Energy Balance Based Attack Detection for Cyber Physical Systems

Abstract

In this article, the deficiency of a widely-used observer-based attack detection scheme for cyber physical systems (CPS) is analyzed and then a novel dissipativity-based attack detector is proposed. Firstly, the definition of kernel attacks is introduced to capture the stealthy attacks which can not be detected through well-established observer-based techniques. Further, a data-driven design of kernel attacks based on subspace identification methods deploying input and output data is presented. To detect kernel attacks, a dissipativity-based attack detector is formulated in light of energy balance equation, providing a complementary strategy to guarantee reliable detection of attacks. Finally, the validity of the presented approach is demonstrated in a numerical example.