On the compilation of statecharts models into target code for embedded systems

Synchronous languages have proven to be very adequate for the modeling of embedded systems. Following the semantics of a synchronous language, a model starts execution from an initial state, receives signals from the environment and iterates until it reaches a stable state. Synchronous models thus react to a stimulus with a chain of actions, where the number of iteration steps in the chain reaction is not explicitly defined. This paper focuses on the language statecharts, and describes how knowledge about the worst-case number of iteration steps (WCNIS) that synchronous models execute to reach a stable state can be used to avoid time consuming run-time detection of stability and significantly improve the quality of programs generated from such models in general. Moreover, the paper describes a method for determining the WCNIS, a given statecharts model executes to reach a stable state. The method described has been integrated in a code generator for STATEMATE models, and experimental results are presented.