Distribution Characteristics Of Self-similar Network Traffic

With the development of network technology and the diversity of traffic, the flow characteristics of modern network have been completely different from the traditional telephone switching network. Research shows that the computer network traffic exhibits self-similarity. To evaluation and modeling network performance, we must take the self-similarity of network traffic into consideration. In recent years, with the rapid development of the smart grid, distribution network is also facing the development of intelligent. As an important support of the distribution network, communication network has various service models, and the load traffic is more and more large as well. Researching self-similarity of traffic in distribution communication network has significant guiding in the construction of the distribution network.Firstly, the characteristics and the causes of self-similarity are analyzed, and a variety of self-similarity traffic generation models are presented. Two methods of generate self-similar traffic in OPNET are introduced, then the self-similar stream and the traditional Poisson traffic’s impact on network performance are analyzed comparatively.Secondly, the overall framework of the distribution network is investigated, and the demand of the distribution network and the information classification of transmission are researched. The data of transmission in the distribution network are divided into periodic, random and burst information, each of which are also modeled and analyzed. OPNET is used for modeling and simulation. The results show that the distribution network system transmission traffic has self-similarity (H=0.657>0.5); meanwhile the influence of self-similarity parameters on the distribution communication network performance is simulated and analyzed.Finally, the distribution network traffic delay bound model based on fractal shaper is proposed, and the traffic bandwidth of the distribution communication network is predicted. Using network calculus deduced the delay bound and the relation between self-similar parameter H and weight W, which is based on the fractal shaper. Using the traditional short-related model to predict bandwidth will result in a waste of bandwidth. To solve the problem, Norros network traffic prediction model is proposed for the distribution network self-similar traffic bandwidth prediction.