Resilient Tube-Based Model Predictive Control With Polytopic Constraints Subject to DoS Attacks

Tube-based model predictive control (TMPC), a well-performed control algorithm in disturbed control scenarios, is widely applied in networked control systems (NCSs). Due to the inherent nature of communication networks, NCSs always are at risk of being attacked. Consequently, the pristine TMPC-based systems may undergo severe destruction. However, the field about resilience of this control method against cyber-attacks in communication links is still in its infancy. In this article, we consider a resilience scheme based on an actuator buffer to eliminate the harmful effects caused by denial-of-service attacks. In addition, we assess the inherent attack tolerance of TMPC by calculating the maximal permissible open-loop steps, the maximal steps with no feedback control while provably maintaining recursive feasibility and input-to-state stability, for arbitrary states. The effectiveness and merits of the proposed resilience scheme and algorithm are demonstrated through numerical comparisons, furthermore, implemented on a three-wheeled omnidirectional robot platform.