A double-layer genetic algorithm for Gm-C filter design

Abstract

Although analog circuits play an important role in Systemson-a-chip, their design is effort and time consuming. Automated design methodologies are elaborated to overcome drawbacks resulting from human design. This paper proposes a double-layer on-line genetic algorithm-based optimization method for use in the automated design of Gm-C filters. To accomplish on-line circuit evolution, a Matlab-Eldo interface is proposed for communication of the GA with the circuit simulation environment. After a presentation of the Gm-C filter with an analysis of filter tunability, the two layers of the evolution are presented: raw filter design and fine-tuning of the filter characteristic. Simulation of the evolutionary algorithm proves the efficiency of the double-layer approach in reducing design time for a GA-only optimization technique. © Springer-Verlag 2010.