Just came back to this paper. The core spanning tree that they describe sounds similar to the transport overlay network in this paper.

So I don't think they are too far off.