Elimination of fixed modes of a decentralized distributed estimator with inter-agent communication

We consider a decentralized distributed estimator for a formation of agents described with a discrete-time LTI system. Due to the decentralized structure, the estimator may contain poles invariant with respect to local agents¿ observer gains. We prove that under mild observability assumptions there exists a communication topology and transmitter-receiver gains which allow assigning all the poles of the distributed estimator. We also propose an algorithm for the design of a minimal communication topology which guarantees closed-loop pole assignability.