An approximation method for sojourn time distributions in general queueing networks

Even though sojourn time distributions are essential information in analyzing queueing networks, there are few methods to compute them accurately in non-product form queueing networks. In this study, the paper models the location process of a typical customer as a GMPH semi-Markov chain and develops computationally useful formula for the transition function and the first-passage time distribution in the GMPH semi-Markov chain. It uses the formula to develop an efficient method for approximating sojourn time distributions in the non-product form queueing networks under a quite general situation. The paper demonstrates its validity through numerical examples. [In Korean.]