On Generating Self-Similar Network Traffic Using Multi-core Processors

Self-similarity is the one of important characteristics of real network traffic. In order to simulate the network traffic accurately, we present a technology of self-similar network traffic generation, and describe it from algorithm and hardware.On algorithm we analyze several typical self-similar traffic models, describe the superiority of multifractal wavelet in detail, and make it our generation model. With this model, we present a generation algorithm of self-similar network traffic. In this algorithm, Hurst (0.5, 1.0) is the key parameter to express the different attribute of self-similar network traffic.On hardware we contrast the performance and flexibility of several programming multi-core processors, choose the Intel IXP2400 to design the platform of self-similar network traffic generation, and implement our algorithm on it. The results show that the algorithm is flexible and accurate.