Discrete-Time Linear Dynamical System Control Using Sparse Inputs With Time-Varying Support

In networked control systems, communication resource constraints often necessitate the use of \emph{sparse} control input vectors. A prototypical problem is how to ensure controllability of a linear dynamical system when only a limited number of actuators (inputs) can be active at each time step. In this work, we first present an algorithm for determining the \emph{sparse actuator schedule}, i.e., the sequence of supports of the input vectors that ensures controllability. Next, we extend the algorithm to minimize the average control energy by simultaneously minimizing the trace of the controllability Gramian, under the sparsity constraints. We derive theoretical guarantees for both algorithms: the first algorithm ensures controllability with a minimal number of control inputs at a given sparsity level; for the second algorithm, we derive an upper bound on the average control energy under the resulting actuator schedule. Finally, we develop a novel sparse controller based on Kalman filtering and sparse signal recovery that drives the system to a desired state in the presence of process and measurement noise. We also derive an upper bound on the steady-state MSE attained by the algorithm. We corroborate our theoretical results using numerical simulations and illustrate that sparse control achieves a control performance comparable to the fully actuated systems.