A Stochastic Model for Software Reliability Assessment by Considering Time-Delay

• Main Authors: Rao Guo-Yu, 饒國煜
• Other Authors: Wang Rong-Tsorng
• Format: Others
• Language: en_US
• Published: 2014
• Online Access: http://ndltd.ncl.edu.tw/handle/21832020053137011223

碩士 === 東海大學 === 統計學系 === 102 === This thesis studies a stochastic model based on non-homogeneous Poisson process (NHPP) for software reliability evaluation when there is time-delay between the fault detection and the fault correction processes. During the past 30 years, many researcher have used stochastic processes to describe software reliability growth model (SRGM). These models have been applied to evaluate software development status during testing phase. Many SRGMs assume that a detected fault will be eliminated immediately. In reality, a detected fault may not always be removed at once, because the correction process depends on the difficulty of the fault, the skill level of engineer, and other factors. Therefore, there is time-delay between the fault detection process and fault correction process. In addition, the fault correction process may result in imperfect debugging. This imperfection will make the number of faults to increase or unchanged. In this thesis, the imperfect debugging NHPPs and queue theory are incorporated to describe the time-delay of the fault detection process and fault correction process. This thesis considers infinite server queue imperfect debugging (ISQ-ID), finite server queue imperfect debugging (FSQ-ID) and one server queue imperfect debugging (OSQ-ID) models. We use maximum likelihood (ML) approach to estimate parameters of the underlying mean value functions. To verify the proposed model, we analyze two actual data sets based on the proposed model. We compare results with several common SRGMs.