Stable stochastic predictive controller under unreliable up-link

This article presents a systematic overview of stable stochastic predictive control, its underlying concepts, and a transmission protocol which does not require any computational or storage facility at the actuator end. We assume that the control channel has i.i.d. packet dropouts. Our results ensure mean square boundedness of the states of the closed-loop plant for any positive bound on control in the presence of potentially unbounded additive noise. The proposed control mechanism stabilizes the system for any non-zero probability of successful transmission.