Development and validation of nature inspired test case selection and prioritization techniques

Abstract

Software testing accomplished to maintain one s confidence in the software after updates is referred to as Regression Testing [1]. Researchers have been rigorously working towards the development and validation of various regression testing techniques. Regression Test Prioritization and Regression Test Selection are the key regression testing techniques focused in this thesis. Regression test selection and prioritization activities when performed in a time-constrained environment, tend to become NP complete combinatorial optimization problems [2]. The way God has created the nature and its living beings is astonishing. Numerous nature-inspired approaches have already been used in solving test case selection and prioritization problem [3] [4]. This thesis contributes to the field of regression test selection and prioritization by improvising the existing algorithms and proposing new algorithms that execute within a time constrained environment. Experiments and analysis have been done using C++ and Python. The existing Ant Colony Optimization approach for selecting and prioritizing test cases [5] was analysed and it was observed that the existing technique had limitations like converging to local optimum solution [6, 7]. Thus, an improved ACO_I_TCSP algorithm [8] has been proposed and implemented. The experiment for its comparison with old technique proves the increased cost effectiveness of the algorithm and also the increased correctness. Results obtained by the comparison of ACO with other reference techniques using Average Percentage of Faults Detected (APFD) metrics, highlight close proximity to optimal solution and show the time reduction achieved by the technique