Analyzing And Forecasting Traffic Characteristics Between Backbone Networks

For the past few years, the Mobile Internet in China has already become one of the most important parts of global Internet. Up to the end of December2013, the number of Internet users in China has reached618million, the Internet penetration rate is45.8%. And the number of Mobile Internet users is more than500million. As the rapidly developing of the Mobile Internet, the problems are also increasingly exposed. Firstly, the increasing of network is rapidly aggravate due to the increasing of user numbers and the increasing of new applications. And the Mobile Internet users require for higher quality of service and network stability. Secondly, the Mobile Internet architecture is becoming more and more complicated, many problems are lack of deep research and accurate description, which seriously influenced the usage rate of network resource and the further development of Mobile Internet, such as network traffic characteristics, user behavior characteristics and traffic variation trend. Lastly, the exponential growth of network traffic makes traditional traffic analyzing methods face challenge of dealing with huge amount of data. Therefore, new methods which are more efficient and reliable are needed. The Hadoop framework, whose core is MapReduce calculating module, has gradually become a basic distributed massive data processing architecture in cloud computing technology.In this thesis, we introduce the significance of network traffic analyzing and forecasting firstly, include the research status of network traffic analyzing and forecasting.Secondly, we introduce Hadoop distributed platform and Hadoop based network traffic processing system. Moreover, we detailed explain the principle and technological process of the network traffic processing system in detail, which was develop by ourselves.Thirdly, we analyze the traffic characteristics between backbone networks using the network traffic processing system developed by ourselves. And show our analyze result.Finally, we introduce the principle of SARIMA (Seasonal Auto-regressive Integrated Moving Average Model). We draw this model into the area of network traffic forecasting, and forecast the traffic characteristics between backbone networks. Also, we evaluate the forecasting result.