How do I go about identifying an unknown digital signal?

Question

I'm playing around with an SDRPlay. I'm using the CubicSDR app, and I'm tuning around the HF bands. I get to 5.650 MHz and I see this:

This looks like a bunch of discrete RTTY signals. How would I go about identifying and decoding some of these signals?

• Audio sample 1
• Audio sample 2

Things I tried:

1. I looked through the Globaltuners Frequency Database. 5600 kHz appears to be an aeronautical band:

According to this site:

MWARA is the term for Major World Air Route Areas, which support HF Radio communications to aircraft outside of VHF range.

2. I looked around for aeronautical data formats, and found HFDL, however these signals look and sound nothing like that format.

3. I opened up the Signal Identification Wiki and looked for signals that appeared to have the same frequency spectrum. Each of these signals seems to have a clear signal of two tones, which confirmed my suspicions that it might be RTTY.

4. I ran fldigi in RTTY mode and tried to decode one of the signals, but it didn't produce anything intelligible. But I may be in the wrong mode, or this is a proprietary protocol, or ... ?

How can I systematically what this signal might be? For example step 1 is it FSK? Step 2 how many tones does it use? Step 3 what is the frequency separation? Step 4 what known protocols have these parameters?

Are there any software packages that can do this?

Answer 1

Unfortunately, I don't know of any free and easy to use software that does all the signal identification you need.

I do know of some free programs that can help you with some of the steps involved.

One is Visual Analyzer. It can show you the spectrum of your audio over time. By looking at the spectrum of your signal, you'll be able to see the number of tones and their individual frequencies, if your settings are correct.

Another option, which I personally didn't use all that much is Spectrum Lab. It provides similar environment to Visual Analyzer, but has some extra options related to very fast and very slow Morse code.

Using the programs I mentioned, you'll have a chance of seeing what type of modulation it is. In your case, it's a 2-FSK, which is visible from two clear peaks in the spectrum. The frequency separation can be directly read from the spectrum views programs provide.

One piece you didn't ask, but is important, is the symbol rate. As of yet, I don't know a smart way to determine it. One thing that helps, when you have high signal-to-noise ratio is to use matched filters in FL-DIGI. Usually, the tone filter size on the spectrum will match the bandwidth, but this is still very rough method. Another thing I used was to load the files in Audacity and try to guesstimate the symbol period by looking at the waveforms. The bad side of both is that they're very rough.

Don't forget that it's possible that your high and low frequencies could be inverted, compared to what the demodulator in FL-DIGI expects.

For the part 4, usually you'll need to search the Internet using the shift, baud rate and frequency and hope that a protocol will come up.