Research And Application Of Traffic Sampling Methodology Based On Self-Similarity

IP networks,Internet as one of their representations,are becoming basic communication facilities of our society. As the scale of IP network expands continuously,network bandwidth increases in double and its structure becomes more and more inhomogeneous and complex,it is necessary for topology structure and network behavior to be understood and analyzed for the Internet correct and effective measurement is very important for Internet,getting available and reliable data evaluate network capability,take some pertinent measure. At the same time, network traffic measure is the key link in the traffic engineering and network behaviors. With the development of Gigabit Ethernet and other high-speed network technology, it becomes very hard to collect all the packets, And the analysis of network behaviors becomes very hard due to the large volume of traffic log. Sampling technology now becomes the focus of the research of network traffic measure.This thesis firstly analyses and researches IP network measurement and its related technologies,and then studies the sampling technology of network traffic. On this basis we improved the traditional network traffic measure system, and designed an adaptive sampling network traffic measure system. And besides, we designed a real-time algorithm developed from the method to estimate Hurst Parameter. This algorithm can calculate H value more efficiently. By this algorithm, we analyzed the influence on H value by the network anomaly, and a based on self-similarity adaptive sampling method is proposed. By this method, sampling strategy adjusts automatically according to the variation of H value in order to capture more network anomaly information,that lays a firm foundation for analysis and treatment at late stage.The adaptive sampling traffic measure system dynamically adjusted the sampling interval with the change of the Hurst Parameter. It not only made network traffic measurement have intelligence, but also improved the percentage of sampling, advanced the precision of the sampling measure obviously.