This paper addresses the optimization problem of minimizing the total construction cost, constrained by a reliability threshold. The existing algorithm is not efficient in the situation where the number of edges is much greater than the number of nodes. In order to resolve this situation, we firstly propose a method to identify the network with maximum all-terminal reliability among the networks that have the same number of edges. For the cases of e = n + 2 and e = n + 3, we discover a group of networks that possess greater all-terminal reliability than the other groups. Then, the exact solution of the maximum all-terminal reliability is derived as a closed form. Our method is implementable even when the number of edges is two or more greater than the number of nodes. This finding directly leads to a significant improvement in the overall efficiency of solving the optimization problem.
A one-shot system is a system that can be used only one time during its lifetime. Its reliability deteriorates with time even when it is in storage, and its failure is detected only through inspections. Thus, an appropriate inspection policy for such systems is required. In this paper, we deal with a one-shot system that consists of a single unit, where minimal repairs are performed when failures are detected by inspections. When the nth failure of the system is detected, the system is replaced by a new one. A periodic inspection policy and a non-periodic inspection policy for the system are proposed. The objective is to find the optimal number of failures before replacement and the inspection intervals that minimize the expected total cost per unit of time ensuring a predetermined mean availability. The results of the two policies are also compared.
Even when estimating software reliability with any nonparametric method, it is required to predict the number of software failures that may occur in the future after releasing the software to the market or users. In this paper, we consider a software release decision on when to stop software testing by minimizing the expected total software cost under the assumption that a probability law for software fault detection is unknown. We focus on a nonparametric prediction method of a nonhomogeneous Poisson process (NHPP) of Miller and Sofer (1991), and apply it to the optimal software release problem. We calculate the predictive confidence interval as well as the point estimate of the optimal software release time. Since our method is based on a predictive approach in spite of its nonparametric nature, it is useful to make exible decisions on when to stop software testing.
Considering an actual software testing phase, we have no doubt that the software reliability growth process depends on test environment factors, such as testing coverage, the number of test-runs and debugging skills, which affect the software failure occurrence or fault detection phenomenon in the testing phase. In this paper, we propose a software reliability modeling approach that considers the effects of the testing environment factors based on a program size that depends on a discrete binomial-type software reliability model which is consistent with software reliability data collection and enables us to consider the effect of the program size. Finally, we compare the accuracy of our model through a software reliability assessment of our model with the existing corresponding model by using actual data.
The restart is one of the typical environmental diversity techniques in dependable computing, and is quite effective to rejuvenate software systems at low cost. In this paper we generalize the seminal results on restart mechanisms by van Moorsel and Wolter (2004, 2006) and analyze optimal restart policies under more general conditions. We further develop a statistical algorithm to estimate the optimal restart policies from the empirical data of task processing time.
In a multi-state consecutive-k-out-of-n:F system, both the system and its components are allowed to be in two or more possible states. One of the most important problems for this system is to obtain the optimal arrangement that maximizes the expectation of the system state. Many researchers have studied the optimal arrangement problem in a multi-state consecutive-k-out-of-n:F system. However, the optimal solution for this problem is obtained by calculating the expectation of the system state as an enumeration method. As the number of n increases, the number of calculations becomes too many to obtain the optimal solution within a reasonable time even if a high-performance computer is used. Therefore, to solve optimal arrangement problem quickly, simulated annealing (SA), which is a kind of metaheuristics, was applied to the optimal arrangement problem in a multi-state consecutive-k-out-of-n:F system. However, although it is possible obtain a solution for the optimal arrangement problem using a simulated annealing algorithm, there is no guarantee that the solution is optimal. Therefore, the simulated annealing algorithm applied to an optimal arrangement problem in a multi-state consecutive-k-out-of-n:F system must be improved in order to search more efficiently. In this paper, we propose three types of simulated annealing algorithms with the aim of obtaining an optimal arrangement efficiently. We execute numerical experiments to compare their performances and investigate the efficiencies of these algorithms.
Modern assembly products are often manufactured and transported as a part of a global supply chain, where suppliers are selected to reduce procurement costs among multiple industrialized and emerging countries in Asia. Additionally, for environmental conscious manufacturing, it is important to visualize and reduce greenhouse gas (GHG) emissions for productions and logistics in whole global supply chains in order to resolve global warming problems. According to life cycle assessment (LCA), which quantitatively measures GHG emissions, even if the same types of materials are used in parts and components production, each part has a different level of GHG emissions and procurement cost depending on the country where it is manufactured. Therefore, the designing of a low-carbon supply chain requires the selection of appropriate material based low-carbon suppliers and making economic decisions using a life cycle inventory (LCI) database. To construct low-carbon supply chains that combine production of each part in various Asian countries, this study proposes a low-carbon and economic supplier selection method that utilizes an estimation of the GHG emissions and an affordable cost increment for each part using a LCI database developed incorporating Asian international input-output (I/O) tables. First, the GHG emissions for the production of each part in multiple Asian countries are estimated using the LCI database. Next, a bill of materials (BOM) with the GHG emissions and the procurement costs for each part is constructed in the cases of multiple production countries. Finally, the suppliers for producing each part are selected among the countries by integer programming in order to balance the reduction of GHG emissions and procurement costs.
Engineer-to-order (ETO) manufacturing is required for designing a product that matches customer requirements. However, firms face challenges when the proposed product specifications need to be changed after the contract has been signed. These changes lead not only to additional costs, but also to lost time for the design, production, and parts supply departments. To correct the product specifications, we propose a product functional structure model for accurately determining the product functional specifications. Using the product functional structure model, the customer requirements can be defined clearly, and unrealizable specifications of a product can be avoided. As a result, the firm can reduce parts inventory costs.
The turnover rate in the accommodation industry is higher than that of any other industry in Japan. The long business hours in ryokans make it difficult for employees to work there. To achieve a work-life balance, it is necessary for ryokans to improve operations. Furthermore, service industries are expected to earn considerable benefits by using time and motion study. In this paper, we apply time and motion study to the operation of cleaning up after banquets in ryokans. First, the problems encountered during operations are identified by time and motion study. Second, the proposed operation intended to solve these problems is developed. Third, the proposed operation is evaluated from the total operation time calculated by the methods-time measurement system. Finally, the usefulness of the proposed operation is evaluated in terms of its practical application. Consequently, it is found that the proposed operation is useful from the perspective of the total operation time and workload.
This paper develops an eye-tracking analysis framework for attention processes while viewing print advertisements. The framework consists of a scheme for classifying information in advertising and data interpretation procedures from attention process perspectives based on measurements introduced. Based on a feasibility study in which the eye-tracking data of 20 female participants looking at insurance advertisements, as well as questionnaire responses, were collected, the potential of our framework and implications for effective advertising design are discussed.