Strange TUBIN Mobitex-NX decodes

[janvgils]

Good evening,

I tried to decode the signals from TUBIN on both gr-satellites master and maint-3.8. Master isn't working because the Mobitex-NX OOT module source isn't migrated to GNURadio 3.9.

When I try to decode the signals with a maint-3.8 installation and using the following Yaml file I also ran in to issues.

name: TUBIN
norad: 99678
data:
  &tlm Telemetry:
    unknown
transmitters:
  1k2 FSK downlink:
    frequency: 435.950e+6
    modulation: FSK
    baudrate: 1200
    framing: Mobitex-NX
    data:
    - *tlm
  2k4 FSK downlink:
    frequency: 435.950e+6
    modulation: FSK
    baudrate: 2400
    framing: Mobitex-NX
    data:
    - *tlm
  4k8 FSK downlink:
    frequency: 435.950e+6
    modulation: FSK
    baudrate: 4800
    framing: Mobitex-NX
    data:
    - *tlm
  9k6 FSK downlink:
    frequency: 435.950e+6
    modulation: FSK
    baudrate: 4800
    framing: Mobitex-NX
    data:
    - *tlm

Decode example:

# Decoding Satellite 99678.
* MESSAGE DEBUG PRINT PDU VERBOSE *
((transmitter . 4k8 FSK downlink) . 5)
pdu_length = 0
contents = 
***********************************
Error while submitting telemetry: HTTP Error 400: Bad Request
* MESSAGE DEBUG PRINT PDU VERBOSE *
((transmitter . 9k6 FSK downlink) . 5)
pdu_length = 0
contents = 
***********************************
Error while submitting telemetry: HTTP Error 400: Bad Request
* MESSAGE DEBUG PRINT PDU VERBOSE *
((transmitter . 9k6 FSK downlink) . 5)
pdu_length = 0
contents =

***********************************
Error while submitting telemetry: HTTP Error 400: Bad Request
* MESSAGE DEBUG PRINT PDU VERBOSE *
((transmitter . 1k2 FSK downlink) . 1)
pdu_length = 520
contents = 
0000: 3b f1 58 64 2e 0c 3f f7 f6 80 39 c8 b0 d6 d5 f3 
0010: 84 93 91 07 20 b9 fd 68 32 16 ec 6a b8 86 2b 1b 
0020: b1 9f 17 43 10 e2 56 6e 2c be d7 70 af 11 9c e6 
0030: 02 c5 28 d7 18 bc 6a d9 57 97 32 c7 c4 ce 61 37 
0040: 86 12 9f 2b 20 73 07 78 51 b8 68 14 50 58 85 9b 
0050: c4 90 9a 65 b7 79 75 d4 85 58 4d ec 49 ee 99 e2 
0060: fe d0 75 d8 1c 6a 69 bc 4b 64 76 88 7a 8f 0b 38 
0070: a9 66 34 3d 42 33 8b c7 6c 51 27 1e 39 b0 53 21 
0080: bb 0a 04 5f 67 a8 2e 6d cb 98 82 28 eb b2 94 df 
0090: 77 02 2b 50 b3 b4 dc 34 be 07 f3 5d 4e 55 25 2f 
00a0: aa 61 ac 01 3c 60 2d c3 64 12 e4 c6 e6 ba f2 d4 
00b0: 7e 93 7b 17 01 2a 7e 4e 94 c6 4e 8f 0a f7 ed 08 
00c0: c4 dc cf 0c ef 2c 3c 38 3d 5a 3d f3 d2 5e 28 f0 
00d0: a9 29 09 39 12 75 70 06 0c 63 a1 d8 f4 15 da 96 
00e0: 29 96 f5 7d f6 77 05 33 93 05 e7 0e 2d 72 7a d4 
00f0: de e5 82 ab 6d 95 b9 d8 30 11 9a 1f c0 39 b5 dc 
0100: 52 ac 71 ab b3 9a 29 8a 94 6c 69 4e 14 6b a8 0b 
0110: 4c 4e d5 46 f3 11 15 73 0b be 4a b4 05 60 3a 39 
0120: 24 c3 21 6b 2f 8c 4c 15 0a 25 13 3b 99 07 f8 71 
0130: a5 01 84 a8 14 80 a8 6c 15 be 33 cc cb a2 8b 02 
0140: f8 5a 8c dd 3a bd 8b 42 18 4c 60 f3 13 29 d7 a1 
0150: da 5b b3 ae 92 e2 f2 41 89 79 e3 46 1c c8 8f a4 
0160: d4 e7 ae 39 4c 49 8f 62 1c c8 ea 44 45 4a df 9d 
0170: c2 59 44 74 1a 43 39 e0 2d f8 3b b3 60 66 fd 3b 
0180: 4b 18 62 88 2c a2 96 96 45 bc b5 ad 82 fb f0 4b 
0190: da 6b ed 6f 3d 38 2f 94 28 e0 69 ea 9e 9b 98 3a 
01a0: 90 a2 a6 d8 66 8a 65 71 db da a3 e5 4a 2f 79 91 
01b0: d4 fa 6e f9 5a bd af 26 b6 c6 65 75 6b 9b 47 62 
01c0: 67 9f 98 bc f0 01 3d fb 2c 9b 54 0e f0 5b 78 dd 
01d0: ea 0e 51 a2 27 fd 70 75 56 83 87 1e 34 42 0f f5 
01e0: 7e 14 45 a9 07 c5 29 68 cb 45 6f 9f ff 30 c2 57 
01f0: 43 38 26 7e 1c 43 cc 8a 03 82 de f3 07 4e a8 87 
0200: 3b 2a aa ff ff ff 0f bb

