A Pilot Study on Ranking the Critical Components of a System

Abstract

Components of complex systems are interdependent in various dimensions including physical, functional, and social dimensions. Identifying and understanding critical components of a system have many implications in terms of system analysis, design, and operation. Systems consist of many components some of which are more significant than the others. A component is deemed significant if it is part of the structural core or architecture of a system, and proper operation of the rest of the system relies on the correct functioning of this component. Identifying such significant components is important since change is inevitable as a normal course of evolution in any system, and critical components of the systems must not only be designed to be reliable but also designed to minimize their impact on change. Testability of these components is also critical for ease of diagnosis and troubleshooting for quality assurance. In addition, complex systems carry socio-technical characteristics where understanding critical elements of an organizational structure is important for successful system design and operation. Several different component ranking strategies exist that rank components using their own unique perspective. As such, these ranking strategies behave differently in their ranking of significant components within a system. In this paper, we propose a discrete-time Markov chain-based algorithm for ranking the components of the system and compare its ranking with three commonly used ranking strategies, namely, closeness, betweenness, and eigenvector centrality. In our pilot study on three different systems, Markov chain outperformed other ranking strategies in highly ranking the significant components of the system.