Measurement and Causes of Self-Similarity in Computer Networks Traffic

It is now well known that network traffic exhibits self-similarity, which cannot be described by traditional Markovian models such as the Poisson process. Efficient and effective techniques for traffic self-similarity measurement are not only desirable, but also increasingly becoming a necessity because it is important to gain knowledge about the characteristics of the traffic. Also the causes of self-similarity of network traffic must be identified because understanding the nature of network traffic is critical to properly design and implement computer networks. This thesis proposes adaptive multistage filter algorithm for self-similar traffic measurement, the proposed algorithm uses multistage filter and adaptation mechanism to concentrate on large flows. Using analysis and simulation, the proposed algorithm effectively reflects the self-similarity of network traffic and as well as reduces the sampling error compared to the Multistage Filter algorithm. The thesis also investigates the effects of TCP’s congestion control and CSMA/CD retransmission algorithm on the self-similarity of network traffic. The result indicates that the congestion control and CSMA/CD retransmission algorithm modulate the network traffic to be self-similar