Research On Software Multi-fault Localization Based On Swarm Intelligence Algorithm

During the development and maintenance of software,a significant amount of effort is needed to determine the root causes of failures.A program may contain one or more faults,but most spectrum-based fault localization approaches perform well for programs with only one fault.Thus,a method is proposed to locate not only single faults but also multiple faults.In this paper,we apply improved Particle Swarm Optimization for Multi-fault Localization(PSOMFL).This method uses the idea of search based software engineering to transform the multi-fault localization into an optimization problem that can be solved by particle swarm optimization algorithm.Through the evolutionary behavior of the algorithm,the optimal solution set of the problem is obtained.Finally,through the analysis of the optimal solution set,the detection sequence containing multiple suspicious entities is obtained.In addition,we apply improved Ant Colony algorithm for Multi-fault Localization(ACOMFL).ACOMFL uses spectrum information to create a special form of search graph,and obtains the optimal solution set of the problem through the unique evolutionary behavior of individuals in the ant colony algorithm.By analyzing and counting the entity information contained in the optimal solution,the suspect sequence containing multiple defective entities can be obtained.Compared with PSOMFL,ACOMFL can improve the efficiency and accuracy of the method through distributed computing and parallel computing,making it more suitable for defect location in large software.In the empirical research stage,we select four large Linux programs as the evaluation object,and shows the effectiveness of the method by comparing with the performance of seven SBFL methods under various indicators.Experimental results show that PSOMFL and ACOMFL have excellent performance in mutil-fault localization and single fault localization,also the execution time and efficiency of the method are acceptable.