[Lwip] Neighbor Cache Management policy Experiment result

[Rabi Narayan Sahoo <[e-mail]>]

Hi All
We have done an experimental study of neighbor cache management policy
uisng whitefiled simulation framework.

We analyzed two primary points, impact on the PDR and the convergence time.
PDR was relatively easy to calculate but the convergence time, turned out,
is not so easy to identify. We defined convergence as the point at which
the BR becomes aware of all the routes to the nodes in the network.But its
difficult to estimate when the corresponding routes are stable enough so as
to declare the network is stable.

Without neighbor management policy, with lower neighbor table size, we
found the network never stabilizes. The reasons are highlighted in the
draft and we have relevant data.The PDR also extremely bad without neighbor
management policy, because the network never stabilizes and we keep getting
forwarding errors.

Please find the attached document which explains the configuration and the
test data.

Please let us know if there is any other data we can augment?

Cheers
Rabi