Design and optimization of digital circuits by artificial evolution using hybrid multi chromosome cartesian genetic programming

Abstract

Traditional digital circuit design techniques are based, for the most part, on top-down methods, which use a set of rules and restrictions to assist the construction of the project. Genetic algorithms, on the other hand, haven proven themselves to be a very useful tool for solving high complexity problems, relying on a bottom-up methodology. This paper proposes a new design algorithm, named HMC-CGP, which operates by first finding a functional solution by using the MC-CGP method. Then, optimizes it by using standard CGP approach. Test circuits used include 1 and 2 bit full adders, 2 bit multiplier and 7 segment hexadecimal decoder. Obtained results show that by making use of faster convergence granted by the MC-CGP mechanism together with an optimization strategy generates novel approaches for those circuits, with results showing a logic gate and transistor usage reduction of up to 60.8%.