Software Fault Detection Research Articles

Predicting software reliability via completely monotone nonparametric estimator with grouped data

Nonparametric software reliability analysis is a challenging issue to predict software reliability under incomplete knowledge on software fault-detection time distribution, because the underlying stochastic model is a function of only software fault data and is not predictable in principle for unknown patterns in future long term. Sofer and Miller (1991) develop a unique approach based on a completely monotone nonparametric estimator, but just focus on the fault-detection time data. However, such data sets are seldom available in practice, and their approach is not applicable to many actual software development processes. In this paper, we revisit the seminal completely monotone estimator by Sofer and Miller (1991) to use in the major case with grouped data, and develop both estimation and prediction methods of software fault count. We investigate the potential performance of our distribution-free method with thirteen actual software development project data, and compare it with the existing parametric models known as nonhomogeneous Poisson process-based software reliability models.

A general modeling and analysis framework for software fault detection and correction process

SummarySoftware reliability growth modeling plays an important role in software reliability evaluation. To incorporate more information and provide more accurate analysis, modeling software fault detection and correction processes has attracted widespread research attention recently. In modeling software correction processes, the assumption of fault correction time is relaxed from constant delay to random delay. However, stochastic distribution of fault correction time brings more difficulties in modeling and corresponding parameter estimation. In this paper, a framework of software reliability models containing both information from software fault detection process and correction process is studied. Different from previous extensions on software reliability growth modeling, the proposed approach is based on Markov model other than a nonhomogeneous Poisson process model. Also, parameter estimation is carried out with weighted least‐square estimation method, which emphasizes the influence of later data on the prediction. Two data sets from practical software development projects are applied with the proposed framework, which shows satisfactory performance with the results. Copyright © 2016 John Wiley & Sons, Ltd.

Exploration of neural network models for defect detection and classification

In order to overcome the software development challenges like delivering a project on time, developing quality software products and reducing development cost, software industries commonly uses defect detection software tools to manage quality in software products. Defects are detected and classified based on their severity, this can be automated in order to reduce the development time and cost. Nowadays to extract useful knowledge from large software repositories, engineers and researchers are using data mining techniques. In this paper, software defect detection and classification method is proposed and neural network models such as generalised regression neural network (GRNN) and probabilistic neural network model (PNN) are integrated to identify, classify the defects from large software repository. Based on defects, severity proposed method discussed in this paper focuses on three layers: core, abstraction and application layer. The performance accuracy of the proposed model is compared with MLP and J48 classifiers.

An Automated Approach for Software Fault Detection and Recovery

Autonomic software recovery enables software to automatically detect and recover software faults. This feature makes the software to run more efficiently, actively, and reduces the maintenance time and cost. This paper proposes an automated approach for Software Fault Detection and Recovery (SFDR). The SFDR detects the cases if a fault occurs with software components such as component deletion, replacement or modification, and recovers the component to enable the software to continue its intended operation. The SFDR is analyzed and implemented in parallel as a standalone software at the design phase of the target software. The practical applicability of the proposed approach has been tested by implementing an application demonstrating the performance and effectiveness of the SFDR. The experimental results and the comparisons with other works show the effectiveness of the proposed approach.

Investigation of Software Defect Prediction Using Data Mining Framework

A software defect is a error, failure, fault in a computer program or system producing an incorrect or unexpected result, or causing it to behave in an unintended way. Software Defect Prediction (SDP) locates defective modules in software. The final product should have null or minimal defects to produce high quality software. Software defect detection at the earliest stage reduces development cost, reworks effort and improves the quality of software. In this study, the efficiency of different classifiers such as Naive Bayes, Support Vector Machine (SVM) and K-Nearest Neighbor (KNN) are evaluated for SDP.

