Analysis of multiple-server polling systems by means of the power-series algorithm

Abstract

We consider a polling model with multiple servers, each of which visits the queues according to its own service order table. In general, such a model is not tractable by means of analytical techniques. In this paper, we show how the model can be analyzed by the power-series algorithm (PSA), a tool for the numerical evaluation and optimization of the performance of a broad class of multi-queue models. Various numerical experiments with the PSA are performed, providing new insights into the behavior of multiple-server polling systems