Sliding DFT with an IQ stream

[Harvs]

I'm looking at using a sliding DFT for demodulating a FSK signal on a uC.  The Real input single bin DFT is simple and there's loads of examples on the net. But I have a hardware system that already has both the I & Q available from RF down-conversion that I'd like to use.

Has anyone seen an explanation of the Sliding DFT or similar for a quadrature signal?  It feels like it shouldn't be terribly difficult, but google is returning blank.

[moffy]

Not sure if it's helpfull but the following article has an analysis of the DFT with respect to IQ signals: http://www.hyperdynelabs.com/dspdude/papers/quadrature%20signal%20processing.pdf

[Nominal Animal]

You do complex DFT, with I being the real part, and Q the imaginary part.

Unlike in the real-input DFT, this time the frequency domain does not have a conjugate, and you want to look at all N bins (instead of N/2 bins, as you do with real input).

Because you have twice the data compared to plain real inputs, you get twice the results, and the calculation takes more MCU/CPU time.

[jbb]

You could also consider a Phase Locked Loop (PLL) based detector for FSK demodulation…

[Harvs]

Thanks, after reading those links and pondering Nominal Animal's response it sunk in that all it was is the same DFT equation with the complex stream as an input.

That's all quite straightforward to process, only adds another couple of multiplies to get the full Fs bandwidth.