Periodic Reporting for period 3 - COSMOS (Game theoretic Control for Complex Systems of Systems)
Periodo di rendicontazione: 2022-02-01 al 2023-07-31
Our society is based on "systems of systems", namely, large-scale and interconnected infrastructures, such as power grids and transportation systems, with network structure and interacting subsystems. These systems are "complex', due to the rationality and autonomy of their subsystems, and in fact require advanced coordination and control actions for their safe and efficient operation, for example to mitigate grid overloading and traffic congestion. Unfortunately, standard multi-agent optimisation is inappropriate and ineffective for systems with noncooperative (selfish) subsystems, virtually all modern systems of systems. In order to ensure safe and efficient operation despite the presence of selfish parties, this project will develop dynamic game theory towards a unifying framework for the analysis and control of complex systems of systems and for supporting automated decision making.
We are devising a unifying mathematical framework for the analysis and control of complex systems of systems. Technically, we are developing a novel operator-theoretic methodology and integrating methods within and across dynamic game theory, networked control systems, statistical learning and stochastic optimisation. From a more practical perspective, we are building up a mathematical theory, algorithms and automated software that can ensure safe and efficient operation of systems of systems, such as modern power grids and automated road traffic, populated by self-interested subsystems and autonomous decision makers.
The project has developed the field of dynamic game theory for control of complex systems of systems beyond the state of the art. In fact, the devised mathematical framework allows the system designer to consider systems of dynamical systems, rather than static systems. The most relevant expected results are the development of a theory and of computational algorithms to model, analyse and solve game equilibrium problems for systems of autonomous systems subject to environmental uncertainty in selected non-standard settings, such as non-convex and non-monotone regimes.