Re: chirp sounding

[Hassan Khalili]

Hi Fabian,

Thanks for your response. Yes, context would help. 🙂

I'm thinking about a wireless channel sounder where a transmitter (probably a USRP) is put in one corner of a room, constantly transmitting chirp signals. A separate receiver is moved around the indoor space and uses the chirps to estimate the impulse/frequency response. I've done something similar in the past in LabVIEW.

Nick Foster suggested I use the Python/numpy in GRC and I think I'm on the right path now! (until my next obstacle 😄)

Kind regards,

Hassan

---

From: Discuss-gnuradio <discuss-gnuradio-bounces+hkhalili=live.com@gnu.org> on behalf of Fabian Schwartau <fabian@opencode.eu>
Sent: 13 November 2020 18:01
To: discuss-gnuradio@gnu.org <discuss-gnuradio@gnu.org>
Subject: Re: chirp sounding

 

Hey Hassan,

sorry to be the "bad guy" to ask: Why not do it differently?
Honestly, I did not fully understand what your plan is, but do you know
the concept of FMCW radar? They also send a chirp, receive the echo
(which is exactly what you want to do as far as I understood) and then
you just have to multiply the received signal with the transmitted
signal (i.e., mix the signals), perform an FFT over the result and you
have your impulse response. Seems much less complex compared to your idea.
Typically you add a window function on the mixed signal (Hamming, Hann,
...) to reduce side lobes in the time/distance domain and maybe add some
zero-padding to the end of the signal (after windowing) to increase the
sample rate in the time/distance domain.
Keep in mind that Rx and Tx need to be coherently synchronized, i.e.,
when implemented in read hardware they need to (at least) share the same
reference osicalltor.

Best regards,
Fabian

Am 13.11.20 um 18:14 schrieb Hassan Khalili:
> Dear Gnu Radio gurus,
>
> I am trying to create a channel sounder using chirp signals where:
>
> 1) Tx side: repeatedly send chips with some guard time
> 2) Rx side: use block-wise matched filter to derive impulse response.
>
> I'm really struggling to find out if I can do this purely with Gnu Radio
> blocks (I come from a LabVIEW background). This is what I had in mind:
> 1) Tx side: Build an N-point vector (linearly increasing - one cycle of
> a triangular waveform) and feed to VCO. Concatenate with M-point zero
> vector. Repeat somehow. But I cannot find a concatenate block and I
> don't know how to create the N-point vector.
>
> 2) Rx side: Break stream into (M+N)-point vectors and correlate with
> chirp signal, use peak to adjust delay and apply matched filter. I
> cannot find a correlation block either.
>
> Any help in the right direction would be great. I cannot seem to find a
> detailed Gnu Radio tutorial and there are a lot of TODOs in the block wikis.
>
> Kind regards,
> Hassan