A novel strategy for automatic test data generation using soft computing technique

Abstract

Software testing is one of the most crucial and analytical aspect to assure that developed software meets prescribed quality standards. Software development process invests at least 50% of the total cost in software testing process. Optimum and efficacious test data design of software is an important and challenging activity due to the nonlinear structure of software. Moreover, test case type and scope determines the quality of test data. To address this issue, software testing tools should employ intelligence based soft computing techniques like particle swarm optimization (PSO) and genetic algorithm (GA) to generate smart and efficient test data automatically. This paper presents a hybrid PSO and GA based heuristic for automatic generation of test suites. In this paper, we described the design and implementation of the proposed strategy and evaluated our model by performing experiments with ten container classes from the Java standard library. We analyzed our algorithm statistically with test adequacy criterion as branch coverage. The performance adequacy criterion is taken as percentage coverage per unit time and percentage of faults detected by the generated test data. We have compared our work with the heuristic based upon GA, PSO, existing hybrid strategies based on GA and PSO and memetic algorithm. The results showed that the test case generation is efficient in our work.

A Hybrid Particle Swarm Optimization and Differential Evolution based Test Data Generation Algorithm for Data-Flow Coverage using Neighbourhood Search Strategy

“…The fitness function is derived from the work by Ghiduk et al [6]; it is augmented with branch distance to produce a smoother landscape for guiding the search and also takes into account that a definition may be killed by another definition before the associated use is reached. The performance of the proposed approach is compared with random search and earlier studies on test data generation for data-flow dependencies of a program by Girgis [7], Ghiduk et al [6] and Girgis et al [21]. The proposed GA-based approach guided by the novel fitness function outperformed random search and the earlier studies [6,7,21] to generate test data for data-flow coverage of a program.…”

“…The performance of the proposed approach is compared with random search and earlier studies on test data generation for data-flow dependencies of a program by Girgis [7], Ghiduk et al [6] and Girgis et al [21]. The proposed GA-based approach guided by the novel fitness function outperformed random search and the earlier studies [6,7,21] to generate test data for data-flow coverage of a program.…”

“…Girgis et al [21] proposed a hybrid Genetical Swarm Optimization (GSO) Technique to generate a set of test paths that cover the all-uses criterion for data-flow coverage. The authors have claimed that the set of paths generated by the proposed GSO can be passed to a test data generation tool to find program inputs that will execute them to complete the data flow paths testing of the program under test.…”

“…Hybridization of search-based algorithms for test data generation has also been reported in literature. GA with a local search algorithm [15] and more recently, GA with PSO has been applied for test data generation in some studies [16,17,18,19,20,21].…”

Framework for cloud‐based software test data generation service

“…The existing cloud‐based testing models have been analyzed and described in the work of Chana and Chawla . In our previous work, a hybrid particle swarm optimization (PSO)–genetic algorithm (GA)–based sequential automated TDG for object‐oriented programs was presented. Later, in our previous work, Apache Hadoop MapReduce and Pareto‐optimal‐based test data generation framework was devised that facilitated efficient generation of test data with better coverage and fault detection capability.…”