The Pareto or Truncated Pareto Distribution? Measurement-Based Modeling of Session Traffic for Wi-Fi Wireless Internet Access

We have analyzed measurement traffic statistics for high-speed wireless Internet access sessions collected in a public nationwide Wi-Fi network. By ranking session lengths and traffic volumes we have found a law implying the truncated Pareto distribution. This finding is analogous to the relationship between Zipf's law and the Pareto distribution. By applying this new law we have developed a model for key session traffic variables. It has been successfully validated by linear regression. Some empirical session traffic statistics have been given for reference.