Understanding IP Traffic Via Cluster Processes

In this paper we investigate the characteristics of network traffic via the cluster point process framework. It is found that the exact distributional properties of the arrival process within a flow is not very relevant at large time scales or low frequencies. We also show that heavy-tailed flow duration does not automatically imply long-range dependence at the IP layer. Rather, the number of packets per flow has to be heavy-tailed with infinite variance to give rise to long-range dependent IP traffic. Even then, long-range dependence is not guaranteed if the interarrival times within a flow are much smaller than the interarrival times of flows. In this scenario, the resulting traffic behaves like a short-range dependent heavy-tailed process. We also found that long-range dependent interflow times do not contribute to the spectrum of IP traffic at low frequencies.

[1]  W. J. Thron,et al.  Encyclopedia of Mathematics and its Applications. , 1982 .

[2]  Walter Willinger,et al.  Self-similarity and heavy tails: structural modeling of network traffic , 1998 .

[3]  Walter Willinger,et al.  Self-Similar Network Traffic and Performance Evaluation , 2000 .

[4]  R. Adler,et al.  A practical guide to heavy tails: statistical techniques and applications , 1998 .

[5]  Patrice Abry,et al.  Wavelet Analysis of Long-Range-Dependent Traffic , 1998, IEEE Trans. Inf. Theory.

[6]  M. Bartlett.  The Spectral Analysis of Point Processes , 1963 .

[7]  Patrice Abry,et al.  Cluster processes: a natural language for network traffic , 2003, IEEE Trans. Signal Process..

[8]  Richard E. Barlow,et al.  Statistical Theory of Reliability and Life Testing: Probability Models , 1976 .

[9]  Paul Embrechts,et al.  A limit theorem for the tails of discrete infinitely divisible laws with applications to fluctuation theory , 1982, Journal of the Australian Mathematical Society. Series A. Pure Mathematics and Statistics.

[10]  P. Lewis.  A Branching Poisson Process Model for the Analysis of Computer Failure Patterns , 1964 .

[11]  David R. Cox,et al.  The statistical analysis of series of events , 1966 .

[12]  C. Klüppelberg,et al.  Tail behaviour of the busy period of a GI/GI/1 queue with subexponential service times , 2004 .

[13]  Jan Beran,et al.  Statistics for long-memory processes , 1994 .

[14]  S. Resnick,et al.  Is network traffic approximated by stable Levy motion or fractional Brownian motion , 2002 .

[15]  Abraham O. Fapojuwo,et al.  Modeling Wireless TCP Connection Arrival Process , 2006 .

[16]  B.K. Ryu,et al.  Point process approaches to the modeling and analysis of self-similar traffic .I. Model construction , 1996, Proceedings of IEEE INFOCOM '96. Conference on Computer Communications.

[17]  Gordon E. Willmot,et al.  LUNDBERG BOUNDS ON THE TAILS OF COMPOUND DISTRIBUTIONS , 1994 .

[18]  Abraham O. Fapojuwo,et al.  Stochastic processes for computer network traffic modeling , 2005, Comput. Commun..

[19]  C. Klüppelberg,et al.  Subexponential distributions , 1998 .

[20]  D. J. Daley,et al.  The Hurst Index of Long-Range Dependent Renewal Processes , 1999 .

[21]  San-qi Li,et al.  Queue response to input correlation functions: continuous spectral analysis , 1993, TNET.