Re: What's required to make wireless reliable?

[Mark Knecht <markknecht gmail com>]

On Mon, 28 Mar 2005 09:18:09 -0800, Rick Stevens
<rstevens vitalstream com> wrote:
> Mark Knecht wrote:
> > On Fri, 25 Mar 2005 19:29:27 -0600, Otto Haliburton
> >>>
> >>>However it's sort of strange but the results from iwlist wlan0 scan
> >>>are not valid unless I do ifdown wlan0 ifup wlan0. Unless I do that
> >>>the value stays the same no matter where I am in the house.
> 
> That is a bit strange.  

Yes, it is, and it only happens on my laptop with is using one of the
Broadcom chips. On my wife's machine and my son's machine, both of
which use the same D-Link boards, each iwlist command gives me updated
values. They don't vary much, but they do move around by 1 or 2 db.

> I have found that the order in which things are
> fed to iwconfig is significant.  It may be that the script sets things
> in a manner that causes problems.  The
> /etc/sysconfig/network-scripts/ifup-wireless script sets up the
> wireless stuff in this order:

I'll have to look at the scripts I have here. Thanks!

> 
>         mode
>         nickname
>         network ID (wireless domain)
>         frequency/channel (if mode is Ad-hoc)
>         sensitivity
>         rate
>         WEP key
>         RTS/CTS
>         frag
>         ESSID
> 
>
> 
> Uh, isn't that "r squared"?  I can't recall my power calculation stuff.
> Decibels (dB) is a logarithmic measurement as well, so things may look
> a bit weird.

Yeah, you're probably right, but I was thinking that r-squared was
from a line source and r-cubed was from a point source but looking up
equations quickly proved me wrong. Sorry!

- Mark