Energy-Aware Radio Chip Management for Wireless Control

This paper addresses energy-aware control in a wireless control system. The goal is to save energy in a smart sensor node by codesigning its use of the radio chip together with the control policy. The focus is on exploiting the fact that in many radio chips used in wireless nodes the depth of sleep can be manipulated dynamically. The choice of the mode involves a tradeoff, since deeper sleep is cheaper to stay in, but leads to higher future transition costs. The opportunities arising from the use of multiple nontransmitting radio modes have not been studied in the systems control literature. This paper presents a first step toward filling this gap. We propose a control scheme that manages the radio modes of a single sensor node with computation capabilities. This smart sensor node is in charge of sensing the system state, computing the control law to be sent to the actuator, and managing its own radio chip. The joint optimization problem of finding the best switching policy for the radio-chip mode and the best feedback control law is obtained off-line using dynamic programming. The optimal policy is formulated over a finite horizon and implemented according to the model predictive control paradigm. Practical stability of the resulting control scheme is assessed, and careful simulation studies document the potential energy savings that can be obtained with this technology.