基于GRU神经网络的移动应用软件流量检测

Mobile application software traffic detection based on GRU neural network

针对移动应用软件流量呈动态变化趋势,导致检测结果不精准的问题,提出基于GRU神经网络的移动应用软件流量检测方法。基于GRU神经网络多层结构构建GRU神经网络检测模型,将恶意软件和非恶意软件流量特征转化为区域特征向量,根据特征转化结果降维处理入侵威胁数据,获取待检测恶意软件入侵训练集和测试集,完成移动应用软件异常流量辨识。根据辨识结果建立病毒库,设计GRU神经网络隐层单元结构,有效地捕获流量动态变化信息,划分异常流量入侵等级,实现移动应用软件流量检测。由实验结果可知,该方法在对应用软件进行流量检测时得到的结果与实际结果一致。在应用软件40~80天时检测出流量使用异常情况,使用流量为1050 MB,其余时间的流量使用正常,具有精准的软件流量检测效果。

A mobile application software traffic detection method based on GRU neural network is proposed to address the issue of inaccurate detection results caused by dynamic changes in mobile application software traffic.Build a GRU neural network detection model based on the multi-layer structure of the GRU neural network.The traffic features of malicious and non malicious software are transformed into regional feature vectors,and intrusion threat data is processed by dimensionality reduction based on the feature transformation results.The training and testing sets for detecting malicious software intrusion are obtained,and the identification of abnormal traffic in mobile application software is completed.Based on the identification results,establish a virus library,design a hidden layer unit structure of GRU neural network,effectively capture dynamic changes in traffic information,classify abnormal traffic intrusion levels,and achieve mobile application software traffic detection.Based on experimental results,it is evident that this method yields consistent outcomes with the actual results when conducting traffic detection for application software.Anomalies in traffic usage were detected within 40~80 days of software application,with a usage of 1050 MB,while the rest of the time exhibited normal traffic usage.This method demonstrates precise effectiveness in detecting software traffic.The rest of the time was normal traffic usage,which is consistent with the actual statistical results and has accurate software traffic detection effect.