Robust Output Feedback MPC for Networked Control Systems with Two-Channel Random Packet Dropouts

In this paper, we focus on the robust output feedback Model Predictive Control (MPC) design for linear constrained Networked Control Systems (NCSs) subject to disturbances, observation noise and random packet dropouts in both Sensor-Controller (S-C) and Controller-Actuator (C-A) channels. The proposed control scheme consists of an observer to estimate the state and a robust model predictive controller to stabilize the disturbed system. In the observer design, we extend the Luenberger observer to estimate the state in two communication scenarios. The resulting dynamics of estimation error can be described by a switched system. With this, a Generalized Robust Positive Invariant (GRPI) set can be developed, providing an explicit bound of estimation errors in the presence of admissible disturbances and packet dropouts. Similarly, a GRPI set is established to bound the prediction error in the MPC framework under the proposed state estimator. These two GRPI sets are further used to develop tightened constraints in the proposed robust output feedback MPC scheme to ensure robust constraint satisfaction. It is rigorously proved that the proposed robust MPC algorithm is recursively feasible and the system state converges to a compact set around the origin. Finally, simulation results are provided to verify the effectiveness of the proposed robust output feedback MPC scheme.