Optimal Stealthy Attack on Cyber-Physical Systems and Its Application to a Networked PMSM System

Abstract

This paper investigates the stealthy attack on cyber-physical systems (CPSs) under the stochastic communication protocol (SCP), which is used to avoid data collisions. The attack stealthiness is characterized by the Kullback-Leibler divergence (KLD) and the attack performance is measured by the corrupted error covariance. This paper aims to maximize the attack performance for a certain stealthy level. Different from the existing results, our proposed attack strategy takes the SCP-introduced effects into account and exploits incomplete measurement outputs at each time instant to construct attack model. Then, a non-convex optimization problem is formulated to achieve the desired attack performance for a given stealthy level. By applying the monotonic optimization technique, the non-convex problem is converted to two optimization problems with equality constraints. Then, the analytical expressions of the attack parameters are derived. Finally, simulations on a three-tank system and experiments on a networked permanent magnet synchronous machine (PMSM) prototype system are conducted to validate our proposed theoretical results.