An Enhancement of Coverage-Based Test Case Prioritization Technique Using Hybrid Genetic Algorithm

Abstract

Test Case Prioritization plays a defining role in regression testing by optimizing the ordering of test case execution for early fault detection. This ensures that modifications to the code do not adversely affect existing functionalities. TCP is considered an effective approach in regression testing that optimizes test execution by ordering test cases according to a criterion.  While previous research works are based on a single criterion using Genetic or Cuckoo search to solve optimization problems in TCP, these single metrics limit their effectiveness. This study proposes a novel hybrid evolutionary algorithm approach that addresses this limitation by significantly improving the Average Percentage Fault Detection (APFD) rates in regression testing. Our adopted approach enhances the strengths of Cuckoo Search Optimization (CSO) to overcome the limitations of premature convergence in GA by parameter tuning helps to overcome the limitations of single algorithm solutions. We evaluate the performance by incorporating APFD and CE for improvement in the rate of fault detection and an increase in coverage effectiveness for a more comprehensive test suite application. Statistical evaluation using ANOVA strengthens our resolve for the adopted approach, with significant results tabulated. Experimental evaluations on Siemens Test Suite datasets demonstrate how our hybrid approach achieves an improvement in APFD of 0.271% over both GA and CSO and an improvement in CE of 6.35% over GA and 6.62%. over CSO across all test suite datasets. These findings highlight the applicability of a well-thought-out hybridized evolutionary algorithm that can be used in TCP to advance software testing practices.