网络化软件交互式行为可信性监测分析仿真  

Networked Software Interactive Behavior Credibility Monitoring Analysis Simulation

作　　者：杨柳[1] YANG Liu(School of Science, Ningxia Medical University, Yinchuan Ningxia 750019, China)

机构地区：[1]宁夏医科大学理学院,宁夏银川750019

出　　处：《计算机仿真》2018年第6期304-307,312,共5页Computer Simulation

摘　　要：通过可靠的监测手段,可实时监测软件交互过程运行状态,及时发现并排除软件故障,提高软件的可信性。现行软件行为分析方法利用隐变量来表示影响行为可靠性的多种因素,监测结果受节点任务波动干扰影响较大,通过监测隐变量的状态变化难以作为交互行为是否可信的依据,导致监测结果与实际结果不符。提出一种基于支持向量机的网络化软件交互式行为可信性监测分析方法。对软件交互式行为散列时间有效性可检验条件分析,通过验证值对比与散列时间有效性统计结果判定行为节点完整性状态,根据节点完整性状态对行为可信性进行检验。采用支持向量机对交互式行为进行分类,使用已获得的行为度量基对交互行为可信性进行评价,通过对交互过程中的行为迹的监测实现了软件交互式行为可信性分析。实验结果表明,所提方法受节点任务波动干扰影响较小,在软件交互式行为可信性监测分析方面较其他方法具有明显优势。The aim of this article is to overcome defect of traditional monitoring method for credibility of interac- tive behavior of software, such as inconformity between monitoring results and actual results. Based on support vector machine （SVM）, a new analytical method for the credibility monitoring is proposed. Firstly, testable condition of hash time efficiency of the interactive behavior was analyzed and integrity state of behavior node was judged via comparison of variation value and statistical result of the hash time efficiency. According to the integrity state, behavior credibility was tested. Then, using the SVM, the interactive behavior was classified and the credibility of interactive behavior was evaluated using the obtained measurement base of behavior. Finally, via monitoring on behavior trace during the interactive process, credibility analysis of the interactive behavior was achieved. Simulation results show that the method has less disturbing influence of node task fluctuation. It has more obvious advantage than other methods of the credibility of interactive behavior.

关 键 词：网络化软件 交互式行为 可信性 监测分析 

分 类 号：TP393.08[自动化与计算机技术—计算机应用技术]