This program can be used to transmit a specified file between 2 computers, using a simple audio cable (for better SNR and higher speeds) or a simple headset, allowing true air-gapped communication (via a speaker and a microphone).
The sender side uses send.py script to modulate the input data into an 32kHz
audio file, which is played to the sound card, using aplay Linux utility.
The receiver side uses arecord Linux utility to record the transmitted audio
to an audio file, which is demodulated concurrently by the recv.py script.
The process requires a single manual calibration step: the transmitter has to find maximal output volume for its sound card, which will not saturate the receiving microphone.
The modem is using OFDM over an audio cable with the following parameters:
- Sampling rate: 32 kHz
- BAUD rate: 1 kHz
- Symbol modulation: 64-QAM
- Carriers: (1,2,3,4,5,6,7,8) kHz
This way, modem achieves 48kpbs bitrate = 6.0 kB/s.
A simple Reed-Solomon ECC is used, with (255,245) rate = ~3.9% overhead.
Make sure the following Python packages are installed:
$ sudo pip install numpy reedsolo bitarray
For graphs and visualization (optional), install:
$ sudo pip install matplotlib
Connect the audio cable between the sender and the receiver, and run the following scripts:
- On the sender's side:
~/sender/amodem $ ./calib.py send
- On the receiver's side:
~/receiver/amodem $ ./calib.py recv
The sender computer's output audio level should be increased, until the received amplitude and peak values are not higher than 0.5, while the coherence is 1.0 (to avoid saturation).
See http://youtu.be/iCg1tepGz10 for calibration demo.
See http://youtu.be/94yS3IZmtho for usage demo.
- Prepare the sender:
~/sender/amodem $ dd if=/dev/urandom of=data.send bs=125kB count=1 status=none
~/sender/amodem $ ./send.py <data.send >audio.pcm
2014-08-01 21:00:06,723 INFO Running MODEM @ 48.0 kbps
2014-08-01 21:00:06,773 INFO 3.210 seconds of training audio
2014-08-01 21:00:07,712 DEBUG 8.886 seconds of data audio
2014-08-01 21:00:08,714 DEBUG 14.941 seconds of data audio
2014-08-01 21:00:09,712 DEBUG 20.994 seconds of data audio
2014-08-01 21:00:10,381 INFO 21.846 seconds of data audio, for 125.000 kB of data
~/sender/amodem $ sha256sum data.send
008df57d4f3ed6e7a25d25afd57d04fc73140e8df604685bd34fcab58f5ddc01 data.send
- Start the receiver:
~/receiver/amodem $ ./wave.py record rx.int16 &
~/receiver/amodem $ ./recv.py <rx.int16 >data.recv
2014-08-01 21:00:30,068 INFO Running MODEM @ 48.0 kbps
2014-08-01 21:00:30,068 DEBUG Skipping first 0.100 seconds
- Start the sender:
~/sender/amodem $ ./wave.py play audio.pcm
- The receiver should print similar messages during demodulation:
2014-08-01 21:00:34,847 INFO Carrier detected at ~4910.0 ms @ 1.0 kHz: coherence=99.991%, amplitude=0.485
2014-08-01 21:00:34,847 DEBUG Buffered 1000 ms of audio
2014-08-01 21:00:34,883 INFO Carrier starts at 4909.844 ms
2014-08-01 21:00:35,447 INFO Prefix OK
2014-08-01 21:00:35,656 DEBUG Current phase on carrier: 0.260
2014-08-01 21:00:35,656 DEBUG Excepted phase on carrier: 0.250
2014-08-01 21:00:35,657 INFO Frequency error: -13.24 ppm
2014-08-01 21:00:35,657 INFO Sampling error: 0.31 samples
