Entropy as a Metric in Critical Infrastructure Situational Awareness

In this paper, we expand our previously proposed critical infrastructure (CI) model with time dependent stochastic elements. In the model, CI is presented as a directed graph where each vertex represents a discrete system and directed edges dependency relations between the systems. Each node is associated with a finite state machine (FSM) which represents the operational status of the system in question. In this paper we associate a probability distribution to each FSM, which accounts for the flow of time and previous confirmed sensor reading. As time passes, the uncertainty about the state of the system increases. By relying on statistical probabilities that have been previously observed or known, it, is possible to make predictions about the current state of Cl. We present a dependency graph modelling a subset of Finnish electric grid and mobile networks. CI components are modelled using FSIVI structure augmented by probabilistic elements for entropy-based calculations. The proposed model provides an estimate about the state of the critical infrastructure when only limited information is available, while taking into account the increasing uncertainty created by the passage of time.