In this paper we use the reversed process to derive expressions for the steady state probability distribution of a class of product-form PEPA models. In doing so we exploit the Reversed Compound Agent Theorem (RCAT) to compute the rates within reversed components of a model. The class of model is, in essence, a generalised, closed, queueing network that might also be solved by mean value analysis, if full distributions are not needed, or approximated using a fluid flow approximation. A general formulation of RCAT is given and the process is illustrated with a running example, including several new variations that consider effects such as multiple servers, competing services and functional rates within actions.
pubs.doc.ic.ac.uk: built & maintained by Ashok Argent-Katwala.