A Rule-Based Approach for Identifying Failure Inducing Combinations in Combinatorial Testing

Abstract

In complex software systems, identification of parameter combinations that lead to failures is important for effective debugging and robust software quality assurance. Combinatorial testing (CT) can greatly reduce the number of test cases (TC) used for testing a complex system by generating TCs based on various parameter combinations, while still maintaining high fault-detection capabilities. In this paper, a rule-based approach is presented that will aid in identifying Failure Inducing Combinations (FIC) of parameters that caused a fault in CT generated TCs. The approach uses heuristics to identify rules to methodically split parameter combinations into more likely or less probable combinations causing failure, thus reducing the set of failure sources. This approach is tested on two different case studies i) Three factor authentication system and ii) Existing literature-based input. The results obtained show significant accuracy and time saving for program debugging, thereby indicating applicability of the approach to real-world problems. The approach was found to be successful in identifying the pair-wise combinations of parameters that are likely to cause failure.