Chaotic Time Series Forecasting Based on Wave Echo State Network

Abstract

The chaotic time series forecasting oriented to the network traffic forecasting is put forward in order to analyze the behavioral traits of the network traffic and make forecasting through modeling. Firstly the time series of one-dimensional network traffic is reconstructed into a multi-dimensional time series and then the support vector machine is taken as a position of bird’s nest to find the optimal parameters through the simulation of parasitism and reproduction mechanism of cuckoo species and finally the network traffic forecasting model is to be established based on the optimal parameters and the performance of chaotic time series forecasting will be tested through the simulation experiment. The simulation result shows that, compared with the reference model, the chaotic time series forecasting improves the forecasting accuracy of the network traffic and more accurately demonstrates the complex change trend of the network traffic and provides the chaotic network traffic with a new research tool.