• python (2.7)
• numpy (scipy)
• perl (5.x)

Unless otherwise noted in a file, everyting here is (c) Sec & schneider and licenced under the 2-Clause BSD Licence

Note: The rad1o has to be in hackrf-mode

hackrf_transfer  -r /dev/stdout -f 1627000000 -a 1 -l 40 -g 20 -s 2000000 | python2 multiprocessing-sec.py -c 1627000000 -r 2000000 -f hackrf --jobs 2 | grep "A:OK" | tee outfile

This writes to outfile. Pager messages can be decoded with

python2 iridium-parser.py outfile

record with usrp. So far we've used two settings:

The catch-all interesting stuff

• center frequency: 1626270833
• sample rate: 2000000 (2M)

or to just catch pager channel stuff and smaller files:

• center frequency: 1626440000
• sample rate: 250000 (250k)

The output files are named

<date>-vX.raw

X is the X'th file of the day the v is replaces with an s on the "narrow" receive settings

To process them, there are three stages:

detector-fft.py <rawfilename>

this searches through the file in 5ms steps to scan for activity and copies these parts into snippets called <rawfilename>-<timestamp>.det

cut-and-downmix.py <detectorfile>

this mixes the signal down to 0Hz and cuts the beginning to match the signal exactly. Output is <detfile>-f<frequency>.cut

demod.py

this does manual dqpsk demodulation of the signal and outputs <cutfile>.peaks (for debugging) <cutfile>.data the raw bitstream and on stdout an ascii summary line.

gather all the ascii bits in a single file <rawfilename>.bits

To simplify running these tools, there is doit.pl.

it runs stage1 (if requested) then it runs stage2/3 per output of stage1, up to $ncpu times in paralell then run stage4 (if requested)

run it as doit.pl [-1234] rawfilename

if you give no options, it tries to autodetect what stages haven't yet run