303 309 0004/Investigation/Multiple PSK31 Channels in Buzz

From otp22 db
Jump to: navigation, search

Note: Buzz recording, spectogram, and final analysis of contents are on [1] forum.

Original investigation

As of 2012-09-18 11:00 UTC, a persistent "Buzz" is present on both known "Agent System" dial-in numbers.

After hours of intensive searching, it has been determined that this audio contains multiple PSK31 channels, with strong valid binary phase shifting, but only a continuous string of spaces as data.

The audio in question is already segregated into 40hz channels. Many of these 40hz channels have strong PSK31 signals, many do not.

Although an exhaustive search has not been entirely completed, it appears that there is no additional data, besides the spaces, encoded in any individual PSK31 channel.

The current working theory is that the pattern of valid channels (compared to invalid ones) may yield a clue. For example, the sequence of valid and invalid channels may form a binary string, providing some valuable clue.

The following are examples of BPSK31 channels:

340Hz - Very Strong
420Hz - Strong
780Hz - Strong
820Hz - Strong
1100Hz - Very Strong
1220Hz - Strong
1700Hz - Strong
1860Hz - Moderate
2340Hz - Strong
2620Hz - Moderate
3460Hz - Moderate
3660Hz - Moderate

This list is by no means complete.

I strongly believe there exists a pattern to the presence of these channels. —The preceding unsigned comment was added by Mwally (talkcontribs) . Please sign your posts with ~~~~!

Binary and Phone number

possible lead on this from irc logs thanks to zaseda_

[21:45] <zaseda_> i converted the spacing into a simple binary with 1 for strong
signal and 0 for no signal, starting with 340 hz and got this which decodes to gibberish
[21:46] <zaseda_> 101000000001100000010010000000000010001000000000001000000100000000000000000000 100001
[21:46] <zaseda_> ignore that space oops, i also inverted the digits and still gibberish
[21:47] <zaseda_> but then i counted the actual 0's and got this
[21:47] <zaseda_> 8 1 8 6 2 11 3 11 6 20 4
[21:51] <zaseda_> and that phone number if you add the 207 area code is registered to
a verizon cell phone in ellsworth maine, right down the road from blue hill

the number has yet to be called as the time there right now is late.

Which number? The number described here has too many digits for a phone number. Please clarify -- Crash demons (talk) 07:11, 12 October 2012 (UTC)
If we consider the first '8' a typo (not in the binary), and ignoring 11 and 20, he may be referring to (207) 186-2364 -- Crash demons (talk) 08:51, 12 October 2012 (UTC)
However that number is invalid -- Crash demons (talk) 10:07, 12 October 2012 (UTC)

Analysis of the binary

The binary message is of great interest and should be looked at closely. It is important to note that the string has 84 bits (or 21 nibbles) and is 4 bits too short to be respresented as bytes.

It is also interesting to note that counting the zeroes does result in what zaseda_ produced except for the first digit. This is shown below:

101000000001100000010010000000000010001000000000001000000100000000000000000000 100001
 1 12345678  123456 12 12345678901 123 12345678901 123456 12345678901234567890  1234
 1        8       6  2          11   3          11      6                   20     4

My personal theories are that either zaseda_ miscounted or posted the incorrect bitstring -- Crash demons (talk)

Strike that, found my mistake and corrected the above - however, this does not account for the first '8' that he gave us. -- Crash demons (talk) 08:42, 12 October 2012 (UTC)


Representing the nibbles as decimal values results in the following:

1010 0000 0001 1000 0001 0010 0000
  10    0    1    8    1    2    0

0000 0010 0010 0000 0000 0010 0000
   0    2    2    0    0    2    0

0100 0000 0000 0000 0000 0010 0001
   4    0    0    0    0    2    1

Further analysis 4 February 2013

[10:46] <wrkq> Well, if you still want a closer look: http://i.imgur.com/pSTdpgW.png http://i.imgur.com/aOeDgIs.png http://i.imgur.com/FV8Ss7a.jpg
[10:46] <wrkq> In order - full wav's "lower half", the odd-looking area between the cues zoomed in, and full wav.
[10:48] <wrkq> Anyway, to me it looks like lots of noise, and that one bit at 1000 Hz carrying any data.
[10:51] <Xkeeper> yeah, it looks like the only point of interest is on that freq and the rest is bupkis
[10:51] <wrkq> I scrolled all over it at high zoom. Nothing that seems non-random anywhere.
[10:52] <wrkq> Except that parts at 1000 Hz between the cues, of course.

Spectograms

pSTdpgW.png aOeDgIs.png FV8Ss7a.jpg