Detecting associative shift faults in predicate testing

Fault-based predicate testing is a promising strategy for generating effective software test data. The test target is faults that may appear in predicates found in the specification or implementation of a computer program. A hierarchy of fault classes has been recently established, identifying associative shift faults (ASFs) as one of the strongest fault classes in predicate testing. The upto date proposed approaches resolve adequately the issue of test generation for weaker fault classes but do not guarantee the detection of all ASFs. In this paper we define a suitable fault model to represent this type of faults and propose a heuristic test strategy, while trying to keep low the number of required tests by fault simulation. Empirical results of the algorithm application on the TCAS-II expression suite are encouraging and suggest that it is easy to detect a significant subset of ASFs by considering only few additional tests.