Author Topic: 1200 baud data transfer over audio passband of cellphone. Is that possible?  (Read 5898 times)

0 Members and 2 Guests are viewing this topic.

Offline sv3oraTopic starter

  • Regular Contributor
  • *
  • Posts: 84
  • Country: gr
1200 baud data transfer over audio passband of cellphone. Is that possible through the use of their codecs? Has anyone tried it? There are videos about comms on 300 baud but how about 1200 baud, has anyone tried it?
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 8526
Tried, yes. Successfully, probably not. GSM audio codecs compress the signal a lot.
 
The following users thanked this post: sv3ora

Offline NiHaoMike

  • Super Contributor
  • ***
  • Posts: 9321
  • Country: us
  • "Don't turn it on - Take it apart!"
    • Facebook Page
The modern "HD Voice" codecs should give a better chance of working. But if you have that, why not just write an app to interpret the tones on the smartphone itself?
Cryptocurrency has taught me to love math and at the same time be baffled by it.

Cryptocurrency lesson 0: Altcoins and Bitcoin are not the same thing.
 
The following users thanked this post: sv3ora

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
1200 baud data transfer over audio passband of cellphone. Is that possible through the use of their codecs? Has anyone tried it? There are videos about comms on 300 baud but how about 1200 baud, has anyone tried it?
Most of the speech codecs used for cellular assign a single pitch for each 5ms of speech, which is a fast enough update rate for smooth speech. This is true whether you use the narrow band or the wide band HD voice codecs. However, even the Bell 103 or V.21 modems at 300bps need a pitch update every 3.3ms, so the cellular codecs can't cope. The standard FSK and PSK modems for 1200bps (V.23, Bell 212, V.22) need a pitch update every 1.7ms or 0.83ms, so they are even more heavily corrupted by the speech codecs. You could design a custom modem that will get some data through fairly reliably, by being adapted to the qualities of low bit rate speech codecs. I don't know if anyone has tried to do that.
 
The following users thanked this post: sv3ora

Offline BrianHG

  • Super Contributor
  • ***
  • Posts: 8275
  • Country: ca
    • LinkedIn
1200 baud data transfer over audio passband of cellphone. Is that possible through the use of their codecs? Has anyone tried it? There are videos about comms on 300 baud but how about 1200 baud, has anyone tried it?
Most of the speech codecs used for cellular assign a single pitch for each 5ms of speech, which is a fast enough update rate for smooth speech. This is true whether you use the narrow band or the wide band HD voice codecs. However, even the Bell 103 or V.21 modems at 300bps need a pitch update every 3.3ms, so the cellular codecs can't cope. The standard FSK and PSK modems for 1200bps (V.23, Bell 212, V.22) need a pitch update every 1.7ms or 0.83ms, so they are even more heavily corrupted by the speech codecs. You could design a custom modem that will get some data through fairly reliably, by being adapted to the qualities of low bit rate speech codecs. I don't know if anyone has tried to do that.
Being a digital system, there is a lack of analog noise, so, you might be able to take advantage of that.  For example, try a 100hz sine wave or square wave tone with 64QAM levels.  At this low frequency, you are relying on the cell networks speech codec just to cope with 100hz, but, modulate the amplitude to 16 levels delivering 4 bits.  Scope the in and out, if it works, you have 600 baud.  If you get lucky enough, operating on the peaks and troffs of the sine wave with 64QAM, just maybe you can squeeze out 1200 baud.  Such a dumb codec shouldn't be too difficult just to try and play with.  It should also be easy enough at this point to try 150Hz, 16QAM to achieve the same 600/1200 baud.
« Last Edit: October 29, 2019, 03:27:13 am by BrianHG »
 
The following users thanked this post: sv3ora

Offline Psi

  • Super Contributor
  • ***
  • Posts: 10385
  • Country: nz
1200 is still pretty slow.

I'm sure that is possible if you picked the right codec and optimized for the audio frequency band that are passed by cellphone audio.

I wonder how fast you could get with a TCM3105 FSK MODEM IC.
People use them on FPV planes to encode a UART onto the audio side channel that 5.8Ghz video transmitters have.
« Last Edit: October 29, 2019, 04:01:42 am by Psi »
Greek letter 'Psi' (not Pounds per Square Inch)
 
The following users thanked this post: sv3ora

Offline sv3oraTopic starter

  • Regular Contributor
  • *
  • Posts: 84
  • Country: gr
It seems to be a bit of yes and no from people I have asked, both on other forums and here.
Is it the "single pitch" problem mentioned in this thread? Is it the heavy compression of some codecs? It is the audio passband?
How about trying different mark/space frequencies (which will make the modem non-standard)? How about increrasing the audio volume to distortion levels? How about passing fake voice signals to the passband other than the mark and space frequencies to trick the codec?

I think that I will have to test it myself to be sure, as it seems to me not many tries have been made.
I will setup two PCs with MixW, set for packet radio. MixW modem, allows the standard rates and mark/space frequencies, but it also allows defining custom rates and mark/space frequencies. So I could play around with these settings and connect the audio on the hands-free jacks of two smartphones.
I believe that if it works for me, will work on every country, since the cellphone networks are compatible? (eg. if I visit another country roaming works).

