ers through the traffic data. 2)Web server monitors the query requests,which are committed by users from the browser,and relay them to network traffic data server,then return results that was presents users generated by network traffic data server.Wi廿1 this structure. an efficient and friendly traffic data query service。 a Basing on the traffic data gathered by preceding application,this paper presents time—series—based network on anomaly detection novel method.This method,which is based sliding windows in which the data is the FCD(flow connection density),can find out the'anormal fluctuation of network,realize real-time network III anomaly detection and 基于NetFlow的网络流量监测研究与应用 enhance the quality of network management through monitoring the time series that is formed from network traffic. To prevent an anomaly from persistent announcement and disturbing following detection,the data of the anomaly have to be replaced with assuming offers all data.This an work approach that establishes autoregressive model of FCD data by training Neural ANN be (Artificial Networks),it generates more accurate assuming data and also can used to predict trafffic data values. The system based results on preceding methods has run in the network of University of Jinan, show that it is efficient for many kinds of network breakdown and network anomalies caused by network viruses and network attacks. Key Words:NetFlow,network flow,traffic monitoring,anomalies detection IV 原创性声明 本人郑重声明:所呈交的学位论文,是本人在导师的指导下,独立 进行研究所取得的成果。除文中已经注明引用的内容外,本论文不包含 任何其他个人或集体己经发表或撰写过的科研成果。对本文的研究作出 重要贡献的个人和集体,均已在文中以明确方式标明。本人完全意识到 本声明的法律责任由本人承担。 论文作者签名:!丝鱼!红 日 期: 竺Z!:竺: 关于学位论文使用授权的声明 本人完全了解济南大学有关保留、使用学位论文的规定,同意 保留或向国家有关部门或机构送交论文的复印件和电子版,允许论文被 查阅和借鉴;