Network Model Construction and Cascading Effect Analysis for Software Systems

As the functional requirements of software keep growing, the structure and scale of software systems become more and more complicated. In order to analyze the topology and quality of complex software systems, the theory of complex networks has been introduced to model and solve software engineering problems. This paper regards functions in the source code of the software as nodes, function-calls in the source code of the software as directed edges, and defines the weight of edges, then presents a method of constructing the weighted software dynamic execution network model. The results on the statistical analysis of the networks obtained from three software programs, TAR, GEDIT and EMACS, show that the weighted networks of the software execution process fit in with the small-world effect and the scale-free property of complex networks. Based on that, we further take advantage of the CML model in complex networks to simulate and analyze the cascading effect for software systems and discover the main factors that influence the final scale and spread speed of cascading failures in software systems, which will give an important support for the research of software quality assurance.