测度属性关系分析的分布式系统异常检测  

Anomaly Detection for Distributed System Based on Measurement Attributes Analysis

作　　者：张乐君[1] 国林[1] 张健沛[1] 杨静[1] 夏磊[1] 

机构地区：[1]哈尔滨工程大学计算机科学与技术学院,哈尔滨150001

出　　处：《北京邮电大学学报》2013年第6期69-73,共5页Journal of Beijing University of Posts and Telecommunications

基　　金：国家自然科学基金项目(61100008;61073041;61073043);中央高校基本科研业务费专项资金项目(HEUCF061002)

摘　　要：针对分布式系统的特点,提出了一种基于网格空间的测度属性关联度计算方法,并以此建立了网络模型来刻画分布式系统测度属性的关系;提出了基于隐马尔可夫模型的测度属性关系网络划分方法;利用最新监测的属性关联度信息,设计了基于测度属性关系网络划分的分布式系统异常检测算法;对多种攻击进行异常检测实验,从而验证了提出方法的有效性.The anomaly detection methods based on measurement attributes relationship analysis is pro- posed. The correlation between the measurement attributes in space grid structure is calculated, and hereby the relationship network model of measurement attributes is established. The network segmentation method is put forward based on hidden Markov model. The anomaly detection algorithm for distributed systems is designed by using the newest monitoring attributes correlation. The anomaly detection experi- ments demonstrate the effectiveness of the proposed method.

关 键 词：分布式系统 测度属性 关系网络划分 异常检测 

分 类 号：TN929.53[电子电信—通信与信息系统]