A simplified framework for stochastic workflow networks

This paper presents a novel method to simplify stochastic workflow networks for their performance analysis under a unified computable framework. This method is based on two techniques: (1) module simplification, and (2) PH equivalence and PH approximation. In the first technique, simplified procedures for at least four crucial modules: sequential routing, parallel routing, selective routing and iterative routing are given, respectively; while in the second technique, the closure properties and the two-order approximation for the PH distributions are discussed. Using this method, we analyze several examples for the stochastic workflow networks and illustrate that performance evaluation of complicated stochastic workflow networks can be obtained by means of subsystems which are clearly constructed by some of the four structured modules. Numerical examples indicate that the method of this paper can tackle large-scale and complicated stochastic workflow networks with both effective approximation and low computational complexity.