I think reliable 1Gbit RF transmission is the biggest problem here.
Personally, I would take let's say 5-10x WiFi 11n transcievers in 5Ghz band so that you can have multiple 40Mhz channels there without interference.
So, I would start from cheapest FPGA board and single transciever, and once you master that, you will see which bigger board you would need, it's totally depend on huw much logic you want to implement there.
Another way to think is compressing video, or going stright to DLNA (which is basically compressed video over WiFi, standardized).
As with any radio system the bottom line is what can be achieved without the hindrance of obstacles. In free-space the baseline is dependent on what frequency you transmit at and how sensitive your receiver can be. The power path loss equation is this:
Loss (dB) = 32.45 + 20\$log_{10}\$(f) + 20\$log_{10}\$(d)
Where f is in MHz and d is in kilometres. This equation tells you how many dB of power loss you can expect at a given distance with a given carrier frequency.
It's formulated from work by Harald T. Friis and a decent reference is here. It doesn't assume any gain in the antennas, so it is a real baseline. Antenna gain brings lower path loss but greater directionality, and for a dipole the gain is about 1.76 dB.
If you are transmitting 10 dBm (10 mW) at 100 MHz over a distance of 10 km, you can expect to receive a power of:
10 dBm - (32.45 + 40 + 20) dBm = -82.45 dBm (5.7 nW)
How much power does your receiver need? A useful equation is this:
Power (dBm) needed by a receiver is -154 dBm + 10\$log_{10}\$(data rate) and as you have a data rate that is quite low you can expect better performance than Wi-Fi!
The question states 30 characters every ten minutes. I'm going to assume that this is transmitted as a burst of 30 x 10 bits in 10 seconds plus a preamble of 100 bits to get the receiver locked in - that's 400 bits in 10 seconds or 40 bits per second.
The receiver power required is therefore -154 dBm + 10\$log_{10}\$(40) dBm = -138 dBm
This assumes that the transmitter and receiver are bespoke items designed to run with this low data rate. It's not easy to get a receiver sensitivity below -120 dBm, so whatever radio system you use read the small print and investigate. Off-the-shelf items are probably incable of transmitting at very low data rates, so they should be avoided.
Anyway, you need to receive -138 dBm and, over 10 km with a 100 MHz carrier, you can expect to receive -82.45 dBm. This sounds good enough considering that you may get a few more dBs from antenna gain.
But on earth, no matter what the terrain appears to be, there will be added attenuations that are really difficult to account for and describe here. There's a thing called fade margin and this, as a rule of thumb basically says - try to ensure your received power is at least 20 dB greater than its baseline sensitivity - this means that if you designed a receiver requiring -120 dBm you should expect to receive -100 dBm on a good day.
Given the terrain I'd hazard a wild guess that you need to have 20 dB more up your sleeve and this just about brings you to being able to get 10 km from a 10 mW transmission.
Hopefully you now have the formulae to work out what power you need to get 15 km. Another helpful fact is that the target may not be moving very fast and can be tracked from one end using a high gain antenna such as a Yagi-Uda antenna. You might have seen wild-life films where a bear or a puma has a radio transmitter attached to them and some guy in a field is pointing the Yagi-Uda antenna this way and that in order to locate the direction the animal is in. This might work to improve things.
Best Answer
You'll find that microwave ovens, aluminum foil, metal garbage cans etc aren't really reliable attenuators.
What you need is a channel model that tells you how much your signal will be attenuated after 1km, but also how bad your multipath will be like, together with a signal model that tells you what effect that has in your reception.
I don't know the devices you're referring to, but usually I'd expect an antenna connector to be present in anything that I'd expect to work over such distances, so instead of an antenna, connect your transmitter and receiver with coax cable and attenuators to emulate the attenuation. It's much harder to emulate the multipath environment and temporal effects like fading, but with your signal model you might be able to calculate an attenuation that would be "just as bad" as an actual multipath channel.