An Evolutionary Strategy Based State Assignment for Area-Minimization Finite State Machines

Most published results show that area reduction of the finite-state machines (FSMs) is achieved by optimizing the state assignment. In order to minimize two-level and multilevel area of FSMs, an evolutionary strategy based state assignment, called ESSA, is proposed in this study. Two cost functions (i.e. Fitness functions) are defined for two-level and multilevel area minimization. A new selection strategy and a new mutation are proposed in HES, which are specifically designed based on the analysis of the search space and individual´s distribution. The selection strategy sorts out parental individuals based on the crowding distance and fitness, and mutation uses ´replacement´, ´2-exchange´ and ´shifting´ operators, which is controlled by the hamming distance constraint, to generate offspring from the parental individuals. Experimental results show ESSA achieves a significant reduction of area to the previous publications in terms of number of cubes and literals in most benchmarks.