Poll stations from AP transparently
Tue Oct 22 04:09:59 PDT 2002
As far as I know, there is no response to beacon frames, so you can't
listen for them (it would generate too much traffic for no useful purpose).
If you want to know if a station can reach other APs, I think the only
solution (other than adding software on the clients) would be to have your
AP communicate with neighbouring APs, asking them to poll them somehow. But
I'm not sure there are ways for an AP to poll a station associated with
another AP that will always work (it depends on CF-Pollable settings and
similar things). Alternatively, if the clients are using active scanning,
the neighboring APs could have a cache of recently seen probe requests. But
again, that won't work with clients using passive scanning only.
A slightly different question: why do you want your AP to reject clients? A
scalability issue to achieve some form of load-balancing?
At 12:38 PM 10/22/2002, Victor Aleo wrote:
>I already know, from multiple emails with this list and especially with
>Jouni, that is not possible to know from the AP side if a station (client)
>can reach other APs in the same ESS. The reason to get this knowledge is
>very simple: if an AP knows that a STA can reach other APs it will be
>potentially possible to reject this station "safely". Thus, with the
>security that the station will be able to reassociate with at least another AP.
>However, I still think that this can be implemented without the aid of any
>special software in the station side (transparent to the clients). Thus,
>only the AP will be "modified". The key question is how? It is clear to me
>that there are not specific methods to perform this task (mainly because
>the standard specified mobile based handoff but not network based). But I
>think there are "indirect" methods to find out the answer:
>1) If the AP can poll, by means of beacon frames for instance, the
>stations. In this case, the APs in the same ESS send beacon frames to a
>specified STA. If the STA answer to this beacon frames, it means that can
>be associated to more than one AP. Therefore, if it is rejected by its
>current AP it will not have problems to reassociate with another.
>My questions are: is this feasible with Prism2? Have you thought about
>other mechanisms to find out this information?
>Sorry to write such a long email, but I think this is a key issue to
>implement load balancing algorithms in WLAN networks and I am working in it :)
>Thanks a lot for your comments or suggestions,
>HostAP mailing list
>HostAP at shmoo.com
-- Jacques Caron, IP Sector Technologies
Join the discussion on public WLAN open global roaming:
More information about the Hostap