I left it running overnight, and have tried a few capture files
selected at random. Some produce a garbage message, with symbol
errors typically 200 or so out of 512 but this one at 0200 gave a
correct decode immediately with just slightly fewer symbol errors at
187 All the contents of the .txt file shown below
(unless the one chance in 50 ^ 5 = 315 million threw the correct
callsign up at random :-)
One thing that puzzles me:
The EbNaut documentation states there is an outer checksum for message
validity. Why does this allow the garbage decodes to get through -
surely that must be failing the checksum. Or are they the few out
of millions of possible garbage decodes that pass the checksum
criteria?
Andy G4JNT
ebnaut-rx V0.3b
input file C:\CohLFRx\151130_0159.WAV
sample rate 31.250000 per second
rx frequency 0.000
file start time 2015-11-30 01:59:59.000
polynomial 8K19A
number of chars 5
block size 46
symbol period 2.000
number of symbols 512
list length 6000
start offset 1.000
freq offset 0.000000
cores 1
initial reference phase 7.0
---------------------------------------------
found VO1NA
list rank 5354
reference phase 180 180 180 180
carrier RMS 3.401e+002 noise RMS 5.160e+001
carrier S/N 16.38 dB in 976.6 uHz
carrier Es/N0 -10.71 dB
carrier Eb/N0 1.83 dB
info bit period 35.93 seconds
symbol error rate 187/512 = 36.523 %
Es/N0 from symbol errors -12.2 dB
Eb/N0 from symbol errors 0.1 dB
Shannon capacity 211.5 bits/hour
Shannon efficiency 47.4 %
symbols file C:\CohLFRx\151130_0159-symbols.csv
elapsed 15 seconds
---------------------------------------------
found VO1NA
list rank 29
reference phase -150 -150 -150 -150
carrier RMS 3.401e+002 noise RMS 5.160e+001
carrier S/N 16.38 dB in 976.6 uHz
carrier Es/N0 -10.71 dB
carrier Eb/N0 1.83 dB
info bit period 35.93 seconds
symbol error rate 175/512 = 34.180 %
Es/N0 from symbol errors -10.8 dB
Eb/N0 from symbol errors 1.5 dB
Shannon capacity 211.5 bits/hour
Shannon efficiency 47.4 %
symbols file C:\CohLFRx\151130_0159-symbols.csv
elapsed 29 seconds
On 29 November 2015 at 23:30, Paul Nicholson <[email protected]> wrote:
>
> VO1NA 2015-11-29 8K19A 2S 5C
>
> 20:00 6.2 dB
> 20:30 7.0 dB
> 21:00 3.2 dB
> 21:30 1.4 dB
> 22:00 5.4 dB
> 22:30 no decode
> 23:00 8.7 dB
>
> Edgar wrote:
>
>> I had been looking at vtmatch.
>
> vtmatch monitors a stream and does a continuous
> cross correlation against the template: in other
> words a matched filter.
>
> It would be the one to use if you knew what the
> message was but not the start times.
>
> Generate a template file for vtmatch with
>
> echo 'TNX PAUL' |
> ebnaut -e -r100 -S2 -p8K19A -N8 |
> awk '{print $1}' > ref.dat
>
> Here, I've dropped to sample rate 100.
>
> I should then be able to do something like
>
> vtread -T2015-11-28_20:00,+8h /raw | # Get 8 hours of signal
> vtfilter -h bp,f=2777,w=2 |
> vtmult -f 2777 | # Mix 137777 to baseband
> vtresample -r 100 | # Baseband I/Q at 100 samples/sec
> vtmatch -t ref.dat > match.vt
>
> Then
>
> vtplot match.vt
>
> should show a good spike whenever the message appears in that 8
> hour stream. The spike should be surrounded by auto-correlation
> 'sidebands' which ought to be quite low compared to the peak.
> For some reason it's not quite working, getting the sidebands
> but no central spike, even on a simulated signal. Maybe I'm
> too tired or there's a bug in vtmatch. I'll look at it again
> tomorrow.
>
> --
> Paul Nicholson
> --
>
|
