Search results

The purpose of this paper is to present a novel method to estimate the optimal software testing time which minimizes the relevant expected software cost via a refined neural network approach with the grouped data, where the multi-stage look ahead prediction is carried out with a simple three-layer perceptron neural network with multiple outputs.

To analyze the software fault count data which follows a Poisson process with unknown mean value function, the authors transform the underlying Poisson count data to the Gaussian data by means of one of three data transformation methods, and predict the cost-optimal software testing time via a neural network.

In numerical examples with two actual software fault count data, the authors compare the neural network approach with the common non-homogeneous Poisson process-based software reliability growth models. It is shown that the proposed method could provide a more accurate and more flexible decision making than the common stochastic modeling approach.

It is shown that the neural network approach can be used to predict the optimal software testing time more accurately.

In this paper, the authors propose two construction methods to estimate confidence intervals of the time-based optimal software rejuvenation policy and its associated maximum system availability via a parametric bootstrap method. Through simulation experiments the authors investigate their asymptotic behaviors and statistical properties.

The present paper is the first challenge to derive the confidence intervals of the optimal software rejuvenation schedule, which maximizes the system availability in the sense of long run. In other words, the authors concern the statistical software fault management by employing an idea of process control in quality engineering and a parametric bootstrap.

As a remarkably different point from the existing work, the authors carefully take account of a special case where the two-sided confidence interval of the optimal software rejuvenation time does not exist due to that fact that the estimator distribution of the optimal software rejuvenation time is defective. Here the authors propose two useful construction methods of the two-sided confidence interval: conditional confidence interval and heuristic confidence interval.

Although the authors applied a simulation-based bootstrap confidence method in this paper, another re-sampling-based approach can be also applied to the same problem. In addition, the authors just focused on a parametric bootstrap, but a non-parametric bootstrap method can be also applied to the confidence interval estimation of the optimal software rejuvenation time interval, when the complete knowledge on the distribution form is not available.

The statistical software fault management techniques proposed in this paper are useful to control the system availability of operational software systems, by means of the control chart.

Through the online monitoring in operational software systems, it would be possible to estimate the optimal software rejuvenation time and its associated system availability, without applying any approximation. By implementing this function on application programming interface (API), it is possible to realize the low-cost fault-tolerance for software systems with aging.

In the past literature, almost all authors employed parametric and non-parametric inference techniques to estimate the optimal software rejuvenation time but just focused on the point estimation. This may often lead to the miss-judgment based on over-estimation or under-estimation under uncertainty. The authors overcome the problem by introducing the two-sided confidence interval approach.

To determine the optimal software warranty period in continuous and discrete circumstances where the difference between the software testing environment and the operational environment can be characterised by an environment factor.

Software reliability models based on continuous and discrete time non‐homogeneous Poisson processes are assumed to describe the failure occurrence phenomena under both environments. Based on the idea of accelerated life testing for hardware products, the operational profile of the software is modeled, and the total expected software cost incurred in both testing and operational phases is formulated.

Under a milder condition, the optimal warranty period which minimizes the total software cost is derived analytically.

This paper introduces the operational profile of software to model the difference between the testing environment and the operational environment.

The determination of the release schedule for a new software product is the most important issue for designing and controlling a software development process. In fact, the optimal software release problem based on some software reliability growth models has been studied by many authors. In this paper, we propose a new method to estimate the optimal software release time under an alternative cost criterion. More precisely, two kinds of artificial neural networks are used to estimate the fault‐detection time observed in both testing and operation phases. As a cost criterion, we adopt the expected cost rate (the expected total software cost per unit testing time). Then, it is shown that the optimization problem to obtain the optimal release time can be reduced to a graphical one to minimize the tangent slope from a point to an (estimated) empirical curve in two‐dimensional space. Through numerical examples using actual fault‐detection time data, it is illustrated that the method proposed is a very useful device to estimate the optimal software release time precisely.

To determine the optimal spare part order‐replacement policy for any high cost single unit complex system in a discrete‐time circumstance.

The expected total discounted cost over an infinite planning horizon is taken as a criterion of optimality as it allows us to put emphasis on the present behavior of the system.

The problem under consideration is a two‐dimensional discrete optimization problem (regular ordering time and inventory time limit for the spare are decision variables) which is difficult to handle, in general. However, it is explored that the problem can be reduced to a simple one‐dimensional one and the optimal ordering time is to be determined under the two extreme situations: no replacement of the spare until the original unit fails and replacement of the spare as soon as it is delivered.

For modeling simplicity, deterministic lead time is considered for both regular and expedited orders. A more appropriate assumption would be to consider randomized lead time for both the orders.

The research provides a useful order‐replacement strategy for a single‐unit system where the failure of the unit is better measured by the number of cycles completed before failure rather than the instant of failure.

The work done in this paper carries certain values as any continuous time model for the problem under consideration can be regarded as only an approximate model.

The software reliability models to describe the reliability growth phenomenon are formulated by any stochastic point process with state‐dependent or time‐dependent intensity function. On the other hand, to deal with the environmental data, which consists of covariates influencing times to software failure, it may be useful to apply the Cox’s proportional hazards model for assessing the software reliability. In this paper, we review the proportional hazards software reliability models and discuss the problem to determine the optimal software release time under the expected total software cost criterion. Numerical examples are devoted to examine the dependence of the covariate structure in both the software reliability prediction and the optimal software release decision.

This paper proposes a new approach source node exclusion method (SNEM) to evaluate terminal pair reliability of complex communication networks.

The proposed approach breaks a non‐series‐parallel network to obtain its sub‐networks by excluding the source node from rest of the network. The reliabilities of these sub‐networks are thereafter computed by first applying the series‐parallel‐reductions to it and if any sub‐network results into another non‐series‐parallel network then it is solved by the recursive application of SNEM.

The proposed method has been applied on a variety of network and found to be quite simple, robust, and fast for terminal pair reliability evaluation of large and complex networks.

The proposed approach is quite simple in application and applicable to any general networks, i.e. directed and undirected. The method does not require any prior information such as path (or cut) sets of the network and their pre‐processing thereafter or perform complex tests on networks to match a predefined criterion.

The proposed approach provides an easy to develop and easy to use tool to determine terminal pair reliability of a communication network. The approach is particularly useful for communication network designer and analysts.

The aim of this paper is to propose a software reliability growth model based on stochastic differential equations for the integration testing phase of distributed development environment.

A client/server system (CSS), which is a new development method, has come into existence as a result of the progress of networking technology by UNIX systems. On the other hand, the effective testing method for distributed development environment has only a few presented. The method of software reliability assessment considering the interaction among software components in a distributed one is discussed.

Conventional software reliability growth models for system testing phase in distributed development environment have included many unknown parameters. Especially, the effective estimation method in terms of these unknown parameters, which means the proportion of the total testing‐load for the software component, has never been presented. This software reliability growth model can be easily applied in distributed software development, because the model has a simple structure.

This model is very useful for software developers in terms of practical reliability assessment in the actual distributed development environment.

The method of software reliability assessment considering the interaction among software components in distributed development environment is proposed. Additionally, several numerical examples for the actual data are presented.