MODELING SYSTEMS BY PROBABILISTIC PROCESS ALGEBRA - AN EVENT STRUCTURES APPROACH

Performance and reliability analysis of distributed systems based on formal specifications is an important and widely recognized issue. This paper treats a probabilistic version of (a subset of) the process algebra LOTOS. It incorporates a probabilistic choice assigning a probability of occurrence to each of its alternatives. Opposed to the traditional interleaving semantics used for existing probabilistic process algebras the presented language is based on a true concurrency semantics. This enables us to distinguish between nondeterminism and parallelism, to reduce the state explosion problem and, moreover, to analyse part of the system without considering other (irrelevant) parts. In this paper the language is presented and the formal semantics is defined by using an extension of bundle event structures. A short example illustrates the novelties of the language and links the language to stochastic analysis based on semi-Markov chains.