Technical discussions about the TI C6000 DSPs (including the c62x, c64x and c67x DSPs).
|
Hi everyone!.. I'm trying to compute a cross correlation of two signals; one is a MLS signal..a pseudo random signal ,periodic with length (2^n)- 1 ..the other one is the system's response to this mls signal; I'd like to cross correlate this two signals to get the impulse response ..but I think that the usual cross correlation is too slow (N^2)..someone told me about the Fast Hdamard Transform (NlogN)to speed up the computation,but I can find a way to get this work out..every ''transform'' requires a signal with length of L=2^N.. how can I do this with my MLS signal??..notice that I can not add a zero to the MLS sequence..it would change everything..loosing every advantage of MLS... Please ,I'm desperate... :-( Thanx very much!! fran |
|
|
|
Hi Fran, I would use FFT based cross correlation. Good book for details is David and Stearns'book Signal processing Algorithms in Fortran and C. But I think TI's web site has examples for "fast-crosscorrelation". Basically, the coross correlation will be carried out by a frequency domain multiplication of the FFT of both signals. Regards, /Khalid |