E/E/PE safety-related systems are applied in various industrial fields as safety systems that mitigate the hazardous risk of the entire system. Safety assessment of the E/E/PE safety-related systems provides important information for determining the need for risk mitigation activities. In particular, because the safety functions of the E/E/PE safety-related systems are controlled by software embedded in those safety-related systems, certain safety concerns exist not only for the hardware system but also for the software system. In this paper, we apply the concepts of existing software reliability modeling techniques to derivation of the target failure measures for quantitative software functional safety assessment of the E/E/PE safety-related systems. Furthermore, simulation-based interval estimation method for those target failure measures is discussed for considering the uncertainty of those measures based on the bootstrap method and bootstrap confidence interval. Finally, we present numerical examples of the proposed method to demonstrate our simulation-based interval estimation approach for quantitative software functional safety assessment.

We compare the methods of OSS performability evaluation by the deep multimodal learning and multitask one. This paper focuses on the Web-based OSS such as cloud computing, edge computing, and Web server. Also, several practical execution examples based on the proposed deep models by actual big fault data. Furthermore, this paper discusses several deep multimodal and multitask learning for the OSS performability analyses.

The demand of open source software is increasing because of the low cost, high quality, and short delivery. In particular, open source software is managed by using the bug tracking system. This paper focuses on the method of reliability assessment based on the deep learning. Then, the Wiener process is applied to the output value of objective variables. Moreover, several sensitivity analyses of the parameter of Wiener process are shown as several numerical examples.

From the perspective of software quality management, the software reliability growth model (SRGM) is frequently used by many project managers. Especially, the nonhomogeneous Poisson process model is used for the commercial software, because it can express the fault convergence status of software. On the other hand, the software reliability growth curve of open source software (OSS) tends to show the distorted curve due to multiple factors such as the number of the users and the software update. It is difficult to evaluate the software reliability for OSS. Thus, it is needed to develop SRGM considering the factors of OSS. In the past, the number of the downloads was used to the OSS’s SRGM considering number of users. However, several software distributors only record the sum of the number of the downloads for all version. To perform evaluation with high accuracy, it needs to use the latest version only for calculating execution time because the older version contributes very little to debugging the latest version OSS. In this paper, we analyze the migration of downloads of the latest version from urllib3 which can analyze the number of downloads by detail version. Moreover, we propose the SRGM considering the package manager behavior.

Non-homogeneous Poisson process (NHPP) based modeling has gained lot of importance in the field of Software Reliability Engineering. Many related models have been proposed in the literature and several are in the pipeline. Yamada's two stage fault removal modeling framework has been extensively worked in the past; wherein last decade has also seen the impact and importance of modeling related to imperfect debugging phenomenon. But, a major limitation of Yamada's two stage modeling framework has been its limitation to capture the time dependent nature of faults in the software. In the present proposition, an attempt has been made to reach to this requirement by making use of the approximation method as a solution methodology. The employment of this alternative solution methodology yields solutions to all the three possible situations; that is, pure error generation, exponential form and linear form of error generation. In order to facilitate the analysis, proposed cases have been verified on real life data sets and results obtained are quiet encouraging.

Previously, many researchers proposed various software reliability growth model (SRGM) to evaluate software reliability. Also, they applied SRGM to open source software (OSS). On the other hand, almost all OSS continuing to develop. In this case, OSS’s latent faults fluctuates frequently. Moreover, it is difficult to evaluate accurately because this doesn’t consider by SRGM assumptions. It is known that as the number of software users increases, the number of the detected faults tends to increase. Especially, it is seen in like the number of the users changes fluctuates and the software update. In this paper, we analyze the impact of fluctuations in the number of software users on OSS using actual data from a fault tracking system. Moreover, we review the assumptions of non-homogeneous Poisson process (NHPP) based SRGM and propose the corrected time for model considering the number of number of users. Furthermore, we applied propose model to existing general NHPP-based SRGM’s mean value function. As a result, we confirmed that the number of users increases effect to the number of the detected faults. Furthermore, the propose model can evaluate software reliability accurately. Especially, logarithmic Poisson SRGM shows the highest estimation result.

