Parameter estimation of software reliability growth models using hybrid genetic algorithm

Software reliability is a quantifiable metric, which is defined as the probability of software to operate without failure for a particular period of time in a specific environment. Various SRGMs (software reliability growth models) have been proposed to predict the reliability of software. The reliability is predicted by calculating the attributes of the software reliability growth models. But the parameters of SRGMs are generally in nonlinear relationships, which is a great problem in finding the optimal parameters by using traditional ways or the techniques for attribute calculation like Maximum Likelihood and least Square Estimation. The following paper is proposing a new approach for calculating the verticals or parameters of SRGM using a hybrid genetic algorithm. This algorithm is the hybridization of GA (Genetic Algorithm) which is real valued g and PSO (particle swarm optimization). Each chromosome is defined as a group of real values in RGA and then the operators of the real valued genetic algorithm are used in directly modifying these chromosomes. While particle swarm optimization is different optimization technique which is an alternative to genetic algorithm because of its simplicity and equal accuracy. The approach which is proposed has several benefits over conventional GA in the vertical or calculations of verticals of SRGM.