I cannot follow your explanations. I start with cleaning your code:
simulations = 10000;
n_vec = 1:2:20;
R = zeros(1, simulations); % Pre-allocate
% Move constants out of the loop:
B = 0.3;
PRn = 0.0002224 ;
Bbn = 20*1000;
Kn = 0.6;
Bandwidth = 100*1000;
Psi = Kn*Bandwidth;
No = -174;
for in = 1:numel(n_vec)
n = n_vec(in);
phi = (1 - B) ./ n;
tau = B ./ n;
Rbn = phi * Bbn;
BR = n .* Rbn;
Ehn = (1 - B - phi) * PRn;
Ptr = Ehn/tau;
count = 0;
for sim = 1:simulations
TR = n .* tau * Psi * log2(1 + rand * Ehn / tau / No);
RTotal = BR + TR;
count = count + RTotal;
end
S = count ./ simulations;
R(in) = S;
end
plot(n_vec, R)
Yes, this looks almost like a straight line, but is is none:
diff(R)
You see, the differences are from -0.076 to -0.085. This is exactly what the given code instructs Matlab to calculate. The effects of the rand value are damped massively, by dividing it by a large value and adding 1 - and by takeing the log2 afterwards:
TR = n .* tau * Psi * log2(1 + rand * Ehn / tau / No)
= n .* tau * Psi * log2(1 + rand * ([-5.9e-6 to -5.9e-5]))
So if the value of rand is 0 or 1 changes the value of TR marginally only.
There is no chance to guess, why you expect another result or why you assume, that there is a mistake.
