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.
Keywords: Network traffic
forecasting; Support vector machine; Cuckoo search algorithm; Parameter optimization
Author: Liu Jun-xia, Jia
Zhen-hong
Journal Code: jptkomputergg160079
