Application-driven Internet traffic analysis - a simulation approach

Internet traffic analysis and prediction is important for network operators to prepare for future demand. The traditional approach of curve-fitting on historical data becomes less effective as technologies advance at a fast pace. To address the problem, a simulation framework is developed with flexibility and modularity in mind so that simulations can be quickly constructed for new applications and scenarios. We show how to use the framework to simulate Web traffic demand over the Internet. Our approach provides important new insights on traffic properties as the effects of different parameters on traffic demand are explored. We verify the simulation models and parameters using current Internet traffic measures and demonstrate the use of the models to predict future Web traffic demand.