diff --git a/ISS.md b/ISS.md
new file mode 100644
index 0000000..69d1dba
--- /dev/null
+++ b/ISS.md
@@ -0,0 +1,32 @@
+![Raspberry NOAA](header_1600.png)
+
+This project now includes full reception and decoding of SSTV transmissions from the ISS.
+
+# ISS reception and scheduling
+Reception is quite easy: The ISS TX works on 145.800 Mhz, Narrow FM. So we just need to tune our RTLSDR receiver and that's all.
+
+Scheduling is governed by the `SCHEDULE_ISS` inside `.noaa.conf`. Setting it to `true` will schedule future ISS passes.
+
+# ISS decoding
+This is the difficult part: Decoding PD120 needs custom software. Right now the best Linux software for SSTV (in mu humble opinion) is [QSSTV](http://users.telenet.be/on4qz/qsstv/index.html). QSSTV is great, works with a lot of modes, has auto slant and signal detection. On the other hand, QSSTV only works in GUI mode as it's written in QT (hence Q-SSTV) and that's not situable for this project as we don't run any Window Server.
+
+Surfing the Internet I've found Martin Bernardi's (and team) [final work about a PD120 modulator - demodulator](https://github.com/martinber/rtlsdr_sstv) so I put hands on slightly modifying it to adjust the project to my needs.
+
+## Decoding behavior
+There are a few things to take care about
+
+1. A single ISS pass during a SSTV event could have zero to three SSTV transmissions (based on an average pass time of 10 minutes) so we need to find a way to detect the transmission header to fire up the decoder and store each image.
+
+![PD 120 header](pd120_header.png)
+
+2. A simple way to detect the header on an audio file (once digitized with scipy.io) is to sample the header as an absolute numeric representation, apply a threshold during certain amount of time. So if the header is present, mark the stream and do the same for each new header that ocurrs in about `this_timestamp + 120 seconds` so we don't get any false positives in the middle of the PD120 carrier.
+
+3. A Satellite transmission to the earth has doppler efect so the frequency changes over the time. This is a problem when you plot audio as data. So I had to include a high pass filter to remove the DC offset from the audio recorded by `rtl_fm`.
+
+# Upgrading procedure
+There's a migration script that handles the required modifications over the database as well as the web files. Take in mind that you need to have the web panel version of raspberry-noaa running.
+
+1. `cd /home/pi/raspberry-noaa`
+2. `git fetch -pt && git pull`
+3. `cd migrations/`
+4. `./20201292-iss.sh`
diff --git a/pd120_header.png b/pd120_header.png
new file mode 100644
index 0000000..dfb71b3
Binary files /dev/null and b/pd120_header.png differ