Characteristics Analysis of Web Traffic with Hurst Index

Today’s computer networks are highly dynamic to change especially Internet. Simulating Internet behavior is therefore a difficult and cumbersome task. The web remains the largest application class on wide-area links. Generation of natural web traffic behavior is an essential component for simulation models. Failure of Poisson and Markov models has led to development of new models. In the past decade selfsimilarity has been proved to be an important feature in the network traffic. Hurst index is used to test for self-similarity of a process. Recently, PackMime has been introduced to NS-2 for HTTP traffic generation. In this paper statistical analysis method has been used to estimate the Hurst index. Simulations have been carried out using PackMimeHTTP. The results obtained are tested for self-similarity by finding the H index from the slope of the least-squares fitting line of the variance-time plot.