基于小波分析与卡尔曼滤波的网络流量估计和预测方法

The Traffic Estimating and Forecasting Method Based on Kalman Filtering and Wavelet Analysis

将目标状态的小波变换系数向量描述为卡尔曼滤波方法的状态变量,进而建立了网络流量估计和预测模型,能够实现周期内的实时跟踪和动态多步预测.利用CERNET华中地区主干网的实测流量数据对该模型进行检验,所有检验周期网络流量预测值的相对误差均值为4.58%,表明网络流量估计和预测模型具有较强的适用性.

By describing the wavelet coefficient as the Kalman filtering state variable, we propose the traffic estimating and forecasting method. This paper, with this method, realize, the real-time tracking and dynamic multistep forecasting in one cycle. Then this method is tested by the real CERNET traffic data. It is clearly that this new method is suitable to the traffic estimating and forecasting.