Elemental patterns of verification strategies

Abstract

This paper proposes a set of seven elemental patterns of verification strategies. These patterns can be useful in modeling verification strategies in a wide range of engineered systems. They form the building blocks under which any verification strategy can be modeled. The patterns lead to a fundamental understanding of the interplay between system parameters and verification activities, as well as an understanding of the mechanisms by which verification evidence builds up. For each pattern, we provide a description and a few examples of its application. A few important theoretical properties of the corresponding set of patterns are discussed, such as belief update, inferential properties, and graph disconnection, as well as some practical guidance to be taken into account when applying them to authentic verification problems. These patterns are intended to be a useful tool for researchers, practitioners, and educators, by formalizing the application of Bayesian networks to verification problems, hence facilitating instruction and communication among verification engineers and with researchers from other domains, particularly statisticians and Bayesian analysts.

An evaluation of the optimality of frequent verification for vertically integrated systems

“…To address these limitations, we derive optimal verification strategies using a belief‐based approach to model the system development process. To the best of our knowledge, only the recent works by Salado et al 3,39–41 have adopted the approach of capturing the epistemic uncertainty in the design process by using belief distributions to model verification strategies. In their work, verification strategies are derived based on the organization's changing belief in the system design meeting the system requirements.…”

A Concept for Set‐based Design of Verification Strategies

“…In this paper, a verification strategy is understood to be a set of verification activities organized as an acyclic directed graph where the verification activities are modeled as nodes and the edges represent their information influence (Salado and Kannan, 2018a). A verification activity is understood to be the collection of information about a specific aspect of the system under development and verification evidence refers to such information (Salado and Kannan, 2019).…”

“…A verification strategy can be modeled as a Bayesian network to capture the way engineers build confidence on the state of the system as verification evidence becomes available (Salado and Kannan, 2019). The basic structure of such Bayesian model is given by three subgraphs (Salado and Kannan, 2019):…”

“…Probability assignments use synthetic data and are given in the Appendix. Following the modeling approach presented in (Salado and Kannan, 2019), prior beliefs are assigned to system parameter nodes, which capture the initial belief on the state of the system (i.e., being absent of errors), and conditional probability tables are created for the verification activity nodes. Posterior beliefs are calculated for system parameters through Bayesian update of the outcomes of the verification activity nodes.…”

Coordination of verification activities with incentives: a two-firm model

“…As a result, verification activities consume a significant amount of resources during systems' lifecycle throughout the industry (Tahera et al 2017). Hence, the importance of discovering the scientific foundations of verification activities in systems engineering has been recognized by multiple authors, and research on developing robust decision-making frameworks for verification activities is an active research area (Engel and Barad 2003;Shabi and Reich 2012;Salado and Kannan 2019;Xu and Salado 2019).…”