City Research Online

Conservative reliability assessment of a 2-channel software system when one of the channels is probably perfect

Popov, P. T. ORCID: 0000-0002-3434-5272 (2021). Conservative reliability assessment of a 2-channel software system when one of the channels is probably perfect. Reliability Engineering and System Safety, 216, 108008. doi: 10.1016/j.ress.2021.108008


In this paper we subject to scrutiny some recent advances in conservative reliability assessment of 2-channel fault-tolerant software, based on the probability of perfection of one of the channels. Our approach extends the previous works by looking in detail at the implications of the assumptions made in these previous works about the relationships between the probability of failure of the channels and of the system, which have not been explored before. We demonstrate that the assumptions made by others impose significant constraints on the epistemic uncertainty of the probability of system failure and explore the implications of these constraints to derive new conservative bounds. An important difference of this work from the prior works is that we use a white-box model of a 2-channel system, while in the previous works a black-box system model was used. We discuss the limitations of an assessment based on a black-box model and compare our conservative results with those, derived by others using a black-box system model.

Publication Type: Article
Additional Information: © 2021. This manuscript version is made available under the CC-BY-NC-ND 4.0 license
Publisher Keywords: Bayesian inference, protection system, two-channel software system, black-box and white-box models, probability of failure on demand, probability of perfection, epistemic uncertainty
Subjects: H Social Sciences > HF Commerce
Q Science > QA Mathematics
Departments: School of Science & Technology > Computer Science
Text - Accepted Version
Available under License Creative Commons Attribution Non-commercial No Derivatives.

Download (860kB) | Preview



Downloads per month over past year

View more statistics

Actions (login required)

Admin Login Admin Login