Evolutionary algorithm for state assignment of finite state machines

Abstract

The paper proposes an evolutionary algorithm (EA) for the state assignment problem (SAP). Two original crossover operators are presented. They are experimentally compared with other known crossovers for SAP using a set of benchmark finite state machines. Solutions generated by EA (using different crossover operators) are compared with the random ones and with the state assignments generated by the MAX+PLUS II system. Experimental results show that the solutions found by EA are significantly better (up to 55%) than those from MAX+PLUS II. Moreover, EA equipped with proposed crossover operators found better results than those obtained with the use of other compared crossovers.