Evolvable Metaheuristics on Circuit Design
Felipe Padilla, Aurora Torres Soto, Julio Ponce, Mara Dolores, Sylvie Ratt, Eunice Ponce-de-Le · InTech eBooks · 2011
genetic algorithms and genetic programming, electronic hardware implies not only digital but analog circuits also.This field has earned importance since the early 1990´s because of the advent of reconfigurable hardware.The ultimate objective of this field is to design and construct intelligent hardware, capable of online adaptation (Yao and Higuchi, 1999).The first classification of evolvable hardware can be found in (De Garis, 1993).In this work De Garis established there are extrinsic and intrinsic EHW.While Extrinsic EHW simulates evolution by software and downloads to hardware only the best configuration; intrinsic EHW simulates evolution directly in hardware.Nowadays the scope of this discipline has grown vastly.According to Zebulum (Zebulum, 1996), evolvable hardware can be classified by several criterion like hardware evaluation, evolvable computation approach, application area and evolvable platform.In regard to its application area EHW in divided in: Circuit design, robotics and control, pattern recognition, fault tolerance and very large scale integration (VLSI).We are interested in discuss about the first one.Circuit design is the art of constructing a sized circuit from user specifications (Das and Vemuri, 2009).This task is divided according to the kind of circuits that are handled in digital and analog circuit design.Nowadays there are different algorithms that can be used to solve problems of optimization of circuits like: Genetic Programming, Genetic Algorithm, Estimation of the Distribution Algorithms, Ant Colony Optimizations, Others.The more amenable nature of digital circuits made researchers like Louis (Louis, 1993) and Koza (Koza, 1992) to focus first on the production of functional logic circuits.Afterwards, the goal was not only to obtain functional circuits, but optimum ones.The work of Louis (Louis, 1993) was pioneer on the use of genetic algorithms on the design of combinational circuits; Thompson et al (Thompson et al., 1996) were the first in coding logic gates and its connections.Other outstanding researches on digital design are Higuchi et al. (Higuchi et al., 1996) specially focused on intrinsic evolution based on neural networks; Hernández and Coello (Hernández and Coello, 2003) first worked with genetic algorithms and later with genetic programming and Information Theory.A very interesting case is the use of ACO on the optimization of combinatorial circuits (Mendoza, 2001).The analog synthesis world also has numerous successful implementations of different