Analysis And Study Of The Network Traffic Characteristic

Analysis to the network traffic characteristic is still the key of the fields related with network measurement,network behavior study,network call admission control and network performance analysis.With the rapid development of communication network involving computer network,the network traffic comes to show different properties compared with the traditional network traffic.Accordingly,new challenge has been brought to the analysis and study of the wired and wireless network traffic. At the same time,new fields has also been introduced which will absorb swarm of attention and investigators.This paper is under the National Nature Fund "IP network topology and analysis of performance based on network detection" and "Research on the Key Technologies of Performance Measurement of NGI". New ideas and methods have been presented and large amount of work has been practiced through referring to large amounts of materials and certain improvement has been obtained in this project.The content of this paper includes:first,the traditional typical traffic models have been introduced,followed by the presentation of prevalent self-similarity traffic model.Then,The classification of self-similarity traffic generation methods is given with the description of some noted generation algorithms in every class as examples.Subsequently,the traffic forecasting means which is the foundation of Charpter 3. are presented including the classical time series forecasting and some up-to-date algorithms base on computation intelligence.;In charpter 3,through the analysis to the statistic properties of the wlan traffic trace,the seasonal and cyclic character of wlan traffic has been certified.And then,a multiply season forecasting model is given ground on the previous conclusion which outperforms the other models in the forecasting performance.;In charpter 4,the study to the relationship between file size,protocol and hurst parameter is approved by simulation at first.Then the connection of hurst to traffic source aggregation level is validated subsequently.At last,a traffic generator coded with java and linux c has been finished base on the previous conclusions.