Steady-State Performance Evaluation of Totally Open Systems of Markovian Sequential Processes

Totally open systems of Markovian sequential processes are deened as a subclass of stochastic Petri nets. They can be viewed as a generalization of a subclass of queueing networks in which complex sequential servers can be synchronized according to some particular schemes. Structural analysis of these nets is considered for avoiding the state explotion problem of the embedded Markov chain. Some qualitative properties interesting from a performance point of view are presented. In particular, a \potential ergodicity" property is characterized by means of two structural properties: consistency and synchronic distance relation. Necessary and suucient ergodicity conditions and the computation of steady-state performance measures are studied.