A customized genetic algorithm with in-loop robustness analyses to boost the optimization process of analog CMOS ICs

Author(s)

de Lima Moreto, Rodrigo Alves ^[1] ; Thomaz, Carlos Eduardo ^[1] ; Gimenez, Salvador Pinillos ^[1]

Affiliation

^[1] FEI Univ Ctr, Dept Elect Engn, Sao Paulo - Brazil

• Document type: Journal article
• Source: Microelectronics Journal; v. 92, OCT 2019.
• Web of Science Citations: 0

Abstract

The traditional optimization processes of analog complementary metal-oxide-semiconductor (CMOS) integrated circuits (ICs) are very complex, slow, and based on the designers' experience. To obtain robust potential solutions, it is necessary to perform robustness analyses (RAs) through SPICE simulations. However, this approach represents a huge bottleneck in the optimization processes due to the significant increase of time of the SPICE simulations concerning the RAs. Therefore, this work proposes an innovative customized genetic algorithm (GA) to boost the optimization process of analog CMOS ICs. The main results obtained showed that all designs of analog CMOS ICs reached a yield of 100% and a remarkable reduction of the optimization time (from 23% to 79%) in comparison with the standard optimization process with the GA, without reducing the random samples number considered in the RAs, and consequently preserving their robustness accuracy. (AU)

• FAPESP's process: 18/21341-4 - Prototype of interactive computational intelligence for the design and optimization of analog CMOS integrated circuits
• Grantee: Rodrigo Alves de Lima Moreto
• Support Opportunities: Research Grants - Innovative Research in Small Business - PIPE