International Journal of Reliability, Quality and Safety Engineering

Burn-in procedure is a manufacturing technique that is intended to eliminate early failures. In the literature, assuming that the failure rate function of the products has a bathtub shape the properties on optimal burn-in have been investigated. In this paper burn-in problem is studied under a more general assumption on the shape of the failure rate function of the products which includes the traditional bathtub shaped failure rate function as a special case. An upper bound for the optimal burn-in time is presented under the assumption of eventually increasing failure rate function. Furthermore, it is also shown that a nontrivial lower bound for the optimal burn-in time can be derived if the underlying lifetime distribution has a large initial failure rate.

We apply (n, k, m) convolutional codes in which elements in generator matrices are the ones over a finite field to Hybrid ARQ combining with retransmission mechanisms, and evaluate the number of transmitted packets and the number of transmissions. We assume star topology and independent model for a transmission model, and compare two retransmission strategies: (1) one which has been applied with Reed-Solomon (RS) code, and (2) one which choose transmitted packets considering the constraint length m. We use computer simulation for (14, 7, m) and (6, 3, m) convolutional codes to observe the effects of four parameters, that is the number of receivers, constraint length m, packet loss probability, and redundancy at initial transmission PF, to the number of transmissions and transmitted packets. Simulation results showed that the number of transmitted packets can be reduced by the retransmission mechanism which consider m. Also, a PF which minimize number of transmitted packets exists for given packet loss probability, number of receivers, and constraint length.

This paper proposes an algorithm to compute marginal reliability importance for network systems with k-terminal reliability efficiently. Marginal reliability importance is an appropriate quantitative measure on a system component against system reliability and it contributes to design of reliable systems. Computing marginal reliability importance in network systems is time-consuming due to its NP-hardness. This paper extends the algorithm proposed in our last study to deal with k-terminal reliability and incorporates an extended factoring theorem to improve the algorithm. Numerical experiments compare the proposed algorithm with a traditional method to reveal efficiency of the algorithm. The algorithm helps to construct efficient algorithm to reliable network design problems.

Firmware is embedded software in hardware devices and they play important role for many critical systems' function. Firmware failure rate in operation should be quite lower than the application software which is operating on it. Most of the study on software reliability deals with systems during development, and it is also important to study the integrated system during operation. Complex systems usually have a bathtub-shaped failure rate over the lifecycle of the product. This paper discusses the parametric analysis of model given by Haupt and Schäbe (1992), exponentiated Weibull distribution and models generated from Pareto and Weibull distribution, as well as their possible application to modeling firmware system failure. In addition, the Safety Integrity Levels (SIL) stipulated in IEC 61508 are taken into account in the modeling since the safety-critical systems in general are firmware-dominated.

A fundamental but impossible to be addressed problem in repairable system modelling is how to estimate the system repair improvement (or damage) effects because of the large-sample requirements from the standard statistical inference theory. On the other hand, repairable system operating and maintenance data are often imprecise and vague and therefore Type I fuzzy sets defined by point-wise membership functions are often used for the modelling repairable systems. However, it is more logical and natural to argue that Type II fuzzy sets defined by interval-valued membership function, called interval-valued fuzzy sets (IVFS), should be used in characterizing the underlying mechanism of repairable system. In this paper, we explore a small-sample based GM(1,1) modelling approach rooted in the grey system theory to extract the system intrinsic functioning times from the seemly lawless functioning-failure time records and thus to estimate the repair improvement (damage) effects. We further explore the role of interval-valued fuzzy sets theory in the analysis of the system underlying mechanism. We develop a framework of the GM(1,1)-IVFS mixed reliability analysis and illustrate our idea by an industrial example.

Architecture-based techniques for reliability assessment of software applications have received increased attention in the past few years due to the advent of component-based software development paradigm. Most of the prior research efforts in architecture-based analysis use the composite solution approach to solve the architecture-based models in order to estimate application reliability. Though the composite solution approach produces an accurate estimate of application reliability, it suffers from several drawbacks. The most notable drawback of the composite solution approach is that it does not allow an analysis of the sensitivity of the application reliability to the reliabilities of the components comprising the application and the application structure. The hierarchical solution approach on the other hand, has the potential of overcoming the drawbacks of the composite approach. However, in the present form, the hierarchical solution approach produces an estimate of application reliability which is only an approximation of the estimate produced by the composite approach since it does not take into consideration the second-order architectural statistics. Also, although the hierarchical solution approach can be used for sensitivity analysis, mathematical techniques to perform such analysis are lacking. Development of an accurate hierarchical solution approach to estimate application reliability based on its architecture is the focus of this paper. Using the approach described in this paper, an analytical application reliability function which incorporates second-order architectural statistics can be obtained. Sensitivity analysis techniques and expressions to determine the mean time to failure of the application are developed based on this analytical reliability function. We illustrate the reliability prediction, sensitivity analysis, and mean time to failure computation techniques presented in this paper using two case studies.