Efficient suboptimal state assignment for large sequential machines

Optimal state assignment is one of the most important problems in the automatic synthesis of sequential machines. State assignment approaches can be divided into: structural (that construct ´suboptimal´ assignments) and statistical (that generate assignments). The author has developed and implemented a new structural method. Analyzing the factors that influence the quality of assignments and performing experiments, it is concluded that the distribution of solutions is close to a normal distribution. Some probabilistic quality measures for assignments and assignment methods are introduced and applied to the structural and random approach. Both approaches are compared and it is indicated that the difficulty in evaluating the quality of an assignment by statistical approach is the main reason why structural approaches can be more efficient than statistical ones.