Analyzing the Long Memory for Traffic Data and Developing the Network Security System

• Main Authors: Pan, Wei- Jiun, 盤維鈞
• Other Authors: Wu, Ji-Cheng
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/90907126351501830629

碩士 === 國立屏東科技大學 === 工業管理系所 === 97 === Nowadays the internet is more and more convenient, but people will worry about the security when they are surfing the internet. If the website is attached by the hacker, it will reduce the efficiency of the network, and the speed will be slow down, even paralyze the website. The worse thing is, your personal information will be stolen for the evil purpose. As a result, the network security is become the most important issue of Network Management. Practically, the network manager through observing the flow of organized or the unit network, usually can find that if the use of population, habit, and network system did not have the great changes, basically the flow of network will appear a kind of similarity cycle and a specific pattern. This study is based on the flow history, and use the theory of quality control statistics, to build the appropriate control chart. Offer the network manager a tool to control the change of network flow, and expect can detect the unusual condition in advance. However, traditional control chart data products with the assumption that the same independent identical the normal distribution, but the network traffic information is clearly with the traditional control chart does not match the assumptions, with autoregressive features even of long memory. In this paper, first of all, the history of a site to collect traffic information, the use of Hurst's index to verify whether the information has a long memory, and network traffic with Hurst parameters results validate the distinction between, respectively, fitting the appropriate autoregressive integrated moving-average(ARIMA) and autoregressive fractional moving-average(ARFIMA), the model will be fitted to monitor network traffic parameters used to construct time series of abnormal control chart.