Abstract
Measurements of network traffic have shown that self-similarity is a ubiquitous phenomenon spanning across diverse network environments. Previously, the researchers designed network traffic simulation system with sequential methods to generate a small number of network traffics in a short time scale. But there is one difficulty in simulating network traffics using parallel methods with high accuracy in a large time scale. In this paper, we present a parallel method of generating self-similar network traffic based on parallel algorithm and implement it via TCP/IP. We also carry out some experiments to show that this parallel method can generate highly accurate network simulation traffics with specified Hurst parameters.