sending data over gsm voice like bell 202 modem

Hi,

I already opened a few topics but I thought it was better to create a new one as it is another subject.

The aim of this project is to control a device which is very far using a gsm but without GPRS. This because, mobile data package (for instance in France) aren't cheap or not convenient at all. SMS are neither convient here for what i want to do.

The idea is to send basic unix commands to a linux board and retreive the output of the shell terminal coming from the device through gsm voice. I know this is quite ambitious project..! The point here is that only few data need to be exchanged.

After some research, I found that originally some modem like the Bell 202, were sending data through audio frequency range using audio frequency-shift keying (AFSK) at a rate of about 1200 bit/s).

So basically we would encode the data into audio, send the audio through gsm, and decode the audio strean when received.

I know this can sounds crazy, but do you think there is any way it could work? or there will be troubles because of signal quality poorness, ...

wat about using DTMF? is it possible to use DTMF for wat u wanna do?

I think it can be a better idea.

http://electrosofts.com/dtmf/index.html

http://www.gsm-modem.de/dtmf-audio-codec.html

http://www.activexperts.com/xmstoolkit/atcommands/dtmf/

Hey! Thanks for your help.

I checked your links which I found iinteresting... I wonder which rate we could reach using DTMF?
Actually, AFSK seems limitated to me, I wonder why it only uses 2 frequencies to represent bits.

I was thinking of using a base 2^8 instead of base 2, with 256 different frequencies representing ASCII characters. For instance frequencies would be taken between 2000 and 4000 Hz which is probably the best quality range (as it corresponds to human voice).

What do you thnik? Is 15 Mhz gap between frequencies enough to be recognized? What should be the sampling rate?

Actually, AFSK seems limitated to me, I wonder why it only uses 2 frequencies to represent bits.
I was thinking of using a base 2^8 instead of base 2, with 256 different frequencies representing ASCII

mutonic

These days creating and decoding tones is a *lot* easier, but fewer is simpler.
If I were you I'd assume those guys knew what they were doing and just copy them

For instance frequencies would be taken between 2000 and 4000 Hz which is probably the best quality range (as it corresponds to human voice).

mutonic

I think phone networks tend work between 300Hz and 4kHz. (worth checking).

Is 15 Mhz gap between frequencies enough to be recognized?

mutonic

15 Mhz ??

DTMF is rite choice for your project because there is no issue of voice spectrum attenuation and distortion in DTMF

u can understand DTMF generation and sampling rate in the following link.

wat u wanna do with data rate? why r u asking about dis?

http://forums.techarena.in/off-topic-chat/1234038.htm

If I were you I'd assume those guys knew what they were doing and just copy them

ExperimenterUK

That's what I think too. If they could have done it different, they would have to
Actually I see one big problem wit my method, the ambiant noise will much likely contains those frequencies and trigger unwanted signals. So it is better to use only 2 frequencies or the DTMF method described by coolmirza.

15 Mhz ??

ExperimenterUK

Oups sorry my mistake, 2000Hz divided by 256 gives 7.8 hz gap between frequencies

wat u wanna do with data rate? why r u asking about dis?

coolmirza143

I was a bit suspicious about the rate we could reach with DTMF but from your link:

DTMF tones 8 kHz sample rate at 8 bit resolution

Does it mean we can send one character (8 bits), 8 000 times per seconds?? which would give 8 Kbyte rate. I really doubt it, I must have misunderstood something...Actually, I don't see where the 8 bits comes from. We only use one matrix of 4x4 tones, wich gives 16 differents possibilites. Could you explain more in details?

Any way thanks for ur help guys... it seems after all this thing could actually work ??!:)

yes data rate for DTMF using PSTN is 8 Kbyte

http://www.cs.cf.ac.uk/Dave/Multimedia/node150.html

Complete detail for DTMF signal generation and Recognization

http://focus.ti.com/general/docs/lit/getliterature.tsp?baseLiteratureNumber=slaae16

DTMF coding and Decoding in C

http://www.phrack.com/issues.html?issue=50&id=13&mode=txt

Thanks for those relevant links.
I checked TI doc and according to them:

Depending on the complexity of the filter and the necessary sampling rate, signals of up to 5kHz can be acquired with this system

if this is true that would be awesome.
But I really doubt we can reach same rates as a 56k traditional modem..??