DoS attack in centralised sensor network against state estimation

Abstract

This study considers a model that a centralised sensor network is attacked by an invader, who launches the denial-of-service (DoS) attack on the network. To understand the behaviour of the invader and propose necessary protection accordingly, the authors study how the invader optimises his attack. In the model, the sensors take a measurement of a process and send the measurements to a remote estimator for state estimation. The invader intends to block the communication channels from the sensors to the estimator by the DoS attack in order to degrade the estimation performance. Constrained by a power budget, the invader needs to decide which sensors and at which time instances to attack, with the target that the estimation performance is mostly deteriorated. In this study, two scenarios that the system has a single sensor and has multiple sensors are investigated, respectively. For the system with a single sensor, the analytical result of the optimal attack schedules is given. For the system with multiple sensors, numerical methods are proposed, where the problem is relaxed and transformed into a convex optimisation problem which can be solved by efficient numerical algorithms.