Insights into peer to peer traffic through nonlinear analysis

The enormous growth in popularity of peer-to-peer applications has recently introduced great interest in understanding the associated traffic workload and behavior. The goal of this work is determining the fundamental dynamics characterizing such traffic that can be used to develop simple and effective prediction models and to illustrate and describe fundamental performance issues. The discovery of nonlinear traffic dynamics, due to the very complex characteristics of the involved time series, led us to use several nonlinear analysis techniques and tools evidencing the presence of chaos-related structures together with self-similarity and long-range dependence features.