Quantitative Analysis of System Based on Extended UML State Diagrams and Probabilistic Model Checking

Abstract

If probabilistic model checking is applied in software architecture, function validation and quantitative analysis for Markov process based real-time model can be automatically performed in model refinement, which will improve software quality. In this paper, the exact definitions and mapping rules between UML state diagrams and probabilistic Kripke structure semantics are proposed, as well as the general translation algorithm of formal semantics. An asynchronous parallel composited DTMC system is illustrated, the key non-function properties of system are described by PCTL, which is automatically validated and analyzed by PRISM model checker. The key system properties are also manually deduced and proved, and compared with the experiment results. The mapping rules we proposed are bi-direction, so it can be used in both forward and reverse software engineering.