[Soekris] interference from two radios - revisited

Jim Thompson jim at netgate.com
Thu Oct 13 11:57:54 UTC 2005

On Oct 12, 2005, at 11:50 PM, Mitja Muženič wrote:

> Hi!
> I've taken the liberty of forwarding this to you as you seem to be  
> the most
> authoritative person on WiFi on Soekris mailing list and my  
> original mail
> references your explanation.
> Can you help me with the RF issues?

comments in-line

> Regards, Mitja
> (probably Netgate's future client :)
> -----Original Message-----
> From: soekris-tech-bounces at lists.soekris.com
> [mailto:soekris-tech-bounces at lists.soekris.com] On Behalf Of Mitja  
> Muženič
> Sent: Saturday, October 08, 2005 12:37 AM
> To: soekris-tech at lists.soekris.com
> Subject: [Soekris] interference from two radios - revisited
> As I've been told offlist, my mail has been caught in the spam  
> filters. I'll
> try to modify it slightly to see if it gets through this time...
> --------------------------
> Hi all!
> I know this topic has been discussed a lot and I've read the  
> archives. I
> found Jim Thompson's explanation on http://www.netgate.com/ 
> zz_faq.php#67
> excellent and I understand the propagation issues behind it. But I  
> still
> need a bouncing board for my thoughts, so please bear with me.
> I'm working on a project to provide indoor wifi coverage into a  
> building
> mostly made by 30cm concrete with rebar.. Think nuclear shelter,  
> only it's
> not government or military, but a huge private mansion designed by  
> a mad
> architect. The rebar is most probably grounded as well since I've  
> never seen
> so much LOS signal loss - the walls seem to suck the signal out of  
> the air.
> Put an AP in the middle of the corridor, walk a few meters away and  
> watch
> the signal level sink as a stone - that bad.

sounds about like what I'd expect.   for the Yanks/Brits in the  
audience, 30cm is approximately 12".

> Jim's explanation on why two AP's in close proximity won't work  
> assumes (as
> I understand it) free space around the two AP's and the antenna on  
> or near
> the AP itself. In my case, I'll probably have two antennas, one on  
> each wing
> of the building, and the signal loss in structure is so big that  
> the signal
> from one AP antenna is undetectable (as far as my measuring  
> equipment can
> tell) at the other location. I wonder how this changes the situation?

my testing (on other 'mother' boards) indicates very little (if any)  
leakage between the cards
on a non-RF path, so, in this case, should you actually find that  
you've got very low (or zero)
signal left from radio 'a' at the antenna for radio 'b', you'll be  
"cooking with gas".
(translation, it will be fine).

> The second problem is that for architectural reasons the AP  
> equipment must
> be in close proximity (one telco cabinet) and the two antennas  
> separated by
> a longish cable, one into each wing. I'll be having approx. 10m of  
> 0.5dB/m
> cable from AP (net4526 running OpenBSD + 5354 Aries MP in 11b mode)
I think you'll find that the Atheros cards work better under FreeBSD  
or linux.

If you're going to run 11b-only, why not just use a pair of 2511MP  

> to the
> antenna (4 dBi omni). My question is: assuming no interference from  
> the
> radiating elements - antennas (they should be in decently shielded  
> areas of
> the building) - , how much leakage will there be from the soekris  
> boxes in
> physical vicinity?

Very little.  Likely too little to matter, but I never directly  
measured it on a soekris board.

> How will this affect the working of two AP's? Do I even
> gain anything by shielding the two cards in separate soekris boxes  
> or can I
> simply use both in the same case, thus saving the cost of one net4526?

I'd use one box, personally.

> The third problem is that I've decided to go with ath(4) based  
> cards in
> order to try Reyk Floeter's latest hostapd(8) tools to help with  
> roaming.

roaming is driven by the STA (client) in any case, so this seems of  
little value.

Heck, put the cards on the same channel, set the ESSIDs the same and  
try it, I think you
might find that it works really well.

> Does anyone have success stories or practical examples including  
> hostapd(8)?
> I went all over the man pages and it looks like there will be some
> tutorials/papers by the author but all those conferences are  
> scheduled for
> next months.
> Thanks for any insight you might give me.
> Regards, Mitja
> _______________________________________________
> Soekris-tech mailing list
> Soekris-tech at lists.soekris.com
> http://lists.soekris.com/mailman/listinfo/soekris-tech

More information about the Soekris-tech mailing list