International Journal of Reliability, Quality and Safety Engineering

The resource allocation problem has been studied in a variety of applications. This problem usually has only one constraint, i.e., the amount of resource to be allocated is constant. Considering its application areas, however, it is important to treat multiresource problems. In this paper we consider a two-resource allocation problem with an exponential objective function. Though this problem is a nonlinear programming problem, we show that it can be solved in strongly polynomial time.

For repairable systems, aging properties of the time to first failure cannot adequately reflect its degradation over time — which is influenced by repairs. For such systems, exploring notions of aging under repairs would provide a more relevant description of deterioration in performance. A framework to formulate appropriate notions of aging relative to repair strategies and examine their implications is indicated. Included among such strategies are renewals (perfect repair or, replacements) and minimal repairs, as are the repair schemes of Brown-Proschan and Block-Borges-Savits. We examine some specific notions of repair relative to aging and briefly review recent results on two aging properties under perfect repairs. The concept of efficiency of repair strategies is introduced and illustrated by simple examples. Our final results concern an induced distribution related to perfect repairs and its realizability.

Ensuring software reliability is one of the most important issues in software development. High-precision fault rate prediction is a powerful method for ensuring reliability. A variety of highly accurate methods have been proposed to date, the most widely used being software reliability growth models (SRGMs) based on how the cumulative number of faults detected varies over time. This paper presents a manifold model that unifies existing SRGMs. This model, in addition to reducing the labor in selecting the most suitable model for each growth curve, makes it possible to predict the number of remaining faults for complicated fault growth curves with higher accuracy than previously. Using the manifold model, this paper clarifies the relationships between the existing SRGMs. Then, using actual data, it compares the fault estimation accuracy of the manifold model and representative SRGMs, showing the usefulness of this model.

Using an arrangement monotonicity property of the parametrized family of hazard rate ordered random variables, and a bivariate characterization of the hazard rate ordering, we obtain some new stochastic arrangement inequalities for the random variables that are hazard rate ordered. Some applications of optimal allocation and assembly are also discussed.

This paper presents a multi-stage software design approach for fault-tolerance. In the first stage, a formalism is introduced to represent the behavior of the system by means of a set of assertions. This formalism enables an execution tree (ET) to be generated where each path from the root to the leaf is, in fact, a well-defined formula. During the automatic generation of the execution tree, properties like completeness and consistency of the set of assertions can be verified and consequently design faults can be revealed. In the second stage, the testing strategy is based on a set of WDFs. This set represents the structural deterministic test for the model of the software system and provides a framework for the generation of a functional deterministic test for the code implementation of the model. This testing strategy can reveal the implementation faults in the program code. In the third stage, the fault-tolerance of the software system against hardware failures is improved in a way such that the design and implementation features obtained from the first two stages are preserved. The proposed approach provides a high level of user-transparency by employing object-oriented principles of data encapsulation and polymorphism. The reliability of the software system against hardware failures is also evaluated. A tool, named Software Fault-Injection Tool (SFIT), is developed to estimate the reliability of a software system.

The lifetime of a system is measured in two principal time scales, L and H. We consider a family of time scales of type T_a=(1−a) · L+a · H, for which the lifetime is T=(1−a) L+aH, a∈[0, 1], where L and H are the lifetimes in the principal time scales. The optimal time scale, by the definition, provides the minimal value of the coefficient of variation of system lifetime. We consider the age replacement model in time scale T_a and develop a nonparametric numerical procedure for finding the optimal weighting parameter a* that provides the smallest value of the corresponding cost function. This procedure is applied to fatigue test data, and it is demonstrated that the best time scale with respect to the age replacement cost function is very close to the optimal time scale. The same is true for a modified age replacement scheme in which the replacement age equals the p-quantile of system lifetime.

A family of regular graphs of degree 3, called chordal rings is presented as a possible candidate for the implementation of a distributed system and for fault-tolerant architectures. The symmetry of graphs makes it possible to determine message routing by using a simple distributed algorithm. Arbitrary data permutations are generally accomplished by sorting. For certain classes of permutations, however, there exist algorithms that are more efficient than the best sorting algorithm. One such class is the Bit Permute Complement (BPC) class of permutations.

In this paper, we first develop algorithms requiring two token storage registers in each node to realize an arbitrary BPC permutation. We next evaluate its ability to realize BPC permutations in networks of arbitrary size by estimating the number of required routing steps when a single fault is present and when not.

We give recursive equations and closed form solutions for two kinds of reliability for linear and circular consecutive-2 link systems under a general reliability model where nodes and the two types of links can all fail and with different probabilities. The results can be extended to the case when the reliabilities of nodes (respectively, the two types of links) are not necessarily constant.

The effects of mental stress on fault-generation are quantitatively determined through a controlled experiment. Two teams develop the same software program under the same conditions, except for mental stress. One team is mentally stressed during the functional design phase, while the other one is not. The degree of stress is measured by two types of metrics: “inner metrics” and “outer metrics”. The inner metrics, which measure the degree of stress, show that the “stressed team” generates more faults than the “nonstressed team”. Among the inner metrics, the “schedule pressure” and “workload” metrics are the most effective for predicting fault generation. Among the outer metrics, the fatigue meter value is related to all inner metrics and is therefore useful for verifying the inner metric values.

In this paper we develop a pipeline processing system called NMR pipeline. In an NMR pipeline, processing elements at each stage are replicated to achieve high fault tolerance. Each stage of the pipeline is an N-Modular Redundant (NMR) System. This design combines the benefits of pipeline processing and NMR systems and is suitable for high speed safety-critical computing. Reliability analysis of this system is presented taking into account possible failures of processing and communication elements. The optimality of the system is defined in terms a suitably defined cost function that accounts for the component costs and the costs associated with system unreliability.

This paper presents a new concept, complex common cause failure, for common cause failure analysis. A common stress model is developed, in which the common cause events are described as common stresses that will affect directly and simultaneously on the internal aging process of each system component and further change their failure probabilities, while the internal aging processes of components are still considered mutually independent. The common stress model can be used to estimate the reliability of systems with complex common cause failures.