Stochastic statecharts and rapid prototype software architecture

Stochastic statecharts are introduced as an attractive means for precisely representing the reactive behavior of complex nondeterministic systems. Such representation centers on absolute real time, concurrency, and stochastic effects. The characterization of stochastic effects entails modest yet significant higher-level statechart notational changes to both the statechart graphical conventions and the transition rule syntax extensions. Appreciable latitude is afforded in expressing differing probability distributions and parameters, for the primary emphasis is on modeling power rather than on analytical tractability. An example of an on-line transaction processing system is used to illustrate the inclusion of stochastic effects. These are in the form of absolute timing variability and probabilistic process decisions. The richness and complexity of such models, moreover, suggest the need for prototyping as a practical necessity in their assessment. Although the modularity of statecharts yields naturally partitioned specifications, their mapping to highly modular prototypes can be rather problematic for higher-level statecharts. This stems largely from the appreciable coupling among distinct threads of control, and in turn the complexity of prototype software interfaces. To redress such difficulties, a new software prototype structuring approach and architecture are described. They facilitate software prototype development as well as software component reuse