基于小波和ARIMA模型的业务流预测方法

摘要

Aiming at the problem of the traffic flow performance by node failure, a novel status prediction algorithm Trous-based ARIMA Prediction(TAP) of Ad hoc network is proposed. In this algorithm, wavelet transform is adopted to ease up the long dependence of actual traffic, and the status prediction formula is built by Autoregressive Integrated Moving Average(ARIMA) model and Kalman filter. A simulation is conducted to study the accuracy between TAP and ARIMA, as well as FARIMA. Results show that TAP has better performance, and the residual is 18.23%.%针对因节点失效而造成的业务流性能变化问题,提出一种新的Ad hoc网络状态预测算法TAP.该算法利用小波变换减弱实际业务流的长相关特性,并结合自回归移动平均(ARIMA)模型和Kalman滤波建立状态预测方程.通过仿真实验对比分析ARMA和FARIMA的预测精度,结果表明,TAP算法业务流性能较优,其残差为18.23％.