Maybe if time allows it, you can have a look.

Observation used: https://network.satnogs.org/observations/4351767/

[daniestevez]

Hi Jan,

These don't seem good decodes. The errorcode aa ff ff ff 0f bb indicates that most of the FEC blocks couldn't be decoded. Probably the framing is not correct. gr-satellites supports the Mobitex and Mobitex-NX variants that are implemented in gr-tnc_nx. It might happen that they've changed something with TUBIN, so I'll try to ask some folks and take a look at the documentation.

[janvgils]

Thanks Daniel.

[daniestevez]

Hi Jan, I've been taking a more careful look at this, and I've found out that the quality of most of the SatNOGS recordings of TUBIN is not so good. Also, using the --disable_dc_block parameter with gr_satellites seems to help.

I found this observation, which is a bit better than others, and produces a valid decode by running

gr_satellites TECHNOSAT --wavfile satnogs_4351795_2021-07-01T14-25-30.wav --disable_dc_block

The decode looks like this:

* MESSAGE DEBUG PRINT PDU VERBOSE *
((transmitter . 4k8 FSK downlink) . 1)
pdu_length = 592
contents = 
0000: 3f 0e 44 50 30 54 42 4e 15 c0 05 11 01 0f 00 1c 
0010: 03 28 70 8e 5a 05 15 c4 0c 00 00 00 54 00 5f 28 
0020: 70 8d cd 05 11 11 03 00 00 00 02 ff f9 28 70 7b 
0030: 8b 1e 00 e5 de 00 05 03 28 70 8e 5a 02 28 44 00 
0040: 6e 59 5c c2 60 00 11 00 28 70 8e 5a 06 3c 44 17 
0050: 00 0a 3c e1 31 f6 a4 13 99 fa fe 00 29 02 36 f7 
0060: d8 00 41 00 28 70 8e 5a 06 83 00 03 ff a0 ff f3 
0070: f6 e8 ff fc 57 c1 00 07 a7 ea fa aa fe 32 ff 8d 
0080: ff fe ff fd 00 01 00 80 02 b8 fe 76 00 b0 fc 46 
0090: 00 ee 02 0e fe 59 fe 79 ff 54 03 87 01 87 f3 40 
00a0: 0a 31 15 77 ff fd 00 04 ff fc 00 58 f2 00 03 03 
00b0: 28 70 8e 5a 0a f9 00 00 03 06 d3 00 0e 00 28 70 
00c0: 8e 59 1e 5a 44 01 7f ff 7f ff 7f ff ff ff ff ff 
00d0: 00 00 fb 83 00 10 05 28 70 8e 5a 00 10 46 00 01 
00e0: 10 0d 00 00 0a 00 04 14 40 00 08 a6 a7 fe 53 00 
00f0: 41 15 28 67 c4 3f 06 83 00 07 2a ad 00 08 f2 e1 
0100: 00 06 39 c2 00 07 e9 f7 01 17 00 7d 04 d2 ff f2 
0110: 00 04 00 02 02 19 fe 85 fe 5f 03 bf 00 11 fe e8 
