Variance reduction for sensitivity estimates obtained from regenerative simulation

The authors apply the method of concomitant control variables to reduce the variance of derivative estimates obtained from regenerative simulations using the likelihood ratio method. They restrict the present attention to systems that contain a Poisson process. The control variables that the authors use arise naturally from likelihood ratios. After deriving an expression for the optimal coefficients of the control variables, they numerically investigate their performance. Using six controls with an M/G/1 queue the authors achieve a reduction in the standard deviation of the derivative estimate of up to 60%.