Fault Detection Analysis And Research Based On Code Structure

Mutation testing is a kind of software testing technique based on fault detection,which used to measure the error detection ability of a test suit and to structure a complete test suit.The research shows that mutants can represent the real faults in a program.But the large number of redundant mutants increase the cost of testing.The mutant is hard to detect has a better utility.It is important to produce mutant has a good utility as little as possi-ble in mutation testing.This paper studied and discussed how to improve the production quality of the mutant.The study is conducted in two aspects.First,analysis the distribution feature of mutant utility in statement subsumption graph of the program.The results show that the mutants generated in the root onear the root nodes of the graph,are easier to have a high utility.That is,if a mutant is covered by less test cases,it is more difficult to be killed.Second,analysis the distribution feature of subsuming mutant in mutation operator.Subsuming mutants all have a high utility.In this paper we selected the operator which generate a higher percentage of subsuming mutants.The results show that the mutant generated by some particular mutation operators are more difficult to be killed.The paper shows that mutants selected by analysis characteristics has smaller quantity and higher utility,and could get a high mutation score,compared to the randomly selected mutants.