A new era for networked dynamical systems
Networked dynamical systems are instrumental for operation of numerous engineering applications such as infrastructure grids, sensor networks and security systems. However, their sheer size and complexity render decision-making difficult, requiring better approaches to improve the required decentralised estimation, detection and adaptive control. The EU-funded DEMAND (Decentralized monitoring and adaptive control for networked dynamical systems) project rose to the challenge. It developed new methods, algorithms and techniques that render networked dynamical systems more robust in several ways, enhancing their decentralisation and distribution capabilities. In more technical terms, the project team created tools and algorithms for decentralised overlapping system identification and decentralised adaptive target tracking. The tools also provide distributed macro-calibration and time synchronisation in sensor networks, as well as distributed reinforcement and supervised learning. Other important achievements include robust fault detection in networked systems as well as obstacle avoidance control, with several practical applications specifically targeting wireless sensor networks. Another advantage of the new tools is their power to offer stochastic extremum seeking control on manifolds with applications related to adaptive distance-based synchronisation of oscillators and rigid bodies. This is especially important, for example, in improving brake system control and optimising bioprocesses and flow control in real time. Overall, the project fulfilled all its goals and even went beyond its initial scope in developing the new tools, techniques and algorithms. The results were distributed through numerous journals and conference publications, and are set to positively impact a variety of fields. Real-world applications such as sensor networks are set to benefit from the research findings, as will the Internet of Things and networked cyber physical systems. New and upcoming technologies such as smart grids, smart societies and the industrial Internet of Things are also set to benefit, along with existing technologies such as drone surveillance, biological systems and social networks. As computers, communications and control converge, more powerful networked dynamical systems will be useful in upgrading numerous large systems, from a network of air conditioning units to a nationwide electric grid. The new tools and approaches developed under DEMAND will no doubt be pivotal in advancing the vision of streamlined, interconnected networks.
