New Results on the Numerical Stability of the Stochastic Fluid Flow Model Analysis

Abstract

The stochastic fluid flow model (SFF) is one of the leading models in performance evaluation for tele-and datacommunication systems, especially in fast packet-switching networks and ATM. However, the numerical analysis of the SFF is widely considered to be unstable. In this paper, some investigations and results are presented concerning the numerical stability of the SFF analysis also for large systems with finite buffer. We identify the main source of the numerical problems and give hints how to circumvent them. The usefulness of different solution methods are compared and the most robust methods for systems with large numbers of sources and large buffer sizes are identified.