If the boats anchor, they will surely swivel around with the wind and water currents.
So this would mean an omni that can go 9 miles would be needed. Thats quite difficult, unless you were to use an array of grid antennas on the shore.
A little impractical.
An old WRT54G or Ubiquiti PowerStation EXT used to have a feature where it was an 802.11g device, but had two antennas and was able to automatically use whichever antenna had/gave the best signal. So one of these hooked up to a pair of 180 degree sectors would work. Unfortunatley, also impractical because its just old.
So next I would look at a dual chain 802.11n device such as a ubiquiti rocket which will do the same thing - though it would only be one chain operating at a time so you wouldnt expect huge speeds. This is the best option I think.
Although you could probably use mikrotik or another brand of radio, I would try out the following.
1) Set up a 5ghz sector using a rocket and use horizontal polarity only. Set the rocket to operate in single chain mode, so it doesnt try to use the vertical polarity.
Aim the sector from your high site out to sea.
If you use the 120 degree high gain rocket antenna, which is really a 90 degree, you could use two and provide 180 degree coverage with two rockets on different channels.
2) On the client end, use 4x 120 degree sectors- could probably use the lower gain ones as they are half the size.
Again you only use the vertical polarity.
3) Connect the back-to-back antennas to a signal splitter such as this one.
»
www.l-com.com/bandpass-f ··· combinerIt will split the signal coming in/out, and therefore you would need to turn up the power level by an extra 3db to compensate or whatever your local laws allow while staying under your eirp limit.
4) Connect the North+south antennas to the chain 1 output on the rocket via the splitter, and then connect the east+west to the chain 2 output on the rocket via another splitter.
So my theory is, that by also putting the client into single chain mode, it will only use the chain that gives the best signal. Need confirmation on this. So it will use whichever pair has an antenna facing the high site, with the other antenna broadcasting out to sea and not picking up noise from the shore.
Power levels would need to be as such that you are compensating for the splitter.
This means the one client radio will need to be connected to the high site, and you are not trying to use a fancy router with failover mode and 4+ wan ports to try and get a connection from whichever radio has a signal - and not necesarily the best signal as the boat rotates.
The one client radio can stay connected, and maybe have a momentary drop as it needs to switch antenna chains every ten minutes as the boat rotates.
Another option as you suggested is to use an array of nanostations. The problem there is how do the computers within the boat know which one to use to connect to the internet?
Just found a 180 degree sector antenna. Its only 10dbi though.
»
www.ebay.com/itm/5-Pack- ··· e28a737cWHT or someone may want to tell us -
A splitter being fed with a 25db signal, will that mean 22db or 12.5db per output? Less the additional .5db insertion loss?
The answer to that question - I think its 22db per output, will mean that two 90 degree 19dbi antennas split would be still more effective than a single 180 degree 10dbi antenna .
We are also assuming that the sea is rather flat. The CPE antenna beam height may shoot over or under the high site as the waves rock the boat. So in that respect, a lower gain 180 degree antenna would be better than a higher gain one because the higher gain has less beamheight.