International Journal of Reliability, Quality and Safety Engineering

Most of the previous studies on developing an optimal burn-in policy for semiconductor products only deal with the burn-in process itself and little is concerned with utilizing the information on the quality levels of chips before being subjected to burn-in. Developed in this paper is a dual burn-in policy in which the number of chips (d) which do not pass the wafer probe (WP) test and lie in the neighborhood of a reference chip is utilized as an indicator on the quality level of that reference chip. The dual burn-in policy first classifies the chips which pass the WP test into two groups using a boundary value of d, and then each group is subject to burn-in for its own duration. For a certain type of 256M DRAM product, the performance of the proposed dual burn-in policy is compared to that of the single burn-in policy in which all chips are subjected to the burn-in of the same duration without considering d. The analysis results show that, for the cases considered, the proposed dual burn-in policy is more cost-effective than the single burn-in policy, implying that the additional information from the WP test is beneficial to establishing an efficient burn-in policy in semiconductor manufacturing.

Our overall objective is to develop a multilayered model that can identify system vulnerabilities and detect anomalous system behavior. Once these model components have been developed, we could link the resulting multilayered SCADA systems (e.g., substation networks linked by the Internet) to determine system interactions, and then test, refine, and validate these models and data through increasingly sophisticated experiments to discover and record additional dependencies and system interactions. What do we want to trust? We want to have confidence that the system is available and reliable to deliver power to the utility's customers as their needs demand. Furthermore, we want a secure system that is not vulnerable to cyber attacks. Why might we have lack of trust in critical infrastructure systems? External threats to the electrical power infrastructure are increasing over internal threats. One reason is the increasing use of the Internet for connecting substation networks to other parts of the electric grid.

We consider systems whose time–dependent behaviour can be described by a Markov chain with a finite state space. There is a finite set of subsets of the state space given and every point in the time interval under consideration is associated with one of these subsets, thereby definig a demand (or usage) pattern. Closed form expressions are derived for the probability of a demand pattern being satisfied, in both, discrete and continuous time. Known reliability measures are identified as special cases. We illustrate the theory on two examples: the first is a system comprising three power transmission lines, the second is a small computer system consisting of four units, each in one of the states up or down. The demand patterns cover a four–week period.

Software rejuvenation is a preventive and proactive solution that is particularly useful for counteracting the phenomenon of software aging. In this article, we consider the similar periodic software rejuvenation model to Garg et al.¹³ under the different operation circumstance. That is, we model the stochastic behavior of telecommunication billing applications by using a discrete-time Markov regenerative process, and determine the optimal periodic software rejuvenation schedule maximizing the so-called cost effectiveness, in discrete-time setting. Also, we provide a statistically non-parametric method to estimate the optimal software rejuvenation schedule, based on the discrete total time on test concept. Numerical examples are devoted to illustrate the determination/estimation of the optimal software rejuvenation schedule and to examine the asymptotic behavior of the estimator developed here.

In this paper, we investigate a dynamic software quality model that incorporates software process and software product measures as covariates. Furthermore, the model is not based on execution time between failures. Instead, the method relies on data commonly available from simple problem tracking and source code control systems. Fault counts, testing effort, and code churn measures are collected from each build during the system test phase of a large telecommunications software system. We use this data to predict the number of faults to expect from one build to the next. The technique we use is called time series analysis and forecasting. The methodology assumes that future predictions are based on the history of past failures and related covariates. We show that the quality model incorporating testing effort as a covariate is better than the quality model derived from fault counts alone.