sending data over gsm voice like bell 202 modem

I thought about DTMF today, I tried to find some ideas to enhance DTMF.

If you check the link posted by Coolmirza http://www.twopaths.com/Software/index.htm, there is a wave file ( DTMF_Decoder.wav) where you can ear some DTMF tones sent. The guy is sending 12 tones in about 2 sec.I think we can send a bit faster, let's say 10 tones/sec.

The DTMF is the combination of 2 different frequencies (or harmonics). There are 4 different harmonics, so the quantity of information sent with 1 tone is 4*4=16 (or 4 bits).
I think the DTMF was originally designed like this because the phone's keyboard can be seen as a matrix (2 dimensions) of 12 keys.

But in our case we can think of using more dimensions, such as 3-dimension instead of the usual bi-dimensional signal. In other words, instead of using a matrix, we would use a cube with 4 more frequencies. Thus we would have 4*4*4 = 64 possibilities every time a signal is sent.

But this is not all. We can have even more information if we consider that a signal can be the combination of only one harmonic, or 2 or 3. The total of different harmonics is 12.
So this gives: 12 possibilities ( for a single tone) + 3*16 (for a dual tone) + 64 (for a combination of 3 tones) = 124 possibilities ~ 7 bits
With a rate of 10 signals/sec, this gives approximatively 70 bauds.

I wonder how far we could go like this, by allowing the combination of more harmonics like 4 or 5..

I'm not sure if it makes sense. Sorry if the explanation isn't clear enough.

The difficulty with this method will remain in the ability to detect the different harmonics of the signal.
With DTMF you decode 2 harmonics but here it will be 3 or even more.

I'm not an exert in signal processing, does anyone know about this domain?

Sorry I just realized harmonic is not the right term here. Just consider harmonic = frequency.

DTMF Decoder in VB
http://www.fullspectrum.com/deeth/

But in our case we can think of using more dimensions, such as 3-dimension instead of the usual bi-dimensional signal. In other words, instead of using a matrix, we would use a cube with 4 more frequencies. Thus we would have 4*4*4 = 64 possibilities every time a signal is sent.

But this is not all. We can have even more information if we consider that a signal can be the combination of only one harmonic, or 2 or 3. The total of different harmonics is 12.
So this gives: 12 possibilities ( for a single tone) + 3*16 (for a dual tone) + 64 (for a combination of 3 tones) = 124 possibilities ~ 7 bits
With a rate of 10 signals/sec, this gives approximatively 70 bauds.

The difficulty with this method will remain in the ability to detect the different harmonics of the signal.
With DTMF you decode 2 harmonics but here it will be 3 or even more.

mutonic

u conceive a very good of using 3 or more frequencies for generating DTMF signal
i will reply u later after renewing my knowledge, i think it is a little bit difficult to do

chek dis work of Matlab u can edit dis for generating 3 or more frequencies
hope this will help u to understand the limitations of generating TTMF instead of DTMF

http://rapidshare.com/files/339762042/DTMFwithGUI.zip.html

I have some good news!

I tried the AFSK method by making some waveforms using WaveLab and by following the specifications of the Kansas City Standard http://en.wikipedia.org/wiki/Kansas_City_standard and it seems to work!

I sent the waveform through my first phone and received with the second plugged to the input of my sound card and then compared the two waveforms.
There are still some imperfections but globally the received waveform seems quite correct and I think could be exploited.

I think some improvements can be made: I couldn't manage to stream the ouptut of my sound card into the hands free kit of my 1st phone. So I had to use the microphone instead which introduces some noise.

Hopefully with this method we could reach 300 bauds.

I have to do further testing with real data and using a real encoder/decoder to ease the task.
Also if you know how to send the audio without the microphone that would be great.