This paper presents a process algebra for specifying soft real-time constraints in a compositional way. For these soft constraints we take a stochastic point of view and allow arbitrary probability distributions to express delays of activities. The semantics of this process algebra is given in terms of stochastic automata, a variant of timed automata where clocks are initialised randomly and run backwards. To analyse quantitative properties, an algorithm is presented for the on-the-fly generation of a discrete-event simulation model from a process algebra specification. On the qualitative side, a symbolic technique for classical reachability analysis of stochastic automata is presented. As a result a unifying framework for the specification and analysis of quantitative and qualitative properties is obtained. We discuss an implementation of both analytic methods and specify and analyse a fault-tolerant multi-processor system.
Pedro R. D'Argenio, Joost-Pieter Katoen, Ed Brinks