Topological network design of state-dependent queueing networks

Topological network design problems often contain nodes and arcs where the stochastic flows of customers, occupants, and messages are state-dependent. A Gilbert–Steiner problem is formulated using a convex cost function to design networks where state-dependent flows occur. The properties of the problem along with an algorithm, complexity analysis, and experimental results for two example applications – one an open queueing network model and the other a closed queueing network model – are presented.