Driver and Sensor Node Selection Strategies Optimizing the Controllability Properties of Complex Dynamical Networks

In recent years, complex networks have attracted the attention of researchers throughout the fields of science due to their ubiquity in natural and artificial settings. While the spontaneous emergence of collective behavior has been thoroughly studied, and has inspired researchers in the design of control strategies able to reproduce it in artificial scenarios, our ability to arbitrarily affect the behavior of complex networks is still limited. To start filling this void, in the past five years, researchers have focused on the preliminary condition of selecting the nodes where input signals have to be injected so to ensure complete controllability of complex networks. Unfortunately, the scale of complex networks is such that more often than not too many input signals are required to arbitrarily modify the behavior of all the nodes of a network. Departing from the idea that achieving complete controllability of complex networks is a chimera, in this thesis, we present a comprehensive toolbox of input selection algorithms so to ensure controllability of the largest number of nodes of a network. Then, we complement this toolbox with algorithms for sensor placement so to also guarantee, when possible, observability of these nodes, thus allowing the implementation of feedback control strategies. Finally, an outlook on the topics that are currently being investigated by researchers working on controllability of complex networks is provided.