Here is the problem.
I am not able to get rid of some of the subscripts since they make the original problem non-convex. But I tried to simplify that as much as possible and I have highlighted the variable of interest (or the functions that contain this variable) in the objective function and constraint.
the variable of interest is pitilda with the size of (K,M,J,i) and the following optimization problem is supposed to find the optimal value for ptilda_{k,m,j,l}.
.
My question is, can I use the cvx in the following loops (since I have K * M* J * I objective function now)
for k = …
for m = …
for l = …
for i = …
cvx
end
end
end
end
or there is a better way to implement that?
Is it correct to use rel_entr for converting objective function and constraint in the above problem as follows
objective function = \alpha * \beta * -rel_entr(1,1 + \chi*SINR/\beta) + a ptilda
constraint = sum(\gamma * \lambda * -rel_entr(1,lambda /(lambda +\chi_tilda sinr)))
Thanks in advance