What's the difference between a spectrogram of the Doppler domain and the Doppler Spectral Density?

Started by ReachRF 4 years ago2 replieslatest reply 4 years ago391 views

Hello colleagues, 

  Background: I've performed some measurements of Doppler shift in a mobile wireless channel with both transmitter (continuous wave) and receiver moving. In the spectrogram, I can clearly see specular and NLOS frequency components shifted due to the movement.

  Quick question here: 

  I've been using interchangeably these terms:

  Doppler Spectral Density

  Spectrogram of the Doppler Domain

  Doppler Profile

  Doppler Channel Response

  Are these terms the same thing? 

  Which phrase would you prefer?  Why? Or In What Context?

[ - ]
Reply by jbrowerDecember 2, 2018


One comment I might make would be to be careful using the word "domain".  To see Doppler shift you are transforming data from time to frequency domain ... I don't see in the literature a carefully defined "Doppler domain".  For example the wikipedia page on "Pulse-Doppler signal processing" only mentions "domain" in context of time and frequency.

Spectrogram and spectrograph are two commonly used terms for viewing a series of frequency transforms over time, maybe if there is something "Doppler unique" about the frequency transforms you make to get that data, it might be a transformation into "the Doppler domain".  I do see mention of "range Doppler domain" so maybe that's it, but in that case it looks like FFT plus a combination of windowing and filtering before, and mapping *after*, the transformation into the frequency domain.

I could easily be wrong on this, but does seem to jump out at first glance.


[ - ]
Reply by ReachRFDecember 5, 2018


  My sincere gratitude for your reply. Thoughtful, and insightful.

  This is where I have struggled, too. In the literature, I first heard of "Doppler domain" from Skolnik, Introduction to Radar Systems (2nd Edition).

  Where we can see that the origin is always 0Hz, and any increment (+/-) is the relative shift in carrier.

  Yet, I have yet to see this used in other literature.

  Your response has me leaning towards, "time-variant Doppler Spectral Density" (DSD), since I have seen that used recently in channel measurement literature (albeit circa 2009).

  I have also come across, "Frequency-time Doppler profile" as recently as 2013, but that's a lot to write out each time, and FTDP looks weird.

  So I'll stick with calling it: the time-variant DSD, even though the Density aspect doesn't really make sense to me since I'm using a sliding FFT (after removing carrier frequency offset). To me density makes more sense when computing probabilities.