PSO-Based Test Case Generation: A Fitness Function Based on Value Combined Branch Distance

Abstract

All-path coverage of software under test is a NP-complete problem. At the time of automatic test case generation, there is some path whose probability of coverage is low. To overcome this problem, we propose a fitness function, value combined branch distance function (VCBDF) on path coverage criteria and is applied with particle swarm optimization (PSO) algorithm to automate test case generation. The objective is to achieve maximum path coverage with the challenge of covering a target path. We have conducted experiment on a bench mark case study, triangle classification problem. Two more existing fitness functions, branch distance (BD) based fitness function and combined fitness function (CFF) are also applied for comparison. Experimental result shows VCBD function gives better result in terms of number of test cases generated for target path and number of iterations as compared to the said functions.