effects During Test Executionfor Advised Explorative Testing 2020-01-0017
The main objective of testing is to evaluate the functionality, reliability, and operational safety of
products. However, this objective makes testing a complex and expensive stage in the development
process. This is particularly true for complex and large systems, such as trains or aircrafts, which require
maximum operational safety. From the perspective of an aircraft manufacturer, the checks are carried
out via test cases on the integration, system and application levels. Thus, they certify the products
against the requirements using black box testing approach. In doing so, a test plan defines a sequence
of test cases whereby it sets up the environment, stimulates the fault, and then observes the system
under test for each case. Subsequently, the post processing of the test execution classifies the test
plan in passed or failed.
The ongoing digitization and interconnectedness between aircraft systems is leading to a high number
of test cases and a multitude of reasons why a specific test-case fails. A corresponding error analysis
and adaptation of the test plan is a complex and lengthy process, which starts repeatedly after the
execution of a manual updated test plan. The goal of the developed approach is to reduce the amount
and the duration of the time-consuming try and error approach. This paper proposes the advised “free
testing” as an extension of existing test processes to meet the requirements.
An essential aspect of adaptive free testing is the adaptive test feedback loop. It enables the recursively
adaption of a test under execution as soon as a faulty behaviour is detected. The adaption is based on
proposals which are calculated using a knowledge base and selected by an test engineer.
A new tool landscape enabling this adaptive test feedback loop is proposed in this paper. The tool
landscape consists of a test automation suite, an autopilot and a test advisor. Their required
functionality and corresponding deployment process for a test bed is described. In addition, this paper
evaluates the benefits of an adaptive test feedback loop based on an avionics test scenario.
Marco Franke, Sergej Krause, K. D. Thoben, Andreas Himmler, K. A. Hribernik
Bremer Inst. Für Produktion Und Logistik, dSPACE GmbH, University of Bremen