On Input Pro le Selection For Software Testing

This paper analyzes the e ect of input pro le selection on software testing using the concept of fault detectability pro le. It shows that optimality of the input prole during testing depends on factors such as the planned testing e ort and the error detectability pro le. To achieve ultra-reliable software, selecting test input uniformly among di erent input domains is preferred. On the other hand, if testing e ort is limited due to cost or schedule constraints, one should test only the highly used input domains. Use of operational pro le is also needed for accurate determination of operational reliability.