The interdependencies among critical infrastructure (CI) systems create a new layer of vulnerability that has not been anticipated. Results obtained from recent studies trying to address the issue in this area are either too abstract for engineering studies or require extensive computational efforts. Therefore, a practical and scalable tool to study large and complex interconnected CI systems needs to be developed. This thesis identifies the most appropriate methodology for interconnected CI system modelling and proposes a modelling framework based on Complex Network Theory (CNT) to holistically model the intra-/inter- relationships within/between CI systems. The significance of this research is that it is able to identify the criticality of components with regard to the vulnerability of interconnected CI systems with uncertainty. This research contributes to the following seven areas. First, based on the CNT, a three-dimensional model has been developed for the interconnected power grid and Information and Communication Technology (ICT) network with both unidirectional and bidirectional (inter)dependencies. Then CNT indices have been implemented to identify the critical components in the system under different operating conditions. Second, the three-dimensional model is extended to a weighted form to account for both systemsâ engineering parameters. Third, a Vulnerability-weighted Node Degree (VWND) combining Degree and Efficiency is proposed to comprehensively evaluate the dependency and importance of each system component. The proposed modelling framework and indices have been applied to a 14-bus test network with supporting ICT infrastructure, and results suggest that VWND, compared to conventional CNT indices, is more effective in identifying critical interconnection nodes (i.e., nodes are critical for the interdependency of the interconnected CI system). Fourth, a Weighted Small-Worldness (WSW) index is proposed to extend the conventional Small-Worldness (SW) index to account for real-world engineering behaviour. Fifth, using a 1326-bus power system with 88-node ICT test network, Pearson and Spearmanâs correlation studies are performed and identified Degree Centrality to be the most robust index among all the CNT indices. Sixth, a two-step CNT modelling framework has been developed considering cyber componentsâ reliability and operational time constraints. Finally, the impact of the uncertainties of cyber componentsâ reliabilities and latencies on the overall reliability of the cyber network is studied, and the criticality of cyber components in different ICT network hierarchical layers is identified.
|Date of Award
|31 Dec 2019
- The University of Manchester
|Jovica Milanovic (Supervisor) & Luis(Nando) Ochoa (Supervisor)
- Complex Network Theory
- Critical Infrastructure System
- Cyber-physical System