[pskmail] Re: Need PSKMail for Disaster Relief Operations

  • From: "karel Fassotte" <karel.fassotte@xxxxxxxxx>
  • To: pskmail@xxxxxxxxxxxxx
  • Date: Fri, 4 May 2007 00:16:12 -0500

Thanks for the information on the compression Per.
Does this mean that the speed is 3 time faster using the paq comression?

Per we got the RFSM2400 working on a Gentoo platform running Wine,
using only 1 soundcard !

WDYT
wrote:
We really need to focus on improving coverage and
efficiency and reliability using likely modes rather
than getting too distracted by the bells and whistles
of modes that are too rare and unreliable to be
useful in a genuine disaster.

Yes I agree we should improve coverage, reliability and efficiency,
but I dont agree with the statement of modes that are to rare and
unreliable etc.

I dont think milstd are rare and are unreliable. It will more be a
question how to integrate the waveforms in the same model as the
PSKmail. If higher bitrates are possible they should be used. The
system should have a fallback mecanism to slower modes if propagation
requires this.

73, Karel, hc1akp


2007/5/3, Per Crusefalk <per@xxxxxxxxxxxx>:
Hi,

I have seen 706 rigs with wider filters used here, they seem to work
fine. Also there are Skanti rigs in use that have too narrow filters
(I believe 2.8 kHz) and they seem to work pretty good anyway (displayed
SNR on Harris RF-5710 is slightly worse). But, many have rigs with
narrow filters and it would be good to handle that. Experiments are of
course always necessary (and fun).

The compression we have been testing is the one already existing in
pskmail. If you click on the download button, instead of read, to read
an email then that email will be compressed during transport and the
client will decompress it when its completely downloaded.
First there is sort of a context based huffman where it looks for words
and compresses, then its compressed agn using gzip and last but not
least made ready for transport as base64 (the process is reversed in the
client). What we have tested now is paq7 and paq8l instead of gzip and I
have added swedish words to the word_table file in .cbh and tested that
(a lot of tests...).

The paq compression is much better than gzip and we save a lot by using
it. More on that here: http://cs.fit.edu/~mmahoney/compression/

Rein tested "daily reports" examples that Walt sent us and without
compression it took 9 minutes 38 seconds to transfer (real on air
tests). With the new compression it took 3 minutes and 28 seconds,
pretty good.

73 de Per, sm0rwo





Other related posts: