Research On NHPP Software Reliability Growth Models Of Fault Detection And Correction Processes

As the development of the computer technology, software systems are widely applied in various fields. People pay more and more attention to the issue of software quality. Software reliability is the most important content among the quality attributes of software, and the software reliability growth models (SRGMs) are the main methods for evaluating and predicting software reliability by means of statistics. Among various SRGMs, the SRGM of the type of non-homogeneous Possion process (NHPP) is very important and widely used. It has been a quite successful tool in practical software reliability engineering. The fault detection and correction are the two main processes in the software testing. Most existing SRGMs only consider the software fault detection process in the testing stage, assuming immediate fault correction activity. However in practice, each detected fault is reported, located and then corrected. The software fault correction process plays a significant role to improve software reliability and ensure software quality. Thus, the fault correction process can not be simply treated as an immediately achieved process. In addition to predicting and assessing software failure occurrence, it is also meaningful for analyzing and modeling the software fault correction. Therefore, the fault correction process can not be simply ignored in the software reliability modeling. This paper studies the correction lag between the fault detection and correction processes. Then, the the fault detection and correction processes SRGMs are proposed. Furthermore, the optimal release problem is also discussed based on these SRGMs.Since the fault detection process and the fault correction process can be related from the aspect of the testing time, most existing models of these two processes are from this viewpoint. However, there are many kinds of time delay functions in these SRGMs which are hypothesized by the researchers. It is very difficult for the test manager to select the most appropriate model. This paper discusses the existing types of the time delay functions. Then, the variation of time delay is analyzed and concluded in three kinds. A new time delay function is proposed which is more general and can characterize all three kinds of time delay circumstance. Based on the new time delay function, the fault detection and correction process SRGM is proposed which named BTD-SRGM. The experimental results on two real data sets show that BTD-SRGM has better goodness of fit and predictive ability than other SRGMs. Moreover, the new model can also estimate the varying trend of the time delay accurately.Compared with the fault correction time, the fault number is easier to be recorded. As a fault must be corrected and removed after its detection, the number of corrected faults is naturally depended on the number of detected faults. The number of remaining uncorrected faults on some real data sets is investigated and found to have an"increasing and then decreasing"feature. The Gamma curve is used to characterize the varying trend of the remaining uncorrected faults. Based on this lag function, we propose a new SRGM which named BCL-SRGM, and the fault detection and correction processes can be successfully modeled. Numerical examples are presented and the comparison results show that the proposed model not only has better performance than other SRGMs on both fault detection and correction processes, but also can characterize the correction lag exactly.The imperfect debugging is the main reason why the fault correction process needs to be departed from the fault detection process in the software reliability modeling. The fault removal efficiency is an important factor in the SRGMs which consider imperfect debugging. Through analyzing some real data sets, the fault removal efficiency reflects the relationship between the fault detection and the fault correction from the aspect of the fault number, and the logistic function is found to be appropriate to characterize it. Based on the fault removal efficiency function, a new fault detection and correction processes model is proposed which named RDC-SRGM. This model not only can be used in evaluating the fault detection and correction processes, but also can provide the mean value of the fault removal efficiency which reflects the effectiveness of the testing.The correction lag is not considered in the traditional software cost model which can not derive the actual software optimal release time. Based on the BCL-SRGM, the effect of correction lag can be introduced into the software cost model. Then, the effect of the imperfect debugging on the software cost is also studied. A new software cost model which has considered the imperfect debugging is proposed. By using the RDC-SRGM, the correction lag is also integrated into the new software cost model which has considered the imperfect debugging phenomenon. Finally, the cost model which integrated correction lag and the imperfect debugging is also validated on a real data set.