A novel defect prediction method for web pages using k-means++

With the increase of the web software complexity, defect detection and prevention have become crucial processes in the software industry. Over the past decades, defect prediction research has reported encouraging results for reducing software product costs. Despite promising results, these researches have hardly been applied to web based systems using clustering algorithms. An appropriate implementation of the clustering in defect prediction may facilitate to estimate defects in a web page source code. One of the widely used clustering algorithms is k-means whose derived versions such as k-means++ show good performance on large-data sets. Here, we present a new defect clustering method using k-means++ for web page source codes. According to the experimental results, almost half of the defects are detected in the middle of web pages. k-means++ is significantly better than the other four clustering algorithms in three criteria on four data set. We also tested our method on four classifiers and the results have shown that after the clustering, Linear Discriminant Analysis is, in general, better than the other three classifiers.

Effective confidence interval estimation of fault-detection process of software reliability growth models

The quantitative evaluation of software reliability growth model is frequently accompanied by its confidence interval of fault detection. It provides helpful information to software developers and testers when undertaking software development and software quality control. However, the explanation of the variance estimation of software fault detection is not transparent in previous studies, and it influences the deduction of confidence interval about the mean value function that the current study addresses. Software engineers in such a case cannot evaluate the potential hazard based on the stochasticity of mean value function, and this might reduce the practicability of the estimation. Hence, stochastic differential equations are utilised for confidence interval estimation of the software fault-detection process. The proposed model is estimated and validated using real data-sets to show its flexibility.

CodeSonar를 이용한 지역 SW개발 업체의 결함 유형분석

Recently, various static analysis tools for software defect detection are becoming widely used in practice. However, there is little public information of the most frequent defects in commercial areas until now. In this paper, we analyze the defects found by CodeSonar, a static analysis tool that finds defects in C/C++, Java programs. So we report the most frequent defects by various aspects in Dongnam area, Korea.

Mining Software Repositories for Defect Categorization

Early detection of software defects is very important to decrease the software cost and subsequently increase the software quality. Success of software industries not only depends on gaining knowledge about software defects, but largely reflects from the manner in which information about defect is collected and used. In software industries, individuals at different levels from customers to engineers apply diverse mechanisms to detect the allocation of defects to a particular class. Categorizing bugs based on their characteristics helps the Software Development team take appropriate actions to reduce similar defects that might get reported in future releases. Classification, if performed manually, will consume more time and effort. Human resource having expert testing skills & domain knowledge will be required for labeling the data. Therefore, the need of automatic classification of software defect is high.This work attempts to categorize defects by proposing an algorithm called Software Defect CLustering (SDCL). It aims at mining the existing online bug repositories like Eclipse, Bugzilla and JIRA for analyzing the defect description and its categorization. The proposed algorithm is designed by using text clustering and works with three major modules to find out the class to which the defect should be assigned. Software bug repositories hold software defect data with attributes like defect description, status, defect open and close date. Defect extraction module extracts the defect description from various bug repositories and converts it into unified format for further processing. Unnecessary and irrelevant texts are removed from defect data using data preprocessing module. Finally grouping of defect data into clusters of similar defect is done using clustering technique. The algorithm provides classification accuracy more than 80% in all of the three above mentioned repositories.

NHPP-Based Software Reliability Model with Marshall-Olkin Failure Time Distribution

A new modeling approach for the non-homogeneous Poisson processes (NHPPs) based software reliability modeling is proposed to describe the stochastic behavior of software fault-detection processes, of which the failure rate is not monotonic. The fundamental idea is to apply the Marshall-Olkin distribution to the software fault-detection time distribution. The applicability of Marshall-Olkin distribution in software reliability modeling is studied. The data fitting abilities of the proposed NHPP-based software reliability model is compared with the existing typical ones through real software project data analysis.

Software defect detection by combining bounded model checking and approximations of functions

In recent years, the study of a software quality assurance technique called bounded model checking (BMC) has been increasingly intensified, because it makes it possible to successfully detect both functional and nonfunctional defects in real software. In this paper, we propose an original approach to the implementation of BMC based on combining the results of several recent studies in this field, namely, the use of the LLVM compiler infrastructure for parsing and transformation of the source code, the use of SMT-solver Z3 for the verification of the correctness of the properties, and the improvement of the analysis efficiency using the approximation of functions. The experimental results show that the approach can be applied to real projects.

