Can modems be used to communicate with RC Submarines?

Question

With reference to this video"R/C Sub, Scratch-built From Sewer Pipe" when the sub went to deep it lost the signal to the RC receiver.

Question: Since water transfers sound better than radio waves would it be possible to connect two 33.3kb/s modems via a water medium?

The thinking is that the modems will be connected to hydrophones that will sit in the water and in the sub. They will negotiate a link at start-up and maintain that link (full duplex). The some simple uC will interoperate the digital data into actionable commands.

Answer 1

It could work, but there would be some major disadvantages to the approach:

• Achieving a sufficient signal-to-noise ratio (SNR) will be very difficult, as you likely cannot control sources of noise. You'll probably need high-power directional transducers to achieve a strong enough signal to drown out the noise. Sorry for the pun.
• A commercial off-the-shelf modem has been designed to work with a copper transmission line and electrical signals. Although the speed of sound in water is higher than in air, it still comes nowhere close to the propagation rate of electrical charges in copper. The modem will definitely have trouble dealing with the added latency.

Using pressure waves to communicate across water can be done, but you'll be much better off if you go ultrasonic. Here's why:

• For starters, high power directional ultrasonic underwater transducers are cheap and very easy to find. Just look for the business end of a discarded echo sounder.
• You'll get a better SNR with less effort, since there are simply fewer ultrasonic noise sources around. Unless you'll be operating your submarine in dolphin territory, of course ;-).
• As a rule of thumb, you'll require less power to transmit a higher frequency signal, which is probably nice unless you're building a nuclear sub.

Answer 2

Sending pressure-wave signals underwater could be an interesting communications method, but I don't think any off-the-shelf POTS modems are going to be a good fit for the task. High-speed modems require a bidirectional link with stable characteristics. Any time the characteristics change significantly, the modems will have to spend a few seconds retraining each other. Such changes could occur when the sub approaches obstacles--not a good time to lose communication. Lower-speed modems are less sensitive to transmission characteristics, but noise at certain key frequencies will render them completely unusable; they have zero ability to adapt and select other frequencies.

Digital signal processing is sufficiently cheap that I would suggest rolling your own communications method. I would suggest using something like quadrature phase modulation on many independent frequencies at once with a low data rate at each frequency (maybe 50 baud), and design your communication scheme with enough redundancy so that even if roughly half the frequencies are unavailable you will still pass enough information to be useful (e.g. use 48 frequencies from 10KHz to 18Khz to send 16 bits of data and a 32-bit CRC, and assign the data from each frequency a "confidence" value; if the CRC comes out wrong, find the combination of bits with the lowest confidence values which, if flipped, would make it right).