Network Traffic Research Based On Wavelet

The original paper on the self-similar nature of network traffic appeared 10 years ago at SIGCOMM 1993. Since then, research on self-similar traffic has generally thrived, but has also seen its fair share of wrong turns, road blocks, dead ends, and specious claims. In regard to recent such claims such as self-similar,multifractal,long range dependent, the multiscale nature of the network traffic is presented. The wavelet analysis is introduced into network traffic for the nework design, nework dimension, network performance evaluation. With the scale nature of wavelet, it is easy to analyze the fractal networks traffic in wavelet domain with accomplishment in parameter estimation, traffic analysis and model, queue analysis and forecasting. The main contributions of the paper are:1. Present an adaptive, efficient unbiased estimation method of Hurst index based on wavelet. Hurst index is the key value of traffic model representing the burstiness of traffic. We analysis them using discrete wavelet transform, and describe the nature of the wavelet coefficients and their statistical properties. Then present an adaptive, efficient unbiased estimation method of Hurst index based on multiresolution wavelet analysis and weighted regression. Simulation results based on fractal Gaussian noise and real traffic data reveal the proposed approach shows more adaptiveness, accuracy and robustness than traditional methods which has only O(N) computation. Thus this adaptive method can be applied to the application of traffic enforcement and congestion control in high-speed networks.2 Present the multifractal nature of the MPEG-4 encoded video traffic. MPEG-4 encoded video traffic is expected to account for a large portion of the traffic in future wireline and wireless networks. We explore the viability of multifractal analysis in modeling the MPEG-4 video traffic. First, prove that the MPEG-4 video traffic exhibit the multifractal characteristic through multifractal formalism and spectra analysis. Then apply the multifractal wavelet model to the MPEG-4 video traffic analysis and synthesis. The statistical numeric results and fit procedure show this model's flexibility and accuracy.3 Present a connection-level traffic model based on wavelet. Recent research shows network traffic exhibits drastically different statistics according to scales. Structure network researchers show there are two components of the traffic according to the speed of connections, one component holds most traffic and being mostlyGaussian, the other absorbs virtually all the small scale bursts. For better understanding of this phenomenon, propose a traffic model based on wavelet. Simulation results with the real traffic show this model is flexible and parsimonious to accommodate Gaussian as well as bursty behavior on different scales.4 Present the multiscale network traffic queuing analysis and forecasting based on wavelet. Research on the queue analysis with multifractal traffic play an important role in network dimention and control. Based on multifractal wavelet analysis, we compare the efficient queue tail analysis with multifractal input to the analysis results. With the scale nature of wavelet, it is more convenient and accurate to perform the tail asymptotics analysis in wavelet domain. These results validated by queueing simulation of measured network traffic. The capacity planning theory for IP network requires accurate modelling of the incoming traffic, as well as accurate predictions of its future behaviour. With the ARIMA linear prediction in the wavelet domain, present a novel network traffic prediction method based on the wavelet model, which shows to be the accurate model of the multiscale network traffic. The simulation results with the real traffic traces show the accuracy and efficiency of the method.