On sparse controllability of graph signals

Controlling the behavior of a signal defined over a graph by acting on a limited set of nodes is a problem that finds application in many fields. In this paper, we merge recently developed tools in graph signal processing with control theory of complex networks and consider the reconstruction of bandlimited graph signals from their samples through a diffusion process properly driven by a subset of control nodes. Then, we propose an optimization algorithm aimed at minimizing the control energy incorporating a regularization term whose goal is to promote sparsity across time and nodes jointly.