Optimizing discrete event systems with the simultaneous perturbation stochastic approximation algorithm

Stochastic approximation is one method that has been applied to the optimization of discrete-event systems requiring simulation. We investigate the use of simultaneous perturbation stochastic approximation (SPSA). This technique requires only two simulations per gradient estimate, regardless of the number of parameters of interest. We apply the technique to an open queueing network optimization problem.<
>