Secure estimation in fuel cells: Eavesdropping prevention and attack detection performance

Abstract

This work presents a nonlinear masking signal protocol to privatize data against eavesdropping attacks in a nonlinear remote estimation problem. The protocol is based on the addition of a masking signal at the source node and a removal filter at the end node. The stability and robustness of the overall architecture are formally proven through incremental passivity arguments. Then, the effects of such protocol on the estimation, attack detection, and closed-loop observer-based control are discussed. Finally, such methodologies are applied in the state estimation scheme of a proton exchange membrane fuel cell model through numerical simulations and experimental validations.