A Generalized Software Fault Classification Model

Most non-homogenous Poisson process (NHPP) based software reliability growth models (SRGMs) presented in the literature assume that the faults in the software are of the same type. This assumption implies that the fault removal rate per remaining faults is independent of the testing time. However, this assumption is not truly representative of reality. It has been observed that the software contains different types of faults and each fault requires different strategies and different amount of testing-effort to remove it. This paper proposes a generalized model based on classification the faults in the software system according to their removal complexity. The removal complexity is proportional to the amount of testing-effort required to remove the fault. The testing-effort expenditures are represented by the number of stages required to remove the fault after the failure observation / fault isolation (with time delay between the stages). Therefore, it explicitly takes into account the faults of different severity and can capture variability in the growth curves depending on the environment it is being used and at the same time it has the capability to reduce either to exponential or Sshaped growth curves. Such modelling approach is very much suited for object-oriented programming and distributed development environments. Actual software reliability data have been used to demonstrate the proposed generalized model. Key-Words: software engineering, software testing, NHPP, SRGM, fault severity.