Abstract

This research addresses the issues of safety-critical systems verification and validation. Safety-critical systems such as avionics systems are complex embedded systems. They are composed of several hardware and software components whose integration requires verification and testing in compliance with the Radio Technical Commission for Aeronautics standards and their supplements (RTCA DO-178C). Avionics software requires certification before its deployment into an aircraft system, and testing is mandatory for certification. Until now, the avionics industry has relied on expensive manual testing. The industry is searching for better (quicker and less costly) solutions.
This research investigates formal verification and automatic test case generation approaches to enhance the quality of avionics software systems, ensure their conformity to the standard, and to provide artifacts that support their certification.
The contributions of this thesis are in model-based automatic test case generations approaches that satisfy MC/DC criterion, and bidirectional requirement traceability between low-level requirements (LLRs) and test cases.
In the first contribution, we integrate model-based verification of properties and automatic test case generation in a single framework. The system is modeled as an extended finite state machine model (EFSM) that supports both the verification of properties and automatic test case generation. The EFSM models the control and dataflow aspects of the system. For verification, we model the system and some properties and ensure that properties are correctly propagated to the implementation via mandatory testing. For testing, we extended an existing test case generation approach with MC/DC criterion to satisfy RTCA DO-178C requirements. Both local test cases for each component and global test cases for their integration are generated. The second contribution is a model checking-based approach for automatic test case generation. In the third contribution, we developed an EFSM-based approach that uses constraints solving to handle test case feasibility and addresses bidirectional requirements traceability between LLRs and test cases. Traceability elements are determined at a low-level of granularity, and then identified, linked to their source artifact, created, stored, and retrieved for several purposes. Requirements’ traceability has been extensively studied but not at the proposed low-level of granularity.