Wavelet-based Network Traffic Characteristics

The traffic behaviors influence network modeling, service providing and traffic engineering. So it has been the focus of many researches.Long-range dependence(LRD) and self-similarity are the fundamental statistical characteristics of the network traffic. In this paper, we introduce some Hurst parameters estimate algorithms, such as Aggragated variance method, Residual variance method, R/S method, Periodgrabl method and Wavelet methode to evaluate the LRD characteristics of the traffic.In order to study the fractal characteristics of the traffic, an extensive analysis is performed to the MPEG-4 video traffic. The results show that: (1) The MPEG-4 video traces have LRD characteristics. The stronger the burstness, the stronger the LRD characteristics. (2) MPEG-4 video traces have monofractal and multifractal behaviois. (3) The vanishing moment has no effect on the performance of wavelet estimatealgorithms. (4) The q^th moment has no effect on the performance of multifractal estimator.