Cross-Correlation between signal and the delay version

[arash rezaee]

Hi everyone. I have an OFDM signal and I simulated my feedback signal with delay version of input signal. So now I want to calculate the amount of delay. I know we can do it with matlab function xcorr but I want to do it with FIR filter which the coefficients are the reverse of input signal. I calculate with both Xcorr function and with FIR filter but the FIR filter answer is not correct!!!!!!!!!! what can I do? Please help me guys. Here is my code:

T = (7e-6/64)/2;
Fs = 1/T;

ofdm_symbol = func_ofdm_generate('16qam', '8k', 1/32, 1);
x = abs(ofdm_symbol)/max(abs(ofdm_symbol)) .* exp(j*phase(ofdm_symbol));
R_input = real(x);
I_input = imag(x);
h = dfilt.delay(5); %make delay version
FB = filter(h,x);
R_FB = real(FB);
I_FB = imag(FB);

xcorr_signal=filter(conj(R_input(64:-1:1)),1,R_FB(1:1:64));
%D = finddelay(R_input,R_FB);
K=xcorr(R_input(1:64),R_FB(1:64));%calculate by matlab function
delay1 = 64-(find(K==max(K)));
delay2 = 64-(find(xcorr_signal==max(xcorr_signal)));

Best Regards,

Arash

 

[milind.a.kulkarni]

Hi Arash,

I try to run your code on my machine but got the error 'func_ofdm_generate' not found.....Can you post the code for this function such that I can run it and give you the feedback

Also, you are using a function 'finddelay' ....however it is in comments... can you put that code too.....

with regards,

milind

 

[arash rezaee]

Hi here is the code for OFDM Function. By the way the finddelay is matlab function.

function y = func_ofdm_generate(qam_mode, ofdm_mode, GI, N_symbol)
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
switch qam_mode
case 'qpsk'; M = 4; case '16qam'; M = 16; case '64qam'; M = 64;
end

switch ofdm_mode
case '2k'; N_fft = 2048*2; n = 1705;
case '4k'; N_fft = 2048*4; n = 3409;
case '8k'; N_fft = 2048*8; n = 6817;
end

scale = modnorm(qammod(0:M-1,M),'peakpow',1) ;
data = randi([0,M-1],1,N_symbol*n);
data_qam = scale * qammod(data,M);
x = reshape(data_qam,n,N_symbol);
y = [x((n-1)/2+1:end,;zeros(N_fft-n,N_symbol);x(1n-1)/2,];
Y = ifft(y, N_fft);
Y1 = [Y(end-GI*N_fft+1:end,;Y];
y = reshape(Y1,1,N_symbol*N_fft*(1+GI));

Thanks

 

[milind.a.kulkarni]

hiii Sorry once again ....


I could not found function 'modnorm' and 'qammod' in the function func_ofdm_generate....... Please post your complete code then only one can run this and give you some feedback.....


Sorry for that I went to mathworks site I think these are the function from "Communications System Toolbox" ...... I don't have that with me.....I will not be able to help you out

Also you are using 'dfilt' from 'Signal Processing Toolbox'.... that also I even do not have.....

Only one thing I can tell from my signal processing background and after going through your code...... you want to do cross-correlation..... and as I understand your code the input and feedback both signal are complex number array...why are you only operating on the real parts of it..... I mean if you want to do cross-correlation ..... I think what you can do is flip the signal ( not the complex number signal but the real value signals generated before even converting it to complex number array ) and then convert it to complex number based transform and them do the dot multiplication ...... as correlation will be done by means of convolution in frequency domain..... I hope you understand what I mean.....

With regards,

Milind

 

[arash rezaee]

Ok. just ignore the ofdm generation and just add random matrix and then delay it. again you can see the output of filter is not equal to delay I just added. Please comment OFDM generator and try to add one matrix by your self.

Regards
Arash