Software Defect Distribution Prediction for BOSS System

Effective detection of software defects is an important activity of software development process. In this paper, we propose an approach to predict residual defects for BOSS project, which applies defect distribution model. Experiment results show that this approach can effectively improve the accuracy of defect prediction.

An empirical study based on semi-supervised hybrid self-organizing map for software fault prediction

Software testing is a crucial task during software development process with the potential to save time and budget by recognizing defects as early as possible and delivering a more defect-free product. To improve the testing process, fault prediction approaches identify parts of the system that are more defect prone. However, when the defect data or quality-based class labels are not identified or the company does not have similar or earlier versions of the software project, researchers cannot use supervised classification methods for defect detection. In order to detect defect proneness of modules in software projects with high accuracy and improve detection model generalization ability, we propose an automated software fault detection model using semi-supervised hybrid self-organizing map (HySOM). HySOM is a semi-supervised model based on self-organizing map and artificial neural network. The advantage of HySOM is the ability to predict the label of the modules in a semi-supervised manner using software measurement threshold values in the absence of quality data. In semi-supervised HySOM, the role of expert for identifying fault prone modules becomes less critical and more supportive. We have benchmarked the proposed model with eight industrial data sets from NASA and Turkish white-goods embedded controller software. The results show improvement in false negative rate and overall error rate in 80% and 60% of the cases respectively for NASA data sets. Moreover, we investigate the performance of the proposed model with other recent proposed methods. According to the results, our semi-supervised model can be used as an automated tool to guide testing effort by prioritizing the module’s defects improving the quality of software development and software testing in less time and budget.

Neural Network Parameter Optimization Based on Genetic Algorithm for Software Defect Prediction

Software fault prediction approaches are much more efficient and effective to detect software faults compared to software reviews. Machine learning classification algorithms have been applied for software defect prediction. Neural network has strong fault tolerance and strong ability of nonlinear dynamic processing of software defect data. However, practicability of neural network is affected due to the difficulty of selecting appropriate parameters of network architecture. Software fault prediction datasets are often highly imbalanced class distribution. Class imbalance will reduce classifier performance. A combination of genetic algorithm and bagging technique is proposed for improving the performance of the software defect prediction. Genetic algorithm is applied to deal with the parameter optimization of neural network. Bagging technique is employed to deal with the class imbalance problem. The proposed method is evaluated using the datasets from NASA metric data repository. Results have indicated that the proposed method makes an improvement in neural network prediction performance.

Monitoring applications: An immune inspired algorithm for software-fault detection

Large-scale software systems are in general difficult to manage and monitor. In many cases, these systems display unexpected behavior, especially after being updated or when changes occur in their environment (operating system upgrades or hardware migrations, to name a few). Therefore, to handle a changing environment, it is desirable to base fault detection and performance monitoring on self-adaptive techniques.Several studies have been carried out in the past which, inspired on the immune system, aim at solving complex technological problems. Among them, anomaly detection, pattern recognition, system security and data mining are problems that have been addressed in this framework.There are similarities between the software fault detection problem and the identification of the pathogens that are found in natural immune systems. Being inspired by vaccination and negative and clonal selection observed in these systems, we developed an effective self-adaptive model to monitor software applications analyzing the metrics of system resources.

The Analysis of Faults Detection Software Based on Improved Neural Network Algorithm

The large-scale software is consisted of the components which are quite different. The detection accuracy of the traditional faults detection methods for the large-scale component software is not satisfactory. This paper proposes a large-scale software faults detection methods based on improved neural network combining the features of the large-scale software by computing the stable probability and building the neural network faults detection models. The proposed method can analyze the serial faults of the large-scale software to determine the positions of the faults. The experiment and simulation results show that the improved method for large-scale software fault detection can greatly improve the accuracy.

