Compensator Synthesis for Unsecured Cyber-Physical Systems

A controller-observer compensator is proposed for Cyber-Physical Systems (CPSs) subject to sensor and actuator faults and communication network disturbances. The observer combines a norm approximator for sparse malicious attacks recovery with a bank of unknown input observers (UIOs) to estimate the CPS state. Convergence analysis of the state estimation error of the proposed UIO architecture is given. To enhance the observer's performance, sensor and actuator fault filters constructed using error correcting code (ECC) techniques are used. A model reference controller with a computable performance level is given. A simulation of the controller-observer compensator in a self-driving ground vehicle is used to demonstrate its effectiveness.