Improving MC/DC and Fault Detection Strength Using Combinatorial Testing

Abstract

Software, in many different fields and tasks, hasplayed a critical role and even replaced humans to improveefficiency and safety. However, catastrophic consequences can becaused by implementation bugs and design defects. Modifiedcondition/decision coverage (MC/DC), required by the FederalAviation Administration on Level A (the most safety criticalsystem), has been shown to be effective in detecting softwarebugs. However, generating tests to achieve high MC/DC can bevery expensive and time consuming. Recently, many studiesshowed that combinatorial testing (CT) could generate highqualitytest cases in a cost-effective way. Can CT generate testcases to achieve high MC/DC? In this paper, we conduct anempirical study on two real-life programs to evaluate theefficiency and effectiveness of using combinatorial testing toimprove MC/DC coverage achievement, as well as the faultdetection strength.