Hi everyone, I am working on Space-Time trellis codes over frequency selective channel. Currently, my decoder section for STTC uses constant quasistatic fading channel parameters which are known by the receiver. Here is a part of the code of this decoder procedure.
X = STTC_Modulator(Tr, M, Scheme);
hX = (H * X);
BM = zeros(num_of_Symbols , numOfStates * M);
for k = 1 : num_of_Symbols
for n = 1 : Nr
BM(k, :) = BM(k, :) + ( abs( r(n, k) - ( hX(n, :) ) ) ).^2;
When I use constant channel parameters, my H matrix is Nr x Nt size and hX matrix is defined according to perfect channel estimation with the use of example STTC encoding steps. With this, for 4 states, I have 16 trellis symbols to follow from X. H is Nr x Nt size and X is Nt x 4^2 size. After this, there is no problem to find Branch metrics BM matrix and going further. However, my problem here is as I use frequency selective channel, I have no longer constant size of such as Nr x Nt size H matrix. So, I have a problem to decode further. I get channel parameters individually. For example Nr=1, Nt = 2. That is why, I have H1 and H2 in simple terms. As I pass through frequency selective channel, I have impulse response of H1 and H2 with the equal size of transmitted signals and in this case, I am stuck. Help me on this please. How can I implement decoding over 2taps-3taps channels?