An Artificial Neural-Network-Based Approach to Software Reliability Assessment

In the software industry, two software engineering development best practices coexist: open-source and closed-source software. The former has a shared code that anyone can contribute, whereas the latter has a proprietary code that only the owner can access. Software reliability is crucial in the industry when a new product or update is released. Applying meta-heuristic optimization algorithms for closed-source software reliability prediction has produced significant and accurate results. Now, open-source software dominates the landscape of cloud-based systems. Therefore, providing results on open- source software reliability - as a quality indicator - would greatly help solve the open-source software reliability growth- modelling problem. The reliability is predicted by estimating the parameters of the software reliability models. As software reliability models are inherently nonlinear, traditional approaches make estimating the appropriate parameters difficult and ineffective. Consequently, software reliability models necessitate a high- quality parameter estimation technique. These objectives dictate the exploration of potential applications of meta- heuristic swarm intelligence optimization algorithms for optimizing the parameter estimation of nonhomogeneous Poisson process-based open-source software reliability modelling. The optimization algorithms are firefly, social spider, artificial bee colony, grey wolf, particle swarm, moth flame, and whale. The applicability and performance evaluation of the optimization modelling approach is demonstrated through two real open-source software reliability datasets. The results are promising

Software has become a vital factor in the fourth industrial revolution. Owing to the increase in demand for software products in various fields (big data, artificial intelligence, the Internet of Things, etc.), the software industry has expanded more than ever before. Therefore, software reliability has become very important, and efforts are being made to increase it. One of these efforts is the development of software reliability models (SRMs). SRMs have been studied for a long time as a model that predicts software reliability by using the number of software faults. Software failures can occur for several reasons, including independent software faults such as code errors and software hangs, as well as dependent cases where code errors lead to other software faults. Recently, due to the diversity of software operating environments, software faults are more likely to occur in a dependent manner, and, for this reason, they are likely to increase rapidly from the beginning and progress slowly to the maximum number thereafter. In addition, many large companies have focused on open-source software (OSS) development, and OSS is being developed by many users. In this study, we propose a new SRM that considers the number of finite faults and dependent faults, and examine the goodness-of-fit of a new SRM and other existing non-homogeneous Poisson process models based on the OSS datasets. Through numerical examples, the proposed model demonstrated a significantly better goodness-of-fit when compared to other existing models, and it also exhibited better results on the newly proposed integrated criteria.

This paper explores a new approach for predicting software faults by means of NARX neural network. Also, a careful analysis has been carried out to determine the applicability of NARX network in software reliability. The validation of the proposed approach has been performed using two real software failure data sets. Comparison has been made with some existing parametric software reliability models as well as some neural network (Elman net and TDNN) based SRGM. The results computed shows that the proposed approach outperformed the other existing parametric and neural network based software reliability models with a reasonably good predictive accuracy.

Software is widely used in various fields. There is no place where it is not used from the smallest part to the entire part. In particular, the tendency to rely on software is accelerating as the fields of artificial intelligence and big data become more widespread. Therefore, it is extremely important to evaluate the reliability of software because of the extensive damage that could occur if the software fails. Previously, software reliability models were developed based on mathematical and statistical grounds; however, immediate response was difficult. Therefore, in this study, a software reliability model was developed that depends on data using deep learning, and it was analyzed by replacing the activation function previously used in deep learning with the proposed software reliability model. Since the sigmoid function has a similar shape to the software reliability model, we utilized this to propose a deep learning software reliability model that replaces the activation function, the sigmoid function, with the software reliability function. Two datasets were compared and analyzed using 10 criteria, and the superiority of the proposed deep-learning software reliability model was proved. In addition, the results were compared by changing the parameters utilized in the proposed deep-learning software reliability model by −10%, −5%, 5%, and 10%, and it was found that the larger the parameters, the smaller the change.

