Proactive attack detection scheme based on nonlinear transformation and moving target defense*

This paper proposes a novel active attack defense strategy for cyber-physical systems(CPS) with unknown-but-bounded (UBB) noises. A resourceful and veteran adversary can design a sequence of perfect injection on channel of sensors and remain undetected. To confront such an attacker, a hybrid attack detection approach is designed by combining moving target(MT) and nonlinear transformation to change the original output to increase the uncertainty of the system and the estimation residues under stealthy false-data-injection (FDI) attacks. Furthermore, the attack detection rate can be improved by the hybrid attack detection approach. Finally, an unmanned aerial systems as an example is provided to demonstrate the effectiveness of the proposed method.