I have a link over 17 miles of open ocean. 5Ghz/10Mhz The link fluctuates significantly in signal strength. One second its -60 and a few seconds later can be as high as -90. Thats pretty extreme. I am confident its not interference -induced because if the history with the link (hard to explain).
Any way, does anyone have links over open water near this distance that see the same behavior? The water is not flat (it is the ocean) and is fairly choppy with constant smaller swells and of course tidal changes of a few feet twice a day.
To complicate matters its shooting parallel with the beach so I also believe I have temperature inversions changing drastically between land & water.
Its perfect LOS from tower to tower (150-200 feet) but it sure has been a bugger
We find our over-water links vary with the tides up to 20dB throughout a day. The only way to keep such links up is to throw antenna gain at them.
I have seen the really rapid changes in signal strength you describe not on a water link, but on a long across a very flat area. There’s 300M of elevation difference between the two ends - which are 60km apart. 900mm dishes at each end. The RSSI charts are completely spiky, no matter what channel we use. Variations are however between -50 and -70, so nothing likely to cause us grief.
Linking over water is a huge pain in the ass, if you ask me.
The root of the issue is that water surfaces function as an RF mirror at these frequencies.
So, what you’re going to see it lots of randomized reflections causing both constructive and destructive self-interferance at different moments in time - as a swell goes up it might reflect right back at you in-phase and cause a huge problem, or maybe it might be out of phase, or maybe it’ll be on the right shift to increase signal, or on just the wrong one to decimate it…
The point I’m getting at is, though I haven’t gone over open ocean like this, it’s my opinion that going over water period (any water, any distance) is going to cause you SERIOUS problems with signal variations as the water level and “calmness” of the surface fluctuates.
As the others have mentioned, probably the only solution is to throw more gain at it; you might also try switching polarizations too, perhaps the opposite of what you’re using might work better? I would suspect something like that is highly dependent on specific situations.
Has anyone tried Circular Polarization? Any reflections will cause a reverse direction of polarization I understand and will not be detected. Worth a try.
It does help to not have the antennas near the water too. If you can move them back a mile away from shore, the fluctuations will be much smaller. If either end is < 1/4 mile from the shore, you will definitely have 30db fluctuations guaranteed.
The link is on the beach on both sides. I may try to shorten the path a few miles to another tower I have also on the beach and see how it behaves there… Its too bad such a beautiful LOS shot has such crappy rf characteristics
Hi there,
This is an old problem to me.. In a past life I engineered microwave links to and between offshore oil production platforms.
It is cause by reflections from the water surface. When the reflection arrives in phase with the main signal then the signal is amplified, when it arrives out of phase the signal is diminished by as much as 50dB. If you can move the antennas so that they cannot “see” the water, it will help. Or you will need to build in some diversity; there are 3 main types Space, Frequency, and Polarisation. With space diversity you use two antennas at one or both ends of the link, this is favoured and there are formulas for working out the optimum separation between antennas (google it like http://www.softwright.com/faq/engineering/MICROWAVE%20SYSTEM%20EQUATIONS.html ) it´s a little complex and changes if you move up or down a channel, however something more than 20 wavelengths usually works well;-)
Here is a suggestion. Why not make one link on 2.4GHz Hpol and one on 5GHz V pol, put both wireless cards in the same router the combine both interfaces set as slaves in a bonding interface http://www.mikrotik.com/testdocs/ros/3.0/interface/bonding.php using broadcast mode. Broadcast mode - Broadcasts the same data on all interfaces at once to provide fault tolerance.
Result should be a link with space, frequency and polarisation diversity, and should have much higher availability. For example if each link is 99% available, then the bonded link should be 99.99% . you could add more radios, and antennas to make the ultimate link if you wished!
This is just a suggestion, and I have not bench tested it, but would like to hear back from anyone who tries it.
PS:
it´s better if you can Keep the water out of the first and third fresnel zones http://en.wikipedia.org/wiki/Fresnel_zone
so higher antennas and 5GHz should be better than 2.4 (small fresnel zone)
