WDS and TX rates, signal measurements

[Sergio M. Ammirata]

Jouni,

Back on June 08 you wrote:

> Anyway, I think I will change the driver to process beacon frames even
> when hostapd is running

Have you gotten a chance to make this change in CVS?

Thanks,

Sergio

> -----Original Message-----
> From: hostap-bounces at shmoo.com [mailto:hostap-bounces at shmoo.com] On
Behalf
> Of Jouni Malinen
> Sent: Sunday, June 08, 2003 9:16 PM
> To: hostap at shmoo.com
> Subject: Re: WDS and TX rates, signal measurements
> 
> On Sun, Jun 08, 2003 at 09:29:24AM -0400, Sergio M. Ammirata wrote:
> 
> > There are a few issues that are still pending in the latest CVS
code:
> > 1) The default value for basic_rates is 3. Shouldn't it be 15?
> 
> No, the default basic rate set should be 1 Mbps and 2 Mbps. This is
> required for compatibility with IEEE 802.11 (i.e., non-802.11b)
> stations. Basic rates are used for multicast/broadcast, ACKs, RTS/CTS,
> but most unicast frames can use higher rates.
> 
> > 2) WDS link autorate negotiation does not work with hostapd (it
stays
> > stuck at 2Mbit) but it does in kernel mode. When I look at the stats
of
> > the neighbor in /proc/.. I see that that the supported_rates are
blank
> > when you use hostapd and they are filled in correctly in kernel
mode.
> 
> Which method do you use to add WDS links? Kernel driver is not
> processing beacon frames when hostapd is used, so it cannot
> automatically get the supported rates list. However, when the station
> entry for WDS peer is added with iwpriv wds_add or based on incoming
WDS
> frame, supported rates are set to 1, 2, 5.5, and 11 Mbps. I did not
see
> them blank in any of my tests with hostapd running.
> 
> Anyway, I think I will change the driver to process beacon frames even
> when hostapd is running
> 
> > 3) Did the signal measurement conversion change? Before, the noise
level
> > was around -100Dbm, now it is around -30Dbm.
> 
> If I remember correctly, I changed at least one of the signal quality
> conversion to use new function. This should match with the numbers
> reported by latest firmware versions. It will most probably give
> incorrect values for older firmware versions.
> 
> Unfortunately, these numbers seem to vary a lot based on both hardware
> model and firmware version. In addition, they are not documented very
> well. I do not have enough interest and energy in going through all
the
> possible combinations in search of the correct value..
> 
> --
> Jouni Malinen                                            PGP id
EFC895FA
> _______________________________________________
> HostAP mailing list
> HostAP at shmoo.com
> http://lists.shmoo.com/mailman/listinfo/hostap