Technical Research Of Cost Estimation And Quality Assurance In Software Maintenance

Over the past a few decades, software cost estimation has been the focus of software engineering research. With the development of software engineering, more and more attention has been paid on software quality. In the mean while, as the evolution of software development, maintaining software gradually becomes much more important in software industry. The research on cost estimation and quality assurance in software maintenance is significant for guiding contemporary software management. However, there is little related known research in this area.The main topic of this paper is the research on cost estimation and quality assurance of maintain software. The research is based on the scenario of software maintenance, which includes the analysis on the overall project cost, the testing cost, the uncertainty measurement and the strategies in testing process. New cost estimation models are created and new technologies are used for estimating cost and improving strategies of testing process. The major contributions of this paper include:●Proposing an extended use case points model which is based on Bayesian network to address the problem of cost estimation at early stage of development. As a result of apply the new model, integrating expert judgments and uncertainties into use case points model by Bayesian network, gives the distribution of cost estimation, which can be used to help to human resource allocation.●Proposing a new model for measuring the uncertainty in testing cost. The model is used to predict software testing cost deviation. Risk in testing is classified into five categories. A Bayesian model is built for predicting testing cost deviation. The deviation distribution, generated can be used as a basis to adjust the buffer time of projects.●Proposing a procedure to revise the original testing cost estimation. This procedure which is based the VaR(value at risk) model, integrates with ratings from experts to propose a confidence level, and then use the confidence level to adjust the original cost estimation according to the result of uncertainty measurement. This procedure can improve the estimation of testing cost.●Proposing a dynamic re-assignment strategy and a minimum cost release strategy for removing defect. The reasonable assumptions are based on the real projects to simulate defect occurrence and removal process. The dynamic re-assignment strategy is for improving the efficiency of defect removal and the minimum cost release strategy is for decreasing the cost of defect removal. These two strategies can help the project manager make better decisions.The case studies show that the proposed cost estimation models have better applicability and accuracy than the previous methods, and the strategies in quality assurance are great innovations.