Font Size: a A A

The Applied Research Of Multivariate Chaotic Time Series Analysis On Network Traffic Prediction

Posted on:2014-11-09Degree:MasterType:Thesis
Country:ChinaCandidate:X C SuiFull Text:PDF
GTID:2268330398490510Subject:Computer technology
Abstract/Summary:PDF Full Text Request
Along with the technology of network attacks continues to improve, network security is threatened as never before, and it is imperative to improve intrusion detection technology. The network flow has a great sensitivity to the network attacks. This paper depend on chaotic dynamics characteristics of the network traffic, apply the multivariate chaotic time series analysis theory to the prediction of network traffic for improving the intrusion detection’s accuracy and performance of real-time. The main work of this paper is as follows.First, author makes a phase space reconstruction to the network traffic data, focus on the selection method of two reconstruction parameters which are embedding dimension and time delay. This paper has improved correlation integral method, and has applied the method to multivariate time-series analysis, and finally has adopted the method after comparing different selected method of parameters. The experimental results show that this method has a better effect in the prediction of network traffic.Secondly, the paper use local polynomial prediction method, the local average prediction method and local linear prediction method in the network traffic prediction. Having compared these three methods, author found that the local polynomial prediction method is more accuracy. Subsequently, the local polynomial prediction method is extended to the multivariate time series. The experimental results show that multivariate timing of local polynomial prediction method has a better prediction effect.Finally, the author prepared a network traffic data analysis procedure. By using that procedure, network data has been acquired and collected. These data is used in the prediction model to analyze network packets and network traffic. Moreover they are used to do multivariate chaotic time series reconstructed and predicted. The experimental results show that the method of multivariate chaotic time series prediction, using network traffic and network packets as variables, has improved univariate chaotic time series prediction results to be accuracy and has got a better real-time performance.
Keywords/Search Tags:Multivariate chaotic time series, Embedding Dimension, Phase SpaceReconstruction, Local polynomial prediction method, Correlation integral method, NetworkTraffic Prediction
PDF Full Text Request
Related items