基于UML的软件Markov链使用模型的构建  

Deriving Markov Chain Usage Model from UML Model

作　　者：吴彩华[1] 刘俊涛[2,3] 彭世蕤[4] 李海鸿[1] 

机构地区：[1]空军雷达学院信息对抗系信息作战指挥教研室,武汉430010 [2]军械工程学院计算机工程系软件工程教研室,石家庄050003 [3]华中科技大学电子与信息工程系,武汉430074 [4]空军雷达学院信息对抗系,武汉430010

出　　处：《计算机研究与发展》2012年第8期1811-1819,共9页Journal of Computer Research and Development

基　　金：河北省自然科学基金项目(F2009001435)

摘　　要：构建软件的使用模型是进行软件可靠性测试及软件可靠性评估的基础.近年来,如何由软件的UML模型构造软件的使用模型成为研究热点.对于大型的软件系统来说,应用现有方法构建的软件Markov链使用模型的状态空间过于庞大,模型描述困难,不利于测试用例的自动生成及软件可靠性评估.针对以上问题,提出了一种由UML模型构建Markov链使用模型的方法.该方法将场景的前置条件和后置条件作为Markov链使用模型的状态,将场景的执行及执行概率作为状态之间的转移及转移概率.与现有方法相比,新方法构建的Markov链使用模型的状态空间小且无需人为干预,而且可以很方便地生成测试输入从而进行可靠性测试.针对UML模型的有效性,提出了经过可靠性评估扩展的UML模型生成Markov链使用模型的验证算法.最后通过一个卫星控制系统的实例对新方法的性能进行了验证.Constructing software usage model is basic for software reliability testing and software reliability evaluation. In recent years, how to derive Markov chain usage model from UML model has gained much attention. For large software system, the state space of Markov chain usage model using the existing methods is too large to describe, which is bad for generating test cases automatically and evaluating software reliability. For the above problems, a new method deriving a Markov chain usage model from UML model is proposed. The preconditions and postconditions of scenario are taken as the states of Markov chain usage model, and the performance and the performance probability of scenario are taken as the transition driver and transition probability. Compared with other methods, the state space of gained Markov chain usage model by proposed method is smaller and the constructing process is direct and automatic. Furthermore, the test cases can be generated easily by using the above method. In order to validate the validity of the UML model, the validity algorithm from UML model, which is extended by reliability evaluation to Markov chain model, is proposed too. Finally, the performance of the proposed method is verified by an example.

关 键 词：UML MARKOV链 使用模型 可靠性评估 测试用例 

分 类 号：TP311.5[自动化与计算机技术—计算机软件与理论]