A systematic EHW approach to the evolutionary design of sequential circuits

Abstract

The main difficulty in the evolutionary design of finite state machines (FSMs) is lack of effective systematic EHW approach. To accomplish the evolutionary design of FSMs, a systematic EHW method named genetic programming–evolutionary strategy (GP–ES), which is a combination of ES and GP, is proposed. ES optimizes the state assignment and provide them to GP for population generation; GP is responsible for evolving the combinational part of FSM, and feeding the fitness of population back to ES for the evaluation of corresponding state assignments. GP–ES is tested extensively on twenty FSMs from MCNC Library. The results demonstrate that the GP–ES-derived state assignments are more efficient than the ones of Xia, Ali, Almaini and NOVA in the evolutionary design of FSMs. The results also illustrate that the GP–ES is superior to conventional synthesis tools in terms of complexity reduction for the design of small and middle FSMs. GP–ES also performs well in comparison with 3SD-ES in most cases.