Interval Observers for Secure Estimation in Cyber-Physical Systems

Kwassi Holali Degue (1) , Denis Efimov (2) , Jerome Le Ny (1) , Eric Féron (3)

1 EPM - École Polytechnique de Montréal
2 NON-A-POST - Non-Asymptotic estimation for online systems
3 ENAC - Ecole Nationale de l'Aviation Civile

Stealthy attacks on the sensors and actuators embedded in cyber-physical systems could hinder the safe operation of these systems if the state estimators monitoring them cannot detect such attacks in time. In this paper, we study stealthy attacks in the framework of interval observers. We consider two classes of attacks: when a malicious agent compromises the sensors and when it is able to alter the system's actuators. For each type of attack, we design a dedicated interval observer for the system's state and we construct bounds for the attack signal. We investigate the ability of such interval observer to provide accurate estimates when the system is under the attack. Numerical simulations for a lateral model of an aircraft illustrate the capabilities of the synthesized observers.

Domaines: Automatique / Robotique

Kwassi Holali Degue, Denis Efimov, Jerome Le Ny, Eric Féron. Interval Observers for Secure Estimation in Cyber-Physical Systems. CDC 2018 - 57th IEEE Conference on Decision and Control, Dec 2018, Fontainebleau (FL), United States. ⟨hal-01888558⟩