摘要
针对移动通信网络流量预测由于预测冗余,导致预测的RMSE和MAE高的问题,文章提出了基于深度学习的移动通信网络流量预测方法.该方法将网络流量视为时间序列,分析网络流量组成结构的自相似性,在图结构上定义网络流量特征,提取流量自相似性特征.引入深度学习算法,将图信号矩阵表示预测节点特征,利用多层神经网络实现卷积运算,构建流量预测算法.更新带有注意力信息的空间隐藏特征,实现移动通信网络流量预测优化.实验结果表明,该方法的RMSE在0.000～0.009范围,MAE 在 0.000～0.070 × 10-3 范围,并且使用H_9函数时,预测误差为0,说明该方法预测精准度较高,具备良好的预测性能.
Abstract
In view of the problem of high RMSE and MAE due to mobile communication net-work traffic redundancy prediction,the paper proposes a mobile communication network traffic prediction method based on deep learning.The method regards the network traffic as a time se-ries,analyzes the self-similarity of the network traffic composition structure,defines the net-work traffic features on the graph structure,and extracts the traffic self-similarity features.The deep learning algorithm is introduced,the graph signal matrix represents the predicted node characteristics,and the multi-layer neural network is used to realize the convolution operation and construct the traffic prediction algorithm.Spatially hidden features with attention informa-tion are updated to realize the optimization of mobile communication network traffic prediction.The experimental results show that the RMSE of the method is in the range of 0.000～0.009 and the MAE is in the range of 0.000～0.07010-3,and the prediction error is 0 when using H_9 function,indicating that the method has high prediction accuracy and good prediction performance.
NETL
NSTL
万方数据