Unknown Input Observers for Discretized Systems With Application to Networked Systems Corrupted by Sparse Malicious Packet Drops

Abstract

Digital estimators for networked systems are proposed where the plant output measurements and control signals are corrupted by independent sparse malicious packet drop errors. The estimators comprise output sparse error approximators and unknown input observers (UIOs). The advantage of using digital estimators is that in many cases the condition for the existence of an UIO fails for continuous-time (CT) plants while it holds for discretized plants. A class of CT systems for which the existence condition for the UIO holds in the discrete-time (DT) domain is characterized. The proposed digital estimators are tested on a nonlinear networked control system (NCS). As part of our analysis, we study the existence of unknown input observers for discretized systems and show that discretization does not negatively impact the existence of UIOs and in some cases can facilitate the construction of UIOs when they do not exist for the CT systems.