Is this a valid way of computing autocorrelation?

Question

I am a beginner in Digital Signal Processing and I was trying to compute and plot an autocorrelation.

I've wrote this piece of code:

r = [zeros(2,1); y(1:98,1)];
r = r.*y;

and I wished to know if this is a valid way of computing an autocorrelation.

Answer 1

• Your code contains magic numbers, avoid them
• To calculate the autocorrelation, you do not add zeros. y(1+lag:end).*y(1:end-lag) would match the definition.
• Multiplying requires your data to be binary -1 or 1, not 0 or 1. I would use a version which processes logic arrays.

Answer 2

How about xcorr(y)? The xcorr() function is part of the Signal Processing Toolbox.

• Mauren: I was trying to calculate by hand, actually.

Answer 3

Are you trying to calculate the correlation factor or the cross correlation function?
The correlation factor is basically a normalized dot product of two vectors.
The cross correlation could be calculated using the conv function (By flipping one of the vectors).