[quote user="Ilya Averin"]
The association process does involve both the MAC and NWK levels. Actually, two stages are performed at the NWK level to join a network. The first stage is the network discovery with the active scan (the generation of beacon requests and processing of received beacons). The information extracted from the beacons is stored in the neighbor table. The second stage is the joining a network. The NWK picks one of the neighbors that meets some criteria (PAN ID, depth, LQI etc) and starts the association procedure through the MAC level functionality.
This means that any node has the single parent node and some number of the neighbors. Actually, the parent node doesn't play any specific role after the joining is completed. The data can be transferred via any suited neighbor.
[/quote]
Sorry to trouble you. I have a couple of questions relating to this specific post. You have said that neighbor table entries are added from the beacons when a scan is executed. I have read that a Mesh network does not support beacons, but I still understand that neighbor tables are used and that they are updated for any kind of packets received. Do you happen to know at what point neighbors are actually added in the case of Mesh?
Also in the case of Mesh and non-Mesh, how are the neighbors selected for entry if there is a limited table size? Is it first come first served or is there some selection from all the beacons (or something else in the case of mesh)?
Finally am I right in understanding that the neighbor table affords a kind of topology control in that RREQ broadcasts which arrive from a non-neighbor are discarded?