Software stability analysis based on change impact simulation

It's difficult to evaluate the software stability because of the complexity of software and the uncertainty of future change requirements. In this paper, various change requirements are regarded as the combination of a series of "atomic change requirement". The modification of software, which is used to satisfy the "atomic change requirement", is regarded as: firstly, modify a randomly selected "initial element"; secondly, a ripple effect caused by the change of it. Then we proposed a software stability evaluation method based on change propagation simulation. A change propagation model and a set of change impact metrics are defined. In order to simplify the calculation of the metrics, simulation technology is introduced into software stability evaluation instead of the conditional probability calculation. The experiments result indicates that decreasing the propagation probability or improving software architecture could effectively resist the happening of broad ripple effect and enhance software stability. And the connections between in-degree hubs result in broad change. The existence of a mount of successors of in-degree hubs results in the frequent happening of broad changes.