An unreplied comment on a previous submission on HN was questioning the remote feasibility of running TCP over FM with the Raspberry Pi. I too am curious about the state of the art in general, if any, and how it might possibly be exploited for long-range networking between small computers like the Pi.
Yes, but it isn't practical or effective. You would be re-inventing 802.11 very poorly.
1) You could use the given technique to create the MOdulation side of a modem[1]. You would have to solve the receiver and DEModulation side. Not likely to be do-able with a bare Raspberry Pi so you will end up adding a receiver and probably a demodulator (software demodulation is theoretically within the grasp of a Raspberry Pi).
2) Your data rate would be dependent on how well you implement your modem. Assuming you limit yourself to audio range frequency modulation (i.e. 10-3000Hz), you will have a hard time exceeding 300 baud using frequency shift keying (e.g. modulate with two tones, one indicating '0' and the other indicating '1'). If you use phase shift keying, you may achieve 2400 baud, but (a) I really doubt that a Raspberry Pi can do phase shift keying, (b) demodulating PSK is much more difficult that frequency shifting, and (c) in my limited experience, attempting phase shift modulation on top of a FM carrier doesn't work for crap (FM can be viewed as a type of phase shift modulation, so you end up phase shift modulating a phase shift modulator which doesn't demodulate + demodulate reliably into your baseband signal).
Higher data rates over phone lines encode multiple bits per baud (signaling unit), making the encoding and decoding tasks go through the roof if done in software. Getting to e.g. 56Kbaud of the end-of-the-road phone modem is impossible - 56Kbaud only worked when the phone company digitized the phone audio, and rather iffy even then.
It's already possible[1], but legislation varies by country. Line-of-sight, 50m from buildings and people (other than the operator and those under the operators control) and a 150m height limit are common to reduce the risk to commercial aviation, and of property damage or injury.
This is something I've been wondering about after seeing https://www.youtube.com/watch?v=2P17LPYnAAg - they seem to be able to use a signal relay that works on a long range with both the control and video channels. So it's possible with some equipment, but not sure if it's custom built or off the shelf.
There are UHF radio modems that are commonly used for radio telemetry for amateur UAVs (typically RC planes or multi-rotors), transmitting on 900MHz or 433Hz using spread-spectrum technology.
Based on HopeRFs HM-TRP module, they present a serial port, either TTL, or as a virtual COM port over USB using a (defacto-standard) FTDI USB to serial chip. You can use two serial, two USB, or one of each - the radio modem in them is identical, it's just the presentation that differs.
Line-of-site range depends on the data rate, but is in the region of 1 - 2km, but up to 24km is possible with amplification. The data rate is up to 128kb/s.
Lots of technical details can be found on the ardupilot wiki[1].
They were designed as a lower cost, longer range alternative to Xbee [2]. Xbee support ZigBee, multipoint, mesh and other protocols, but not the 433MHz frequency legal in Europe.(900MHz is fine for the US, but the alternative, 2.4GHz, is the same frequency commonly used for spread-spectrum radio-control, so risks interference or reduced range.)
They can be purchased in the US for $75 a pair [2]. Clones can be found can be found from Chinese suppliers for less than $40 a pair [3].
It should be entierly possible to modulate packet radio using the GPIO hack, however you would probably need a real-time kernel extension to ensure proper transmission. Decoding is easily possible on the CPU as well, provided you attach soeme RF demodulation circuitry, e.g. to a sound card input.
