As requested.

?

I've been running RTLSDR_Airband from the day I found out about it. I was trying to write code for a multi-channel FM receiver and it was proving quite difficult, and I dropped the effort once I found out about it. It uses a very efficient algorithm mix and filter multiple frequencies to baseband RF, where AM and FM demodulation can be applied. FM demodulation is very efficient as well. On my Raspberry Pi I'm running 10 channels at present, with 4 recording audio from repeater outputs, 4 recording baseband RF from repeater inputs, and 2 FM audio channels piped into Direwolf. This consumes about 10% of one of the 4 processor cores on a Raspberry Pi 4. It is supposed to work on a Raspberry Pi 2 and Pi 3, using the on-board GPU to accelerate computation, but I've never tried it. It is also quite reliable, and I've had it running over a month at a time on the Pi.

?

The reason I was doing this was to hunt a repeater jammer in the area, and APRS was an afterthought. The jammer is now gone but new ones come from time to time so I record 4 repeaters 24 hours/day. The audio recordings are used for evidence, and the IQ recordings for analysis, especially experimentally with transmitter profiling. Experience using Direwolf made it obvious it would be a good fit - and it almost was. I now have about 40 GB of recordings. I also encouraged some fellow club members to run similar setups and devised a method by which they could upload their recordings to my system, making them all available for review or analysis. It still has some rough edges but basically works. The most difficult thing is managing security. I called the system Dragnet, which if you look at the files you'll see this as the user account for the programs.

?

The biggest issue I had, which I was dealing with today, was that Airband emits 32-bit floating point samples and Direwolf wants 8 bit unsigned or 16 bit signed integers. There are programs that would do the conversion but they are rather complex, so I decided to write one for the purpose to act as an intermediary between Airband and Direwolf. Initially I tried reading from Airband using an Internet socket, and writing to a socket for Direwolf to read but I ran into difficulties and just wrote the output to stdout and piped it into Direwolf. That worked well for APRS traffic on 144.390 and a couple of weeks ago I added the ISS APRS RS0ISS as a second channel. Yesterday I checked the logs and found that I was receiving the ISS about once per day with several transmissions during the pass and was pleased.

?

I started thinking that because I had two instances of Direwolf I would have 2 igate connections and thought that wasteful, so I decided to pursue the socket to socket sample format conversion again. This time I used Microsoft Copilot which sped things up a lot. The last thing I asked Copilot to do was add GPL V3 terms and conditions to make the code open source. After much back and forth I now have a version which will forward samples to another socket or to stdin based on the command line options. I wrote simple system services for airband, the two instances of the format converter (Terrestrial and ISS APRS) and one instance of Direwolf. It's all running nicely as I write.

?

The main components are

• Configuration files for airband and Direwolf
• Systemctl .service files for the 4 system services.
• The sample converter code, which I called f32_i16.c due to lack of imagination (maybe I should call it AirWolf)

I'm happy to share all of it and will post some stuff in the replies. Use at your own risk of course.