Evaluation of Node Importance Based on Information Theory in Cyber-Physical System

The Cyber-physical system (CPS) will soon become the basic technology of contemporary industry, which consists of two interacting networks where the cyber network overlays the physical network. Due to the high interdependence, the CPS are more fragile in general when facing attacks. In order to avoid the occurrence of catastrophic accidents while taking into account the cost of construction, it is critical to find and protect key nodes whose failure will lead to serious cascading failures. In this paper, we present a way to evaluate the node importance in CPS based on information theory, where node states are treated as discrete random variables, and their importance is interpreted as the information gain. After research, we find that self-information of a node state can reflect its ability to influence the whole system. In particular, the result of this method is affected by the probability of attacking nodes, which helps to assess the node importance in different scenarios and environments.