Over time, software has become increasingly important in various fields. If the current software is more dependent than in the past and is broken owing to large and small issues, such as coding and system errors, it is expected to cause significant damage to the entire industry. To address this problem, the field of software reliability is crucial. In the past, efforts in software reliability were made to develop models by assuming a nonhomogeneous Poisson-process model (NHPP); however, as models became more complex, there were many special cases in which models fit well. Hence, this study proposes a software reliability model using deep learning that relies on data rather than mathematical and statistical assumptions. A software reliability model based on recurrent neural networks (RNN), long short-term memory (LSTM), and gated recurrent units (GRU), which are the most basic deep and recurrent neural networks, was constructed. The dataset was divided into two, Datasets 1 and 2, which both used 80% and 90% of the entire data, respectively. Using 11 criteria, the estimated and learned results based on these datasets proved that the software reliability model using deep learning has excellent capabilities. The software reliability model using GRU showed the most satisfactory results.

Management technologies for improving software reliability are very important for software TQM (Total Quality Management). The quality characteristics of software reliability is that computer systems can continue to operate regularly without the occurrence of failures on software systems. In this chapter, we describe several recent developments in software reliability modeling and its applications as quantitative techniques for software quality/reliability measurement and assessment. That is, a quality engineering analysis of human factors affecting software reliability during the design-review phase, which is the upper stream of software development, and software reliability growth models based on stochastic differential equations and discrete calculus during the testing phase, which is the lower one, are discussed. And, we discuss quality-oriented software management analysis by applying the multivariate analysis method and the existing software reliability growth models to actual process monitoring data. Finally, we investigate an operational performability evaluation model for the software-based system, introducing the concept of systemability which is defined as the reliability characteristic subject to the uncertainty of the field environment.KeywordsHuman factor analysisDesign-review experimentOSS reliabilityStochastic differential equationDiscrete modelingDifference equationSoftware managementSoftware project assessmentSoftware performability modelingSystemability assessment

In current software reliability modeling research, the main concern is how to develop general prediction models. In this paper, we propose several improvements on the conventional software reliability growth models (SRGMs) to describe actual software development process by eliminating some unrealistic assumptions. Most of these models have focused on the failure detection process and not given equal priority to modeling the fault correction process. But, most latent software errors may remain uncorrected for a long time even after they are detected, which increases their impact. The remaining software faults are often one of the most unreliable reasons for software quality. Therefore, we develop a general framework of the modeling of the failure detection and fault correction processes. Furthermore, we apply neural network with back-propagation to match the histories of software failure data. We will also illustrate how to construct the neural networks from the mathematical viewpoints of software reliability modeling in detail. Finally, numerical examples are shown to illustrate the results of the integration of the detection and correction process in terms of predictive ability and some other standard criteria.

In reality, the fault detection (or correction) phenomenon and software reliability estimation in the operational phase are different from that in the testing phase. The fault removal continues at a slower rate during the operational phase. In this paper, we will investigate some techniques for software reliability prediction and measurement in the operational phase. We first review how some software reliability growth models (SRGMs) based on non-homogeneous Poisson processes (NHPPs) can be readily derived based on a unified theory for NHPP models. Under this general framework, we can not only verify some conventional SRGMs but also derive some new SRGMs that can be used for software reliability measurement in the operational phase. That is, based on the unified theory, we can incorporate the concept of multiple changepoints into software reliability modeling. We can formularize and simulate the fault detection process in the operational phase. Some numerical illustrations based on real software failure data are also presented. The experimental results show that the proposed models can easily reflect the possible changes of fault detection process and offer quantitative analysis on software failure behavior in field operation.

Non-homogeneous Poisson process (NHPP) model with typical reliability growth patterns is an important technology about evaluating software reliability. Two parameters which affect software reliability are original failures number and failure-detected rate. The paper firstly defines the software failure distributed and discusses a few kinds' software reliability models. Non-homogeneous Poisson process model is presented. The models with different parameters are discussed and analyzed. The model's restrict condition and some of parameters, original failures number and failure-detected ratio, are extrapolated. Assessment methodology about key parameters is given.

Software reliability modelling (SRM) is a mathematics technique to estimate some measures of computer system that relate to software reliability. One group of existing models is using non-homogeneous Poisson process (NHPP) whose fault-number and failure-rate are constant or time-dependent functions. A few studies have been manipulated S-shaped curve to construct their models. However, those works remain some limitations. In this study, we introduce a new model that is based on a generalised S-shaped curve and evaluate it by real data set. After installing it in real code of Matlab and using MLE method to estimate parameter with a range of initial solution, we prove that our model converge to the most basic model of NHPP group, Goel-Okumoto model.

