Simulation Based Analysis of Queueing Property for Continuous Self-Similar Traffic

The research for self-similar traffic is classified into four categories; measurement-based traffic modeling, physical modeling, queueing analysis and traffic control. These works for queueing analysis are important in the establishment of basic performance boundaries by investigating queueing behavior with long-range dependent input. The purpose of this research is to explore queueing properties for self-similar traffic using the network simulator with different conditions, and clarify the causes of different features focusing on the bottleneck link. We generated two typical traffics with Pareto distribution on ns-2 simulator and observed queue size, then following properties were extracted. When the On/Off traffic transmissions obeys the Pareto distribution, self-similar traffic consumes more queue resources than short-range dependent traffic. On the other hand, when the transmitted file size obeys the Pareto distribution, self-similar traffic consumes less queue resources than short-range dependent traffic.