A Study on the Measurement and Forecasting Models of Network Traffic

• Main Authors: Lee,Menp-Jean, 李夢珍
• Other Authors: Chen,Yen-Wen
• Format: Others
• Language: zh-TW
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/fh295m

碩士 === 中央警察大學 === 資訊管理研究所 === 90 === All kinds of voice , image and video prevail over the whole internet environment increasingly ,result in serious overload on wide and local area network. Therefore, how to maintain the quality of network service is the important responsibility of network management. Forecasts on accurate network traffic will be the only way to supply multimedia service in the future. This study used the time series analysis on past network traffic data and established fitted models for forecasting future network traffic’s trend. Accordingly , the network management adjust the manage rule based on the aforesaid models dynamically .In the mean time the models can balance network loading and suffice the requirement of network users. Consequently, the network users can avoid congestion before executing internet applications takeover lengthy waiting. Additionally, we also run the outlier models to detect and analyse the network traffic data. By outlier detection , the network management can judge the network status , for example network bursting or delay, conduce to solving bug and delay state. By means of above-mentioned procedure, the network administrators can carry out effectively bandwidth resources allocation.