一种新的基于ARIMA-SVM网络流量预测研究

研究网络流量预测精度问题,网络流量受多种因素的综合影响,其变化具有周期性、非线性和随机性等特点,将ARIMA模型和SVM模型相结合建立一种网络流量预测模型。采用ARIMA预测网络流量周期性和线性变化趋势;然后采用SVM对网络流量非线性和随机性趋势进行拟合;最后将两者结果再次输入SVM进行融合,得到网络流量最终预测结果。采用具体网络流量数据对模型性能进行测试,仿真结果表明,ARIMA-SVM提高了网络流量预测精度,降低了预测误差,能更全面刻画网络流量变化规律。

Application of ARIMA-SVM model in network traffic prediction

Network flow is affected by many factors, its change is cyclical, nonlinear and stochastic characteristics. This paper used ARIMA model and SVM model to establish a network traffic prediction model. It used ARIMA forecasting network traffic cyclic and linear trend, and then used SVM to network flow nonlinear and stochastic trend fitting. At last two results again entered the SVM integration, network traffic had been the final prediction results. By using network flow data for model performance test, simulation results show that, ARIMA-SVM increases network traffic prediction accuracy, reduces the prediction error, and it can fully characterize network traffic variation.