基于小波变换的网络流量预测模型

目前研究发现实际网络流量具有明显的分形特性，流量的多重分形特性对网络性能有着非常重要的影响，有必要建立一个基于多重分形特性的可以同时预报长相关和短相关特性的实际网络业务模型。利用AR, ARMA等模型对短相关数据能较好地预测而对长相关数据预测精度不高的特点，并结合小波变换能够去除实际数据相关性，建立新的预测模型，使其对长相关数据同样具有比较高的预测精度。改进后的模型克服了FARIMA模型计算量比较大的缺点，保持了算法的简单性。

Network Traffic Prediction Models Based on Wavelet Transform

More recent studies have proposed that fractal is a ubiquitous property of real-traffic, and the multi-fractal of traffic has great impact on network performances. So it is important to build the forecasting model based on multi-fractal for service quality of network, which is a good traffic model capable of capturing both Long-Range Dependence(LRD) and short-range behavior of a network traffic stream. As the AR and ARMA models have a good prediction to the short-range dependence data except for the LRD, it proposes a new prediction model which has the great accuracy for the LRD, it is based on Wavelet transformation which can release the correlation of data. So the new model overcomes the defect of the FARIMA model, and maintains the simplicity of the algorithm.