Improving Fault Localization through Fault Propagation Analysis

Abstract

Lots of studies have been done to pinpoint program faults almost to the exact line where different types of abnormal behaviors occur. Among them there is the promising testingbased fault localization (TBFL) technique, the aim of which is to locate the faults in the program automatically when the program is executed with test suite. This paper proposes an improved TBFL method augmented with fault propagation. It localizes faults in three major steps: generate the execution path and dependency information of program, modify the existing fault state of program, and ultimately infer statement suspicious score. Our empirical results show that, for the subject we studied, our technique can get better effectiveness than others. Empirical results show that our method can help to debug efficiently and get better effectiveness than other competitors.