关键词: 有效性, 预测, 长相关, 小波, 自回归滑动平均模型

Abstract: To mitigate the network congestion by node failures, a new effectiveness evaluation method was proposed with reference to wavelet transform. In this method, the long-range dependence of actual traffic was depicted at first, and the prediction algorithm of arrival traffic was established with Auto-Regressive and Moving Average (ARMA) model, of which standard deviation between the prediction data and actual data was 6.89. Then, a simulation was conducted to research the relationship between the effectiveness and other influential factors. The results show that the effectiveness is of positive correlation with average velocity, and negative correlation with buffer size.

Key words: availability, prediction, long-range dependence, wavelet, Auto-Regressive and Moving Average (ARMA) model