Automatic data flow class testing based on 2-step heterogeneous process using evolutionary algorithms

Abstract

Software testing is a systematic process to identify the presence of errors in the developed software performed using test data. Manually generating test data is ineffective in terms of cost, time and code coverage. Past three decades automation of test data generation has been a research problem of interest and a wide range of work has been done to apply evolutionary meta-heuristics. Much work in past is devoted to apply evolutionary algorithms for data flow testing on procedural programs. Generating test cases of testing classes is more challenging. Presented work identifies these challenges and proposes a systematic data flow class testing based on 2-step heterogeneous process using Genetic Algorithm (GA) and Particle Swarm Optimization (PSO) evolutionary techniques. A prototype is systematically implemented and explained on an example class. A set of classes are further tested to study efficiency of the proposed work in terms of coverage percentage and execution time.