Stability of Nonlinear Model Predictive Control Under Consecutive Packet Losses

In this paper, we study the stability of a discrete-time nonlinear networked system controlled by model predictive control without explicit terminal constraints. The system is subject to input constraints and random packet losses on the actuation communication channel between the controller and the plant. We consider the scenario where a buffer-intended to store transmitted control sequences and provide some robustness to packet losses-is present, and also the scenario where a buffer is not present. We analyse the stability of the closed-loop system in these stochastic scenarios, employing the assumption that the terminal cost is merlye a local, rather than global, control Lyapunov function (CLF). We develop conditions that characterize an upper bound on the number of consecutive packet losses in order that stability is maintained.