Variance Reduction in Simulation of Multiclass Processing Networks

TL;DR

This paper introduces two novel control variate-based estimators that significantly reduce variance in steady-state performance simulation of multiclass queueing networks, especially under heavy load conditions, with minimal extra computation.

Contribution

The paper presents two new variance reduction estimators for simulation of multiclass queueing networks, improving accuracy in both moderate and heavy traffic scenarios.

Findings

Substantial variance reduction in moderately-loaded networks

Significant variance reduction in heavily-loaded networks

Minimal additional computational cost for both estimators

Abstract

We use simulation to estimate the steady-state performance of a stable multiclass queueing network. Standard estimators have been seen to perform poorly when the network is heavily loaded. We introduce two new simulation estimators. The first provides substantial variance reductions in moderately-loaded networks at very little additional computational cost. The second estimator provides substantial variance reductions in heavy traffic, again for a small additional computational cost. Both methods employ the variance reduction method of control variates, and differ in terms of how the control variates are constructed.