Application of generalized stochastic Petri net to quantitative evaluation of software process

As the complexity and size of software increases, project management becomes important to develop reliable software cost-effectively within specified time constraints. In order to manage a software project effectively, it is necessary to establish a control method based on the objective and quantitative evaluation data on the current status of the software project. Furusawa (1994) presented quantitative evaluations of the software development process described by a generalized stochastic Petri net. However, the evaluation in the above paper cannot deal with the development period sufficiently because of restrictions on the GSPN model. This paper defines a new hierarchical model for software projects by extending the previous GSPN model and by introducing the concept of workload, and proposes a method to evaluate the software process from the viewpoints of the quality, cost and delivery date. As a result, the new model can take the influence of human factors into account, and thus can evaluate the dynamic aspects of software projects. Finally, the results of a case study of the proposed method provide us with some implications on software project management.