| Distributed estimation is where a network of agents is tasked to estimate the state of a dynamical system. Agents only communicate over a sparse communication network. Recently, consensus-based estimation has been proposed as a distributed solution of this problem where the agents implement a large number of information exchanges between every two successive time-steps of the system dynamics. For optimal performance, this consensus-based estimator requires a consensus to be reached first. When the network is unable to implement a consensus due to, e.g., resource-constraints or faster system dynamics, distributed solutions have been proposed with single-time information exchanges. In this scenario both system dynamics and distributed estimator evolve at the same time-scale. This scenario requires the system to be observable at every estimator/agent, implying the new concept of distributed observability. Given this background, this thesis is devoted to (1) formulation of distributed observability in single-time scale estimation, (2) partitioning the necessary set of state measurements based on their role in distributed observability, and (3) characterization of necessary and sufficient connectivity of the underlying communication network topology among the agents. Employing structure-based generic methodology instead of algebraic approaches motivates application in power systems and social networks. |
