摘要
研究网络流量准确预测,针对网络优化控制问题,由于网络数据拥塞严重,网络流量变化具有高度自相似性、非线性和多尺度等特点,线性数据的传统预测方法无法准确刻画网络流量的非线性变化规律,导致预测准确率低。为了提高网络流量的预测准确率,在分析网络流量变化特征的基础上,提出一种小波分析的网络流量混合预测模型。混合模型首先利用小波分析将网络流量分解线性和非线性部分,然后分别采用ARIMA模型和BP神经网络模型对其进行预测,最后采用小波分析对线性和非线性部分预测结果进行重构,得到混合模型最终预测结果。仿真结果表明,混合模型比其它网络流量预测模型具有更高的预测准确率,为网络优化控制提供了有效分析方法。
Network optimization control problem is researched. Network traffics have highly selfsimilarity, nonlinear and muhi-scale features, the traditional forecasting methods based on linear data cannot accurately depict net- work traffic of nonlinear variation, and prediction accurate is low. In order to improve the network traffic prediction accurate, this paper proposed a network traffic hybrid prediction method based on wavelet analysis. The hybrid model used wavelet analysis to decompose the linear and nonlinear parts from network traffics, and then used ARIMA model and the BP neural network model respectively to prediction them, lastly used wavelet analysis to reconstruct the linear and nonlinear partial results and eventually to get prediction result from the hybrid model. Simulation results showed that the hybrid model is better than other network traffic prediction model; it has higher prediction accurate and is an effective analysis method for network optimization control.
出处
《计算机仿真》
CSCD
北大核心
2011年第9期158-160,202,共4页
Computer Simulation
关键词
小波变换
神经网络
自动回归模型
Wavelet transform
Neural network
Self-regression mode