Software Optimal Release Time Decision-making Research

The rapid development of IT industry made the software plays an important role in military, astronautics, and our daily life. However, software product inevitably has defects or faults which will lead to software failure when released to the market. The only way to assure a sufficient reliable software product after coding process is through constant testing. Then a problem arises on the testing phase during the software development process: when to stop testing software and release it to the market. An earlier release of the software product may reduce the development cost, but increase the maintenance cost and may incur damage to company's reputation. A delay in the release may cause a loss of market share which finally leads to reduce profit of the software product. Thus it is of great importance to decide an optimal software release time, and the determination of software release time should consider both software reliability aspect and testing cost aspect.The traditional research on optimal software release time problem focuses on the software reliability model and testing cost model and make sure the model fits the practical more accurately. So far, only the software reliability growth model exist more than one hundred. However, the application of optimal software release time is still not optimistic. Therefore, in this paper, we focus on the optimization model, instead of software reliability model and testing cost model. The main work of this paper can be summarized as follows:Firstly, we analyze the uncertainty in the software testing cost model. The testing cost is in natural a random variable, and the traditional optimization model only considers the expectation aspect. However, the variation of a random variable should not be ignored. Therefore, in this paper, we propose three risk functions, and three optimization models, and with abundant numerical example to illustrate the uncertainty and its influence on optimal software release time.Secondly, the traditional optimization models are single objective, and it is very difficult for the decision-maker the set appropriate optimization parameters. In this paper, we propose a fuzzy multi-objective optimization which reflects the decision maker's preference through fuzzy set and optimizes the expected testing cost and reliability simultaneously. In addition, we also propose a weight combination method to for decision maker to reflecting the relative importance of two objectives. We also derive the sufficient and necessary condition for the existence of optimal solution.