Comparisons of nonhomogeneous Poisson process software reliability models and its applications

Abstract

In this paper, existing software reliability models based on an nonhomogeneous Poisson process (NHPP) are summarized. A recently developed NHPP model with imperfect debugging and time-dependent fault detection rate is examined. T he results show that this proposed model provides improved descriptive and predictive power when compared with other existing NHPP models. T he model can also be incorporated into a software cost model for reliability assessment. T he optimal release policies which minimize the expected total software cost is also presented. Nomenclaturea…t † time-dependent fault content function b…t † fault detection rate per unit of time C 1 testing cost coe cient, that is software testing cost per unit time C 2 error removal cost coe cient, that is cost of removing a fault per unit time during testing phase C 3 risk cost coe cient, risk cost due to software failure m…t † mean value function, that is expected number of faults detected by time t m…t k † estimated cumulative number of failures at time t k obtained from the model R…x=t † software reliability function, that is conditional probability that no faults occur during …t;t ‡ x † given that the last failure occurred at time t y k total number of failures observed at time t k according to the actual data P k sum over k from 1 to n

Reliability analysis and optimal version-updating for open source software

“…Given different expressions and explanations of a(t) and b(t), many NHPP SRGMs can be developed [15]. The basic assumption illustrated by the above formulation can also hold in the context of OSS [9].…”

“…The total expected cost including both testing cost and operation cost is a crucial factor for such determination [14][15][16]. In fact, the software cost estimation is also important in software development [17].…”

Integration of Imperfect Debugging in General Testing-Domain Dependent NHPP SRGM

“…MVF's are obtained and tabulated in Table 1 for the fault detection rate functions appeared in the literature. [1][2][3][4][5][6][15][16][17][18][19][20] The MVF's in Table 1 are superscribed with i in order to distinguish them from the MVF's of the general T-D dependent NHPP SRGM, 13 which are denoted by m j (t). The fault detection rate functions in Table 1 can be classified into two categories: The scale-variant fault detection rate function and the scale-invariant fault detection rate function.…”

Multi-software reliability allocation in multimedia systems with budget constraints using Dempster–Shafer theory and improved differential evolution

“…In order to achieve the reliability goal of software, many researchers have devoted to various software reliability growth models (SRGMs) [3][4][5][6]. However, accurate software reliability estimation is not usually available until the software has been tested by a large number of failure data for a longer period of time.…”