Negativizability for Nonlinear Estimation in Cyber–Physical Systems

Abstract

This letter introduces a novel fully distributed estimation scheme for nonlinear continuous-time dynamics over directed and strongly connected graphs. Leveraging on the assumption of local negativizability, the proposed approach performs the estimation of the interdependent subsystems of a cyber-physical system, despite the presence of nonlinear dependencies on the dynamics. This transforms the intricate task of nonlinear state estimation by each agent into more manageable local negativizability problems for the design of the estimation gains. A pivotal aspect of the approach is that each agent should be aware of an upper bound on the Lipschitz constant of the overall nonlinear function that characterizes the dynamics. To face this issue, we developed a novel distributed methodology for the estimation of the global Lipschitz constant, starting from the local observations of the system’s nonlinearities. The effectiveness of the proposed scheme is numerically demonstrated through simulations.