Research On Rule-based Team Software Process Simulation Modeling

In recent years, people have gradually realized the decisive action of software process to the quality of software product. Therefore, software production enters into a software engineering era which focuses on software process improvement. People have bring forward a lot of software process methods that more suitable to guide modern software development project. At present the representative method is Capability Maturity Model(CMM)/Personal Software Process(PSP)/Team Software Process(TSP) software improvement system designed by USA Carnegie Melton University Software Engineering Institute (CMU/SEI).However, software process improvement is a comprehensive activity. It needs long-term practice and involves methods of software process engineering and key technologies of software process field. The effectiveness and reliability of software process frame is obtained after the software project completed, this kind of lag reduces TSP efficiency consumedly. Therefore, in order to obtain the data of project plan, project track, progress management, quality control, resource assignment and so on before software project developed, and analyzes the effect of TSP implementation in advance, establishing simulation model for TSP is an important theory and practical application subject. The significance of this subject is to improve the software project management and control capability and boost the development of software industry.In this dissertation the main work includes:1. A simulation model based on rule of TSP is proposed. Considering TSP is constituted by development activity, activity is one of the most basic ingredients of software process. Therefore, the TSP simulation model defined in this dissertation is a activity-centered process model. Also, considering modeling method based on rule provides active dynamic link mechanism and naturally describes the unpredictable character of process, which makes people control process flexibly, so in this dissertation modeling method based on rule is used.2. Dynamic programming algorithm is used to implement model. Considering simulation model should provide reliable support to project management decision-making, and the results of simulation model should satisfy the request of user to the software product. But the solution to the question must make a good balance between both of them, this instance is just fit to use dynamic programming for implementation.3. A concrete example to validate simulation model is given. Through the example, it is testified that this model may both understand TSP qualitatively and trace software development process quantitatively. It proves that it may help to analyze factors such as person allotment, work progress, process control and policy-making support, which provide the effective way to forecast and analyze software process.The simulation model proposed in this dissertation solves the problem of forecasting software process that can make developers carry through process management and resources assignment more reasonably. It is fit to guide and control software process implementation, and may be used as a scientific method to improve software process quality.