About link-layer handoff

[Leonardo Maccari]

On Wed, Jul 20, 2005 at 06:50:59PM +0800, JIE XIONG wrote:
> How can we decrease the layer 2 handover (it depends on OS, kernel,
> driver, wireless card??)
> 

a couple of documents I suggest over the topic:

http://winternet.sics.se/workshops/sncnw2003/proceedings/30T-Techniques%20to%20reduce%20IEEE%2080211b%20handoff%20time.pdf

http://portal.acm.org/citation.cfm?id=956990

expecially the second document gives an insight view of the phases of
handoff in 802.11b networks. Actually they have made an experiment with a
prism 2.5 card as a client and prism 2.5 with hostap driver as AP and they
measured average handoff time around 200 ms.
I still have to read it in deep (it looks a too little gap).

basically it depends on a mix of the causes you named, for example, the
protocol doesn' really specify when to leave current AP and pass to a new
one, it should happen when the signal strenght goes under a certain value,
but it is not clear how to measure that value. So as reported from first
document different nics perform differently. Then there is a scan phase,
to select new AP, this is controlled normally by firmware (see
host_roaming options in hostap driver) and then authentication,
association, eventually 802.1X stuff, llc layer procedures ... 

see also http://www.drizzle.com/~aboba/IEEE/ there are some links to
documentation on handoff procedures. 

ciao,
leonardo.

-- 
   Key fingerprint = 3129 C583 F03B 2E73 0115  C040 3489 0185 B592 19FE
 Obviously -thisaintpartofmyaddress- is not part of my real email address