From: Nitram on
Hello,

I am trying to correlate a circularly-shifted sequence in frequency
domain: i.e. I take a sequence that has been circularly shifted in time
domain, do an FFT on it and try to correlate it in frequency domains. The
time domain sequence is shifted by a random amount of samples.

Since a circular shift in time domain is a phase shift in frequency
domain, one thing I could do is multiply my correlation sequence by all the
possible phase shifts. This seems pretty expensive (if the sequence has N
possible circular shifts, I would have to do N correlations).

Therefore I was wondering if there was a faster algorithm to correlate in
frequency domain for circular shifts in time domain.

Thanks.