MODELING OF MEASURED SELF-SIMILAR NETWORK TRAFFIC IN OPNET SIMULATION TOOL

The Modeling, analysis and simulation of self-similar traffic has become the main goal of much research work around the world, over the last 15 years. In our research we measured many different types of real traffic in different networks and classified it on the basis of analysis in the sense of self-similarity and long-range dependence. We used estimated statistical parameters for measured network traffic in order to model this traffic in simulation tool OPNET. We used the following statistical criteria for successful modeling: average bit rate, average packet rate, Hurst parameter, and histograms of statistical network traffic processes. During measurements and simulations we discovered that the shape parameter of Pareto distribution has a great impact on simulated traffic, and also that classical estimation usually leads to significant discrepancies between measured and simulated traffic in the sense of average bit rate and also bursts, which are characteristic of self-similar traffic. So, we developed a novel method for estimating the shape parameter of Pareto distribution which shows successful results regarding the chosen criteria, during the testing process.