Development of Novel Architectures for Control and Diagnosis of Safety-Critical Complex Cyber-Physical Systems

Abstract:

The project is developing novel architectures for control and diagnosis of complex cyber--physical systems subject to stringent performance requirements in terms of safety, resilience, and adaptivity. These ever--increasing demands necessitate the use of formal model--based approaches to synthesize provably--correct feedback controllers. The intellectual merit of this research lies in a novel combination of techniques from the fields of dynamical systems, discrete event systems, reactive synthesis, and graph theory, together with new advancements in terms of abstraction techniques, computationally efficient synthesis of control and diagnosis strategies that support distributed implementations, and synthesis of acquisition of information and communication strategies. The project's broader significance and importance are demonstrated by the expected improvement of the safety, resilience, and performance of complex cyber--physical systems in critical infrastructures as well as the efficiency with which they are designed and certified. The original approach being developed is based on the combination of multi-- resolution abstraction graphs for building discrete models of the underlying cyber-- physical system with reactive synthesis techniques that exploit a representation of the solution space in terms of a finite structure called a decentralized bipartite transition system. The concepts of abstraction graph and decentralized bipartite transition system are novel and open new avenues of investigation with significant potential to the formal synthesis of safe, resilient, and adaptive controllers. This methodology naturally results in a set of decentralized and asynchronous controllers and diagnosers, which ensures greater resilience and adaptivity. Overall, this research will significantly impact the Science of Cyber--Physical Systems and the Engineering of Cyber--Physical Systems.