Multi-Objective Genetic Algorithm Using Class-Based Elitist Approach

Abstract

An approach named Class -Based Elitist Genetic Algorithm is presented in this paper. The test data is being generated for object-oriented programs using evolutionary techniques. A class -based algorithm derived from class control-flow graph (CCFG) is used fo r testing object-oriented software. Evolutionary techniques have been used for solving most of the software engineering problems. Evolutionary testing technique which is based on theory of evolution like reproduction, mutation, recombination, and selection is used to generate test cases in CBEGA. Evolutionary Algorithm applies these techniques repeatedly toa set of individuals called population to obtain optimal solution in a global search space with minimum search time. Multi-objective optimization involves optimizing a number of objectives simultaneously like time, cost and fault detection capability. The objectives considered here for optimization are maximum path coverage, minimum test suite size, and minimum execution time. For experiments, the data are taken from Siemens’ test suite which shows better path coverage results like 9 6% in CBEGA and are co mpared with simple GA which gives only 88% path coverage for a set of sample java classes.