We focus on the noise analysis by using stochastic differential equation model as training data of deep learning. Then, we use the solution process of stochastic differential equation for the software fault analysis to the training data. Also, several numerical examples are shown in this paper. Moreover, this paper shows several sensitivity analyses for the actual big fault data.

IEC 61508 is used for target failure measures, such as the average probability of a dangerous failure on demand, PFDave, for low-demand operation mode and the probability of a dangerous failure per hour, PFH, for high-demand operation mode when assessing the functional safety of E/E/PE safety-related systems. It is known that this regulation leads to discontinuous functional safety assessment and the problem of evaluation accuracy around the boundary of these operation mode. We newly propose a target failure measure which enables us to conduct the functional safety assessment continuously regardless of the operation mode for a single channel structure of the E/E/PE safety-related systems. A numerical experiment is conducted to consider the difference between our approach and the IEC 61508-based approach.

In recent years, open source software (OSS) has become ubiquitous in every field of our daily life. Hence, the OSS’s reliability is a significant challenge. The traditional models, like the software reliability growth model, can not handle a large scale of data efficiently, while the deep learning provides an effective method. This paper proposes a multi-input multi-output deep neural network to predict the fault detection time intervals and the fault modification time intervals simultaneously. Additionally, we present several numerical examples with a cross validation conducted with 3 types of data splits.

The fault big data sets of many open source software (OSS) are recorded on the bug tracking systems. In the past, we have proposed the effort assessment method under the assumption that the fault detection phenomenon depends on the maintenance effort, because the number of software fault is influenced by the effort expenditure. The past research in terms of the effort assessment method of OSS is based on the effort data sets. On the other hand, we propose the deep learning approach to the OSS fault big data. In the past, the existing method without Wiener process cannot estimate within the range of existing data only. The proposed method assumes that the fault detection process follows the Wiener process such as the imperfect debugging and Markov property. Thereby, the proposed method can estimate the exceeding values by adding the white noise based on the Wiener process. Then, the proposed method make it possible for the OSS managers to assess the values exceeding from the existing data. Then, we show several reliability assessment measures based on the fault modification time based on the deep learning. Moreover, several numerical illustrations based on the proposed deep learning model are shown in this paper.

This paper focuses on the reliability of open source software (OSS). In particular, the proposed method is based on the deep learning by using the fault big data obtained from the bug tracking system. The immune system of human body takes on the role of the protection for the pathogenic microorganisms. Recently, the source code and complexity of OSS system become large year by year. Then, we assume that the OSS fault removal system is similar to the pathogenic microorganisms. We propose the OSS fault removal system based on deep learning inspired by immune system in this paper.

Recently, open source software (OSS) has been widely used in many fields due to the spread and development facilitated by networks. The characteristics of OSS include no cost and high performance, which make it a significant component of modern society. However, the number of reported faults is increasing due to its vulnerabilities. Detecting these faults requires substantial costs, and correcting the growing number of faults necessitates a large workforce. In this paper, we propose a method for Reliability Assessment of OSS using deep learning based on the human immune system. Additionally, we present several numerical examples based on the proposed method.

Maintenance activities for safety-related systems are generally required to ensure that the systems are working as intended. Regarding the maintenance activities, proof-testing is known as scheduled inspections and maintenance activities for detecting dangerous undetected faults which cannot be detected by diagnostic testing systems installed in the safety-related systems. However, the proof-testing needs a lot of cost and provokes decreasing of the availability for the whole system because the whole system is needed to shut down for proofing that the whole system is working as intended. We discuss analytical methodologies for obtaining optimal proof-testing interval with harmful risk and proof-testing cost by describing the behavior of the safety-related system based on a continuous-time Markov chain. Further, an analytical optimal policy for obtaining economic proof-testing interval is proposed in this paper.

This paper focuses on the sustainability based on the effort by using the fault big data of open source software (OSS). The fault detection phenomenon depends on the maintenance effort, because the number of software fault is influenced by the effort expenditure. Actually, the software reliability growth models with testing-effort have been proposed in the past. In this paper, we apply the deep learning approach to the OSS fault big data. Also, we propose the reliability assessment measure of sustainability. Then, we show several sustainability assessment measure based on the deep learning. Moreover, several numerical illustrations based on the proposed deep learning model are shown in this paper.