Connection state overhead in a dynamic linear network

We consider a dynamical linear network where nearest neighbours communicate via links whose states form binary (open/closed) valued independent and identically distributed Markov processes. Our main result is the tight information-theoretic lower bound on the network traffic required by the connection state overhead, or the information required for all nodes to know their connected neighbourhood. These results, and especially their possible generalisations to more realistic network models, could give us valuable understanding of the unavoidable protocol overheads in rapidly changing Ad hoc and sensor networks.