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

[Bernard Aboba <aboba@internaut.com>]

> Here is my doubt: unless I am wrong, this is what is done in AP
> implementation, but nothing is said on VLAN in 802.11 specification.
> The implementation that I tested (cisco AP) uses a different SSID per
> VLAN. But the SSID is the identifier of the ESS. So that means that two
> MNs that are attached to the same AP, but in two different VLAN are in
> the same BSS but not the same ESS. I think this is going against the
> specification, isn't it ??

No. There is nothing that requires all APs advertising an SSID to place
all authenticating stations on the same network.  In reality the SSID is a
service provider identifier *not* a network identifier.  So an operator
like TMOBILE uses the "tmobile" SSID worldwide, even though the APs at the
different locations are part of a world-wide routed network.

This means that the SSID is *not* the identifier of an ESS, but rather an
identifier of the operator.

> For the moment (if I am not wrong !), IEEE 802.11 does not specify which
> information to put in beacon frame when several VLANs (ie SSIDs) are
> configured on a single AP.

There is no need for such a specification.  In practice, Beacon frames
need to be small enough to avoid fragmentation, and attempting to cram
advertisements for multiple networks into a single Beacon frame would push
those limits. In any case, there is no "grouping" capability in IEEE
802.11-1999 so this could not be done in any case.

Instead, the specification allows a single physical AP to send multiple
Beacons, each with its own SSID and unique capabilities. The links
below describe what is done by current implementations:

http://www.drizzle.com/~aboba/IEEE/virtual-APs.ppt
http://www.drizzle.com/~aboba/IEEE/11-03-154r1-I-Virtual-Access-Points.doc

> In current implementation, as soon as
> multiple SSIDs are on the AP, no SSID are advertised in beacon.

I have not seen an implementation that does this. Most implementations
either have a single 'primary' SSID used in the Beacon, or they advertise
multiple SSIDs, each with their own BSSID.  There are even some that
advertise multiple SSIDs with the same BSSID, but some drivers do not
react well to this, so it's inadvisable unless you have NICs and APs from
the same vendor.

> a STA scans the channel with a particular SSID, Probe Response from AP
> contains the information related to this SSID.

This is the 'secondary SSID' approach.  One of the problems with this
approach is that the SSIDs need to be configured beforehand; it is not
possible to discover a previously unknown SSID.

> So if we want to consider mutliple VLANs on a single AP, maybe we can
> start with the same concept: no information in Beacon and specific
> information in Probe Response.

I fail to see how this is relevant to the topic at hand: DNA.  Proposals
for changing IEEE 802.11 behavior are best handled with in IEEE 802.11,
or in the WFA Public Access group which is debating this issue, not in an
IETF WG that has no authority to create standards on the subject.