Any comments are welcome.
« Last Edit: October 29, 2019, 09:19:11 am by sv3ora »
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Most of the speech codecs used for cellular assign a single pitch for each 5ms
Then use OFDM modulation with symbol length > 5ms.
 
The following users thanked this post: Someone

Offline jhpadjustable

  • Frequent Contributor
  • **
  • Posts: 295
  • Country: us
  • Salt 'n' pepper beard
Meanwhile, in these first years of the 21st century ;) mobile phones don't pass "audio". They send a ~13:1 compressed description of audio in 20ms frames, which include the least data necessary to synthesize something that sounds sort of intelligible and recognizable to a human ear on the far end. They also interpolate any missing packets on the far end, and might dynamically shift to a half-rate version by the network or terminal equipment. These days, it's probably even more complex and more attuned to decoding the human vocal tract. (All the better for mass speech-to-text to hear you, my dear!)

Random university PowerPoint PDF illustrating what OP is up against.

The academic work in the area is certainly worth a google. Researchers have recently managed to get a "subliminal" channel through the GSM codec. Perhaps not all is lost.

I'd suggest a $5 GSM module, if the goal is to get data from point A to point B through the GSM network, rather than an exercise in signal processing and fighting the algorithm. :)
"There are more things in heaven and earth, Arduino, than are dreamt of in your philosophy."
 
The following users thanked this post: tooki

Offline sv3oraTopic starter

  • Regular Contributor
  • *
  • Posts: 84
  • Country: gr
Sure they are digital! But the end result is the same (or mostly). Every digital system does not carry the real waveform, but a representation of that waveform. But the end result must be the same (or a good representation), if humans need to hear it.
I am thinking it that way:

Can you pass a single continuous tone of 1200Hz through GSM?
Can you pass a single continuous tone of 2200Hz through GSM?
Can you pass alternatively switched 1200/2200 tones in low speed through GSM?
Can you switch these tones more frequently and how much more?

This is the end result and this is what's of interest at the end.
 

Offline TheUnnamedNewbie

  • Super Contributor
  • ***
  • Posts: 1211
  • Country: 00
  • mmwave RFIC/antenna designer
Audio on a cellphone tends to also be compressed in very specific ways that are optimized for voice (which is why the 'please wait' music with callcenters is so horrible). I suspect you will find it very hard to send any data through it as it too will be crazy deformed by the codecs.
The best part about magic is when it stops being magic and becomes science instead

"There was no road, but the people walked on it, and the road came to be, and the people followed it, for the road took the path of least resistance"
 
The following users thanked this post: sv3ora

Offline sv3oraTopic starter

  • Regular Contributor
  • *
  • Posts: 84
  • Country: gr
I have encountered this situation you mention, only on remote areas where there is no good signal, or when I used very old phones (1998 era). Dropouts of audio parts or "pixelated" sounds in some audio parts. I have not come across similar situations when there is a good signal.

Note, for FSK, audio amplitude compression should be less critical, as it is the frequency of the tones that matter.
« Last Edit: October 29, 2019, 12:16:18 pm by sv3ora »
 

Offline TheUnnamedNewbie

  • Super Contributor
  • ***
  • Posts: 1211
  • Country: 00
  • mmwave RFIC/antenna designer
Note, for FSK, audio amplitude compression should be less critical, as it is the frequency of the tones that matter.

Don't most audio compression algorithms do a lot of stuff in frequency/wavelet domain where frequencies/wavelets that are considered secondary to the main few peaks are just removed?
The best part about magic is when it stops being magic and becomes science instead

"There was no road, but the people walked on it, and the road came to be, and the people followed it, for the road took the path of least resistance"
 
The following users thanked this post: sv3ora

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
I suspect you will find it very hard to send any data through it as it too will be crazy deformed by the codecs.
Not true. Speech indeed is spoken *data*. Question here is how to modulate digital data such a way that speech codecs do not affect audio signal during transmission.
 
The following users thanked this post: sv3ora

Offline sv3oraTopic starter

  • Regular Contributor
  • *
  • Posts: 84
  • Country: gr
Question here is how to modulate digital data such a way that speech codecs do not affect audio signal during transmission.
Precisely!

I think I am going to do the test with the MixW modem to see what will happen. We may come across pleasant discoveries...or not.
« Last Edit: October 29, 2019, 12:37:44 pm by sv3ora »
 

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
Audio on a cellphone tends to also be compressed in very specific ways that are optimized for voice (which is why the 'please wait' music with callcenters is so horrible). I suspect you will find it very hard to send any data through it as it too will be crazy deformed by the codecs.
Yep. Speech codecs are so speech specific that most other things, including multiple voices speaking at the same time, sound awful.
 

