Simulating Application Level Self-Similar Network Traffic Using Hybrid Heavy-Tailed Distributions

Many networking researches depend on an accurate simulation of network traffic. For example, Intrusion Detection Systems generally require tuning to be effective in each new environment. It follows that researchers need to produce traffic backgrounds for laboratory testing that accurately reflect the characteristics of organizations of interest. Because self-similarity is a common feature in today\u27s network traffic, simulations which can produce the same degree of self-similarity as the original traffic are desired. The authors discovered that modeling some important protocol characteristics has required the use of hybrid modeling and heavy-tailed distributions. These include protocols like HTTP that account for a large percentage of traffic today although they were not present for studies done a few years ago. In this paper hybrid and heavy-tailed modeling techniques are used to build detailed models of major Internet protocols. NS-2 is used to simulate the Internet traffic captured at University of Central Florida and the result is compared against original traffic. Copyright 2005 ACM