7
$\begingroup$

If I were to post a question with a screenshot of a waveform, or a table of datapoints, and ask for suggestions as to what protocol that signal was using, would that be on-topic here?

I'm not sure if your community would consider it too localized to be a good question. If it would be permissible under certain conditions, please let me know what data you'd want in a question of that type.

$\endgroup$
4
$\begingroup$

I think it's borderline, but what I'd expect to see in a good question like that is the picture / data (I personally like having the data or part of it available) and an explanation of what you've tried so far in terms of answering your question... and why that doesn't satisfy you.

Just dumping a signal on the community and asking "What's the protocol?" makes it closable as "too broad" for my liking.

The more information the better.

When composing such a question for myself, I've often found it clarifies my thinking and, while not allowing me to answer my own question, it helps others to answer me in a short space of time.

$\endgroup$
3
$\begingroup$

I personally think this is the signal processing StackExchange site, not the process my signal one.

Therefore, I'd say: Everyone should be very encouraged to ask on how to identify a signal; I think it's perfectly possible to show a signal, give all important information about what it is, how you've got it, and then ask a question, explaining your own attempts and problems, that asks how to classify that signal.

Note that I'm not asking for the demonstration of an attempt just because it weeds out the "do my homework for me" folks (it certainly does that, too), but because it'll allow people to answer precisely, use knowledge that you already have, and introduce only the concepts that are new to you.

For example, it'd be very different if someone asked:

below's a screenshot of my signal. What is it?

and

Below is a plot of the I and Q components [plot with sensible axes labels] of a baseband signal that I mixed down from 1872 MHz.

The sampling rate was 2 MS/s, and as far as I can tell, things should have happened within the linear range of all components, and without significant aliasing.

Looking at that, I first thought that it might be some FSK variant, so I made a waterfall plot (1024-FFT, 50% overlap, below), but it's nearly a rectangle in frequency domain, but not quite. In an attempt to find rough temporal characteristics, I tried to calculate an autocorrelation of the mag-square signal, but that led to nowhere, as it only showed a DC peak and nearly white behaviour around that, plot below.

Because the latter not only rules out a lot of things that one might consider first, and, even more importantly, explains the tools that OP knows how to use. It also shows things that an expert my spot as "mistake" or misconception, and make it very much easier to answer the question.

Don't fret if your toolset isn't that rich (or you don't understand many of the terms used in the second example I gave) – it's fair to ask if you don't know something! But then, it's probably even better for you if someone shows you how to see something yourself. That's a win-win!

Also, if possible, share samples. I'm personally on a small crusade against sharing millions of values as CSV, but if that is your format of choice, it's certainly better than nothing, and I'd be happy, and no-one will downvote you for sharing too much info or info in a suboptimal package :D

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .