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

[Paul Tan <tanpaul@i2r.a-star.edu.sg>]

Title:

Hi Bernard,

Bernard Aboba wrote:

I can believe that the reception of the prefix as well as the
router IP address and MAC within a beacon frame or probe response
is a significant boost to performance (it will beat everything else).

Is it something which IETF can standardize on now?
    

The Beacon and Probe Response are owned by IEEE 802.11, not by
the IETF.  As a result, an IETF draft submission on the subject cannot
have an effect, not because it is unwelcome but because it is being
submitted to the wrong organization.

my original intention of this draft was to present an alternative to solving some L3 problems (e.g. MD). Obviously, the proposal (or sometime any other protocol) required something (cross-layer) which is outside the scope of IETF, but I thought maybe (later) somebody who have experience with IEEE 802 can bring it further.

Our prime objective was to identify various alternatives and finally proposed an 'universal' solution. Pardon me for using the wrong vocal.

IEEE 802 operates much like IETF in that a charter (known as a PAR) is
required before a standard can be created.  IEEE 802 submissions that do
not relate to a PAR are much like individual submissions not taken up as
an IETF WG work item -- they are not part of the standards process.  To
get the submission "welcomed" it is necessary to get it accepted as relevant to a
PAR, or if there is no such PAR, to create a PAR that is relevant to it.

Thanks for the explaination on the workings. I probably guess it is must easier if an collective consensus is reached in IETF, and then present our requirement for the standard to IEEE802

That said, here are some issues that can arise from this proposal:

a. It assumes that the AP only offers access to a single network. APs
(like switches) are capable of supporting VLANs, in which case the AP can
offer access to multiple networks.  Which prefixes should it advertise in
this case?  If a network is advertised, can the host hearing the
advertisement then conclude it will be able to obtain access to that
network?

Good point! Imagine several WISPs connecting/sharing a single AP.
Pardon me, but in VLAN, if a (mobile) station moves across the VLAN but still within the VLAN, does it trigger mobile IP (i.e. change of IP) ? I need to read the specification in detail.

b. In practice, VLANs are used in quite a few scenarios.  APs which need
to transition from one access mechanism to another now often use VLANs for
that purpose. So there can be a VLAN for Web Portal, another for WEP, for
WPA, for RSN, etc.  Also Guest VLANs are being implemented; dynamic VLANs
to enable L2 mobility;  shared use APs offer access to VLANs from multiple
providers; etc.  In this case the VLAN to which a host will get access is
determined after authentication and possibly association, it cannot be
known a priori.  This implies that advertisement in the Beacon and Probe
Response could be misleading.
  

Assume we include all prefixes or maybe a single VLAN-prefix (if there is no IP change when MN moves within a VLAN), isn't these information contained inside the periodic beacons received by the stations before any operations ?

For all these reasons, if prefix(es) are present in the Beacon and Probe
Response they need to be treated as *potential* networks to which the host
*might* be connected. This means that this particular advertisement would
act as a *weak* hint.

We need to investigate all possible scenarios which VLAN (or 'futureLAN') introduces, and YES, I agreed with your classification of 'hints'. At least, we are at the right frequency ( using L2 to provide 'useful & reliable' hints ).