I primiarily did paper survey on the papers following Diffusion Kalman Filter, in order to characterize its performance in the persecptive of network topology. On the way of surveying, I found that there are two main research groups in this field.
The first on is lead by Dr. José M. F. Moura. One the of papers in his group (or his student's group) called ''On the Genericity Properties in Distributed Estimation: Topology Design and Sensor Placement,'' basically summarizes the observability criterion of distributive estimation.
The other line of work is lead by Dr. Reza Olfati-Saber. One of the highlight is the work on switching topology, trying to explain the behavior observed in this paper.
Supposedly, we can characterize the topology behavior of diffusion Kalman filter first, and discuss the effect of varying topology, which is unavoidable in multirobot systems. One of the main challenges is that most of the works are based on 1st order consensus fusion. The way to apply on Kalman filter should be further explored.