modelling AWGN question
In my simulations up to now using BPSK signaling, the AWGN channel was modelled by simply adding a gaussian real value to the transmitted symbol. However, I came across some matlab AWGN modelling like this one which add a complex value to the transmitted symbol
function Y =awgn(X, sigma)
% AWGN GENERATOR
% 07-31-1993
% Awgn is an m-file that adds awgn noise to the matrix X
% Where sigma is standard deviation of the noise
% Ec/No- 1/(2*sigma^2).
[rr,ccJ =size(X);
W =randn(rr,cc)+ i*randn(rr,cc);
Y=X+sigma.*W;
disp('AWGN IS ADDED TO SIGNAL');
(From Mehmet Kutlu: Kalman Filtering Approach to blind equalization).
My question is the following: Why is it necessary to add a complex noise value while we are dealing with BPSK symbols which are +1 or -1? Is it because of the subsequent filtering? How does this translate into physical (real electrical) quantities? Any link to a book or an article that explains this in details will be appreciated.
Thanks GA