[DSP-10] More EME2 post processing, possible result.

[Courtney Duncan]

In the process of cleaning up my EME2 post processing program so I 
could call it Version 1.0 and put it away for a while, I added some 
things to the summary report and played around some with the 
settings.  Here is another output:

DSP-10_EME2_Post has exited with status 0.
[Session started at 2006-06-06 23:09:27 -0700.]

Parsing file 5_21_set.dat
Parsing file 5_26_set.dat
Parsing file 5_26_ris.dat
Parsing file 5_27_set.dat
Parsing file 5_27_ris.dat
Parsing file 5_28_ris.dat
Parsing file 5_29_set.dat
Parsing file 5_30_ris.dat
Parsing file 6_01_set.dat
Parsing file 6_02_set.dat
Parsing file 6_03_set.dat
Points processed:          564
Points skipped:            0
Points skipped for noise:  10832
Biggest noise value:       1999.99

values fixed:      5880 280
values not fixed:  5964 284

Noise Bins:  9.39332 0.143633

bin  mean  sigma   -noiseMean  sigma
noise 1829.36 150.25
-10 9.50564 4.44338 0.112324 0.782024
-9 9.55609 4.55871 0.162767 1.13321
-8 9.43886 4.62095 0.0455424 0.317076
-7 9.54192 4.22536 0.148604 1.03461
-6 9.32072 4.20358 -0.0726018 -0.505468
-5 9.46358 4.42613 0.0702649 0.489198
-4 9.41645 4.62733 0.02313 0.161035
-3 9.2668 4.3803 -0.126517 -0.88084
-2 9.14984 4.38219 -0.243478 -1.69514
-1 9.6926 4.5962 0.299284 2.08367
0 9.89637 4.69772 0.503053 3.50236
1 9.31409 4.24443 -0.0792265 -0.551591
2 9.24561 4.10516 -0.147706 -1.02836
3 9.40869 4.25121 0.0153656 0.106978
4 9.35783 4.37424 -0.0354878 -0.247073
5 9.2257 4.08965 -0.167622 -1.16702
6 9.316 4.62622 -0.0773189 -0.538309
7 9.47706 4.33566 0.0837359 0.582986
8 9.65168 4.4999 0.258365 1.79879
9 9.26132 4.0263 -0.131995 -0.918979
10 9.25589 4.51932 -0.137428 -0.956803



DSP-10_EME2_Post has exited with status 0.

For values "fixed" and "not fixed" are from a routine I wrote to 
reverse engineer the bin calculation, test the most significant bit 
of the least significant byte, correct the calculation, and reform 
the observable.  Since I have about 14 bits of action here, that bit 
is in the noise and about half of the values get "corrected."  The 
first column is values corrected or not, the second is equivalent 
rows.

The signature is still there without the "fix" but it is stronger 
with it, which you'd expect if the signal was real and the fix was 
real.

"Noise Bins" is the mean and standard deviation of the 20 bins (not 
counting bin 0).

The first line below the little header is the mean and standard 
deviation of 'pave.'

The five columns, per bin, are:
- bin number, 2.3 Hz +/- from expected signal in bin 0.
- the mean value reported in that bin, from points used.
- the standard deviation in that bin

Then, comparing with the "Noise Bins" values,

- the difference between the level in this bin and the mean of all bins
- the number of standard deviations (of all bins) that difference represents.

What we see here is that when I set the maximum noise level (which I 
still don't fully understand mathematically) to 2000, it throws out 
95% of my points collected over two weeks, but makes a unique peak on 
bin 0 where the signal is supposed to be.  The 3.5 on bin 0 is the 
biggest number and, being positive, might be signal.

I'm a little skeptical that this is a detection since this behavior 
is pretty sensitive to that maximum noise level setting (if I go much 
lower I have nearly no data points left and if I go much higher all 
the noise bins come up), but it is encouraging enough for me to go 
back and look at AOS and LOS times, path losses per file, and such to 
see if other reasonable constraints lead to results like this one.

I'm saying "reasonable" because it's "reasonable" that if you average 
only the "quietest" points, any actual signal should be more visible. 
Similarly, I ought to be able to exclude data when the moon was out 
of beamwidth or blocked and see similar improvement, if this really 
is something.

I realize I'm on a thin edge between properly editing the data and 
just fooling around until I get lucky here.  But a little more 
fooling around seems reasonable.

- n5bf/6