【摘要】 云网融合是云计算与通信网络的结合。云计算技术为动态调控网络资源提供支持。为了能够及时、按需调控网络资源，就需要提前预知网络流量信息。因此对网络流量进行预测，成为动态管理网络资源的前提和保障。提出一种并行流量实时预测模型，通过并行运行GARCH和LSTM模型，达到对未来网络流量的实时预测。该模型结合线性模型和非线性模型的优点，既能够对周期性、有规律的流量值进行预测，也能对随机、突发大流量值进行预测。通过对并行流量预测模型、单一GARCH模型和单一LSTM模型进行对比实验，结果表明并行流量预测模型比单一LSTM模型的均方根误差（RMSE）降低了4.403%，比单一GARCH模型的RMSE降低了5.833%。除此之外的平均绝对误差（MAE）和平均绝对百分比误差（MAPE）均有所降低。更多还原

【Abstract】 Cloud network convergence is the combination of cloud and network. Cloud intelligence technology provides support for dynamic regulation of network resources. In order to control the network resources in time and on demand,it is necessary to predict the network traffic information in advance. Therefore,the prediction of network traffic has become the premise and guarantee of dynamic management of network resources. A parallel traffic real-time prediction model is proposed. By running GARCH and LSTM models in parallel,the real-time prediction of future network traffic can be achieved. The model combines the advantages of linear model and nonlinear model. It can not only predict the periodic and regular flow value,but also predict the random and sudden large flow value. Through the comparative experiment of parallel traffic prediction model,single GARCH model and single LSTM model,the results show that the root mean square error(RMSE)of parallel traffic prediction model is 4.403% lower than that of single LSTM model and 5.833% lower than that of single GARCH model. In addition,the average absolute error(MAE)and the average absolute percentage error(MAPE)are reduced.更多还原