How far are we from steady state? On-line error bounds for steady state approximations

By studying performance measures via reward structures, on-line error bounds are obtained by successive approximation. These bounds enable one to determine when to terminate computation with a guaranteed accuracy. Furthermore, they provide insight into steady-state convergence in practical situations. the method is tested for a number of simple capacitated queuing networks. The results obtained indicate that the method provides a practical tool for numerically approximating performance measures of queuing networks