This paper introduces a simulation-based differential evolutionary optimization algorithm for design of analog integrated circuits. This algorithm happens to be much faster than other none derivativebased algorithms mentioned in literature. A 2.4GHz CMOS Gilbert Mixer has been chosen in this case. In order to evaluate the time of convergence, the algorithm is compared with a GA optimization procedure, using automated iterations of MATLAB-APLAC in a 0.18μ process. The differential evolutionary algorithm proves to be several times faster, even for smaller population sizes.
Hamidreza Rezaee-D. , Amin Ibrahimy-K. , Mohammad Danaie
Department of Electrical Engineering,
Ferdowsi University of Mashhad