Forward-Propagation Approach for Generating Feasible and Minimum Test Case Suites From Cause-Effect Graph Specifications

Abstract

Cause-effect graphs are a popular black-box testing technique. The most commonly used approach for generating test cases from cause-effect graph specifications uses backward propagation of forced effect activations through the graph in order to get the values of causes for the desired test case. Many drawbacks have been identified when using this approach for different testing requirements. Several algorithms for automatically generating test case suites from cause-effect graph specifications have been proposed. However, many of these algorithms do not solve the main drawbacks of the initial back-propagation approach and offer only minor improvements for specific purposes. This work proposes two new algorithms for deriving test cases from cause-effect representations – an algorithm that uses forward-propagation of all possible cause values in order to get all feasible test cases and an algorithm that minimizes the feasible test case suite while taking multiple effect activations into account. Cause-effect graphs previously used for testing algorithms based on the backward-propagation approach were used for evaluation and comparison, as well as the newly introduced test effect coverage metric and fault detection rate effectiveness. The evaluation shows that the proposed algorithms work in real time even for a very large number of cause nodes. The results also indicate that the first newly proposed algorithm generates a larger test case suite, whereas the second newly proposed algorithm generates a smaller test case subset than the originally proposed approaches while ensuring the maximum effect coverage, fault detection rate effectiveness and a better test effect coverage ratio.