Design And Implementation Of Network Traffic Monitoring System Based On Time Series

Network traffic management is an important part of network management, network traffic is the data transmitted over the network to reflect the performance and operating status of the network. The analysis of network traffic can not only play a monitoring role of network traffic, but also be able to have a certain network behavior rules to do the implementation of network traffic forecast. It can timely warn for traffic anomalies to help network managers to troubleshoot and locate network problems, and provide a scientific basis for optimizing the network topology and network planning.This paper focuses on network traffic, designs and implements a network traffic monitoring system based on B/S model, which has good usability, visibility and scalability. It intergrates with network traffic related technologies, including network traffic monitoring technology, network traffic prediction technology and network traffic anomaly detection technology. The system implements the data collection by the network traffic monitoring technology based on SNMP protocol, and improves the collection efficiency by a multi-threaded data collection method. For collected historical data of network traffic, the system implements the mathematical modeling by time series analysis methods. The modeling process uses the classic Box-Jenkins method, including several major steps contains stationary test for time series, non-stationary time series smoothing, model order determination, model parameter estimation, model verification and model prediction.The established ARMA model has been used to achieve the short-term forecasting of network traffic online and provides a source to determine the boundaries for the network traffic adaptive threshold anomaly detection method. The system can have a good real-time monitoring of network traffic for the target network. It can display the network traffic in a clear graphical form, predict trends in the coming period of network traffic accurately, and detect and alarm timely for possible network traffic anomalies.The implemented network traffic monitoring system has been deployed on the server of Communication & Computer Network Key Laboratory of Guangdong of South C hina University of Technology. After testing, the system is running well and achieves the desired goals. It can meet the actual network traffic management needs for the campus network.