0120: fe 9d 02 52 fe 97 ff c3 03 93 01 8d fa e8 10 96 
0130: eb 54 00 03 ff fc ff fc 00 a2 3e 00 2a 12 28 67 
0140: c4 40 00 a9 00 41 00 20 00 24 4e 00 00 00 00 34 
0150: 00 00 0f 0e 08 09 0a 0c 1d 1a 3e 1c 18 00 00 00 
0160: 00 00 00 0a 16 18 14 10 0d 0e 10 13 3b e5 a8 96 
0170: 00 41 15 28 67 c4 7b 06 83 00 01 d9 95 ff f7 78 
0180: 8b ff ff 10 37 ff f3 84 c5 00 c6 00 a7 04 74 00 
0190: 05 ff fe 00 03 ff 11 fd 22 ff 39 00 37 fc 1f 00 
01a0: 63 fe aa 02 42 fe 39 ff c3 03 93 01 8d fb a7 0f 
01b0: 62 ea 41 00 03 ff fb ff fc 00 57 fe 00 41 15 28 
01c0: 67 c4 b7 06 83 00 06 b1 9f ff f8 b8 10 00 02 19 
01d0: d9 ff f4 92 78 00 52 00 da 04 87 00 00 00 01 00 
01e0: 05 fd 4f ff be fe 18 fe 89 fc cd 01 b1 fe bc 02 
01f0: 25 fd de ff c3 03 93 01 8d fc 6c 0e 1c e9 46 00 
0200: 03 ff fb ff fd 00 ab c8 00 2a 12 28 67 c4 b8 00 
0210: a9 00 41 00 1f 00 22 00 00 26 0c 00 00 00 00 0f 
0220: 09 07 09 0c 0f 22 1e 3e 9b 00 00 00 10 21 13 00 
0230: 00 17 15 10 0d 0c 0f 13 15 3d b6 99 10 ff ff ff 
0240: ff ff ff ff ff ff ff ff ff ff aa 00 00 00 00 bb 
***********************************

Note that between the aa and bb at the end of the frame we have zeros. This indicates that all the FEC codewords were successfully decoded. Also the callsign DP0TBN appears somewhere in the beginning of the frame.

The decoder also produces some empty frames marked with the flag 5 by the gr-tnc_nx decoder (I don't remember what this means):

* MESSAGE DEBUG PRINT PDU VERBOSE *
((transmitter . 4k8 FSK downlink) . 5)
pdu_length = 0
contents = 
***********************************

So apparently TUBIN uses exactly the same protocol as TECHNOSAT, which gr-satellites calls Mobitex-NX. This matches the online documentation and the reports of people on Twitter using UZ7HO's soundmodem for TECHNOSAT to decode TUBIN.

I think this solves the matter (besides #238). I'll now add a SatYAML file for TUBIN. For the time being I'll only include the 4k8 mode unless we have an indication that other baudrates are used.

[janvgils]

Thanks for taking the time to investigate this, I will use the suggestion and see if I can get better decodes.