Research On Hybrid Simulation Model And Application Of Software Process

Recent years, as the development of IT industry, people begin to recognize that scientific software process management is the important method for the improvement of software productivity and the assurance of software quality. The software process simulation problem stems from the issue above is increasingly being concerned by the academic and the IT industry.This dissertation describes the hybrid simulation model concerning one kind of software process. It is the hybrid model that combines the discrete, continuous and analytical method according to the characteristics of software process.This dissertation firstly studies the The main parts of software process; analyses and compares several popular software process models. Software cost model and risk management are introduced, especially the main theory and application of COCOMOII are emphasized, preparing for the building of simulation model.Then the reason, content and method of software process simulation are introduced. Two modelling method are compared:discrete event simulation and continuous simulation. These two methods discribe the software process from different aspects. The continuous model is good at obtaining the detailed activities and progress of the process; discrete event model is feedback-oriented, expert in solving high-level problem, but can not be used for analysing the details of the process. Therefore, the best way for this problem is to combine the two methods, reflecting software process from different aspects and different levels.Next, the hybrid simulation model of software process is builded:on the base of waterfall model, the software process is abstracted as two level model. The outer level, simulates the activities, interaction between activities and the exchanged artifacts with dicrete event queuing network; the inner level, adopts the analytical method and system dynamics, building the dynamic COCOMOII model, in the mean time, combines the progress, cost and risk model, supporting the decision making in the aspects of progress, cost and risk. The simulation of the hybrid model is introduced, the validation is done to suggest the credibility of the model.At last, this hybrid simulation model is applied to a software development program, suggesting the rationality and feasibility.