Online Perturbation Analysis and Controloftandem 2-Class Buffers Using Stochastic Fluid Models

Abstract

In this paper, we apply infinitesimal perturbation analysis (IPA) to packet loss and buffer workload-related performance metrics in tandem networks of two-class stochastic fluid models (SFM). We consider these performance metrics at a downstream SFM node as a function of the controlled traffic stream’s threshold parameter at an upstream SFM node. The derived IPA gradient estimators are unbiased and nonparametric in the sense that they are computable directly from online measurements of real-life traffic processes as well as offline simulations, without any knowledge of underlying stochastic characteristic of the traffic and service processes. The efficiency of the derived IPA gradient estimators has been demonstrated in tandem buffers control problem by simulation.