Re: [DNA-BOF] Using L2 to provide Instantaneous Movement Detection andNeighborhood Discovery

[Bernard Aboba <aboba@internaut.com>]

> Is it possible to learn a potential secondary SSID by
> using one from the previous AP (as the only non-broadcast
> SSID used)?

Generally not, because an organization will typically only Beacon the
primary SSID on all its APs.  However, it is reasonable behavior for an
implementation to Probe for all previously encountered SSIDs in order to
find the secondary ones, if present.  However, this prolongs the scan
time.

> I know this isn't useful as a movement hint, but may be used
> as a definitive non-movement hint, if it means that we can
> use the same VLAN, as on the previous AP.

Use of the Beacon vs. Probe Request/Response doesn't affect movement
detection; this only relates to the effectiveness and timeliness of active
vs. passive scanning.  Generally implementations do both.

> Considering the number of intra-subnet movements with
> such a topology, this could be a boon to reduce L3
> handovers or L3 message exchanges.

It really has no effect on L3 handovers or message exchanges. This is a
pure L2 issue.  If an SSID is not Beaconed and isn't configured, it just
won't be found with the primary/secondary scheme.  That's one of the
reasons why this scheme is inherently limited -- and why the industry
moving away from it towards APs with multiple BSSIDs.

With multiple BSSIDs, it is possible for a single AP to Beacon multiple
SSIDs along with the corresponding capabilities.  The station sees all the
"virtual APs" and available networks, and can choose to connect to them
without preconfiguration.

It's even possible to connect to multiple networks at the same time.  See:
http://research.microsoft.com/~bahl/ms_projects/multinet/default.htm

> It helps to have a clear understanding of how things
> work (and current proposals in 802 or WFA) so that
> we don't keep re-debating issues though.

The most important thing to take away from all this is that L2 "hints" are
generally fallible and that an implementation MUST be robust in the face
of misleading hints.