Bert, thanks for the comments, the link you posted at IETF doesn't work; this one does:
I believe the author does distinguish between 'mesh' or P2P and star networks by stating the fact the most mesh networks function as star's most of the time. You are also right in your observation that star networks are not self-healing while P2P are. I also believe that most M2M implementations in industrial automation will end up being star networks.
It is very much required protocol which is expected in the talk above, this is also a major research area in ad-hoc networks. Routing over wireless is a stringent matter that will require few years to become optimal.
I was having a difficult time understanding this, until I figured that perhaps "mesh" and "star," to describe network topologies, were being used interchangeably?
A network configured as an RF mesh is a network in which each client node can behave as a router, filling in areas that would otherwise not have coverage. (See for example the MANET working group of the IETF, http://datatracker.ietf.org/wg/manet/.)
Instead in a star network, as the article points out, each client node needs to be within range of the hub (server node?), or within range of a router (which in turn is within range of the hub).
I certainly agree that stars are simple. However they are not self-healing, as a mesh would be, and they limit bandwidth of the network to whatever the single hub can carry. As the article describes.
So, the M2M functions such networks can carry are quite limited. I continue to believe that this recent popularization of the term "M2M" only describes a very small subset of actual machine to machine comms that exist. For instance, any kind of distributed control system.