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
Most of the speech codecs used for cellular assign a single pitch for each 5ms
Then use OFDM modulation with symbol length > 5ms.
What does the modulation have to so with the codec being used? They are completely unrelated issues. Most forms of OFDM can add a lot of latency to the channel, but that's about the only specific effect it has on the signal.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Most of the speech codecs used for cellular assign a single pitch for each 5ms
Then use OFDM modulation with symbol length > 5ms.
What does the modulation have to so with the codec being used? They are completely unrelated issues.
You talked about standard FSK and PSK modulations, their short symbol length (update every 1.7ms or 0.83ms). I offer OFDM modulation because it has slong(er) symbols that can be stretched depending on FFT order used
« Last Edit: October 29, 2019, 04:49:56 pm by ogden »
 

Offline borjam

  • Supporter
  • ****
  • Posts: 908
  • Country: es
  • EA2EKH
Old telephone modems based on FSK were the V.21 (300 bps in full duplex) or the asymmetric V.23 (1200/75) intended for videotex systems.

The problem with voice compression algorithms is that they degrade the signal in a way that is mostly harmless for speech intelligibility but they can make it impossible for a data modem to work. Especially if you use more complex modulation systems.

I remember when some telco in Spain begun using compressors in some of their voice trunks. 28800 bps modems were unable to link beyond 9600 bps.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
The problem with voice compression algorithms is that they degrade the signal in a way that is mostly harmless for speech intelligibility but they can make it impossible for a data modem to work. Especially if you use more complex modulation systems.

BS. Many research papers prove you wrong, achieves >1200 bps data transmission rates over GSM-FR (worst case codec, rarely used in modern networks). There are some "secure phone" products that uses not only encrypted VoIP but audio channels as well.  You may start here:

https://www.researchgate.net/publication/304296823_Data_transmission_via_GSM_voice_channel_for_end_to_end_security
 

Online coppice

  • Super Contributor
  • ***
  • Posts: 10034
  • Country: gb
Most of the speech codecs used for cellular assign a single pitch for each 5ms
Then use OFDM modulation with symbol length > 5ms.
What does the modulation have to so with the codec being used? They are completely unrelated issues.
You talked about standard FSK and PSK modulations, their short symbol length (update every 1.7ms or 0.83ms). I offer OFDM modulation because it has slong(er) symbols that can be stretched depending on FFT order used
Oh, sorry, I was looking at what you said in the wrong way. OFDM won't really help, though. It gets high throughput through parallelism. Good luck getting any parallel signals through a speech codec.
 

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
OFDM won't really help, though. It gets high throughput through parallelism. Good luck getting any parallel signals through a speech codec.
Agreed. I mentioned OFDM just because it kinda solves "too short symbol problem". Modulation shall be tuned for LPC-based speech codec to pass through. Anyway paper I mentioned and many references in it, not mentioning commercial "secure phone" products, proves that it is possible indeed.
 

Offline sv3oraTopic starter

  • Regular Contributor
  • *
  • Posts: 84
  • Country: gr
Anyway I have just tested it. I connected a smart phone from it's hands free port to the PC mic and phones. then I called the smart phone from a land line at home.

It could not even send 110 baud. It started sending them ok but after 1-3 seconds they were suppressed. Continuously changing the AF volume to 0% and then back to a percentage tricked the codec somehow.

Sending pulses of a single tone passed ok.
Also, sending a dual tone passed ok.
But not FSK at 110, 300 or 1200baud, they all seemed to have this problem I mentioned.
Also when sending the above and tried to also send FSK at the same time, again everything was ceased at the channel and nothing heard at the other end, even my sound card was output tones.
I have also tried SSTV and the result was that some tines refused to be transmitted.

Now, I do not know if this is due to the mobile phone codecs or the landline phone codecs, as I have not tried it with two mobile phones yet.

I may have a more detailed analysis on it, maybe note which tones pass ok and which not, or if it is not a tone problem, identify why the codecs cause silence on the channel.
« Last Edit: October 29, 2019, 09:52:59 pm by sv3ora »
 

Offline borjam

  • Supporter
  • ****
  • Posts: 908
  • Country: es
  • EA2EKH
The problem with voice compression algorithms is that they degrade the signal in a way that is mostly harmless for speech intelligibility but they can make it impossible for a data modem to work. Especially if you use more complex modulation systems.

BS. Many research papers prove you wrong, achieves >1200 bps data transmission rates over GSM-FR (worst case codec, rarely used in modern networks). There are some "secure phone" products that uses not only encrypted VoIP but audio channels as well.  You may start here:

https://www.researchgate.net/publication/304296823_Data_transmission_via_GSM_voice_channel_for_end_to_end_security
Amazing work, thanks :)

I agree that I shouldn't say "impossible" but "very hard".
 
The following users thanked this post: ogden

Offline ogden

  • Super Contributor
  • ***
  • Posts: 3731
  • Country: lv
Now, I do not know if this is due to the mobile phone codecs or the landline phone codecs
Landline usually have 8ksps @8bit PCM, no codecs. Who uses landline today anyway  :-//
 
The following users thanked this post: sv3ora


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf