Parameter Estimation of Software Reliability Growth Models by A Modified Whale Optimization Algorithm

Kezhong Lu,Zongmin Ma
DOI: https://doi.org/10.1109/dcabes.2018.00076
2021-01-01
Journal of Algorithms & Computational Technology
Abstract:Software reliability growth models (SRGMs) are non-linear in nature, so they are difficult to estimate the proper parameters. An estimation method based on modified Whale Optimization Algorithm (MWOA) in which parameters are estimated is discussed in this paper. The proposed MWOA shows significant advantages in handling variety of modeling problems such as the exponential model (EXPM), power model (POWM) and delayed S-shaped model (DSSM). However, the fitting error of a single model is relatively large. In view of the different growth characteristics at each stage of the model, a three-stage software reliability growth model is proposed in this paper. MWOA is used to estimate the parameters of three-stage software reliability models. Experimental results show that the fitting accuracy of three-stage model is significantly less than that of a single model. MWOA with the three-stage model can provide a better estimate of the software faults.
What problem does this paper attempt to address?