The safety-critical software of Reactor Protection System (RPS) plays a significant role for the safe operation of the nuclear power plant (NPP). However, it also brings challenges both to the reliability analysis of the RPS and to the Probabilistic Risk Assessment of the NPPs. The reliability analysis of safety-critical software is also expected by the nuclear regulation agencies and the software development groups for test evaluation and optimization. The detected faults during the software test process are regarded to have close connection with the software reliability and there have been hundreds of test-based software reliability models. Due to the particularity of its function, the safety-critical software of an NPP is especially sensitive to the faults with higher severity levels which should be paid special attention to. Severity levels of faults are commonly taken into account during software reliability modelling. In this paper, a novel software reliability growth model considering actual severity data was built based on a non-homogeneous Poisson process. Ratio of fatal faults with highest severity level to all accumulated faults was modelled with logistic curve. Then the mean value functions of both fatal and general faults were derived. A belief factor was employed to describe the trend of severity classification. In the end, the test data of a practical project was used to verify this model. The analysis results showed that this model had better prediction effect than Goel-Okumoto model and Inflection S-shaped model even when limited test data were collected. This novel model can be used as a helpful tool to evaluate both the reliability and the release time of the safety-critical software of an NPP.

The importance of software has increased significantly over time, and software failures have become a critical concern. As software systems have diversified in functionality, their structure has become more complex, and the types of failures that can occur in software have also diversified, stimulating the development of diverse software reliability models. In this study, we make certain assumptions regarding complex software, thereby proposing a new type of non-homogeneous Poisson process (NHPP) software reliability model that considers both dependent cases of software failure and cases that originate from the differences between the developed and actual operating environments. In addition, a new multi-criteria decision method (MCDM) that uses the maximum value for a comprehensive evaluation was proposed to demonstrate the effectiveness of the developed model. This improves the judgment of model excellence through multiple criteria and is suitable for multiple interpretations. To demonstrate the effectiveness of the proposed model, 15 NHPP software reliability models were compared using 13 evaluation criteria and three MCDM methods across two datasets. The results showed that one dataset performed well for all the criteria, whereas the other dataset performed well for the newly proposed a multi-criteria decision method using maximum (MCDMM). The sensitivity analysis also showed a change in the mean value function with a change in the parameters. These results demonstrate that an extended structure for complex software can lead to improved software reliability.

In software reliability modeling, the parameters of the model are typically estimated from the test data of the corresponding component. However, the widely used point estimators are subject to random variations in the data, resulting in uncertainties in these estimated parameters. Ignoring the parameter uncertainty can result in grossly underestimating the uncertainty in the total system reliability. This paper attempts to study and quantify the uncertainties in the software reliability modeling of a single component with correlated parameters and in a large system with numerous components. Another characteristic challenge in software testing and reliability is the lack of available failure data from a single test, which often makes modeling difficult. This lack of data poses a bigger challenge in the uncertainty analysis of the software reliability modeling. To overcome this challenge, this paper proposes utilizing experts' opinions and historical data from previous projects to complement the small number of observations to quantify the uncertainties. This is done by combining the maximum-entropy principle (MEP) into the Bayesian approach. This paper further considers the uncertainty analysis at the system level, which contains multiple components, each with its respective model/parameter/ uncertainty, by using a Monte Carlo approach. Some examples with different modeling approaches (NHPP, Markov, Graph theory) are illustrated to show the generality and effectiveness of the proposed approach. Furthermore, we illustrate how the proposed approach for considering the uncertainties in various components improves a large-scale system reliability model.

Performance analysis of software reliability growth models with testing-effort and change-point

Software reliability is one of the most important characteristics of software quality. Its measurement and management technologies during the software life-cycle are essential to produce and maintain quality/reliable software systems. In this chapter, we discuss software reliability modeling and its applications. As to software reliability modeling, hazard rate and NHPP models are investigated particularly for quantitative software reliability assessment. Further, imperfect debugging and software availability models are also discussed with reference to incorporating practical factors of dynamic software behavior. And three software management problems are discussed as an application technology of software reliability models: the optimal software release problem, statistical testing-progress control, and the optimal testing-effort allocation problem.