Hello everyone,

My Name is Kevin Karhan, I'm currently enlisted as Student for
IT/Information Science [B.Sc.] at the University of Wuppertal, NRW, Germany.


I would like to share some troughts I'm having during the process of my
own project, which is intended to become topic of my bachelor thesis
since they seemed to be useful.



As maintainer of the Cryptofon - Project, I've taken a look on the needs
of an Voiceband Modem (which in difference to FDMDV does not use
multiple 50 baud - links, but instead a single 300 baud link with ~48 Hz
difference)

Using a 64-FSK (enabling an 8 bit : 1 baud ratio, the full 2400 bit/s
for Codec2 are available - so you can either use 2 (syncronized) modems
at 300 baud, each using 150 baud in a Time Division Duplex.

Regarding a 2400 bit/s transmission, buffered in frames, either useage
standard 20ms frames, having 48 bit per frame (mean: 6 baud per frame)
at 50 frames/s, or half the frame rate at 96 bit per frame (mean: 12
baud per frame) seems useful.

In the case of the Cryptofon, I use a total bandwith of 2400 bit/s -
including FEC 3/4 - meaning 1800 bit/s of data. Using 300 bit/s for Sync
and other 300 bit/s for additonal Short-burst Data, the system should
self-syncronize.

In detail, I intent to create frame packets alike these:

bits  1 - 48: (48 bits) Codec2 Frame (1200 bit/s - 40ms frame)
bits 49 - 60: (12 bits) Short Burst Data
bits 61 - 72: (12 bits) Sync I/0 (could be likewise the
Mancester-Sequence of the FDMDV-Modem)
bits 73 - 96: (24 bits) FEC 3/4 (additional Forward Error correction)

When regarding this 96 bit frame as buffered before transmission, it
seems useful to have a similar balance at the "subframes" for the baud
conversion in order to cover lost or incomplete bursts.
By picking each 12th bit (could be easily pipelined with a LFSR!),
starting at bit 1, you get 12 well-balanced "subframes", each consisting of:

4 bit Codec2
1 bit SBD
1 bit Sync
2 bit FEC 3/4

It seems obvious that creating the FEC 3/4 at this point would make
sense instead of the frame itself, but I just calulated them in the
frames for reminder...
Also instead of FEC 3/4, a similar FEC 6/8 could be used.



Regarding the primary use of Codec2 for shortwave narrowband / SSB
amateur radio, the FDMDV modem is still the best choice.
(Since multipath problems do not encounter on PSTNs, this problem does
not appear for the Cryptofon in the first place).



Anyway, I may cross-examine my troughts on this by reading the
documentation about the FDMDV modem with much more attention than before.


If the FDMDV modem would offer a total troughput of 2400 bit/s per
direction and work on PSTN/POTS, too - which I am not shure about since
it was not designed for it, (in theory, this would generate frequency
overlays making it hard or even impossible to decode the symbols) I
could directly integrate it.


But ATM I'm focussing of keeping baud rate limited to 300 baud (instead
of 700 baud) to ensure maximum compactbility even in suboptimal
constellations (e.g. reversible non-damaging use on a public phone w/
graphite mic/speaker via self-made acoustic coupler).
All in all: The Cryptofon has a very bad SNR envoirment it needs to fit
in...




I'm keeping an eye on the mailinglist...



Greetings,
Kevin Karhan

Attachment: 0x3062688F.asc
Description: application/pgp-keys

Attachment: signature.asc
Description: OpenPGP digital signature

------------------------------------------------------------------------------
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Reply via email to