Am 2012-03-05 08:43, schrieb David Rowe:
> Hello List,
> 
> To gain some understanding of the problems facing Codec 2 over the HF
> channel I have been coding up an Octave simulation of the FDMDV modem
> that has been used with some success in the past for digital voice over
> HF.   It's a 1400 bit/s modem with 14 carriers, each carrier being DQPSK
> at 50 symbols/s.

With Codec 2 hopefully doing 1000 bit/s in the future that would give
some headspace for channel coding.

> Just starting to test it with an open source Windows/GUI based HF path
> simulator called PathSim.
> The idea is to get my head around the modem/codec requirements for the
> HF channel the tune it for best performance.

I see different requirements for proposed transport methods like
SIP/VoIP, HF, V/UHF. I think Codec 2 should do what it can do best:
source coding of speech. The modem/channel coding should make do the
best to transport the bits from source to destination.

> I have attached a spectrogram of my simulated modem data through a "CCIR
> good" channel at 4dB SNR (equivalent to DQPSK Eb/No=7dB over AWGN
> channel). 
> I can see segments of several seconds where the signal is wiped out.

Deep fades are nasty, they tend to get more the narrower the HF channel
is. You can guard against fades with interleaving. Phil Karn Ka9Q used
16 seconds interleaving for his BPSK1000 scheme which was used at
Arissat-1, but that seems to be a bit exagerated for voice communication ;-)

I guess with the new VQ approach not only lower bitrtes are possible but
also even better quality at higher bitrates. Any plans to have different
quality grades with differnt bitrates? With some metadata about the
content of a transmission  the receiver could accomodate to different
modulation and coding rates. At good conditions (not much fading, good
SNR, low noise) one would use all the badwidth for the raw unprotected
data and get high quality sound. If conditions get worse, you would
switch to higher compression with lower speech quality and put in some
FEC to compensate for the bad channel. You would write the parameters in
the header anyway.

> Also visible are some "clicks" due to clipping of the input signal when
> multipath actually amplified the modem signal.

This is where multiple carriers have a strenth. You can use this
multipath constructive interference if you spread your bits over
frequency and time.
I expect this to bee same for VHF/UHF channels, where you can even use
more bandwidth.

I think it would be nice to have a modulation scheme that could be
parametrized: Bandwidth, datarate, code rate, modulation (BPSK, QPSK
etc.), SNR requirements (which is a result of the other parameters). The
same encoder/decoder could be used for HF, VHF, UHF or even higher
frequencies. For every band you would have say 3 recomended parameter sets.

I know this is ambitious, but we see that digital transmission in in
that profile can be handled for several years now, and Codec 2 brings
new opportunities and requirements.

Regards

Patrick

------------------------------------------------------------------------------
Try before you buy = See our experts in action!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-dev2
_______________________________________________
Freetel-codec2 mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/freetel-codec2

Reply via email to