Re: [Discuss-gnuradio] Decoding 2FSK Compensating for carrier jitter/skewing (CFO)

[Andy Walls]

From:   Cinaed Simson
Date:   Sun, 9 Jul 2017 23:04:26 -0700
________________________________________________________________________
> On 07/09/2017 12:15 PM, Andy Walls wrote:
> > On Sat, 2017-07-08 at 21:38 -0400, Andy Walls wrote:
> >>> Date: Fri, 7 Jul 2017 19:50:55 +0300
> >>> From: HLL
> >>> Hi all,
> >>>
> >>> I'm relatively new to DSP and gnuradio but I tried tons of stuff
> >>> and
> >>> I couldn't decode a fairly simple FSK data.
> >>> baudrate seems around 600-700 bps and fsk deviation is less then
> >>> 3k.
> >>
> >>
> >> Hmmm.  I took a look at your signal and tried building a coherent 2-
> >> FSK 
> >> demodulator.  Under the assumption that it was straight 2-FSK, the
> >> signaling tones looked to be at +/- 1200 Hz when properly centered.
> 
> Just to be clear - doesn't 2FSK have 4 tones? I'm not sure what you mean
> by signaling tones.

"2 Frequency Shift Keying" uses 2 frequencies (aka tones) to send symbols.
When I write "signaling tones" for a 2-FSK, I mean the "mark" frequency
and the "space" frequency tones.  (Assign symbol values of -1 and 1 to
mark and space as appropriate for the encoding scheme in use.)


> I just looked at it in inspectrum and there are 4 visually identical
> structures - and I would guess the fact there are 4 structures is just a
> coincidence.
> 
> They're roughly 1.45 seconds long with blank spacers varying from 116 ms
> to 200 ms.

Those are the actual RF bursts for each packet.  There are 4 packets in
the file.


> I must be doing something wrong.

The modulation scheme for this unit appears to be Audio Frequency Shift
Keying: 2-FSK performed with 2 (real) square wave audio tones (at 350 Hz
and 940 Hz fundamental frequencies) that are then FM modulated.
The baud rate appears to be 233 symbols/second.  It does not appear to
me that the bits are Manchester encoded.


Using square waves for the FSK tones is a unusual, unless a cheap
digital microcontroller is generating the tones on a digital output pin.

Look at the blue signal trace in the first image in this mailing list
post, and ignore the level wobble the author is focusing on:
http://lists.gnu.org/archive/html/discuss-gnuradio/2017-07/msg00071.html

The blue line is not showing bits directly.  It is showing intervals of
square waves at two different frequencies: 350 Hz and 940 Hz.  The
frequencies are what are encoding the symbols, not the amplitudes.  This
is 2-FSK with the tones generated by a really cheap square wave
generator.


Given the FCC documents cited, I doubt this signal was generated by the
devices in the FCC documents.  Probably some other device in the same
ISM band generated this signal.

-Andy

> -- Cinaed