A SPC-Based Comprehensive Model For Quantitative Software Project Control And Its Implementation

As a result of the development of software industry and software project management, quantitative software project control is demanded. In recent years, there are a lot of research activities and practices in this field, and many achievements have been made. It is well worth exploring how to employ those research results to implement quantitative software project control.Software process is the object of quantitative software project control. The characteristics and the improvement model of software process are demonstrated via analyzing process model and software process maturity. After analyzing the duty of project management and criteria for selecting process parameters to be controlled, a basic parameter set is determined according to results from software engineering research. Based on the analysis of statistical process control and COCOMO, a comprehensive model for quantitative software project control is developed, and its composition and inner relationship is detailed. In addition, the functions of software process, statistical process control, and estimation models are analyzed. Lastly, the requirement, design, and technical plan of a real system are used to explain the implementation of this model.The comprehensive model for quantitative software project control includes four parts. They are process disintegration, estimation & planning, measure & monitoring, and analysis & adjust. Software process model and statistical process control are used to guide macro and micro process disintegration respectively. Accordingly, estimation models and statistical process control are used to estimate software size, effort, schedule, defects, and their macro and micro distribution. According to the estimation, a plan can be devised. In the section of measure & monitoring, the principles of measuring and supervising are discussed, and specific practices are defined. In the section of analysis & adjust, the methods that can be used to detect and locate process exceptions by analyzing the data gathered in the process are discussed, and suggestions to prevent exceptions are provided. The structure and composition of the model are designed by combining the idea of the PDCA cycle and the need of quantitative software project control.As a classic quantitative control technique, statistical process control has been widely recognized by software industry for its effectiveness. However, its collaboration with software engineering can be improved by employing results from software engineering research. Therefore, establishing a comprehensive model for quantitative software project control can take advantage of statistics process control and software engineering, and thus achieve the best possible result.