Mitigating the effects of equivalent mutants with mutant classification strategies

Mutation Testing has been shown to be a powerful technique in detecting software faults. Despite this advantage, in practice there is a need to deal with the equivalent mutants' problem. Automatically detecting equivalent mutants is an undecidable problem. Therefore, identifying equivalent mutants is cumbersome since it requires manual analysis, resulting in unbearable testing cost. To overcome this difficulty, researchers suggested the use of mutant classification, an approach that aims at isolating equivalent mutants automatically. From this perspective, the present paper establishes and empirically assesses possible mutant classification strategies. A conducted study reveals that mutant classification isolates equivalent mutants effectively when low quality test suites are used. However, it turns out that as the test suites evolve, the benefit of this practice is reduced. Thus, mutant classification is only fruitful in improving test suites of low quality and only up to a certain limit. To this end, empirical results show that the proposed strategies provide a cost-effective solution when they consider a small number of live mutants, i.e., 10–12. At this point they kill 92% of all the killable mutants.

Software fault detection and recovery in critical real-time systems: An approach based on loose coupling

Remote handling (RH) systems are used to inspect, make changes to, and maintain components in the ITER machine and as such are an example of mission-critical system. Failure in a critical system may cause damage, significant financial losses and loss of experiment runtime, making dependability one of their most important properties. However, even if the software for RH control systems has been developed using best practices, the system might still fail due to undetected faults (bugs), hardware failures, etc. Critical systems therefore need capability to tolerate faults and resume operation after their occurrence. However, design of effective fault detection and recovery mechanisms poses a challenge due to timeliness requirements, growth in scale, and complex interactions. In this paper we evaluate effectiveness of service-oriented architectural approach to fault tolerance in mission-critical real-time systems. We use a prototype implementation for service management with an experimental RH control system and industrial manipulator. The fault tolerance is based on using the high level of decoupling between services to recover from transient faults by service restarts. In case the recovery process is not successful, the system can still be used if the fault was not in a critical software module.

The Design Process of Defect Detection Method for Large Industrial Automatic Control Software

The current difference trend of automatic control software structure is more obvious, software integration makes linking correlation of overall structure appeared to play down. No significant linking characteristics exist between structural ports. The traditional detection methods of software defect are utilized to locate software defect under the differentiation structural framework, due to the lack of clear articulation feature to indicate the location region, resulting in unidentified area which software defect signal belong to and position deviation. In this paper, software defect detection method was proposed on the basis of difference structure fusion algorithm. The signal fusion technology was applied to mix the signal of software defect detection effectively, and chaos particle swarm algorithm optimization was employed to support vector machine parameters, establish the optimal models of software defect prediction, thus completing the software defect detection. Experimental results show that the algorithm for software defect detection, can greatly improve the accuracy of detection.

Multi up-gradation software reliability growth model with learning effect and severity of faults using SDE

In recent years, the dependence on a computer system has become large in our social life. Therefore, it becomes more important for software developers to produce highly reliable software systems. Due to timely demand and competitive nature of the market of software product, firms are frequently launching their upgraded versions of the base software. Many software reliability growth model have been developed by software developers and managers in tracking and measuring the growth of reliability. As the size of software system is large and the number of faults detected during the testing phase becomes large, so the change of the number of faults that are detected and removed through each debugging becomes sufficiently small compared with the initial fault content at the beginning of the testing phase. In such a situation, we can model the software fault detection process as a stochastic process with continuous-state space. In this paper, we derive a stochastic differential equation of \(i\hat{t}o\) type based multi-up-gradation model with severity of faults and effect of learning. Moreover, we discuss the identification of the faults left in the software when it is in operational phase during the testing of the new code i.e. developed while adding new features to the existing software. We examine the case where there exists two types of faults in the software; simple and hard and during testing the simple faults are removed by exponential rate whereas hard faults are removed by Yamada with learning effect function. Results are supplemented by a numerical example.