2014-08-01 21:00:36,623 INFO 1.0 kHz: Noise sigma=0.0030, SNR=50.4 dB
2014-08-01 21:00:36,627 INFO 2.0 kHz: Noise sigma=0.0025, SNR=52.0 dB
2014-08-01 21:00:36,631 INFO 3.0 kHz: Noise sigma=0.0021, SNR=53.5 dB
2014-08-01 21:00:36,636 INFO 4.0 kHz: Noise sigma=0.0029, SNR=50.8 dB
2014-08-01 21:00:36,643 INFO 5.0 kHz: Noise sigma=0.0038, SNR=48.3 dB
2014-08-01 21:00:36,649 INFO 6.0 kHz: Noise sigma=0.0046, SNR=46.7 dB
2014-08-01 21:00:36,657 INFO 7.0 kHz: Noise sigma=0.0053, SNR=45.5 dB
2014-08-01 21:00:36,664 INFO 8.0 kHz: Noise sigma=0.0066, SNR=43.6 dB
2014-08-01 21:00:36,840 INFO Demodulation started
2014-08-01 21:00:37,591 DEBUG 6.0 kB, realtime: 75.04%, sampling error: +0.052%
2014-08-01 21:00:38,576 DEBUG 12.0 kB, realtime: 86.78%, sampling error: +0.069%
2014-08-01 21:00:39,522 DEBUG 18.0 kB, realtime: 89.38%, sampling error: +0.085%
2014-08-01 21:00:40,669 DEBUG 24.0 kB, realtime: 95.72%, sampling error: +0.085%
2014-08-01 21:00:41,510 DEBUG 30.0 kB, realtime: 93.39%, sampling error: +0.083%
2014-08-01 21:00:42,587 DEBUG 36.0 kB, realtime: 95.79%, sampling error: +0.082%
2014-08-01 21:00:43,586 DEBUG 42.0 kB, realtime: 96.37%, sampling error: +0.087%
2014-08-01 21:00:44,547 DEBUG 48.0 kB, realtime: 96.33%, sampling error: +0.064%
2014-08-01 21:00:45,594 DEBUG 54.0 kB, realtime: 97.27%, sampling error: +0.071%
2014-08-01 21:00:46,652 DEBUG 60.0 kB, realtime: 98.12%, sampling error: +0.056%
2014-08-01 21:00:47,600 DEBUG 66.0 kB, realtime: 97.81%, sampling error: +0.055%
2014-08-01 21:00:48,565 DEBUG 72.0 kB, realtime: 97.71%, sampling error: +0.038%
2014-08-01 21:00:49,618 DEBUG 78.0 kB, realtime: 98.29%, sampling error: +0.029%
2014-08-01 21:00:50,568 DEBUG 84.0 kB, realtime: 98.06%, sampling error: +0.019%
2014-08-01 21:00:51,645 DEBUG 90.0 kB, realtime: 98.70%, sampling error: +0.011%
2014-08-01 21:00:52,588 DEBUG 96.0 kB, realtime: 98.42%, sampling error: +0.002%
2014-08-01 21:00:53,600 DEBUG 102.0 kB, realtime: 98.59%, sampling error: -0.004%
2014-08-01 21:00:54,562 DEBUG 108.0 kB, realtime: 98.46%, sampling error: +0.006%
2014-08-01 21:00:55,541 DEBUG 114.0 kB, realtime: 98.43%, sampling error: -0.002%
2014-08-01 21:00:56,582 DEBUG 120.0 kB, realtime: 98.71%, sampling error: -0.006%
2014-08-01 21:00:57,545 DEBUG 126.0 kB, realtime: 98.60%, sampling error: -0.005%
2014-08-01 21:00:58,371 INFO EOF encountered
2014-08-01 21:00:58,371 DEBUG Demodulated 131.070 kB @ 21.531 seconds (98.6% realtime)
2014-08-01 21:00:58,372 INFO Received 125.000 kB @ 21.531 seconds = 5.806 kB/s
- Verify correctness and stop the recording:
~/receiver/amodem $ sha256sum data.recv
008df57d4f3ed6e7a25d25afd57d04fc73140e8df604685bd34fcab58f5ddc01 data.recv
~/receiver/amodem $ killall arecord
See https://www.dropbox.com/sh/2yai1xmntdqlwf1/AACvfzasKEHK0zVxdzI4jF7pa for complete sender and receiver snapshot data.
Make sure that matplotlib package is installed, and run (at the receiver side):
~/receiver/amodem $ PYLAB=1 ./recv.py <rx.int16 >data.recv
The following figures are be plotted:
- Prefix symbol: https://www.dropbox.com/s/lrk2ghastf7rbue/figure_1.png
- Training symbols: https://www.dropbox.com/s/3jnmdjgd3zavcod/figure_2.png
- Actual data symbols (full 64-QAM): https://www.dropbox.com/s/z0ca6tfvf8b8y8k/figure_3.png
