Test based safety-critical software reliability estimation using Bayesian method and flow network structure

Abstract

System safety is closely related to system reliability. Safety requirements many times are translated to reliability requirements. Nowadays, software systems exist in many engineering systems. However, there is no consensus method for software reliability estimation. On the contrary, there is an increasing interest in estimating the software reliability due to concerns for safety-critical systems. In this article, we try to close the gap by proposing a systematic and probabilistic method to estimate the software reliability based on software test data.

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

“…Another relevant work is [2], in which the author takes into account the structure of software code (as a "flow network structure") with a claim that this method is a significant improvement over the alternatives, which disregard the control flow. A similar work was conducted by May et al, e.g.…”

“…In other literature sources, e.g., dealing with software testing, the term "white-box" is often used to refer to different degrees of knowledge about the tested software, e.g., whether the source code is available or not, etc. [2] makes use of the structure of the code to develop a Bayesian method of reliability assessment.…”

An AdaBoost-based tree augmented naive Bayesian classifier for transient stability assessment of power systems

“…However, the artificial intelligence method does not consider the operating mechanism of the power system, which replaces the mechanism model with a purely data-driven approach. The Bayesian network method combines the advantages of the datadriven and mechanism model methods [15][16][17] ; that is, the Bayesian network method for assessing the transient stability of the power system is not only a data-driven method but also considers the causal relationships between the attributes of the power system. The tree augmented naive Bayesian (TAN) classifier is an application of the Bayesian network, and one of the biggest advantages of the AdaBoost-based tree augmented naive Bayesian (ATAN) classifier is that the TAN classifier allows specifying the structure that represents domain knowledge in the form of variable relationships.…”