Binary Information Press
The current researches have confirmed that network traffic signals have such multidimensional properties as self-similarity, emergency and long-range dependence. Spurred by the need of network monitoring and security, network traffic modeling is a much discussed research topic. However, the classical Markov ow model is no longer suitable for packet-switched network because of such network properties as self-similarity and long-range dependence. This paper presents a network traffic model based on chaotic time series which can predict the network traffic with smaller error than other models through experiments.