服务器集群网络中纠缠疑似攻击检测技术研究  

Research on Detection Technology for Entanglement Suspected Attack in Server Cluster Network

作　　者：杨云峰[1] 汤鹏杰[2] 

机构地区：[1]河池学院计算机与信息工程学院,广西宜州546300 [2]井冈山大学数理学院,江西吉安343009

出　　处：《计算机仿真》2015年第5期315-317,349,共4页Computer Simulation

基　　金：河池学院课题(2014QN-008;2014ZD-N002);河池学院"计算机网络与软件新技术"重点实验室(院科研【2013】3号)

摘　　要：传统的直推式服务器集群网络攻击检测方法具有高检测率和低误报率的优点,但由于算法中采用了欧氏距离的计算使得计算量巨大,且对纠缠疑似攻击的检测效率不高。采用粗糙集的直推式网络异常数据检测方法对服务器集群数据包进行异常监测,在对异常数据奇异值的计算中,引入相似度概念,避免欧式距离出现的"维度灾难",对检测出的异常数据特征加入属性权重进行攻击识别,提高了对服务器集群网络纠缠疑似攻击的检测率和及时性。仿真结果表明,提出的算法对服务器集群网络攻击中纠缠疑似攻击检测具有较高的检测率和较为及时的响应速度。Traditional method for transductive server cluster network attack detection has the advantages of high detection rate and low false alarm rate, but in the algorithm, calculation of Euclidean distance is adopted, making a huge amount of calculation and low efficiency of entanglement suspected attack detection. The method of abnormal data detection in transduetive network based on rough set was applied to make abnormal monitoring for the data packets of server cluster. In the calculation of singular values of abnormal data, the concept of similarity was introduced to a- void the ＂dimension disaster＂ of Euclidean distance. The attribute weight was added to the detected abnormal data feature to identify the attack, which improved the detection rate and timeliness of entanglement suspected attack in server cluster network. The simulation experiment shows that the proposed algorithm has higher detection rate and a more timely response speed.

关 键 词：服务器集群网络 直推式检测 粗糙集 视情推理 

分 类 号：TP211[自动化与计算机技术—检测技术与自动化装置]