Wireless bridge

John van Oppen john at vanoppen.com
Thu Jun 18 16:34:19 UTC 2009

To come up with an accurate recommendation one really needs to know your
budget, on that distance speeds up to 1 gbit/sec are possible if you
spend enough on the radios...    Do you have some cost and desired
throughput parameters to guide everyone's recommendations?

-----Original Message-----
From: Tim Huffman [mailto:Tim at bobbroadband.com] 
Sent: Thursday, June 18, 2009 9:27 AM
To: nanog at nanog.org
Subject: RE: Wireless bridge

> The line of sight is all clear, no trees. Only one building along the
> has a rooftop of similar height, but the antennas are extended far
> the
> roofline. We have used a rifle scope to confirm line of sight is all
> at all angles.

Unfortunately, you can't necessarily rely on visual line of sight. At
800meters, the Fresnel Zone on your radio is about 14ft in diameter at
the midpoint. You need to make sure that this is free of obstructions.

> Oh I know. Luckily it's located in an industrial area just on the
> outskirts
> of the city. There isn't a lot of other WiFi (in my opinion); 3-5
> SSIDs spread across 2 of the 3 physical channels (1,6,11) depending on
> which
> rooftop you measure from.

Make sure you're using the channel that doesn't have an AP on it!

> Bandwidth requirements aren't too picky. If it can handle minimum 9
> full-duplex everyone will be happy. Of course, the faster the better.
> I don't know if it makes a difference or not but this is all taking
> in
> Canada. I don't know of any regulations drastically different from the
> U.S's
> regarding frequency use here. The biggest problem I've ever had though
> just been payment/shipping depending on the supplier (some don't ship
> Canada or are very specific about payment methods!).

Canadian and US regulations are very similar in the unlicensed bands.
I'd still pick 5.2GHz if you were replacing the radio. 

> Just to answer a few more questions I've been getting, the access
> are
> located inside, connected to a small UPS. The antenna wire is a very
> coax up to the roof, BNC connectors to the access point and I'm fairly
> certain BNC connectors on the antenna end as well. I'll double check
> grounding on the poles but I'm somewhat afraid to turn it into a
> rod. I'm fairly certain that the ground in the antenna wire is clean
> again, something to double check.

How long is your cable run, and what kind of cable is it? It's probably
LMR-400 (the most common) loses about 6.6dB of your signal for every 100
feet. Also, you should check the waterproofing on the connector at the
antenna. We normally use a 'courtesy wrap' of electrical tape, followed
by a thick layer of Mastic tape, followed by another layer of electrical
tape. Also, check your cable for nicks or kinks.

> Rain/moisture doesn't seem to cause problems. In fact the connection
> more
> reliable through the winter. The last 2 months here have been
> dry/wet and there's been no pattern to the stability issues. The only
> correlation between weather and stability that they have noticed there
> lightning related.

Moisture in the cables doesn't necessarily show up during rain! That
moisture can seep throughout the cable, and cause attenuation when it
gets cool and the moisture condenses, for example.

You haven't said what kind of antennas you are using, but if they are
yagi's, they probably have very poor back-to-front ratios, which means
that you could be picking up interference from behind you, or on the
sides, especially if the antennas are up above the tree cover. You might
try horizontal polarization on the antennas (just rotate them 90
degrees, but make sure you do it on BOTH sides!) to see if that helps.
Cross-polarization is usually good for about 20dB of noise rejection.

The fact that there doesn't seem to be any pattern to your loss means
that it's probably either interference (somebody changing channels),
hardware failure, or software failure.

Hope this helps.

Tim Huffman
Director of Engineering
Business Only Broadband, LLC
O (630) 590-6012
C (630) 340-1925
tim at bobbroadband.com

More information about the NANOG mailing list