Problem With Links Over Water

[butchkemper]

I have a tower on a penenusla that sticks out in the middle of the lake. On the tower is a 5.8 trango AP radio with a 120* horizontal polarized sector. The clients are either a Fox with a built in antenna or an Atlas in a dish reflector.

One Atlas is 6.9 miles away and the performance is inconsistent - sometimes it is great with a -65 rssi and sometimes it is poor with a -85 rssi. The other clients have similar experiences.

I have changed APs, antennas, changed the polarization between V and H, replaced client radios and the problem persists.

I am quite certain the problem is associated with the links being over the lake but I am not sure if the problem is related to temperature or lake surface conditions.

Does anyone have any suggestions of where I could read about how the signal is affected by the lake and what I could do to correct the problem?

Butch

[Stratolinks]

Oh that's an easy one, just drain the lake. ;-)

I seem to remember reading that at higher microwave frequencies that horizontal was better over bodies of water, but I don't recall what the "higher" numbers were. Maybe it was 20GHz, I don't know for certain. I did a quick Google but didn't find anything very relevant to the situation. I am sure the info is out there, you just have to find the right combination of words to Google.

[nickwhite]

There were some discussions about this on the isp-wireless list... 10 months ago... the discussion continues for several weeks, so check July archives as well:
http://isp-lists.isp-planet.com/isp-wireless/0606/msg00801.html

I faintly remember another discussion on the WISPA list, but I can't find it in my email.

-Nick

[therealboss]

I think the answer to your problem is to use Diversity, I had a P2P llink over a lake and found Diversity worked best with about 10' between both grids. It worked ok but was never 100%, I now have a path around the lake so we took down the Diversity link.

[jcollery]

As this is a lake I'm not sure if this is your problem, however if it happens twice a day, regularly, it's tidal. Use directional antenna, and point both ends slightly upwards to resuce the lower sidelobes and this should solve your problem.

Thanks,
JC

[valenti]

Butch,
Maybe you are already doing this, but one thing I would do is use SNMP and graph the signal levels using MRTG (or similar). Trango's website has a tech note on setting this up. I'm only graphing my backhaul link, but should start doing all the clients too.

You could then compare the graphs to weather charts, time of day, etc and see if there is a correlation. (how to fix what you might find is another problem)

[handyman]

We once got advice from experienced techs at a vendor to point the antennas a bit too high. You lose some signal strength, but also less of the Fresnel zone is hitting the water. I don't remember whether it helped.

We have one PtP shot over a lake that *seems* to work better in vertical than horizontal polarization.

[nbright]

The problem you're having is that as the level of the water changes, it impacts your fresnel zone, probably what is happening is that it's changing between being less than 1.0 and more than 1.0. When your fresnel zone is more than 1.0, you will destructively interfere with your own signal.

This is intensified over water, because it's basically an RF mirror, whereas over land the ground will absorb most of the signal. While over ground some is reflected, over water it is nearly all reflected. Also because the surface isn't completely smooth you can also run into problems with signal phasing causing destructive interference as well.

[0ldman]

We once got advice from experienced techs at a vendor to point the antennas a bit too high. You lose some signal strength, but also less of the Fresnel zone is hitting the water. I don't remember whether it helped.

We have one PtP shot over a lake that *seems* to work better in vertical than horizontal polarization.
Seems horizontal vs vertical would depend on the H vs V beamwidth, wouldn't you think? Probably more an issue with a grid than a dish, right?

[handyman]

Seems horizontal vs vertical would depend on the H vs V beamwidth, wouldn't you think? Probably more an issue with a grid than a dish, right?

In our case, they were flat-panel antennas. If the H beamwidth differed from the V beamwidth, then, yes, that could account for it. I cannot find the specs...

My impression was that the way that the H-polarized waves reflected off the water differed from the way that the V-polarized waves did. We need a physicist...