Subsample Delay Estimation: y=fft_circshift(u,d) - File Exchange

Code covered by the BSD License

Highlights from
Subsample Delay Estimation

[e f_e G P]=zzb(s,n1,n2, f_s,... calculates the modified Ziv-Zakai lower bound for time delay estimation.
[x_max y_max A]=crit_interp_p...
delay=delayest_3point(u2,u1,m... delay=delayest_3point(u2,u1,method,estimator,parameter);
delay=delayest_fft(y,u); delay=delayest_fft(y,u);
delayest_iterative(u2,u1,N_i_...
x=rand_white(N);
y=fft_circshift(u,d) y=fft_circshift(u,d)
y=fourier_series(Y,t);
compare_delay_methods_6_03.m
simple_example.m
View all files

from Subsample Delay Estimation by Travis Wiens
Demonstrates a number of methods of estimating the delay between two signals to subsample accuracy.

y=fft_circshift(u,d)

function y=fft_circshift(u,d)
%y=fft_circshift(u,d)
%This performs a circular shift of d samples on u, using a linear phase
%shift in the Fourier domain. The shift, d, need not be an integer.  Some 
%information is lost if there are an even number of samples and there is 
%frequency content at the Nyquist frequency, since this frequency cannot 
%be shifted and is disregarded.

%There is probably room for speedwise improvement in this code.

%    Copyright Travis Wiens 2009 travis.mlfx@nutaksas.com

U=fft(u);

N_p=numel(u);

if (mod(N_p, 2) == 0)
    U(length(U)/2+1)=0;%disregard Nyquist frequency for even-sized dft
end

f=(mod(((0:N_p-1)+floor(N_p/2)), N_p)-floor(N_p/2))/N_p;
c_shift=exp(-2i*pi*d .*f);
Y=U.*c_shift;
y=ifft(Y);

Highlights from Subsample Delay Estimation

Highlights from
Subsample Delay Estimation