Improved Multistep Model Predictive Control for Nonlinear Networked Systems With Stealthy Attacks and Actuator Saturation

In this paper, an improved multi-step model predictive control (MPC) algorithm is investigated for nonlinear networked control systems (NCSs) subject to stealthy denial-of-service (DoS) attacks and actuator saturation. First, the impact of stealthy DoS attacks on data transmission is quantitatively characterized through a Signal to Interference plus Noise Ratio (SINR) metric, while the packet dropouts of the corresponding channel are described by a Bernoulli random process. Actuator saturation is presented via a convex hull representation and explicitly addressed by incorporating saturation constraints into the MPC optimization problem. An interval type-2 Takagi-Sugeno (IT2 T-S) fuzzy model is integrated to robustly handle system nonlinearities and model uncertainties. Second, the improved multi-step strategy employs intersecting multiple ellipses to construct the initial feasible set for multi-step control laws, simultaneously enlarging the initial feasible set, improving the control performance, and reducing conservatism. Under the framework of the improved multi-step strategy, a model predictive controller is constructed via designing a sequence of feedback control laws before the fixed control law, driving the state from the initial feasible set to the neighborhood of zero. The recursive feasibility and closed-loop stability are demonstrated. Finally, simulation results prove the effectiveness of the algorithm under stealthy DoS attacks and actuator saturation.