Chinese Journal of Computers   Full Text
Title	Study on Software Reliability Growth Model Considering Failure Dependency
Authors	ZHAO Jing ZHANG Ru-Bo GU Guo-Chang
Address	(School of Computer Science and Technology, Harbin Engineering University, Harbin 150001)
Year	2007
Issue	No.10(1713—1720)
Abstract &   Background	Abstract Software reliability growth model (SRGM) is one of the important tools to evaluate and predict software reliability. However, software failure dependency, the differences of testing environment and operational environment are not considered by most SRGMs. This paper proposes a software reliability growth model that consider software failure dependency and differences of testing and operational environment. The experimental results based on two failure data sets show that the proposed model has better predictive power and curve fit than other NHPP SRGMs. keywords software reliability growth model; software failure dependency; testing environment; operational environment background NHPP (nonhomogeneous Poisson process) models, as a class of SRGMs, are extensively used. NHPP SRGMs have been quite successful tools in practical software reliability engineering. Many software reliability growth models (SRGMs) have been proposed, and most SRGMs assume that the successive software failure is independent and testing environment and operation environment is the same. However, the fault dependency is universal, and the differences of testing environment and operational environment exist in software testing. This paper incorporates fault dependency and differences of testing environment and operation environment. The relations of dependencies of fault can be classified into three types according to digraph theory, and many failure-node graphs can integrate into a multi-failure-node graph. Moreover, the decreasing emerging ratio of dependent faults is analyzed. Supported by the scientific Research Foundation of Harbin Engineering University (HEUST07021), the authors have studied the differences of testing environment and operational environment. This paper focuses on the evaluation accuracy and prediction precision of software reliability growth model incorporating fault dependency. The authors have worked on the evaluation algorithms for fault-tolerance and dependability. Some progress has been achieved. Furthermore, they will develop evaluation algorithm for dependability of very large scale network system.