[Discuss-gnuradio] Need help with calculation of channel impulse response

[Rush]

Hi @all.

Maybe I should introduce my project first: I want to make a flowgraph that
prints the channel impulse response of an OFDM(DAB) ensemble on a Scope
Sink.

CIR = IFFT {Received PRS x PRS*}

What I have done is:
1. wrote a matlab script that generates a binary file with samples of the
phase reference symbol described in
http://www.etsi.org/deliver/etsi_en/300400_300499/300401/01.04.01_40/en_300401v010401o.pdf
page 147

2. made a FlowGraph that takes the samples from the file and does the
calculation described above. For testing purposes I made CIR = IFFT {PRS x
PRS*}. My supposition was, that I will get a signal of 0 because the
multiplication eleminiates it. But there is one high and zwo smaller peaks
on the right and left.

3. The sample file contains 1536 float samples, one for each carrier
frequency. So, if I put all these samples into one Vector, does the FFT,
multiplication and IFFT my Scope Sink should print a line at 0 right ?
The same but with a real received phase reference symbol should generate one
single peak when there is no phase difference. Or am I wrong?

What am I doing wrong. Why don't I get a constant 0 signal at the Scope? And
how to calculate the time axis on the scope that it does not "roll"?

prs_float_ReIm.bin is the sample file of the PRS. first 4 bytes -> real
part; next 4 bytes -> imaginary part and so on

prs_mal_prs.grc
<http://gnuradio.4.n7.nabble.com/file/n60375/prs_mal_prs.grc>  
prs_float_ReIm.bin
<http://gnuradio.4.n7.nabble.com/file/n60375/prs_float_ReIm.bin>  




--
View this message in context: 
http://gnuradio.4.n7.nabble.com/Need-help-with-calculation-of-channel-impulse-response-tp60375.html
Sent from the GnuRadio mailing list archive at Nabble.com.