Re: [Madwifi-users] signal/link quality wrong?

[Sven <sr12 duke edu>]

thanks for the info. i will try to look into this issue, but as my
programming skills are virtually nonexistent i will have to leave
patches to others. there is a great whitepaper on this issue:
www.wildpackets.com/elements/whitepapers/Converting_Signal_Strength.pdf as seen at
http://mail.gnome.org/archives/networkmanager-list/2005-February/msg00031.html
which incidentally includes a patch for Ipw2100.

thanks again, 

Sven

On Thu, 2005-02-03 at 08:44 -0800, Sam Leffler wrote:
> Sven wrote:
> > hi, 
> > 
> > there has been some discussion on the NetworkManager mailing list about
> > the madwifi driver (and others) and the way the driver reports "Link
> > Quality" (eg, in iwconfig or in /poc/net/wireless). 
> 
> 	...stuff snipped..
> 
> At present you can only get rssi data that is reported relative to the 
> current noise floor.  The noise floor is "uncalibrated"; i.e. real world 
> units are not available (unless you know certain info about the 
> construction of the final product).  We're likely to add some form of 
> noise indication derived from looking at the noise floor data for other 
> channels (found while scanning) but otherwise that's what's there. The 
> current code assumes a specific noise floor and calculates data based on it.
> 
> When I look for signal quality/strength I monitor rssi and implicitly 
> derive link quality based on a range of 0-63 (or 60 for some cards).
> 
> Given the above please suggest a patch to improve the current code and 
> we'll consider it.  Do not assume there is noise data as I've no idea 
> when that'll be available.
> 
> 	Sam
> 
> PS. A separate issue is that the rssi data is potentially calculated 
> with "uninteresting frames"; e.g. in station mode the driver 
> could/should ignore frames from stations not in the bss.