Parametric Design of Resilient Complex Networked Systems

Huy T. Tran, Jean Charles Domercant, Dimitri N. Mavris

Research output: Contribution to journalArticle

Abstract

Engineered systems are becoming increasingly complex and interconnected, requiring consideration of their resilience to potential network threats. We therefore explore a parametrically defined network design space to further understand properties and design choices that promote resilience in complex networked systems. This design space focuses on three aspects of a network: its initial topology, adaptation strategy, and threat faced. We assess network resilience using complex network properties and a resilience engineering approach. The proposed methods are applied to an agent-based information exchange network simulation, developed to represent a general class of networked systems relying on message passing among nodes. We find that resilient networks tend to have more random, rather than scale-free initial topologies, regardless of the threat faced. However, the optimal adaptation strategy shows a sharp transition from fully random to fully degree-based once a critical point is reached for threat randomness. These results demonstrate the value of thoroughly exploring the network design space, expanding upon insights gained from existing point design analyses. This work further confirms the importance of understanding fundamental behaviors of complex networked systems.

Original languageEnglish (US)
Article number8353456
Pages (from-to)1496-1504
Number of pages9
JournalIEEE Systems Journal
Volume13
Issue number2
DOIs
StatePublished - Jun 2019
Externally publishedYes

Keywords

  • Agent-based modeling (ABM)
  • complex networks
  • complex systems
  • network design
  • resilient systems

ASJC Scopus subject areas

  • Control and Systems Engineering
  • Information Systems
  • Computer Science Applications
  • Computer Networks and Communications
  • Electrical and Electronic Engineering

Fingerprint Dive into the research topics of 'Parametric Design of Resilient Complex Networked Systems'. Together they form a unique fingerprint.

  • Cite this