Don't care sequences and the optimization of interacting finite state machines

The authors consider the nature of incomplete specifications for a finite state machine embedded in a network of sequential machines. They show how limited controllability and observability of component machines are expressed in quite different ways. For the input don't care sequences, a general solution was known. The authors present extensions to it, both in terms of topologies contemplated and in terms of applicability to larger designs. For the output don't care sequences, they provide a general theory based on the concept of information lossyness and present algorithms to address the related optimization problem in practical cases. The implementation of the proposed techniques in a program called SEQUOIA (sequential optimization of interacting automata) shows that the proposed approach is viable and effective.<<ETX>>