Title : Nearest Neighbour Modelling for Finite State Processes indexed on a Graph Presenter : Lang White Abstract : This presentation examined modelling of finite state stochastic processes defined on a graph. We firstly consider the one dimensional case (ie the graph is a chain) and then extend the results. We show that the state evolution of the system can be described by equations which are linear in certain Kronecker products. Recursive realisations are then derived. We also indicate how these stochastic realisation results lead to optimal estimation which is inherently decentralised. Joint work with Francesco Carravetta (AISI-CNR Italy).