Hungarian-German researchers exchange program 2000-2001 (DAAD-MÖB)
Participants:
Friedrich-Alexander-University Erlangen-Nuremberg, Inst. of Computer Science (Computer Structures)Objective:
(Dr. M. Dal Cin, D. Kreische, C. Kosmidis)
Budapest University of Technology and Economics, Dept. of Measurement and Information Systems
(dr. Pataricza A., dr. Majzik I., dr. Csertán Gy., Jávorszky J. K., Szász Cs. D., Huszerl G.)
The aim of the project is harmonising the design process of embeddedindustrial control systems and the modern software engineering methods, aswell as promoting the reliability of these systems by automatic analysis of the effects of potentially arising faults. Our goal is the extension of the current knowledge in the field of fault modelling, together with contribution to the qualitative and quantitative reliability analysis.Further information:The present project focuses on the UML (Unified Modelling Language), a new technology of software design. The main goal is the qualitative and quantitative estimation of the performance and reliability of so-called embedded industrial control systems, since these systems raise - as compared to pure information processing systems - new and theoretically unsolved questions.
The diversity of components being used in these systems exceeds the diversity of the components used in common IT systems significantly. Accordingly, the typical models of these systems have strongly heterogen structures, whatalso makes the numerical evaluation structurally complex. The complexityproblem gets more dominant, when considering the fault models necessary tothe computation of reliability attributes.
Modelling of the effects of faults of components of a typical embedded system (various sensors and actors) is executable realistically only if the individual fault classes are strongly distinguished. The large number of potential faults and fault classes enlarges the complexity of the model to work with by at least an order of magnitude. From the user’s point of view the timing behaviourof the system is of great importance, which has to be analysed additionally assuming faults. Accordingly, true mapping of timing relations within the models is of emphasised importance.
The project aims at developing methods for building peculiar fault models of industrial processes in UML so that an expert of the target technology can model these systems without special knowledge in informatics, and sothat the models describing the nominal performance of the system in faultfree cases and the degraded performance in case of any supposed faults canbe managed efficiently, and the effects of potential faults can be foretoldas accurately as possible.
In the first year we work on the issues of modelling embedded industrialcontrol systems, and in the second year we implement and evaluate the transformations for qualitative and quantitative analysis.
András Pataricza , Ph.D.