Performance Analysis of System Model Based on UML State Diagrams and Continuous-time Markov Chains

Abstract

If software architecture is assigned with formal semantics, then automatic verification and validation can be performed during the process of model refinement. In this paper, we emphasized on the formal semantics of UML state diagrams oriented performance analysis. The exact definitions of the basic elements and composition mechanism of UML state diagrams are proposed, UML state diagrams is abstracted as a multi-tuple, CTMC models are abstracted as stochastic Kripke structure, mapping rules between the above two mathematics models are proposed, furthermore the corresponding formal semantics are generated. Finally, an asynchronous parallel composition queuing network is presented to illustrate how the theory is applied to formalize UML state diagrams. The key properties of system are manually deduced and validated. The results are analyzed and compared with the automatic executing results through model checker, which validated the practicability and validity of the theory.