Research On Network Traffic Based On Mwm

With the development of technology, internet traffic has become complex and diversiform; it has become a very arduous task to research on the network behavior and analyzes complex flowing data. However, in order to improve and guarantee the service quality of existing networks, to construct further rigorous and safe network architecture, we need have a clearer and deeper understanding of the network traffic.It is quite not feasible in making an analysis of the characteristics of the network traffic without traffic model. Firstly, we introduced several basic characteristics of network such as LRD, SRD, self-similarity and multifractal, and also with some H estimation methods which are frequently used by researchers. And then the paper introduced several common traffic models: Traditional models and self-similarity ones. With the experiment exited, wavelet model was better than other ones in the check of self similarity. Based on above conclusions, this paper mainly analyzed multi-fractal model (MWM).Firstly, paper added another two kinds of MWM, which are point-mass MWM and hybrid MWM, to synthesize the network traffic. With the analysis and comparison between the real traffic and synthesized traffic, the conclusion is that symmetricβdistribution is not the only method of network traffic synthesis, that is, the hybrid MWM could obtain a better match in the relativity of target signal. Therefore, according to the different characteristics of signal, different distributions should be chosen to adjust the fine scales and the coarse scales after wavelet transform, which is to gain a closet synthesis traffic comparison with other models.The paper provided some new theories and experimental basis for the network traffic analysis and application in the future's work, also had practical significance for managing network effectively and supplying high quality of service for us.