软件故障定位是软件测试实践中最重要的活动之一。目前，利用人工方式通过调试器进行故障定位仍然占据主导地位，故障定位能力常常依赖于人员的知识和经验，调试效率低，调试支持工具相对较弱，对程序员要求高。因此，无论在软件开发阶段还是软件投入使用以后，软件调试和故障定位的自动化技术越来越成为一个有重要价值的研究课题。研究了多种成熟的基于谓词的统计学错误定位技术，分析了典型的参数化和非参数化故障定位统计模型，提出了一种新的基于谓词统计的故障定位算法：频率差异积分算法（Frequency Difference Integration ，简称为 FDI），突破了现有算法的部分限制；最后对 FDI 算法进行了验证，并和已有算法进行了对比，验证了其有效性和高效性。借助该研究成果，可以有效提高故障定位的准确率和覆盖率，这对于推动程序自动化调试技术、缩短软件开发周期、降低软件的维护成本具有一定的意义和实用价值。%Locating software faults manually with the help of debugging tools often relies on the knowledge and ex -perience of a programmer, has poor debugging efficiency and high requirements for the programmer .Therefore, the automation of software debugging and fault location is increasingly urgent and important .However, due to the com-plexity of software faults and complicated processes , the existing fault location methods have strong dependence and weak versatility, being difficult in analyzing the causes of the faults correctly .We analyze the existing non-paramet-ric and parametric statistical fault location algorithms and propose our fault location algorithm for the frequency difference integration (FDI) of predicate statistics, which we believe is more effective and efficient .Finally, to verify the effectiveness and efficiency of our fault location algorithm , we perform experiments on the Siemens suite ;the experimental results, given in Figs.3 and 4 and Table 1, and their analysis show preliminarily that : (1) our fault location algorithm can effectively locate software faults no matter whether the probability density function of the predicate′s truth value conforms with any known parametric models ; (2) compared with other fault location algo-rithms, our fault location algorithm reduces computational costs and computational complexity .
展开▼
