Railway control systems are safety-critical systems as failures may endanger human life or at least the environment. Hence, software development for these systems has to follow strict guidelines which are captured in the relevant, norm CENELEC EN 50128. Inter alia, the standard demands verification at each level in the software development life cycle and validation of the whole system. One means for verification and validation is testing. The basic functionality of the software has to be tested both by verification and validation to ensure that the software ia working correctly. A special emphasis in validation is testing the behaviour of the software in safety-critical situations. In addition to testing, validation can be based on models and simulations of the software. Currently, railway control systems are developed as a generic piece of software which is tailored to specific applications by configurations. The reason for this is that verification and validation have to be performed only for the generic software which saves time and money in comparison to developing several small controllers. The disadvantage of this proceeding is that testing and maintenance become difficult due to the amount of possible configurations. Problems due to untested or insufficiently tested configurations or unintended side-effects cannot be excluded. In this thesis, a new approach for the development of railway control systems has been developed which is based on ideas of model-driven and language-driven development. The basis of this approach is a domain-specific language that has been exclusively designed for the development of railway control systems software. The language incorporates domain knowledge in its static find dynamic semantics as well as domain notation in its concrete syntax. In this way, modelling is simplified as only relevant details fur a particular system have to be explicitly modelled while domain-specific constraints rhat have to be fulfilled are automatically checked. Moreover, domain experts that are not software experts are able to design models without further help. Most importantly, the incorporated domain knowledge and the rigorous language specification allow for the automation of several verification and validation tasks First, each model can be automatically validated with respect to its con formal ice to the constraints of the domain. It is ensured that all further steps in the software development life cycle are based on a valid model. Second, a simulation is automatically generated for each model. The correct functionality can be checked at the model level before software has been developed. In addition, test data is automatically generated which can he used in the simulation as well as in softwmi; or hardware/software integration trating. The selection of the test data is based on the domain knowledge. Hence, one can be sure that the basic functionality of the software is tested as well as its correct behaviour in hazardous situations. The applicability of the test suite is ensured by mutant testing, i.e. the test suite is used to test mutatis! code in the simulation. The test suite is applicable as critical errors in the mutant code are detected. We believe that the approach that has been developed in this thesis contributes in several ways to an efficient development of railway control systems according to CENELEC EN 50J28. Models can be easily designed by software or domain experts as the newly developed language allows for concise and precise models. Domain-specific constraints are implicitly incorporated. Model validation is automated as well as simulation generation and test data generation. Simulation and mutant testing support the validation process and increase confidence in the system. Test data generation facilitates both verification and validation. All in all, the simplicity of model design and the automation of several important tasks for the development of railway control systems software decrease development time and hence costs. Moreover, the lack of figurations reduces the probability of unintended side-effects in the software and increases testability. In this way, the new approach is a serious alternative to the current development process.
Domain-specific modelling of railway control systems with integrated verification and validation
Bereich-spezifische Modellierung von Eisenbahn-Steuersystemen mit integrierter Verifizierung und Validierung
2010
156 Seiten, Bilder, Tabellen, Quellen
Theses
English
Automated validation and verification of railway specific components and systems
Tema Archive | 2008
|Towards a Tool-Based Domain Specific Approach for Railway Systems Modeling and Validation
British Library Conference Proceedings | 2019
|