Discussion Forums

baudline analysis of Kepler-4b redux

8 replies [Last post]
sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733

I did a detailed signal analysis of the new Kepler-4 re-observation data that was uploaded a couple days ago and posted it to my blog.

http://baudline.blogspot.com/2010/09/setiquest-kepler-4b-redux.html

The original Kepler-4 analysis discovered an interesting FSK modulated signal about -480 kHz left of Hydrogen. The baudline analysis of these new Kelper-4 data sets discovered an odd drifting zigzag modulated signal that is -483 kHz left of Hydrogen. Different modulation schemes, different baud rates, same location.  A third observation of Kepler-4 should be scheduled but I end the conclusion section with a skeptical "Hmmmm."

gerryharp
Offline
Joined: 2010-05-15
Posts: 365
nice analysis

Hi sigBlips

This is very good work. Congratulaions!

Your identfication of a drifting signal in the data is very interesting. We'll go back and think about why we decided this was a red herring. Snice there are two mirror-image copies of the drifing signal, doesn't that suggest an artifact? It may well be related to the strong tone you observe that  is visible in the wide-band spectrum.

In any event, keep in mind that these data (from May) are corrupted by some digital processing problems which have meanwhile been fixed. To remove ambiguity, last Friday we made more observations on Keplper04 at 1420 and 1619 MHz. They are not uploaded yet (this process takes days with our slow bandwidith out of Hat Creek). When they are available, I'll post them. These datasets are over longer time periods too.

Gerry

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
> This is very good work.

> This is very good work. Congratulaions!

Hi Gerry,  Thanks.
 

> Snice there are two mirror-image copies of the drifing signal, doesn't that suggest an artifact?

I wouldn't call them mirror-image copies because they are two very different signals that are in mirror-positions (sidebands) around Hydrogen and they are drifting at different rates. These sidebands are present in the kepler04-3 data set but not in the kepler04-4 data set. The zigzag modulated signal is present in both data sets.

Yes the mirroring does seem artifact-like but the kepler04-4 data set does not have the mirroring and most of the other setiQuest signals I've found have had some sort of mirrored-sidebands.  So is the signal causing the mirroring or is it caused by it? I don't know but it doesn't seem to happen all the time.
 

> It may well be related to the strong tone you observe that  is visible in the wide-band spectrum.

The non-drifting strong tone at the suspicious frequency is only in the kepler04-3 data set.  It could be causing the mixing. Do you have any ideas what its source could be?  Internal or external?  I'm going to guess that it is internal.

Did you change the tuner frequency between the kepler04-3 and kepler04-4 data sets or did you just extract a different section from the 104 Mbps stream?

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
Discussion about the zigzag

Discussion about the zigzag modulated signal regarding candidate verification / confirmation procedures and matching modulation markers has branched here:

http://setiquest.org/forum/topic/discussion-roadmap-wiki#comment-1207

Those topics are probably best replied to over there and everything else Kepler-4b redux related over here.

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
The zigzag modulated signal

The zigzag modulated signal has a baud rate of 0.2221 symbols/second. Assuming that it is a binary coding, here is the demodulated bit stream:

10101011000001010100000001101100000110(1000010101101010

The parenthesis represents a possible demodulation error but it is likely a "1" bit.  Another way of looking at this strange frequency wiggle is as a delimiter. Since the signal has two of these delimiting shapes the demodulated bit stream becomes:

1010101 1000001010100000001101100000110 1000010101101010

What is interesting to me are the long stretches of 0's, the repeating 01 sequence, and a couple instances of 11. This doesn't look random to me and it doesn't look like it was created by a simple cyclic-type of machinery. The structure here seems more complicated than what we saw in the original Kepler exo4 signal analysis even though that signal had twice the number of bits. Again, it's most unfortunate that the data set ended when it did because it would of been great if we had ten times the number of bits to decode.

Any thoughts?  Does anyone else see any other patterns?

jpsd0510
Offline
Joined: 2010-04-28
Posts: 2
Question about very weak signal on 8 bits resolution data

I have followed discussions about FSK modulated signal on Kepler-4 data set.  I understand that signal is very weak and the data set is 8 bits resolution.  Is it possible that the FSK pattern observed come from insuffisant resolution of data set in regard of amplitude of signal detected?   It will be interesting to obtain, if it is possible, same data set but with 16 bits resolution for comparison.

Excuse my english so i am a french language poeple.

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
That is a very good question.

That is a very good question. There are a couple reasons why I don't think the FSK signal is caused by the conversion to 8-bit samples:

* Measuring the distortion that's caused by a 16 to 8-bit sample conversion is very easy to do. I'm sure the ATA engineers measured and tested this thoroughly.

* Bad sample bit depth conversion usually creates harmonically related artifacts. I've seen many distortion harmonics in the setiQuest data but not with the FSK signals.

* I've never seen bad bit depth conversions create a modulated signal but that doesn't mean it's impossible! Normally some sort of mixing operation would be required.

That said, this is all speculation and I'm not sure how the bit depth conversion is actually being performed so I could be wrong with my assumptions. I don't think this code is scheduled to be open sourced either.

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
Here is a great video of my

Here is a great video of my Astrophysics teacher Geoff Marcy giving a talk about exoplanets.  Jump to 1:07 near the end for a discussion about Kepler-4.

http://www.youtube.com/watch?v=bFlrfTvmMwk

Choice quote:

"Now here's my favorite of the Kepler 5 planets, I'm going to spend a couple minutes on this one, cause this is the best. This star Kepler-4, yes it dims periodically, check it out, dim dim dim ..." (:

Some Kepler-4 stats:

  • 4.1 times the radius of the Earth.
  • 20 Earth masses
  • 1.9 gm/cc density (water = 1 gm/cc, Earth = 5 gm/cc)
  • it's almost certain that Kepler-4 has water
  • 3.2 days to go around it's star
  • blow torched surface 1500°C

Note the SETI slide of a picture of the ATA at the end of the talk.

Thanks Geoff, you rock!

sigblips
sigblips's picture
Offline
Joined: 2010-04-20
Posts: 733
I have finished my analysis

I have finished my analysis of the 5-14-2010 Kepler-4 data sets and rewrote the conclusion. There are now 21 target signals. This is an incredible number of signals and many have periodic patterns which suggest some form of modulation. Unfortunately I don't have the time to give each one an in-depth investigation.  I also added an #anchor tag for the most interesting zigzag modulated signal at +2044295 Hz:

http://baudline.blogspot.com/2010/09/setiquest-kepler-4b-redux.html#+204...

There are still many open questions.