A rainbow net simulator with a dependability application

The authors present an efficient rainbow net (RN) implementation that capitalizes on RN's multiple modeling paradigms, thus yielding a natural and convenient way of expressing system behavior models. A powerful extension to Petri nets, RNs are a modeling technique that combines graphical and state variable based specification for discrete event simulation. RNs were designed especially for dependability modeling. Previous modeling techniques suffer from an explosion in the number of states required to model realistically-sized systems. The efficacy of the RN simulator is demonstrated by a dependability analysis of an n-modular redundant system. This model is shown to be constant in specification complexity, linear in simulation space, and polynomial in simulation time.<
>