Optimal sensor scheduling under intermittent observations subject to network dynamics

TL;DR

This paper develops optimal sensor scheduling strategies for linear Gaussian systems over dynamic lossy networks, ensuring stability and near-optimal performance in various cost settings.

Contribution

It characterizes stationary optimal policies for different cost criteria and extends stability conditions to multidimensional measurement loss rates in dynamic networks.

Findings

Convergence of value iteration to average cost solutions.

Near-optimal policies via rolling horizon truncation.

Extended stability conditions for multidimensional loss rates.

Abstract

Motivated by various distributed control applications, we consider a linear system with Gaussian noise observed by multiple sensors which transmit measurements over a dynamic lossy network. We characterize the stationary optimal sensor scheduling policy for the finite horizon, discounted, and long-term average cost problems and show that the value iteration algorithm converges to a solution of the average cost problem. We further show that the suboptimal policies provided by the rolling horizon truncation of the value iteration also guarantee stability and provide near-optimal average cost. Lastly, we provide qualitative characterizations of the multidimensional set of measurement loss rates for which the system is stabilizable for a static network, significantly extending earlier results on intermittent observations.