International Journal of Reliability, Quality and Safety Engineering

Service Reliability is an important consideration for new service deployment. Traditional system-oriented measures are no longer adequate to describe the reliability perceived by the user. In this paper we propose a general service reliability analysis approach based on user behavior, and derive formulas to compute service reliability from the user model and service models. The derived user-perceived service reliability incorporates the user task reliabilities, the dependencies of different user tasks, and various types of user behavior besides failure and recovery of system hardware and software components. This approach is applied to the service reliability computation for an example fault tolerant cluster hosing two services. Factors from both the system side and the user side are analyzed for the user-perceived service reliability of the cluster, and the results are compared with the system availability and service reliability lower bound.

The classical Gumbel probability distribution is modified in order to study the failure times of a given system. Bayesian estimates of the reliability function under five different parametric priors and the square error loss are studied. The Bayesian reliability estimate under the non-parametric kernel density prior is compared with those under the parametric priors and numerical computations are given to study their effectiveness.

Improving power of goodness-of-fit tests is an important research topic in statistics. The goal of the goodness-of-fit test is to check whether the underlying probability distribution, from which a sample is drawn, differs from a hypothesized distribution. Numerous research papers have been published in this area. It has been shown that the power of the existing goodness-of-fit tests in the literature is unsatisfactory when the alternative distributions are of V-shape or when the sample sizes are small. This motivates the development of more powerful test statistics. In this research, a new test statistic is proposed. The result can be used to test whether the underlying probability distribution differs from a uniform distribution. By applying the probability integral transformation, the proposed test statistic can be used to check whether the underlying distribution differs from any hypothesized distribution. The performance of the method proposed in this research is compared with the Kolmogorov–Smirnov test, which is a widely adopted statistical test in the literature. It has been shown that the test proposed in this proposal is more powerful than the Kolmogorov–Smirnov test in some cases.

This paper discusses an new approach for discrete-time software reliability growth modeling based on an discrete-time infinite server queueing model, which describes a debugging process in a testing phase. Our approach enables us to develop discrete-time software reliability growth models (SRGMs) which could not be developed under conventional discrete-time modeling approaches. This paper also discuss goodness-of-fit comparisons of our discrete-time SRGMs with conventional continuous-time SRGMs in terms of the criterion of the mean squared errors, and show numerical examples for software reliability analysis of our models by using actual data.

As a result of the technological progress, software development environment has changed into development paradigm based on client/server systems by using network computing technologies. Network technologies have made rapid progress with the dissemination of computer systems in all areas. These network technologies become increasingly more complex in a wide sphere. Especially, open source software systems which serve as key components of critical infrastructures in the society are still ever-expanding now.

In this paper, we propose a method of software reliability assessment based on stochastic differential equations. Especially, we derive several assessment measures in terms of imperfect debugging. Also, we analyze actual software fault-count data to show numerical examples of software reliability assessment for an embedded open source software. Further, it has been necessary to manage the software development process in terms of reliability, effort, and release time. Then, we find the optimal release time based on the total expected software maintenance effort.