The deployment of Infinitesimal Perturbation Analysis

This paper considers congestion-related performance metrics in tandem networks of Stochastic Fluid Models (SFMs), and derives their IPA gradient estimators with respect to buffer sizes. Specifically, the performance met-rics in question are the total loss volume and the cu-mulative buffer workload (buffer contents), and the con-trol parameter consists of buffer limits at both the node where the performance is measured and at an upstream node. The IPA estimators are unbiased and nonparamet-ric, and hence can be computed on-line from field mea-surements as well as off-line from simulation experiments. The IPA derivatives are applied to packet-based networks, where simulation results support the theoretical develop-ments. Possible applications to congestion management in telecommunications networks are discussed. Key words. IPA, stochastic flow networks, telecom-munications networks, congestion management