Model Predictive Control for Networked Control Systems with Double-Sided Packet Loss

In this paper, a Model Predictive Control (MPC) strategy for networked control systems with double-sided packet loss is proposed. A predictor and a buffer are respectively designed in controller and in actuator. If data packet dropout happens between sensor and controller at one sampling time, the predictor will work to predict current state of system and control actions will be computed using this predict state. The buffer is used to store newest control sequence transmitted successfully from controller to actuator. When control sequence is lost at one sampling time, the control sequence corresponding to current time in buffer will be applied to actuator. A sufficient condition guaranteeing stability of closed loop system is also provided. A simulation example illustrates the effectiveness of the proposed approach.