Empirical Performance of Weibull Self-Similar Tele-traffic Model

J. Popoola, R. A. Ipinyomi

The stringent memoryless assumption in general traffic modeling is often violated due to concurrent or batch arrivals. Batch arrivals often leads to self-similarity or long range dependency problems in Internet traffic. Modeling tele-traffic data in this condition requires the use of heavy tailed distribution. In this paper, we propose a general form of Weibull tele-traffic model for self-similar Internet traffic data. The model empirical performance was observed via Monte-Carlo simulation with the aid of discrete event simulation. Performance analysis for the proposed model alongside standard Poisson/ Exponential model was also achieved. The results from the analysis established the strength of Weibull model over the existing model.