基于强化LSTM的网络高隐蔽性入侵轨迹预测研究  

Research on network high concealment intrusion trajectory prediction based on enhanced LSTM

作　　者：徐李阳 王晨飞 穆松鹤 杨自兴 马建勋 XU Liyang;WANG Chenfei;MU Songhe;YANG Zixing;MA Jianxun(State Grid Customer Service Center,Tianjin 300300,China;State Grid Slji Testing Technology(Beijing)Co.,Ltd.,Beijing 100192,China)

机构地区：[1]国家电网有限公司客户服务中心,天津300300 [2]国网思极检测技术(北京)有限公司,北京100192

出　　处：《电子设计工程》2024年第21期104-107,112,共5页Electronic Design Engineering

基　　金：国家电网科技项目(16ER63857)。

摘　　要：网络高隐蔽性入侵信息的维度难以确定,导致入侵轨迹预测困难增加,因此研究基于强化LSTM的网络高隐蔽性入侵轨迹预测方法。设置强化LSTM预测模型基础架构,根据历史数据特征取值结果,求解标记参数,利用这些参数标记入侵数据轨迹节点。确定高隐蔽性入侵行为的表现强度从而确定入侵向量。结合入侵信息维度实现网络高隐蔽性入侵轨迹预测。实验结果表明,在强化LSTM模型的作用下,高隐蔽性入侵信息维度的预测结果完全属于该信息所处轨迹维度参数实际取值范围之内,说明该方法的预测结果更为精准。The dimensionality of highly covert intrusion information in networks is difficult to determine,leading to increased difficulty in predicting intrusion trajectories.Therefore,research is being conducted on a network highly covert intrusion trajectory prediction method based on enhanced LSTM.Set up an enhanced LSTM prediction model infrastructure,and based on the historical data feature value results,solve the labeling parameters,and use these parameters to label the intrusion data trajectory nodes.Determine the intensity of highly covert intrusion behavior to determine the intrusion vector.Combining the dimension of intrusion information to achieve high concealment intrusion trajectory prediction in the network.The experimental results show that under the strengthening of the LSTM model,the predicted results of the high concealment intrusion information dimension are completely within the actual value range of the trajectory dimension parameters where the information is located,indicating that the prediction results of this method are more accurate.

关 键 词：强化LSTM模型 网络入侵轨迹 历史数据 入侵行为 入侵向量 信息轨迹维度 

分 类 号：TN911[电子电信—通信与信息系统]