Considering the Dependency of Fault Detection and Correction in Software Reliability Modeling

Abstract

Most existing software reliability growth models (SRGMs) focused on the fault detection process, while the fault correction process was ignored by assuming that the detected faults can be removed immediately and perfectly. However, these assumptions are not realistic. The fault correction process is a critical part in software testing. In this paper, we studied the dependency of the fault detection and correction processes in view of the number of faults. The ratio of corrected fault number to detected fault number is used to describe the dependency of the two processes, which appears S-shaped. Therefore, we adopt the logistical function to represent the ratio function. Based on this function, both fault correction and detection processes are modeled. The proposed models are evaluated by a data set of software testing. The experimental results show that the new models fit the data set of fault detection and correction processes very well.