A Petri net-based approach to modeling and analysis of component-based distributed real-time systems

Research, development, and applications of distributed real-time control systems have grown immensely in recent years. However, increasing complexity, rising costs, and demands for high reliability have made it necessary to adopt component concepts in traditional distributed realtime systems, an approach which can also drastically improve system reusability. In this paper, we propose a formal approach which combines Generalized Stochastic Petri Nets (GSPN) and G-nets (a Petri Net extended with object-oriented concepts) to model and evaluate the performance of component-based distributed real-time systems. The components and their modules are standardized and modeled with G-nets, whereas the whole system is modeled with GSPN in order to describe its stochastic and concurrent characteristics. We apply the approach to a new Distributed Railway Interlocking System (DRIS), which is a real-time, safety-critical control system.