Research Andapplication Of Statistical Analysis In Network Traffic Monitoring System

With the rapid development of network technology, network provides a fast way for information exchange and resource sharing. Network becomes an indispensable part in people's life and work. With the rapid increase of user groups, complex and diverse of services category, which lead to the tasks undertaken by network more and more heavy. Data grows exponentially, at the same time, complexity and heterogeneity of network increases more and more. Network congestion and network failures occur frequently which cause the huge losses for social and economic. Therefore, it is urgently to enhance network monitoring and optimize network structure.Network traffic monitoring is a part of performance management, it plays a critical role in researching characteristics of network behavior and changes of network behavior. On the basis of monitoring network traffic, it researches network behavior and operation rules by analyzing distribution of network traffic and resource utilization.By statistical analysis of traffic data to establish network performance baseline, which brings to monitor network more effectively, allocate network resources more reasonably, locate network faults more quickly and even provide scientific basis for planning and optimizing the network structure. Therefore, it is of great significance in designing and implementation of network traffic monitoring system.Network traffic is the amount of data in network transmission, it reflects the operating state of network. Network traffic is the critical data which determine whether the network is running normally. Network traffic monitoring monitors the network by the continuous collecting network data .It gets performance of network and main component by statistical analysis of traffic data in order to maintain database logs and historical data of network traffic. By modeling of historical data, it can predict traffic value in the future. The predictive value of traffic can help network traffic monitoring to determine the boundary values which make the traffic anomaly can be detected.Firstly, the paper researched concepts, analytical methods and procedures about statistical analysis and time series analysis. The paper focused on the process of establishing auto regressive moving average model which belonged to time series analysis. The process consisted model order, model parameter estimation, model testing and model prediction. Secondly, the paper studied related technologies of network traffic, including network traffic monitoring technology, simple network management protocol and network traffic anomaly detection technology. Again, the paper analyzed the characteristics of network traffic and used the auto regressive moving average as network traffic model .By gathering traffic data, the paper established prediction model of network traffic which helped anomaly detection to determine the boundary value in the adaptive threshold method. Finally, the paper designed and implemented a network traffic monitoring system which combined the above techniques.The paper designed network traffic monitoring system which combined with time series analysis of statistical analysis. Network traffic monitoring system established prediction model to simulate and predict traffic changes in the future period. At the same time, it used adaptive threshold method to detect traffic anomaly. Adaptive threshold method belonged to anomaly detection technology. It established the detection boundary by network traffic prediction model. It identified traffic anomaly rely on calculating the possibility of real-time traffic value over detection boundary value. Network traffic monitoring system could monitor network traffic of statistical analysis in the specified time, and then graphically displayed network traffic. It also could detect traffic anomaly and send alarm information.