Artificial Neural Network Applications for Software Reliability Prediction

Manjubala Bisi is currently an Assistant Professor in the Computer Science and Engineering Department, Kakatiya Institute of Technology and Science, Warangal, Telengana, India. She received her PhD from the Indian Institute of Technology Kharagpur in Reliability Engineering in 2015. Her research interests include software reliability modelling, artificial neural networks and soft computing techniques.Neeraj Kumar Goyal is currently an Associate Professor in Subir Chowdhury School of Quality and Reliability, Indian Institute of Technology Kharagpur, India. He received his PhD from IIT Kharagpur in Reliability Engineering in 2006. His major areas of research are network /system reliability and software reliability. He has completed various research and consultancy projects for various organizations, e.g. DRDO, NPCIL, Vodafone, ECIL etc. He has contributed research papers to refereed international journals and conference proceedings.

• Preface xiAcknowledgement xvAbbreviations xvii1 Introduction 11.1 Overview of Software Reliability Prediction and Its Limitation 61.2 Overview of the Book 81.2.1 Predicting Cumulative Number of Software Failures in a Given Time 91.2.2 Predicting Time Between Successive Software Failures 111.2.3 Predicting Software Fault-Prone Modules 131.2.4 Predicting Software Development Efforts 151.3 Organization of the Book 172 Software Reliability Modelling 192.1 Introduction 192.2 Software Reliability Models 202.2.1 Classification of Existing Models 212.2.2 Software Reliability Growth Models 252.2.3 Early Software Reliability Prediction Models 272.2.4 Architecture based Software Reliability Prediction Models 292.2.5 Bayesian Models 312.3 Techniques used for Software Reliability Modelling 312.3.1 Statistical Modelling Techniques 312.3.2 Regression Analysis 352.3.3 Fuzzy Logic 372.3.3.1 Fuzzy Logic Model for Early Fault Prediction 382.3.3.2 Prediction and Ranking of Fault-prone Software Modules using Fuzzy Logic 392.3.4 Support Vector Machine 402.3.4.1 SVM for Cumulative Number of Failures Prediction 412.3.5 Genetic Programming 452.3.6 Particle Swarm Optimization 492.3.7 Time Series Approach 502.3.8 Naive Bayes 512.3.9 Artificial Neural Network 522.4 Importance of Artificial Neural Network in Software Reliability Modelling 542.4.1 Cumulative Number of Software Failures Prediction 552.4.2 Time Between Successive Software Failures Prediction 582.4.3 Software Fault-Prone Module Prediction 602.4.4 Software Development Efforts Prediction 642.5 Observations 672.6 Objectives of the Book 703 Prediction of Cumulative Number of Software Failures 733.1 Introduction 733.2 ANN Model 763.2.1 Artificial Neural Network Model with Exponential Encoding 773.2.2 Artificial Neural Network Model with Logarithmic Encoding 773.2.3 System Architecture 783.2.4 Performance Measures 803.3 Experiments 813.3.1 Effect of Different Encoding Parameter 823.3.2 Effect of Different Encoding Function 833.3.3 Effect of Number of Hidden Neurons 863.4 ANN-PSO Model 883.4.1 ANN Architecture 893.4.2 Weight and Bias Estimation Through PSO 913.5 Experimental Results 933.6 Performance Comparison 944 Prediction of Time Between Successive Software Failures 1034.1 Time Series Approach in ANN 1054.2 ANN Model 1064.3 ANN- PSO Model 1134.4 Results and Discussion 1164.4.1 Results of ANN Model 1164.4.2 Results of ANN-PSO Model 1214.4.3 Comparison 1255 Identification of Software Fault-Prone Modules 1315.1 Research Background 1335.1.1 Software Quality Metrics Affecting Fault-Proneness 1345.1.2 Dimension Reduction Techniques 1355.2 ANN Model 1375.2.1 SA-ANN Approach 1395.2.1.1 Logarithmic Scaling Function 1395.2.1.2 Sensitivity Analysis on Trained ANN 1405.2.2 PCA-ANN Approach 1425.3 ANN-PSO Model 1455.4 Discussion of Results 1485.4.1 Results of ANN Model 1495.4.1.1 SA-ANN Approach Results 1495.4.1.2 PCA-ANN Approach Results 1525.4.1.3 Comparison Results of ANN Model 1555.4.2 Results of ANN-PSO Model 1625.4.2.1 Reduced Data Set 1625.4.2.2 Comparison Results of ANN-PSO Model 1636 Prediction of Software Development Efforts 1756.1 Need for Development Efforts Prediction 1786.2 Efforts Multipliers Affecting Development Efforts 1786.3 Artificial Neural Network Application for Development Efforts Prediction 1796.3.1 Additional Input Scaling Layer ANN Architecture 1816.3.2 ANN-PSO Model 1836.3.3 ANN-PSO-PCA Model 1866.3.4 ANN-PSO-PCA-GA Model 1886.3.4.1 Chromosome Design and Fitness Function 1896.3.4.2 System Architecture of ANN-PSOPCA-GA Model 1906.4 Performance Analysis on Data Sets 1926.4.1 COCOMO Data Set 1946.4.2 NASA Data Set 2026.4.3 Desharnais Data Set 2066.4.4 Albrecht Data Set 2097 Recent Trends in Software Reliability 215References 219Appendix Failure Count Data Set 231Appendix Time Between Failure Data Set 235Appendix CM1 Data Set 241Appendix COCOMO 63 Data Set 283Index 289