Topology change detection in networks due to false data injections: a priori assessment based on graph matching techniques

Abstract

This paper analyzes how to detect changes in the topology of a network under a false data injection (FDI), when an intruder inserts false information into the system so as to modify the structure of the adjacency matrix describing the connections in the network. Detecting an FDI is of paramount importance for maintaining the resilience and performance of a network, whether it is a computer network, power system, or any other interdependent systems. Indeed, an FDI can cause incorrect decisions or actions to be taken based on inaccurate information, leading to operational errors and potential disruptions. For example, detecting false data injection helps preserve the integrity of the energy supply of an electric power system. Among the possible types of FDI, we are interested in the assessment of a special type of attack that shuffles the labels of the nodes of the network causing a permutation on the adjacency matrix, thereby giving the false impression that the network has a topology different from reality. We propose a simple but effective iterative procedure based on the use of graph matching algorithms to derive a possible perfect correspondence between the nodes of the actual topology and the (false) perturbed topology. The R packages igraphMatch and netcom were used for the evaluation. The procedure is applied to two different topologies with promising results.