Empirical Data and Analysis of Developing New Simulation Model of Software Evolution Process

Abstract:

This paper presents a new system dynamic (SD) simulation model for software evolution process. Previous SD models have been based on actor-network theory (ANT) to develop a general purpose simulation environment. The model presented in this  paper utilizes agent based simulation environment (Repast). The new model will be compared with the original ANT based model.The Repast model simulation model presented in this paper shows more stable behaviour compared with the increasing oscillations behaviour of the ANT based SD simulation model. Also it shows that the evolution health of the system can be calibrated quantitatively. The results shows that the new Repast simulation model accurately presents real-world processes of software evolution given that the particular impact of a participant on the health of system is accurately drawn. The results in this paper will be of interest to the software development and evolution research community, and contribute to improving the understanding of the software evolution process and the factors that influence its behaviour.