International Journal of Reliability and Safety (5 papers in press)

Regular Issues

• Reliability analysis using change-point concept and optimal version-updating for open source software  
  by Ajay Kumar 
  Abstract: Our main objective is to analyse the reliability of open source software (OSS) using a change-point concept and estimate the optimal time for version updating of OSS. The focus is to provide an efficient parametric decomposition that will result in an analytical expression for the mean value function for predicting the OSS reliability that could be used by the software developer as well as software user to judge the reliability of OSS. A reliability growth model for OSS has been proposed on the basis of a modified NHPP which is developed by incorporating the change-point concept and imperfect debugging with constant Fault Reduction Factor (FRF). Owing to the varying nature of bugs and frequent software releases in OSS, the fault detection process cannot be smooth. To overcome the non-smoothness in the fault detection process, the change point concept has been incorporated in the model to assess OSS reliability. Such a reliability growth model further can be applied to estimate optimal time for version updating of OSS by considering the two important factors in the context, involvement of the volunteers and newcomers and reliability to ensure the quality of the software using multi-attribute utility theory. The proposed model has been tested on datasets from the public release of Apache open-source software and it has been found that the proposed model provides more accurate reliability estimation. Moreover, the proposed reliability growth model for OSS helps management to determine the optimal version-updating time for OSS using multi-attribute utility theory. The proposed model is practically implicated since it considers the multiple releases of the software to be common by taking the reliability growth model for each release individually into consideration. Furthermore, multiple change points can be considered in order to make the model more accurate.
  Keywords: open source software; fault reduction factor; change-point; imperfect debugging; software reliability growth model; multi-attribute utility theory; non-homogeneous Poisson process.
  
  
• A multi-criteria decision-making approach for optimal selection of software reliability growth models  
  by Aakash Gupta, Rakesh Garg 
  Abstract: The software industries are major stakeholders in the economy of any country. Most business processes are now implemented through software both online and offline. Before this software is deployed and delivered to the client, it is essential for the solution provider to extensively check its reliability. Reliability is an important factor that directly affects the future of any software in the market. The reliability of software could be enhanced by the use of software reliability growth models (SRGMs). In the present research, a hybrid multi-criteria decision making (MCDM) approach, namely Entropy-Evaluation based on Distance from Average Solution (EEDAS), is proposed and implemented to rank 16 SRGMs by selecting nine criteria using a failure dataset. Further, the EEDAS method results are compared with two well-known methods, WEDBA and TOPSIS, and validated by Kendall's tau correlation test. The present research results based on EDAS show that the Pham Zhang IFD model is the best one whereas the Gompert model is the worst.
  Keywords: software reliability; SRGMs; EDAS; multi-criteria decision making.
  
  
• A discrete modelling framework for fault prediction with logistic fault reduction factor  
  by Avinash K. Shrivastava, Ruchi Sharma 
  Abstract: Many software reliability growth models were developed by considering constant and time-dependent (increasing and decreasing) fault reduction factors (FRF). All these models were developed by considering the continuous-time approach. However, sometimes the appropriate unit of fault detection is the number of test runs or the number of test cases executed. Software reliability growth models developed for such cases are termed as discrete ones. In this work, we have developed two software reliability growth models with FRF by considering the perfect and imperfect debugging environment. We have further extended our discrete modelling approach for the multi-release of software. FRF is modelled by using logistic distribution. The results obtained for the proposed discrete FRF-based software reliability growth models are compared with the existing discrete models. On comparing the results, we see that the proposed models fit better on the three datasets used for numerical illustration.
  Keywords: software reliability; modelling; discrete; prediction; fault reduction factor.
  
  
• Code-based CRUD analysis for prioritising test cases  
  by Tomohiro Takeda, Satoshi Masuda, Tohru Matsuodani, Tsuyoshi Yumoto, Kazuhiko Tsuda 
  Abstract: When software is modified, an impact analysis is conducted to determine the effect of these modifications on other functions. However, the current impact-analysis techniques cannot identify such impact analysis. To compensate for this, comprehensive test cases are created. Therefore, impact analysis faces problems when increasing the true-positive ratio, which denotes the impacted implementations, and when reducing the false-positive ratio, which denotes the non-impacted implementations. To address this, Impact-Data-All-Used (IDAU) can be used to create and prioritise test cases based on CRUD information contained in design documents. We herein propose a code-based-IDAU (CB-IDAU) that applies IDAU to the source code using the control-graph and call-graph analysis. Based on a performance comparison of CB-IDAU to that of its previously proposed version, CB-IDAU, we observed an increase in the true-positive value by 157% and a reduction in the false-positive value by 60% when the full-coverage-test performance was used as the benchmark.
  Keywords: software testing; impact analysis; test case creation; call flow; control flow; data flow; testing automation; test case prioritisation; graph search; intermediate language.
  
  

Special Issue on: ICMTEA2020 Current practices in System Reliability Modelling

• Testing effort based SRGM and release decision under fuzzy environment  
  by Palak Saxena, Naveen Singh, Avinash K. Shrivastava, Vjay Kumar 
  Abstract: Software reliability is an important measure of software quality and is described by a mathematical model known as the software reliability growth model (SRGM). Several SRGMs have been developed for software engineers and testers to evaluate software reliability, the remaining number of faults, and the testing release time. In this study, we have proposed testing effort based software reliability growth models (SRGMs) with generalised modified Weibull (GMW) distribution under a fuzzy environment. During the testing process, it is likely that new faults are added, which is known as error generation, or the rate of error detection is poor owing to a low degree of skilled technique known as imperfect debugging. Further, we have incorporated a time-dependent rate function in our model and have calculated model parameter using Statistical Package for Social Sciences (SPSS). Moreover, we have also discussed the release time of the software using SRGM and the optimum value of reliability of the software under uncertainty in parameters. Numerical illustration on a real life dataset is done to validate the results.
  Keywords: SRGM; fuzzy set theory; testing; software reliability; release time.