Prediction in LAN Traffic Flow Based on Chaos Theory

There exists widely the self-similarity in LAN traffic flow, and there is a close relationship between the self similarity characteristics and chaotic phenomena. The LAN time series of traffic flow were reconstructed in phase space theory. The embedding dimension and the delay time were computed by the C-C algorithm, and the largest Lyapunov exponent was then calculated via the small data method to determine its chaotic level. The weighted neighborhood prediction method was proposed and conducted considering the only decisive role of the nearest point on the center point based on the largest Lyapunov exponent while ignoring its neighborhood points on the predicting affection. The validation of the method was done by predicting the actual LAN traffic flow.