You re not setting up an LMI as you appear to think.
You have bilinear stuff in Sys1 and Sys2 coming for instance from the product Bf'*P12
>> D'*Bf1'*P12
Bilinear matrix variable 1x4 (full, real, 14 variables)
Coeffiecient range: 0.1 to 0.1
Hence, you cannot solve this problem (bilinear semidefinite programming is much much harder than linear, and there is basically no tractable way to solve them unless you develop your own dedicated solver).
BTW, once you actually want to solve a linear SDP, install a better SDP solver and don't use lmilab.Best you can do is to install mosek.