Research On MDE Based Model Transformation

Nowadays model-driven technology is widely used in software development and a large number of visual modeling languages, represented by UML, keep emerging. But there are no strict formal semantics among most of them and there are also many other potential inconsistent problems existing in the established models, which greatly affect the reliability of the systems. Therefore, how to provide effective verification and simulation methods for non-(semi-)formal models in model-driven engineering has been attracting much attention in fields of both academic and industry.In this paper, we conduct in-depth research in UML-MARTE and SysML, which are widely used visual modeling languages and standards in real-time and embedded fields currently. We transform these two types of models into formal system and simu-lation platform by MDE-based model transformation technology. Then, we can verify the models and provide simulation support taking the advantage of formal verification tools and simulation platform. At last, we achieve the goal of checking and verifying the models of system at an early stage of system design. Specifically, our works focus on the following aspects:This paper firstly studies two ways of transformation from SysML state machine diagram to Timed Automata and transformation from SysML sequence diagram to Timed Automata all based on MDE. For the result of transformation, we indirectly take advantage of the current verification tools for Timed Automata to solve problem of verification of SysML state machine diagram and sequence diagram.Secondly, this paper studies the technology of generation from model to simula-tion code based on MDE. In this part, we study the semetic mappings between MARTE and SystemC and then construct the model transformation rules based on ATL and QVT-R, which are the mainstream in model transformation languages. At last, we implement the whole model transformation rules based on different execution engins, mapping MARTE models into simulation platform, getting dynamic simulation and verification supports for system models from simulation platform.Finally, we study two commonly used model transformation languages:ATL and QVT-R and make a comparison between them based on the implementation of transfor-mation. We show the advantages and disadvantages of two transformation languages in different scenarios. Also we propose a graphical presentation of declarative trans-formation rules in order to improve the model transformation rules comprehensibility.