Ordinal statistics for computation reduction in control of discrete event systems

A stochastic optimization problem for discrete event dynamic systems is addressed and solved by means of concurrent estimation and ordinal comparison. A performance function defined on the system must be optimized by choosing an appropriate resource allocation. An analytical expression of the objective function is not available, and concurrent estimation is used to estimate it. The approach proposed in this paper, which can be applied more easily than the methods presented previously, exploits the fast convergence rate of order statistics in order to reduce computation effort