Distributed state estimation for nonlinear systems without sharing control inputs

This paper, accepted at ASME Letters in Dynamic Systems and Control, tackles distributed state estimation for discrete-time nonlinear systems with unknown dynamics over directed networks. Each sensing agent independently estimates the source state using only local measurements and data from neighbor agents, eliminating the need for shared control inputs or excitation signals. This is a significant departure from prior work, which often assumes agents can pool all input information. The authors introduce a normalized adaptive estimation scheme that simultaneously identifies unknown linear and nonlinear components while ensuring robust adaptation in discrete time.

Key technical contributions include a Lyapunov-based input-to-state stability (ISS) proof, bounding estimation errors under disturbances and guaranteeing asymptotic convergence in noise-free cases. To handle network coupling, the paper develops explicit norm-based and LMI-based Schur stability conditions—including a robust version for model uncertainties. Numerical simulations on star, cyclic, and path communication topologies validate the approach, showing accurate distributed estimation and efficient scaling with network size. This work is highly relevant for multi-agent sensor networks in autonomous systems, robotics, and infrastructure monitoring where input sharing is impractical or insecure.