Discrete-state and fluid stochastic petri net models for open-loop video-on-demand systems:a comparative case study

Abstract

We compare discrete-state stochastic Petri nets (SPNs) and fluid stochastic Petri nets (FSPNs) with a mixed discrete-continuous state space in their application to openloop video broadcasting systems. For both classes, we present models of the Video-on-Demand client-server system, which generally capture all relevant system aspects in a concise and graphical way. Differences in model design as well as solution techniques for various variants of these models are discussed. By simulation, we obtain measures related to blocked video playout, which allow us to evaluate customer satisfaction against bandwidth requirements. While the FSPN model has advantages in model specification and evaluation efficiency, all presented models excel in their flexibility – both in defining various performance measures and in their extensibility to include, for instance, full VCR functionality and arbitrary user behavior.