Post by gabriel on Dec 27, 2009 13:26:48 GMT -5
I am working on spatial channel modeling. My problem is that I generated a channel matrix per tap (=path) : H_tap of dimension [Tx, Rx], where Tx and Rx is the number of Tx and Rx antennas respectively. For a system level simulation I have to calculate the simulated OFDM channel H with dimension [Tx Rx NoCarrier] (without taps). I did the following: I put all complex pathgains into a vector Hn and calculated: H = fft(Hn, noCarrier). Is this correct? The problem I have is that the H_tap have different path delays; i.e. the path delays are not equidistant. Do I have to account for that? What is the impact of the sampling frequency? I read that I have to interpolate the values. This is not clear for me. What happens if 2 or more paths fall into the same sampled tap? Do I have to add up the corresponding channel coefficients? What about unused sampled taps? Should I put a zero in it?