why is fourier transform of AWGN is of the form complex

37 Views Asked by At

This is related to this Nice answer by the thread

What is the Fourier transform of $f(x)=e^{-x^2}$?.

Considering this derivation of fourier transform of Gaussian with mean 0 and unit variance, why do we get complex data type when I take fourier transform of AWGN (Addititive white gaussian noise) ?. (You can try in matlab by taking FFT of awgn function.)

"""

no_sig = zeros(128,256);

no_sig_noise = awgn(no_sig,0);

no_sig_fft = fft2(no_sig_noise);

"""

if you observe no_sig_fft, it has complex form (a+jb). My question is : As per the derivation, the Fourier transform of Gaussian distributed data is : $e^{-w^2/4}$ , which will give real data. But when i do fft of AWGN, i get complex data ($a+jb$). Why is this so. Anything i have missed in understanding?