Optimising trade-offs between cyber reliability and cyber-physical risk for water distribution systems

• Novel analysis of trade-offs between cyber-physical risk and cyber-reliability • Introduction of a practical Cyber Connectivity metric to aid decision making • Application to two cyber-physical water systems, C-Town and Richmond • Novel process to identify redundancy strategies in cyber-physical water systems • Development of an optimisation framework for this risk/reliability trade-off Water distribution systems (WDSs) are critical for human survival. The ability to control these systems remotely via cyber networks has made it possible to increase their functionality significantly. The degree of effectiveness of this control is a function of the reliability of the cyber network (i.e. cyber reliability), which can be increased via the addition of redundant cyber components. However, this also has the potential to reduce WDS reliability by increasing opportunities for cyber-attacks (i.e. increasing cyber physical risk). Consequently, it is imperative to consider the potential trade-off between increasing cyber reliability and decreasing cyber-physical risk in the design of WDSs. In this paper, we introduce novel approaches for quantifying and optimising this trade-off, including a novel cyber-connectivity metric for estimating WDS cyber-physical risk in a practically efficient manner. Application of our approaches to two complex WDSs shows that the degree to which an increase in cyber network reliability increases cyber-physical risk is highly variable, depending on the configuration of the redundant cyber components. This suggests that there are opportunities to design cyber networks in a way that minimises increases in cyber-physical risk for a given increase in cyber reliability. The ability of our proposed optimisation framework to achieve this is demonstrated for the two WDSs considered. While the quantification and optimisation approaches presented in this paper are tailored to WDSs, the underlying principles are also applicable to cyber-physical systems in other problem domains.