Phase: A Stochastic Formalism For Phase-Type Distributions

Models of non-Markovian systems expressed using stochastic formalisms often employ phase-type distributions in order to approximate the duration of transitions. We introduce a stochastic process calculus named PHASE which operates with phase-type distributions, and provide a step-by-step description of how PHASE processes can be translated into models supported by the probabilistic model checker PRISM. We then illustrate our approach by analysing the behaviour of a simple system involving both non-Markovian and Markovian transitions.