ABSTRACT
This chapter explores the prevalence of heavy-tailed probability distributions in the Internet traffic. These distributions are present in: files requested by the users, files transmitted through the network, transmission durations of files, and files stored on the servers. The chapter provides a framework for the study of self-similarity in Internet traffic. The Internet traffic is characterized by long-range dependence and slow-decaying variances. Network traffic can be considered to be a time series representing bytes or packets per unit time which flow on a communication channel. This traffic was traditionally carried on circuit-switched networks. The telephony traffic is described by Poissonian arrival streams, and exponentially distributed hold times. Traffic burstiness occurs due to both the self-similar sources and the self-similar behavior of the users. Sources of traffic in the Internet alternate between “on” and “off” states. In addition, the Internet traffic appears to be bursty across an extremely wide range of time scales.
