A plan for distributed SDR decoding

Foundations of Amateur Radio

Yesterday I finally discovered the missing piece of information that will allow me to create a project that I've, if not outright spoken about, at least hinted at.

In an ideal world by now I'd have built a proof concept and would be telling you that I've published a GitHub repository under my callsign for you to explore. If wishing made it so. Unfortunately, currently sitting at a keyboard for anything longer than ten minutes or so makes it nigh on impossible to stand up, so you'll have to make do with hand waving and gesticulation rather than actual code, but for now, that's all I have. Consider this a design specification if you're so inclined.

So, big idea.

Imagine that you have a device that can listen to radio frequencies. This device is connected to a network and it shares the data to any number of different listeners which might each do something different with the information.

If you were to do this in the way we watch YouTube or listen to streaming audio, each listener would get their own unique copy of the data. If you have ten listeners, you'd have ten streams crossing your network, even if everyone was enjoying the exact same video or audio at the exact same time.

Instead I want the data coming from the device to have only one stream on the network and for as many different listeners or clients to access it as required at the same time.

Let's get specific here for a moment. I'm talking about using a software defined radio, could be a $25 RTL dongle, could be any SDR, that is tuned to a part of the spectrum, let's say the entire 40m band, and sends that radio information digitally onto the network. This network could be your local network, or it could theoretically be the internet, for now, let's just put it out on our own network.

So, you have a copy of the entire 40m band streaming across your network. Great, now what? Well imagine that you want to decode RTTY on 7.040. You fire up your decoder, point it at the network stream and decode RTTY.

Then you want to decode a WSPR signal, at 7.0386. You fire up your WSPR decoder, point it at the network stream and decode WSPR.

Then you want to decode FT8 on 7.056, same deal, fire up your decoder, point it at the network stream and decode FT8.

Now you want to compare two different RTTY decoders. Fire them both up, point them both at the same stream, decode both, simultaneously.

Of course you could do this with CW signals, with SSB signals, with any decoder you have lying around, Olivia, Hellschreiber, AM, Packet, whatever. All these decoders could be running independently but together on the same band.

You could add a tool that shows a waterfall display of the same data on a web page, or play some of the decoded data to your headphones, or record it to disk, or do spectral analysis, all at the same time.

The information that you're processing is on the network once. You don't have to flood your network with multiple copies of the 40m band, the only limit is how much CPU power you can throw at this and to be frank, most computers on the globe today spend much of their time waiting for you to do something, so processing a bit of data like this is not going to tax anything built in the past 20 years or so.

The missing ingredient for this was a Linux tool called netcat, or nc. It allows us to distribute the information across the network using a technique called broadcasting.

So, RTL dongle, data extracted by a tool called rtl_sdr, distributed across the network using netcat and used by as many clients as you can think of.

The proof of concept I'm working on uses Docker to build a bunch of different containers, or clients if you like, that each can do a different task with the same stream. When I've got something to show and tell, you'll find it, predictably, on my GitHub page.

Oh, if you want to run the same thing for say the 80m band, you can. Now you have two network streams, one for 40m, one for 80m and as many decoders on your network as you have CPU cycles to play with.

If all this sounds like magic, you've seen nothing yet.

I'm Onno VK6FLAB