Modeling and analysis of software fault detectability and removability with time variant fault exposure ratio, fault removal efficiency, and change point

Abstract

Software reliability growth models have been proposed to assess and predict the reliability growth of software, remaining number of faults, and failure rate. In previous studies, software faults have been mainly categorized into two categories based on its severity in removal process: simple faults and hard faults. In reality, fault detectability is one of the crucial factors which can influence the reliability growth of software. The detectability of a software fault depends on how frequently the instructions containing faults are executed. However, fault removability of a software fault depends on fault removal efficiency of debugging team. The main motive of this article is to incorporate the fault detectability in software reliability assessment. Fault exposure ratio is an essential factor for software reliability modeling that controls the per-fault hazard rate. It is strongly dependent on fault detectability. In this article, the effect of fault detectability, fault removability, fault exposure ratio, and fault removal efficiency has been considered simultaneously in software reliability growth modeling. Moreover, a logistic fault exposure ratio has been introduced. The effect of change point is incorporated in the proposed software reliability growth model. Two illustrative examples with software testing data have been presented.

Software reliability and cost models with warranty and life cycle

“…Zhu and Pham 13 introduced a two-phase fault removal process by categorizing faults into independent and dependent types with imperfect fault removal phenomenon. Chatterjee et al 14 proposed a framework to develop fault detection and correction model with time-dependent fault introduction rate, fault removal rate and change point. Later, Chatterjee and Shukla 15 developed a unified approach to develop testing coverage-based SRGM by incorporating fault detection probability, imperfect debugging, and change point.…”

Modelling software reliability growth through generalized inflection S-shaped fault reduction factor and optimal release time

“…Chatterjee and Shukla 43 also proposed fault identification and correction process SRGM for reliability estimation. Chatterjee et al 6 developed SRGM for reliability assessment with the incorporation of fault detectability. The fault exposure ratio, fault removability and fault removal efficiency have been studied together during the formulation of SRGM.…”

“…4 Over the past four decades, many SRGMs have been introduced for fault prediction and reliability growth estimation of software. 5,6 Several studies on software fault prediction have recently gained much attention because they assist the software testing team in improving the software system's quality and productivity. Since the software errors or faults can be detected and corrected, the SRGMs measure the reliability growth.…”

“…Some researchers have attempted to explain the dependency among FDP and FCP by various time lag functions such as exponential, linear and constant. 6,20 They believe that fault correction begins when faults are detected.…”

Software reliability prediction: A survey

“…In the design and implementation phase, which deals with code and its features, supervised methods and fuzzy inference system are the basic methods for evaluating reliability, and in the test phase, if the failure data is based on the number of failures observed in each time period or is recorded according to the time between failures, statistical models indicating the growth rate of reliability can be used to identify the data recording process 21,22 . These are known as software reliability growth models (SRGM) or growth models in general, 23–28 , which are used for both prediction and assessment of software reliability.…”