• Proceedings of the KIEE Conference
• /
• 2003.11c
• /
• pp.818-822
• /
• 2003

We propose a software-reliability growth model incoporating the amount of uniform and Weibull testing efforts during the software testing phase in this paper. The time-dependent behavior of testing effort is described by uniform and Weibull curves. Assuming that the error detection rate to the amount of testing effort spent during the testing phase is proportional to the current error content, the model is formulated by a nonhomogeneous Poisson process. Using this model the method of data analysis for software reliability measurement is developed. The optimum release time is determined by considering how the initial reliability R(x|0) would be. The conditions are $R(x|0)>R_o$, $R_o>R(x|0)>R_o^d$ and $R(x|0)<R_o^d$ for uniform testing efforts. Ideal case is $R_o>R(x|0)>R_o^d$. Likewise, it is $R(x|0){\geq}R_o$, $R_o>R(x|0)>R_o^{\frac{1}{g}$ and $R(x\mid0)<R_o^{\frac{1}{g}}$ for Weibull testing efforts. Ideal case is $R_o>R(x|0)>R_o^{\frac{1}{g}}$.

• Proceedings of the Korean Reliability Society Conference
• /
• 2002.06a
• /
• pp.273-273
• /
• 2002

In this paper, we investigate a software release policy with software reliability growth factor during the warranty period by assuming that the software reliability growth is assumed to occur after the testing phase with probability p and the software reliability growth is not assumed to occur after the testing phase with probability 1-p. The optimal release policy to minimize the expected total software cost is discussed. Numerical examples are shown to illustrate the results of the optimal policy. And we consider a Bayesian decision theoretic approach to determine an optimal software release policy. This approach enables us to update our uncertainty when determining optimal software release time, When the failure time is Weibull distribution with uncertain parameters, a bayesian approach is established. Finally, numerical examples are presented for illustrative propose.

• International Journal of Reliability and Applications
• /
• v.4 no.1
• /
• pp.1-12
• /
• 2003

Many software reliability growth models (SRGM's) based on a nonhomogeneous Poisson process (NHPP) have been proposed by many researchers. Most of the SRGM's which have been proposed up to the present treat the event of software fault-detection in the testing and operational phases as a counting process. However, if the size of the software system is large, the number of software faults detected during the testing phase becomes large, and the change of the number of faults which are detected and removed through debugging activities becomes sufficiently small compared with the initial fault content at the beginning of the testing phase. Therefore, in such a situation, we can model the software fault-detection process as a stochastic process with a continuous state space. In this paper, we propose a new software reliability growth model describing the fault-detection process by applying a mathematical technique of stochastic differential equations of an Ito type. We also compare our model with the existing SRGM's in terms of goodness-of-fit for actual data sets.

• Journal of Information Technology Applications and Management
• /
• v.11 no.4
• /
• pp.87-94
• /
• 2004

The software reliability growth model(SRGM) has been developed in order to estimate such reliability measures as remaining fault number, failure rate and reliability for the developing stage software. Almost of them assumed that the faults detected during testing were evetually removed. Namely, they have studied SRGM based on the assumption that the faults detected during testing were perfectly removed. The fault removing efficiency. however. IS imperfect and it is widely known as so in general. It is very difficult to remove detected fault perfectly because the fault detecting is not easy and new error may be introduced during debugging and correcting. Therefore, the fault detecting efficiency may influence the SRGM or cost of developing software. It is a very useful measure for the developing software. much helpful for the developer to evaluate the debugging efficiency, and, moreover, help to additional workloads necessary. Therefore. it is very important to evaluate the effect of imperfect dubugging in point of SRGM and cost. and may influence the optimal release time and operational budget. I extent and study the generally used reliability and cost models to the imperfect debugging range in this paper.

