关于网络数据库传输中异常数据检测仿真研究

摘要

This paper proposes an abnormal data detection method of network database based on cosine clustering.Firsdy,the structure model of abnormal network database is built,and reasons for formation of abnormal data in depth taking structure model as basis is analyzed,then feature of abnormal data in network database is extracted,and cosine clustering is used to build a behavior model of abnormal data in network database.Finally,through measuring deviation distance between characteristics of abnormal data and normal behavior model,abnormal data detection in network database is completed.Simulation results show that the proposed method has high sensitivity and efficiency for detecting abnormal data in network database.%对网络数据库传输中异常数据的检测,能够有效提高数据库的安全性以及稳定性.对异常数据的检测,需要得到异常数据的特征,求取其与正常行为模型之间的偏差距离,完成对异常数据的检测.传统方法将数据库数据流量矩阵高维数据映射至正常子空间和异常子空间,但忽略了求取数据点特征与正常行为之间的偏差,导致检测精度偏低.提出基于余弦聚类的网络数据库异常数据检测方法.上述方法先建立异常网络数据库的结构模型,并以构建的结构模型为依据深入分析异常数据形成的原因,提取网络数据库异常数据的特征,采用余弦聚类方法建立网络数据库异常数据行为模型,通过度量异常数据特征与正常行为模型之间的偏差距离,以此完成网络数据库异常数据检测.实验结果表明,所提方法对网络数据库异常数据的检测具有较高的敏感性和高效性.