05885898

a

1226.60128 van Dijk, Nico M. Error bounds and comparison results: the Markov reward approach for queueing networks. Boucherie, Richard J. (ed.) et al., Queueing networks. A fundamental approach. New York, NY: Springer (ISBN 978-1-4419-6471-7/hbk; 978-1-4419-6472-4/ebook). International Series in Operations Research \& Management Science 154, 397-459 (2011). 2011 New York, NY: Springer EN 60K25 90B22 60E15

doi:10.1007/978-1-4419-6472-4_9

The chapter presents an approach to the comparison of two queueing networks. One of the networks is of practical interest, but unsolvable. The other network is a modification of the first one that is solvable. The approach is essentially based upon evaluating steady state performance measures by a cumulative reward structure; it is referred to as Markov reward approach. The approach leads, among other things, to analytic error bounds for the discrepancy between the actual and the modified networks. In order to illustrate the Markov reward approach, it is applied to a breakdown example, to a finite tandem queue and to a finite Jackson network. Moshe Shaked (Tucson)