• Journal of Information Technology Application
• /
• v.3 no.3
• /
• pp.91-106
• /
• 2001

We propose a software-reliability growth model incoporating the amount of uniform and Weibull testing efforts during the software testing phase in this paper. The time-dependent behavior of testing effort is described by uniform and Weibull curves. Assuming that the error detection rate to the amount of testing effort spent during the testing phase is proportional to the current error content, the model is formulated by a nonhomogeneous Poisson process. Using this model the method the data analysis for software reliability measurement is developed. The optimum release time is determined by considering how the initial reliability R($\chi$ 0) would be. The conditions are ($R\chi$ 0)＞$R_{o}$ , $P_{o}$ ＞R($\chi$ 0)＞ $R_{o}$ $^{d}$ and R($\chi$ 0)＜$R_{o}$ $^{d}$ for uniform testing efforts. deal case is $P_{o}$ ＞($R\chi$ 0)＞ $R_{o}$ $^{d}$ Likewise, it is ($R\chi$ 0)$\geq$$R_{o}$ , $R_{o}$ ＞($R\chi$ 0)＞R(eqation omitted) and ($R\chi$ 0)＜R(eqation omitted)for Weibull testing efforts. Ideal case is $R_{o}$ ＞ R($\chi$ 0)＞ R(eqation omitted).

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.5
• /
• pp.1419-1425
• /
• 2000

When we predict of software reliability, we can't know the testing stopping time and how many faults be residues in software the (the maximum value of data) during these software testing process, therefore we assume the maximum value and the training result can be inaccuracy. In this paper, we present neural network approach for software reliability prediction with unnormalized (actual or original collected) data. This approach is not consider the maximum value of data and possible use the network without normalizing but the predictive accuracy is better. Also, the unnormalized method shows better predictive accuracy than the normalized method given by maximum value. Therefore, we can make the best use of this model in software reliability prediction using unnormalized data.

• Journal of Korea Society of Industrial Information Systems
• /
• v.14 no.5
• /
• pp.61-73
• /
• 2009

Many software reliability growth models(SRGM) have been proposed since the software reliability issue was raised in 1972. The technology to estimate and grow the reliability of developing S/W to target value during testing phase were developed using them. Most of these propositions assumed the S/W debugging testing efforts be constant or even did not consider them. A few papers were presented as the software reliability evaluation considering the testing effort was important afterwards. The testing effort forms which have been presented by this kind of papers were exponential, Rayleigh, Weibull, or Logistic functions, and one of these 4 types was used as a testing effort function depending on the S/W developing circumstances. I propose the methology to evaluate the SRGM using least square estimater and maximum likelihood estimater for those 4 functions, and then examine parameters applying actual data adopted from real field test of developing S/W.

• Journal of the Korea Computer Industry Society
• /
• v.3 no.7
• /
• pp.837-844
• /
• 2002

In general, software test is carried out to detect or repair errors in system during software development process. Namely, we can evaluate software reliability through collecting and removing the faults detected in testing phase. Software reliability growth model evaluates reliability of software mathematically. Many kinds of software reliability growth modeling which modeling the processes of detecting, revising and removing the faults detected in testing phase have been proposed in many ways. and, it is assumed that almost of these modeling have one typed detect and show the uniformed detection rate. In this study, significance levels of the faults detected in test phase are classified according to how they can affect on the whole system and then the fault detection capability of them is applied. From this point of view, We here by propose a software reliability growth model with faults detection capability according considering fault significance levels and apply some fault data to this proposed model and finally verify its validity by comparing and estimating with the existing modeling.

• Journal of the Korean Society for Railway
• /
• v.8 no.6 s.31
• /
• pp.597-601
• /
• 2005

Korea Railroad Research Institute(KRRI) developed the Driverless Rubber Tired Korea-AGT(Model: K-AGT) from 1999 to 2004. We have finished the safety and performance tests of K-AGT. Data obtained from this testing can be used to evaluate the growth of reliability. The most widely used traditional growth tracking model is included as IEC International standard. With the tracking model all corrective actions are incorporated during test, called test-fix-test. In the test-fix-test strategy problem modes are found during testing and corrective actions for these problems are incorporated during testing. In this paper, we demonstrated reliability analysis using growth model of driverless rubber tired K-AGT system to prove reliability of development system. Therefore, we introduce the well-known NHPP model and analyze a reliability growth using ReliaSoft's RGA software.

• The KIPS Transactions:PartD
• /
• v.9D no.4
• /
• pp.659-666
• /
• 2002

Park, Seo and Kim (12) developed the input domain-based SRGM, which was able to quantitatively assess the reliability of a software system during the testing and operational phases. They assumed perfect debugging during testing and debugging phase. To make this input domain-based SRGM more realistic, this assumption should be relaxed. In this paper we generalize the input domain-based SRGM under imperfect debugging. Then its statistical characteristics are investigated.