Research And Application On Software Reliability Growth Model Considering Imperfect Debugging

With the continuous development of computer technology,software is widely used in various fields, its scale is bigger, and its complexity is higher. Social demand for reliable software is also growing rapidly, especially software applied to the frontier and high real-time demand and its reliability is becoming more and more important. Improving software reliability has become a very important research direction of software quality assurance and evaluation activities. Software reliability growth model is the main tool for quantitative evaluation and expectation of software reliability.Software testing phase mainly includes the fault detection and fault removal process. In the fault removal process the imperfect debugging phenomenon exists, and the influence factors involved in software reliability modeling are the fault removal rate function and the fault introduction rate function. At the same time, there are change points in the process of fault detection, the fault detection rate is neither constant nor continuous changing over time,and it will change at some points. The precise modeling on the fault detection and fault removal process can predict and evaluate the software reliability better. This paper focuses on the impact of imperfect debugging and change point phenomenon on software reliability. The main research content is as follows:Firstly, the fault removal rate in the software testing phase is not 100%, and new faults will be introduced by the detected fault which remained in the software, meaning the fault removal rate and fault introduction rate exist which both are reduction function with different downward trend. In order to solve this problem, this paper puts forward a new form of the fault introduction rate function, and combines with the existing fault removal rate function to propose a new software reliability growth model considering imperfect debugging. With two sets of software failure data published, the model presented is simulated and verified, and the experimental results show that the model has better fitting effect and prediction ability.Secondly, as the existing software reliability growth model considering imperfect debugging ignores change point in the software fault detection process,a software reliability growth model based on change point considering imperfect debugging is proposed. By two groups of software failure data published validating the model, the experimental results show that the model has excellent fitting and prediction ability.Finally, the electronic chart system is introduced, and its reliability is assessed and predicted by two models proposed. Then, the problem of the optimal software release is analyzed. Aiming at the shortcomings of the existing decision, this paper chooses the method optimizing the software cost and software reliability both. The fuzzy mathematics theory is introduced to build the fuzzy set of the software cost and reliability, using the aggregation operator, and achieving the purpose of the multi-objective optimization. Combined with the failure data of the electronic chart system, the optimal release time is discussed.